The Immune Checkpoint Inhibitors Journey: From Early Promise to Lasting Impact

Abstract

Immune checkpoint inhibitors (ICIs) have transformed cancer therapy, yet their clinical impact is limited by immune-related adverse events (irAEs), therapeutic resistance, and the lack of reliable predictive biomarkers, contributing to a shift from early promise to a therapeutic plateau. irAEs affect multiple organ systems and may lead to treatment interruption and significant morbidity. Emerging strategies emphasize phenotype-driven and steroid-sparing approaches to control toxicity while preserving antitumor efficacy. Concurrently, primary and acquired resistance remain major challenges, driven by tumor-intrinsic mechanisms, immune microenvironment alterations, and host factors. Furthermore, current biomarkers, including programmed cell death ligand 1 (PD-L1) expression and tumor mutation burden, demonstrate inconsistent predictive performance across tumor types. Advances in immune profiling, genomics, and microbiome research are enabling more precise patient stratification and informing novel therapeutic strategies, including rational combinations and targeted immunomodulation. This position article synthesizes key barriers in ICI therapy while highlighting emerging opportunities to refine patient selection, improve safety, and enhance therapeutic durability. Together, these advances position the field to move beyond the current plateau toward a more precise, effective, and patient-centered era of immuno-oncology.

Graphical Abstract

INTRODUCTION

Although the role of the immune system in recognizing and responding to foreign antigens, including cancer, has been appreciated for more than a century, the clinical exploitation of immune regulatory “brakes” to treat cancer became a reality only just over a decade ago with the advent of immune checkpoint inhibitors (ICIs).[1] Antibodies targeting the programmed cell death-1 (PD-1) pathway and its ligand, programmed cell death ligand 1 (PD-L1), have fundamentally transformed the landscape of cancer therapy and are now used across multiple malignancies, both as single agents and in combination with chemotherapy, targeted therapies, and other immunomodulatory approaches.[2] Therapeutic agents targeting additional immune checkpoints, including cytotoxic T-lymphocyte–associated antigen 4 (CTLA-4) and lymphocyte activation gene-3 (LAG-3), have also received regulatory approval, further expanding the immunotherapy armamentarium.

The remarkable clinical success of checkpoint blockade initially generated extraordinary enthusiasm, fueling an explosion of clinical trials and unprecedented investment in immuno-oncology drug development. Despite the evaluation of numerous immunotherapeutic agents across a wide range of tumor types, many trials failed to meet expectations.[3] As enthusiasm within the biotechnology sector has tempered, the field has entered a phase often described as a transition from promise to plateau.[4]

In this position article, we focus on ICIs and examine the factors that have shaped this trajectory. The challenges associated with immune checkpoint blockade reflect a central paradox: the same immune activation that can generate durable tumor control may also produce clinically meaningful toxicity, whereas in other patients, it fails to elicit sufficient antitumor activity. Understanding the current plateau in ICI therapy, therefore, requires examining the major clinical and biological barriers that limit its impact. In this position article, we explore these challenges, including immune-related adverse events, mechanisms of primary and acquired resistance, limitations of existing biomarkers, and considerations in special patient populations. We also discuss emerging strategies to optimize treatment delivery through modifications in dose, interval, and duration, as well as novel approaches aimed at maximizing efficacy while minimizing harm. Together, these topics provide the foundation for an integrated framework for precision immuno-oncology and the next phase of immune checkpoint-based therapy. Given the expanding role of ICIs across cancer types, the perspectives discussed herein are particularly relevant to practicing oncologists and other clinicians involved in the care of patients receiving immunotherapy, while also informing researchers and other stakeholders interested in advancing precision immuno-oncology.

IMMUNE-RELATED ADVERSE EVENTS

ICIs have transformed the treatment landscape for several cancers.[5] By targeting regulatory pathways, including PD-1, PD-L1, and CTLA-4, these ICIs release the tumor-induced brakes on the immune system, unleashing cytotoxic T cells to attack the tumor. However, this unrestrained T cell activation triggers a unique spectrum of side effects called immune-related adverse events (irAEs).[6] The clinical presentation varies from mild dermatitis to life-threatening pneumonitis and myocarditis. The development of irAEs affects patient outcomes in several ways.[7,8] The symptom burden associated with these irAEs may lead to treatment discontinuation, impairing cancer care and the potential for treatment benefit. Furthermore, severe irAEs are associated with high rates of hospitalization and emergency room visits and poor quality of life (QOL). As increasing numbers of ICIs have entered routine clinical practice, and as ICI-based combinations are increasingly being used, the occurrence of severe irAEs is on the rise.[9]

Cutaneous Immune-Related Adverse Events (irAEs)

Introduction and clinical spectrum of cutaneous irAEs (cirAEs)

Cutaneous irAEs (cirAEs) are the earliest and the most frequent irAE that recapitulate common dermatologic conditions, including pruritus, psoriasiform, lichenoid, and eczematous dermatitis, along with severe manifestations including immunobullous disease and severe cutaneous adverse reactions (SCARs). Reviews and meta-analyses suggest that cirAEs are frequent across tumor types and ICI classes and often prompt treatment delays or early discontinuation, reinforcing the importance of management strategies that preserve anticancer therapy when safe. In parallel, baseline corticosteroid exposure and higher peak steroid doses used to manage treatment-related adverse events have been associated with inferior ICI outcomes, supporting steroid-sparing, and stepwise escalation when feasible.[10–14]

Existing guidelines for management of cirAEs

Consensus recommendations and society guidelines converge on several practical principles, including graded assessment and phenotype-directed care:

• Grade severity using standardized criteria, then confirm whether the presentation is a cirAE phenotype vs infection, drug eruption, or other mimicker.[15–18]

• Standardized adverse event grading remains central to communication across oncology and dermatology, trial reporting, and decisions to escalate therapy (Table 1). Common Terminology Criteria for Adverse Events (CTCAE) version 6.0 (Table 2), released in 2025, introduces updated terminology and mappings that will affect how skin toxicities are recorded and compared across studies.

• Workup and biopsy are phenotype-dependent (e.g., blistering disease, vasculitic morphology, and atypical or severe eruptions), and early dermatology involvement improves diagnostic precision and facilitates targeted management.[19–21]

• Treatment to allow ICI continuation when feasible: supportive care and topical anti-inflammatories for mild disease; systemic corticosteroids for more severe disease; and escalation to additional immunomodulators for steroid-refractory or steroid-dependent cases.[15–22] Evidence on the efficacy of topical and systemic treatments remains limited to case series and indirect experience from managing primary cutaneous diseases unrelated to ICIs.

Table 1.

Management recommendations for cirAEs and available level of evidence and grade of recommendation[7,23,49]

cirAE	CTCAE v5 Grade 1	CTCAE v5 Grade 2	CTCAE v5 Grade 3*
	Morbilliform eruption or eczematous dermatitis	Continue ICI (IV, A)# Medium to high potency TCS (IV, A)	Continue ICI (IV) High potency TCS (IV)	Hold ICI until grade 0–1 (II, B)
	For pruritus: Oral antihistamines (IV, A) Gabapentinoids (IV, B)		Systemic CS (IV, A) Biologics (omalizumab, dupilumab) (IV, B)
			For pruritus: Aprepitant (V, C)
Lichenoid dermatitis	Acitretin (IV, A) NbUVB (V, B/C)		Systemic CS (IV, A) Methotrexate (V, B/C) Biologics (IL-17-, IL-23-, TNFα inhibitors) (IV, B)
Psoriasiform dermatitis	Consider adding Acitretin, Apremilast, Methotrexate, and/or NbUVB phototherapy, (IV, A)		Anti-IL-23**/IL-17/TNFα (IV, B)
Pruritus	Oral antipruritics (IV, B) Gabapentinoids (IV, B) Aprepitant (V, C) NbUVB (V, author opinion)		Dupilumab, omalizumab (IV, B)
Bullous pemphigoid	Continue ICI (IV, A) High potency TCS (IV, A)	For grade 2 or 3, depending on response to TCS, BSA, and impact (IV, A): Hold ICI Systemic steroids Dupilumab Omalizumab Rituximab
SCARs (SJS/TEN, DIHS)	Stop ICI Hospitalization and supportive care Systemic steroids Steroid-sparing alternatives: IVIG, TNFα inhibitor, cyclosporine, tocilizumab (IV/V)

^*Similar regimen can be pursued for a refractory grade 2 eruption.

^**Targeted biologics are preferred over broader range of immunosuppressant agents. For patients on ICI, IL-23 inhibitors are preferred over IL-17 inhibitors due to risk of colitis from ICIs and IL-17 inhibitors.

^#Level of evidence and grades of recommendation follow the European Academy of Dermatology and Venerology Task Force Position Statement and are adapted from the Infectious Diseases Society of America—United States Public Health Service Grading System (Supplemental Table S1).[23]

BSA: body surface area; cirAE: cutaneous immune-related adverse events; CTCAE: Common Terminology Criteria for Adverse Events; DIHS: drug-induced hypersensitivity syndrome; ICI: immune checkpoint inhibitor; IL: interleukin; IVIG: intravenous immunoglobulin; nbUVB: narrow band ultraviolet B; SCAR: severe cutaneous adverse reaction; SJS/TEN: Steven Johnson syndrome/toxic epidermal necrolysis; TCS: topical corticosteroids; TNF: tumor necrosis factor.

Table 2.

Common Terminology Criteria for Adverse Events (CTCAE) v6 cirAE severity grading

CTCAE Grade	Grade 1	Grade 2	Grade 3	Grade 4	Grade 5
Rash (maculopapular)	Asymptomatic	Mild symptoms	Macules/papules covering > 50% BSA; moderate or severe symptoms	Life-threatening consequences; urgent intervention indicated	Death
Pruritus	Mild or localized; topical intervention indicated	Widespread and intermittent; skin changes from scratching (e.g., edema, papulation, excoriations, lichenification, and oozing or crusts); oral intervention indicated; limiting instrumental ADL or mild/moderate impact on age-appropriate normal daily activity (pediatric)	Widespread and constant; systemic corticosteroid or immunosuppressive therapy indicated; limiting sleep or self-care ADL or severe impact on age-appropriate normal daily activity (pediatric)	—	—
Hypopigmentation	Present	Associated with psychosocial impact	—	—	—
Lichenoid dermatitis	Present; topical intervention indicated	Oral intervention indicated	IV intervention indicated	—	—
Psoriasiform dermatitis	Present; topical intervention indicated	Oral intervention indicated	IV intervention indicated	—	—
Eczematous dermatitis	Present; topical intervention indicated	Oral intervention indicated	IV intervention indicated	—	—

ADL: activities of daily living; BSA: body surface area; cirAE: cutaneous immune-related adverse events; CTCAE: Common Terminology Criteria for Adverse Events.

Table 1 summarizes these recommendations. Where available, the level of evidence and grade of recommendation follow the European Academy of Dermatology and Venerology Task Force position statement on the management of cirAEs (Supplemental Table S1, available online).[23] For The University of Texas MD Anderson Cancer Center’s (MD Anderson) cirAE management algorithm, which closely follows published guidelines, we invite readers to review Supplemental Table S2).[24]

Updates on management options

Why steroid-sparing matters more now

As ICIs are being moved earlier in disease courses and are being administered for longer durations, clinicians are seeing more persistent, relapsing, and steroid-dependent cirAEs. Chronic systemic steroids carry cumulative risks, may complicate cancer care, and may blunt ICI effect; therefore, there is a strong push toward phenotype-directed, steroid-sparing approaches that control symptoms while enabling ICI continuation.[15–22]

Selected updates in phenotype-guided treatment

• A. Pruritus and eczematous–spongiotic patterns.

• First-line: emollients, topical corticosteroids (TCS) ± topical calcineurin inhibitors; nonsedating antihistamines for itch; consider neuropathic itch strategies when appropriate.[19–21]

• Escalation: dupilumab has increasing supportive evidence from retrospective series as a steroid-sparing option for chronic eczematous and pruritic cirAEs, including patients requiring repeated steroid tapers.[25] Although limited in evidence, immunologic profiling of eczematous dermatitis also supports the use of T_H2 (helper T 2 cells) inhibitors, such as dupilumab.[26]

• Ali et al demonstrated that among 102 patients with metastatic cancer who received ICIs, HLA-DRB1*11:01 was significantly associated with pruritus, suggesting an underlying genetic predisposition.[21,27]

• B. ICI-associated bullous pemphigoid (irBP).

• Standard approaches include high-potency topical steroids, systemic corticosteroids for extensive disease, and conventional steroid-sparing agents in selected cases.[19–21]

• Update: emerging clinical evidence from case series supports dupilumab as a steroid-sparing strategy in select irBP patients.[28]

• C. Psoriasiform or lichenoid dermatitis.

• Phenotype-tailored topical therapy remains foundational; systemic escalation is individualized based on severity, body surface area (BSA) involvement, and cancer context.[20,21]

• Histopathological features of lichenoid dermatitis include hyperkeratosis, a dense band-like lymphocytic infiltrate, interface changes, and keratinocyte apoptosis in the basal layer of the epidermis.[29] Unlike classic lichen planus, epidermal spongiosis, parakeratosis, eosinophils, and necrosis can be observed in ICI-induced lichenoid eruptions.[30]

• Update: MD Anderson experience, expert consensus, and evidence from small series support the use of targeted immunomodulators for psoriasiform dermatitis (e.g., psoriasis-directed biologics or small molecules) when conventional measures fail. These therapies are best used with careful multidisciplinary discussion.[20,21]

• D. Severe cutaneous adverse reactions.

• A Steven–Johnson syndrome and toxic epidermal necrolysis (SJS/TEN)-like eruption characterized by less profound mucosal involvement and a more favorable prognosis, called progressive immunotherapy-related mucocutaneous eruption (PIRME), has been described. PIRME often arises from a morbilliform, urticarial, or lichenoid dermatitis that slowly progresses and abruptly desquamates. PIRME may include a second drug trigger. The activation of the immune system, including T cells, antigen presenting cells, and the inhibition of the suppressive function of T_regs by PD-1 blockade may accelerate a concurrent medication-induced drug eruption.[20,21,31–35]

Interpreting cirAE outcome associations

Meta-analytic data suggest associations between cirAEs and improved survival in some cohorts.[11] This should not lead to therapeutic nihilism; instead, it supports controlling toxicity with focused immunomodulation avoiding oncologically active T cells to reduce unnecessary discontinuation, prevent escalation to severe disease, and maintain ICI therapy.[10,11,15–18]

Practice gaps

Managing moderate to severe and chronic cirAEs

Most cirAEs are low-grade, self-limiting toxicities that are managed conservatively without ICI modification. A subset of patients with eczematous, lichenoid, psoriasiform, or irBP phenotypes evolve into chronic cirAEs.[36] In particular, irBP and lichenoid dermatitis have a propensity to persist or be present after ICI discontinuation, suggesting a sustained state of immune activation.[20] Following initial stabilization with corticosteroids and temporary ICI cessation, there is limited evidence-based consensus on managing CTCAE version 5 grade 3–4 cirAEs or persistent and chronic dermatoses. In attempts to avoid systemic corticosteroids and broad immunosuppression, targeted immunomodulators should be utilized whenever possible. Evidence on the use of these agents derives from retrospective studies and expert opinion. At MD Anderson, management of high-grade, refractory, and chronic cirAEs prioritizes escalation with steroid-sparing strategies, ideally agents with low T cell impact. Close follow-up intervals of 1–2 weeks are generally pursued until control, at which point management can be tailored with longer follow-up periods, especially for patients on biologics with a favorable side effect profile (Supplemental Table S2).

Decision to rechallenge

Current guidelines recommend permanent ICI discontinuation upon development of a SCAR; however, for other cirAEs, rechallenge decisions should weigh severity of the event, response to therapy, and oncologic need. Chronic or late cirAEs are increasingly recognized and require longitudinal planning.[22] For patients with refractory CTCAE version 5 grade 2 or grade 3 cirAEs who would benefit from ongoing ICI therapy, rechallenge can be considered upon cirAE severity improvement to grade 1 and multidisciplinary discussion of toxicity recurrence.[37] The incidence of all-grade irAEs after ICI rechallenge ranges from 27.5% to 57.5%,[38,39] with cirAEs being the most frequent to recur. In this setting, the introduction of a corticosteroid-sparing agent or targeted immunomodulator should be considered for toxicity prevention and symptom management.[37]

Late and chronic irAEs: the long tail

As survival improves and ICI indications expand, clinicians must remain vigilant of delayed-onset irAEs that may present years after ICI exposure.[22] In the case of cirAEs, management strategies should emphasize expectation counseling and monitoring for persistent, recurrent, or delayed-onset toxicities in cancer survivors. Cohort data underscore the need for durable management plans, survivorship-aware monitoring, and clearer rechallenge frameworks.

What is next? Preclinical, translational, and clinical work in progress

From “rash” to endotypes: mechanistic and histologic profiling

A major limitation in cirAE care is the historically coarse labeling of heterogeneous eruptions. Recent work emphasizes clinicopathologic patterns (e.g., spongiotic, interface and lichenoid, psoriasiform, and blistering) linked to distinct immune programs.[20,21] Pilot studies into immunologic phenotyping demonstrate polarizable inflammatory pathways that can guide phenotype-driven therapy.[26,40] Lichenoid and psoriasiform dermatitis demonstrate a predominant type 17 inflammatory response, marked by elevated interleukin (IL)-17 mRNA expression in lesional skin, whereas a skewed type 2 inflammatory axis mediates eczematous eruption and irBP.[26,41] Translational studies into immune profiling of cirAEs remain limited; however, available data suggests that the immune microenvironment in cirAEs parallels inflammatory pathways that mediate the idiopathic counterparts of these diseases and lends support to the use of targeted biologics implicated in inflammatory axes.

Biomarkers and predisposition

Host factors likely shape susceptibility and phenotype. Genetic predisposition signals—such as human leukocyte antigen (HLA) associations with checkpoint inhibitor-induced skin adverse events—support the concept that cirAEs may be partially predictable and biologically patterned rather than purely idiosyncratic.[27] Specifically, the HLA-B*51:01 allele was associated with a higher risk of any cirAE in a retrospective review of melanoma patients.[42]

Clinical trials: toward toxicity-preserving interventions

Parallel efforts are refining preclinical and clinical platforms to study irAEs and de-risk toxicity-sparing interventions.[37] Trials evaluating targeted immunomodulators alongside PD-L1 blockade aim to demonstrate that controlling select irAEs can reduce interruptions without negating antitumor efficacy. For example, NCT05013450 evaluates dupilumab and anakinra with PD-L1 blockade.[43]

Recent work from high-volume oncodermatology programs has expanded the evidence base for cirAE endotyping and steroid-sparing care, including prospective clinicopathologic immune characterization, synthesis of mucocutaneous management, and MD Anderson experience with biologic immunomodulators and systemic treatment algorithms.[44–48]

Conclusion

cirAEs are common, heterogeneous, and increasingly chronic as ICIs expand. Guidelines provide a graded framework, but the field is shifting toward phenotype-driven, steroid-sparing management supported by growing clinical experience—particularly for eczematous and pruritic patterns and irBP. Translational studies are clarifying mechanisms and endotypes, enabling biomarker discovery and rational trial design. Near-term priorities include standardized phenotyping (including histology when appropriate), improved adverse event capture under CTCAE updates, and prospective trials of targeted therapies to support continued life-prolonging immunotherapy with minimized dermatologic morbidity.

Introduction and clinical spectrum of endocrine irAEs

Endocrine irAEs are among the most common toxicities associated with ICI therapy and span a broad clinical spectrum, ranging from frequent thyroid dysfunction to rare but fulminant complications such as ICI-induced diabetes mellitus (ICI-DM). Recent consensus efforts have improved disease definitions and classification of endocrine irAEs, enabling greater clarity and standardization in reporting and evaluation.[50] Thyroid dysfunction is the most prevalent endocrine irAE, with hypothyroidism occurring in 3–15% of treated patients and hyperthyroidism in 0.6–8%.[51,52] ICI-associated hypophysitis (ICI-HP) occurs less frequently, with a higher incidence reported with CTLA-4 inhibitors (3–13%) and a lower incidence with PD-1/PD-L1 inhibitors (approximately 1%). ICI-DM is rare (0.2–1.9%) but often severe, frequently presenting with diabetic ketoacidosis (DKA) and requiring hospitalization in more than 50% of cases.[51,53–55] Over 95% of ICI-DM occur in the setting of with PD-1/PD-L1 exposure.[51,56] Less common endocrine toxicities include primary adrenal insufficiency (PAI), hypoparathyroidism, and lipodystrophy.[57–60]

The risk of endocrine irAEs varies by treatment regimen, with combination CTLA-4 and PD-1/PD-L1 blockade conferring a higher incidence than monotherapy.[52,54] Although differences across ICI classes are increasingly well described, whether cancer type independently modifies irAE risk remains incompletely understood. The timing of onset also varies by endocrinopathy and ICI class. Thyroid dysfunction typically manifests early, most often within 3–8 weeks of treatment initiation, and multiple studies have demonstrated that baseline thyroid autoantibodies are associated with an increased risk of thyroid irAEs, supporting a role for pre-existing autoimmune susceptibility, although their role in routine clinical risk stratification remains undefined.[61,62] In contrast, ICI-DM demonstrates a more heterogeneous time course, with reported onset ranging from weeks to many months or even years after exposure, and the presence of islet autoantibodies has been associated with more rapid onset.[53,54] Hypophysitis likewise shows treatment-specific patterns, occurring earlier with CTLA-4 inhibitors and later with PD-1/PD-L1 inhibitor monotherapy.[63,64] Importantly, endocrine irAEs can develop even after ICI discontinuation, necessitating continued surveillance beyond the active treatment period.[65]

Endocrine irAEs exhibit several distinguishing features that differentiate them from other immune-related toxicities and have important implications for long-term management and outcomes. In reviewing these features, we will highlight their importance in precision oncology.

Permanent nature and treatment approach

Endocrine irAEs typically reflect irreversible glandular damage rather than transient inflammation. Management relies on lifelong hormone or insulin replacement, underscoring the permanent nature of most endocrine irAEs and their implications for survivorship. ICI-induced thyrotoxicosis is a partial exception, as the hyperthyroid phase often resolves spontaneously but frequently evolves into permanent hypothyroidism.[66]

The apparent irreversibility of endocrine irAEs may partly reflect limitations in clinical detection. Subclinical or mild abnormalities, such as transient thyroid function changes, mild hyperglycemia, or pituitary enlargement without overt hormone deficiency, may go unrecognized or resolve before formal diagnosis.[67] However, in those with severe presentations or without confounding exposures, attribution to ICI therapy is typically clear, and recovery of endogenous function once clinically established is quite uncommon. In these settings, patients can be counseled that endocrine dysfunction is likely to be permanent.

This permanence has important implications for management and precision strategies. Therapeutic options remain largely limited to hormone or insulin replacement, without established immunomodulatory options once glandular injury has occurred. This likely reflects the rapid onset of immune-mediated damage and the absence of well-defined, targetable preclinical disease states. Because endocrine irAEs are rarely immediately life-threatening once appropriately treated, aggressive immunosuppression aimed at organ salvage is uncommon. Ongoing research in this area is informed by safety lessons from more fatal irAEs, such as myocarditis, as well as from conventional autoimmune endocrinopathies including type 1 diabetes.

Association with survival

Across multiple cancer types and ICI regimens, irAEs have been linked to survival benefit. The development of thyroid irAEs, in particular, has been consistently associated with improved progression-free survival (PFS) and overall survival (OS).[61] This association persists across multiple studies after adjustment for prognostic factors including age, sex, metastatic burden, and time-dependent survival bias.[68] In contrast, the prognostic implications for ICI-DM and ICI-HP are less clear. For ICI-DM, interpretation is limited by the rarity of the condition and lack of statistical power.[53,55,69] For ICI-HP, the data are heterogenous just as the forms of hypophysitis seem to be.[70,71] Although some studies report improved OS among those who develop ICI-HP, this association is not consistently observed after adjusting for survival bias and other confounders such as age, sex, and cancer type.[72] Notably, outcomes may be influenced not only by the development of ICI-HP but also by its management, particularly of the adrenal insufficiency (AI) that it nearly always includes. Among patients with ipilimumab-induced hypophysitis, higher doses of glucocorticoids have been associated with reduced survival.[73]

These findings highlight a central challenge for precision oncology: distinguishing immune responses that drive durable tumor control from those representing off-target autoimmunity and determining whether survival associations are consistent across cancer types. From a translational perspective, endocrine irAEs provide a valuable model to interrogate the links between immune activation, autoimmunity, and therapeutic efficacy.

Existing guidelines for management of endocrine irAEs

Guidelines from the National Comprehensive Cancer Network (NCCN) and Society of Immunotherapy of Cancer (SITC) provide a standardized framework for the evaluation and management of endocrine irAEs.[74,75] Table 3 summarizes the current guidelines across the natural history of the three major endocrine irAEs.

Table 3.

Comparison of guideline recommendations and expert considerations for endocrine immune-related adverse events

Endocrinopathy	Clinical Phase	NCCN Guidelines[74]	SITC Guidelines[75]	Considerations and Practice Gaps
Hypothyroidism/ thyrotoxicosis	Pretreatment with ICI	Baseline TSH, FT4	Baseline TSH, FT4	Consider baseline thyroid autoantibodies to identify patients at higher risk; rarely done in practice despite predictive value
	Monitoring while on ICI	TSH every 4–6 wk during therapy, then follow up every 12 wk as indicated	TSH, FT4 every 4–6 wk during therapy; continue monitoring every 6–12 mo after ICI cessation	Thyroid irAEs may present between scheduled labs; symptom-triggered testing remains important. At times, 6 wk between TFTs is too long, as evolving thyroiditis can progress from subclinical hyperthyroidism to overt, severe hypothyroidism in this time frame
	At irAE diagnosis	Supportive care for thyrotoxicosis; antithyroid drugs not recommended. Initiate levothyroxine for hypothyroidism. Check serum cortisol to exclude concurrent adrenal insufficiency prior to treatment	If symptoms of hyperthyroidism, give β-blockers as needed; consider Graves’ disease if persistently low TSH, high FT4 Initiate levothyroxine for hypothyroidism. If symptoms of hypothyroidism and/or with elevated TSH and low FT4, test AM cortisol to identify possible concurrent adrenal insufficiency	Hyperthyroidism often evolves to hypothyroidism; anticipate transition rather than reactively managing once severe hypothyroidism occurs
	Post-irAE diagnosis	Continue thyroid hormone replacement indefinitely. Continue ICI if/once stable	For hypothyroidism, levothyroxine 1.5–1.6 µg/kg/d for young, healthy patients and 25–50 mcg/d for age ≥ 65 or history of heart disease	If low/normal TSH and low FT4, consider nonthyroidal illness including hypophysitis
Hypophysitis/ adrenal insufficiency (AI)	Pre-treatment with ICI	No routine baseline pituitary testing; consider AM cortisol and thyroid function testing as above	If low TSH and normal/low FT4, check ACTH and AM cortisol	Assessment of steroid history prior to initiation of ICI
	Monitoring while on ICI	Symptom-driven assessment; consider repeating AM cortisol every 4–6 wk during therapy, then follow up every 12 wk as indicated	Symptom-driven assessment: If symptoms of acute hypophysitis are observed, immunotherapy should be interrupted and corticosteroids administered	Early symptoms often nonspecific, overlap with cancer-related fatigue Cortisol and ACTH interpretation can be confounded by exogenous steroids or opioids and timing of draws Consider preoperative cortisol and ACTH for screening
	At irAE diagnosis	AM cortisol, ACTH, TSH, FT4, Na; consider LH, FSH, and sex hormones; MRI brain with pituitary cuts if mass effect symptoms or concern for metastatic disease	If low ACTH and AM cortisol, check TSH, FT4, FSH, LH, and sex hormones; replacement hydrocortisone 10–12 mg/m2/d	Primary adrenal insufficiency (high ACTH, low AM cortisol) is rare, consult endocrine if concern
	Post-irAE diagnosis	Long-term glucocorticoid replacement; continue ICI if stable	Long-term glucocorticoid replacement; encourage to have medical alert device, stress dose steroid education	Determining optimal steroid dose is challenging due to overlapping cancer symptoms
ICI-induced diabetes mellitus	Pretreatment with ICI	Baseline comprehensive metabolic panel (CMP); pancreatic baseline testing is not required	Baseline CMP	Baseline A1c may help distinguish new-onset ICI-DM from progression of pre-existing dysglycemia and may prepare providers for steroid induced diabetes; areas under study: utility of islet autoantibody testing, genetic risk, and continuous glucose monitors
	Monitoring while on ICI	Repeat CMP prior to each treatment or every 4 wk during treatment, then in 6–12 wk or as indicated	Repeat CMP intermittently throughout course of therapy; if beginning corticosteroid therapy, counsel on hyperglycemia	Abrupt onset limits opportunity for early detection; education on warning symptoms may reduce severe presentations or improve early detection
	At irAE diagnosis	Check C-peptide with glucose, evaluate for DKA (manage DKA per institutional guidelines), consider measuring islet autoantibodies; initiate insulin therapy, urgent endocrine consultation	Check HbA1c, C-peptide, islet autoantibodies; If severe hyperglycemia or DKA, should be hospitalized	C-peptide should be interpreted with concurrent serum glucose; fasting values may underestimate residual β-cell function Repeat testing when out of hyperglycemic crisis recommended although usually consistent with initial testing
	Post-irAE diagnosis	Lifelong insulin therapy, consider early use of continuous glucose monitoring	Counsel on diabetes management including diet and lifestyle changes, glucose monitoring, and insulin treatment	High rates of DKA at presentation highlight need for patient and provider education on symptoms for awareness. A1c goals should be individualized based on patient prognosis.

ACTH: adrenocorticotropic hormone; CMP: comprehensive metabolic panel; DKA: diabetic ketoacidosis; FSH: follicle-stimulating hormone; FT4: free thyroxine; ICI: immune checkpoint inhibitor; ICI-DM: ICI-induced diabetes mellitus; irAE: immune-related adverse event; LH: luteinizing hormone; MRI: magnetic resonance imaging; NCCN: National Comprehensive Cancer Network; SITC: Society of Immunotherapy of Cancer; TFT: thyroid function test; TSH: thyroid-stimulating hormone.

Updates on management options

Management of endocrine irAEs differs in several important ways from both other immune-related toxicities and traditional autoimmune endocrine disease. Treatment emphasizes hormone replacement or insulin use rather than immunosuppression, and ICI therapy can often be continued once patients are clinically stable. However, these approaches are largely reactive, highlighting the need for precision strategies for early risk identification. Alternative immunomodulatory approaches in place or in addition to hormone replacement are under investigation in preclinical stages[76–78] and emerging into clinical trials.

Practice gaps

Gaps in real-world diagnosis and management

Despite the availability of consensus guidelines,[74,75] important gaps remain in the real-world monitoring and management of endocrine irAEs (Table 3). Several endocrine-specific nuances contribute to variability in practice, particularly in diagnostic interpretation and treatment optimization, and access to onco-endocrinologists is severely limited.

Obstacles in diagnosis

The nuanced normal reference ranges for endocrine lab tests pose challenges for non-endocrinologists and for electronic medical record systems.

In suspected ICI-DM, interpretation of C-peptide requires concurrent glucose measurement, as fasting or euglycemic stages may obscure residual beta-cell function. Although hemoglobin A1c is recommended at diabetes diagnosis, baseline assessment prior to ICI initiation is rarely performed and may help distinguish new-onset ICI-DM from progression of pre-existing dysglycemia, particularly in patients with recent steroid exposure or anemia.

Similar challenges arise in pituitary and adrenal evaluation. Biochemical testing is frequently confounded by glucocorticoid or opioid use, both of which can suppress adrenocorticotropic hormone (ACTH) and cortisol levels. In patients with concurrent hypothyroidism and AI, appropriate sequencing of testing and therapy is critical, as AI can cause sick euthyroidism and glucocorticoid replacement must precede thyroid hormone initiation to avoid precipitating adrenal crisis.

Challenges in treatment optimization

Optimizing hormone replacement remains challenging even once endocrine irAEs are established. Determining optimal glucocorticoid replacement doses in ICI-HP and ICI-PAI is complicated by symptom overlap with malignancy and other cancer therapies, underscoring the need for individualized dosing and consistent counseling regarding stress dosing during illness or procedures. Emerging evidence also suggests that patients with ICI-hypothyroidism may require different levothyroxine doses compared to those with Hashimoto’s thyroiditis, highlighting possible limitations of extrapolating dosing strategies from non-ICI-related endocrine disease.[79,80]

What is next? Preclinical, clinical, and translational work in progress

Beyond management, increasing evidence supports the feasibility of predicting endocrine irAE risk, representing a promising and rapidly evolving area of investigation. However, most risk prediction strategies remain investigational and are not yet ready for routine clinical implementation. Thyroid autoantibodies and genetic variants associated with lifetime thyroid autoimmunity have been linked to increased risk of thyroid irAEs,[61,62,81,82] yet are rarely incorporated into baseline assessment despite their low cost and availability. Emerging data also implicate genetic susceptibility in hypophysitis and ICI-DM, including associations with specific HLA haplotypes such as HLA-DQ0602 for hypophysitis.[64] Polygenic risk scores for autoimmune disease have been associated with early ICI discontinuation due to irAEs, further supporting a potential role for genetic risk stratification.[83]

Despite these advances, key barriers to implementation remain, including uncertain clinical utility, lack of prospective validation, and limited evidence that risk stratification alters management or outcomes. As such, incorporating these tools into routine care is premature, and their use should currently be considered exploratory. Importantly, this is an area of active investigation, with ongoing studies aimed at defining how risk prediction can be integrated into clinical decision-making.[84]

Conclusion

Although most prediction strategies remain investigational, several practical interventions can be implemented now to improve clinical outcomes. These include improved patient education regarding endocrine risks, enhanced symptom awareness, and closer monitoring to enable earlier diagnosis and reduce complications. Patients with ICI-HP or ICI-DM should receive consistent counseling on medical alert identification and sick day management, areas that remain inconsistently addressed in practice.

Looking forward, progress will depend on a deeper mechanistic understanding of endocrine irAEs and the development of safe, targeted immunomodulatory strategies capable of preserving antitumor efficacy while limiting irreversible endocrine damage. The path toward precision prevention and early intervention is increasingly visible.

Gastrointestinal irAEs

Introduction and clinical spectrum of gastrointestinal irAEs

Inflammation of the gastrointestinal (GI) system is one of the more common toxicities of immune-checkpoint inhibition. The luminal GI tract consists of the esophagus, stomach, small intestine and colon with the lower GI tract most frequently involved, presenting as immune-mediated diarrhea and colitis (IMDC). Although upper GI irAEs occur in 0.8–3% of ICI-treated patients,[85,86] IMDC is typically seen in 8–30% of patients receiving immunotherapy.[87–89] This prevalence is increased in the presence of risk factors such as CTLA-4 inhibition,[90] melanoma,[91] obesity,[92] nonsteroidal anti-inflammatory drug,[93] and proton-pump inhibitor use.[94]

Existing guidelines for management of gastrointestinal irAEs

Most of the existing guidelines for IMDC were published between 2020 and 2021 and make similar recommendations regarding the evaluation and treatment of this disease.[16,95–98] Traditionally, clinical symptomatology is used to guide management with testing of stool biomarkers reserved for select cases to determine the need for endoscopic evaluation. Patients with milder symptoms may be treated conservatively with anti-diarrheal medications whereas those with more severe symptoms are initially treated with steroids followed by biologic agents for steroid-refractory disease. Currently recommended biologic agents include Tumor Necrosis Factor-α Inhibitor (infliximab [IFX]) and α4β7 Integrin inhibitor (vedolizumab,VDZ), and options for third-line treatments are limited, but evolving. That said, much data has been published in the intervening years that expands on our understanding of the evaluation and treatment of IMDC.

