Toxicity of Immune Checkpoint Inhibitors: Considerations for the Surgeon

Abstract

Background.

The use of immunotherapeutic agents, specifically immune checkpoint inhibitors (ICIs) for solid malignancies, is rapidly rising, and many new agents and treatment combinations are in development. However, ICIs have a unique side-effect profile of immune-related adverse events (irAEs) compared with chemotherapeutic agents or targeted therapies.

Methods.

In this report the diverse spectrum of irAEs is highlighted using two patients with metastatic melanoma undergoing treatment with ICIs. We supplement these case reports with a brief literature review of the data regarding the safety of surgical intervention in patients taking irAEs.

Results.

The report describes the basic approach to the detection and management of irAEs, notes important perioperative considerations, and discusses the safety of surgical intervention for these patients.

Conclusions.

Overall, these irAEs represent a diverse group of pathologies with variable timing and sometimes subtle presentation requiring careful monitoring and heightened clinical suspicion for potential toxicity by all providers, including surgeons.

Immunotherapy is being used across all cancer types and spectrums of disease with significant improvements in outcomes, even for patients with advanced metastatic disease. Immune checkpoint inhibitors (ICIs) represent the most common form of immunotherapy used currently, with approved agents being applied in the treatment of patients with numerous disease types, sometimes as front-line therapy (Table 1). Additionally, these agents currently are used for patients with earlier-stage disease, including disease amenable to surgical resection in the neoadjuvant and adjuvant setting. Furthermore, unique combinations of ICIs with traditional anti-cancer therapies, including radiation and/or chemotherapy and/or additional new agents that affect other immune pathways, are becoming increasingly common.¹ Approved agents currently include anti-CTLA-4 (ipilimumab/Yervoy), anti-PD-1 (nivolumab/Opdivo, pembrolizumab/Keytruda, cemiplimab/Libtayo), and anti-PD-L1 (atezolizumab/Tecentriq, avelumab/Bavencio, durvalumab/Imfinzi) (Table 1), with many more in clinical trials.

TABLE 1.

FDA approvals for immune checkpoint inhibitors

Target	Agent	Disease	Year approved
CTLA-4	Ipilimumab (Yervoy)	Metastatic melanoma	2011
		Stage 3 melanoma (adjuvant)	2015
PD-1	Nivolumab (Opdivo)	Metastatic SCLC	2018
		HCC previously treated with sorafinib	2017
		MSI-hi or dMMR CRC	2017
		Advanced urothelial carcinoma	2017
		Stage 3 melanoma (adjuvant)	2017
		Recurrent/metastatic HNSCC	2016
		Relapsed cHL	2016
		Advanced RCC	2015
		Advanced non-squamous NSCLC	2015
		Metastatic squamous NSCLC	2015
		Advanced metastatic melanoma	2014
	Pembrolizumab (Keytruda)	Metastatic SCLC	2019
		HNSCC	2019
		Advanced RCC (with axitinib)	2019
		Stage 3 (not amenable to surgery or radiation) or metastatic NSCLC	2019
		High risk stage 3 melanoma (adjuvant)	2018
		Recurrent or locally advanced MCC	2019
		Advanced HCC	2018
		Advanced squamous NSCLC (in combination with chemo)	2018
		Advanced non-squamous NSCLC (in combination with chemo)	2018
		PMBCL	2018
		Metastatic/refractory cervical cancer	2018
		Advanced/recurrent stomach and gastroesophageal cancer	2017
		All metastatic solid tumors classified as MSI-hi or dMMR	2017
		Advanced urothelial carcinoma	2017
		Advanced non-squamous NSCLC (in combination with chemo) irrespective of PD-L1	2017
		Refractory/relapsed cHL	2017a
		Recurrent/metastatic NSCLC (no EGFR or ALK mutations with high PD-L1 levels)	2016
		Recurrent/metastatic HNSCC	2016
		Squamous and non-squamous advanced NSCLC with PD-L1 expression	2015
		Advanced or unresectabie melanoma	2014
	Cemiplimab (Libtayo)	Cutaneous SCC	2018
PD-L1	Atezolizumab (Tecentriq)	Advanced SCLC (in combo with chemo)	2019
		Unresectable or metastatic TNBC (in combo with chemo)	2019a
		Metastatic non-squamous NSCLC (in combination with avastin and chemo)	2018
		Urothelial carcinoma unable to receive cisplatin	2017a
		Metastatic resistant NSCLC	2016
		Urothelial carcinoma	2016
	Avelumab (Bavencio)	Advanced RCC (in combo with axitinib)	2019
		Locally advanced/metastatic urothelial carcinoma	2017a
		MCC	2017
	Durvalumab (Imfinzi)	Unresectable, stage 3 NSCLC	2018
		Urothelial carcinoma	2017a
Combination	Nivolumab and ipilimumab	MSI-hi or dMMR relapsed/refractory CRC	2018
		Advanced RCC	2018
		Advanced melanoma	2015

