Immune checkpoint inhibitors work by releasing inhibitory pathways that normally restrain immune activation. The same biological mechanism that enables durable antitumor responses can also disturb immune tolerance and produce inflammatory injury across almost any organ.
This creates one of the central therapeutic dilemmas of modern immuno-oncology. When immune-related adverse events develop, clinicians may need to suppress the immune system that checkpoint therapy was intended to activate. Corticosteroids remain essential for many clinically significant toxicities, but growing evidence suggests that the intensity, timing and biological specificity of immunosuppression may matter.
The field is therefore moving beyond simply asking how to stop an irAE. The more ambitious question is whether we can identify the immune pathway responsible for injury in a particular tissue and control it while preserving as much antitumor immunity as possible.
Navigating Immunotherapy Side Effects: Understanding Immune-Related Adverse Events
Immune Toxicity Is More Complex Than Excessive Immune Activation
Immune-related adverse events were initially understood largely as consequences of generalized immune activation. Their biology now appears considerably more heterogeneous. Checkpoint inhibition can disturb peripheral tolerance, activate autoreactive and tissue-resident lymphocytes, alter B-cell responses, recruit inflammatory myeloid populations and amplify cytokine networks, with different mechanisms dominating in different organs.
This helps explain why checkpoint inhibitors can produce clinically distinct syndromes involving the skin, colon, liver, lungs, endocrine organs, kidneys, joints, nervous system, skeletal muscle and heart. Even within the same organ, apparently similar toxicities may represent biologically different inflammatory states.
The heterogeneity also creates an important therapeutic possibility. If the immune populations and pathways responsible for tissue injury are not completely identical to those responsible for tumor control, toxicity might sometimes be suppressed without eliminating the antitumor response.
One of the most interesting demonstrations of this biological separation came from Steven Blum, Daniel Zlotoff, Alexandra-Chloé Villani and colleagues in the 2024 Nature study “Immune Responses in Checkpoint Myocarditis Across Heart, Blood and Tumour.” Using single-cell RNA sequencing, T-cell receptor sequencing, microscopy and proteomics, the investigators analyzed cardiac tissue, blood and tumors from patients with checkpoint inhibitor-associated myocarditis. Cytotoxic T cells, conventional dendritic cells and inflammatory fibroblasts were enriched within affected cardiac tissue, but TCR clones expanded in the heart were largely distinct from those enriched in paired tumors.
The finding does not mean that antitumor immunity and autoimmunity are completely independent. They may share upstream mechanisms of checkpoint activation and immune expansion. However, it suggests that the immune response damaging normal tissue can diverge from the immune response operating within the tumor. That distinction provides much of the biological rationale for more selective approaches to irAE management.
Corticosteroids Remain Essential, but Dose May Matter
Systemic corticosteroids remain the foundation of treatment for many moderate and severe irAEs because they rapidly suppress multiple inflammatory pathways. For potentially life-threatening pneumonitis, hepatitis, myocarditis, neurological toxicity and other severe immune-mediated syndromes, delaying necessary immunosuppression in an attempt to preserve antitumor immunity can allow irreversible organ damage.
The more difficult question is whether every toxicity requires the same intensity of systemic immune suppression. A particularly important analysis came from Rik Verheijden and colleagues in the 2024 Journal of Clinical Oncology study “Corticosteroids for Immune-Related Adverse Events and Checkpoint Inhibitor Efficacy: Analysis of Six Clinical Trials.” The investigators analyzed individual patient data from six international phase II and III trials of combined PD-1 and CTLA-4 blockade: CheckMate 067, 142, 214, 648, 743 and 9LA.
Among 1,959 patients treated with combination immunotherapy, 834 received systemic immunosuppression for treatment-related adverse events. Higher peak corticosteroid doses were associated with worse outcomes. Compared with a peak prednisolone-equivalent dose of 0.5 mg/kg, a dose of 1 mg/kg was associated with an adjusted overall-survival hazard ratio of 1.21, while 2 mg/kg was associated with an adjusted hazard ratio of 1.66. Higher peak doses were also associated with worse progression-free survival. Interestingly, cumulative corticosteroid exposure was not significantly associated with survival.
These findings require careful interpretation. The analysis was post hoc, and patients requiring higher corticosteroid doses may have had more severe toxicities or other unfavorable characteristics. The results therefore cannot establish that high-dose corticosteroids directly caused poorer cancer outcomes.