Despite a heavy reliance on symptom severity to direct clinical decision-making in IMDC based on current guidelines, a growing body of research suggests that symptoms poorly correlate with endoscopic findings,[99–102] cannot reliably predict the need for more advanced treatment,[103,104] and have not been associated with any clinical outcomes, while high-risk endoscopic features correlated with the need for more aggressive treatment. In contrast to symptomatology alone, stool biomarkers such as fecal lactoferrin and calprotectin have achieved sensitivities of 80–90% for endoscopic inflammation[105] and have been associated with clinical outcomes.[106] Altogether, these findings support the use of stool biomarkers for the initial evaluation of IMDC over clinical symptoms alone as they have higher sensitivities for detecting active disease. Moreover, endoscopic findings may allow for earlier introduction of biologic agents, which have been tied to better outcomes.[107]

The role of imaging in this disease is currently limited. Computed tomography of the abdomen has been shown to have low sensitivity and negative predictive value for diagnosing IMDC.[108,109] That said, intestinal ultrasound (IUS) is a readily available and cost-effective alternative that is seeing increasing use in the realm of inflammatory bowel disease.[110] Although its application is more common in Europe, more centers in the United States have adopted IUS as both a diagnostic and monitoring tool. To date, only three studies have reported its use in IMDC.[111–113] One showed that IUS findings significantly correlated with the degree of endoscopic inflammation and involved features from both ulcerative colitis and Crohn’s disease.[111] The remaining two studies demonstrated its utility for monitoring therapeutic response.[112,113] It is a promising imaging modality that will likely be incorporated into future management guidelines once more evidence supporting its use emerges.

Updates on management options

To date, IFX and VDZ remain the best-supported first-line treatment for steroid-refractory disease. They have comparable efficacy with 30-day steroid-free remission rates of 75% and overall remission rates of up to 90% and similar safety profiles.[105,114,115] That said, IFX has been associated with shorter time to clinical improvement while VDZ may be associated with shorter steroid courses and fewer hospitalizations.[105,115] There is increasing data to support alternative second- and third-line options for patients who fail either of the aforementioned agents. The interleukin-12/23 (IL-12/23) inhibitor (ustekinumab, UST), for instance, has gained traction as an effective alternative among patients who either failed or could not tolerate treatment with IFX and VDZ, achieving remission in up to 68.4% of the 19 patients studied.[116] Similarly, the Janus kinase (JAK) inhibitor tofacitinib shows promise in IFX- and VDZ-refractory IMDC, achieving remission in all nine cases reported in the literature, either alone or in conjunction with fecal microbiota transplantation (FMT).[117–122] A recently published international multicenter study revealed that calcineurin inhibitors such as tacrolimus may prove effective in managing IFX-refractory IMDC with clinical improvement in 74% of the 31 patients treated, but potentially worse oncologic outcomes.[123] Meanwhile, anti-metabolites such as mycophenolate mofetil (MMF) have a modest success rate of 50% in improving clinical symptoms across the 38 patients reported in the literature.[123,124] Treatment with Interleukin-6 inhibitor (tocilizumab, TCZ) led to improvement in 8 of the 10 participants of the open-label COLAR trial.[125] However, all participants of this trial were biologic-naïve and only 3 of the 10 participants had steroid-refractory IMDC, raising some questions regarding its utility in more recalcitrant disease. Finally, procedural interventions such as extracorporeal photopheresis (ECP; clinical remission in 3/3 patients reported in the literature[126–128]) and FMT (remission in 51/64 [80%] patients[129]) have proven to be surprisingly effective treatment modalities that may spare patients the typical immunosuppressive regimens.

Practice gaps

Long-term outcomes and surveillance

Although most patients with checkpoint inhibitor colitis typically have timely improvement in their symptoms, this may not always be the case. Although the data are limited, one study suggests that around 16% of patients may have a more chronic disease course lasting up to 6 months or longer, with an even higher proportion of patients having persistent endoscopic or histologic inflammation, even in the absence of symptoms.[130] Alternatively, a small subset of patients (∼10%) with initial symptom remission may develop recurrent symptoms.[130] This risk is increased among patients who resume immunotherapy, particularly with anti-CTLA-4 agents, and may be mitigated by concurrent administration of selective immunosuppression therapy such as VDZ.[131,132] Whether these disease patterns differ based on the indication for ICI has yet to be determined, making it a promising avenue for future research. To date, only one study explores this, finding that toxicity is less severe in more advanced melanoma.[133]

The decision to rechallenge patients with ICI should be highly personalized based on the patient’s overall prognosis, their response to prior immunotherapy, and the severity of their IMDC. Current guidelines do not give clear instruction on the suitable timing for rechallenge, but typically recommend restarting ICIs once symptoms improve to CTCAE grade 1 or below, ideally with PD-1/L1 monotherapy rather than combination or CTLA-4 therapy.[16,95–97] Guidelines specific to colitis additionally advise considering rechallenge once clinical or endoscopic remission is achieved, depending on initial symptom severity.[97] Although useful, these algorithms fail to capture the intricacies of cancer care. For instance, some patients with advanced malignancy who previously responded to immunotherapy may have a more pressing need to re-initiate ICI as opposed to patients with better prognoses or poor initial response to ICIs. The benefit of ICI re-introduction must therefore be carefully weighed against the risk of recurrent IMDC on a case-by-case basis. For patients in whom the decision is made to restart ICIs, either IFX or VDZ can be administered concurrently with checkpoint inhibition, with a reported absolute risk reduction in recurrent IMDC of 13.6% (34.4% recurrence rate in control vs 20.8% in patients with concurrent selective immunosuppressive therapy).[132]

Little is known about the long-term outcomes and recommended surveillance strategies for IMDC. Although current guidelines broach the subject, they offer very limited recommendations to guide clinical reasoning. Repeat colonoscopic evaluation remains essential to ensure mucosal healing while also screening for pre-malignant polyps, which these patients are at increased risk for.[134] However, the appropriate timing of this remains unclear. Fecal calprotectin, on the other hand, is a key tool for monitoring inflammation in inflammatory bowel disease (IBD), its role in the longitudinal monitoring of IMDC disease activity is also emerging and remains to be further elucidated.[106] Longitudinal studies are therefore needed to help establish better-defined follow-up intervals. Nonetheless, 85–94% of patients ultimately achieve clinical symptom remission.[104,105]

With regard to oncologic outcomes, there is a complex interplay between ICI toxicity, irAE treatment, and survival outcomes. IrAEs and their severity have traditionally been associated with improved progression-free survival, likely corresponding to the potency of the checkpoint inhibitor response.[135–137] This same response is hypothesized to be blunted by immunosuppression, which is the mainstay of irAE treatment. Although the literature is conflicting on this topic, several large-scale studies seem to suggest that high doses and longer durations of systemic corticosteroid therapy as well as the use of second-line immunosuppression with Tumor Necrosis Factor-alpha inhibitors specifically may all contribute to worse overall and progression-free survival.[137–140] This underscores the importance of balancing effective IMDC treatment against the oncologic consequences of immunosuppression to maximize patient outcomes and highlights the need for alternative treatments for IMDC that avoid such consequences.

What is next? Preclinical, clinical, and translational work in progress

Multiple efforts are currently underway to further enhance our understanding of this complex disease. Preclinical models involving immune phenotyping,[141] murine treatment models,[142–144] and gut microbiomal analysis[145–147] may help identify key molecular targets that we can use to guide treatment. Moreover, given its overlap with IBD, studies on the efficacy of proven IBD treatments in IMDC will serve as an important proof of concept for adopting additional treatment strategies. Current animal models focused on targeted IL-23 blockade show great promise in ameliorating colitis severity while importantly maintaining antitumor efficacy in ICI-treated mice.[148,149] Risankizumab, a monoclonal antibody to IL-23 and highly successful IBD treatment,[150] may therefore be one such treatment option worth investigating. Ongoing clinical trials on novel treatments for IMDC are currently limited to exploring the use of ECP, UST or FMT. Future studies that explore the aforementioned treatment options in larger samples are necessary. Particularly, larger case series and multicenter studies will help provide much needed clinical data to justify their more consistent use in difficult-to-treat IMDC cases and allow for head-to-head comparisons in the future.

Conclusion

IMDC is a common and often severe toxicity to checkpoint inhibitors that significantly affects patient’s QOL. Although management is traditionally guided by clinical symptomatology, a growing body of evidence supports reliance on stool biomarkers and endoscopic evaluation instead of determining the need for systemic immunosuppression and biologic therapy. Early immunosuppressive treatment is associated with better outcomes, encouraging more aggressive treatment of confirmed cases of IMDC, reminiscent of the top-down approach favored in inflammatory bowel disease. Although standard therapies such as IFX and VDZ are highly effective, there is a need for alternate treatments for management of IFX- and VDZ-refractory disease, especially given concerns for infection and reduced anti-tumor immunity with these drugs. Multiple biologic agents have been explored. Currently, UST is the best supported, followed closely by tofacitinib. IL-23 inhibitors are a promising future avenue for research with established success in IBD as well as few preclinical models and case reports. FMT is a safe and effective alternative to immunosuppressive treatment with high clinical remission rates and a preferable safety profile as well. Studies exploring alternate treatment options with larger sample sizes as well as long-term IMDC outcomes investigating recurrence, ICI resumption, and the effect of different treatments on cancer outcomes are needed to further enhance our understanding of this disease entity.

Introduction and clinical spectrum of hepatobiliary irAEs

Immune-mediated hepatobiliary toxicity remains a well-recognized potential irAE in patients who undergo treatment with ICIs. As ICI use expands, including introduction of anti-LAG-3 checkpoint inhibitor in addition to interval approvals of anti-CTLA-4 and anti-PD-1/PD-L1 agents for a wide range of malignancies, the hepatobiliary toxicity will need to be managed promptly and effectively to ensure optimal outcomes in cancer treatment. In the current literature, the abbreviations IMH (immune-mediated hepatotoxicity), ILICI (immune-mediated liver injury caused by checkpoint inhibitors), and ChILI (checkpoint inhibitor-induced liver injury) have been introduced.[151–155] For ease, we will use “IMH” to refer to this general class of irAEs.

Variable statistics have been reported as to the incidence of IMH in patients who undergo ICI treatment; the incidence differs depending on use of anti-CTLA-4, anti-PD-1/PD-L1 or combination therapy. The composite data suggest that overall, hepatotoxicity of any grade may occur up to 30% of patients undergoing combination ICI treatment in general.[156–158] Data from a single-center study at a cancer center reported IMH of any grade with an incidence of 5.9% for anti-PD-1/PD-L1 agents, 9.5% in anti-CTLA-1 agents, and 18.7% in combination ICI; high-grade IMH occurred in 1.1%, 1.7%, and 9.2% for the respective categories.[159]

The pathophysiology of IMH to date has not been adequately delineated, although putative mechanisms have been proposed. IMH is classified as an “indirect type” of drug-induced liver injury (DILI), as opposed to direct hepatotoxic effects on hepatocytes. Cytokines and signaling pathways involving IL-6/IL-6R, JAK, and signal transducer and activator of transcription 3 (STAT3) pathways may play a role.[154,160–162] Proinflammatory cytokines mobilize NK (natural killer) cells and macrophages that contribute to acute liver injury.[154] In idiopathic autoimmune hepatitis (iAIH), T cell–mediated responses that involve T_reg and T_H17 cells lead to an inflammatory response that includes plasma cell and lymphocytes, leading to interface hepatitis.[154] However, unlike iAIH, the histology of IMH is arguably dissimilar. Although not specific, IMH tends to manifest in lobular lymphocyte inflammatory infiltration and less so plasmacytic populations; concomitant endotheliitis and (rarely) granulomas may occur.[153,163,164] Immunohistochemical stains may include CD3⁺ and CD8⁺, rather than CD4⁺ and CD20⁺ as might be seen in iAIH.[153,163,165–167] Bile duct injury on histology may represent an additional phenotype which extends into the cholangiopathic variants of IMH.

Cholestatic or cholangiopathic phenotypes (immune-mediated cholangiohepatitis [IMCH]) are important to recognize because of their association with anti-PD-1 agents, approved for a broad range of malignancies, and have been attributed to suboptimal or even poor steroid-response in the majority of cases.[154,168] Recognition of the phenotype, then, becomes paramount as a prognostic indicator to be prepared to escalate immunosuppressive treatment. Cholecystitis, vanishing bile duct syndrome (VBDS), sinusoidal obstruction syndrome (SOS), and development of chronic biliary strictures (sclerosing cholangitis) are rare manifestations of cholangiopathic irAEs.[168–171]

Existing guidelines for management of hepatobiliary irAEs

Guidance documents from multiple oncology societies (American Society of Clinical Oncology [ASCO], SITC, NCCN, European Society For Medical Oncology [ESMO]) and one gastroenterology society (American Gastroenterological Association [AGA]) have traditionally endorsed disruption of ICI treatment once the patient enters liver injury categorized as CTCAE grade 2 or higher.[18,49,95,97,172–174] In high-grade IMH, these documents endorse the use of high-dose systemic corticosteroids (1–2 mg/kg/d), although this particular weight-based dosing did not have an evidence-grounded basis.[97,173,174] The analysis for assessing liver injury generally did not emphasize the interpretation of the alkaline phosphatase (ALP) in relation to bilirubin (important to screen for possible IMCH phenotypes). Although international normalized ratio (INR), bilirubin, and presence of hepatic encephalopathy and/or ascites, were included in the analysis, the incidence of acute liver failure associated with mortality from acute decompensated liver disease is considered rare, and thus, those four parameters have to be assessed contextually before ascribing their abnormality to the CTCAE grading of the irAE itself. NCCN’s irAE guidelines recently included ALP in the liver injury analysis explicitly, although the updated CTCAE version 6.0 no longer ascribes grading severity to ALP levels. This further emphasizes the need to make a conscientious effort to make a clinical interpretation of ALP. Society guidance documents also generally endorse the use of introducing MMF as a second-line treatment, likely adapted from practices of iAIH.[49,97,175] Unlike iAIH, the practice of upfront diagnostic liver biopsy before introduction of immunosuppression is not part of the routine recommendations, except in situations where initial high-dose steroids seemingly fail to provide adequate improvement after the first 3 days.[172,176] Also unlike iAIH, no scoring system has been developed guide diagnostic confidence for IMH, as about 1 in 10 liver biopsies could disprove the presence of what was initially suspected to be IMH.[177] After an adequate course of induction steroids, the steroids are to be tapered over 4–6 weeks, although societies have not defined at which point in the treatment the taper should be initiated. Li et al[178,179] have attempted to define a threshold by which steroid responsiveness could be assessed; the latest experience suggests that, after 7 days of treatment with systemic corticosteroids, the alanine aminotransferase (ALT) level should exhibit a decrement of at least 30%, and suboptimal ALT improvement acts as a cue that the patient warrants escalation to adjunctive agents. Currently, a variety of adjunctive agents have been proposed, although lacking are systematic studies (let alone prospective data or even case-control retrospective data) to suggest favorability of one agent over another. However, case reports and published case series continue to emerge for the use of TCZ, with additional retrospective data that include analyses of patients who have been treated with MMF.[180–193] Limited clinical experience has been published with regards to other adjunctive strategies involving tacrolimus, antithymocyte globulin, plasma exchange (PLEX), and intravenous immunoglobulin (IVIG).[194–202]

Some patients with pre-existing chronic liver disease such as steatotic liver disease (SLD), chronic hepatitis B, and chronic hepatitis C, whether or not they have cirrhosis, may undergo ICI treatment and develop IMH. The decision to use any immunosuppressive agents, especially systemic steroids, would require important consideration in this population. Systemic steroids may worsen SLD, and prophylactic antiviral agents (such as entecavir or tenofovir) would need to be considered before initiating steroids to avoid reactivation or flare; this should include patients who also only have positive testing for hepatitis B core antibody without active chronic disease. Periodic hepatitis B quantitative DNA viral load monitoring can be helpful during the immunosuppressive phase. In clinical practice, because prescribing direct-acting antiviral agents for treatment of chronic hepatitis C treatment is generally deferred until after completion of cancer treatments, consultation with a specialist in infectious disease can be useful to provide guidance during the immunosuppressive phase for IMH treatment.

When patients managed for IMH successfully achieve biochemical remission (normalization of all liver enzymes, including ALT level < 40 U/L in general, without evidence of liver dysfunction), the clinician should monitor for sustained steroid-free biochemical remission at least for the duration of the next month. In clinical practice, when patients complete the final day of steroids, subsequent once-weekly liver biochemistry labs for 4 weeks can be performed to ensure that there is no short term resurgence of liver injury. Because of relatively long half-lives of several ICI (ranging from 2 to 4 weeks for agents such as nivolumab, pembrolizumab, tremelimumab, dostarlimab, atezolizumab, and relatlimab), it is important to keep in mind that the window for resurgent IMH remains open for several months, even if ICI are not resumed. As such, periodic liver enzyme monitoring even after cessation of ICI can be considered to identify any unexpected significant changes in liver enzymes. However, no studies are available to delineate the frequency of a spontaneous flare of IMH within the first year of cessation of ICI after patients successfully recover from the initial IMH episode.

Historically, permanent discontinuation of use of ICI was advised for patients who experience (and recover) from high-grade IMH. In practice, the clinician should be precise in defining the aspect of liver injury based on laboratory data and clinical presentation. The interpretations affect implications of the acute clinical presentation and of prospects for future ICI rechallenge. For example, a case exhibiting CTCAE grade 3 ALT and aspartate aminotransferase (AST) elevations alone bears different clinical implications than a case whose CTCAE is considered grade 3 based on bilirubin elevations (concurrently with high ALP levels). These two scenarios also differ from a situation where a patient exhibits acutely elevated INR level due to liver dysfunction, with or without presence of hepatic encephalopathy. In clinical practice, when patients resume ICI, liver biochemistry labs can be monitored initially once a week for the first 4 weeks before resuming standard monitoring before each ICI cycle.

Updates on management options

IMH should remain in the differential diagnosis of new-onset progressive liver injury, even after 90 days of last exposure to an ICI, especially in patients treated with ICI with longer half-lives such as nivolumab and pembrolizumab. IMH presenting as a delayed immune-related event (DIRE) have been reported.[203,204]

Either during initial evaluation of high-grade acute liver injury or a working diagnosis of IMH is established, the clinician is to make an important decision about whether immunosuppression is warranted. If so, systemic corticosteroids, such as oral prednisone, have been the mainstay of initial treatment. The most recent data challenge whether high-dose steroids (greater than 1 mg/kg/d, let alone greater than 60 mg/d equivalent of prednisolone) is warranted, let alone steroids at all.[205–208] A recent prospective study showed that half of patients with high-grade IMH initially who are observed to demonstrate improved ALT continue to improve toward resolution without any steroids.[206] A diagnostic parenchymal liver biopsy performed in the other half of patients actually show variable severity of histologic inflammation, and even those with mild inflammation can also be spared of steroid treatment altogether. This meets or exceeds prior retrospective experience that 30–50% of IMH do not require steroids.[163,206,209,210] In clinical practice, patients with newly identified CTCAE grades 2–4 transaminase elevations without liver dysfunction can be observed with serial liver biochemistry lab tests as an outpatient for a “grace period” of 7 days to prognosticate the need for steroids. Patients who demonstrate consistent improvement within this week can be spared of steroids as long improvement is ongoing. Patients with up to CTCAE grade 1 elevation of ALT with ALP > 1.5× upper limit of normal can be empirically started on ursodiol without steroids initially if the suspicion of a cholestatic or cholangiophatic variant of IMH is high. Li et al demonstrated that systemic corticosteroids greater than 1 mg/kg/d is not associated with faster time to ALT improvement. In iAIH, initial induction doses generally range from 40 to 60 mg/kg/d with later introduction of adjunct (such as azathioprine or MMF) to allow for reduction of steroid doses across time.[175,211] Also in iAIH, oral budesonide has already been established as a viable first-line alternative option of a steroid, which is favorable in patients who may have comorbidities disfavoring prednisone or develop significant adverse event while on prednisone.[211,212] The use of budesonide was initially introduced in 2017 as a possible secondary prophylactic strategy.[213,214] Recent abstracts and case reports demonstrate budesonide as a therapeutic option, sometimes without any need for using systemic corticosteroids; budesonide can also be prescribed bridge from systemic corticosteroids when patients demonstrate improvement or if ongoing use of systemic corticosteroids is disfavored.[215–219] From a tertiary cancer center, preliminary data from a case series of patients from a cancer center demonstrated that budesonide without systemic corticosteroids is capable of optimally treating 90% of high-grade IMH, with optimal response in 93% of those patients transitioned to budesonide from systemic corticosteroids.[219]

Recognition of cholangiopathic phenotypes has garnered additional attention in the past several years. As anti-PD-1 agents becomes more widely used, not surprising is the ongoing reporting in the medical literature and case reports of this particular irAE phenotype.[220–223] The type of response is variable. Although associations are made with suboptimal steroid response or even steroid-resistance, some patients with appreciable biochemical cholestasis (such as high ALP) may also improve with ursodiol monotherapy without use of steroids.[220,224] This clinical observation supports an approach where ursodiol can be empirically initiated even without steroids to ascertain degree of initial improvement, in cases of early identification of cholestatic patterns and/or new bile duct injury. However, more severe phenotypes can lead to sclerosing cholangitis with biliary strictures, entailing the need for serial endoscopic retrograde cholangiopathy procedures with biliary stenting.[222] The cholangiopathic phenotype may also manifest with chronic biochemical cholestasis or chronic radiographic abnormalities on imaging.[221,223,225] Longitudinal studies would be meaningful to understand the natural history of chronic cholangiopathic phenotypes. In clinical practice, long-term ursodiol may be considered for patients who have not normalized levels of hepatobiliary ALP and/or of gamma-glutamyl transferase, including patients with persistent biliary tree abnormalities (such as chronic ductal dilatation or intrahepatic biliary ductal beading) in efforts to attenuate effects of this irAE.

The use of TCZ for refractory IMH requiring additional therapy was initially reported in 2017.[189] Since then, additional case reports, abstracts, and case series have corroborated the potential therapeutic role to help patients achieve biochemical remission.[181,182,184–186,190–193] However, although many cases of TCZ were described in the context of treating IMCH phenotypes, there has not been evidence that TCZ is able to mitigate development of chronic severe sequelae such as biliary strictures. Use of tofacitinib has been reported in three patients.[226] The use of IFX, for which society guidance has suggested against due to risk of DILI (albeit low), has also emerged recently as a therapeutic option for steroid-refractory IMH.[227] Basiliximab (anti-IL-2R antagonist) was administered as a series of infusions in a patient with a significantly refractory form of IMH who exhibited a protracted clinical course.[228] ECP is yet another emerging strategy applied for treating severe refractory IMH.[229,230] Li et al[180] demonstrated that early consultation with a specialist in hepatology is beneficial and is associated with earlier time to achieve biochemical remission.

Since initial studies published in 2018 and 2020 suggested that about 70–80% of patients who undergo ICI rechallenge do not experience recurrence after ICI rechallenge, additional data from four additional retrospective studies (suggesting similar outcomes) and one prospective multicenter study in 2022 that demonstrated 65% of cases of high-grade IMH did not experience recurrence when ICI was resumed.[38,151,152,159,231,232] As such, this provides significant flexibility to individualize the treatment plans after patient recovers from high-grade IMH, rather than endorsing permanent discontinuation of ICI treatment, as suggested in prior society recommendations.

Practice gaps

In relation to the available medical literature that may influence evolving practices, four important topics that represent ongoing practice gaps should be recognized:

1. Routine incorporation of ALP and gamma-glutamyl transferase (GGT) as part of the initial diagnostic analysis (which may influence decisions about steroid-refractory cases and upfront use of ursodiol);

2. Differentiation between biochemical “high-grade” liver toxicity categorization (which relies predominantly on CTCAE grading based on magnitude of ALT and AST levels) from clinically significant hepatic dysfunction (i.e., CTCAE grade 4 ALT without jaundice and normal INR vs CTCAE grade 3 ALT with jaundice and elevated INR);

3. Implementation of a routine practice that incorporates a period of initial observation without immunosuppression for patients with high-grade ALT and AST values without liver synthetic dysfunction; and

4. Prioritization of early escalation to adjunctive strategies, such as MMF, TCZ, azathioprine, tacrolimus, IFX, or ECP, which requires early recognition of a steroid-refractory or steroid-dependent case.

What is next? Preclinical, clinical, and translation work in progress

Clinical predictors for development of IMH as well as predictive biomarkers should be studied. Development of a diagnostic scoring system that ideally emphasizes analysis of parenchymal liver histology to differentiate IMH from other immune-mediated hepatobiliary diseases (such as iAIH, primary biliary cholangitis, or primary sclerosing cholangitis) can help the clinician improve diagnostic accuracy to ensure appropriate decisions with regard to immunosuppression. Clinical trials for patients with pre-existing autoimmune hepatobiliary diseases (ideally during an optimized state) are necessary to establish the safety of initiating ICI treatment in this population to determine additional candidates that could benefit from immunotherapy for cancer treatment. An important area of unmet need includes identification of biomarkers that can be helpful for predicting onset of IMH or steroid-responsiveness. A diagnostic scoring system (such as that which already exists for iAIH) could be helpful to the clinician to ensure a confident diagnosis of IMH before embarking on immunosuppressive strategies, whether in presence or absence of available liver biopsy histologic information. Case-control or prospective studies that compare common management strategies; this may include comparing observation without steroids, prednisone monotherapy, prednisone induction at various doses such as 40 mg/d, 60 mg/d, or 1 mg/kg/d if higher, ursodiol monotherapy, prednisone with one adjunctive agent, budesonide induction without systemic corticosteroids, and early upfront steroid-sparing cytokine-directed treatments such as TCZ. Like immune-mediated colitis, management of IMH should shift its current paradigm to minimizing the cumulative systemic corticosteroids dose and duration, while leveraging steroid-sparing strategies or biologics to overcome steroid-refractory or steroid-dependent cases early, with the goal of making permissive the opportunity for ICI rechallenge. The role of secondary prophylaxis with oral budesonide or intravenous (IV) TCZ is also worth exploring, provided, predictors of recurrent IMH can be delineated with greater precision in select patients, because more than half of studied cases do not experience such recurrence. Finally, because cholestatic or cholangiopathic phenotypes are increasingly recognized as a manifestation especially in patients who have been exposed to anti-PD-1 treatment, longitudinal studies are necessary to better understand and predict scenarios that would prognosticate steroid-resistance or risk of developing severe chronic sequelae such as secondary sclerosing cholangitis.

Conclusion

The understanding and management of hepatobiliary toxicities are evolving. This irAE may present on a spectrum of phenotypes, both biochemically (ALT and ALP elevations), histologically (hepatocellular inflammation with or without bile duct injury), and radiographically (severe cholangiopathic disease with biliary stricture). CTCAE grading that largely focuses on ALT and AST values may limit an accurate clinical assessment of the degree of liver injury, which can vary based on liver histology and by liver synthetic function measures (such as INR elevation with concurrent jaundice, regardless of CTCAE grade of the ALT and AST). A significant proportion of patients with initial “high-grade” ALT elevations are capable of natural resolution in a reasonable timeframe without need for systemic steroids (as is the case with non-ICI-associated DILI). As such, without endorsing a practice that allows for the opportunity for observation in select patients has already led to overtreatment of many cases with unnecessary immunosuppression. Steroid “intensity” upon induction should also be examined closely alongside the effects of adjunctive therapies. As a priority in this field of hepatobiliary irAEs, an effort should be made to steer the field into minimizing use of steroids in general toward use of early steroid-sparing strategies.

Introduction and clinical spectrum of pancreatic irAEs

Pancreatic toxicity, specifically type 3 autoimmune pancreatitis (AIP) secondary to ICIs, is a rare but increasingly recognized irAE in oncology. This section provides a high-level, evidence-based summary of current knowledge, recent advances, and ongoing research, with a focus on the most clinically relevant and recent literature, including emerging survivorship considerations as durable responses to ICIs become more common.

Type 3 AIP, also termed ICI-induced pancreatitis, is a distinct, drug-induced, immune-mediated chronic inflammatory disease of the pancreas. The incidence is low (≤1–2% of ICI-treated patients), and clinical presentations range from isolated, asymptomatic pancreatic enzyme elevations to overt pancreatitis with abdominal pain and, less commonly, exocrine or endocrine insufficiency.[233–242] Treatment-related risk factors include cumulative exposure to PD-1 or PD-L1 inhibitors and the use of combination ICI therapy.[234,239,240,243] Patient-related factors include cancer type and a prior history of pancreatitis.[240] Diagnosis is one of exclusion and is based on a temporal relationship to ICI exposure, fulfillment of clinical criteria for acute pancreatitis when present, exclusion of alternative etiologies, and, when available, biopsy findings demonstrating T cell–predominant inflammation.[233,234,236,240,242,244] Cases are commonly graded according to the degree of serum amylase or lipase elevation.

Existing knowledge and guidelines

Current NCCN guidelines recommend careful assessment for true pancreatic symptoms in patients with mild enzyme elevations, as biochemical abnormalities alone frequently do not reflect clinically meaningful pancreatitis.[241,245] When diagnostic criteria for pancreatitis are met, management is generally conservative with intravenous fluids and analgesia. For moderate to severe cases, ICIs should be held, and in persistent or severe presentations, corticosteroids may be considered if there is no improvement with supportive care alone.[233–242,245–248]

In routine clinical practice, isolated and asymptomatic elevations of amylase or lipase are not uncommon during ICI therapy. In the absence of abdominal symptoms or imaging abnormalities, close observation with repeat laboratory testing is often appropriate. Further evaluation—including cross-sectional imaging, temporary interruption of ICI therapy, or referral to gastroenterology—is generally reserved for patients with persistent enzyme elevation, development of symptoms, or concern for evolving pancreatic injury.

All patients with ICI-associated pancreatic injury warrant longitudinal monitoring of pancreatic function. Practical follow-up strategies may include periodic assessment for symptoms of exocrine insufficiency, routine screening for dysglycemia or new-onset diabetes, and selective imaging in patients with persistent biochemical abnormalities or recurrent symptoms to assess for chronic changes such as pancreatic atrophy.[235,236,240] Early recognition of evolving dysfunction is critical, as glandular injury may progress even after ICI discontinuation.[235,240]

Updates on management options

Regarding diagnostic evaluation, magnetic resonance cholangiopancreatography is particularly useful for differentiating AIP from pancreatic malignancy.[237,249] Positron emission tomography and computed tomography (PET/CT) can identify increased metabolic activity within the pancreas and may be more sensitive than conventional CT for detecting inflammatory changes; together, these modalities provide complementary information for diagnosis, differential diagnosis, and management planning.

Endoscopic ultrasound with fine-needle biopsy (EUS-FNBx) plays an important role in select cases, particularly when imaging reveals focal or mass-like lesions. Histopathology typically demonstrates patchy, lymphocyte-rich inflammation with ductitis and venulitis in the absence of malignancy, supporting the diagnosis of type 3 AIP.[244]

Type 3 AIP is often mild and may resolve with ICI withdrawal alone. Available evidence suggests that corticosteroids do not reliably mitigate abdominal pain, prevent rapid pancreatic atrophy, or reduce the risk of subsequent exocrine or endocrine insufficiency.[235,236,240] Holding ICIs appears to be associated with improved biochemical resolution compared with steroids alone, while relapse rates are similar across management strategies.[235,240] In steroid-dependent or recurrent cases, data supporting the use of additional immunosuppressive agents such as azathioprine or tacrolimus remain limited to isolated case reports.[235]

Rechallenge with ICIs following pancreatic toxicity remains a frequent and challenging clinical question. Lower-risk scenarios, such as transient, asymptomatic enzyme elevations without structural pancreatic changes, may be approached differently from higher-risk contexts that include clinically overt pancreatitis, recurrent episodes, or evidence of progressive glandular dysfunction. Given the limited evidence base and the availability of enzyme replacement therapy, rechallenge decisions though encouraged in general should be individualized, incorporating prior toxicity severity, treatment intent, alternative therapeutic options, and patient preferences.

Practice gaps

Importantly, pancreatic injury may evolve beyond the acute phase. Longitudinal imaging studies demonstrate that more than 50% of affected patients experience greater than 20% pancreatic volume loss within 1 year, and up to 18% develop new-onset diabetes within 2 years, regardless of treatment strategy.[235,240] These delayed metabolic sequelae underscore the need for ongoing surveillance even in patients with apparent clinical resolution.

As ICIs are increasingly incorporated into neoadjuvant and adjuvant treatment paradigms, pancreatic irAEs may carry distinct implications in curative-intent populations compared with metastatic disease. In these settings, tolerance for long-term metabolic complications and the threshold for ICI rechallenge may differ, necessitating careful, individualized risk–benefit assessment.

Despite identified associations, validated tools for predicting pancreatic irAEs are lacking, representing an important unmet need for future research.

What is next? Preclinical, clinical, and translational work in progress

The mechanisms underlying ICI-associated pancreatic injury remain incompletely defined. Preclinical and translational studies implicate T cell–mediated injury as the dominant pathway, characterized by increased CD8⁺ and CD4⁺ T cell infiltration, immune-mediated ductal and acinar damage, and islet involvement.[236,240,242] Histopathologic analyses consistently demonstrate acinar-centric inflammatory infiltrates, atrophy, fibrosis, and stromal remodeling, overlapping with features of autoimmune and chronic pancreatitis.[236,240,244]

Ongoing clinical efforts, including large registries and systematic reviews, aim to better characterize the clinical spectrum, risk factors, and long-term outcomes of ICI-induced pancreatic injury.[233–242] Notably, recent data suggest that most pancreatic irAEs classified as pancreatitis by CTCAE criteria represent asymptomatic pancreatic injury under Atlanta criteria, with only a minority progressing to clinically significant acute pancreatitis.[236]

Conclusion

Type 3 AIP is a rare but clinically meaningful complication of ICI therapy, with heterogeneous presentation and a substantial risk of chronic pancreatic dysfunction. Most cases are mild, and the benefit of corticosteroids appears limited.[235,236,240] As long-term survival improves, delayed endocrine and exocrine sequelae are increasingly relevant survivorship issues. A coordinated, multidisciplinary approach involving oncology, gastroenterology, and endocrinology is essential to optimize early recognition, guide management, and ensure appropriate long-term surveillance for metabolic and digestive complications.[235,240,242]

Introduction and clinical spectrum of pulmonary irAEs

Although relatively less common than other irAEs, the acute and chronic impact of ICI-related pneumonitis (ICI-P) can be significant. Estimates of the incidence in those treated with PD-L1 inhibitors is 4% and in those on anti-CTLA-4 inhibitors is 1%,[250] but combination therapy with PD-L1 and CTLA-4 inhibitors increases the incidence to 7–10%[250] and approaches 10% in real-world cohorts, particularly in non–small cell lung cancer (NSCLC).[251,252] ICI-P increases all-cause mortality, likely due to a combination of acute lung injury and exclusion of ICIs from future regimens. The long-term morbidity is less clear, but one study found that up to one-third of survivors from ICI-P experience residual hypoxemia, and the majority had persistent or worsening dyspnea compared to their clinical status prior to ICI-P.[253]

ICI-P occurs more frequently in cancers associated with tobacco smoking, such as NSCLC and renal cell carcinoma (RCC).[16,254,255] Pre-existing interstitial lung abnormalities (ILAs), which are present in as many as 10% of people who smoke, have emerged as one of the main risk factors for the development of ICI-P. In one study, patients with pre-existing ILAs had a 29% incidence of ICI-P as compared to 10% in those without ILAs.[256] Furthermore, ILAs often progress into interstitial lung disease (ILD), but this progression may take several years to manifest. Nevertheless, the underlying cellular processes that contribute to ILA development may be subtly present when ICIs are first initiated.

Symptoms of ICI-P range from asymptomatic to increasing gradations of dyspnea, persistent cough, fever, chest pain, and/or hypoxemia. Radiographic patterns include peripheral or subpleural patchy or confluent consolidations, ground glass opacifications, or interstitial findings such as increased interstitial markings and interlobular septal thickening.[75] The median time to onset is about 12 weeks but can occur at any time after ICIs are first given and can even occur months after ICI treatment is stopped.[257]

Existing guidelines for management of pulmonary irAEs

ICI-P is typically graded using the CTCAE, which is based upon symptoms, activity limitation, and the requirement for respiratory support.[16] Recent ASCO guideline updates in 2021 recommend including radiographic scoring based on the amount of lung parenchymal involvement.