FDA Food and drug administration, SCLC small cell lung cancer, HCC hepatocellular carcinoma, MSI-hi microsatellite instability high, dMMR mismatch repair deficient, CRC colorectal cancer, HNSCC head and neck squamous cell carcinoma, cHL classical Hodgkin’s lymphoma, RCC renal cell carcinoma, NSCLC non-small cell lung cancer, MCC Merkel cell carcinoma, PMBCL primary mediastinal B-cell lymphoma, EGFR epidermal growth factor receptor, ALK anaplastic lymphoma kinase, SCC squamous cell carcinoma, TNBC triple-negative breast cancer

^aIndicates accelerated approval; blue indicates approved front-line therapy

Coincident with these developments is an increasing need for surgeon engagement (e.g., in neoadjuvant strategies and consolidative surgery after an effective systemic response) as well as an ongoing need for multidisciplinary approaches to the care of these patients. Although the use of immunotherapy, specifically ICIs, has met with great success, these agents also are associated with a substantial rate of immune-related adverse events (irAEs). Because these irAEs often occur perioperatively, a critical need exists for surgeons, especially surgical oncologists, to have an appreciation for the toxicity of ICIs and targeted therapies as well as their management.

To highlight the presentation and management of irAEs, we provide two cases exemplifying such issues (Fig. 1).

FIG. 1.

Timeline of events for two patients with metastatic melanoma who developed IRAEs. Diagnosis, surgical intervention, pathology and irAE are denoted as indicated in the legend. Purple rectangle indicates immunotherapy regimen. Red triangle indicates steroid taper

Case 1

Pyrexia, Pneumonitis, and Acute Interstitial Nephritis in a Patient With Melanoma Who Underwent Combined Ipilimumab and Nivolumab (Fig. 1a).

A 51-year-old woman with a history of acral lentiginous melanoma in the left middle finger presented with metastatic disease in the left axilla. She underwent axillary lymph node dissection (ALND) and was treated with adjuvant nivolumab. This was well-tolerated, but recurrent disease developed in the axilla. Her case was discussed in a multidisciplinary setting, and the recommendation called for three cycles of combined nivolumab and ipilimumab followed by surgical resection. During her preoperative therapy, the woman experienced fevers, shortness of breath, and acute renal insufficiency requiring hospital admission. Infectious and cardiopulmonary evaluations were negative for acute pathology, and her kidney function improved with hydration. Pyrexia and pneumonitis secondary to ICIs were diagnosed. Further treatment was withheld. The woman recovered, underwent redo-ALND, and did well until postoperative day 2, when she experienced recurrent fevers and worsening renal function. She had no evidence of an infectious etiology. Renal biopsy demonstrated pathology consistent with acute interstitial nephritis secondary to ICI. Steroid therapy using high-dose intravenous (IV) hydrocortisone was initiated, with improvement of the woman’s kidney function and resolution of her fevers, which were ultimately diagnosed as ICI-associated pyrexia. She was discharged home receiving a steroid taper.

Case 2

Hypothyroidism, Dermatitis, Sarcoid-Like Lymphadenopathy, and Hypophysitis in a Patient Treated With Nivolubmab and Subsequently With a Combination of Ipilimumab and Nivolumab (Fig. 1b).

A 61-year-old man with a diagnosis of acral lentiginous melanoma in the great toe underwent amputation and sentinel node biopsy, which was positive. He elected to proceed with nodal basin surveillance and adjuvant nivolumab. After three cycles of nivolumab, his thyroid-stimulating hormone (TSH) was undetectable, and his free T4 was elevated, consistent with hyperthyroidism. He subsequently experienced hypothyroidism, and ICI-related destructive thyroiditis was diagnosed. He later experienced a rash on his upper extremities and chest, and a biopsy was consistent with ICI-associated dermatitis. Approximately 11 months later, the man had recurrence in the left groin, and his treatment regimen was switched to combination ipilimumab and nivolumab, with plans for subsequent surgical resection. Within 2 weeks after his first infusion, he experienced bulky palpable cervical lymphadenopathy. The biopsy was negative for melanoma and demonstrated non-necrotizing granulomas consistent with sarcoid-like reaction and attributed to the ICI. The man subsequently underwent left inguino-femoral, iliac, and obturator lymph node dissection. He was readmitted 3 weeks postoperatively with fevers (103°), chills, rigors, and groin cellulitis, with significant leukocytosis noted. Intravenous antibiotic therapy was initiated, and a fluid collection in the surgical bed was drained percutaneously. The man experienced acute renal insufficiency thought to be multifactorial (ICI, IV contrast, vancomycin use), which resolved with supportive care. However, the patient had ongoing abdominal pain, nausea, vomiting, and general malaise. Extensive evaluation for the etiology of his gastrointestinal symptoms was performed. His adrenocorticotropic hormone (ACTH) was undetectable, and ICI-associated hypophysitis was diagnosed. After initiation of IV steroids, his symptoms resolved almost immediately. He has since done well on maintenance oral steroids.