Nevertheless, they challenge the assumption that more immunosuppression is automatically better once an irAE develops. The therapeutic objective should be adequate control of potentially damaging inflammation while avoiding unnecessarily intensive systemic immunosuppression when a lower dose or more selective strategy is clinically appropriate.
The Future May Be Steroid-Sparing Rather Than Steroid-Free
Concerns surrounding prolonged systemic corticosteroid exposure have accelerated interest in selective immunosuppressive therapies. The goal is not to eliminate corticosteroids from irAE management, but to reduce dependence on broad systemic suppression when the inflammatory mechanism can be targeted more precisely.
Immune-mediated colitis provides one of the clearest examples. Patients with steroid-refractory or steroid-dependent ICI colitis can receive infliximab, which inhibits TNF, or vedolizumab, which targets the α4β7 integrin involved in lymphocyte trafficking to intestinal tissue.
The difference between these strategies is biologically important. Infliximab produces systemic TNF inhibition, whereas vedolizumab has a more gut-selective mechanism because α4β7 contributes to lymphocyte homing into the gastrointestinal tract. For inflammation predominantly confined to the bowel, this creates an attractive possibility: control pathogenic immune trafficking within the affected organ while limiting systemic interference with immune activity elsewhere.
A 2025 systematic review and meta-analysis comparing infliximab and vedolizumab included six retrospective cohorts and 645 patients with ICI-associated colitis. Vedolizumab was associated with lower recurrence of colitis and approximately 17 fewer days of systemic corticosteroid exposure, although it generally required more doses. Remission rates between vedolizumab and infliximab monotherapy were not significantly different.
The underlying evidence remains retrospective, so the findings should not be interpreted as proof that one agent is universally preferable. More importantly, the comparison illustrates a broader direction in immunotherapy toxicity management: the possibility of replacing prolonged nonspecific immunosuppression with interventions matched to the biology and location of the inflammatory process.
Colitis Is Becoming a Model for Precision Toxicity Management
ICI-associated colitis also demonstrates why clinical symptoms alone may not adequately characterize an irAE. Diarrhea can vary substantially in severity and underlying pathology, and patients with similar symptoms can have very different endoscopic and histological patterns.
Endoscopy, histopathology and inflammatory biomarkers can therefore provide information beyond symptom grading, particularly in persistent or severe disease. This becomes clinically relevant when deciding whether corticosteroids are sufficient, whether selective immunosuppressive therapy should be introduced and how quickly treatment can be tapered.
The larger principle extends beyond the gastrointestinal tract. As irAE biology becomes better defined, management may increasingly depend on identifying the inflammatory phenotype rather than treating every toxicity affecting the same organ as a single disease.
Myocarditis Shows Why Preserving Antitumor Immunity Cannot Always Be the First Priority
The effort to preserve antitumor immunity must never delay treatment of a potentially fatal irAE. Checkpoint inhibitor-associated myocarditis is uncommon, but it remains among the most dangerous immune toxicities because patients can deteriorate rapidly through ventricular dysfunction, malignant arrhythmias, conduction abnormalities or overlap syndromes involving myositis and myasthenia-like neuromuscular disease.
In this setting, rapid control of immune-mediated cardiac injury takes priority. Checkpoint therapy is interrupted, corticosteroids are initiated promptly and additional immunomodulatory treatment may be required when inflammation continues despite initial therapy.
Myocarditis has nevertheless become one of the most scientifically interesting areas of irAE research because investigators are beginning to identify specific immune pathways responsible for the toxicity. This creates the possibility of replacing sequential empiric immunosuppression with mechanism-directed rescue.
Abatacept Introduces the Idea of Mechanism-Based Rescue
One of the most interesting agents being investigated in severe checkpoint myocarditis is abatacept, a CTLA-4 fusion protein that binds CD80 and CD86 on antigen-presenting cells and interferes with CD28-mediated T-cell costimulation.
Its mechanism is particularly relevant to checkpoint toxicity. Whereas checkpoint inhibitors are designed to release inhibitory signals and promote T-cell activity, abatacept can restore an inhibitory influence on T-cell activation through the CD28 costimulatory axis.
A highly influential observational study evaluated a strategy incorporating high-dose abatacept, the JAK1/2 inhibitor ruxolitinib and systematic screening for respiratory muscle failure in patients with severe ICI myocarditis. Mortality associated with myotoxicity was dramatically lower after implementation of the intensified strategy than in the preceding conventionally treated group.
