Severe Immune-Related Adverse Events Associated with Durable Antitumor Responses Following Immune Checkpoint Inhibitor Therapy: Two Case Reports

Abstract

Background

Immune checkpoint inhibitors (ICIs) have achieved remarkable success in treating advanced cancers, yet the accompanying immune-related adverse events (irAEs) pose a major challenge to their clinical application. Growing evidence suggests an association between irAEs and improved clinical outcomes; however, the heterogeneity in this relationship warrants further exploration.

Case Presentation

This article reports two cases of patients with advanced cancer who achieved long-term disease control after experiencing severe irAEs. Case one involved a patient with heavily pretreated Luminal B1 advanced breast cancer who developed Grade 3 immune-related arrhythmia and heart failure following treatment with Toripalimab combined with chemotherapy. Case two involved a patient with ROS1-fusion advanced lung adenocarcinoma who, after developing resistance to Crizotinib, received chemotherapy and Tislelizumab, leading to Grade 3 immune-related hypopituitarism, adrenal insufficiency, and cutaneous toxicity. Both patients were managed with corticosteroids with or without immunosuppressants and discontinued all antitumor therapies. Despite this, they have achieved a partial response (PR) with a duration of response (DoR) of 22 and 37 months, respectively, and both responses are ongoing at the time of this report.

Conclusion

In this series, two patients with advanced solid tumors achieved durable tumor responses after developing severe irAEs and discontinuing treatment, suggesting that severe irAEs may herald the establishment of long-term immune memory. However, the association between irAEs and efficacy is not straightforward, as the type and severity of irAEs and the use of immunosuppressive intervention all influence the ultimate clinical benefit. In the absence of reliable predictive biomarkers, clinical decisions should not equate irAEs directly with efficacy markers; rather, the benefits and risks of immune activation should be carefully weighed.

Introduction

Immune checkpoint inhibitors (ICIs), by blocking negative regulatory signals on T cells, have revolutionized the treatment paradigm for advanced cancers by reinvigorating antitumor immunity.^1,2 However, the overactivation of the immune system acts as a “double-edged sword”, potentially harming the host while attacking the tumor, leading to immune-related adverse events (irAEs).³ In the clinical practice of managing these toxicities, an intriguing observation has emerged: patient cohorts experiencing irAEs often demonstrate superior tumor shrinkage rates and longer survival.^4,5 This phenomenon raises a central scientific question: are irAEs merely a side effect of immune activation, or are they themselves a potent predictive signal for efficacy?

Although multiple retrospective studies and meta-analyses support a positive correlation between irAEs and improved outcomes, the clinical utility of this association remains substantially uncertain due to several key gaps: the predictive value varies markedly by affected organ (eg, skin/endocrine toxicities showing stronger links than myocarditis or pneumonitis);^6–8 evidence in certain tumor subtypes, particularly heavily pretreated hormone-receptor-positive breast cancer, remains scarce.⁹ However, a recent study in early-stage triple-negative breast cancer (TNBC) treated with neoadjuvant pembrolizumab showed that irAEs were significantly associated with pathological complete response (pCR) (35.4% vs 19.7%, p=0.01), supporting irAEs as a potential marker of immune activation;¹⁰ and it is unclear whether greater severity linearly correlates with deeper efficacy or merely reflects excessive, potentially harmful immune activation. Furthermore, whether high-dose corticosteroids or immunosuppressants used to rescue severe irAEs inadvertently abrogate long-term antitumor immune memory is an unresolved concern.⁸ In real-world practice, the occurrence of Grade 3–4 irAEs often creates a clinical dilemma—whether to continue, pause, or permanently discontinue ICIs—making cases with durable responses after treatment cessation particularly instructive for unraveling this complex association.

Notably, cases experiencing severe irAEs (Grades 3–4) who achieve long-term disease control after management provide an excellent model for investigating the intrinsic link between efficacy and toxicity. This article describes in detail two such typical cases—one with breast cancer and one with lung cancer—who achieved long-term remissions after developing immune-related adverse events and discontinuing all antitumor therapy. Based on these cases, we provide an in-depth discussion of potential immunological mechanisms and the incomplete concordance between autoimmunity and antitumor efficacy, and we explore the implications of these observations for the use of biomarkers in managing severe irAEs, with the goal of offering new insights and generating testable hypotheses in this field.

Case Reports

Case 1

Advanced Breast Cancer with Long-term Remission after Immune-related Arrhythmia and Heart Failure (The diagnostic and treatment process is shown in Figure 1).

Figure 1.

Treatment timeline of Case 1 (breast cancer).