These recent cases highlight not only the frequency and complexity of irAEs, but also the variability in the presentation of these toxicities and the need for heightened awareness during the perioperative period. In this report, we discuss the basics of ICI-associated toxicity and approaches to detection and therapy. We further discuss the safety of surgical intervention and perioperative considerations for patients treated with ICIs.

IMMUNE CHECKPOINTS

A typical immune response against an infection (or cancer) requires the engagement of an antigenic peptide in the context of a major histocompatibility complex (MHC) on the antigen-presenting cell (APC) with a T cell receptor (TCR). However, a second signal is required to generate an effective immune response (e.g., the interaction between co-stimulatory molecules CD80/86 on the APC and CD28 on the T cell), inducing the expansion of antigen-specific T cells.

In normal physiology, additional molecules are expressed on the surface of T cells (immune checkpoint molecules such as CTLA-4 and PD-1) that function to downregulate the response after the threat is controlled. However, these immune checkpoints also can be exploited by tumors via primary or adaptive mechanisms.² In this setting, engagement of checkpoints causes a deleterious downregulation of the immune response, promoting immune evasion and allowing tumor growth, metastasis, or both.

Although CTLA-4 and PD-L¹³ are commonly lumped together as immune checkpoint molecules, they have distinct mechanisms of action in the priming and effector phases of an immune response (Fig. 2). The CTLA-4 molecules compete with CD28 for binding to co-stimulatory molecules CD80 and CD86 on the APC, preventing T cell activation. In addition, CTLA-4 is expressed on regulatory T cells (Tregs), which are responsible for the dampening of a generalized immune response but are especially important for immunomodulation within the gut. Meanwhile, PD-1 is expressed on circulating T cells and engages PD-L1, which is inducibly expressed by myeloid cells in areas of chronic inflammation and/or tumor cells, leading to negative feedback. Differences in the location and function of these molecules are critical to an understanding of their role in irAEs.

FIG. 2.

Immune activation and immune checkpoints. a A primed antigen-presenting cell (APC) can stimulate a naïve T-cell through the simultaneous engagement of antigenic peptide in the context of major histocompatibility complex (MHC) with a T-cell receptor (TCR) in addition to co-stimulatory molecules CD80/86 on the APC with CD28 on the T-cell. Upon T-cell activation, CTLA4 is upregulated and outcompetes CD28 for CD80/86 binding, activating signaling pathways that function to down-regulate the host immune response. Within the tumor, upregulation of PD-L1 on the tumor can engage PD-1 on the activated T-cell, further dampening anti-tumor immunity. CTLA-4 is also expressed on immunosuppressive regulatory T-cells (Tregs) which can generally dampen immune responses through the production of immunosuppressive cytokines. b Monoclonal antibodies blocking these immune checkpoints work both in the tumor-draining lymph node (anti-CTLA-4) and within the tumor immune microenvironment (anti-CTLA-4 and anti-PD-1/L1) enabling robust anti-tumor immune responses with decreased immunosuppressive cytokine production and increased recruitment and activity of activated T-cells leading to tumor cell death

RATES AND TYPES OF TOXICITIES

Because they represent nonspecific off-target inflammatory responses, irAEs are markedly diverse in spectrum and severity and quite unique compared with toxicities related to traditional cytotoxic chemotherapeutic or targeted agents.^1,3,4 The toxicities number more than 70, with dermatologic, gastrointestinal (GI), endocrine, pulmonary, and rheumatologic toxicities as the most common. More rare irAEs include cardiovascular, renal, hematologic, ophthalmologic, and neurologic conditions. Given that CTLA-4 and PD-1/PD-L1 have unique and non-redundant functions in the regulation of the immune system (Fig. 2), it is not surprising that toxicity profiles vary among these agents. Colitis and hypophysitis are commonly associated with anti-CTLA-4 therapies, whereas thyroiditis and pneumonitis are more often associated with anti-PD-1/PD-L1 therapies.⁴ Although combinatorial regimens using these two agents increase overall toxicity rates, their combined use does not result in a unique toxicity profile.⁴ Furthermore, the type of toxicity has not been linked to tumor type.^1,4