Grade 1 ICI-P can be managed with monitoring. Guidelines recommend treating patients with grade 2 ICI-P with a 4–6-week taper of 1–2 mg/kg/d of prednisone. For grade 3 or 4 ICI-P, patients receive upfront therapy with 1–2 mg/kg/d of methylprednisolone, followed by a tapering course of oral prednisone. Patients with grade 2 ICI-P can be re-challenged with ICI if symptoms resolve and there is radiographical improvement. In those with grade 3 or higher, rechallenge is not recommended.[75]

Updates on management options

ICI-P unresponsive to corticosteroid therapy occurs in up to 18.5% of cases of ICI-P.[258] In cases of ICI-P that do not improve after 48–72 hours of corticosteroid treatment, second-line treatment options can be considered.[75] IFX 5 mg/kg IV as a single dose is commonly used to treat steroid-refractory pneumonitis. If there is no response to IFX after 1–2 weeks, an additional dose of IFX or TCZ 4 mg/kg IV once can be considered. TCZ has shown benefit in smaller studies of grade ≥ 3 pneumonitis. Recent studies have shown possible improvement with IVIG and IFX in combination with IVIG.[258] Cyclophosphamide may also be of benefit, but this is based on its use in other immune-mediated lung diseases.[75] However, there is no study that clearly demonstrates superiority of one agent over another.

Practice gaps

Clinical practice patterns have not significantly changed since the last set of guidelines, but several gaps in knowledge remain.

Selection of patients for ICI therapies

ICIs are increasingly used in front-line, neoadjuvant, and adjuvant regimens, but the exclusion criteria have not been well-validated and largely mirror those used in pivotal phase 3 studies. For example, patients with autoimmune diseases are often excluded from ICI-containing regimens, but the actual risk for increased toxicity may differ depending on the specific underlying autoimmune condition.[259] A similar paradigm may exist with pre-existing lung diseases, particularly in patients with early ILAs of unknown significance. In these patients, there are no validated biomarkers to understand whether the ILAs are of any clinical significance, or might instead represent prior injury, such as from an antecedent pneumonia or prior therapies. Furthermore, the pulmonary workup of pre-existing ILAs is not often performed. Pulmonary function testing is generally safe and could help identify patients with subclinical ILD who have not yet developed any revealing symptoms, such as shortness of breath or cough, yet these studies are rarely performed prior to ICI initiation and are not endorsed in society guidelines.

Precision of pneumonitis grading

The current guidelines rely on CTCAE grade to manage ICI-P, but key questions remain. First, because there are no validated biomarkers of ICI-P activity, particularly those that are measurable in peripheral blood for patients who cannot undergo bronchoscopy, it is unclear which patients may benefit from a shorter course or more rapid taper of corticosteroids. Second, the categories of ICI-P grade remain broad and unable to differentiate nuances. For example, a patient with mild symptoms may be considered grade 2 if they do not require supplemental oxygen, but the same patient may be considered grade 3 if they have mild hypoxemia. Similarly, a patient with more severe dyspnea without hypoxemia could be considered grade 2 or grade 3, depending on how the clinician views the impact of dyspnea on the ability to perform activities of daily living (ADL). Because the intensity and duration of treatments depend upon the grade of ICI-P, a more precise grading system, either incorporating more quantitative clinical information or with serum or bronchoalveolar lavage (BAL) biomarkers, could better guide the intensity and duration of corticosteroid therapy. Finally, though the recent 2021 ASCO guidelines recommend incorporating radiographic scoring in the grading of ICI-P,[16] there is no evidence that radiographic scoring improves risk categorization. Furthermore, subjectivity may exist due to the semiquantitative nature of quantitative radiographic tools. Incorporation of radiographic imaging into the clinical practice of ICI-P still requires further study.

Diagnosis of pneumonitis

Diagnostic markers for ICI-P with bronchoscopy are lacking, and the procedure is mainly performed to evaluate for infection prior to immunosuppression.[16,260,261] A recent study showed that BAL lymphocytosis was associated with ICI-P, but this marker was not sufficiently sensitive or specific to be used routinely. Another study suggested that BAL and serum CCL18 levels were markedly higher in ICI-P, but this marker also requires validation.[262] Thus, diagnosis relies on clinical suspicion which can result in delays for treatment initiation, potentially worsening ongoing lung injury. Biopsy is not routinely recommended but could help in clinical scenarios concerning malignant disease progression. Further study is needed to understand the potential role for biopsy to meaningfully improve the diagnosis of ICI-P.

Steroid-unresponsive and steroid-dependent pneumonitis

Newer guidelines have defined clinical features of irAEs that do not have a typical response to corticosteroids, including those unresponsive to corticosteroids and those that require persistent corticosteroids for a prolonged period to control inflammation.[263] Steroid-unresponsive pneumonitis is associated with high mortality,[258,264] but key questions remain unanswered. First, while guidelines recommend additional immunosuppression for patients with grade 3 or grade 4 ICI-P who do not improve within 48 hours, it is not clear whether upfront immunosuppression (i.e., immediately upon ICI-P diagnosis) may further improve outcomes. Second, in those who do not improve within 48 hours, it is unclear how much of the pulmonary impairment is due to lung injury as opposed to ongoing inflammation that remains untreated by high doses of corticosteroids. Third, some centers advocate for “pulse dose” corticosteroids upfront,[265] but the timing and dosing of the corticosteroid pulse is not standardized. Fourth, there is no evidence to recommend which additional immunosuppressive agents should be instituted, and practice patterns in different centers exhibit substantial heterogeneity.

Steroid-dependent ICI-P poses significant challenges because prolonged immunosuppression may preclude patients from receiving additional cancer treatment and potentially rechallenge with ICI therapies if they had developed grade 2 pneumonitis. Strategies to manage this condition with steroid-sparing agents such as MMF are used, but the duration of therapy and the ability to restart ICI while on steroid-sparing immunosuppression remain unclear.

Monitoring for resolution of pneumonitis

Patients with ICI-P are typically re-evaluated after a few weeks of corticosteroid treatment to ensure resolution. One challenge imposed by this time interval is the possibility of treatment failure, particularly with more rapid corticosteroid tapers. The use of patient-reported outcomes (PROs) to measure symptom improvement may help ensure that patients have an adequate response to corticosteroids but may be confounded by the higher overall symptom burden following the development of ICI-P. Once patients are re-evaluated, repeat imaging is usually recommended to ensure radiologic resolution, but a significant proportion of patients have ongoing opacities which may represent scar tissue or ongoing inflammation.[253] Because biomarkers of ICI-P activity are lacking, often the decision to continue or halt corticosteroid treatment will depend on the clinician’s overall assessment, which should include a careful review of symptom trajectories. Even in the hands of ICI-P experts, the qualitative nature of these assessments may result in erroneous conclusions for whether ICI-P has resolved completely.

ICI rechallenge after pneumonitis

Rechallenge with ICI-containing regimens is only recommended in patients who have complete resolution of grade 2 ICI-P, potentially excluding up to 40% of patients who survive the initial episode of ICI-P.[251] In many cases, toxicities that occur with ICI rechallenge are not the same irAE as in the first instance, and therefore the concern for recurrent ICI-P, potentially of a higher grade, is not well substantiated by observational data.[38] Patients with an excellent upfront response to ICI may benefit from subsequent rechallenge despite grade 3 toxicity, but no studies to date have examined the safety of this approach. This gap in knowledge particularly impacts patients with metastatic disease, for they may have limited treatment options, and a durable remission may be difficult to achieve without continued ICI treatment. Strategies to mitigate the risk of recurrent ICI-P after rechallenge should be studied and may potentially pave the way for rechallenge scenarios in those with higher grade pneumonitis.

What is next? Preclinical, clinical, and translational work in progress

Biomarkers that denote ILA activity are actively under investigation by ILD researchers, and studies are underway to examine whether some markers of ILD activity can help differentiate which patients with pre-existing ILAs who are at higher risk for ICI-P.

Pragmatic randomized controlled trials, potentially utilizing adaptive designs, are needed to quickly enroll patients and test many agents simultaneously, but this would require a multicenter consortium of trial sites. Further insights into the biology of ICI-P may reveal additional therapeutic targets, and in our view represent an urgent gap in knowledge. Although some cytokines can be measured by most hospital laboratories (e.g., IL-6, IL-1β, and tumor necrosis factor [TNF]-α) and elevations in these cytokines could guide the choice of immunosuppressive agent (e.g., TCZ, anakinra, or IFX), this type of “precision medicine” approach has not been proven to be superior to a more empiric choice of immunosuppressive agent. Furthermore, certain agents that may modulate lung inflammation (e.g., JAK kinase inhibitors) may lack such a diagnostic biomarker which may guide treatment. A better understanding of the pathobiology of ICI-P, utilizing modern translational scientific methods, will also help guide the treatment of high-grade and steroid-refractory pneumonitis.

Conclusion

Despite the growing use of ICI-containing regimens, the management of pneumonitis has not changed significantly over the last decade, though the number of patients who are potentially at risk for ICI-P will likely continue to increase. These critical questions must be addressed with rigorous scientific studies to treat or prevent this potentially fatal outcome.

Introduction and clinical spectrum of neurologic irAEs

Neurologic irAEs (n-irAEs) have been extensively documented as potentially life-threatening complications of ICIs. The fatality rate for severe n-irAEs, particularly in the setting of myasthenia gravis and encephalitis phenotypes, remains among the highest of all organ-specific toxicities, necessitating rapid identification and aggressive intervention.[266] Numerous management guidelines are available with the core diagnostic concepts revolving around clinical recognition of subacute onset neurologic and systemic symptoms, subsequently separating the symptoms into central nervous system and peripheral nervous system phenotypes.

A critical challenge in clinical practice continues to be distinguishing true n-irAEs from disease progression, paraneoplastic neurological syndromes, or alternative explanations such as metabolic encephalopathies.[267] There can be substantial clinical overlap, and reaching a true n-irAE diagnosis continues to rely on the exclusion of other possibilities.

Current guidelines for n-irAEs do not yet include a role for any biomarkers. In this context, the prevailing protocol remains centered on a comprehensive diagnostic assessment relying on imaging, laboratory testing, cerebrospinal fluid (CSF) analysis, and electrodiagnostic studies such as electromyography (EMG).[268] Identifying high-risk clinical scenarios such as anti-PD-1/CTLA-4 combinations, an appropriate constellation of symptoms, and the presence of overlap syndromes such as myositis, myasthenia gravis, and myocarditis is a necessary part of the current diagnostic approach.

Significant heterogeneity in the tissue composition, immune microenvironment, and exposure to circulating blood factors of the various components of the nervous system underlies the broad clinical diversity seen in n-irAE. Given the current lack of generalizable biomarkers across n-irAE syndromes, and the high incidence of comorbid non n-irAE neurological conditions in patients with cancer, identification of these conditions requires close collaboration with consultative neurology. A high index of suspicion is appropriate for patients who have or are experiencing other non-neurologic irAE, and subsequently develop neurological symptoms not readily explained by an alternative known cause. Most severe n-irAE begin within the first 8–12 weeks of treatment, although notably neuromuscular toxicities (such as myositis) can appear earlier with a more fulminant course.[269] These syndromes can be broadly categorized by anatomic compartment (either central or peripheral nervous system)[270] that guides diagnostic testing (Table 4). The components of the nervous system affected, mechanism and severity of dysfunction, as well as the risk of irreversible neurological deficit and morbidity, all influence the selection of immunomodulatory therapy (Fig. 1).[16]

Table 4.

Neurologic irAE classified by anatomic compartment, syndrome, and pathophysiology

	Presenting Signs/Symptoms	Typical Diagnostic Evaluation	Immune Axis
Central nervous system (CNS) syndromes
Meningitis/encephalitis	Altered mentation, confusion, headache, particularly with neck stiffness, photophobia, or phonophobia, seizures	MRI brain with and without contrast, lumbar puncture with opening pressure, cell count, meningoencephalitis panel, cytology, and other studies as appropriate, EEG if fluctuating consciousness or frank seizure, autoimmune or paraneoplastic antibodies in serum/CSF as appropriate, rheumatologic evaluation	T cell mediated parenchymal inflammation, autoimmune encephalitis may be either antibody-mediated (to cell-surface or synaptic antigen) or T cell driven (to intracellular antigen)[270–273]
Vasculitis	Subacute focal neurological deficits, multifocal strokes, headache, stepwise cognitive decline	MRI brain with and without contrast, vascular imaging (CT or MR angiography, with direct catheter angiography as gold standard), exclusion of conventional cerebrovascular risk factors, lumbar puncture (inflammatory profile), rheumatologic evaluation	T cell mediated vascular inflammation
Demyelinating syndromes (akin to multiple sclerosis, acute disseminated encephalomyelitis)	Optic neuritis (pain, vision loss, color perception loss), transverse myelitis (sensory level, bowel and bladder habit changes, paraplegia or quadriplegia), multifocal neurological deficits	MRI brain and total spine with and without contrast, Lumbar puncture (inflammatory profile, cerebrospinal fluid IgG index, oligoclonal bands, myelin basic protein can be helpful), consider testing for aquaporin-4 or myelin oligodendrocyte glycoprotein antibodies as appropriate, paraneoplastic antibody testing and rheumatologic testing as appropriate	Mixed humoral and cellular mechanisms resulting in immune-related demyelination
Peripheral nervous system syndromes
Neuropathy (demyelinating or axonal)	Ascending (length-dependent) weakness, loss of previously present reflexes, neuropathic pain, sensory loss, occasionally autonomic dysfunction	Lumbar puncture (inflammatory profile, rule out mimics such as leptomeningeal disease and infections), neurophysiology (electromyography with nerve conduction studies), metabolic/nutritional and rheumatologic evaluation usually indicated	Mixed humoral and cellular mediated nerve root or axon injury, immune mediated demyelination
Myopathy/myositis	Proximal limb muscle pain and weakness, dysphagia, ocular muscle weakness	Serum creatine kinase, aldolase, neurophysiology (electromyography with nerve conduction studies), muscle MRI, muscle biopsy in select cases	T cell mediated, frequent overlap with systemic immune activation and other irAE (myocarditis, myasthenia gravis)—the “Triple M” syndrome is a fulminant and frequently morbid presentation
Neuromuscular junction (ICI-associated myasthenia gravis)	Fluctuating/fatigable weakness, ptosis, diplopia, bulbar, or respiratory involvement	Myasthenia antibody panel (acetylcholine, MuSK, LRP4), neurophysiology (electromyography with repetitive nerve stimulation or single-fiber EMG), pulmonary function testing	Antibody mediated with prominent T cell dysregulation, high risk for overlap

CSF: cerebrospinal fluid; CT: computed tomography; EEG: electroencephalogram; EMG: electromyography; ICI: immune checkpoint inhibitor; irAE: immune-related adverse event; LRP4: low-density lipoprotein receptor-related protein 4; MRI: magnetic resonance imaging; MuSK: muscle-specific receptor tyrosine kinase.

Figure 1.

Representative approach to stratifying neurologic immune-related adverse event subtype and severity.

IV: intravenous; IVIG: intravenous immunoglobulin; PLEX: plasma exchange.

Existing guidelines for management of neurologic irAEs

Guidelines for the management of n-irAEs universally recommend high-dose corticosteroids as the first-line treatment for grade 2 or higher n-irAEs. The standard protocol involves intravenous methylprednisolone (1–2 mg/kg/d) or oral prednisone equivalents.[268,271] For more prominent symptoms and severe (grades 3 and 4) toxicity, guidelines advise permanent discontinuation of the ICI and consideration of pulsed steroids (1 g/d for 3–5 days) as well as weekly intravenous steroids or an oral steroid taper thereafter. Traditional algorithms suggest escalating to IVIG (2 g/kg total dose) or PLEX immediately for severe cases and only after 48–72 hours without a clear steroid response for less severe symptoms. Maintenance treatment options are based on frameworks from any similar idiopathic counterpart condition and are chosen on a case-by-case basis, with substantial variability persisting in clinical practice.

Updates on management options

Separating central nervous system phenotypes (encephalitis, aseptic meningitis, and transverse myelitis) from peripheral nervous system phenotypes (myasthenia gravis, Guillain-Barré syndrome, and myositis) continues to be a core diagnostic concept. Over the past few years, some “subphenotypes” have been described, including ICI-associated facial diplegia Guillain-Barré syndrome. However, the constellation of visible clinical symptoms remains the first step in accurate diagnosis as no biomarkers have been validated for n-irAEs.[268] The management of n-irAEs continues to shift toward more precise approaches focusing on antibodies and phenotypes. Recent literature increasingly emphasizes the early use of biologic agents and small molecules in part to target specific mechanisms but also to try to limit prolonged steroid tapers. Furthermore, a core shift in practice has been to consider early, aggressive therapy in place of a stepwise approach.

In addition to stratifying decisions by anatomic location, algorithms are increasingly emphasizing antibody status as a critical data point in selecting treatments. This is broadly applicable to any neurologic phenotype and is particularly emphasized in encephalitis n-irAEs. A bifurcated treatment algorithm is permeating clinical practice: patients testing positive for cell surface onconeural antibodies with presumed B cell mechanisms are treated with standardized rituximab protocols, whereas intracellular onconeural antibodies or antibody-negative cases, thought to be driven by T cells, receive cyclophosphamide for broad immunosuppression.[267] Recognition of possible cytokine-driven processes resulting in symptoms, particularly elevated IL-6 in the cerebrospinal fluid, are more recent developments. In this setting, IL-6 inhibition with off-label use of TCZ has been reported.[272]

Treatment protocols continue to be refined based on the severity of symptoms and suspected mechanism of action but also on a more granular level with phenotype-specific recommendations. When selecting therapeutic options, the presence of any idiopathic counterpart continues to provide a framework for any similar ICI-related condition. For example, FcRn antagonists (Efgartigimod), used in classic myasthenia gravis, have recently been used in ICI-related myasthenia gravis (ICI-MG) complicated by myocarditis.[273] This was effective as part of acute, upfront treatment and replaced the traditional role of PLEX. Because of the potential for hemodynamic instability and concurrent myocarditis, the neonatal fragment crystallizable receptor (FcRn) antagonist was used to rapidly reduce circulating IgG levels without intravascular volume shifts. Another example of a recent phenotype-specific change in treatment is notable in the “Triple M” overlap patient population where anti-CTLA-4 therapies such as Abatacept are used to block T cell costimulation.

Broader immunosuppressive treatments such as JAK inhibitors have been used for systemic symptoms but also n-irAEs by suppressing the interferon-gamma signaling pathway.[274] These have been proposed as an option for salvage therapy in severe cases of ICI-encephalitis due to their reported favorable blood–brain barrier penetration profile. This class of drugs is also considered for “bridge” therapy, allowing clinicians to wean high-dose steroids within days rather than weeks. JAK inhibitors have been proposed as a potential treatment option in intracellular onconeural antibodies or antibody-negative cases.

Finally, early aggressive escalation of treatments has been increasingly prominent in clinical practice as recent retrospective analyses suggest that a delay may contribute to the high fatality rate in fulminant cases, particularly in myasthenic crises where respiratory failure can progress rapidly.[266] Limiting the duration of total steroid use is also a priority due to the numerous systemic adverse effects that can occur in the setting of a chronic steroid treatment schedule.[271]

What is next? Preclinical, clinical, and translational work in progress

Current translational and clinical work is moving toward predictive biomarkers and risk-stratified rechallenge protocols. The lack of specific biomarkers has added to the challenge of n-irAE diagnosis and management. Attempting to predict toxicity before visible clinical symptoms is a focus of ongoing research. Some studies have identified Neurofilament light chain (NfL) as a potential biomarker, offering a highly sensitive but nonspecific measure of nervous system inflammation.[272,275] Levels have been shown to rise in the setting of encephalitis n-irAEs, potentially even preceding progressive neurologic symptoms. Protocols are exploring routine NfL monitoring in high-risk patients (e.g., those on combination checkpoint inhibitor therapies) with rising values triggering “pre-emptive” holding of immunotherapy before n-irAEs result in symptoms. However, due to overlap with other causes of nervous system inflammation, a cytokine panel or profile may ultimately provide a more accurate biomarker assessment. With this perspective in mind, some teams are attempting to combine NfL with specific cytokines (e.g., CXCL13 and CXCL10) to create a “neuro-toxicity signature” that distinguishes irAEs from other etiologies.[272]

Work continues on accurate risk stratification in patients with pre-existing neurologic autoimmune conditions, high-risk cancers, and in rechallenge decisions following n-irAEs. Some pre-existing neurologic autoimmune conditions, particularly myasthenia gravis, appear to have a high risk of severe flares in the setting of ICI therapy as they can present with neuromuscular respiratory failure. Other conditions, such as multiple sclerosis, have a more modest risk of exacerbations.[276] It is expected that overtime risk stratifying pre-existing autoimmune conditions will be increasingly accurate. Important oncologic variables, including the type of primary malignancy, have also emerged, and some teams have recommended baseline autoantibody screening in patients with a high paraneoplastic risk malignancy such as small-cell lung cancer.[277]

Neurologic irAE are associated with higher early mortality than other irAEs; however, overall impact on survival and long-term outcomes are heterogeneous and confounded by cancer type and stage. Some components of the nervous system are more readily amenable to recovery, such as remyelination of peripheral nerves, restoration of acetylcholine receptors to the neuromuscular synaptic junction, reversal of toxic-metabolic or cytokine mediated encephalopathy or decrease in vasogenic cerebral edema. Conversely, T cell–mediated damage to brain or spine parenchyma results in more durable neurological deficits. Residual neurological deficits after recovery from n-irAE may impact every speech, memory, personality, strength, sensation, and coordination, all of which significantly impact QOL. In addition, more than half of the patients in a retrospective multicenter study who survived the initial toxicity went on to experience long-term neurological deficits and were at higher risk of death due to cancer progression.[278]

There are also indications that patients who receive ICIs are at risk for long-term neurocognitive sequelae, and that even in absence of frank n-irAE immunotherapy may potentially compound cancer related cognitive impairment (CRCI).[279] Generally, studies of n-irAE outcomes use generalizable neurological functional endpoints such as the modified Rankin scale (mRS), but deficits in cognition (and other symptoms) may fall below the resolution that can be captured by such metrics which are static and emphasize physical disability. Further long-term analysis may require the additional use of disease-specific endpoints such as the quantitative myasthenia gravis (QMG) score and inflammatory neuropathy cause and treatment (INCAT) score, depending on the phenotype. In addition, there is emerging evidence that PROs can provide insight into the burden of delayed neurotoxicities such as cognitive impairment.[280]

After a n-irAE occurs, increasingly granular rechallenge data are available to guide multidisciplinary clinical decision-making. Longer-term data demonstrate that n-irAEs still have the lowest rechallenge rate of any organ system (only 9.5% rechallenged).[281] Although rechallenge is likely feasible for the majority of patients, the recurrence rate of symptoms was approximately 30–40%, significantly higher than previous estimates. Recent studies have identified central nervous system and peripheral nervous system contraindications for rechallenge.[266,267] These include certain phenotypes such as encephalitis, myelitis, and myasthenia gravis due to the risks of severe flares and permanent disability. However, successful rechallenge has been documented even after severe events such as Guillain-Barre syndrome, provided that a medication class switch is used.[282]

Current best practice includes a multidisciplinary approach, early recognition of symptoms, and timely initiation of immunosuppression. Overall grading of neurologic adverse events has been adequately standardized, but the wide range of possible neurologic manifestations has made recognition, granular grading, and selection of optimal treatment challenging. Standard acute management continues to revolve around high-dose steroids with escalation to IVIG or PLEX. With increasing data and clinical experience, the management of neurologic adverse events is increasingly tailored toward more targeted therapies, reflecting a similar trend seen in other organ systems. Some algorithms have been updated to distinguish between autoantibody-mediated syndromes (e.g., MG) and T cell–mediated syndromes (e.g., encephalitis), with attempts being made to move away from a “one-size-fits-all” approach toward increasing use of novel agents, most prominently in steroid-refractory cases.

Introduction and clinical spectrum of rheumatic irAEs

About 10% of patients with cancer receiving ICIs will experience rheumatic irAEs. Inflammatory arthritis, polymyalgia rheumatica (PMR), sicca syndrome, and myositis are the most frequently reported irAEs.[283] Here, we review the clinical presentations and the latest treatment evidence for the most common rheumatic irAEs.

Inflammatory arthritis

ICI-induced inflammatory arthritis (ICI-IA) occurs in approximately 7% of patients receiving ICI, based on the latest studies.[284,285] The pathogenesis of ICI-IA is not completely understood, but it is thought to involve autoreactive B and T cells, which can produce pro-inflammatory cytokines such as vascular endothelial growth factor A (VEGF A), TNF–, IL-6, and IL-17).[286,287] Furthermore, ICIs may induce T cell epitope spreading increasing the diversity of T cell epitope specificity.[288] The potential effect on B cells is less recognized, but ICI may prevent their apoptosis. Interestingly, most patients with ICI-IA do not have the autoantibodies associated with primary rheumatoid arthritis, rheumatoid factor or anti-citrullinated protein antibodies.

ICI-IA can present as polyarthritis (rheumatoid arthritis-like) with symmetrical joint involvement, or oligoarthritis, sometimes with features of spondyloarthritis, such as back pain.[289–291] Unlike other irAE which can be short-lived, ICI-IA can linger for months, even after ICI cessation, often requiring long-term therapy to avoid flares.[292] For instance, a recent retrospective cohort reported that 87% of the patients with ICI-IA were still symptomatic even after 3 months of ICI cessation.[293]

Polymyalgia rheumatica

A polymyalgia rheumatica-like syndrome (ICI-PMR) has been reported in about 1% of patients receiving ICI, although the true incidence might be higher.[294] ICI-PMR’s inflammation appears to be primarily steered by T cell infiltration.[295] In contrast, primary PMR appears to be driven by macrophage inflammation.[295]

Similar to primary PMR patients, ICI-PMR patients will typically complain of bilateral shoulder pain, stiffness and restriction in range of motion, often accompanied by similar symptoms in the pelvic girdle.[296] Patients with ICI-PMR may also develop arthritis in peripheral joints which is not seen in primary PMR.[297] Patients with ICI-PMR do not usually develop giant cell arteritis, which can present with primary PMR.[298] Most studies suggest that ICI-PMR have parallel symptoms to primary PMR, but one recent study also found that ICI-induced PMR patients may have less intense symptoms compared to that of classical PMR.[296,297,299]

Sicca syndrome

Sicca syndrome is estimated to occur between 3% and 24% s receiving ICI based on clinical trial data.[300] The clinical features resemble those of primary Sjogren syndrome, but they tend to occur much more acutely, and the oral symptoms are more common and severe than ocular ones. The pathogenesis of ICI-sicca seems somewhat different from primary Sjogren’s syndrome. Many patients are seronegative, whereas in Sjogren’s the majority have anti-SSA/SSB antibodies.[301] The histopathology reveals major differences between the two diseases. In primary Sjogren’s syndrome, the affected exocrine glands is characterized by periductal mononuclear infiltration mainly composed of T cells (CD4⁺ and CD8⁺) and CD20⁺ B cell.[302–304] In contrast, the histopathology of ICI-induced sicca syndrome’s characteristically shows only a few CD20⁺ B cells and mostly T cells with CD4⁺ predominance.[40,44]

Myositis

Myositis induced by ICIs is rare, with an incidence of 0.4–1.6%.[284,305–307] Similar to other rheumatic irAEs discussed, the pathogenesis of ICI myositis is also not completely understood. However, it is thought that the autoreactive T cells may recognize peptides that are similar or shared with tumor antigen, leading to destruction of healthy tissues.[288,308] For instance, CD4⁺ and CD8⁺ lymphocytes have been found in necrotic fibers in muscle biopsies of ICI-myositis patients.[288]

Myositis induced by ICI presents with weakness in proximal muscles as seen in idiopathic inflammatory myopathy, but its onset can be acute, often accompanied by significant myalgia, and occasionally with rhabdomyolysis. Myositis can also present as inflammation of the extraocular muscles and orbital soft tissues, mimicking symptoms of thyroid eye-disease.[309] Despite its rarity, clinicians should be wary of the disease given its high mortality estimated to be around 22–24%.[306] Furthermore, ICI-induced myositis can co-present with myasthenia gravis and myocarditis, raising its mortality rate to 27% and 51–57%, respectively.[306,307] One systematic review found that 41% of patients with ICI-myositis have myocarditis, which significantly increases mortality.[310] These patients also present with severe myopathy with ptosis and respiratory muscle paralysis.

Existing Guidelines for Management of Rheumatic irAEs

Although corticosteroids remain the mainstay of initial treatment for grade 2 or higher rheumatic irAEs, disease modifying anti-rheumatic drugs (DMARDs) are becoming more important in their treatment, as evidenced by the latest guidelines.[16,49,95,174]

Inflammatory arthritis

Current guidelines recommend nonsteroidal anti-inflammatory drugs (NSAIDs) for mild or grade 1 ICI-IA.[16,49,95,174] Systemic or intraarticular corticosteroids are typically used as a first line treatment for grade 2 or above ICI-IA.[16,49,95,174] However, when patients are unable to taper corticosteroids to a prednisone-equivalent dose of 10 mg daily or below, DMARDs need to be considered given emerging evidence that corticosteroids may reduce ICI’s antitumor efficacy.[13,311] There are some discrepancies among the guidelines on when DMARDs should be considered. For example, while NCCN recommends consider starting DMARDs when prednisone cannot be tapered below 30 mg daily after 1 week of use, other guidelines suggest considering DMARDs when prednisone cannot be tapered below 10 mg daily after 6–8 weeks of use or have no specific recommendations on when to start DMARDs.[16,49,95,174] DMARDs that can be started to help taper glucocorticoids for ICI-IA include methotrexate, leflunomide, sulfasalazine, hydroxychloroquine, IL-6 inhibitors (IL-6Ri), and TNF inhibitors (TNFi).[16,49,95,174]

Polymyalgia rheumatica

All current guidelines recommend starting prednisone 10–20 mg daily with a slow taper anywhere from 4 to 8 weeks.[16,49,95,174] When symptoms are severe, such as those in grade 3 or above, a higher dose of prednisone at 30–40 mg daily can be considered.[16,49,95,174] The inability to taper glucocorticoids or nonresponse to glucocorticoids warrants the use of methotrexate or IL-6 receptor inhibitor (IL-6Ri).[16,49,95,174]

Sicca syndrome

The most recent guidelines recommend supportive measures such as saliva substitutes, salivary stimulants, and systemic sialagogues to treat sicca syndrome.[16,49,95,174] However, prednisone 20–40 mg daily or an oral rinse containing dexamethasone can be used when symptoms become severe.[16,49,95,174]

Myositis

Because of its significant mortality rate, ICIs are typically discontinued after patients develop ICI-induced myositis.[16,49,95,174] With mild or grade 2 myositis, prednisone at 0.5–1 mg/kg daily is recommended. When grade 3 myositis occurs, aggressive treatment with prednisone at 1–2 mg/kg/d is typically started for patients.[16,174] More severe myositis, such as when cardiac or respiratory muscles are involved, requires IV methylprednisolone 500 mg daily to 1 g daily for 3 days followed by a prednisone taper.[16,49,95,174] Additional therapies to be considered include MMF, methotrexate, azathioprine, TNFi, IL-6Ri, rituximab, IVIG, and PLEX.

Updates on management options

Inflammatory arthritis

Conventional DMARDs such as methotrexate and hydroxychloroquine can be used as steroid sparing agents in mild cases, although there is a scarcity of data on the efficacy of these agents. A small retrospective case series of 11 patients reported that hydroxychloroquine was effective in treating ICI-IA and facilitated tapering of corticosteroids.[312] In another small retrospective study of 16 patients, methotrexate adequately controlled ICI-IA without any cancer progression.[313] Biologic DMARDs (bDMARDs) are typically reserved for more severe ICI-IA. Bass et al[314] conducted a multicenter retrospective cohort study evaluating the safety and effectiveness of TNFi, IL-6Ri, and methotrexate in ICI-IA and found that TNFi achieved the fastest response in controlling arthritis, followed by IL-6Ri, and then methotrexate. In another retrospective cohort of 26 patients with ICI-IA, IL-6Ri improved arthritis in all patients and prevented arthritis relapse during ICI rechallenge in most.[315]

Given the potential deleterious effect of corticosteroids and the chronic nature of ICI-IA, the current expert consensus is to start DMARDs if corticosteroids cannot be tapered within 1 month to control symptoms and to allow continuation of ICI therapy. Long-term close rheumatology follow-up is warranted given the chronic nature of ICI-IA and DMARDs’ own toxicity profile (e.g., immunosuppression, blood cytopenias, and metabolic abnormalities).

When considering whether ICIs can still be administered after developing ICI-IA, current data suggest that they typically do not have to be permanently discontinued but may be temporarily withheld when symptoms are severe and adversely affecting ADL.[49,95,174] One multicenter observational study showed that about half of ICI-IA will have arthritis flares after ICI rechallenge and will develop symptoms earlier than when the initial ICI-IA occurred.[316] The same study, albeit small, also showed that immunosuppressive therapy did not prevent ICI-IA flare.[316] Yet, interestingly, in other studies mainly based on retrospective cohort data, IL-6 blockade has shown promise in mitigating irAE development, also after ICI rechallenge, suggesting a possible role in ICI rechallenge prophylaxis to be evaluated in future studies.[315,317] Overall, based on current data, DMARDs can be given during ICI rechallenge to control ICI-IA.[314,318]

Polymyalgia rheumatica

The course of ICI-induced PMR may vary; therefore, treatments may also differ. Some patients may only have symptoms for a short period of time, but some patients will have persisting symptoms even after discontinuing ICIs.[296] Other than corticosteroids, methotrexate and IL-6Ri, which is the only bDMARD approved for primary PMR, have been used as steroid-sparing agents.[182,314,315] Finally, ICIs are usually not permanently discontinued for ICI-PMR.[49,95,174,300]

Sicca syndrome

When diagnosing ICI-induced sicca syndrome, other potential causes of dry eyes and dry mouth must be excluded such as medication or radiation induced.[319] After the diagnosis of ICI- induced sicca syndrome has been established, symptomatic relief involving artificial tears, punctual plugs, saliva substitutes, and sialogogue remain the frontline treatment.[320] However, when symptoms are more severe and affecting patient’s QOL, corticosteroids can be used to restore salivary function.[290,301,321] One case series found that with treatment, 87.5% of the patients can have partial or complete resolution of their symptoms.[322]

Myositis

The choice of a steroid sparing agent often depends on the co-existence of myasthenia gravis or myocarditis with myositis. When myasthenia gravis presents with myositis, IVIG can be used to treat both conditions.[323] When myositis is co-existing with myocarditis, abatacept has been shown to be promising in reducing mortality.[324] Other steroid sparing medications that have been used include azathioprine, methotrexate, MMF, IFX, TCZ, rituximab, cyclophosphamide, and tacrolimus.[305,324–326]

When ICI-myositis only compromises the musculoskeletal system, without cardiac or neurological complications, mortality appears to be low.[305] Patients with myositis will require long-term follow-up by rheumatology and/or neurology, as approximately half of them may have residual weakness requiring longer immunosuppression.[327] Of note, patients who have received high doses of corticosteroids for prolonged periods of time may also develop steroid-induced myopathy, which needs to be considered in the differential diagnosis of patients with persistent weakness despite immunosuppression.

Practice gaps

Inflammatory arthritis

The safety of bDMARDs with respect to cancer outcomes has been highly debatable. Bass et al[314] found that patients with ICI-IA receiving tumor necrosis factor inhibitor (TNFi) had a significantly shorter time to cancer progression than those receiving methotrexate. A similar trend was observed for IL-6Ri which did not reach statistical significance. Yet, other studies have found that bDMARDs did not significantly worsen tumor response and progression, especially in melanoma, and it has been suggested that IL-6Ri might be the treatment of choice for irAE.[328–330] Other bDMARDs such IL-17 inhibitor (IL-17i) and IL-23 inhibitor (IL-23i) have also been proposed for treatment of ICI-IA on the basis of case reports and small case series, but their safety profile remains poorly defined.[331–334]

In summary, both conventional and bDMARDs have shown efficacy for the treatment of ICI-IA. However, because of the nature of retrospective cohort studies and often small sample sizes, whether bDMARDs may affect tumor response needs to be evaluated with larger prospective studies.

Polymyalgia rheumatica

When comparing ICI-PMR to primary PMR, there have been conflicting data about the dosage of corticosteroids needed for treatment. Although previous studies based on case series and a retrospective cohort have found that ICI-PMR required higher corticosteroid doses, one recent single center retrospective cohort study reported the opposite results.[297,299,335] One case series suggested that up to 25% of the patients may need continuing DMARD treatment with drugs such as methotrexate or IL-6 inhibitors.[297] As with ICI-IA, when patients with ICI-PMR are rechallenged with ICIs, about 50% of them may flare.[316] Given the unpredictable chronicity of ICI-PMR, long-term rheumatologic follow-up is recommended to ensure its proper management.