Patient Profile

The patient is a 65-year-old female with Luminal B (hormone receptor-positive / human epidermal growth factor receptor 2-negative) breast cancer. In April 2014, she self-discovered a right breast mass, and breast ultrasound suggested possible right breast cancer with right axillary lymph node metastasis. On May 14, 2014, she underwent modified radical mastectomy for right breast cancer. Postoperative pathology showed invasive ductal carcinoma, histological grade II, with cancer metastasis in 2 out of 16 right axillary lymph nodes; estrogen receptor (ER) 3+, progesterone receptor (PR) 3+, CerbB2 0, Ki67 50%. After surgery, she received routine chemotherapy with the “Epirubicin + Cyclophosphamide → Docetaxel” (EC-T) regimen (cumulative epirubicin dose: 560 mg/m²). She subsequently received adjuvant radiotherapy and has been on adjuvant endocrine therapy with oral letrozole. No other cardiotoxic agents were used throughout her treatment course. In September 2020, follow-up upper abdominal MRI first revealed liver metastases, and subsequently lung and bone metastases also appeared. To dynamically monitor the molecular subtype, two liver metastasis core needle biopsies were performed. On October 21, 2020, the pathology of the liver metastasis biopsy showed: metastatic breast carcinoma, ER 60%, PR 0, CerbB2 1+, Ki67 35%. On October 31, 2023, the pathology of the liver metastasis biopsy showed: metastatic breast carcinoma, ER 60%, PR 0, CerbB2 1+, Ki67 70%. The patient received multiple lines of palliative therapy, including regimens with albumin-bound paclitaxel, docetaxel, capecitabine, eribulin, utidelone, palbociclib, abemaciclib, dalpiciclib, exemestane, fulvestrant, and apatinib, in addition to radiotherapy to the liver metastases. The main adverse reactions were neutropenia, nausea and vomiting, and abnormal liver function (Common Terminology Criteria for Adverse Events (CTCAE) Grade 1–2). On April 13, 2024, follow-up imaging showed enlargement and an increase in the number of liver metastases. Although the pathology from the liver core needle biopsy on October 31, 2023, was not triple-negative breast cancer, the patient requested treatment with immune checkpoint inhibitors. Her past medical history includes type 2 diabetes mellitus. In 2000, she underwent radiofrequency ablation for Wolff-Parkinson-White syndrome with no subsequent discomfort.

ICI Treatment Regimen

From April 14, 2024, the patient received a 21-day cycle regimen of Toripalimab 240 mg IV on Day 1 combined with Gemcitabine 1.2 g IV on Days 1 and 8.

irAEs Onset and Management

The patient received the first day of cycle 2 on May 5, 2024. Beginning May 7, 2024, she developed fever (max temp 38.3°C), palpitations, chest tightness, exertional dyspnea, orthopnea, and facial and bilateral lower limb edema. Physical examination revealed heart rate 120–150 bpm, irregular rhythm, varying heart sound intensity, and bilateral pulmonary rales. A baseline ECG on February 18, 2024 (prior to ICI initiation), was normal. On May 4, 2024 (before the second cycle of ICI), an ECG showed sinus rhythm with occasional ventricular premature beats. On May 7, 2024, an ECG indicated sinus tachycardia, frequent atrial premature beats (in salvos), and short runs of atrial tachycardia. Further 24-hour Holter monitoring revealed paroxysmal atrial fibrillation with rapid ventricular rate and frequent premature ventricular contractions. Bedside echocardiogram on May 7, 2025 showed a left ventricular ejection fraction (LVEF) of 60%; a repeat echocardiogram on May 9, 2025 revealed left atrial enlargement, moderate mitral regurgitation, and an LVEF of 64.8%. Laboratory tests on May 7, 2025: White blood cells 3.19×10⁹/L, neutrophils 85.6%, hemoglobin 76.0 g/L, C-reactive protein 16.3 mg/L; Brain natriuretic peptide (BNP) was 598.40 pg/mL (compared with 150.9 pg/mL on May 4, 2024, with a peak of 716.41 pg/mL after the cardiac event); high-sensitivity cardiac troponin I (hs-cTnI) was 0.007 µg/L (peak 0.012 µg/L after the cardiac event). D-dimer was 1.42 mg/L on May 4, 2024, and peaked at 2.27 mg/L during the event, but pulmonary embolism was excluded by clinical assessment and subsequent imaging. Other parameters including myocardial enzyme spectrum were essentially normal. The diagnosis was heart failure and arrhythmia (CTCAE Grade 3). Based on the diagnostic criteria from the Clinical Diagnosis and Treatment Recommendations for Immune Checkpoint Inhibitor-Associated Myocarditis (2026 Edition),¹¹ although the troponin level remained within the normal range and endomyocardial biopsy was not performed, the following elements nevertheless support a clinical consideration of association with ICI: (1) new-onset cardiac symptoms (palpitations, dyspnea, edema) and ECG abnormalities (new atrial fibrillation, frequent premature beats) occurring 23 days after ICI initiation, fitting the typical 28–65 day window; (2) significant elevation of BNP from baseline; (3) exclusion of acute coronary syndrome (no ischemic ECG changes or typical chest pain), pulmonary embolism, and infectious myocarditis; and (4) despite prior anthracycline exposure, baseline LVEF was preserved and the acute arrhythmia and heart failure occurred in direct temporal association with ICI administration, arguing against delayed anthracycline cardiotoxicity. Therefore, after excluding alternative etiologies, the cardiac event was attributed to ICI-related myocarditis/arrhythmia. Management included digoxin for inotropy, furosemide for diuresis, bisoprolol, amiodarone, and ivabradine for heart rate control, recombinant human BNP to reduce cardiac load, dapagliflozin for glycemic control and cardioprotection, anticoagulation, anti-infection therapy, and intravenous methylprednisolone 40 mg daily for 3 days. By May 16, 2024, the symptoms resolved, and a repeat ECG normalized, indicating correction of the arrhythmia and heart failure. The patient declined any further antitumor therapy following this event.