Most irAEs are mild and reversible and do not lead to significant disruptions in therapy. However, severe toxicities are not uncommon, and fatalities have been reported.^1,4,5 Some, toxicities, particularly the endocrine toxicities, are permanent. For purposes of consistency in reporting, toxicity severity typically is graded according to the Common Terminology Criteria for Adverse Events (CTCAE) guidelines, which assess the severity of toxicity based on clinical presentation. The overall rates for any grade of toxicity are approximately 70% to 80% with anti-CTLA-4 therapy, 70% to 80% with anti-PD-1/PD-L1 therapies, and up to 95% with combined anti-CTLA-4 and anti-PD-1/PD-L1 therapies.⁴ Generally, grades 1 and 2 toxicities are underreported. Thus, these numbers, although high, are likely an underestimate.⁶ The overall rates for grades 3 and 4 severe toxicity are 20% to 25% for anti-CTLA-4 therapies, 10% to 15% for anti-PD-1/PD-L1 therapies, and up to 50% for combined regimens.⁴ Meta-analysis data suggest the overall incidence of treatment-related death is 1.08% for anti-CTLA-4, 0.36 for anti-PD-1, 0.38% for PD-L1, and 1.23% for combination therapy. Death related to ICI therapy has been attributed to colitis, pneumonitis, diabetic ketoacidosis (DKA), encephalitis, and myocarditis.⁵

Due to the unique mechanism of action (targeting the host, not the tumor) and pharmacokinetics of ICIs, the timing of toxicity compared with traditional chemotherapy or targeted therapy is unique.^7,8 Toxicity typically occurs within weeks after the start of therapy but can happen even months to years after the therapy has stopped.^1,4 Furthermore, the timing of toxicity may differ with the organ system involved and the agent used. For example, whereas rash/dermatitis and diarrhea/colitis typically occur soon after initiation of therapy, endocrine toxicities such as hypophysitis and other gastrointestinal toxicities (e.g., hepatitis) occur somewhat later, and renal toxicity can occur much later (Fig. 3).⁹ Renal toxicity from anti-CTLA-4 therapy most often occurs 2 to 3 months after therapy, whereas renal toxicity resulting from anti-PD-1 therapy occurs 3 to 10 months after therapy (Fig. 3).^8,10 Toxicities associated with combinatorial regimens have a tendency to present earlier.^7,8

FIG. 3.

Timing of development of relatively common immune-related adverse events (irAEs). The timing of toxicity related to immune checkpoint inhibitors (ICIs) is quite unique compared to that of traditional chemotherapeutic or targeted agents, sometimes starting long after the therapy has started and occasionally after cessation of therapy. Further, it is quite variable depending on the organ system involved

The breadth of pathology and unique timing of irAEs complicate diagnosis. An astute clinician must suspect potential irAE in all patients treated with ICIs.

MECHANISMS OF TOXICITY TO IMMUNE CHECKPOINT BLOCKADE

Although these toxicities globally represent an off-target inflammatory response, to date, we do not fully understand their etiology. Often, irAEs mimic idiopathic autoimmune disease. Furthermore, autoimmune disease in humans is associated with single nucleotide polymorphism (SNPs) of immune checkpoint-related genes such as CTLA-4 and PD-1.³ However, idiopathic and provoked autoimmune phenomena have subtle differences in presentation, clinical course, and pathology. Both the similarities and differences between irAEs and idiopathic autoimmune disease likely will be key to elucidating the true mechanism of irAEs.³ Preclinical mouse models, including CTLA-4- and PD-1-deficient mice, also have provided some insight.^3,11,12 However, these particular models have profound immune pathology, and no good preclinical mouse models for irAEs have been developed to date.

Multiple theories regarding the development of irAEs have been proposed, and all may be partially contributory.^1,3 One causal theory relies on the concept of shared antigens. That is, the tumor expresses a neo-antigen recognized by the immune system, but that neo-antigen is similar enough to native protein expressed on host tissue to elicit a significant off-tumor effect. In patients with development of myocarditis, reactive T-cells have been identified in the diseased tissue and tumor that may exhibit cross-reactivity.¹³ It would be expected, however, that tumor type might skew the toxicity profile if shared antigens dominated, but this is not universally true.^3,4 Furthermore, some suggest that the use of ICIs simply unmasks subclinical autoimmune disease. This would be supported by increased rates of irAEs in patients with known mild or quiescent autoimmune disease or irAEs resulting from activities of preexisting antibodies (thyroid disorders).^1,14 However, not all patients with preexisting autoimmune disease experience flares when treated with ICIs.³ Finally, some believe this is secondary not to direct off-tumor but on-target effects. Notably, CTLA-4 and PD-1 are not uniquely expressed by T cells. For example, CTLA-4 is expressed on normal pituitary cells, potentially explaining the hypophysitis more frequently seen with anti-CTLA-4 therapies.^1,3,4

Some debate has focused on whether irAEs may be a marker of overall immune activity, and even a potential indicator of an active anti-tumor immune response. As such, some have proposed that the development of irAEs could serve as a positive prognostic indicator in patients.¹ Certain toxicities may hold particular predictive value. For example, vitiligo may be predictive of a good clinical response in patients treated with ICI for melanoma, in line with the shared-antigen mechanistic model. 1,15At this time, however, we conclude that the presence of toxicity does not guarantee a response, nor does a lack of toxicity mark a poor response.