Sicca syndrome

Evidence for steroid sparing agents remains limited, but hydroxychloroquine has been recommended for the management of ICI-sicca syndrome.[336] Although B cell–targeting biologic therapies such as rituximab and belimumab are used in primary Sjogren’s syndrome, there is little evidence as to their role in ICI-sicca.[301,337,338] Typically, ICIs are not discontinued unless symptoms become severe affecting ADL.

Based on the limited data available, ICI rechallenge in patients with sicca may result in less flares than in those with ICI-IA and ICI-PMR. One case series reported that only one out of seven patients had a sicca flare after rechallenge.[301] However, most of the current data are based on small case series. Long-term management should include follow-up with rheumatology, with dentistry to maintain oral hygiene, and with ophthalmology if patient had ocular sicca complications.[301]

Myositis

Because of the high mortality rate of myositis when associated with myocarditis or myasthenia gravis, ICI-rechallenge is primarily dependent on whether patients have these clinical manifestations, in which case, rechallenge is generally not recommended. However, there have been a few case reports of successful ICI-rechallenge in patients who had developed ICI-myositis with myocarditis.[339,340] A careful assessment or the risk-benefit ratio is needed when considering ICI rechallenge. Finally, the data on most of the steroid sparing agents used for myositis are based on small case series; hence, further studies need to be conducted to determine best practices.

What is next? Preclinical, clinical, and translational work in progress

Although the exact pathophysiology of rheumatic irAEs is not completely understood, emerging evidence suggests they are primarily driven by T cells and related proinflammatory cytokines. T_H1-CD8⁺ T cells are increased in the serum and synovial fluid of patients with ICI-IA.[341] Proinflammatory cytokines such as interferon γ (IFN-γ) and IL-6 have been found to elevated as well, raising the possibility that these cytokines can be used as biomarkers for disease activity, and to predict flares.[341,342] In patients who have developed irAEs, certain HLA haplotypes have also been associated with loss of tolerance in T cells and T cell cross-reactivity.[343,344] Hence, specific genotypes may eventually be used as biomarkers for the development of irAEs, if they show high predictive values. Finally, tumor cells may share homology with self-antigens, triggering a response from the already over-activated cytotoxic T lymphocytes, a pathogenetic mechanism that has been described in patients with myocarditis.[288,345] As research into mechanistic aspects of irAE advances, conceivably, it will be possible to identify specific immune profiles that can identify those patients at high risk of developing specific immune toxicities, monitor their disease activity, and predict who may develop complications after ICI rechallenge. Furthermore, the ultimate goal is to find which immune-related pathways may be uniquely associated with toxicity without hindering tumor immunity, so targeted therapy can be used to treat irAE without adversely affecting the efficacy of ICI and cancer outcomes.

Currently, there are multiple clinical trials going on around the world evaluating the efficacy and safety of bDMARDs in treating irAEs. Specific classes of interested bDMARDs include TNFi and IL-6Ri. Few trials are also evaluating the efficacy of IL-6Ri in preventing irAE development. The data generated from these trials will likely address some of the current dilemmas posted by concurrent use of immunosuppressive medications with ICIs.

Conclusion

In general, management of irAEs should follow a stepwise, severity-driven approach that balances toxicity control with preservation of oncologic benefit. Initial management involves prompt recognition, grading, and initiation of corticosteroids for moderate to severe irAEs, with early involvement of organ-specific subspecialists. Escalation beyond corticosteroids should be considered in patients with steroid-refractory disease, recurrent toxicity during taper, or contraindications to prolonged steroid exposure. The choice of subsequent immunomodulatory therapy should be guided by the affected organ system, the presumed immunopathology (e.g., predominantly T cell–mediated vs B-cell–mediated mechanisms), prior treatment response, and the safety profile of organ-appropriate immunosuppressive agents.

When the dose of corticosteroids cannot be decreased in weeks most experts recommend initiation of DMARD therapy. When considering conventional or biologic and targeted DMARDs, the severity of the irAE symptoms should help with the decision. In contrast to conventional DMARDs, bDMARDs tend to have a more rapid onset of action, resulting in faster reduction of disease activity, which may allow a more rapid reduction in the dose of corticosteroids, thereby allowing for reduced corticosteroid usage.[346] Hence, in mild to moderate rheumatic irAEs, conventional DMARDs may be a reasonable option. However, when symptoms are severe, bDMARDs may be the more suitable option.

The current landscape of management of rheumatic irAEs heavily involves corticosteroids, especially at the initiation of treatment, but more recent studies propose a more aggressive approach with steroid sparing agents used early on, to facilitate a quick tapering of the corticosteroid dose, and continuation of ICI therapy. Much of the evidence is based on case reports, case series, and small retrospective cohort studies, and lacks adequate evaluation of potential harms of immunosuppression with different agents in respect to both tumor progression, and other complications such as infections. Hence, more rigorous controlled studies are needed to determine the best approach to the management of rheumatic irAE addressing both the effectiveness and safety of specific agents. A multidisciplinary approach including regular rheumatology follow-up is recommended as many patients will have persistent symptoms, even after discontinuation of ICI therapy, requiring long-term immunosuppression.

Decisions regarding ICI permanent discontinuation or rechallenge should be individualized, taking into account the severity and potential reversibility of the initial irAE, involvement of vital organs vital, availability of alternative cancer therapies, depth and durability of tumor response, and patient preferences. In short, rechallenge is feasible for ICI-IA, ICI-PMR, and ICI-sicca syndrome, but permanent discontinuation is likely warranted in ICI-myositis given its close association with ICI myocarditis and myasthenia gravis. Overall, close longitudinal monitoring for rheumatic irAEs is essential due to its often chronic nature, particularly in patients requiring prolonged immunosuppression or those undergoing rechallenge.

Introduction and clinical spectrum of renal irAEs

Kidney involvement with ICI use is rare, with acute kidney injury directly attributable to the ICI (“ICI-associated acute kidney injury” [ICI-AKI]), occurring in 2–5% of ICI-treated patients.[347–349] However, the development of ICI-AKI has major repercussions—potential treatment delays, permanent ICI discontinuation, prolonged immunosuppression, and irreversible renal damage affecting eligibility for other cancer treatments. Thus, an understanding of the epidemiology, mechanisms, risk factors, clinical features, and optimal treatment of ICI-AKI is critical to minimizing morbidity and mortality in complex patients with cancer.

Epidemiology of ICI-AKI

AKI is a common complication in patients with cancer in general. Several studies have reported that the incidence of any cause AKI (i.e., hemodynamic AKI, obstruction, etc.) ranges from 15% to 20%.[350–352] AKI directly attributable to ICI represents a smaller proportion of these patients (2–5%), although this may be underestimated in the absence of performing routine kidney biopsies and heterogeneity in definitions for ICI-AKI across studies. The predominant histopathologic pattern of injury in ICI-AKI is acute tubulointerstitial nephritis (ATIN), which is observed in 80–90% of biopsied patients with ICI-AKI.[347,353,354] Up to 20% of patients may have other lesions, including acute tubular necrosis and glomerulopathies such as pauci-immune vasculitis, C3 glomerulopathy, IgA nephropathy, membranous nephropathy, and minimal change disease.[354] Importantly, ICIs are associated with renal complications, aside from ICI-AKI, including electrolyte disturbances such as renal tubular acidosis,[355,356] hyponatremia,[357,358] and hypercalcemia,[359] among many others. However, the literature on such electrolyte disorders is largely limited to single-center case series, highlighting the need for more data in this area.

Mechanisms of ICI-AKI

The pathophysiology of ICI-AKI is not well understood. Preclinical models have demonstrated that the inhibition of immune regulatory checkpoints results in unrestrained immune cell activation and diminished immune self-tolerance.[360,361] Several mechanisms for renal irAEs, and specifically ICI-AKI, have been proposed including loss of immune tolerance to renal antigens via shared structural homology with tumor antigens, abundance of CD4⁺ and CD*⁺ memory T cells generating proinflammatory cytokines, de novo generation of autoantibodies, or reactivation of drug-specific T cells.[308,362–365] In addition, chemokine signaling molecules like CXCL9 and 10 may lead to infiltration of activated T cells in tissues, including the kidneys.[366–369]

Clinical features and risk factors for ICI-AKI

ICI-AKI is challenging to diagnose in the absence of a kidney biopsy, as there are no clinical features that are sufficiently sensitive or specific to differentiate it from non–ICI-AKI. This limitation was highlighted in a multicenter cohort study which examined the utility of clinical and laboratory findings for diagnosing ICI-AKI, and found that pyuria, subnephrotic proteinuria, and peripheral eosinophilia were neither sensitive nor specific for the diagnosis.[353] The same study also highlighted the variability in timing of onset of ICI-AKI. Although most retrospective series describe a median onset of 12–16 weeks after ICI initiation and approximately 3 weeks after the most recent dose, ICI-AKI can occur within the first 3 weeks of initiating ICIs, more than a year after initiation, and even several months after treatment discontinuation.[350,353,370,371] The variability in timing of onset suggests that ICI-AKI is an idiosyncratic reaction, like other forms of ATIN, rather than a dose-dependent nephrotoxicity.

Several risk factors for the development of ICI-AKI have been identified, including prior or concomitant irAEs,[353,372] lower baseline eGFR (< 45 mL/min/1.73 m²),[353,354] use of combination ICI therapy,[370] and exposure to proton pump inhibitors (PPIs).[350,353,370,372–375] PPIs are likely one of the strongest and most important risk factors for ICI-AKI, highlighting the importance of minimizing use of these drugs in ICI-treated patients unless absolutely indicated. PPIs are thought to lead to activation of drug-specific T cells, which become latent over time, and introduction of ICIs may lead to reactivation of these drug-specific T cells, loss of tolerance, and subsequently ICI-AKI.[353,373]

Existing guidelines for management of renal irAEs

The management of ICI-AKI depends on the suspected cause. In most cases, ICIs should be temporarily held in patients with moderate to severe AKI (i.e., a doubling of serum creatinine or more from baseline, or the need for dialysis). In any case of suspected ICI-AKI, early nephrology consultation is recommended to evaluate alternative causes of AKI and to discuss the merits of a kidney biopsy. Because ATIN is the most common lesion observed in ICI-AKI, medications commonly associated with ATIN such as proton pump inhibitors, certain antibiotics, nonsteroidal anti-inflammatory drugs, and allopurinol should be discontinued. Early initiation of glucocorticoids (i.e., within 3 days of ICI-AKI) has been associated with a higher likelihood of kidney recovery in ICI-associated acute interstitial nephritis (ICI-AIN);[353] thus, if kidney biopsy may be delayed, empiric glucocorticoid therapy may be warranted. In some cases, biopsy can be foregone in the absence of a plausible alternative etiology (i.e., other irAEs, concomitant proton pump inhibitor use, new sterile pyuria). For ICI-AIN, expert guidelines recommend systemic glucocorticoids at 0.8–1 mg/kg/d prednisone equivalents, with consideration of a short (1- to 3-day) course of pulse-dose intravenous methylprednisolone (0.25–1 mg/kg) for severe AKI (Kidney Disease: Improving Global Outcomes [KDIGO] stage 3: ≥ 3× baseline sCr, sCr > 4.0 mg/dL, or renal replacement therapy [RRT] initiation) or for hospitalized patients, followed by a transition to oral glucocorticoids.[376–378] In the setting of ICI-associated glomerulonephritis, data on optimal management are limited, and immunosuppression should be tailored to the specific lesion.

At present, the optimal treatment duration for glucocorticoid treatment in ICI-AIN is unknown. Although guidelines suggest glucocorticoids should be tapered over 6–8 weeks,[16,176,377,378] depending on the severity of AKI, two observational studies found that shorter durations of glucocorticoids (< 4 weeks) were not associated with a higher risk of recurrent ICI-AKI.[379,380] Accelerated tapers should be particularly considered in patients at high risk of complications from glucocorticoid therapy (i.e., among patients with diabetes with suboptimal glycemic control, those with refractory hypertension, and elderly patients at risk for fractures).

Overall, ICI-ATIN is quite responsive to glucocorticoids, with studies suggesting that up to two-thirds of patients will have kidney recovery, particularly with early initiation.[353] However, data on the reversibility of other irAEs, such as renal tubular acidosis and other electrolyte abnormalities, are lacking.

Updates on management options

“Second-line” immunosuppressants are emerging as promising treatment options for ICI-AKI, both for refractory or relapsing ICI-AKI, and potentially even as upfront therapy.[381] IFX, for example, has been used for other glucocorticoid-refractory irAEs, such as ICI-colitis and ICI-myocarditis.[378,381–383] Data from randomized clinical trials are needed to directly compare the utility of glucocorticoids vs other agents for ICI-AKI. Upon resolution of ICI-AKI, the risk of recurrent ICI-AKI with rechallenge ranges from 15% to 25%, with no clear association between recurrence risk and the severity of the initial AKI.[353,370,371,384] Prophylactic glucocorticoids are not routinely recommended based on existing data.[377,378] In addition, studies have not observed differences in ICI-AKI recurrence rates when rechallenge with the same vs a different class of ICI.[353,370,371] Use of PPIs is discouraged at the time of rechallenge.[377,378] Data from prospective studies are needed to guide optimal rechallenge strategies. Multidisciplinary discussions may be warranted prior to rechallenge, with consideration of the need for ongoing ICI therapy, the severity and histology of the initial ICI-AKI event, competing oncologic needs, biomarker information when available, and PPI exposure.

Practice gaps

There is an urgent need for novel, noninvasive biomarkers capable of improving diagnostic accuracy and enabling more timely, precise differentiation of ICI-AKI from other causes of AKI. Currently, kidney biopsy is the diagnostic gold standard for diagnosing ICI-AKI. However, the procedure is invasive, carries a meaningful bleeding risk, and can be logistically challenging to organize in a timely fashion. In addition, patients with cancer often have multiple comorbidities, including the need for therapeutic anticoagulation that cannot be discontinued, anemia, thrombocytopenia, and/or a solitary kidney, which can make pursuing a biopsy challenging. Although awaiting a biopsy, patients may be unnecessarily exposed to glucocorticoids and their ICI may be temporarily or permanently held, potentially affecting oncologic outcomes. Although the presence of certain laboratory findings or clinical risk factors may raise the pre-test probability of ICI-AKI, their limited sensitivity and specificity highlight the shortcomings of current diagnostic tools.

Additional practice gaps remain with regards to the optimal treatment for ICI-AKI, and whether there is a role for steroid-sparing agents (e.g., IFX), as upfront therapies. Ultimately, head-to-head randomized clinical trials are needed to compare steroid-sparing agents directly to glucocorticoids.

What is next? Preclinical, clinical, and translational work in progress

Biomarkers

Several blood and urinary-based biomarkers show promise in diagnosing ICI-AKI. Studies have suggested that serum C-reactive protein and soluble IL-2 receptor, as well as urinary markers such as retinol binding protein and tumor necrosis-α, may help differentiate ICI-AKI from AKI from other causes.[371,382,385,386] However, these studies had limitations, including small sample size and in some cases, lack of inclusion of controls who were treated with ICIs and developed AKI from other causes.

CXCL9 is an interferon-γ–induced chemokine secreted by activated T cells that promotes leukocyte chemotaxis, differentiation, and proliferation. Recently, one multicenter study found that urinary CXCL9 reliably differentiated ICI-AKI (specifically, biopsy-proven ICI-ATIN), from patients with non–ICI-AKI, ICI-treated patients with stable kidney function, as well as controls treated with other anticancer agents (e.g., cisplatin and vascular endothelial growth factor inhibitors).[387] Other studies have also suggested that urinary CXCL9 may have utility for the diagnosis of ICI-AKI;[388] however, this test is not routinely available for clinical purposes yet.

In addition to blood and urinary-based biomarkers, radiographic biomarkers such as fluorodeoxyglucose-PET/CTs (FDG-PET/CTs) may help differentiate ICI-AKI from other causes. One study found that the area under the curve for the change in the mean standardized uptake value (SUV_mean; a measure of radiotracer uptake), from baseline FDG-PET/CTs prior to ICI initiation to scans obtained at the time of AKI, was 0.97.[389] However, larger, prospective studies are needed to validate these findings, along with validated prediction tools or risk scores to identify patients at highest risk of ICI-AKI.

Special populations

ICIs can be used in patients with kidney transplants. However, the risk of T cell–mediated and antibody-mediated allograft rejection is 30–40%;[390,391] therefore, several factors must be considered prior to initiation including history of previous rejection, transplant duration, presence of donor-specific antibodies.[392] Recent studies suggest that a combination of glucocorticoids and mechanistic target of rapamycin (mTOR) inhibitors may be a potentially a feasible alternative to calcineurin inhibitors for immunosuppression.[390,391]

Finally, data on use of ICIs among patients with pre-existing autoimmune disease (i.e., systemic erythematous lupus with renal involvement, anti-neutrophil cytoplasm antibodies [ANCA] vasculitis) are limited. However, ongoing trials are underway to assess the safety of ICI use in these complex patients (NCT03816345).[393]

Conclusion

ICI-associated renal complications can lead to temporary or permanent discontinuation of ICIs, along with a higher risk of chronic kidney disease (CKD), which can jeopardize eligibility for additional anticancer treatments. Ultimately, there is often a need to balance and weigh competing interests, such as mitigating kidney toxicity from ICIs vs minimizing cumulative immunosuppression, as well as salvaging remaining kidney function vs resuming ICIs despite the risk of recurrence. This challenge is particularly prescient given the expanding use of ICIs in neoadjuvant and adjuvant settings, underscoring the need for additional data and close multidisciplinary collaboration.

Introduction and clinical spectrum of immune checkpoint inhibitor-related myocarditis

ICI-related myocarditis (ICIMy) is the most rare and fatal of all the irAEs with an incidence of around 1%.[394] However, with over half of the United States cancer population being eligible to receive ICIs their increased use will lead to a significant number of patients developing ICIMy.[395] The initial reports of ICIMy captured the most fulminant cases and reported mortality up to 50%, but as recognition of this condition has improved the mortality has decreased in more contemporary studies.[394,396] Fulminant ICIMy presents with life-threatening arrhythmias, such as advanced atrioventricular block and ventricular tachycardia or fibrillation, and cardiogenic shock. As the recognition of ICIMy has improved a wider spectrum of disease has been recognized with more patients with mild symptoms, typically nonspecific such as fatigue and dyspnea, and troponin elevation.[397] The improvement in ICIMy-related mortality is likely due to two factors;[1] earlier recognition of a wider spectrum of disease and[2] more aggressive immunosuppressive treatment.

In addition, after the acute episode of ICIMy continued monitoring and management is needed for lasting effects. Although most of the morbidity and mortality of ICIMy occur within the first 60–120 days after diagnosis, several chronic and long-term complications may persist among survivors. A large international registry comprising 748 patients with ICIMy found that 33% of patients developed major adverse cardiovascular events (defined as severe arrhythmia, heart failure, respiratory muscle failure, and/or cardiomyotoxicity-related death).[398] Pathologic studies from myocardial biopsy specimens indicate that lymphocytic infiltration of the myocardium and conduction system can have lasting effects leading to myocardial fibrosis and conduction abnormalities.[16] Patients may need persistent heart failure therapy and implantable cardiac device therapy for heart block. Troponin levels may remain persistently elevated even after clinical improvement with median time to normalization of Troponin T being 133 days and Troponin I being 17 days. Prolonged elevation could reflect ongoing myocardial injury and inflammation and may require closer observation.[399] Outpatient follow-up with a cardio-oncologist in a short interval may be needed with serial troponin serologies. If there is myocardial dysfunction, follow-up echocardiogram evaluation may also be useful to assess for clinical improvement while on immunosuppressive therapy and for titration of goal-directed medical therapy for heart failure.

Favorable prognostic factors in late phase ICIMy differ from acute phases which rely heavily on extreme troponin elevation and reduced left ventricular ejection fraction (LVEF). For patients who survive the initial complications of ICI myocarditis, substantial troponin decrease (≥ 42%) by day 8 of immunosuppression, normal QTc interval, normal LVEF at presentation predict favorable long-term survival (> 360 days).[400] Although guidelines recommend permanent discontinuation of immunotherapy after ICIMy, some limited reports have called for cautious consideration of rechallenging patients on a case-by-case basis. For grade 2–4 ICIMy, the NCCN guidelines recommend permanent discontinuation.[96] For nonfulminant ICIMy presenting as asymptomatic troponin elevation without significant clinical cardiac injury in the setting of advanced cancer with no alternative therapies, rechallenge may be considered in a multidisciplinary decision between oncologists and cardiologists. If this is considered, stringent cardiac monitoring should be performed through troponin surveillance and periodic cardiac imaging to ensure early detection of any further myocardial injury.

Existing guidelines for management of ICIMy

The current mainstay of treatment is the use of steroids but the timing, dose, and duration of steroids are not well established. Most guidelines recommend early initiation of steroids when suspicion of ICIMy arises even before diagnosis is established.[16,95,96,401] This is based on a retrospective cohort study of 126 patients from 26 sites that showed when patients received steroids within 24 hours of symptom presentation their mortality was significantly reduced compared to within 24–72 hours and > 72 hours.[402] Early initiation of steroids is especially important when patients are presenting with a concomitant major adverse cardiovascular event, such as life-threatening arrhythmias (advanced atrioventricular block, ventricular tachycardia or fibrillation) and/or heart failure or cardiogenic shock, or with concurrent neuromuscular irAEs such as myositis and/or myasthenia gravis. Once diagnosis has been established, another area of consideration is the starting dose of steroids to be used for treatment. All of the expert consensus guidelines recommend initiation with pulse dose steroids (501–1000 mg IV methylprednisolone daily for 3–5 days) with the exception being the ASCO irAE clinical practice guidelines which recommend 1–2 mg/kg.[16,95,96,401] The recommendation for pulse dose steroids is also based on the aforementioned retrospective cohort study in which patients that received pulse dose steroids had significantly reduced major adverse cardiovascular events compared to intermediate (60–500 mg/d) and low dose (< 60 mg/d) steroids (22.0% vs 54.6% vs 61.9%, respectively, p < 0.001).[402] Finally, the duration of steroid therapy and the optimal taper schedule is unclear. Most expert consensus guidelines recommend a taper over 4–6 weeks; however, the European Society of Cardiology Cardio-Oncology guidelines recommend a much slower taper lasting 17 weeks.[16,95,96,401] Experience at MD Anderson has generally been to taper steroids over 4–6 weeks to allow patients to proceed with their next line of cancer treatment.[403]

Updates on management options

The first case reports of ICIMy were published in 2015, and over the past decade, the treatment has evolved as more research and literature provided further insights.[404] Previous guideline recommendations have only recommended use of immunomodulatory treatment agents beyond steroids for refractory ICIMy.[16,75,174] Refractory ICIMy has loosely been defined as not responding to steroid therapy as assessed clinically or by biomarker response. The European Society of Cardiology Cardio-Oncology guidelines from 2022 were the first to suggest use of steroids plus another immunomodulatory agent upfront in patients with fulminant ICIMy.[401] The choice of which immunomodulatory agent to use is still not established and guidelines give a wide range of options including but not limited to abatacept, mycophenolate, and IFX.[16,18,75,401]

Practice gaps

Several guidelines from a myriad of societies have published management and treatment suggestions, but gaps still exist without any US Food and Drug Administration (FDA)-approved treatment for ICIMy.[16,95,96,401] The following will address these gaps and future directions for managing ICIMy. In patients presenting with milder symptoms or asymptomatic troponin elevation it may be prudent to first perform a comprehensive diagnostic workup and establish a diagnosis of ICIMy prior to initiating treatment.[403] This is a nuance that is not highlighted in current guideline recommendations, but we believe should be considered because mislabeling a patient with ICIMy has significant implications on their future cancer treatment. Despite the recommendation from most guidelines for pulse-dose steroids, a recent multicenter retrospective cohort registry comprising 748 cases of ICIMy demonstrated that, on multivariable analysis for association of 30-day major cardiomyotoxic events, the dose of steroids was not significant (< 1 mg/kg hazard ratio [HR] 0.81, 95% confidence interval [CI], 0.43–1.54; 1–2 mg/kg HR 0.80, 95% CI, 0.51–1.26; and > 2 mg/kg HR 0.88, 95% CI, 0.62–1.25).[398] This contemporary data highlights the previously mentioned fact that recognition of ICIMy has improved and with improved recognition a wider spectrum of disease is now identified as opposed to only the most fulminant cases. This suggests that treatment needs to be tailored to the severity of presentation rather than a “one-size-fits-all” approach. Further research is needed to identify high-risk vs low-risk patients early and to test tailored treatment approaches for these groups. Some of the early work on this has tried to use results of endomyocardial biopsy to identify lower risk groups whereas others have used a variety of clinical factors.[398,405]

Typically, while being treated for myocarditis a patient will not be started on their next line of therapy which can affect overall cancer outcomes. This point is highlighted from a retrospective cohort which showed that severity of myocarditis was associated with increased 1 year cardiovascular related mortality, but OS was not different.[406] It is important to always consider both the cardiovascular and cancer outcomes when determining the best treatment course for patients.

Another trend in the treatment of ICIMy has been the early initiation of nonsteroidal immunomodulatory agents. Several different agents with various mechanisms of action have been reported in case reports and case series. These include but are not limited to abatacept, ruxolitinib, antithymocyte globulin, mycophenolate, intravenous immunoglobulin, PLEX, IFX, and rituximab. The guidelines do not make a specific recommendation which of the nonsteroidal immunomodulatory agents to choose or when to initiate them. Most experts in the field consider early initiation in patients not responding to steroids or with fulminant disease (presence of major adverse cardiovascular events or overlap neuromuscular irAEs). However, the definition of nonresponse is also unclear and there are not established troponin thresholds to guide this decision and is often based on the discretion of the treating clinical team. The current trend in the field is to use abatacept, which is a fusion protein of IgG1 and extracellular CTLA-4 domain that inhibits T cell activation. The predominant inflammatory infiltrate in ICIMy are cytotoxic CD8 T cells making abatacept an attractive already available pharmacologic agent to treat ICIMy. In a CTLA4⁺/CTLA4⁻, PDCD1⁻/PDCD1⁻ murine model that recapitulates the myocardial inflammatory infiltrate and arrhythmogenic disturbances observed in humans, when mice were given abatacept, there was a profound improvement in survival.[407] This was followed by a prospective cohort of 30 patients who received steroids, abatacept, and ruxolitinib (JAK kinase inhibitor) compared to 10 historical controls who received various treatments not including abatacept or ruxolitinib. The treatment cohort mortality decreased to 3% compared with 60% in the historical control group.[396]

What is next? Preclinical, clinical, and translational work in progress

The prior studies with abatacept have led to two ongoing randomized controlled trials evaluating the use of abatacept in ICIMy. The first is the ATRIUM trial (NCT05335928), a phase III multicenter trial comparing abatacept 10 mg/kg and steroids vs placebo and steroids which is evaluating major adverse cardiovascular events as its primary endpoint. The second is the ACHLYS trial (NCT05195645), which is a dose finding study randomizing patients to abatacept 10, 20, and 25 mg/kg and evaluating CD86 receptor occupancy for its effectiveness in inhibiting T cells.[408] The results of these two studies will hopefully clarify the use of abatacept in patients with ICIMy. Even so, the question still remains whether all patients require immunosuppressive therapy beyond steroids or if these agents can be used in the absence of steroids or with dose and duration reduction.

Conclusion

In a little over 10 years since ICIMy was recognized as a rare but highly fatal disease, the understanding of how to manage and treat this disease has evolved. Now, with clinical trials evaluating the effectiveness of novel therapeutic agents, it is expected that the standard of care treatment for ICIMy will continue to change and improve outcomes for patients. Evolving guidelines have helped to improve awareness of this rare disease and to standardize the diagnosis and treatment. Guidelines will need to be updated and revisited especially as ongoing large clinical trials complete and report their results.

Introduction and clinical spectrum of hematologic irAEs

Hematologic irAE associated with ICI treatment may not be recognized when they first occur, despite their potential severity. Although rare, they carry high morbidity and mortality and require prompt multidisciplinary management. These events may go unrecognized as irAE as they may mimic conventional chemotherapy toxicity, marrow infiltration by tumor, sepsis, or cancer progression. Diagnosis and attribution to ICI treatment can be challenging given limited diagnostic tools. Hematologic irAE should be considered in all patients exposed to ICI who develop severe or persistent cytopenias, disseminated intravascular coagulation (DIC), thrombotic microangiopathy, or other hematologic syndromes that can’t be attributed to other etiologies.

Hematologic irAEs are estimated to occur in < 1–5% of patients receiving ICIs, but real-world studies suggest that there is substantial under-reporting.[409,410] True incidences are hard to determine as most of the data are obtained from observational studies, often health claims databases, with results subject to reporting bias. Cytopenias are confounded by other possible etiologies including conventional chemotherapy. The most common events include anemia, thrombocytopenia, and neutropenia, with severe CTCAE grade 3–4 cytopenias more frequently associated with combination ICI or chemotherapy plus ICI regimens. Post-marketing pharmacovigilance data highlight rare but potentially life-threatening hematologic irAE such as thrombotic thrombocytopenic purpura (TTP), autoimmune hemolytic anemia (AIHA), aplastic anemia (AA), hemophagocytic lymphohistiocytosis (HLH), and even acquired FVIII deficiency. Many of these are reported as individual cases, small case series, or reviews of these published cases.[411–413] Median onset ranges from 3 to 10 weeks after treatment, though delayed presentations many months later are reported.[414]

Mechanistically, hematologic irAEs arise from disruption of immune tolerance. PD-1/PD-L1 blockade leads to unchecked T-cell activation and may promote autoantibody production. CTLA-4 inhibition enhances autoreactive T-cell priming, and combined regimens heighten risk. Events may be autoantibody mediated (AIHA, immune thrombocytopenia [ITP]), T-cell mediated (neutropenia, pure red cell aplasia [PRCA]), cytokine-driven (HLH), or represent endothelial cell and immune dysregulation (thrombotic microangiopathy [TMA] syndromes and DIC). Teasing out one specific mechanism responsible for the toxicity is difficult, and in many situations, may be a combination of pathways leading to the hematologic irAE.[6]

Initial evaluation and management

Initial workup of suspected hematologic irAEs includes CBC with peripheral smear review, reticulocyte count, hemolysis labs (lactate dehydrogenase [LDH], total and indirect bilirubin, haptoglobin), direct antiglobulin testing (Coombs test), ferritin, and coagulation studies (prothrombin time [PT], partial thromboplastin time [PTT], fibrinogen, and often D-dimer for suspected TMA or DIC). Bone marrow biopsy should be performed for persistent cytopenias or for unexplained grade ≥ 3 cytopenias, including AA, PRCA, or HLH.

Management of all types of hematologic irAE includes discontinuing or holding ICI, which can be the first step for grade 1 or mild irAE. Corticosteroids are used as first-line therapy for grade ≥ 2 cytopenias or lack of response to discontinuation of ICI. Additional treatments directed at the specific hematologic irAE will be described below. Supportive care is also important as many patients may require aggressive transfusion support and close observation for bleeding, infection, and organ dysfunction.

Therapies directed at the specific hematologic irAE, such as steroids or immunoglobulin G (IgG), are based on severity, with mild grade 1 toxicities usually managed with observation alone after stopping ICI. Findings consistent with grade 2 toxicity may respond to ICI discontinuation and steroids only, with other therapies directed at the specific hematologic irAE reserved for more severe grades or prolonged duration. Treatments in these patients follow the standard treatment approach used for de novo development of the disorder, especially for AIHA, ITP, TTP, and others. In most types of hematologic irAE, more severe manifestations are more frequently associated with mortality.[415]

Existing guidelines for management of hematologic irAEs

Society guidance for ICI associated hematologic irAE are incorporated within their broad documents covering irAE in multiple organ systems. Many of these guidance documents were published early after the introduction of ICI, when emerging toxicities were first recognized. Some have had updates, but most are a few years old. The major societies with published hematologic irAE guidance include ASCO, NCCN, ESMO and SITC.[16,18,49,75] Among these, ASCO provides the most detailed hematology-specific guidance, however even this section is brief with limited scope and admits that hematologic toxicities are “poorly described.” NCCN uses its typical algorithm-based flow chart format for the major hematologic irAE, with all consensus statements classified as category 2 A: uniform consensus based on lower-level evidence. As noted earlier, hematologic irAE are rare compared with other organ system irAE, and are associated with high mortality, therefore data for management are limited. Guidance often suggest consultation with a hematologist and to follow standard hematologic disease management strategies after the general recommendations of holding ICI and/or starting steroids depending on the hematologic irAE severity at presentation. The following updates on management of hematologic irAE are intended to provide a more comprehensive and current approach to hematologic irAE management by including updated data on frequency and presentation; they also incorporate the clinical perspective of a senior hematologist experienced in the evaluation and management of these ICI-associated toxicities.

Updates on management options

Cytopenias

Autoimmune hemolytic anemia

Among the cytopenic hematologic irAE associated with ICI use, AIHA is the most frequently reported. Warm AIHA (WAIHA) predominates, typically arising within 8–12 weeks of therapy. Direct antiglobulin tests are positive in most but not all cases, as with other etiologies of WAIHA.[7,416,417] Other lab findings diagnostic for hemolysis, such as elevated LDH, total bilirubin, and reticulocytes, may be altered by the cancer, infection, or other cancer treatments, often making it hard to confirm the diagnosis. Management of grade ≥ 2 WAIHA anemia starts with holding or stopping ICIs and administering high-dose glucocorticoids. Rituximab is reserved for steroid-refractory disease. Endogenous erythropoietin levels are often high if renal function is normal; for those with CKD or AKI, use of erythropoiesis stimulating agents (ESA) may be of benefit, although the impact of ESA on promoting tumor growth is still unclear.

Immune thrombocytopenia

ITP can present with abrupt severe thrombocytopenia. Evaluation must exclude bone marrow infiltration by primary tumor, which can often be indirectly assessed based on marrow signal on CT or magnetic resonance imaging (MRI) obtained for other indications, as well as TMA, DIC, or antibiotic and other drug mediated thrombocytopenia. A recent large database analysis examined 86,467 patients treated with ICI, with ITP found to occur in just 214 (0.25%) with a median onset of 8 weeks after starting ICI. Predictors for ICI associated ITP included low baseline platelet count, combination ICI treatment, stage 4 cancer, and presence of other irAE.[418] Standard therapy in patients with grade 3–4 toxicity, especially for platelet counts < 20–30,000/µL or with bleeding, includes corticosteroids and IVIG. Although thrombopoietin receptor agonists (TPO-RA) such as romiplostim or eltrombopag are often reserved for prolonged thrombocytopenia, in practice starting TPO-RA early and at maximal dose in those with platelet counts < 10,000/µL may hasten platelet recovery.

Neutropenia and agranulocytosis

Neutropenia and agranulocytosis represent severe but much less frequent hematologic irAE, reported in roughly 0.17% ICI treated patients.[419] As with anemia, it is often initially difficult to determine the etiology of neutropenia or agranulocytosis due to other possible or confounding contributors if ICI is given with conventional chemotherapy. The neutropenia in irAE however is often more severe and will persist past the expected recovery time from chemotherapy. Onset is often abrupt, with rapid development of profound grade 4 neutropenia, defined as ANC < 500/µL; however, patients can develop severe neutropenia with ANC < 50/µL. Fatal agranulocytosis has been reported in 1 of 7626 ICI treated patients.[412] Mechanism is generally thought to be T-cell mediated, especially agranulocytosis. Bone marrow biopsy can be useful, but if recent cytotoxic chemotherapy has been given, it may need to be repeated. Grade 4 neutropenia and agranulocytosis require immediate discontinuation of ICIs, use of broad-spectrum antibiotic prophylaxis, and use of combinations of the following agents based on severity of neutropenia or agranulocytosis: granulocyte colony-stimulating factor, high dose corticosteroids, and IVIG.