Tumor Response and Follow-Up

Imaging reviews every 3 months showed shrinkage of the liver metastases, assessed as PR per RECIST 1.1 criteria (Figure 2). To date, the progression-free survival (PFS) on the ICI combination regimen has reached 22 months (and 21 months since cessation of all antitumor therapy).

Figure 2.

Abdominal CT of Case 1: (a) CT on April 3, 2024, shows progression of the right liver metastasis (pre-ICI) (white arrow); (b) CT on June 26, 2024, shows marked shrinkage of the right liver metastasis with resolution of the solid component after 2 cycles of ICI plus chemotherapy (yellow arrow); (c) CT on February 12, 2026, shows the right liver metastasis stable in its shrunken state, resembling a cyst (green arrow).

Case 2

Advanced Lung Cancer with Long-term Remission after Multiple irAEs (The diagnostic and treatment process is shown in Figure 3).

Figure 3.

Treatment timeline of Case 2 (ROS1 fusion-positive lung adenocarcinoma).

Patient Profile

The patient is a 65-year-old male who presented with chest pain in April 2022. Chest CT at that time revealed a right hilar mass, suggestive of central-type lung cancer, with metastasis and confluence of lymph nodes in the right hilum, mediastinum, and right axilla. On May 6, 2022, a biopsy of the right hilar mass was performed, and pathology confirmed lung adenocarcinoma with Programmed Death-Ligand 1 (PD-L1) expression of 20% (Tumor Proportion Score, TPS); PCR testing demonstrated ROS1 gene fusion mutation, while EGFR, ALK, RET, MET and other genes were wild-type. From May 2022 to May 2023, the patient received targeted therapy with oral Crizotinib 250 mg twice daily, with no significant adverse reactions. In May 2023, he presented with recurrent chest tightness and dyspnea for one month. Repeat chest CT revealed enlargement of the right thoracic inlet mass with compression and narrowing of the superior vena cava. Blood NGS testing indicated wild-type EGFR/ALK/ROS1 genes.

ICI Treatment Regimen

From May 19, 2023, the patient received a combination regimen consisting of Tislelizumab 200 mg, Pemetrexed 800 mg, and Carboplatin 400 mg, all administered intravenously on day 1 of each 21-day cycle, for a total of 6 cycles. From October 18, 2023, the patient received Tislelizumab 200 mg on day 1 of each 21‑day cycle as maintenance therapy for 4 cycles. Treatment was interrupted for approximately 2 months due to irAEs, which improved after management. Given the PD-L1 20% expression, imaging-assessed tumor response, and perceived immune benefit, the regimen was switched to “Sintilimab 200 mg on day 1 of a 21-day cycle”, and the patient received one cycle of this treatment on March 8, 2024. Antitumor therapy was discontinued thereafter due to worsening irAEs.