SUSPECT, DETECT, AND REFER FOR TREATMENT

As our case studies illustrate, irAEs frequently occur in the perioperative setting, either before or after surgical intervention. The latter is perhaps related to altered immune activity in the setting of surgical intervention. The onus is on the surgeon to have a high degree of suspicion for potential toxicities in patients treated with ICI. Vague symptoms in these patients cannot and should not be dismissed because nonspecific ailments can be indicative of severe toxicity. The rheumatoloic toxicities and endocrinopathies are some of the most difficult to recognize given their relatively nonspecific presentation.⁴ Fatigue and poor energy, quite common for patients with metastatic disease, could represent hypophysitis and adrenal insufficiency, which are potentially severe irAEs.⁴ Furthermore, other toxicities are essentially asymptomatic.⁴ Renal and hepatic toxicity generally are detected only on routine labs. This coupled with their relatively late presentation makes them especially deceptive.^4,10

Thus, a comprehensive preoperative workup for immune-related toxicity is imperative (Table 2). A thorough history specifically targeted to potential irAEs with a specific focus on bowel habits and changes since initiation of ICI treatment may identify patients at risk for colitis.⁸ Complete blood work should include complete blood counts, electrolyte panels, liver function tests and thyroid function. Dermatologic manifestations often are the most common and earliest irAEs reported, and a complete skin exam should be performed. Patients with fatigue and poor energy should be evaluated for adrenal insufficiency and hypophititis with morning cortisol and ACTH levels. Renal insufficiency may require renal biopsy for confirmation of ICI-related nephritis.¹⁰

TABLE 2.

Preoperative evaluation and diagnostic test to consider for patients treated with checkpoint blockade undergoing surgery

History

Detailed questioning for autoimmune, infectious disease; for endocrine and organ-specific disease history

History of baseline bowel habit (frequency of bowel movements, usual stool consistency)

Blood tests

CBC

CMP

TSH

HbAlc

Free T4

Total CK

Liver function tests including fasting lipid profile

Dermatologic examination

Full skin and mucosal exam, taking note of the extent and type of lesions present

Pulmonary tests

Baseline oxygen saturation on room air and during ambulation

Cardiac tests

ECG

Troponin I or T

Additional screening tests recommended for patients with preexisting organ disease/at risk of organ-specific toxicity

Endocrine tests

8 am cortisol

8 am ACTH

Cardiac tests

Brain natriuretic peptide (BNP) or N-terminal pro B-type natriuretic peptide (NT pro-BNP)

Pulmonary tests

PFTsa

6MWTa

CBC complete blood count, CMP complete metabolic panel, TSH thyroid-stimulating hormone, HbA1c glycosylated hemoglobin, T4 thyroxine, CK creatine kinase, ECG electrocardiogram, ACTH adrenocorticotropic hormone, PFTs pulmonary function tests, 6MWT 6-min walk test

These tests become very relevant if patients experience immune-related adverse events (irAEs) and require immunosuppressive treatment such as steroids and/or anti-tumor necrosis factor alpha (TNFα) treatment. Adapted from Puzanov et al.⁸

^aGiven the rarity of pulmonary toxicity, pretreatment PFTs and 6MWTs should be considered for patients with preexisting lung disease (e.g., chronic obstructive pulmonary disease, interstitial lung disease, sarcoidosis, pulmonary fibrosis), but may not be feasible for all patients

TREATMENT

Various guidelines for the treatment of irAEs have been published by American Society of Clinical Oncology (ASCO),¹⁶ National Comprehensive Cancer Network (NCCN),¹⁷ European Society of Medical Oncology (ESMO),¹⁸ Society for Immunotherapy of Cancer (SITC)⁸ in a disease site-specific manner and should be followed. Despite these guidelines, treatment of irAEs currently varies significantly.¹⁹

As a general rule, grade 1 irAEs require only symptomatic management and close monitoring. For irAEs classified as grade 2 or higher, systemic steroids should be started and tapered gradually as possible. Grades 3 and 4 toxicity require high-dose systemic corticosteroids, alternative biologic immunosuppressive agents, or both. Biopsy of the involved tissue or organ generally is unnecessary but may be beneficial in the setting of grade 3 or 4 toxicity or when there is diagnostic doubt.