Aplastic anemia and pure red cell aplasia

AA and PRCA are very infrequent, with reported rates of 0.15–0.6% for both, however they are among the most life-threatening hematologic irAEs.[420] AA presents with persistent pancytopenia and hypocellular marrow, while PRCA presents with profound anemia out of proportion to suppression of other cell lines, absent reticulocytes, and no marrow red cell precursors. These are often diagnosed days to weeks after initial presentation, when ICI discontinuation and subsequent treatment with steroids have not been effective. Bone marrow aspirate and biopsy are mandatory to make the diagnosis of either AA or PRCA. Data for management are scarce, given the infrequent occurrence. The pathophysiology of AA or PRCA is likely a T cell mediated process and not autoantibodies. Interestingly, response to a combination of steroids and high dose IVIG is often more rapid than that seen for AA or PRCA not related to ICI treatment; however, close follow-up after steroids are tapered is required in case of relapse. Rapid return of granulocytes for AA and rapid increase in reticulocytes are often seen following IVIG treatment. If there is no response to these treatments, next line therapy includes treatment extrapolated from that given for de-novo presentations of AA or PRCA, including antithymocyte globulin and cyclosporine, although tacrolimus or sirolimus are plausible alternatives.[421,422] There is one recent case report of successful use of eltrombopag for ICI related AA, as is used in de-novo AA.[423] Referral to hematologists experienced in the care of these disorders is required if steroids and IgG do not produce a rapid response, as these cases require expert monitoring and supportive care. Although hematopoietic stem cell transplantation (HSCT) is often treatment for either primary AA or PRCA, the presence of active malignancy often precludes its use in refractory irAE-related cases.

Thrombotic microangiopathies and disseminated anticoagulation

Thrombotic microangiopathies and DIC are uncommon but require immediate treatment with more aggressive approaches than just withholding ICI in most cases as organ dysfunction and even death can occur rapidly. Clinical findings common to both include finding schistocytes on peripheral smear accompanied by increased LDH and increased indirect bilirubin. Elevated reticulocytes, common in non-ICI TTP or DIC, may not be present if conventional chemotherapy has also been given, or if there is marrow involvement by tumor. Thrombocytopenia is also found in both, although often more severe in TTP. Routine coagulation tests should include PT, PTT, fibrinogen, and D-dimer level. Although these will be abnormal in both, they are more significantly altered in DIC, with more severe decrease in fibrinogen and higher elevation of D-dimer. Although scoring systems for predicting TTP and DIC are available, the parameters included in these scores may be affected by other concomitant processes including cancer, and so are sometimes misleading or hard to interpret. ADAMTS13 activity and inhibitor levels should be checked in those suspected of having TTP, ideally before treatment is initiated.

Early recognition of TTP and DIC is essential despite the infrequent occurrence with immediate discontinuation of ICI; just 35 cases of ICI related TTP based on FDA adverse event reporting were identified by 2022.[424] Although ICI associated TTP is an antibody mediated process, steroids alone are insufficient to induce rapid remission before organ damage occurs. Urgent treatment with PLEX is required for suspected TTP.[425] ADAMTS13 activity level and inhibitor assay results can take 5 days to return at most institutions and should not be waited for to initiate lifesaving treatment with PLEX. ICI associated hemolytic uremic syndrome (HUS) although not a hematologic toxicity is a TMA and is differentiated from TTP by elevation of creatinine out of proportion to more modest thrombocytopenia. It has been much less frequently reported, therefore best treatment if stopping ICI and steroids are not successful is not clear, but use of complement inhibition is a rational approach (eculizumab) as it is used for non-ICI HUS treatment.[426,427] As in patients without cancer, DIC is often considered secondary to other irAE manifestations, such as colitis, hepatitis, or pneumonitis, that result in significant inflammation with sepsis-like physiology or from rapid tumor lysis. It requires aggressive transfusion support with coagulation factors and platelets to manage the associated consumptive coagulopathy after withdrawal of the ICI. Treatment of the underlying trigger is critical for resolution.

Hyperinflammatory syndromes

ICI therapy has been associated with hyperinflammatory syndromes including HLH and cytokine release syndrome (CRS). These syndromes share similar clinical and laboratory features with macrophage activation syndrome (MAS), a term which is usually reserved for those with hyperinflammatory syndromes in association with rheumatologic disorders, usually in the setting of single allele inherited mutations resulting in impaired cytotoxic regulation, which also form the basis for familial HLH. HLH, CRS, and MAS are overlapping syndromes; CRS and MAS are often considered forms of HLH. All differ from each other by the type of triggering events and the degree of immune disruption, including variations in released cytokines, and consequently treatment.

HLH occurs much less frequently than the cytopenic irAE. It manifests as severe cytokine-driven systemic inflammation with marked hyperferritinemia, hypofibrinogenemia, hypertriglyceridemia, elevated soluble Il-2 alpha receptor levels (sIL-2R alpha also known as CD25), pancytopenia, organ involvement especially splenomegaly, and hemodynamic instability. Although a scoring system comprising major and minor diagnostic criteria is available for those suspected of having HLH outside of the context of malignancy and ICI treatment, it has not been validated for ICI-mediated HLH.^[427] HLH following ICI treatment can be as severe as in noncancer settings, where it is associated with a 41% mortality.[427]

HLH is frequently suspected but can be a difficult diagnosis to confirm. Any inflammatory condition will result in an elevated ferritin so that the Hscore in almost any patient post ICI treatment for this parameter of ≥ 500 ng/mL will be positive. With HLH, however ferritin is excessively high, with greater than 10,000 ng/mL considered a specific diagnostic finding in pediatric HLH.[428] A prediction score for cancer-related HLH has been developed, using a higher ferritin threshold of 1000 ng/mL combining it with sIL-2R levels; however, application to patients treated with ICI was not performed.[429] As with an elevated ferritin, pancytopenia can be the result of a variety of processes including chemotherapy, marrow infiltration by tumor, infection, sepsis, or other medications such as antibiotics. ICI-related CRS shares some of the clinical manifestations of HLH but is usually not associated with significant suppression of at least two cell lines or complete pancytopenia. Profound hypofibrinogenemia is also not typically seen in CRS. Soluble IL-2R levels can be difficult to interpret, as they can be modestly elevated early on; however, HLH will often have continuously increasing levels that will also persist for a longer period of time. Bone marrow biopsy must be performed to diagnose HLH, as the treatments go beyond stopping ICI and giving high dose steroids, and can include the use of etoposide if HLH is refractory to high dose steroids and cytokine inhibitors.

One study tried to distinguish between HLH, CRS, and sepsis in 35 ICI treated patients with hyperinflammatory syndromes (9 irHLH-like, 18 high-grade irCRS, 8 sepsis) by assessing a large number of biomarkers and circulating leukocyte phenotypes. Twenty-four circulating biomarkers differentiated HLH from grade 3 CRS, with increased levels of ferritin and hepatocyte growth factor (HGF) indicating HLH; these findings were associated with better specificity and sensitivity than the HScore. Sepsis was found to be distinct from CRS by leukocyte phenotypes and cytokine levels (including IL-6, IL-7, fibrinogen, granulocyte macrophage-colony stimulating factor [GM-CSF], and EGF). A high ferritin and elevated HGF were indicative of HLH, while leukocytosis with elevated IL-6 favored sepsis. These findings are of interest; however, they were retrospectively derived from a small number of patients with many of the assessed biomarkers not readily available in clinical practice.[430]

If steroids fail to induce a response, both ICI-related HLH and CRS will respond to anti-cytokine agents but with different efficacy profiles based on pathophysiology and cytokine levels, with the caveat that data are from small numbers of patients and even case reports alone. CRS has been shown to respond well to TCZ (anti-IL-6)[430,431] and HLH less so, while anakinra (IL-1 receptor antagonist) and emapalumab (anti-IFN-γ) have been found to be effective in HLH, in combination with steroids.[432,433] The JAK inhibitor ruxolitinib has also been trialed in HLH with demonstrated responses at the case report level; a trial of ruxolitinib is underway in combination with lower doses of steroids and etoposide in both malignancy-associated and nonmalignancy-associated HLH. (NCT06160791).

Venous thromboembolic events

Conflicting results about the risk of venous thromboembolic events (VTE) due to ICI have been reported, with some results suggesting that ICI confers an increased risk while others have reported no differences.[434,435] There is a high likelihood of under-reporting in the prospective randomized controlled trials of ICI as VTE was not included as a related adverse event. Currently available data are from retrospective observational studies, with reported incidence rates of 2.1% at 6 months, while others have found 6-month cumulative incidence of VTE between 8% and 13% in small numbers of patients, often analyzing single tumor types, single class of ICI.[434,436,437] The most comprehensive analysis of 10,638 VTE-naïve patients treated with ICI found a 6-month VTE incidence of 7.6%, increasing to 11.1% at 12 months, with substantial variation by regimen. Dual CTLA-4/PD-1 blockade was associated with higher VTE risk, while PD-L1 inhibitors were associated with lower risk compared with PD-1 monotherapy.[438] As use of primary VTE prophylaxis in patients with cancer has not been widely adopted, likely due to concern for bleeding, primary VTE prophylaxis cannot be recommended based on use of ICI alone at this time.

Practice gaps

Rechallenge with ICI

Whether patients with a history of ICI irAE can be rechallenged with ICI after resolution of hematologic irAE is a frequent question for hematologists. Data for recurrence rates with re-exposure to ICI are even less confident than for the incidence of hematologic irAE. One WHO database study examined 24,000 patients treated with ICI; of these 6123 were rechallenged with the same ICI, with just 452 providing informative information about recurrence of the index irAE. In 130 cases, rechallenge resulted in recurrence of the initial toxicity (28.8%; 95% CI, 24.8–33.1), with colitis, hepatitis, and pneumonitis predominating. Of those with an initial hematologic irAE and rechallenged, 30% (95% CI, 10–61%) had recurrence of the same irAE.[38]

For those with grade 1 toxicity or grade 2 with rapid response to steroids, as is frequently the case for antibody mediated irAE such as WAIHA or ITP, rechallenge may be considered, but for those with grade 3–4 toxicity or more concerning neutropenia, AA, TTP, or HLH, and other irAE with life threatening potential, it is generally not recommended. No data has been published for rechallenge following successful treatment of AA or PRCA. Reported recurrence rates for WAIHA after rechallenge in were found to be 14.3%, however the numbers were small with a very wide CI (95% CI, ∼0.4–57.9%) as just 14 patients were identified for this analysis.[439] In a study of ICI associated ITP, 30% of a small number were re-exposed to ICI and developed recurrent ITP.[418] However, 80% of those with neutropenia or agranulocytosis were reported to recur after ICI rechallenge.[419] As with any treatment, but especially in those with cancer, the individual benefits and risks need to be carefully addressed.

Relapse and late onset hematologic toxicity

Treating ICI hematologic irAE differs from treating de novo humoral and lymphocyte mediated auto-immune hematologic disorders as just stopping the ICI, i.e., removing the inciting agent, can often be effective treatment for low grade irAE manifestations. Similarly, a short course of steroids to suppress inflammation may be sufficient to induce complete resolution of the hematologic irAE. In other cases, however, more aggressive treatments are required to halt persistent hematologic manifestations of ICI toxicity. Hematopoietic cell count recovery, whether due to autoantibodies or T cell–mediated suppression or destruction, may be rapid due to steroid interference with antibody mediated macrophage clearance of target cells, but may rebound when steroids are withdrawn if immunologic remission has not occurred. Close follow-up and monitoring are required for many patients following hematologic irAE.

Late onset hematologic irAE are extremely rare but interesting phenomena with reports documenting development of hematologic complications ascribed to ICI treatment 1 year or more after discontinuation.[22,440] Enduring immune dysregulation following ICI treatment completion is not surprising, and can be viewed as similar to that seen with rheumatologic disorders or B cell lymphomas, particularly chronic lymphocytic leukemia (CLL). All of these have abnormal B and T cell functions, albeit with different drivers. CLL for example is known to be associated with both WAIHA and ITP that is independent of any treatment.[441,442] Why irAE occurrence can be delayed for up to a year or more after treatment is discontinued is unclear. It is plausible that subtle triggers occur in the ICI reprogrammed immune milieu leading to production of autoreactive T cells or antibody production. Rigorous reporting of these late events is unlikely, therefore true incidence and confounding factors are unknown.

Conclusion

Hematologic irAE following ICI treatment are infrequent but can be associated with high morbidity and mortality. Further investigation of the mediators of hematologic ICI induced irAE is required. More rapid recognition and diagnosis, identification of factors associated with increased risk, and improved understanding of the immunobiology and triggers of irAE are needed to be able to standardize guidance for treatment. As indications for use of ICI expand into earlier disease settings, irAE may have different clinical presentations, as those with advanced stage cancer have the presence of other confounding hematologic problems. The recognition and management of hematologic irAEs will become increasingly important for optimizing the outcomes of treatment with ICI.

Introduction and clinical spectrum of audiovestibular irAEs

Hearing loss is a widely known toxicity of both chemotherapy and radiotherapy with significant impacts on long-term QOL outcomes.[443–445] As immunotherapy has become more commonly employed for different pathologies, reports of audiovestibular dysfunction have surfaced as a notable toxicity but have been largely underrecognized when compared to other irAEs.[446–448] Symptomatology of irAEs can afflict the cochlea, the vestibular system, or both. A crucial aspect of management and identification is appropriate auditory monitoring with baseline audiogram prior to immunotherapy initiation.

Audiovestibular dysfunction (also described as ototoxicity) presents in varied ways. Hearing loss associated with irAEs may be unilateral or bilateral in nature;[447,449] for patients with bilateral hearing loss, it can be asymmetrical or symmetric between the two ears. Hearing loss typically manifests as sensorineural hearing loss (affecting the cochlea or cochlear nerve) as shown by an increase in pure-tone thresholds on audiometric testing; with increasing severity, the patient may also experience decreased word understanding. Tinnitus can accompany hearing loss and is usually described as high-pitched and nonpulsatile.[447] Vestibular dysfunction may emerge as acute vertigo or a general sensation of imbalance and disequilibrium.[448]

Existing guidelines on management of audiovestibular irAEs

The overwhelming majority of published literature regarding audiovestibular irAE has been in case reports and small case series regarding patients receiving immune-checkpoint inhibitor (ICI) therapy.[447,450–454] Hearing loss can occur unilaterally or bilaterally; bilateral hearing loss can be symmetric or asymmetric. Auditory dysfunction is frequently accompanied by tinnitus.[447] Wierzbicka et al and Naples et al combined available case reports and series into more comprehensive reviews focusing on the presentation of ICI-related hearing loss.[455,456] Wierzbicka et al[456] showed that median time from ICI initiation to ototoxicity was 3 months, but the range at which ototoxicity can present is wide and imprecise due to limited data. The current landscape is not nearly comprehensive enough to predict the overall incidence of irAE ototoxicity, partially due to the lack of routine auditory monitoring in many of these patients.[451] Hambach et al[449] published the first retrospective review among 12 patients with adequate auditory monitoring receiving ICI, with one patient (8.3%) experiencing a 10 dB shift in two consecutive frequencies in only one ear. The literature regarding ototoxicity from immunotherapy is sparse in regards to projected incidence, severity of toxicity, and underlying mechanism.

Ototoxicity may arise concurrently with other irAEs due to widespread adaptive cell immunotherapy;[457] similar to the presentation of Vogt-Koyanagi-Harada syndrome, where melanocytic destruction can occur in multiple organ systems, including the inner ear.[458] Tampio et al highlighted a case of combined audiovestibular dysfunction with comorbid uveitis, and Guven et al showed that over 60% of patients with hearing loss had at least one other irAE present.[448,457] It is always important, however, to remember to rule out leptomeningeal involvement or metastases for patients with new hearing loss or vertigo in light of their active cancer diagnosis, which is exemplified by the report by Trach et al[459] of a patient with new onset deafness and acute vestibular function that was thought to be an irAE but was subsequently found to be leptomeningeal metastasis. Hearing loss and vestibular dysfunction due to metastasis is likely not to improve and signifies disease progression, which is important for continued treatment and prognosis. Ototoxicity related to immunotherapy, however, has the ability to improve quickly with treatment with corticosteroids (discussed below); improvement can signify a lower likelihood of metastases as the etiology, but patients with improvement in symptoms should still undergo appropriate imaging to rule out a metastatic lesion.

Updates on management options

Management of audiovestibular irAEs should be individualized based on symptom severity, degree of functional impairment, and oncologic context. In patients with mild or asymptomatic changes on auditory testing without functional impact, close clinical observation with repeat audiometric monitoring every 3 months may be appropriate while immunotherapy is continued. In patients with moderate or severe ototoxicity, management includes consideration of cessation of immunotherapy and treatment with corticosteroids to decrease systemic immunologic activity. Corticosteroids can be administered via intravenous, oral, and intratympanic routes, and timely audiometry should be repeated to confirm treatment response.[451,455] Choi et al reported a case of a patient with significant recurrent bilateral hearing loss and disequilibrium while receiving ICI therapy. During these episodes, the patient experienced marked steroid responsiveness quickly upon initiation of IV methylprednisolone 1 mg/kg/d, which was transitioned to oral prednisone 1 mg/kg/d for 1 week and tapered over 30 days.[450] This particular patient experienced multiply recurrent audiovestibular symptoms and was treated with corticosteroids during each episode with at least partial recovery each time. For patients who cannot receive systemic corticosteroids, one or more intratympanic steroid injections can be offered. Different formulations exist such as methylprednisolone 20–40 mg/mL and dexamethasone 8–24 mg/mL. The steroid diffuses into the inner ear via communication with the round window of the cochlea. Systemic and intratympanic steroids have been able to improve vestibular and tinnitus related symptoms in addition to audiological dysfunction in some cases, but full recovery may not always occur.[450,451,455] Even with symptomatic improvement, patients can continue to have recurrent episodes of audiovestibular dysfunction despite ICI cessation.[450] We have seen in our practice patients with no improvement or partial recovery after appropriate treatment, which highlights the potential for ototoxicity to have lasting, chronic implications for patients.[451] For patients with chronic hearing loss that has stabilized, audiologic testing and clinical evaluation by an otolaryngologist should be completed annually or biannually. Affected patients with resultant long-term hearing loss may require rehabilitation with traditional hearing aids or cochlear implantation to improve communication and QOL.[453]

Tinnitus, which may not resolve with steroid treatment, may be managed via tinnitus retraining strategies in addition to rehabilitation of hearing loss, if applicable. If the tinnitus is causing dysfunction, such as significant anxiety, depression, or insomnia, anti-depressant medication has been shown to be effective for symptoms affecting survivors.[460] Similarly, patients with long-term vestibular hypofunction may require additional vestibular testing and rehabilitation. Patients with prolonged symptoms of ototoxicity from their ICI therapy should follow-up with an otolaryngologist to guide long-term management and optimize QOL, as there are ways to improve chronic symptoms of audiovestibular dysfunction.

Immunotherapy rechallenge can be considered, particularly for patients who experience improvement in hearing or balance function and do not have other viable cancer treatment options. There is currently no available literature to guide determination of which patients are suitable for ICI rechallenge after ototoxicity; the NCCN recommends that resuming ICI after an irAE can be considered for severity grade 2 or 3 irAEs that at least improve to grade 1, but does not make a specific recommendation for ototoxicity.[18] The decision for rechallenge needs to be individualized and discussed thoroughly in a shared decision-making process among the treatment team, the patient, and their loved ones. An assessment of the severity of audiovestibular symptoms should be made by the patient, as they need to determine if the risk of worsening hearing, tinnitus, or disequilibrium is personally acceptable. Close monitoring is needed in these cases with cessation of ICI if the patient experiences worsening symptoms.

Practice gaps

The pathophysiology of irAE ototoxicity is not completely understood, which affects understanding best diagnosis, monitoring, and management. With adoptive cell immunotherapy, melanocytic cells are targeted via T cell receptor signaling, which is currently hypothesized as a mechanism for ototoxicity. Although melanocyte death is desirable for treatment effect in certain pathologies, melanocytic cells in nondiseased organ systems can be affected, which includes inner ear melanocyte-like cells known as intermediate cells. Intermediate cells live in the stria vascularis part of the cochlea. It is a thin, vascularized structure that maintains the cochlear endopotential, which is important for intracochlear electrochemical signaling.[461,462] Disruption of these cells and their function results in hearing loss. The vestibular and cochlear portions of the inner ear are soft tissue spaces of endolymph and perilymph combined within the same osseous mold; disruptions to the cochlear endopotential can affect all functions within the inner ear. Further investigation is required to precisely define the mechanism, which may yield a fine-tuned approach for management of ototoxicity arising from immunotherapy. Heat-shock protein and biomarkers signaling upregulation of T cell immunity (interferon-gamma) have been implicated as markers of autoimmune sensorineural hearing loss but remain unexplored in ICI-related ototoxicity.[463–466]

What is next? Preclinical, clinical, and translational work in progress

In a recently published retrospective data regarding auditory screening in 29 platinum-naïve patients receiving ICI at MD Anderson,[467] hearing loss was defined by three grading scales: the National Cancer Institute CTCAE version 5, the American Speech-Language-Hearing Association (ASHA) criteria, and the TUNE scale.[468,469] We found, depending on the defining criteria, that 20–44% of patients experienced hearing loss after receiving ICI therapy, suggesting that this irAE may occur in a higher than expected number of patients. This finding reinforces the argument that patients receiving ICI should consider undergoing regular auditory monitoring with baseline testing done prior to treatment initiation. Longer time interval from ICI initiation to follow-up hearing testing was predictive of experiencing a hearing loss; otherwise, no additional confounding patient factors were found. Further prospective studies are needed to assess the true risk, optimize management strategies, and determine the mechanism of ICI-related audiovestibular dysfunction. With more complete evidence, true guidelines surrounding this phenomenon can be created.

Conclusion

Given the emerging evidence of audiovestibular dysfunction in patients receiving immunotherapy, our team advocates for baseline audiogram testing in patients initiating immunotherapy. This allows for comparison of pre-ICI and post-ICI audiometric testing in patients who subjectively detect a change in their hearing. Ideally, post-ICI audiometric testing would follow 6–12 months after starting therapy.

Introduction and clinical spectrum of ocular irAEs

Although ocular toxicities are relatively uncommon compared with other systemic side effects including dermatologic, GI, or endocrine irAEs, ocular irAEs can have a significant impact on patient function and overall QOL. Reported ocular irAEs span a broad clinical spectrum, ranging from mild and asymptomatic to potentially vision-threatening inflammation, including uveitis, scleritis, retinal vasculitis, optic neuritis, and neuro-ophthalmic complications. Delayed or insufficient management may result in irreversible vision loss and may necessitate holding or discontinuing immunotherapy.[16]

The diagnosis and management of ICI-associated ocular toxicity present unique challenges for clinicians. Symptoms may be nonspecific, or occur in isolation without concurrent systemic irAEs, leading to delays in referral and treatment. Furthermore, treatment decisions frequently necessitate a balance between preserving vision and ensuring effective cancer management. This highlights the importance of early detection, standardized grading, and timely ophthalmologic evaluation of toxicity. Emphasis is placed on early recognition, risk stratification, and individualized management approaches, as well as the critical role of multidisciplinary collaboration between oncologists, ophthalmologists, neuro-ophthalmologists, and other subspecialists.[49,469]

ICIs can cause a diverse spectrum of ocular toxicities arising from immune dysregulation targeting both tumor and normal eye tissues. By inhibiting key immune regulatory pathways such as PD-1/PD-L1 and CTLA-4, ICIs disrupt peripheral immune tolerance and promote autoreactive T cell activity, leading to inflammatory processes within immune-privileged sites such as the eye. Although ocular irAEs are relatively uncommon, their potential severity and risk of irreversible visual impairment necessitate a high index of clinical suspicion and early intervention.

The mechanisms underlying ICI-related ocular toxicity remain under active investigation. Current evidence suggests a multifactorial process involving both innate and adaptive immune responses: enhanced activation of autoreactive T cells, increased cytokine production, and loss of immune privilege within ocular tissues, ultimately promoting inflammation and tissue damage.

Additional contributing factors may include cross-reactivity between tumor and ocular antigens, unmasking of subclinical autoimmune disease, genetic susceptibility, and preexisting ocular or systemic autoimmune conditions. Definitive predictive biomarkers are currently lacking.[74,470]

Conjunctiva and ocular surface

Ocular surface disease is among the most frequently reported ocular irAEs in patients on ICIs, with dry eye disease (keratoconjunctivitis sicca) and conjunctivitis, meibomian gland dysfunction representing the most reported ocular irAEs. These manifestations often represent early or mild ocular toxicity, but their high prevalence and chronicity make them clinically important. Progressive disease may result in corneal complications, including epithelial breakdown, filamentary keratitis, infection, scarring, and, rarely, vision-threatening sequelae.

Symptoms may be bilateral, insidious, sometimes associated with other systemic irAEs. Ocular surface disease associated with ICIs is thought to result from immune-mediated lacrimal gland dysfunction and direct inflammatory injury to the ocular surface. This immune dysregulation may lead to both aqueous-deficient and evaporative dry eye. Chronic inflammation may persist even after cessation of immunotherapy. Although typically graded as low-grade toxicities, untreated or inadequately managed ocular surface disease can significantly impair QOL and negatively affect adherence to cancer therapy.

Management is guided by disease severity and often begins with preservative-free artificial tears, lubricating ointments, and environmental modifications. Anti-inflammatory therapies, including TCS or immunomodulatory agents such as cyclosporine or Lifitegrast, may be required for persistent or moderate-to-severe disease. In selected cases, punctal occlusion or autologous serum tears may be considered. Frequent ophthalmic monitoring and collaboration with oncologists are crucial for management while facilitating the continuation of immunotherapy when possible.[471,472]

Uveal tract

Uveitis is one of the most clinically significant and potentially vision-threatening oculars irAEs associated with ICIs. It may involve any part of the uveal tract, anterior, intermediate, posterior, or panuveitis. The time of onset is variable ranging from weeks to months after initiation of therapy, delayed presentations have also been reported with or without other systemic immune-related toxicities. Patients are commonly present with ocular pain, redness, photophobia, floaters, and decreased visual acuity, though posterior involvement may be insidious and asymptomatic. Certain phenotypes, including Vogt–Koyanagi–Harada–like or sarcoid-like uveitis, have been described in association with ICIs.

ICI-related uveitis may occur in isolation or with systemic irAEs has been reported with both monotherapy and combination immunotherapy. A thorough evaluation is essential to exclude alternative etiologies, including infectious, metastatic, paraneoplastic, or treatment-related causes, particularly in immunocompromised oncology patients.

Unrecognized and untreated uveitis can lead to significant ocular morbidity, including secondary cataract, and glaucoma, cystoid macular edema, retinal scarring, and irreversible vision loss. Multimodal imaging, including optical coherence tomography (OCT)and fluorescein angiography (FA), may be required to assess posterior segment involvement and guide management. Treatment is tailored based on disease severity and anatomic involvement. Mild anterior uveitis may respond to TCS and cycloplegic agents, whereas intermediate, posterior, or panuveitis often necessitates periocular or systemic corticosteroids. In refractory or recurrent cases, escalation to additional immunosuppressive or steroid-sparing therapies may be required.

Optic nerve and neuro-ophthalmic complications

Optic neuritis and other neuro-ophthalmic irAEs are rare but potentially vision and life-threatening complications of ICIs therapy. These toxicities may involve the optic nerve, chiasm, optic tracts, or higher visual pathways and can present with acute or subacute vision loss, visual field defects, dyschromatopsia, pain with eye movements, and relative afferent pupillary defects (RAPD). Funduscopic findings may range from a normal-appearing optic nerve in retrobulbar involvement to optic disc edema or, in chronic cases, optic atrophy.

Optic nerve involvement may occur in isolation or as part of a broader neuroinflammatory syndrome, including aseptic meningitis, encephalitis, myelitis, or hypophysitis. Bilateral involvement and atypical features such as poor steroid response, recurrent episodes, or severe visual loss should raise concern for alternative or overlapping etiologies, including paraneoplastic syndromes, metastatic disease, infectious processes, or demyelinating disorders such as neuromyelitis optica spectrum disorder (NMOSD) or myelin oligodendrocyte glycoprotein antibody-associated disease. Prompt evaluation is crucial and should involve urgent neuroimaging with contrast-enhanced MRI of the brain and orbits, laboratory testing to exclude competing diagnoses, and CSF analysis. Early initiation of high-dose systemic corticosteroids is critical to maximizing visual recovery.

In severe, bilateral, or steroid-refractory cases, escalation to additional immunosuppressive therapies such as IVIG, PLEX, or other steroid-sparing agents may be required. Decisions regarding interruption, suspension, or permanent discontinuation of ICI therapy must be individualized depending on severity of neuro-ophthalmic toxicity against benefit and should be made in close coordination with the treating oncology team.

Diplopia on the other hand is an important neuro-ophthalmic manifestation of immune motor nerve palsies, neuromuscular junction dysfunction, or inflammatory myopathy. ICI-MG although rare, may represent a severe and potentially life-threatening complication. Patients may present with fluctuating diplopia, ptosis, ophthalmoplegia, fatigable weakness, dysphagia, or respiratory compromise, often with rapid progression and overlap syndromes involving myositis or myocarditis. Onset frequently occurs early after initiation of immunotherapy and may be more aggressive than idiopathic myasthenia gravis. Prompt recognition, exclusion of alternative etiologies, and early initiation of immunosuppressive therapy are critical. Management typically includes high-dose systemic corticosteroids, acetylcholinesterase inhibitors, and, in severe cases, IVIG or PLEX. Given the high morbidity and mortality associated with ICI-MG, immunotherapy interruption or permanent discontinuation is often required, multidisciplinary coordination between neuro-ophthalmology, neurology, and oncology teams is crucial.[18]

Retina and choroid

Retinal and choroidal irAEs, while rare. These manifestations reflect immune dysregulation directed against retinal and choroidal tissues and may occur independently or in association with other ocular or systemic irAEs. This may include retinal vasculitis, chorioretinitis, multifocal serous retinal detachments, Vogt–Koyanagi–Harada–like syndromes, and paraneoplastic- or autoimmune-like retinopathies, including melanoma-associated retinopathy. Clinically, patients may present with blurred or decreased vision, central or paracentral scotomas, photopsias, metamorphopsia, or nyctalopia. Symptoms may be bilateral and can evolve rapidly.

Diagnosis requires multimodal retinal imaging including OCT for identifying intraretinal or subretinal fluid, photoreceptor disruption, and choroidal thickening. FA may demonstrate vascular leakage, capillary nonperfusion, or vasculitis. In select cases, electrophysiologic testing may aid in diagnosing autoimmune or paraneoplastic retinopathies. Management frequently requires systemic immunosuppression, initiated with high-dose corticosteroids, followed by tapering based on clinical response. Refractory or recurrent disease may need immunomodulatory. Close ophthalmic follow-up with serial imaging is essential. Decisions regarding continuation, interruption, or discontinuation of immunotherapy should be individualized, weighing visual prognosis against oncologic benefit, and made through close multidisciplinary coordination.

Mild ocular toxicity

Mild ocular irAEs are the most frequently encountered and are often underreported. Patients may complain of ocular dryness, foreign body sensation, burning, itching, tearing, or intermittent blurred vision. Clinical findings may include conjunctival injection, superficial punctate keratopathy, blepharitis, or mild conjunctivitis. These symptoms may be mistakenly attributed to environmental factors, aging, or preexisting dry eye disease. However, in the context of ICI therapy, such symptoms may represent early immune-mediated ocular surface inflammation or lacrimal gland dysfunction. Although vision is typically preserved, untreated ocular surface disease can progress, significantly impair QOL and compromise visual function over time.

Moderate ocular toxicity

Moderate presentations often involve intraocular inflammation, most commonly uveitis. Patients may report blurred or decreased vision, photophobia, floaters, eye pain, and redness. Depending on the anatomic location, uveitis may present as anterior uveitis with ciliary flush and anterior chamber cells and flare, intermediate uveitis with vitreous inflammation and floaters, or posterior uveitis with retinal or choroidal involvement. Moderate disease may already be associated with early complications such as cystoid macular edema or elevated intraocular pressure. At this stage, prompt diagnosis and escalation of therapy are essential to prevent progression to more severe disease and long-term visual sequelae.

Severe and vision-threatening toxicity

Severe ocular irAEs are uncommon but carry a high risk of permanent vision loss. Optic neuritis and other neuro-ophthalmic complications may present with acute or subacute unilateral or bilateral vision loss, central or altitudinal visual field defects, impaired color vision, and pain with eye movement. Fundoscopic findings may be subtle or absent early in the disease course, making neuro-ophthalmic assessment and ancillary testing essential. Other severe manifestations include posterior uveitis, panuveitis, retinal vasculitis, and chorioretinitis, which may present with scotomas, metamorphopsia, or sudden visual decline. These entities often require urgent intervention with systemic corticosteroids and, in some cases, additional immunosuppressive therapies, as well as temporary or permanent modification of ICI treatment.[233,473,474]

Early warning signs and red flags

Early warning signs that should prompt immediate ophthalmologic referral include any new-onset visual disturbance, persistent eye pain, photophobia, new floaters, visual field defects, or unexplained decrease in visual acuity in patients receiving ICIs (Table 5). Importantly, ocular irAEs may occur in isolation without concurrent systemic irAEs, and onset can range from days to months after initiation of therapy or even following treatment discontinuation. Clinicians should maintain vigilance throughout the entire course of immunotherapy.

Table 5.

Red flags for immune checkpoint inhibitor–associated ocular toxicity requiring urgent referral

The following symptoms should prompt urgent ophthalmology or neuro-ophthalmology referral, as they may indicate vision-threatening immune-related adverse events (irAEs)
New or progressive vision loss
Acute eye pain, particularly pain with eye movement
Acute/marked photophobia
New floaters, flashes, or scotomas
New visual field defects
New diplopia
Acute ptosis or cranial neuropathy
Additional high-risk features include:
Bilateral ocular involvement
Rapid progression
Concomitant neurologic irAEs
History of autoimmune disease
Clinical action: Suspected moderate or severe ocular irAEs warrant immediate specialty evaluation, consideration of temporary immune checkpoint inhibitor interruption, and close multidisciplinary coordination. Delayed recognition may result in irreversible visual morbidity.

This table underscores key ophthalmic and neuro-ophthalmic symptoms in patients receiving ICIs that should prompt immediate ophthalmology or neuro-ophthalmology evaluation. Early identification of vision-threatening irAEs and timely multidisciplinary intervention are essential to prevent permanent visual loss and to guide decisions regarding continuation or interruption of immunotherapy (Supplemental Table S3).

CTCAE grading (Table 6) is based on severity of symptoms, impact on visual function, and need for intervention. Visual acuity thresholds should be interpreted in conjunction with clinical findings and functional impairment.

Table 6.

Ocular toxicity grading according to CTCAE version 5.0

CTCAE Grade	Severity	Clinical Description	Visual Acuity (BCVA)	Recommended Management
Grade 1	Mild	Asymptomatic or mild ocular symptoms not interfering with activities of daily living	BCVA ≥ 20/20 or no clinically meaningful change from baseline	Observation or supportive care; medical intervention not indicated
Grade 2	Moderate	Symptomatic ocular toxicity with moderate decrease in vision or functional impairment	BCVA ≥ 20/40 or < 3 lines decrease from baseline	Medical intervention indicated; consider holding ICI therapy; initiate local corticosteroids; ICI may be resumed if symptoms improve to Grade ≤ 1
Grade 3	Severe	Marked symptoms with significant visual impairment	BCVA < 20/40 or ≥ 3 lines decrease from baseline (up to 20/200)	Invasive intervention indicated; hold or discontinue ICI therapy; initiate systemic corticosteroids if no improvement within 4–6 wk or earlier if vision-threatening
Grade 4	Sight-threatening/life-threatening	Severe visual loss with risk of permanent blindness	BCVA < 20/200	Urgent intervention required; permanently discontinue ICI therapy; initiate high-dose systemic corticosteroids ± local therapy
Grade 5	Death	Death related to ocular toxicity	Not applicable	Not applicable

BCVA: best corrected visual acuity; CTCAE: Common Terminology Criteria for Adverse Events; ICI: immune checkpoint inhibitor.

Existing guidelines for management of ocular irAES

With the expanding use of ICIs, ocular irAEs are increasingly recognized across malignancies. Although less common than systemic irAEs, ocular toxicities may be vision-threatening and can disrupt effective cancer therapy. Multiple professional organizations, including NCCN, ASCO, ESMO, and American Academy of Ophthalmology (AAO), have issued guidance to support clinicians in diagnosis and management. Despite minor variations, these guidelines are highly concordant (Table 7). Shared principles include early symptom recognition, prompt ophthalmologic evaluation, standardized severity grading using CTCAE criteria, severity-based treatment escalation, and multidisciplinary decision-making to balance visual preservation with oncologic benefit.[75,475]

Table 7.