irAEs Onset and Management

From December 2023, the patient developed generalized rash, pruritus, and worsening bilateral lower limb edema. Antiallergic treatment provided some relief. On January 12, 2024, “Tislelizumab” was continued, after which the pruritus worsened, accompanied by depigmentation (vitiligo-like changes). This was diagnosed as immune-related skin toxicity (CTCAE Grade 3). “Tislelizumab” was stopped, and management with Methylprednisolone and antiallergics led to improvement. Intravenous methylprednisolone was started at 40 mg per day and gradually tapered to withdrawal over a period of 4 weeks. Considering the immune benefit, treatment was changed to “Sintilimab” on March 8, 2024. From May 2024, the symptoms worsened, and new symptoms emerged, including extreme fatigue, dizziness, walking difficulty, poor appetite, hyperpigmentation on both upper limbs, and hoarseness, with recurrent episodes leading to multiple hospitalizations. During this period, cortisol dropped to a minimum of 0.63 µg/dl, and adrenocorticotropic hormone (ACTH) to <1 pg/mL. Renal function and thyroid hormones were normal. The diagnosis was immune-related hypopituitarism and adrenal insufficiency (CTCAE Grade 3). Management with Methylprednisolone, Mycophenolate Mofetil, and antiallergics improved the symptoms. Methylprednisolone was administered intravenously at a dose of 2 mg/kg, and then tapered gradually to a maintenance dose over approximately 4 weeks. Currently, the patient still experiences occasional mild dizziness, fatigue, and bilateral lower limb edema, and self-administers Methylprednisolone tablets 8 mg daily intermittently for replacement therapy. Cortisol levels fluctuate, peaking at 1.02 µg/dl; measured cortisol on September 22, 2025, was <0.5 µg/dl.

Tumor Response and Follow-Up

Imaging reviews approximately every 3 months showed shrinkage of the right thoracic inlet metastasis, assessed as partial response (PR) per RECIST 1.1 criteria (Figure 4). To date, PFS on the ICI combination regimen has reached 37 months (and 27 months since cessation of all antitumor therapy).

Figure 4.

Chest CT of Case 2: (a) CT on May 4, 2023, shows a new metastasis in the right thoracic inlet (pre-ICI) (white arrow); (b) CT on May 8, 2024, shows near-complete resolution of the right thoracic inlet metastasis after multiple cycles of chemotherapy and ICIs (yellow arrow); (c) CT on June 2, 2026, shows the right thoracic inlet metastasis maintained in complete radiological remission (green arrow).

Discussion

The two patients reported here both developed severe irAEs following ICI therapy. Notably, the second case presented with multi-system irAEs. Fortunately, through timely recognition and intervention, the symptoms in both patients were effectively controlled. To date, both have achieved prolonged tumor remission, with the potential for further extension, despite having discontinued all antitumor therapy for 21 and 27 months, respectively. This clinical phenomenon raises the question: does the occurrence of severe irAEs predict superior and durable efficacy from ICIs in such patients? And is the degree of clinical benefit correlated with the severity of irAEs?

The “Shared Immunological Basis” Hypothesis for Efficacy and Toxicity

The “shared immunological basis” hypothesis posits that ICIs simultaneously release the brakes on both antitumor immunity and autoimmunity. This leads to the activation of cross-reactive T cell clones that, within a shared systemic inflammatory milieu, attack common antigens (including tumor antigens and self-antigens), manifesting clinically as tumor regression (efficacy) and normal tissue damage (irAEs) in the same patient.^12,13 In other words, the observed antitumor efficacy and irAEs are essentially two different clinical manifestations of the same activated immune system.

When the immune system attacks normal tissues expressing these shared antigens (eg, myocardium, pituitary, or melanocytes), it implies that highly effective T cells capable of recognizing and attacking the tumor have been activated on a large scale. For instance, the myocardial injury observed in Case 1, though critical, is often associated with high-titer autoantibodies and extensive infiltration of cytotoxic T lymphocytes,^14,15 potentially reflecting an unusually intense and broad immune response that drove the sustained regression of liver metastases. In Case 2, we speculate that T cell clones attacking the pituitary (possibly targeting antigens such as the transcription factor PIT-1) and melanocytes (possibly targeting shared melanoma-associated antigens like MART-1 or GP100) significantly overlap with those attacking the lung cancer cells.^16–18 These “imprinted” T cell clones may differentiate into long-lived memory T cells after ICI withdrawal, establishing sustained immune surveillance against the tumor, thereby maintaining disease stability even without continuous medication. Severe irAEs might mark a threshold of immune system activation beyond which the body can establish a powerful and durable antitumor immune memory, achieving a state akin to “functional cure”.

Multiple studies indicate that patients experiencing irAEs often have superior objective response rates, progression-free survival, and overall survival compared to those without irAEs.^19–21 In malignancies such as melanoma, non-small cell lung cancer, and renal cell carcinoma, patients developing irAEs including vitiligo, dermatitis, colitis, and endocrine disorders are often observed to have more significant survival benefits.^7,22,23

A Dialectical View on the Efficacy-irAEs Correlation

Despite considerable clinical evidence supporting a link between irAEs and better outcomes, the existence of a clear positive correlation must be viewed cautiously.