When a biopsy is performed, a pathologist must be in tune to the nature of the question at hand and look for evidence of autoimmune phenomena. Unfortunately, for many of the endocrine toxicities, no treatment exists that reverses the underlying organ dysfunction. Rather, the loss of endocrine function is most often permanent, requiring lifelong hormone replacement therapy which can have profound effects on quality of life (e.g., hypothyroidism and type 1 diabetes mellitus), but patients tend to do well overall.⁴

The use of corticosteroids to treat irAEs does not seem to have an impact on the overall survival of these patients, suggesting that they do not dampen anti-tumor immune activity.^20,21 However, these findings do not account for the fact that these patients might be expected to have even higher efficacy and better outcomes given the presence of irAE.¹ Furthermore, some data suggest that corticosteroids do have an impact on expression of inhibitory molecules on immune cells.^3,22 Although the effects of systemic immunosuppression on ICI efficacy and anti-tumor immune activity remain debated, clearly significant side effects are seen with corticosteroids, which are occasionally severe and can affect the perioperative course. These side effects include hyperglycemia and wound-healing difficulties, generalized immunosuppression and susceptibility to infection/sepsis, and mood alteration.²³

Targeted therapies, when possible, are beneficial to avoid the numerous side effects of systemic corticosteroids. Disease-specific specialists may be able to guide a more targeted treatment of irAEs. For example, dermatologists may, based on the appearance of the rash, be able to treat with alternative targeted or topical therapies rather than with systemic corticosteroids.²⁴ For ICI-associated colitis, there is hope that we can move toward targeted immunosuppressive agents such as tumor necrosis factor (TNF)-blocking agents (infliximab), which currently are reserved for patients with disease refractory to steroids because these agents have been associated with shorter duration of symptoms, less requirement for hospitalization, and less dependence on steroids.²⁵

The optimal length of corticosteroid therapy is unknown,¹⁹ and the question whether ICIs can be re-initiated after development of irAEs remains unsettled.¹ The presence of most irAEs is not an absolute contraindication to resuming therapy. Retrospective data demonstrate that when these therapies have been restarted for patients with irAE, more than half of the patients experienced no further toxicity. Approximately half of those who do experienced recurrent toxicity, although often of lesser severity, with the remaining half experiencing new toxicity.^1,26–28

Overall, the current recommendations can be summarized as follows.^8,17,18 For grade 1 toxicities, no cessation or break in therapy generally is required. For grade 2 toxicities, holding therapy should be considered until symptoms are reduced to grade 1 severity or less and the patient has been weaned to less than 10 mg of oral prednisone equivalent. For grade 3 toxicity, ICI therapy should definitely be stopped, but resuming can be considered, whereas for patients presenting with grade 4 toxicity, recurring grade 3 toxicities, or persistent treatment-refractory grade 2 toxicity, ICI should definitively be stopped. Endocrinopathies are an exception to this rule. For endocrinopathies, ICI therapy can continue throughout treatment for grades 1 and 2 toxicities and may be temporarily held for grades 3 and 4 toxicities.

One toxicity that warrants additional discussion for the surgeon is ICI-associated colitis, which represents an especially virulent form of inflammatory bowel disease that histologically, pathologically, and clinically resembles Crohn’s disease and/or ulcerative colitis.^29,30 As with all other toxicities, early recognition and careful monitoring can, for the most part, prevent poor outcomes. Infectious causes (e.g., Clostridium difficile, Salmonella, cytomegalovirus)²⁹ must be ruled out, and endoscopy should be performed early after presentation. Early endoscopy (preferably full colonoscopy given the sometimes spotty nature of the disease and frequent involvement of the right colon and terminal ileum) is recommended given the lack of correlation between clinical presentation and severity of pathology. The concern for potential perforation should not preclude endoscopic evaluation, and biopsies should be taken to guide therapy.^29–31 Patients presenting with severe ICI-associated colitis require serial abdominal exams and close monitoring. Severe worsening pain or new fevers could indicate perforation/abscess formation and necessitate cross-sectional imaging. Surgical intervention should be a last resort, performed only for perforation with gross contamination, abscesses not amenable to percutaneous drainage, intractable bleeding or toxic megacolon, or unrelenting severe disease.²⁹

In all cases, early intervention for irAEs is associated with improved outcomes.^1,4 Ultimately, the development of specialized immune-oncology multidisciplinary teams for the timely identification and thoughtful management of irAEs will potentially streamline toxicity management and minimize long-term complications. Thus, perioperative complications concerning for ICI-related disease should prompt early specialist involvement.