Shared principles in the management of ICI-related ocular toxicity: Consensus across key guidelines

Guideline	Initial Management	Criteria for ICI Hold or Discontinuation	Steroid Use and Immunosuppressive Strategies	Rechallenge and Long-Term Follow-Up
NCCN	Supportive care for mild symptoms (grade 1) Oral corticosteroids for moderate toxicity (grade 2) Systemic steroids for severe (grade 3) or life-threatening (grade 4) toxicity	Hold ICI therapy for grade 2–3 toxicity until symptoms improve Permanent discontinuation for grade 4 toxicity	Systemic corticosteroids (oral or IV) for moderate to severe cases Steroid-sparing agents (e.g., mycophenolate and infliximab) for refractory cases	Consider rechallenge after resolution to grade ≤ 1 Long-term follow-up for posterior segment involvement or optic neuritis Routine monitoring every 3–6 mo for high-risk patients
ESMO	Mild: Lubricating drops or low-potency topical steroids Moderate: Oral corticosteroids Severe: IV corticosteroids for life-threatening or vision-threatening toxicity	Temporary ICI interruption for grade 2 Permanent discontinuation for grade 4 Discontinue if no improvement within 4–6 wk	Systemic corticosteroids for grade 3–4 toxicity Infliximab, mycophenolate, or cyclosporine as steroid-sparing agents Consider plasmapheresis for refractory cases	Rechallenge in select patients after full resolution to grade 1 Ongoing surveillance for neuro-ophthalmic involvement, especially optic neuritis or uveitis
ASCO	Local therapy (lubricants and topical steroids) for mild symptoms Systemic corticosteroids for moderate and severe cases	Hold or discontinue ICI for grade 3–4 toxicity Discontinue permanently for bilateral or severe optic neuritis, or posterior uveitis	High-dose systemic corticosteroids for severe cases Consider steroid-sparing agents for patients with recurrent or refractory disease	Ongoing monitoring for delayed complications (e.g., optic atrophy and recurrent uveitis) Rechallenge after resolution of grade 3 or lower toxicity on a case-by-case basis
AAO	Artificial tears, corticosteroid eye drops for mild surface symptoms Oral steroids for moderate-to-severe inflammation	Hold or discontinue ICI therapy for grade 3 or 4 toxicity	High-dose IV corticosteroids for grade 3–4 toxicity Infliximab or mycophenolate for persistent symptoms	Regular follow-up for patients with optic neuropathy or uveitis Consider ICI rechallenge after full resolution of symptoms with no relapse

AAO: American Academy of Ophthalmology; ASCO: American Society of Clinical Oncology; ESMO: European Society for Medical Oncology; ICI: immune checkpoint inhibitor; IV: intravenous; NCCN: National Comprehensive Cancer Network.

Our practice recommendations for the management of ICI-related ocular toxicity in managing ICI-related ocular toxicities, we recommend a unified, stepwise approach that emphasizes early recognition, prompt ophthalmologic evaluation, severity-based treatment, and multidisciplinary collaboration. The primary objectives are to preserve vision and ensure the ongoing efficacy of oncologic treatments. Early diagnosis and intervention are crucial to preventing further disability, and we encourage patients to report any visual changes or discomfort early during their treatment. Patients should be educated to recognize and report symptoms such as dry eye, blurred vision, visual disturbances, eye pain, photophobia, double vision, and other visual discomforts, as these may be indicative of ocular toxicity.

For high-risk patients, we recommend a comprehensive baseline ophthalmic evaluation prior to starting ICI therapy. This assessment will help identify any pre-existing ocular conditions and establish a baseline for future comparisons.

We advise regular ophthalmic exams, at least every 1–2 months, especially for patients undergoing prolonged ICI therapy or presenting with early ocular symptoms. This will facilitate early detection of any emerging issues.

Grading and management based on severity (CTCAE)

We follow the CTCAE to grade ocular toxicities and manage them based on severity. Our recommendations are as follows:

• Grade 1 (mild) patients may be asymptomatic or report mild symptoms, such as minor visual disturbances. We recommend conservative management with preservative-free lubricating eye drops (artificial tears) and, if needed, mild TCS. Patients should be monitored regularly (every 1–2 weeks) to ensure no progression of symptoms. Most patients with Grade 1 toxicity can continue ICI therapy with close observation.

• Grade 2 (moderate) patients with anterior uveitis or mild inflammatory eye disease may occur. Management include initiation on TCS, cycloplegic agents and oral corticosteroids (e.g., prednisone 0.5–1 mg/kg/d) if needed. Close monitoring (weekly to biweekly) is essential to assess treatment response and prevent complications like macular edema or optic nerve involvement. ICI therapy may need to be temporarily suspended until symptoms improve. Once symptoms are reduced to Grade 1, ICI therapy may be resumed.

• Grade 3 (severe) patients with posterior uveitis, panuveitis, or severe ocular inflammation. Referral to an ophthalmologist or uveitis specialist is strongly recommended. We recommend starting high-dose systemic corticosteroids (e.g., prednisone 1–2 mg/kg/d) or intravenous methylprednisolone, followed by an oral steroid taper (10 mg every 7–10 d) depending on clinical improvement. For long-term management, steroid-sparing agents (e.g., MMF or cyclosporine) may be considered. Intraocular steroids could be used if there is minimal response to systemic therapy. Frequent ophthalmic evaluations (weekly), including imaging techniques such as OCT and FA, are necessary to monitor for posterior segment involvement. We recommend withholding ICI therapy until symptom resolution, with rechallenge only considered after a thorough multidisciplinary consultation.

• Grade 4 (life-threatening) patients with optic neuritis or optic neuropathy. Management: include immediate referral to an ophthalmologist or neuro-ophthalmologist. We recommend starting high-dose intravenous corticosteroids (e.g., methylprednisolone 1 g/d for 3–5 days), followed by an oral steroid taper (10 mg every 7–10 days). If the condition is refractory, advanced therapies such as PLEX or IVIG may be considered. Continuous monitoring by ophthalmology, neuro-ophthalmology, and oncology specialists is crucial. Because of the severity of the toxicity, permanent discontinuation of ICI therapy is often necessary.

At MD Anderson, management of ICI–related ocular irAEs are anchored in early recognition and proactive multidisciplinary collaboration among oncology, ophthalmology, and neuro-ophthalmology, with immunology or rheumatology involvement for complex or refractory cases. Early cross-specialty communication is prioritized to ensure accurate diagnosis, timely initiation of therapy, and balanced decision-making that preserves visual function while maintaining oncologic efficacy.

Rechallenge with ICI therapy is considered only after complete resolution of ocular toxicity and individualized multidisciplinary risk–benefit discussion. Rechallenge may be reasonable after grade 1–2 events, particularly ocular surface disease or mild anterior uveitis, once symptoms have improved to grade ≤ 1 with close ophthalmic monitoring. Rechallenge after grade 3 toxicity is reserved for select patients with substantial anticipated oncologic benefit, while rechallenge is generally avoided after grade 4 toxicity, especially optic neuropathy or posterior segment disease.

Long-term follow-up is emphasized for patients with posterior uveitis, neuro-ophthalmic involvement, recurrent disease, or bilateral manifestations. Surveillance includes periodic ophthalmologic evaluation with appropriate ancillary testing, cautious corticosteroid tapering, and monitoring for delayed complications such as optic atrophy or persistent visual field defects.

Steroid-refractory or recurrent ocular irAEs require escalation beyond corticosteroids to steroid sparing immunosuppressive agents or in selected cases, biologic therapies. This should be guided by close collaboration between oncology and ophthalmic subspecialists. Patient education is integral to care, with proactive counseling on early warning symptoms, prompt reporting pathways, and adherence to prescribed topical or systemic corticosteroid regimens (Supplemental Tables S4 and S5).[74]

Challenges

Despite increasing awareness, several challenges persist: Unpredictable Onset: Ocular irAEs can occur at any time during or even after ICI therapy. Heterogeneity of Presentation: Toxicities range from mild dry eye to severe optic neuropathy, complicating standardized management. Limited Consensus: Global treatment guidelines are evolving, and local practices vary.

Practice gaps

• There is a lack of standardized protocols across institutions for early detection and grading, with limited data on long-term outcomes of ocular irAEs.

• In addition, there is insufficient guidance on resuming ICIs after ocular toxicity resolution.

• Under-recognition of subtle neuro-ophthalmic manifestations by non-ophthalmologists.

• Limited prospective studies comparing management strategies or immunosuppressive regimens.

Conclusion

Ocular toxicity from ICIs is a significant, potentially sight-threatening complication of cancer immunotherapy. Early recognition, accurate grading, and prompt intervention are essential to preserve vision and maintain QOL. Multidisciplinary collaboration, patient education, and ongoing research will continue to refine treatment strategies, enabling more personalized and effective care for patients receiving ICI therapy.

Introduction and clinical spectrum of oral irAEs

With the unprecedented deployment of novel ICIs, bispecific antibodies, and antibody-drug conjugates (ADCs), well-documented and previously unknown oral irAEs present a unique challenge for patient adherence to treatment protocol, nutritional uptake during cancer therapy, communication hardships, emotional stress associated with decreased oral function, delay in ICI dosing, termination of immunotherapy, and increased emergency visits and re-admissions. Oral irAEs are generally under-reported related in part to insufficient monitoring and the lack of clear guidelines to manage oral irAEs. Improvements in early detection, precise diagnosis, and early intervention of oral irAEs promise to make a transformative impact on a significant population of patient with cancer receiving ICIs. This review will summarize the known oral irAE (Tables 8–10) and also define two emerging oral diseases associated with ICI therapy, including immune-related gingival enlargement (IRGENT) and immune-related osteonecrosis of the jaw (IRON Jaw).

Table 8.

Oral mucosal lesions associated with immune checkpoint inhibitors and targeted anticancer agents

Therapy Type	Tumor Type	Lesion Type (%)	Source
Oral mucositis
Pembrolizumab	BC, GC, NSCLC, UC	Oral mucositis (33.3, stomatitis)	NCT04383938
Pembrolizumab	Endometrial neoplasms	Oral mucositis (12.3, stomatitis)	NCT04865289
Pembrolizumab	Melanoma	Oral mucositis (9.0, stomatitis)	NCT03820986
Pembrolizumab	NSCLC	Oral mucositis (10.2, stomatitis), Oral mucositis (5.3, stomatitis)	NCT04676412 NCT02007070
Pembrolizumab (+ 4-1BB agonist)	Advanced cancer	Oral mucositis (40.0, stomatitis)	NCT02179918
Pembrolizumab (+ anti-CD73)	Advanced cancer	Oral mucositis (16.7, stomatitis)	NCT04148937
Pembrolizumab (+ IL-2 inducer)	HNSCC	Oral mucositis (12.5, stomatitis)	NCT04144517
Nivolumab	Neuroendocrine tumor	Oral mucositis (15.8)	NCT04197310
Atezolizumab	cSCC	Oral mucositis (6.8, mucositis)	NCT04710498
Ipilimumab	RCC	Oral mucositis (1.0, stomatitis)	NCT04513522
Nivolumab	Melanoma, NSCLC	Oral lichenoid reaction with ulceration	Obara[489] 2018
Nivolumab	OSCC	Oral lichenoid reaction	Shazib[521] 2020; Fässler[487] 2020
Nivolumab	Lung adenocarcinoma	Oral lichenoid reaction with ulceration	Enomoto[486] 2019; Bhattacharyya[480] 2020; Shazib[521] 2020
Nivolumab	MM, RCC, GB, Lymphoma, LUSCC	Oral lichenoid reaction	Sibaud[19] 2018
Nivolumab (+ kinase inhibitors)	Melanoma	Oral mucositis (16.1)	NCT04493203
Everolimus, Temsirolimus (mTOR inhibitor)	RCC, PDAC, BC, Lymphoma	Oral mucositis (stomatitis)	Vigarios[528] 2017 Villa[498] 2024
Afatinib, Lapatinib, Gefitib, Erlotinib, Dacomitib, Cetuximab (HER inhibitors)	NSCLC, BC, HNSCC, LUSCC	Oral mucositis (stomatitis)	Vigarios[528] 2017 Villa [498] 2024
Sorafenib, Axitinib, Dovitinib, Pazopanib, Sunitinib, Lenvatinib, Cabozatinib, Bevacizumab (angiogenesis inhibitors)	RCC, HCC, PDAC, thyroid cancer, CML	Oral mucositis (Stomatitis)	Vigarios[528] 2017; Gerber[529] 2021
Imatinib (BCR-ABL inhibitors)	CML, Glioma	Oral mucositis (stomatitis)	Vigarios[528] 2017; Villa[498] 2024
Crizotinib (ALK inhibitors)	LUSCC, NSCLC	Oral mucositis (stomatitis)	Vigarios[528] 2017; Villa[498] 2024
Oral lichenoid reactions: mucous membrane pemphigoid and bullous pemphigoid
Pembrolizumab (+ anti-CD20)	Lymphoma, Bladder cancer	Oral lichenoid reaction	Bhattacharyya[480] 2020
Pembrolizumab	BC	Oral lichenoid reaction	Sibaud[19] 2018; Bhattacharyya[480] 2020
Pembrolizumab (+ RTK)	RCC	Oral lichenoid reaction	Wellen[485] 2025
Atezolizumab	RCC	Oral lichenoid reaction	Sibaud[19] 2018
Pembrolizumab	Melanoma, MCC	Mucous membrane pemphigoid, Oral lichenoid reaction	Fässler[487] 2020; Haug[530] 2018
Pembrolizumab	Melanoma, MCC	Mucous membrane pemphigoid	Zumelzu[483] 2018; Haug[530] 2018; Fässler[487] 2020
Nivolumab (+ anti-CD20)	Metastatic adenocarcinoma	Bullous pemphigoid	Sowerby[531] 2017; Sadik[532] 2020
Rituximab (anti-CD20)	Lymphoma	Oral lichenoid reaction	Kuten-Shorrer[488] 2014

ALK: anaplastic lymphoma kinase; BC: breast cancer; BCR-ABL: breakpoint cluster region–Abelson; CD: cluster of differentiation; CML: chronic myeloid leukemia; cSCC: cutaneous squamous cell carcinoma; GB: glioblastoma multiforme; GC: gastric cancer; HCC: hepatocellular carcinoma; HER: human epidermal growth factor receptor; HNSCC: head and neck squamous cell carcinoma; IL-2: interleukin-2; LUSCC: lung squamous cell carcinoma; MCC: Merkel cell carcinoma; MM: multiple myeloma; mTOR: mechanistic target of rapamycin; NSCLC: non–small cell lung cancer; OSCC: oral cavity squamous cell carcinoma; PDAC: pancreatic ductal adenocarcinoma; RCC: renal cell carcinoma; RTK: receptor tyrosine kinase; UC: urothelial carcinoma.

Table 10.

Temporomandibular joint disorder and osteonecrosis of the jaw (MRONJ) associated with ICIs and targeted anticancer agents

Therapy Type	Tumor Type	Lesion Type (%)	Source
Temporomandibular Joint Disorder
Pazopanib, Sunitinib (RTK inhibitor)	RCC	Trismus	Benhima[492] 2024; Iyer[493] 2017
Pembrolizumab	OSCC	Trismus	Tang[533] 2022
Osteonecrosis of the jaw
Pembrolizumab	Advanced tumor	Osteonecrosis of jaw (2.9)	NCT04082572
Pembrolizumab	NSCLC	Osteonecrosis of jaw (0.3)	NCT03829332
Pembrolizumab	HNSCC	Osteonecrosis of jaw with spontaneous fracture	Subramanian[507] 2024
Pembrolizumab (+PARP inhibitors)	Prostate Cancer	Osteonecrosis of jaw (0.2)	NCT03834519
Pazopanib (RTK inhibitor)	RCC	Osteonecrosis of jaw	Papadopoulou[509] 2022
Pembrolizumab	Malignant melanoma	Osteonecrosis of jaw with spontaneous fracture	Pundole[515] 2020

BC: breast cancer; CML: chronic myeloid leukemia; CRC: colorectal cancer; cSCC: cutaneous squamous cell carcinoma; GC: gastric cancer; HCC: hepatocellular carcinoma; HNSCC: head and neck squamous cell carcinoma; LASN: locally advanced solid neoplasm; LUSCC: lung squamous cell carcinoma; MCC: Merkel cell carcinoma; MRONJ: osteonecrosis of the jaw; NSCLC: non–small cell lung cancer; PC: prostate cancer; PDAC: pancreatic ductal adenocarcinoma; RCC: renal cell carcinoma; RTK: receptor tyrosine kinase; TMJ: temporomandibular joint disorder; UC: urothelial carcinoma.

The development of irAEs is mechanistically distinct from toxicities associated with conventional chemotherapy or radiation therapy.[476] irAEs may involve any organ system, including the oral and orofacial regions, and typically manifest during treatment or within the first year of therapy. Approximately 60% of oral irAEs occur within the first 3 months of ICI initiation.[477–479] Oral irAEs encompass a broad spectrum of clinical manifestations, including mucosal lesions,[477,480–489] salivary gland dysfunction,[301,477,490,491] and temporomandibular joint (TMJ) arthritis and trismus.[492,493] To improve awareness and diagnosis of oral irAE, we provide an up-to-date overview of the known and emerging oral manifestations associated with ICIs.

Oral mucosal lesions

Oral mucosal irAEs represent a heterogeneous group of manifestations associated with ICI therapy, mimicking lichenoid drug reaction, lichen planus, benign mucous membrane pemphigoid, paraneoplastic pemphigus, aphthous ulcers, or generalized nonspecific desquamative mucositis.[477,480–489] (Table 8). Clinically, these conditions present with erythema, ulceration, or blistering, often accompanied by pain, burning sensations, dysgeusia, and dysphagia, which collectively significantly impair oral intake and the overall QOL.[477,483] Notably, up to 20% of patients receiving ICIs develop dermatologic irAEs, often presenting as nonspecific rashes or erosive mucositis.[494] Diagnosis is primarily clinical, supported by histopathologic examination when necessary to exclude infectious, autoimmune, or neoplastic processes.[480,485–489] The proper management of these lesions first requires refined diagnosis. Oral mucosal lesions can represent a complex set of etiologies, including epithelial barrier impairment, dry mouth, inflammation, anemia, and oral infections.

In addition to the aforementioned forms of mucositis, ICI-related oral mucosal changes can also present as a distinct form of gingival enlargement. Previously, generalized gingival enlargement was most commonly associated with the use of immunosuppressants, calcium channel blockers, and anticonvulsants.[495] Emerging evidence shows that treatment-delaying, ICI-induced gingival enlargement can present as gingival swelling similar to calcium channel blockers-associated gingival hyperplasia, but unlike gingival leukemic infiltrates, which often cause discoloration and surface ulceration. The clinical features of IRGENT are also distinct from localized juvenile spongiotic gingival hyperplasia, which usually presents as a solitary lesion with spongiotic epithelial changes. IRGENT tends to bleed with variable levels of pain. In such cases, a custom drug tray could offer an effective topical treatment that does not require high doses of systemic immunosuppressive agents or IVIG. In patients who have dry mouth or confounding oral infections, management of those conditions will help reduce the severity of ICI-induced mucositis. To reduce the risk of confounding oral infections, meticulous oral hygiene, avoidance of mucosal irritants, regular monitoring, and close multidisciplinary collaboration among oncologists and oral medicine specialists are crucial for the timely recognition and management of oral mucosal irAEs.[496–498] To the best of our knowledge, this report includes one of the first cases of the newly defined IRGENT associated with ICI therapy. However, it should also be noted that ICI-related gingival enlargement has been reported previously, although that case involved chemotherapy followed by ICI treatment, with an adriamycin–bleomycin–vinblastine–dacarbazine (ABVD) regimen that is not typically linked to gingival enlargement.[499]

Case #1 (IRGENT): A 50-year-old man with metastatic clear-cell RCC involving the lungs and liver developed progressive gingival enlargement and spontaneous bleeding, causing discomfort, impaired speech, and difficulty with oral intake. Symptoms appeared following initiation of nivolumab (480 mg) and cabozantinib (40 mg) and worsened more than 9 months to a point that cancer treatments must be on hold. Examination revealed diffuse lip swelling, gingival enlargement, and mucosal fissuring, more pronounced in the left maxillary and mandibular arches. Biopsy showed subepithelial blister formation with dense mixed inflammatory infiltrates, while immunofluorescence examination showed a nonspecific staining pattern. This is the first case, to our knowledge, to report a distinct form of gingival enlargement, hereby coined as IRGENT. Management included TCS perioral desonide 0.05% ointment, dexamethasone oral rinse (0.5 mg/5 mL), and clobetasol propionate 0.05% gel twice daily—and an immediate discontinuation of anti-PD-1 therapy after multidisciplinary review. To enhance mucosal healing and drug-tissue contact, a custom mucosal medication carrier was fabricated. Digital models of dental arches were created, gingival surfaces were reduced by 1 mm using Meshmixer software, and the models were 3D printed (Formlabs, Boston, MA, USA) to produce individualized carriers ensuring gentle pressure and sustained topical delivery. Follow-up at 2, 4, 8, and 12 weeks showed progressive improvement, with IRGENT completely resolved by 8 weeks (Fig. 2).

Figure 2.

Clinical appearance and treatment response of immune-related gingival enlargement. (A) Frontal view of gingival tissues with blistering appearance. (B) Lateral view of gingival tissues with blistering. (C) Palatal/lingual surface with blistered appearance. (D) Maxillary and mandibular mucosal carrier. (E) Two weeks after initiation of topical steroid and dexamethasone mouth rinse. (F) Near-complete resolution 6 weeks after continued conservative therapy and a nivolumab drug holiday. (G) Occlusal view at 6 weeks following conservative therapy and nivolumab drug holiday.

Salivary gland dysfunctions and associated secondary infections

Salivary gland dysfunction is one of the most common oral irAEs observed during treatment with ICIs. It is commonly associated with xerostomia, dysgeusia, and dysphagia (Table 9).[281,477,490,491,496] These symptoms can markedly affect oral comfort, mastication, swallowing, and speech, and predispose patients with this condition to opportunistic infections, particularly oral candidiasis.[497] The underlying mechanism involves immune-mediated injury to the salivary glands, characterized by lymphocytic infiltration and destruction of acinar and ductal epithelial cells, resembling autoimmune inflammation of the salivary glands.[500] Histopathological examination of minor salivary glands typically reveals periductal infiltration of mononuclear immune cells composed predominantly of helper CD4⁺ T cells and B cells.[304] However, in ICI-treated patients, while aggregated CD20⁺ B cells and CD3⁺ T cells are present, T cells predominate, with CD8⁺ T cells outnumbering CD4⁺ subsets.[501] These findings suggest a cytotoxic T cell–mediated mechanism, distinct from the classic B cell–driven germinal center pathology characteristic of autoimmune Sjögren’s syndrome.[304,500] Clinically, Sjogren’s syndrome has a strong female predilection. In contrast, ICI-induced salivary gland dysfunction or Sjogren-like syndrome appears to be more common in male with a M:F ratio of ∼1.5:1.[502]

Table 9.

Salivary gland dysfunctions associated with immune checkpoint inhibitors and targeted anticancer agents

Therapy Type	Tumor Type	Lesion Type (%)	Source
Dry mouth/xerostomia
Pembrolizumab	Advanced tumor	Dry mouth (2.9)	NCT04082572
Pembrolizumab	Lymphoma	Dry mouth (14.3)	NCT04317066
Pembrolizumab	Melanoma	Dry mouth (10.8)	NCT03820986
Pembrolizumab	Endometrial neoplasms	Dry mouth (8.2)	NCT04865289
Pembrolizumab	Melanoma	Dry mouth (5.9)	NCT02180061
Pembrolizumab	MCC	Dry mouth (9.0)	NCT03783078
Pembrolizumab	NSCLC	Dry mouth (5.3)	NCT03625323
Pembrolizumab (+ 4-1BB agonist)	Advanced cancer	Dry mouth (20.0)	NCT02179918
Pembrolizumab (+ anti-CD20)	Sarcoma	Dry mouth (23.3)	NCT03899805
Pembrolizumab (+ anti-LAG-3)	NSCLC	Dry Mouth (5.3)	NCT03625323
Pembrolizumab (+ GSK2857916)	Endometrial neoplasms	Dry mouth (9.1)	NCT03884101
Pembrolizumab (+HDAC inhibitors)	Melanoma	Dry mouth (18.2)	NCT03765229
Pembrolizumab (+ NOX inhibitors)	HNSCC	Dry mouth (14.3)	NCT05323656
Pembrolizumab (+oncolytic virus)	Glioblastoma	Dry mouth (14.6)	NCT04479241
Nivolumab	NSCLC, Melanoma, CRC, RCC, PC	Dry mouth (16.7)	NCT00441337
Ipilimumab (+ Nivolumab)	Melanoma	Dry mouth[9]	NCT05116202
Ipilimumab	Melanoma	Dry mouth[5]	NCT01990859 NCT00162123
Atezolizumab	BC	Dry mouth (6.5)	NCT02302807
Atezolizumab (+ anti-TIGIT)	HNSCC	Dry mouth (16.6)	NCT05459129
Secondary oral infections and xerostomia-related symptoms
Pembrolizumab (+ anti-LAG-3)	NSCLC	Oral candidiasis (3.5)	NCT03625323
Pembrolizumab (+ NOX inhibitors)	HNSCC	Oral candidiasis (10.7)	NCT05323656
Ipilimumab	Melanoma	Oral candidiasis (7.9)	NCT00920907
Ipilimumab (+ tumor vaccine)	Melanoma	Oral candidiasis (6.5)	NCT04382664
Pembrolizumab (+ kinase inhibitor)	HNSCC	Dental carries (10.0)	NCT04555837
Pembrolizumab	Endometrial Neoplasms	Periodontal disease (8.2)	NCT04865289
Pembrolizumab	HNSCC	Dysphagia (2.8)	NCT04199104
Pembrolizumab	MCC	Dysphagia (1.8)	NCT03783078
Pembrolizumab (+ anti-LAG-3)	NSCLC	Dysphagia (6.1)	NCT03625323
Pembrolizumab (+ anti-TIGIT)	Melanoma	Dysphagia (0.1)	NCT05665595
Pembrolizumab (+ kinase inhibitor)	HNSCC	Dysphagia (20.0)	NCT04555837
Pembrolizumab (+ IL-2 inducer)	HNSCC	Dysphagia (12.5)	NCT04144517
Pembrolizumab (+ PARP inhibitors)	Cholangiocarcinoma	Dysphagia (7.1)	NCT04306367
Nivolumab (+ kinase inhibitors)	Solid Tumors	Dysphagia (13.3)	NCT04704154
Atezolizumab	NSCLC	Dysphagia (1.4)	NCT01903993
Nivolumab	Neuroendocrine tumor	Dysgeusia (36.8)	NCT04197310
Pembrolizumab (+HDAC inhibitors)	Melanoma	Dysgeusia (18.2)	NCT03765229
Pembrolizumab (+ GSK2857916)	Endometrial neoplasms	Dysgeusia (8.1)	NCT03884101
Pembrolizumab (+PARP inhibitors)	PC	Dysgeusia (29.5)	NCT03834519
Vismodegib (Hedgehog-pathway inhibitors)	BCC	Dysgeusia	Vigarios[528] 2017 Villa[498] 2024
Crizotinib (ALK inhibitors)	LUSCC, NSCLC	Dysgeusia	Vigarios[528] 2017 Vil la[498] 2024

ALK: anaplastic lymphoma kinase; BC: breast cancer; BCC: basal cell carcinoma; CD: cluster of differentiation; CRC: colorectal cancer; HDAC: histone deacetylase; HNSCC: head and neck squamous cell carcinoma; IL-2: interleukin-2; LAG-3: lymphocyte-activation gene 3; LUSCC: lung squamous cell carcinoma; MCC: Merkel cell carcinoma; NOX: NADPH (nicotinamide adenine dinucleotide phosphate) oxidase; NSCLC: non–small cell lung cancer; PARP: poly(ADP-ribose) polymerase; PC: prostate cancer; RCC: renal cell carcinoma; TIGIT: T cell immunoreceptor with immunoglobulin and immunoreceptor tyrosine-based inhibitory motif domains.

Temporomandibular joint (TMJ) disorders

TMJ disorders are uncommonly associated with ICIs and targeted therapies but might be under-reported in this patient population due to insufficient monitoring of arthritis of the TMJ. Although two cases of trismus have been reported following targeted anticancer therapy (tyrosine kinase inhibitor),[492,493] no confirmed cases have yet been linked to ICIs (Table 10). Clinically, drug-induced trismus is characterized by jaw pain, restricted mouth opening, and impaired oral function.[492,493] These manifestations can severely affect nutrition, oral hygiene, and dental care.[492,493] The precise mechanism underlying receptor tyrosine kinase (RTK) inhibitor-associated trismus remains uncertain. It has been hypothesized that inhibition of vascular endothelial growth factor receptor (VEGFR) signaling leads to reduced vascular permeability and angiogenesis within the TMJ. This impaired vascularization may promote muscle atrophy, fibrosis, and apoptosis-mediated tissue remodeling.[492,493] Furthermore, inhibition of VEGFR signaling may suppress osteoclast differentiation and bone resorption, thereby contributing to joint dysfunction.[503,504] However, alternative mechanisms cannot be excluded. Diagnosis requires careful exclusion of mechanical, infectious, and metastatic causes through comprehensive clinical assessment and imaging studies, such as MRI or computed tomography (CT).

Immune-related osteonecrosis of the jaw (IRON jaw)

Bone-modifying agents-related osteonecrosis of the jaw, also referred to as medication-related osteonecrosis of the jaw (MRONJ), is a severe adverse event associated with targeted cancer therapies, predominantly involving the mandible (Table 10). Clinically, MRONJ presents with pain, mucosal ulceration, exposed necrotic bone, swelling, and, in advanced stages, spontaneous pathological fractures.[505–507] Although its pathophysiology is not fully resolved, MRONJ is generally linked to impaired bone remodeling within an inflammatory milieu.[505–507] Immune modulation appears to contribute, as PD-1 blockade has been shown to enhance osteoclast activity and promote T cell–dependent bone loss,[508] while anti-angiogenic agents further exacerbate osteonecrosis by impairing vascularization and bone healing.[505,509] Diagnosis is based on clinical and radiographic evaluation, and management ranges from conservative therapy to surgical intervention depending on disease severity. Furthermore, management depends on disease severity and ranges from conservative measures, pentoxifylline plus tocopherol-based protocols, antimicrobial mouth rinses, antibiotics, and pain control to surgical intervention, including sequestrectomy or segmental resection in advanced cases.[505,506,510]

In this context, IRON jaw has recently emerged as a clinically significant oral irAE associated with ICI therapy across multiple malignancies.[511–515] The clinical impact of IRON jaw is considerable, as it often necessitates interruption or modification of immunotherapy, potentially compromising treatment outcomes. A higher incidence in male patients has been reported, though the overall reported case number remains low.[511] Current evidence supports a multifactorial pathogenesis in which local and systemic factors intersect. Pre-existing oral disease and bacterial infection likely contribute to disease initiation, as indicated by histopathologic findings of necrotic bone containing bacterial colonies.[513] In addition, concomitant use of ICIs with high-risk agents, such as RTK-inhibitors or monoclonal antibodies, may further increase susceptibility. Collectively, these findings support a model in which ICI-driven immune activation disrupts bone remodeling while amplifying local inflammatory responses, predisposing the jawbone to necrosis, particularly in the presence of microbial and pharmacologic stressors.[516] As ICI use continues to expand, systematic documentation of IRON jaw cases will be essential to improve mechanistic understanding and guide preventive and therapeutic strategies. Accordingly, minimizing oral trauma and closely monitoring patients during ICI therapy remain critical. Here, we expand upon a previously reported case of IRON jaw (case #2),[515] together with supplementary orodental details and imaging findings that highlight the severity and clinical progression of this irAE.

Case #2 (IRON Jaw): A 73-year-old man (age at the initial melanoma diagnosis) presented with a T3b malignant melanoma of the left forearm. After wide local excision and sentinel node biopsy of the left axilla, pathology revealed no residual melanoma, but the sentinel node contained tumors. The patient went on to receive a left axillary lymphadenectomy. Imaging studies 2 months later showed new left axillary lymphadenopathy with a maximum diameter of 8.0 cm, subcentimeter findings in the right lung mediastinal lymph nodes, and in the liver. The patient received single-agent nivolumab (240 mg every 2 weeks). The patient experienced jaw pain 1 week after the first dose of nivolumab. After the symptoms became worse and persistent, he was treated with antibiotics for a possible dental infection, which did not improve the symptom control. Thus, the fourth cycle of nivolumab was delayed. After the symptoms were alleviated, the patient resumed treatment with nivolumab. The detailed treatment history has been previously reported.[515] The patient reported pain at 10 of 10 and received opioids for symptom management. About a month after the infusion of the last nivolumab for cycle 8 treatment and 10 days after initiation of 40 mg of prednisone, the patient noticed a loud “pop” while eating. Then, he noticed his maxillary and mandibular teeth were not aligned. The patient was referred to dental oncology for an evaluation. An orthopantogram and cone-beam CT study were performed. Review of the imaging study showed extensive osteonecrosis of the mandible, involving the bilateral inferior border, mandibular body, angle of the mandible, and bilateral ramus. A vertical fracture was present through the left coronoid process (Figs. 3A–E). Given the extensive necrosis and structural compromise, the patient underwent a total mandibulectomy (coronoid to coronoid) with immediate fibula free-flap reconstruction as well as tracheotomy (Figs. 3F, G).

Figure 3.

Clinical presentation and surgical management of immune-related osteonecrosis of the jaw. (A–E) Cone beam computed tomography and orthopantomogram images of the head and neck regions demonstrating jaw involvement. (F) Total mandibulectomy (coronoid to coronoid) with immediate fibula free-flap reconstruction. (G) The postoperative orthopantomogram demonstrating reconstruction of the mandible with a fibula free flap.

Integrating the previously reported cases, all five patients with IRON Jaw did not have a history of bone-modifying agent treatment. Four of them received anti-PD1 therapy, including nivolumab and pembrolizumab, and the other patient received anti-CTLA-4 therapy. Unlike bisphosphonate-related osteonecrosis of the jaw, also referred to as MRONJ, ICI-related IRON jaw can present as spontaneous osteonecrosis in the absence of clear local dental triggers. This distinct presentation suggests that IRON jaw may represent a mechanistically different entity rather than a variant of MRONJ, although larger cohort studies are needed to confirm this distinction.

Existing guidelines for management of oral irAEs

The proper management of oral mucosal lesions first requires refined diagnosis. Oral mucosal lesions can represent a complex set of etiologies, including epithelial barrier impairment, dry mouth, inflammation, anemia, and oral infections. Management strategies are guided by etiology to avoid excessive use of corticosteroids, which remain the mainstay of the first-line management of immune-related oral mucositis. The strategy should focus on control of topical inflammation, elimination of infection or mechanical injury, alleviation of dry mouth, correction of anemia, if possible, and restoration of barrier function.

The application of TCS can provide meaningful symptomatic relief and promote mucosal healing in a subset of patients (Table 11).[496–498] In patients who do not respond well to TCS, topical application of calcineurin inhibitors, such as tacrolimus ointment, can potentially provide symptom relief. In persistent and refractory cases, intralesional injection of triamcinolone may lead to symptom relief, sparing high systemic doses.[517] Intralesional injection of platelet-rich plasma has been tested in some rheumatic diseases and autoimmune oral conditions, such as oral lichen planus, and shows promise in rapid symptom relief.[518–520] The exact mechanism of this treatment remains insufficiently characterized, and its potential application for ICI-induced oral mucositis and side effects require further evaluation. In cases of severe or refractory mucosal disease, it is necessary to examine systemic conditions, such as previously unknown cancers, cancers nonresponsive to treatments, severe anemia, systemic autoimmune reactions, other concurrent chemotherapy or targeted therapy medications with known side effects of oral mucositis, and oral infections, before administering high-potency systemic corticosteroids or other adjunctive immunosuppressive agents. Systemic therapies such as IVIG and systemic corticosteroids may be needed in severe cases with spontaneous bleeding and extensive desquamative mucositis. These therapies can also arguably mitigate the efficacy of ICIs. Thus, early recognition of oral irAEs is essential for successful adherence to treatment protocols.

Table 11.