First, irAE type determines correlation strength. The correlation between different types of irAEs and efficacy varies significantly. irAEs strongly correlated with efficacy include vitiligo and endocrine irAEs (eg, hypophysitis, thyroiditis).^7,8 In contrast, irAEs affecting vital organs such as pneumonitis, myocarditis, and hepatitis are potentially fatal. Immune-mediated myocarditis, for instance, carries a high mortality rate. The occurrence of such severe toxicities may therefore preclude the observation of any meaningful antitumor efficacy; thus, their correlation with efficacy remains unclear.^14,24

Second, there may exist a “clinical benefit inflection point” with respect to severity. Mild to moderate immune activation may fall within an ideal “therapeutic window” effectively killing tumors while remaining manageable. However, when the reaction becomes excessive, leading to Grade 3–4 severe irAEs, it results in treatment interruption, the need for high-dose immunosuppressants, and potentially organ failure or death. To control severe irAEs, the use of high-dose corticosteroids or even potent immunosuppressants such as infliximab or mycophenolate mofetil may inadvertently suppress the nascent antitumor immune response.²⁵ More importantly, severe irAEs in critical organs directly threaten life, rendering discussions about efficacy secondary.²⁶

Third, a “decoupling” phenomenon between antitumor immunity and autoimmunity is frequently observed. Clinical practice commonly sees patients with significant irAEs but poor efficacy, and vice versa. This indicates that while antitumor immunity and autoimmunity share overlapping mechanisms, they are also distinct. They may be mediated by different T cell clones or differentially regulated by factors such as the local microenvironment and antigen expression levels.^27–29 Therefore, a potent autoimmune reaction (irAEs) does not necessarily equate to a potent antitumor response, and the absence of clinically apparent irAEs does not preclude robust antitumor activity.

Clinical Implications and Limitations

The two cases presented here add to the growing body of evidence suggesting that severe irAEs may coincide with profound and durable antitumor responses. However, as a case series, our findings are inherently limited by the small sample size, lack of a control group, and the retrospective, observational nature of the report. We cannot exclude the possibility that the observed tumor responses were attributable to the concomitant chemotherapy or to the natural history of the disease rather than to the ICI-induced immune activation. Furthermore, while we have invoked mechanistic hypotheses to explain the observed association, such speculations cannot be directly validated from two cases and should be considered hypothesis-generating rather than conclusive.

Beyond mechanistic speculation, our observations underscore the urgent need for biomarkers that can dissociate antitumor efficacy from life-threatening autoimmunity. Emerging tools such as T-cell receptor repertoire analysis, cytokine profiling, and gut microbiome sequencing hold theoretical promise in this regard; however, their clinical utility remains investigational and far from routine application.^30–34 For now, the priority lies in recognizing that severe irAEs do not guarantee benefit and may impose substantial risks, despite their potential to signal profound immune activation. Prospective cohort studies with systematic immune monitoring are warranted to validate whether the severity, type, or pattern of irAEs can guide treatment decisions or serve as reliable predictive indicators.

Importantly, the long-term remissions observed in our two patients, with ongoing responses lasting 22 and 37 months despite cessation of all antitumor therapy, provide compelling evidence that severe irAEs should not be viewed solely as “side effects” to be avoided, but rather as biologically informative events that merit careful study. A balanced approach that weighs the potential for durable benefit against the risks of life-threatening toxicity remains essential in clinical practice.

Conclusions

In summary, this case series provides additional clinical evidence supporting an association between severe irAEs and durable antitumor responses. However, our findings are observational and hypothesis-generating, and should not be interpreted as establishing a definitive predictive role for irAEs. Larger prospective studies with detailed immune profiling are needed to clarify the biological mechanisms underlying this association and to determine whether irAEs can be reliably utilized as predictive biomarkers in clinical practice. Until such evidence emerges, clinical decision-making regarding ICI management should remain anchored in patient safety and individualized risk assessment.

Funding Statement

This work was supported by Fujian Provincial Natural Science Foundation of China (No. 2026J0011419 and No. 2023J011356).

Data Sharing Statement

The datasets generated during and/or analysed during the current study are available from the corresponding author, Naying Yu, on reasonable request.

Informed Consent Statement

Written informed consent was obtained from the patients for the publication of this case report and any accompanying images.

Author Contributions

Qunxiang Chen and Xianglan Lin are co-first authors. Naying Yu and Xi Chen are co-corresponding authors. All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors report no conflicts of interest in this work.