TIMING AND SAFETY OF SURGERY

Data evaluating the optimal timing and the safety of surgery after treatment with ICIs are limited (Table 3). However, several retrospective studies have evaluated surgical resection of metastatic disease,^32–34 and data evaluating ICI in the neoadjuvant setting are emerging.^35–38 For patients who have received cytotoxic chemotherapy, elective surgical resection typically is performed 4 to 6 weeks after cessation of therapy to allow for a washout period and bone marrow recovery. Data from patients treated with ICI in the metastatic setting demonstrate that earlier surgical resection is probably safe and reasonable.^32–34 The median time from ICI to surgery ranged from 16 to 75 days, with a significant number of patients undergoing surgery within 2 weeks after ICI administration. Complications were similar to those seen in historical series of these high-risk patient groups. Notably, no anastomotic leaks occurred for patients who underwent bowel resection,³³ and wound infection rates were similarly acceptable for this high-risk population.³⁴ However, complications not previously seen after cytotoxic chemotherapy may exist. For example, in a study by Bott et al.³² evaluating patients undergoing lung resection after ICI, significant pneumonitis in the contralateral lung of a patient who underwent video-assisted thoracoscopic surgery (VATS) wedge resection occurred 2 weeks after surgery. In addition, surgeons in this series noted the presence of significant fibrotic adhesions, sometimes complicating dissection. Although not universal, these data highlight the potential of new complications not previously identified in surgical series.

Table 3.

Studies detailing the safety of surgery after treatment with immune checkpoint inhibitors (ICIs)

Author	No. of patients	Disease	Presurgical therapy	Median time from ICIs to surgery	Procedures	Perioperative complications
Surgery for metastatic disease
Gyorki et al.34	23 Patients, 34 operations	Melanoma	Anti-CTLA-4 (100%)	25 days (range, 3–74); 15% within 7 days	Subcutaneous/lymph node (48%) Laparotomy (35%) SB resection (17%) Craniotomy (15%)	Gr 1/2: wound infection (22%) Gr 3 (0%)
Bott et al.32	19 Patients 22 operations	Lung cancer (47%), melanoma (37%), other (16%)	Anti-PD-1 (64%), anti-PD- L1 (16%) anti-CTLA-4 (16%), anti-PD-l/CTLA-4 (5%)	75 days (range, 7–183)	Lobectomy (36%) Bilobectomy (9%) Pneumonectomy (5%) Wedge (50%)	30- & 90-day mortality (0%) 32% morbidity Gr 4: pneumonitis (5%) *contralateral lung Gr 1/2: air leak (9%) Gr 2: arrhythmia (5%) Gr 2: pneumonia (5%)
Elias et al.33	17 Patients; 22 operations	Melanoma (82%), RCC (12%), Urothelial carcinoma(6%)	Anti-PD-1 (45%), anti-PDL- 1 (9%) anti-CTLA4 (23%), anti-PD-l/CTLA-4 (9%)	16 days (1–32) 36% within 7 days 23% between 8 and 14 days	Cutaneous/lymph node: 50%, Bowel resection: 22% Abdominal wall resection: 14%, Other abdominal: 14%	Mortality: 4% (VFib in patient with h/o CAD) 18% Grl; 18% grade 2-wound infections and anemia
Neoadjuvant therapy
Carthon et al.37	6 Patients	pTl-T2, high grade urothelial carcinoma	2 Preop doses anti-CTLA4 3 mg/kg: 50% 10 mg/kg: 50%	4 weeks	Radical cystectomy (83%) *1 patient did not have surgery due to presence of metastatic disease	3 mg/kg: no surgery delay Grl/2 Rash: 67%, Diarrhea: 33%, Uveitis: 17% Gr3 Uveitis: 17% 10 mg/kg: 33% surgery delay Grl/2 Rash: 50%, Orchitis: 17%, Diarrhea: 50% Gr3 diarrhea: 33% transaminitis: 17%
Amaria et al.35	23 Patients	Stage III or oligometastatic Stage IV Melanoma	4 Preop doses anti-PD-1# (52%), OR 3 doses anti-PD-l/CTLA-4* (48%) 0% dose delay in anti-PD-1 monotherapy; 64% dose delay in anti-PD-1/CTLA-4	2 weeks	Not reported	Anti-PD-1: Gr3: 8%-tumor pain Anti-PD-1/CTLA4: Gr3: 84% transaminitis: 27% colitis: 18% hyperthyroid: 9% pneumonia: 18% arthralgias: 9% myositis: 9% electrolyte abnl: 9%
Blank et al.36	20 Patients	Stage III melanoma	2 Pre-op doses & 2 post-op doses anti-PD- l/CTLA-4* (50%) OR 4 post-op doses anti-PD-1/CTLA-4* (50%) only 10% received all 4 doses	Week 6	Lymph node dissection (100%)	No difference in post-op complications Wound infection: 40% Seroma: 90% No difference in irAE between neoadjuvant and adjuvant All grades: 100% adrenal insuffic.: 25% Gr3: 90% transaminitis: 65% colitis: 30% hyperthyroid: 5% pneumonia: 18% hypophysitis: 10% rash: 25% electrolyte abnl: 30%
Forde et al.38	21 Patients	Stage MIIA NSCLC	2 Doses anti-PD-1 91% completed 2 doses	18 days (11–29)	95% Underwent resection	No treatment-related surgery delays Morbidity: 32% Gr3: Pneumonia: 5% Grl/2: Fever: 5% diarrhea: 10% transaminitis: 5% anorexia: 14%