Management strategies for oral immune-related adverse events (irAEs) associated with anticancer therapy

Category	Therapeutic Approach/Agent (Formulation)	Clinical Objective	Administration and Dosage
Oral mucosal irAEs	Dexamethasone (0.01–0.05%) or Budesonide (0.03–0.1%) mouth rinse	Attenuate mucosal inflammation and promote epithelial recovery	Rinse with 5 mL for 5 min, 1–3 times/d
	Triamcinolone (10–40 mg/mL, injectable)	Manage corticosteroid-responsive or recalcitrant lesions	Intralesional injection at multiple affected sites
	Lidocaine (2% solution), Diphenhydramine (0.15–0.25% solution), Morphine (0.2% solution)	Provide local analgesia and symptomatic relief	Rinse with 5 mL for 5 min, up to 3/d
	Lidocaine (2% gel), Benzocaine (20% gel)	Deliver localized surface anesthesia	Apply thin layer to affected mucosal areas as needed
Salivary gland dysfunction (xerostomia)	Sialagogues (for residual gland function): Pilocarpine (5–10 mg tablet, 3–4/d; 0.01–4% rinse, 2–3/d); Cevimeline (30 mg capsule, 3/d)	Enhance salivary secretion	Oral or topical administration as indicated
	Sialolytics (for viscous or ropey saliva): N-acetylcysteine (200 mg rinse, 2–3/d or 600 mg tablet, 1–2/d); Guaifenesin (200–400 mg, up to 6/d)	Decrease salivary viscosity and improve flow	Oral or topical administration per clinical need
	Saliva substitutes, mucosal lubricants, and hydration devices	Maintain mucosal moisture and comfort	Use as supportive therapy per clinical guidance
	Gustatory and mechanical stimulation or electrostimulation devices	Stimulate residual salivary gland activity	Per specialist recommendation
Dysgeusia	Zinc supplementation, nutritional optimization, GABA analogs, benzodiazepines	Improve taste perception and oral intake	Administer per clinical guidance
Dysphagia	Systemic analgesics, dietary modification (soft diet), speech/swallow therapy	Maintain oral intake and mitigate discomfort	Individualized management per symptom severity
Gingival enlargement	Rigorous oral hygiene, isotonic saline or sodium bicarbonate rinses Dexamethasone (0.5 mg/mL), clobetasol propionate (0.05%)	Reduce inflammation and prevent secondary infection	Rinse as supportive care per clinical protocol
Lichenoid lesions	Dexamethasone (0.5 mg/mL), Fluocinonide (0.05%), Tacrolimus (0.1%, for lip involvement); long-term surveillance via biopsy	Control inflammation and prevent malignant transformation	Topical, intralesional, and/or systemic as indicated

GABA: gamma-aminobutyric acid; irAEs: immune-related adverse events.

Management of salivary gland dysfunction focuses primarily on symptom relief and oral rehabilitation. Supportive strategies include frequent hydration, saliva substitutes, sugar-free lozenges, and, in selected cases, sialagogues such as pilocarpine or cevimeline to enhance salivary flow (Table 11).[496–498] Patients should maintain meticulous oral hygiene and nutritional counseling, emphasizing soft and moist foods, can improve swallowing and comfort.[497] A substantial subset of patients with xerostomia present with confounding oral infections, which should be treated sufficiently to prevent further exacerbation of the symptoms. Early detection of ICI-induced salivary gland dysfunction is arguably the most effective strategy to prevent severe comorbidities. Xerostomia significantly elevates the risk of developing rampant dental decay in a short period of time, complicating cancer treatments. Thus, a close collaboration with dental oncologists is required for patients going through ICIs. For patients who also develop lacrimal gland symptoms, an early intervention and consultation with ophthalmologists is essential to reduce comorbidities. Based on the reported literature, xerostomia was twice as common as xerophthalmia.[502] Please also refer to the ophthalmology section of this manuscript for further details.

Management of TMJ-related disorder focuses on symptom relief and functional recovery, including temporary discontinuation of the causative agent and multidisciplinary care involving oral medicine, oncology, and physical therapy specialists.[492,493] A broad description of the management of oral irAEs is shown in Table 11.

Updates on management options

Oral irAEs that are not adequately managed may lead to premature discontinuation of ICIs,[478,521] but can persist for an extended period of time, even months after the discontinuation of ICIs. The mechanism of protracted oral irAE remains to be characterized but is likely multifold. First, the clearance of ICIs, such as nivolumab, is similar to endogenous immunoglobulins with an average half-life of ∼25 days,[522] although receptor occupancy and immune activation may persist well beyond the dosing schedule.[523] Sustained activation at the oral mucosal barrier and in the salivary glands may therefore contribute to chronic toxicity. In addition, ICIs may expand self-reactive T cell populations that persist as tissue-resident cells before being fully cleared. Because salivary gland acinar cells have limited regenerative capacity, inflammation-associated injury may also extend beyond the treatment window.

These observations support the importance of integrating oral care before ICI initiation. Early involvement of oral medicine can facilitate timely recognition of serious oral irAEs, improve adherence to cancer therapy, and reduce avoidable treatment discontinuation. Oral irAE-related interruption of ICI therapy does not necessarily preclude later rechallenge after symptoms are controlled. For example, in the IRGENT case described above, the patient resumed ICI therapy with concurrent topical treatment of gingival lesions. However, rechallenge should be approached with caution in severe or irreversible oral irAEs, such as IRON jaw. Decisions should be guided by symptom duration, severity, response to treatment, and the overall oncologic benefit of ICI therapy.

Practice gaps

Despite increasing recognition of oral irAEs, important gaps remain in their classification and management. Under the current CTCAE version 6.0 system, severe oral mucositis is broadly grouped as grade 3 when it interferes with oral intake, which limits its usefulness in guiding nuanced decisions about ICI interruption or rechallenge. As a result, many patients referred to oral medicine for mucosal complications are placed in the same category, even when their underlying etiologies and clinical courses differ substantially.

This limitation is particularly problematic because oral irAEs are multifactorial. Their development can be influenced by immune-mediated mucosal injury, chemotherapy, targeted therapy, radiotherapy, xerostomia, poor oral hygiene, oral bacterial infections, oral viral infections, oral fungal infections, local neural damage, anemia, neutropenia, myelosuppression, occlusal trauma, exacerbated reactions to dental restorative materials, enteral nutrition, lack of chewing function, and xerostomia. Newer cancer therapies also produce oral toxicities through distinct mechanisms. For example, trophoblast cell surface antigen 2 (TROP)-targeted ADCs can cause severe oral mucositis through direct on-target, off-tumor injury to the oral epithelium, and human epidermal growth factor receptor 2 (HER2)-targeted ADCs have also been associated with severe oral irAEs.[524–527] These mechanistic differences are not adequately captured by current grading systems.

What is next? Preclinical, clinical, and translational work in progress

The expanding use of ICIs and other immune-modulating agents underscores the need for better predictive and mechanistic tools for oral irAEs. With the new results from phase 3 studies, cancer patients’ eligibility for ICIs has dramatically increased from 1.5% in 2011 to 56.6% in 2023 in the United States alone.[395] As more patients receive these therapies, the frequency and spectrum of oral irAEs are likely to evolve, including differences between patients with primary vs recurrent or metastatic disease.

Future studies should focus on identifying predictive biomarkers and refining risk stratification. Because oral irAEs are multifactorial, effective approaches will likely require integrated immune, imaging, and microbial markers at baseline, followed by prospective validation in clinically well-annotated cohorts. Such work could improve precision grading, support early intervention, and guide decisions about ICI continuation or rechallenge. Thus, timely public and private investment in translational research will uncover immune, imaging, and microbial markers at baseline with predictive value for further validation and precise oral irAE grading, generating returns in precision diagnosis and improved survivorship.

Conclusion

Systematic reviews of phase 1–3 trials indicate that oral mucositis is the most frequent oral irAE, reported in up to 40% of patients receiving ICIs such as pembrolizumab, nivolumab, and atezolizumab, as well as targeted therapies including mTOR inhibitors, HER inhibitors, and anti-angiogenic agents. Dry mouth or xerostomia is reported in 2–23% of ICI-treated patients and may predispose the patients to secondary oral infections such as candidiasis, rampant dental caries, and periodontal disease. Dysgeusia and dysphagia are also frequently observed. IRON Jaw is a significant, previously under-recognized side effect associated with ICIs, that may lead to severe pain, secondary infections, or even spontaneous fractures (Tables 8–10). Collectively, these irAEs can significantly impair oral function, nutrition, and survivorship, highlighting the importance of early recognition, preventive dental assessment, and multidisciplinary management. A brief summary of oral irAEs management is provided in Table 11.

Oral irAEs are currently underdiagnosed and undermanaged, leading to a significant unmet need for patient adherence to treatment protocols, quality in cancer survivorship, and a reduction in patient readmission. Here, we raise awareness of two emerging new oral diseases associated with ICIs, including IRGENT and Iron Jaw. Notably, although dental baseline evaluation has been incorporated into treatment protocols for bone-modifying agents, such as bisphosphonates and denosumab, dental oncology is not currently part of the routine workflow for ICI initiation in many medical centers. Given the promise of neoadjuvant use of ICIs, many more patients will receive this line of therapy. It is prudent to collaborate with dental oncology and oral medicine services to prevent and early diagnose IRON Jaw and other oral irAEs and improve survivorship by reducing oral side effects. Collectively, managing oral irAE requires early recognition, early involvement of dental oncology for pre-immunotherapy baseline evaluation, early intervention for oral irAE, and structured communications between oral health care providers and oncology teams. Evidence-based management to minimize oral co-morbidity while preserving antitumor efficacy is urgently needed. Integration of dental oncology and oral medicine practice into the multidisciplinary oncology team is essential for early recognition and ongoing management. Elucidating the immunopathogenic mechanisms and identifying patient-specific risk factors will be critical for developing more effective grading systems and preventing oral irAEs.

The preceding sections highlight that toxicity (Fig. 4) is not simply a complication of ICI therapy, but a major determinant of treatment continuity, rechallenge decisions, QOL, and survivorship (Fig. 5). However, toxicity represents only one side of the current ICI plateau. Even when treatment can be delivered safely, many patients either fail to respond from the outset or eventually progress after an initial response. Thus, improving outcomes requires not only safer management of immune activation but also a deeper understanding of why antitumor immunity is absent, insufficient, or lost over time. This leads directly to the complementary challenge of primary and acquired resistance.

Figure 4.

Organ-specific spectrum and clinical characteristics of immune-related adverse events. Organ-specific manifestations of immune-related adverse events across major systems, highlighting differences in frequency, clinical presentation, chronicity, and response to corticosteroids.

AIHA: autoimmune hemolytic anemia; AKI: acute kidney injury; ATIN: acute tubulointerstitial nephritis; CRS: cytokine release syndrome; HLH: hemophagocytic lymphohistiocytosis; ICI: immune checkpoint inhibitor; ILA: interstitial lung abnormalities; irAE: immune-related adverse event; ITP: immune thrombocytopenia; PMR: polymyalgia rheumatica; QOL: quality of life; TTP: thrombotic thrombocytopenic purpura.

Figure 5.

Key challenges in the management of immune related adverse events. Overview of the major clinical and system-level challenges associated with the diagnosis, management, and long-term care of patients with immune related adverse events.

irAE: immune-related adverse event; QOL: quality of life.

Resistance to ICIs remains a central challenge in immuno-oncology.[534,535] Broadly, resistance is often categorized as primary (innate) resistance, defined by lack of response from the outset, and acquired (secondary) resistance, defined by disease progression after an initial objective response or prolonged disease stabilization.[534,535] Although these categories provide a clinically useful framework, it fails to capture the underlying mechanisms that govern immunotherapy failure required for the development of strategies to prevent or reverse resistance. Thus, the current definitions represent simplified constructs within a complex and evolving biological continuum, indicating a critical gap between clinical classification and biological understanding.[535]

Resistance Defined by Response and Duration

In clinical practice, resistance is often operationally defined based on response and duration of response,[536,537] metrics that are downstream clinical outcomes rather than direct reflections of immune biology. These measures integrate diverse and interacting factors, including tumor intrinsic features, host immunity, stochastic immune dynamics, prior therapies, and systemic markers without identifying the drivers of the outcome.[535,538]

Atypical response pattern

Interpretation of response to immunotherapeutic agents is complicated by occurrence of atypical immune-related response patterns. Pseudoprogression, while less common than initially anticipated, continues to complicate early response assessment.[539] Pseudoprogression is generally described as a temporary increase in tumor size followed by disease stabilization and response.[540,541] This phenomenon is considered a treatment-related effect, primarily due to immune cell infiltration and associated inflammation, rather than true disease progression.[542] The incidence of pseudoprogression ranges from 0% to 15%, with most studies reporting rates below 10%.[539,543–545] Because conventional imaging cannot reliably distinguish immune mediated inflammatory changes from viable tumor growth, radiographic progression during immunotherapy creates substantial uncertainty in clinical decision-making.

Global versus local immune control

An ideal immunotherapeutic outcome would involve uniform immune activation and tumor regression across all disease sites. In practice, heterogeneous responses, also called dissociated or mixed responses, have been reported in 3–47% of patients treated with ICIs.[546–548] Responding, nonresponding, and new lesions may coexist within the same patient across primary and metastatic lesions. This phenomenon of intrapatient and intratumoral heterogeneity reflects variability in vascularization, stromal composition, immune infiltration, antigen presentation, PD-L1 expression, IFN-γ signaling pathway, and metabolic constraints, resulting in site specific sensitivity or resistance to immune checkpoint blockade.[549–552] The presence of resistant, responding, or new lesions within the same tumor or patient, challenges the traditional binary response classification.

Because resistance is commonly defined by temporal criteria rather than underlying biology, tumors with distinct resistance mechanisms may be grouped within the same category, while tumors sharing similar biological escape pathways may be classified separately based solely on the timing of progression, underscoring the limitations of radiographic endpoints alone.

Tumor microenvironment as the central determinant of immune response

At a mechanistic level, response and resistance to ICIs are increasingly recognized as properties of the tumor microenvironment (TME)—specifically, the presence, functional state, spatial organization, and regulation of immune cells within the tumor.[535,553–555] Effective immune-mediated tumor control requires coordinated antigen presentation, T cell priming, infiltration, and sustained effector function within a hostile microenvironment characterized by hypoxia, metabolic stress, immunosuppressive cytokines, regulatory immune populations, and stromal barriers.[556]

Concepts such as immune-inflamed (hot), immune-desert (cold), immune-excluded, and intermediate or “lukewarm” tumors are frequently used to predict patterns of immune responsiveness.[554] However, these phenotypes lack strict boundaries and may coexist within the same patient or even within the same tumor.[549–552]

Furthermore, immune activation may occur at one tumor site while remaining absent or ineffective at others within the same patient.[549] Organ-specific immune contexts, particularly within the liver, have been associated with diminished responses to ICIs, reflecting regional immune tolerance and distinct stromal and myeloid landscapes.[534,550,557] As a result, uniform systemic responses remain uncommon, and spatial heterogeneity of the TME represents a major obstacle to durable disease control.[550–552]

The TME determines whether antitumor immunity can be initiated, whether effector cells can access tumor cells, and whether immune responses are sustained or suppressed over time. Resistance should be conceptualized as a failure of one or more of these immune processes rather than as a purely temporal clinical event. Therefore, resistance is best understood as a dynamic process shaped by the TME, host immunity, and treatment-induced selective pressures rather than fixed tumor states.[535,538,549,552,556,558]

Mechanisms of Resistance to Immune Checkpoint Blockade

Mechanistic basis of primary resistance

Primary resistance is frequently associated with TME that are noninflamed or immunologically excluded often referred to as “cold” tumors.[549] These TMEs are characterized by impaired antigen presentation, lack of T cell infiltration, deficient immune priming, or dominance of immunosuppressive populations.[555,556,558] Contributing features may include low tumor immunogenicity, loss of major histocompatibility complex (MHC) class I, defects in IFN-γ signaling, dominant myeloid-derived suppressor cells or regulatory T cell populations, exclusionary stromal architecture, abnormal vasculature, and metabolic constraints that impair effector immune function. In such settings, as effective antitumor immunity is never established, immune checkpoint blockade alone may be insufficient to initiate an immune response.[534,559]

Mechanistic basis of acquired resistance

Acquired resistance typically occurs in inflamed tumors and reflects immune escape under therapeutic pressure.[535,549,552,556,558,560–562] Mechanistically, this includes progressive T cell exhaustion, immune senescence limiting clonal renewal and effector expansion, tumor adaptation under immune pressure such as loss of HLA, loss of target antigen expression, impaired IFN-γ signaling, or development of compensatory immune suppressive mechanisms such as adaptive upregulation of alternative inhibitory checkpoints (T cell immunoglobulin and mucin-domain containing 3 [TIM-3], LAG-3) and expansion of immunosuppressive cell populations.[534,535,549,556,558,560–565] These data suggest that acquired resistance reflects a dynamic remodeling of the TME, leading to loss of immune control despite initial response.

Overlapping mechanisms blur the primary-acquired distinction

Importantly, immune dysfunction programs commonly associated with progression, including T cell dysfunction signatures, suppressive myeloid infiltration, and disrupted interferon signaling, are frequently detected at baseline in tumors that never responded, while similar phenotypes can also arise during therapy through immune pressure and adaptive evolution.[550,556,558,560–562] These observations indicate that primary and acquired resistance are not biologically discrete entities but converge on shared immune states despite potentially distinct temporal origins.

Implications for clinical trial design and translational studies

Resistant immune states observed at progression may either pre-exist at baseline at varying intensity or emerge during therapy through adaptive immune and stromal remodeling.[536,549,552,556,558] Therefore, it is important to understand the baseline immune status and on-treatment immune trajectory of the tumor, requiring integration of clinical outcomes with longitudinal and spatially resolved tissue analyses.[536,553,556,558,565–567] To this end, clinical trials should incorporate baseline biopsies prior to treatment initiation to perform deep baseline immune profiling, on-treatment biopsies to assess early immune modulation, and biopsies at the time of progression to define mechanisms of resistance (Fig. 6A). When sequential combination therapies are administered, serial biopsies following each therapeutic intervention are particularly informative, allowing direct assessment of how individual agents reshape the TME over time (Fig. 6B).[568]

Figure 6.

Study schema for correlative studies. (A) ICI monotherapy (B) ICI-based combination therapy * Biopsy #3 will be performed on progression after achieving stable disease ≥ 6 months, partial response, or complete remission.

ICI: immune checkpoint inhibitor; PBMC: peripheral blood mononuclear cell.

Beyond informing clinical trial design and translational research, these principles have direct implications for the clinical management of patients with acquired resistance to ICIs. When clinically feasible, repeat tumor biopsy at the time of progression (Figs. 6A, B) or liquid biopsy approaches should be considered to identify resistance-associated molecular and immune alterations that may inform subsequent therapeutic strategies. Selection and sequencing of combination approaches should be individualized according to the pattern of resistance, prior treatment exposure, biomarker status, and available clinical evidence. Given the diversity of resistance mechanisms and the limited availability of validated resistance-directed therapies, enrollment in a clinical trial should be strongly considered whenever feasible, particularly for patients progressing after standard treatment options. These clinical considerations further underscore the importance of longitudinal tissue sampling to distinguish pre-existing resistant immune states from adaptive mechanisms that emerge during treatment.

Baseline biopsies define immune states associated with high resistance liability such as profound T cell dysfunction, dominant suppressive myeloid niches, or nonfunctional IFN-γ signaling that predicts the likelihood of primary resistance.[568–570] Early on-treatment biopsies assess immune state transitions, enabling early discrimination between productive immune reprogramming favoring response (such as IFN-γ signaling, improved T cell functional state) and trajectories that predict futility (such as myeloid suppression, increased T cell exhaustion).[556,568] Furthermore, emergence of adaptive suppressive pathways identifies patients at risk for acquired resistance despite any initial benefit. Progression biopsies then resolve whether resistance reflects amplification of pre-existing immune dysfunction or de novo adaptation during therapy.[571,572]

Longitudinal tumor biopsies have been demonstrated to be feasible and safe, with high success rates and low complication rates.[561,568,573] Such trial designs enable precise characterization of immune dynamics and transitions between immune-responsive and immune-resistant states.[561,562,568,573] To minimize confounding by spatial heterogeneity, biopsies should ideally be obtained from the same tumor site across time points, whenever clinically feasible. Standardized lesion selection and documentation protocols improve rigor and interpretability in longitudinal sampling.[550,567,574] Investigators at MD Anderson have developed a web-based lesion selection tool to facilitate this effort, enabling structured tracking of biopsy sites and achieving complete consistency in longitudinal lesion sampling compared with partial consistency in historical controls.[575]

Expanded framing: beyond tissue biopsies alone

Tumor biopsies provide irreplaceable spatial and cellular resolution of tumor immune interface and remain central to defining immune states associated with primary and acquired resistance. However, resistance to immunotherapy is shaped by multi scale processes that extend beyond a single tumor site and evolve over time, limiting the ability of tissue based analysis alone to fully capture resistant dynamics.[576,577] Although integration of transcriptomic and single cell approaches enables dissection of cellular heterogeneity, functional state, and lineage trajectories that underlie immune dysfunction, noninvasive longitudinal immune profiling, such as peripheral immune phenotyping, circulating transcriptomic signatures, and soluble mediators, offers a practical means to monitor new trajectories when repeat biopsies are not feasible and to contextualize local tumor findings within systemic immune responses.[556,566]

Importantly, host level factors also modulate both the likelihood and durability of immunotherapy responses.[534] Age-associated immune remodeling, sex, prior therapeutic exposures, diet, body mass index, neutrophil-to-lymphocyte ratio, metabolic state, prior antibiotic use, and microbiome composition can influence T cell function, myeloid polarization, and inflammatory tone, thereby shaping both primary resistance and the evolutionary pathways leading to acquired resistance.[535,536,538,550,553,554,578–580] These systemic determinants are incompletely captured by tumor tissue analyses alone but may critically condition the capacity for immune reprogramming following therapy initiation. A comprehensive understanding of resistance therefore requires integrated analysis of tumor-intrinsic and tumor extrinsic factors across time.

Perspective

Primary and acquired resistance to ICIs represent overlapping, dynamic processes governed largely by the TME and its spatial and temporal heterogeneity.[535,538,550–555] Clinical classifications based on response timing are useful but incomplete.[535,536,539] Progress in this field will depend on trial designs incorporating longitudinal, site-consistent tumor biopsies and deep translational analyses to capture the evolving immune landscape.[536,553,556,565–567] This knowledge reframes immunotherapy resistance from a descriptive, outcome-based classification to a mechanism driven, biologically grounded framework. Furthermore, combining single-cell and transcriptomic profiling of tissue biopsies with noninvasive longitudinal immune monitoring, and assessment of host determinants, resistance can be interrogated as a dynamic, multi compartment process. This approach enables more precise classification of resistant phenotypes, accurate prediction and earlier interception of therapeutic failure, supports rational selection of combination therapies tailored to specific immune defects, and addresses fundamental knowledge gaps regarding immune plasticity, durability, and reversibility.

The diversity of resistance mechanisms illustrates why empiric ICI use cannot fully address the heterogeneity of patient outcomes. Some patients experience durable benefit, others develop early progression, and still others face substantial toxicity without meaningful tumor control. These divergent trajectories highlight the need for biomarkers that can identify, before or during therapy, which patients are most likely to benefit, which patients are at risk for toxicity, and which resistance pathways may be therapeutically actionable. Therefore, the next critical step is to examine the current biomarker landscape and its limitations.

One of the major limitations in the current use of ICIs is the lack of robust and reliable biomarkers capable of predicting therapeutic response, resistance, and toxicity. Over the past decade, a broad spectrum of candidate biomarkers that represent the host-tumor interaction has been proposed.[581] Tumor-intrinsic features such as PD-L1 expression, tumor mutation burden (TMB), microsatellite instability (MSI) status, alterations in antigen presentation machinery, and specific oncogenic drivers;[564,571,582–599] the composition and functional state of the TME, particularly CD8⁺ T cell infiltration, IFN-γ-associated transcriptional programs, and myeloid populations;[595,600–618] and host-related factors such as germline variation and the gut microbiome shape therapeutic outcomes.[619–626] Circulating biomarkers from peripheral blood including circulating tumor DNA (ctDNA) and immune cell subsets have further expanded the landscape of potential predictors.[342,627–638]

The accompanying table (Supplemental Table S6) synthesizes current evidence across these domains, highlighting biomarkers associated with response, resistance, and irAEs to ICIs. Although the field supports these important ICI biomarker domains, many remain exploratory and context-dependent rather than universally validated.

Beyond molecular and cellular biomarkers summarized in Supplemental Table S6, additional clinical, computational, and imaging-derived measures may also provide insight into treatment outcomes with ICIs. PROs can capture changes in symptoms before and during treatment that may serve as an early indicator of either treatment benefit or toxicities.[639,640] Computational approaches, particularly artificial intelligence-based analysis integrating multimodal data such as histopathology and imaging, are emerging tools that may help predict ICI effectiveness by capturing complex tumor-host interactions.[641–645] Baseline tumor characteristics, including tumor size, metabolic tumor volume, metastatic burden, and liver metastasis have been shown to influence response to ICIs. Smaller baseline tumor size, typically measured by Response Evaluation Criteria in Solid Tumors (RECIST) criteria, is associated with higher objective response rates in melanoma and NSCLC.[646,647] Similarly, lower metabolic tumor volume and total lesion glycolysis measured by FDG PET/CT[648] and delta imaging features from the peritumoral ring measured by CD8-PET/CT imaging[649] predict better response with ICIs. Furthermore, administration of ICIs in the neoadjuvant setting produced high pathological response rates and enhanced recurrence-free survival in patients with melanoma compared to adjuvant setting.[650–652] In addition, higher metastatic burden reflected by a greater number of metastatic sites or involvement of organs such as liver or bone has been associated with lower response rate independent of other clinical factors.[653] The presence of liver metastases at baseline has also been associated with poor response to ICIs.[654,655] These findings highlight the value of integrating clinical reporting, image-based assessments, and computational tools to better characterize response, resistance, and toxicity in patients receiving ICIs.

Limitations and Current Challenges

Despite substantial progress, to date, PD-L1 expression, TMB, and MSI status are the most widely used biomarkers for guiding treatment decisions in immuno-oncology.[656] The clinical translation of biomarkers remains constrained by several conceptual and methodological challenges. First, tumor and immune heterogeneity complicate biomarker interpretation.[550–552] Immune infiltration, checkpoint expression, and antigen presentation capacity can vary substantially across tumor regions and metastatic sites, and these features evolve dynamically under therapeutic pressure. Consequently, biomarkers measured from a single tumor biopsy may inadequately capture the spatial and temporal complexity of antitumor immunity.

Second, many proposed biomarkers exhibit context-dependent predictive value. Widely used markers such as PD-L1 expression or TMB demonstrate variable performance across tumor types.[656] Differences in assay platforms, scoring systems, and lack of harmonized thresholds (notably for PD-L1 and TMB) further limit reproducibility and cross-study comparability. These methodological challenges have limited predictive value and hindered the development of universally applicable biomarkers.

Third, the complexity of antitumor immunity, shaped by interactions between tumor cells, immune populations, stromal elements, and systemic host factors[554,556] suggest that single parameter biomarkers are unlikely to fully capture the determinants of therapeutic response or irAEs.

Emerging Concepts and Future Directions

Recognizing these limitations, the field is increasingly moving toward integrative and longitudinal approaches as responsiveness to immune checkpoint blockade appears to reflect a dynamic interplay between tumor genomics, immune contexture, systemic host biology, and treatment-induced evolutionary pressures.[554,556] Composite biomarker frameworks combining genomic features, immune signatures, circulating markers, radiomic features, and microbiome data are emerging as promising strategies to improve predictive accuracy. Advances in single-cell sequencing, multiplex imaging, and spatial transcriptomics enable deeper characterization of the TME, revealing distinct immune niches and resistance states that may shape responsiveness to checkpoint blockade.

Another emerging concept is the importance of dynamic biomarker assessment. Longitudinal monitoring using ctDNA, immune repertoire profiling, and systemic inflammatory markers may provide real-time insights into treatment response and emergence of resistance. Parallel work has highlighted the potential influence of host factors including the gut microbiome, germline factors, and systemic inflammatory states on immunotherapy outcomes, expanding the biomarker landscape beyond tumor-centric paradigms.

Artificial intelligence driven analytical frameworks are increasingly being applied to integrate high dimensional datasets, including radiologic, digital pathology, genomic, and clinical variables to improve prediction of response and toxicity to ICIs.[657,658] Machine learning models derived from radiomic signatures and histopathological image analysis can capture complex spatial patterns of immune infiltration and tumor architecture that reflect the functional state of the TME.[657] As multimodal datasets continue to expand, AI-enabled platforms capable of integrating tumor-intrinsic features, immune contexture, and host factors may facilitate more accurate and dynamic prediction of ICI treatment outcomes moving the field beyond single-parameter biomarkers toward comprehensive, systems-level models. However, careful attention to data quality is essential, as predictive performance is fundamentally limited by the principle of garbage in, garbage out.[659] The reliability of predictive models depends on rigorous analytical and clinical validation of biomarkers as poorly curated data sets, bias, overfitting, and inadequate validation strategies frequently lead to irreproducible biomarker signatures.[660]

For an ideal clinical decision framework to be clinically useful, it must translate complex biological and clinical information into decisions that can be applied at the point of care. The goal is not simply to predict response, but to estimate the overall therapeutic index for an individual patient by weighing expected benefit against toxicity risk, patient vulnerability, treatment feasibility, and available alternatives. One conceptual approach is to stratify patients into groups (Fig. 7) based on the likelihood of benefit, risk for clinically significant toxicity, and baseline patient characteristics:

Figure 7.

Clinical decision framework based on predicted benefit and toxicity risk. Patients are stratified into four groups according to the likelihood of therapeutic benefit (high vs low) and risk for severe irAEs (high vs low) with ICI-based therapy: (1) optimal with high benefit /low toxicity risk (2) acceptable with high benefit/high toxicity risk, (3) suboptimal with low benefit/low toxicity risk, and (4) unsuitable with low benefit/high toxicity risk. This stratification is intended to guide individualized clinical decision-making. irAEs, immune-related adverse events.

1. High likelihood of benefit with low toxicity risk

2. High likelihood of benefit with elevated toxicity risk

3. Low likelihood of benefit with low toxicity risk

4. Low likelihood of benefit with high toxicity risk

This stratification could facilitate more informed clinical decision-making. Patients in the first group (a) could be offered treatment with ICIs with greater confidence, given the favorable balance between expected benefit and risk. For patients in the second group (b), treatment with ICIs may still be appropriate, particularly when effective alternatives are limited, but would warrant closer monitoring to enable early detection and prompt management of potential irAEs. In contrast patients in the third (c) and fourth group (d) may have a relatively low probability of benefit from ICIs, and treatment decisions in these settings may require other considerations. However, for some patients in the third group (c), combination strategies such as ICI administered together with targeted therapies or chemotherapies may overcome resistance mechanisms and provide meaningful clinical benefit. Patients in the fourth group (d) may represent a particularly vulnerable population, as they are unlikely to derive meaningful benefit while facing a substantial risk of toxicity that could delay subsequent therapies or result in life-threatening complications. Importantly, this proposed stratification should be considered a conceptual framework to guide discussion rather than a validated clinical decision tool.

Although biomarkers offer an important path toward more precise ICI selection, clinical decision-making cannot rely on biomarker data alone. Many patients considered for ICIs have baseline features that were underrepresented or excluded from pivotal trials, including pre-existing autoimmune disease, immunosuppression, organ transplantation, advanced age, pregnancy, disability, or major comorbidities. In these settings, the balance between efficacy and toxicity becomes more complex, and standard trial-derived recommendations may not fully capture individual risk. Therefore, after considering the promise and limitations of biomarkers, it is essential to examine how ICI therapy should be approached in special patient populations where biologic vulnerability, competing risks, and patient-centered outcomes may substantially influence treatment decisions.

The remarkable success of ICIs has transformed oncology practice, yet their application in special populations remains a persistent challenge. Patients with pre-existing autoimmune disease, solid organ transplants, pregnant women, and those requiring rechallenge after severe irAEs have historically been excluded from pivotal clinical trials due to theoretical safety concerns. These exclusions created a critical knowledge gap, as such patients represent a substantial proportion of cancer patients in real-world practice. A significant percentage of patients with cancer have concurrent autoimmune conditions, with some estimates as high as 27%.[661–664] Over 48,000 organ transplants occur annually in the United States alone, with the solid organ transplant recipient population particularly at risk for cutaneous malignancies.[665–666a] Meanwhile the rate of patients receiving a cancer diagnosis while pregnant has also been on the rise, in line with increasing age of gestation and the greater use of assisted reproductive technologies in the developed world.[667] All of these trends have led to increased interest in using ICIs in these and other special populations.

Pre-Existing Autoimmune Disease

From the earliest use of checkpoint inhibitors, there was concern about the use of ICIs in patients with pre-existing autoimmune disease. As the side effects associated with these agents closely approximate the phenotype of certain autoimmune diseases in many cases, there was legitimate concern that the use of immune checkpoints would exacerbate these conditions. Retrospective analyses have suggested that ICIs can be safely administered in this patient population, but there was an increased risk of autoimmune disease flares.[668–672]

The AIM-NIVO trial (NCT03816345) represents a landmark prospective evaluation of nivolumab in patients with pre-existing autoimmune conditions and advanced malignancies. This NCI-sponsored, multicenter study, includes disease-specific cohorts for dermatomyositis and systemic sclerosis, rheumatoid arthritis, Sjögren's syndrome, systemic lupus erythematosus, inflammatory bowel disease, psoriasis and psoriatic arthritis, multiple sclerosis, and other conditions. This prospective effort aims to solve some of the known issues with retrospective analyses. For example, autoimmune disease severity cohorts are grouped with prespecified criteria, written by autoimmune disease specialists.[672a]

Each cohort operates through a collaborative model pairing oncologists with subspecialists in rheumatology, gastroenterology, dermatology, or neurology, enabling comprehensive management of both cancer treatment responses and autoimmune disease activity. This approach allows investigators to characterize disease-specific risk profiles for both autoimmune flares and conventional irAEs, providing granular data to guide clinical decision-making. Early evidence suggests that with rigorous patient selection, baseline disease activity assessment, and intensified monitoring protocols, ICIs can be administered to selected patients with autoimmune disease, though the competing risks of disease flare and irAEs must be carefully balanced against oncologic benefit.

Solid Organ Transplant Recipients

Solid organ transplant recipients face disproportionate cancer burden, with two-to-four-fold elevated overall cancer risk and dramatically higher incidence of specific malignancies including cutaneous squamous cell carcinoma (cSCC; > 65-fold increase) and MCC (∼25-fold increase) compared to age-matched populations.[673] Cancer represents the third leading cause of death in kidney transplant recipients, yet chronic immunosuppression and allograft rejection risk have historically precluded ICI use in this population.[665]

A sentinel 2016 case report described a kidney transplant recipient with refractory metastatic cSCC who achieved complete response to pembrolizumab but experienced acute T cell mediated allograft rejection at 8 weeks, ultimately requiring return to dialysis.[674] Critically, immunohistochemical analysis of the rejected kidney demonstrated activated CD8⁺ and CD4⁺ lymphocytes expressing PD-1, with PD-L1 and PD-L2 expression on glomerular endothelial cells and infiltrating immune cells, establishing the mechanistic importance of the PD-1/PD-L1 axis in maintaining allograft tolerance.