CTLA-4, ipilimumab (Yervoy); PD-1, nivolumab (Opdivo), pembrolizumab (Keytruda), cemiplimab (Libtayo); PD-L1, atezolizumab (Tecentriq), avelumab (Bavencio), durvalumab (Imfinzi)

^#nivolumab: 3 mg/kg;

^*nivolumab 1 mg/kg + ipilimumab 3 mg/kg

Gr grade, RCC renal cell carcinoma, NSCLC non-small cell lung cancer

The use of ICI in the neoadjuvant setting is a rapidly expanding field, and data evaluating its safety are starting to emerge (Table 3). The first study to evaluate the safety and efficacy of neoadjuvant ICI treated six patients with resectable urothelial cancer using two doses of neoadjuvant anti-CTLA-4.³⁷ At the 3-mg/kg dose, no surgical delays occurred, and the toxicity profile was consistent with that of previous studies evaluating anti-CTLA-4. Not surprisingly, at the higher dose (10 mg/kg), 33% of the patients experienced a surgical delay, and significantly more irAEs occurred. Simultaneous studies from MD Anderson and the Netherlands Cancer Institute evaluated two different perioperative ICI strategies for patients with advanced melanoma.^35,36 Both studies demonstrated significant irAEs with combined therapy (83–90% grade 3), and the majority of the patients did not complete all the doses of therapy. Importantly, Blank et al.³⁶ noted that neither the postoperative complications nor the irAEs differed between the neoadjuvant and adjuvant therapy groups, demonstrating that the toxicity observed likely was related to ICI itself and not the perioperative dosing schema. Data evaluating neoadjuvant anti-PD-1 monotherapy demonstrate a more favorable safety profile. In the neoadjuvant melanoma study by Amaria et al.,³⁵ no dose delays occurred, and grade 3 irAEs occurred at a rate of only 8% in the anti-PD-1 group. In a recent study of 21 patients with stages 1 to 3A NSCLC treated with neoadjuvant anti-PD-1 therapy, Forde et al.³⁸ demonstrated no surgical delays and low grade 3 toxicity rates.

The development of irAEs preoperatively also has surgical implications given current treatment strategies that include high-dose steroids and sometimes quite long tapers and/or biologic immunosuppressive agents.^8,16–18 Although little data are available at this point, we can extrapolate from other diseases for which surgical intervention occurs during periods of immunosuppression. Although the data that exist for the complications associated with steroid use and/or TNF inhibitors are complicated by significant patient heterogeneity in terms of indication for use, most studies do indicate increased risk of surgical-site infections and overall complication rates.^39–41 Furthermore, the risk for stress-induced adrenal insufficiency is increased in the perioperative period for any patient with a history of steroid use.⁴²

More than 2000 different immune-oncology regimens are being tested in preclinical and early-phase clinical trials.⁴³ Specific agents and/or combinatorial strategies currently being tested may warrant particular consideration when used perioperatively. In particular, combinations involving both radiation (and chemoradiation) and ICI, including their use in neoadjuvant strategies, are continuing to increase, and to date, we do not understand how this might affect the overall immune response, particularly the risk for surgical complications.⁴⁴ In the coming years, we expect results from numerous neoadjuvant ICI studies investigating a wide spectrum of malignancies that will further define the safety and optimal timing of surgery in the neoadjuvant setting with ICI therapy alone and in combination with cytotoxic chemotherapy and other treatment methods.

POTENTIAL BIOMARKERS OF TOXICITY AND FUTURE DIRECTIONS

As use of the aforementioned powerful therapies rises, clinicians are gaining further understanding of the true incidence, spectrum, and timing of irAEs, and detection of irAEs and treatment regimens have improved through the use of standardized protocols and multidisciplinary specialist team approaches. However, significant opportunities for improvement remain.

Meanwhile, scientists are working to understand toxicity further on a mechanistic level, which to date has been limited by a lack of good preclinical models for toxicity as well as prospective clinical trials designed with a focus on toxicity. Such advancements would enable the development of predictive biomarkers, enabling practitioners to identify patients most likely to experience toxicity. Furthermore, understanding the immunologic pathways responsible for individual toxicities will promote the development of more targeted therapies that will not generally suppress immune function, obviating the toxicity while enabling ongoing anti-tumor immune activities. In addition, the separation of toxicity and efficacy suggests that novel agents could be developed that would have efficacy in anti-tumor immune activity without irAEs.

The use of immunotherapeutic agents for cancer will no doubt continue to rise rapidly, and these agents will be used increasingly for patients undergoing operative intervention. All clinicians who treat patients with cancer, especially surgeons, must be cognizant of the potential side effects, not only of the ICIs, but also of the novel immunotherapeutic agents in development.