These observations informed the design of ETCTN 10214, a prospective trial evaluating nivolumab combined with standardized low-dose immunosuppression (tacrolimus with serum trough 2–5 ng/mL plus prednisone 5 mg daily) in kidney transplant recipients with advanced cutaneous malignancies. All eight evaluable patients experienced cancer progression on nivolumab monotherapy. Paired tumor biopsies revealed only mild CD8⁺ T cell infiltration despite ICI administration. When patients with progressive disease received ipilimumab plus nivolumab, 2 of 6 patients (33%) achieved complete responses, though treatment-related allograft loss occurred in 2 of the 8 patients.[675]

A similar clinical trial was conducted with cemiplimab, a PD-1 inhibitor, in twelve patients with cSCC and a pre-existing kidney transplant. Patients enrolled were cross tapered to immunosuppression with a mTOR inhibitor and decreasing pulsed doses prednisone. Among the 11 evaluable patients, there were five responses. None of the treated patients experienced organ rejection.[392]

Allogeneic Hematopoietic Stem Cell Transplant Recipients

In contrast to solid organ transplantation, where ICIs can directly disrupt allograft tolerance, the allo- HSCT setting requires balancing graft-vs-tumor (GVT) activity against the risk of graft-vs-host disease (GVHD). An early case series of 31 patients receiving PD-1 blockade for relapsed classical lymphoma (primarily classical Hodgkin Lymphoma) after allo-HSCT showed impressive tumor regressions. Overall response rates approached 77%, with complete responses in roughly half of treated patients. Unfortunately, this early work also highlighted the risk of severe and steroid-refractory acute GVHD that could emerge concomitantly with disease response, with 55% of patients developing treatment-emergent GVHD.[676]

Prospective data with CTLA-4 blockade came from the phase 1/1b trial of ipilimumab in patients with relapse after allo-HSCT reported by Davids et al in 2016. In that study, 28 patients with high-risk hematologic malignancies received ipilimumab at 3 or 10 mg/kg; no objective responses were observed at 3 mg/kg, whereas the 10 mg/kg cohort achieved a 23% complete response rate, including durable remissions in extramedullary acute myeloid leukemia and other aggressive histologies. Immune-related adverse events occurred in 21% of patients, including one treatment-related death, and clinically significant GVHD that precluded further dosing developed in 14%, underscoring the narrow therapeutic window for CTLA-4 blockade in this setting.[677] A subsequent amendment to that trial led to a multicenter phase 1 study of nivolumab for post-allo-HSCT relapse similarly demonstrated meaningful antitumor activity but identified dose-limiting toxicities, including severe GVHD, prompting de-escalation to a maximum tolerated dose of 0.5 mg/kg.[678] Larger pharmacovigilance and case-series analyses have confirmed a strong association between post-allo-HSCT ICI exposure (both PD-1 and CTLA-4 inhibitors) and GVHD, with GVHD-associated mortality reported in up to one quarter of affected patients.[679] As a result, current practice limits ICI use after allo-HSCT to highly selected patients with relapsed disease. Ideally, this should occur within clinical trials, with risk-mitigation strategies and careful monitoring.

Pregnancy

ICI use during pregnancy raises unique concerns, as pregnancy and cancer both exploit immune checkpoint pathways to maintain tolerance. PD-L1 blockade reduces allogeneic fetal survival in murine models, and CTLA-4 on regulatory T cells is essential for pregnancy maintenance through induction of indoleamine 2,3-dioxygenase.

Mittra et al reported an analysis of pregnancy outcomes in patients exposed to immunotherapy in NCI-sponsored clinical trials. This retrospective analysis of the NCI Cancer Therapy Evaluation Program (CTEP)-Adverse Event Reporting System identified nine female patients who had unexpected pregnancies while receiving immunotherapy or cancer vaccines between 2011 and 2020. Seven patients chose to continue their pregnancies to term, and all seven delivered apparently normal infants via vaginal birth. Notably, these exposures occurred during the critical periods of conception and first-trimester organogenesis, yet no apparent fetal complications were observed.[680]

Evens et al reported a landmark case of intentional nivolumab administration in a pregnant patient with relapsed-refractory Hodgkin lymphoma, which was discovered at 13 weeks gestation. The patient was initially treated with chemotherapy, but her disease continued to progress, necessitating salvage therapy late in her pregnancy. She received nivolumab from 26 weeks through 37 weeks of gestation. This appears to be among the latest gestational ages reported for ICI exposure. A healthy infant was delivered without complications.

Pharmacokinetic studies documented nivolumab levels in both placenta and umbilical cord blood, confirming transplacental drug passage at approximately two thirds of the level in the fetus as compared to maternal circulation. This case provided critical pharmacokinetic and safety data for ICI exposure during late pregnancy and demonstrated that successful maternal cancer control could be achieved without apparent neonatal toxicity.[681]

A 2024 analysis of 91 ICI exposures during pregnancy from the WHO pharmacovigilance database revealed that ICIs were not associated with overreporting of adverse pregnancy, fetal, or neonatal outcomes compared to other anticancer agents. Only 3.3% of ICI-exposed pregnancies demonstrated suspected immune-related complications, including one maternal antiphospholipid syndrome with spontaneous abortion, one possible fetal pneumonitis, and one infant with growth restriction and transient congenital hypothyroidism. Importantly, combination anti-PD-1 plus anti-CTLA-4 therapy was associated with higher preterm birth rates, likely reflecting enhanced anti-fetal rejection mechanisms, and should be avoided when possible, during pregnancy.[682] Similarly, in 2023, a case report of severe congenital enterocolitis was reported in an infant at 4 months of age whose mother was treated with pembrolizumab during gestation.[683] Both of these articles suggest that caution is in order in the use of ICIs in pregnancy.

Current recommendations emphasize avoiding ICI therapy during pregnancy when feasible. When ICI treatment is considered for rapidly progressing cancer with limited alternatives, maternal oncologic benefit must substantially outweigh theoretical fetal risks, with comprehensive counseling about potential neonatal immune-related complications such as congenital hypothyroidism and immune-mediated enterocolitis. The expanding use of immunotherapy creates a growing need for systematic prospective data collection on pregnancy outcomes to inform contraception recommendations and guide treatment decisions for this unique, medically-complex population.[667]

ICIs in Patients with Chronic Viral Infections

HIV infection

Combination antiretroviral therapy (ART) has transformed HIV infection into a chronic condition with markedly improved survival, leading to a growing population of people with HIV (PWH) at risk for a competing cause of death, such as a sporadically occurring, non-AIDS-defining malignancy. Historically, PWH were largely excluded from pivotal ICI trials. A systematic assessment of NCI CTEP-sponsored associated studies showed that advocacy by a public health agency sponsor of trials was substantially more likely to lead to the inclusion of PWH on trials than typical, industry-sponsored trials, underscoring the role of public sponsors in broadening access to immunotherapy to underserved populations.[684]

NCI-sponsored interventional trials provide complementary prospective evidence. Cancer Immunotherapy Trials Network-12 (CITN-12) enrolled PWH with controlled HIV infection (CD4 ≥ 100 cells/µL on stable ART) and advanced malignancies to receive pembrolizumab in a phase 1 trial that was defined by cohorts based on CD4 counts of affected patients.[685] The trial demonstrated an acceptable safety profile, with immune-related toxicities similar to those seen in HIV-negative populations and no evidence of clinically meaningful HIV viral rebound. A dedicated Kaposi sarcoma (KS) cohort of CITN-12 was later published, showing meaningful antitumor activity and durable responses to pembrolizumab in HIV-associated KS. In 29 participants with evaluable KS, the overall response rate (ORR) was 62.1% (95% CI, 42.3–79.3). Patients with KS responded regardless of CD4 count.[686]

The NCI-supported AIDS Malignancy Consortium (AMC) also conducted phase I study of ICIs for PWH in their multicenter network of sites. AMC-095 evaluated nivolumab in PWH with a range of advanced cancers on suppressive ART. Among 36 treated participants (median CD4 315 cells/µL, all with HIV RNA < 75 copies/mL at baseline), nivolumab was generally well tolerated; immune-related toxicities were manageable, treatment discontinuation due to irAEs occurred in a minority of patients, and there was no consistent signal of loss of HIV virologic control. A cohort of patients on the trial were treated with nivolumab and ipilimumab, with no obvious safety or feasibility concerns noted.[687]

More robust clinical data have emerged over the last several years. The international CATCH-IT consortium reported outcomes for 390 PWH with diverse malignancies treated with ICIs while on suppressive ART. Across tumor types, objective response rates and survival were comparable to those of matched HIV-negative patients, and the incidence of immune-related adverse events (irAEs) was similar: any-grade irAEs occurred in roughly 20% and grade ≥ 3 irAEs in about 7–8% of patients. CD4⁺ T-cell counts remained largely stable and HIV RNA levels stayed below clinically significant thresholds, including in patients receiving combination PD-1 and CTLA-4 blockade.[688]

Prospective, national, real-world data from the ANRS CO24 OncoVIHAC cohort further support the feasibility of ICI therapy in PWH. In this French study, 140 PWH with cancer treated with ICIs were followed for a median of 9.2 months; the 1-year cumulative incidence of a first grade ≥ 3 irAE was 15.0% (95% CI 9.6–22.9), with an overall incidence of 26.9 severe irAE episodes per 100 person-years. One treatment-related death due to myocarditis was reported. Multivariable analyses identified low CD4 count, longer duration since HIV diagnosis, positive cytomegalovirus serology, and prior cancer surgery as factors associated with increased risk of severe irAEs, while HIV viral suppression and CD4 counts were generally maintained during ICI therapy.[689]

From a mechanistic standpoint, chronic HIV infection is characterized by T-cell exhaustion and a stable viral reservoir that are closely linked to immune checkpoint pathways. HIV persists on ART within long-lived, clonally expanded CD4 T cells, with reservoir maintenance shaped by integration site biology and ongoing immune activation.[690] HIV-specific CD4 and CD8 T cells frequently co-express PD-1, TIGIT, and LAG-3. These exhausted T cells are enriched for replication-competent virus, and that they contribute disproportionately to the latent reservoir.[691] These observations provide a biologic rationale for studying PD-1 blockade in PWH, both to restore antitumor immunity and potentially to perturb the HIV reservoir, an area of active investigation.

Taken together, CATCH-IT, OncoVIHAC, CITN-12 (including the KS cohort), and AMC-095 provide convergent evidence that ICIs can be safely and effectively administered to appropriately selected PWH with controlled HIV infection and adequate CD4 counts, when managed in close collaboration with HIV specialists.

Hepatitis B and C

Patients with chronic hepatitis B virus (HBV) or hepatitis C virus (HCV) infection have also been systematically underrepresented in ICI trials because of concerns regarding viral reactivation and immune-mediated hepatitis. A 2023 systematic review and meta-analysis of ICIs in patients with chronic HBV or HCV, including roughly 1000 patients across 49 studies, found that objective response rates and survival outcomes were similar to those of uninfected patients, supporting the oncologic efficacy of ICIs in this population. Overall rates of grade 3–4 irAEs were comparable between patients with and without viral hepatitis (21.3% vs 21.8%), although hepatic toxicity was more frequent and somewhat more severe in those with chronic HBV or HCV.[692]

Importantly, several ICIs now carry regulatory approval for the treatment of hepatocellular carcinoma (HCC) in populations largely driven by chronic HBV and HCV, including nivolumab, pembrolizumab, and the atezolizumab–bevacizumab regimen, among others.[693] In the pivotal phase II and III HCC trials that supported these approvals, patients with chronic HBV or HCV were generally eligible provided viral replication was controlled and liver function met prespecified criteria, demonstrating that ICIs can be administered safely in carefully selected patients with viral hepatitis–related cirrhosis.[694,695] Across an MD Anderson analysis of patients receiving ICIs in the setting of chronic viral hepatitis B or hepatitis C, researchers found that HBV reactivation events were uncommon and occurred mainly in patients not receiving appropriate antiviral prophylaxis, with only four documented cases of HBV reactivation and no HCV reactivation events.[696]

Best practice in this setting includes systematic evaluation of baseline liver enzymes and virologic status, initiation or continuation of nucleos(t)ide analog therapy in patients with HBV, and close collaboration with hepatology or infectious disease specialists. Although current evidence supports the safety and efficacy of ICIs in patients with chronic viral hepatitis, including those enrolled in large HCC trials, heightened vigilance for hepatic irAEs and prompt intervention are warranted, especially when multiple ICIs are used in combination or when significant underlying liver disease is present. Treatment-emergent ICI hepatitis can be a diagnostic dilemma in patients with either primary liver cancers or metastases of other cancers to the liver. In patients with chronic hepatitis B and C, reactivation of a chronic viral infection could further complicate the differential diagnosis.

The challenges encountered in special patient populations further emphasize that precision immuno-oncology must extend beyond the binary decision of whether to use an ICI. For many patients, the more clinically relevant question is how ICIs can be delivered in a way that preserves benefit while reducing avoidable harm. This includes consideration of pharmacokinetics, dose, dosing interval, treatment duration, rechallenge, de-escalation, and sequencing with other therapies. As ICIs move into earlier disease settings and are used for longer periods, optimizing treatment delivery becomes a central component of personalized immunotherapy.

Pharmacokinetics, Dose Optimization, and Biological Rationale

Pharmacokinetic characterization has historically played a central role in oncology drug development, guiding dose selection, exposure–response relationships, and safety margins. In cytotoxic and targeted therapies, serum drug concentration is often closely linked to both efficacy and toxicity. However, immuno-oncology represents a fundamental shift in this paradigm.[697–699] For ICIs, clinical activity is driven less by sustained systemic drug exposure and more by effective immune engagement and downstream immune activation.[697–699]

Unlike cytotoxic agents or small-molecule inhibitors, ICIs do not exert direct antitumor effects. Instead, they release inhibitory immune checkpoints, enabling endogenous antitumor immunity.[700–702] As a result, exposure–response relationships for ICIs are often relatively flat across a wide dosing range once a minimal effective exposure threshold is achieved, with no consistent improvement in efficacy at higher serum concentrations, although tumor-specific heterogeneity exists.[697–699] For example, exposure–response relationships for pembrolizumab have been demonstrated in melanoma and non–small cell lung cancer (NSCLC) but not consistently in RCC.[698,703]

Early clinical development of PD-1 inhibitors relied heavily on pharmacodynamic endpoints such as receptor occupancy, which demonstrated near-complete PD-1 saturation on circulating T cells at relatively low doses.[523,704,705] These findings suggested that substantial immune engagement could be achieved without escalating to doses traditionally used in oncology phase I trials. However, receptor occupancy alone is an imperfect surrogate for biological activity.[706–708] Downstream immune activation, including cytokine signaling and T cell functionality, may continue to evolve beyond apparent receptor saturation, and assay methodology, tissue compartment differences, and receptor internalization complicate interpretation.[706,707,709]

Beyond static exposure metrics, dynamic pharmacokinetic parameters have emerged as clinically relevant. Decreasing clearance over time has been associated with treatment response for both nivolumab and pembrolizumab, likely reflecting changes in disease burden and systemic inflammation.[703,710–713] Although the causal relationship between clearance changes and therapeutic benefit remains under investigation, time-varying clearance may represent a more informative pharmacokinetic–pharmacodynamic signal than baseline exposure alone.[703,710–714]

These principles informed the eventual dose selection of nivolumab and pembrolizumab. Initial dose-escalation studies explored substantially higher doses than those ultimately adopted.[523,704,705] Subsequent pharmacokinetic and pharmacodynamic analyses demonstrated that receptor occupancy and immune activation plateaued at relatively low doses, with no meaningful gain in antitumor efficacy at higher exposures.[523] These data supported the transition to simplified flat dosing regimens that maintained clinical efficacy while reducing unnecessary drug exposure.[715–717]

In addition, there is a growing body of data indicating that the time of day (ToD) immunotherapy is administered may impact treatment efficacy.[718] Numerous retrospective studies demonstrated that earlier ToD administration, particularly mornings and early afternoons, are associated with improved OS and PFS across various cancer types including melanoma,[719] NSCLC,[720] RCC,[721] and others.[718] Proposed mechanisms for this time-dependent variability include circadian variations in CD8⁺ T cell tumor infiltration and changes in endothelial cell expression of adhesion molecules.[722]

Collectively, experience with ICIs underscores a broader principle in immuno-oncology: clinical efficacy depends not on maximizing serum drug concentration, but on effectively engaging immune pathways capable of generating durable antitumor responses.[697–699] Although pharmacokinetic assessment remains essential for safety and regulatory purposes, dose optimization for immune-based therapies should prioritize immune function over traditional exposure thresholds.[699,723]

Dose and Interval Adaptations in Resource-Constrained Settings

Despite the proven efficacy of ICIs across multiple malignancies, global access to these therapies remains highly uneven. In many healthcare systems, the cost of prolonged immunotherapy limits availability and continuity of care.[724–727] In this context, clinicians have explored pragmatic dose and interval adaptations, including reduced fixed doses and extended dosing intervals, to improve affordability while preserving clinical benefit.[724,725,727]

Observational studies, institutional experiences, and conference-reported cohorts—particularly from low- and middle-income countries—have described meaningful antitumor activity using reduced-dose or less frequently administered PD-1 inhibitors.[724,725,728–731] A recent systematic review encompassing over 2000 patients reported objective responses and disease control using regimens such as nivolumab 40 mg every 2 weeks or 20 mg every 3 weeks.[724,729] Across studies, radiologic response rates ranged from 5% to 75%, with outcomes comparable to pivotal trials in selected tumor types including Hodgkin lymphoma, lung cancer, and RCC.[724,728,731,732]

Although heterogeneous and frequently limited by retrospective design, these findings are biologically plausible given early target saturation and flat exposure–response relationships.[697–699] Comparative studies in NSCLC have demonstrated no statistically significant differences in OS between standard-dose and low-dose pembrolizumab.[733] Randomized data remain limited, but notable trials include a randomized phase III study in head and neck cancer demonstrating improved survival with low-dose nivolumab added to metronomic chemotherapy, and the randomized PLANeT phase II trial in triple-negative breast cancer showing improved pathologic complete response rates with low-dose pembrolizumab added to neoadjuvant chemotherapy.[734–736] Other strategies may include prioritizing early ToD administration of immunotherapy to maximize the potential circadian benefit. This would require minimal cost and risk to implement, even in resource-limited settings.

Duration of Immune Checkpoint Inhibitor Therapy: Is 2 Years Enough?

The optimal duration of ICI therapy remains unresolved. Many pivotal trials capped treatment at 2 years, a practice that has since been widely adopted in clinical care.[715,717] This approach is pragmatic rather than biologically definitive, and whether 2 years represents sufficient, excessive, or inadequate treatment remains an open question.[715,717]

Long-term follow-up from melanoma and NSCLC trials demonstrates that a subset of patients can maintain durable remission after discontinuing therapy at 2 years.[715,737–739] Five-year follow-up from KEYNOTE-006 showed that 78.4% of patients who completed 2 years of pembrolizumab remained progression-free at 24 months after treatment cessation, with 24-month OS of 95.9%.[739] In NSCLC, patients who completed 2 years of ICI therapy demonstrated 12-month PFS rates of 81.1% and OS rates of 96.4% after treatment completion.[737]

Potential advantages of stopping at 2 years include reduced cumulative immune-related toxicity, mitigation of financial toxicity, improved QOL, and avoidance of diminishing returns once immune memory has been established.[715,717] However, discontinuation introduces significant uncertainty. Among patients who completed 2 years of pembrolizumab in KEYNOTE-006, 26% experienced progression, with a median time to progression of 33.3 months from treatment cessation.[739] In a French real-world cohort of NSCLC patients who discontinued ICIs after at least 18 months, 33% experienced tumor progression after a median time of 10.0 months.[740] Depth of response strongly influences outcomes, with complete responders demonstrating substantially lower relapse rates than patients with partial response or stable disease.[715,738–741]

Rechallenge after relapse is feasible but generally less effective than initial therapy. In KEYNOTE-006, among patients receiving second-course pembrolizumab, objective responses were observed in a subset, with durable disease control in selected cases.[739] Meta-analyses report objective response rates of approximately 20% upon rechallenge with ICIs, with higher response rates observed in patients who previously achieved deep or durable responses and in those with longer treatment-free intervals.[739,740,742]

Considerations of dose, interval, and treatment duration demonstrate that the future of ICI therapy depends not only on identifying the right patient, but also on selecting the right treatment strategy over time. Toxicity risk, resistance biology, biomarker information, patient-specific vulnerabilities, pharmacokinetics, and treatment goals all interact to shape the therapeutic index of checkpoint blockade. These domains should therefore not be viewed as separate challenges, but as interconnected determinants of benefit and harm. The final section integrates these themes into a practical framework for moving beyond the current plateau toward more individualized, evidence-based immuno-oncology care.

The preceding sections illustrate that the current plateau in ICI therapy cannot be explained by any single limitation. irAEs constrain treatment delivery and survivorship, resistance limits efficacy, biomarker gaps impair patient selection, special patient populations expose the limitations of trial-derived evidence, and unresolved questions regarding dose, interval, and duration highlight the need to optimize how ICIs are delivered over time (Fig. 8). Taken together, these challenges support a broader model of precision immuno-oncology—one that integrates tumor biology, host immune state, toxicity risk, pharmacokinetics, treatment schedule, patient comorbidities, and patient goals into a unified clinical decision framework.

Figure 8.

Clinical and biological barriers underlying the current plateau in immune checkpoint blockade. The efficacy of ICI is constrained by irAEs, limitations of existing biomarkers, primary and acquired resistance mechanisms, challenges in treatment optimization, and the complexity of special patient populations. Addressing these interconnected barriers through precision immuno-oncology approaches will be critical to improving outcomes and expanding the therapeutic potential of cancer immunotherapy.

ctDNA: circulating tumor DNA; ICI: immune checkpoint inhibitor; irAEs: immune-related adverse events; MSI: microsatellite instability; PD-L1: programmed cell death ligand 1; TMB: tumor mutation burden.

Heightened awareness of irAEs therefore remains essential. irAEs exhibit marked heterogeneity across organ systems. They differ substantially depending on the affected organ, with variability in presentation, severity, reversibility, and responsiveness to immunosuppression. Some manifestations are frequent and persistent, while others are uncommon but potentially severe or difficult to treat, highlighting the need for tailored, organ-specific care. At the same time, clinical decision-making is complicated by imprecise diagnostic tools, overlapping clinical features, and an unpredictable disease course. Long-term consequences, including ongoing morbidity and the need for continued monitoring, further add to the burden of care. These challenges are compounded by gaps in evidence and variability in clinical practice. Although corticosteroids remain the cornerstone of treatment for many irAEs, treatment approaches may differ according to the affected organ system (e.g., myocarditis vs endocrinopathies), toxicity severity, and available evidence. As a result, decisions regarding corticosteroid dosing, duration, and escalation to steroid-sparing therapies often require individualized clinical judgment. This variability underscores the importance of personalized strategies, multidisciplinary coordination, and continued efforts to refine risk assessment and evidence-based management approaches.

Although substantial progress has been made in recognizing and managing system specific toxicities, current approaches remain largely reactive. Ideally, predictive models capable of estimating both therapeutic benefit and irAE risk would allow clinicians to prospectively identify patients most likely to benefit from treatment while safely tolerating therapy, as well as those for whom toxicity risk outweighs potential benefit. At present, such validated tools are lacking. As a result, vigilance remains the cornerstone of safe immunotherapy delivery. Early recognition of subtle or evolving irAE symptoms, prompt diagnostic evaluation, and timely intervention are critical not only to prevent life threatening complications, but also to minimize treatment interruptions, preserve antitumor efficacy, and reduce the substantial healthcare costs associated with management of severe immune mediated toxicity.

The absence of reliable biomarkers to anticipate primary resistance, detect emerging secondary resistance, or guide rational sequencing and combination strategies further limits the precision of immunotherapy. Biomarkers that integrate tumor intrinsic characteristics, host immune state, and treatment induced immune perturbations would enable more individualized decision-making, improve patient selection, and reduce both ineffective treatment exposure and unnecessary toxicity.

Looking ahead, meaningful progress will require deliberate integration of translational research into routine clinical practice. Eligible patients should be enrolled in thoughtfully designed clinical trials that incorporate systematic biospecimen collection, longitudinal immune monitoring, and real-world toxicity assessment. Patient selection remains a critical yet often underappreciated determinant of trial success. Heavily pretreated and highly refractory patients with significant tumor burden, whose remaining options are limited to experimental immunotherapy or hospice, may not be optimally served by early phase immunotherapy studies in the dose escalation setting, where the objectives are to establish tolerability and identify early signals of efficacy.

In the clinical trial setting, expecting sustained objective responses in such populations and subsequently abandoning further drug development in the absence of these outcomes may not always represent the most informative or constructive approach. Each patient enrolled in a clinical trial provides an opportunity to generate meaningful clinical and translational insights. Systematic learning from both responders and nonresponders can clarify mechanisms of action, resistance, and toxicity, and help refine future patient selection strategies. These insights are essential for determining how investigational agents might be deployed earlier in the disease course or rationally combined with other anticancer therapies when scientifically and clinically feasible, ultimately maximizing therapeutic potential and patient benefit. Immunotherapy should be considered earlier in treatment planning discussions, alongside other systemic options, particularly in settings where both efficacy and toxicity can be meaningfully evaluated.

Taken together, the trajectory of ICI therapy reflects both the power and the complexity of therapeutic immune activation. Toxicity, resistance, biomarker limitations, special population considerations, and treatment-delivery questions are not isolated problems; rather, they are interdependent barriers that determine whether checkpoint blockade produces durable benefit, avoidable harm, or no meaningful clinical improvement. The next phase of immuno-oncology will require integrated clinical decision tools that align the right treatment, dose, schedule, monitoring strategy, and patient population with the biology of each cancer and the priorities of each patient. Ultimately, the goal extends beyond achieving objective responses to inducing durable remission, eradicating minimal residual disease, and overcoming tumor plasticity that drives adaptation, immune escape, and disease recurrence (Fig. 9). This approach is essential to disentangle the complex interplay between patient selection, therapeutic efficacy, resistance mechanisms, and immune mediated toxicity, and to advance predictive frameworks that can meaningfully guide clinical care. Although predictive tools remain an unmet need, the field is well positioned to advance through careful clinical application of ICIs and other immunotherapeutic agents. With continued investment in biomarker development, predictive modeling, and well-designed clinical trials, ICIs, particularly when integrated with other anticancer therapies, is poised to move beyond its current plateau and deliver increasingly durable and meaningful benefit to patients.

Figure 9.

Ultimate goals of next generation immuno-oncology. Future immune checkpoint inhibitor-based immunotherapeutic strategies should aim to induce durable remission, eradicate minimal residual disease, and overcome tumor cell plasticity, thereby reducing immune escape, treatment resistance, and disease recurrence.

APPENDIX

List of corresponding authors by section.

Section	Corresponding Author	Email
Introduction	Aung Naing, MD	anaing@mdanderson.org
Cutaneous immune-related adverse events	Omar Pacha, MD	opacha@mdanderson.org
Endocrine immune-related adverse events	Zoe Quandt, MD, MS	Zoe.Quandt@ucsf.edu
Gastrointestinal immune-related adverse events	Yinghong Wang, MD, PhD	YWang59@mdanderson.org
Hepatobiliary immune-related adverse events	Hao Chi Zhang, MD	HZhang20@mdanderson.org
Pancreatic immune-related adverse events	Anusha Shirwaikar Thomas, MD	ASThomas1@mdanderson.org
Pulmonary immune-related adverse events	Ajay Sheshadri, MD	ASheshadri@mdanderson.org
Neurologic immune-related adverse events	Ivan Darin Carabenciov, MD	Carabenciov.Ivan@mayo.edu
Rheumatic immune-related adverse events	Maria E. Suarez-Almazor, MD, PhD	msalmazor@mdanderson.org
Immune checkpoint inhibitor-associated acute kidney injury	Shruti Gupta MD, MPH	sgupta21@bwh.harvard.edu
Cardiac immune-related adverse events	Nicolas L. Palaskas, MD, MPH	NLPalaskas@mdanderson.org
Hematologic immune-related adverse events	Jean M Connors, MD	jean_connors@dfci.harvard.edu
Audiovestibular immune-related adverse events	Marc-Elie Nader, MD	mnader@mdanderson.org
Ocular immune-related adverse events	Nagham Al-Zubidi, MD	nsal@mdanderson.org
Oral immune-related adverse events	Yu Leo Lei, DDS, PhD	YLeoLei@mdanderson.org
Primary and acquired resistance to immune checkpoint inhibitors	Joud Hajjar, MD, PhD, MS	Joud.Hajjar@bcm.edu
The biomarker landscape of immune checkpoint inhibition: predicting response, resistance, and toxicity	Aung Naing, MD	anaing@mdanderson.org
Recommendations regarding the use of immune checkpoint inhibitors in special populations	Elad Sharon, MD, MPH	elad_sharon@dfci.harvard.edu
Optimizing immune checkpoint inhibitor therapy: dose, interval, and duration	Aung Naing, MD	anaing@mdanderson.org
Integrating toxicity management, understanding resistance and filling biomarker gaps: A path forward	Aung Naing, MD	anaing@mdanderson.org

Conflicts of Interest

Joud Hajjar: research funding from the AstraZeneca, Jeffrey Modell Foundation, Takeda, and Immune Deficiency Foundation; principal investigator (PI) for a Pharming-sponsored clinical trial; advisory board role for Immune Deficiency Foundation, Cogent, Pharming, ADMA Biologics, and CSL Behring; honoraria from WebMD (non-branded educational talk) and RealCME (non-branded educational course development). Ashley E. Aaroe: research funding from NCATS L30TR005243-01 National Institutes of Health (NIH) Loan Repayment Program; consulting fees from M3 Global Research; honorarium for Methodist Advances in Neurology Symposium talk and ADVI/Astellas Delphi Panel Member for the enfortumab vedotin neuropathy management panel; and Society for Neuro Oncology Public Policy Committee Chair (unpaid). Jean M Connors: consulting fees from Abbott Laboratories, Alynylam, Anthos Therapeutics, Bayer Pharmaceuticals, Bristol-Myers Squibb (BMS), Janssen Pharmaceuticals, Novartis, Perosphere Technologies, Pfizer, Inc, Regeneron Pharmaceuticals; Data Safety Monitoring Board or advisory board roles for Abbott Laboratories, Alexion Pharmaceuticals, Bayer Pharmaceuticals, BMS, Cerus Corporation, Janssen Pharmaceuticals, Perosphere Technologies, Regeneron Pharmaceuticals; and committees and leadership roles as AC Forum Board of Directors (unpaid), ISTH Council Member (unpaid), and ASH Committee on Practice (unpaid). Michael A. Davies: research funding from NIH National Cancer Institute (NCI) (P50CA221703), LEAD Pharma, ABM Therapeutics, Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, AIM at Melanoma Foundation, American Cancer Society and Melanoma Research Alliance, Cancer Fighters of Houston, Anne and John Mendelsohn Chair for Cancer Research, Andrew M. McDougall Brain Metastasis Clinic and Research Program, and other philanthropic contributions to the MD Anderson Melanoma Moon Shots Program; consulting fees from Replimmune, Nurix, BMS, Eisai, Iovance, Merk, ABM Therapeutics, Theratrame; and scientific advisory board role for Theratrame. Clifton D. Fuller: research funding from NIH NCI (R01CA257814, P30CA016672); Elekta AB, and NRG Oncology; advisory/consultancy role with the International Atomic Energy Agency; honoraria, travel, meals, or other in-kind support from NIH, GE Healthcare, Varian/Siemens Healthineers, Australian & New Zealand Head and Neck Society, Elekta AB, Philips Medical Systems, European Commission/European Innovation Council and SMEs Executive Agency (EISMEA); American Society for Radiation Oncology (ASRO), European Society for Radiotherapy and Oncology (ESRO), American Association for Physics in Medicine (AAPM), Radiological Society of North America (RSNA), and NRG Oncology; committee service and leadership roles with the American Society for Clinical Oncology (ASCO), ASRO, AAPM, and NIH; royalties or license (via University of Texas System) from Kallisio, Inc; and U.S. patent (11730561) issued to the University of Texas System. Dan S Gombos: honoraria for for lectures, presentations, speaker bureaus at Tulane University, American Board of Ophthalmology, and Stanford University; support for attending meetings and/or travel from Castle Biosciences, American Board of Ophthalmology, and American Association of Ophthalmic Oncology & Pathology. Shruti Gupta: research funding from NIH NIDDK (K23DK125672 and R03DK141708), BTG International, AstraZeneca, Johnson & Johnson, Travere, MGB Department of Medicine, Chen Foundation, American Heart Association; consultant for BTG/SERG, Mersana Therapeutics, and Alexion; and honoraria (speaker fees) from BTG/SERB. Katherine A. Hutcheson: research funding from NIH (R01DE034780, R21CA273984, P01CA285249; P01CA278716; R01CA271223, PAR-20-060 U01), Cancer Prevention & Research Institute of Texas (CPRIT) (RP240125), U.S. Department of Defense (DoD) (Co-PI: 81XWH-20-PRCRP-BHSA), Patient-Centered Outcomes Research Institute (PI: PCS-1609-36195), and an unrestricted educational grant to MD Anderson Head & Neck Collaborative from Atos Medical; payment, honoraria, or other support for travel/attending conferences: International Academy of Oral Oncology, American Cancer Society, Dysphagia Café, American Speech-Language-Hearing Association; and steering committee role for the American Head and Neck Society’s Adoption of Guidelines for Reporting on Functional & Quality Outcomes Task Force. Aye Khine: research funding from NIH T32 Training Grant (5T32DK007418-43). Stephen Y. Lai: research funding from NIH (PO1 CA278716, PO1 CA285249, P30CA016672, RO1 DE032521, RO1 CA280980, R21 CA259839, T32 CA261856, UO1 DE032168, U54 CA274321, RO1 DE028290, RO1 DE025248); and consultant for Cardinal Health. Yu Leo Lei: research funding from NIH (R01 DE026728, R01 DE034406, U01 DE033330); licensing royalties (via University of Michigan) from Kerafast Inc. and Applied Biological Materials Inc.; and co-founder and advisory board role at Saros Therapeutics Inc. Amy C. Moreno: research funding from NIH/NCI/NIDCR (K01DE030524-01A1, P01CA285249-01A1, R21DE031082-01, U01DE032168-01A1, R01DE034780, P01CA278716-01A1, R01DE034406), CPRIT (RP230022), and The University of Texas MD Anderson Cancer Center Physician Scientist Program Award. Marc-Elie Nader: ownership of stock: Palantir, Abbvie. Nicolas L. Palaskas: research funding from CPRIT (RP200670), NIH/NCI (P01CA261669-01), and FDA (U01FD008717-01); consultant for Kiniksa Pharmaceuticals; honoraria for lecture: Novant Health, University of Chicago, ATRIUM Health, University of Virginia. Anisha B. Patel: research funding from Lutris Pharma, Hoth Therapeutics, Azitra; royalties from UpToDateOnline; and scientific advisory board roles: Lutris Pharma, Fortress Biotech, Revolution Medicine, Astellas. Zoe Quandt: research funding from NIH (K08AI193076) and the Breakthrough Type 1 Diabetes Kellogg Career Development Award; consultant and advisory board fees from Sanofi. Elad Sharon: consulting fees from D.E. Shaw Research; participation on Data Safety Monitoring Board or advisory board: FortVita and Mallinckrodt/Therakos; patent pending (Use of Atezolizumab for the Treatment of Patients with Alveolar Soft Part Sarcoma). Ajay Sheshadri: research funding from NIH National Heart, Lung, and Blood Institute (NHLBI) and Gateway for Cancer Research; consulting or advisory board roles: Tract Bio, WCG Biohaven, Pfizer, and Pulmotect. Maria E. Suarez-Almazor: research funding from Novartis (unrelated clinical trial contract); advisory board roles at Syneos Health and Setpoint Medical. Yinghong Wang: consulting for Abbvie, Ferring, and Thornhill; honoraria for lecture from Duke University; royalties or licenses: Up-to-date (panel member for guideline) and Springer (book editor). Hao Chi Zhang: participation on Data Safety Monitoring Board or advisory board at Novartis Pharmaceuticals Corporation. Aung Naing: research funding from NIH/NCI (PI/PD: R01CA279749), (Co-PI: R01CA269622), NCI, EMD Serono, MedImmune, Healios Onc. Nutrition, Atterocor/Millendo, Amplimmune, ARMO BioSciences, Karyopharm Therapeutics, Incyte, Novartis, Regeneron, Merck, BMS, Pfizer, CytomX Therapeutics, Neon Therapeutics, Calithera Biosciences, TopAlliance Biosciences, Eli Lilly, Kymab, PsiOxus, Arcus Biosciences, NeoImmuneTech, Immune-Onc Therapeutics, Surface Oncology, Monopteros Therapeutics, BioNTech SE, Seven & Eight Biopharma, and SOTIO Biotech AG, GV 20 Therapeutics, Sanofi, Orionics, Novatek, BeOne, and Summit; consulting roles: CTI, Deka Biosciences, Janssen Biotech, NGM Bio, PsiOxus Therapeutics, Immune-Onc Therapeutics, STCube Pharmaceuticals, OncoSec KEYNOTE-695, Genome & Company, CytomX Therapeutics, Nouscom, Merck Sharp & Dohme Corp, Servier, Lynx Health, AbbVie Mural Oncology, Merck; payment or honoraria for speaking, attending meetings, or travel: AKH Inc, The Lynx Group, Society for Immunotherapy of Cancer, Korean Society of Medical Oncology, Scripps Cancer Care Symposium, ASCO Direct Oncology Highlights, ESMO, and CME Outfitters; ARMO BioSciences, NeoImmuneTech, NGM Biopharmaceuticals; Founder: KN Foundation USA. The remaining authors report no conflicts of interest.

Supplemental Material

Supplemental materials are available online with the article.