The result generated considerable interest, but it must be interpreted cautiously. The study was nonrandomized, the cohorts were sequential rather than concurrent and several components of management changed simultaneously. It therefore cannot establish that abatacept, ruxolitinib or any single intervention was responsible for the difference.
The study nevertheless introduced an important concept: immunosuppression can potentially be guided by mechanism and pharmacodynamics rather than simply intensified until inflammation resolves. Abatacept dosing was adjusted according to CD86 receptor occupancy, while ruxolitinib was used to interfere with inflammatory JAK-dependent cytokine signaling.
This represents a different philosophy of irAE treatment. Instead of asking which additional immunosuppressant should be added after corticosteroids fail, the question becomes which immune pathway is sustaining the organ injury and how specifically can it be interrupted?
Tissue Biology May Tell Us Which Pathway to Target
The 2024 Nature myocarditis study takes this concept further. Analysis of more than 84,000 cardiac cells demonstrated increased frequencies and spatial colocalization of cytotoxic T cells, conventional dendritic cells and inflammatory fibroblasts within affected hearts. Blood profiling of more than 366,000 cells identified additional systemic immune alterations, while heart-expanded TCR clones could be detected within a cycling CD8+ T-cell population in blood. The presence of these clones was associated with fatal myocarditis case status.
Interestingly, 52 heart-expanded TCR clones tested in the study did not recognize the candidate cardiac autoantigens α-myosin, troponin I or troponin T. This illustrates how much remains unknown about the antigenic drivers of checkpoint toxicity.
The eventual clinical value of this type of work may extend beyond myocarditis. Tissue-level single-cell analysis could reveal which lymphocyte populations, antigen-presenting cells, cytokine programs or stromal interactions distinguish severe toxicity from self-limited inflammation. Circulating counterparts of those populations might then become biomarkers capable of detecting dangerous immune activation before irreversible organ injury occurs.
Endocrine Toxicities Follow a Different Biological Logic
Not every irAE should be approached as active inflammation requiring systemic immunosuppression. Immune-related endocrinopathies illustrate why organ biology matters.
Checkpoint inhibitors can produce thyroiditis, hypophysitis, adrenal insufficiency and insulin-deficient diabetes. In several of these conditions, clinically apparent endocrine dysfunction reflects substantial or irreversible loss of hormone-producing cells. Once the tissue has been destroyed, high-dose systemic corticosteroids cannot reliably restore its function.
Management therefore often centers on replacement of the missing hormone rather than prolonged attempts to suppress the immune system. A patient with permanent checkpoint-associated hypothyroidism, for example, may require lifelong thyroid hormone replacement while immunotherapy can sometimes continue depending on the overall clinical context.
This distinction is important because “immune-related adverse event” is a clinical umbrella covering biologically different processes. Some toxicities represent active, reversible inflammation in which rapid immune suppression can preserve organ function. Others result in permanent functional loss for which physiological replacement becomes more important than additional immunosuppression.
Precision management therefore requires understanding what stage of immune-mediated injury is occurring within the affected organ, not simply identifying which organ is involved.
Top Immunotherapy Trials to Watch at ASCO 2026
Can We Predict Severe irAEs Before They Happen?
The ideal approach would be to identify patients at risk before clinically significant tissue damage occurs. Numerous candidate biomarkers are being investigated, including circulating cytokines, autoantibodies, HLA genotypes, immune-cell phenotypes, T-cell receptor repertoires, microbiome characteristics and early changes in routine laboratory measurements.
No universal biomarker currently predicts clinically important irAEs with sufficient accuracy for routine use across cancers and checkpoint regimens. One reason may be that a universal biomarker is biologically unrealistic. The mechanisms predisposing to thyroiditis are unlikely to be identical to those causing colitis, pneumonitis, nephritis or myocarditis.
Prediction may therefore need to become organ specific. The myocarditis study provides an example: heart-expanded TCR clones detected within cycling peripheral CD8+ T cells were associated with fatal disease, suggesting that circulating immune populations might eventually provide information about an inflammatory process occurring within otherwise difficult-to-sample tissue.
Earlier detection could have an important therapeutic consequence. If dangerous immune activation can be recognized before extensive tissue damage develops, intervention may be possible at an earlier stage, potentially reducing the need for prolonged or increasingly aggressive immunosuppression.
The Microbiome May Influence Toxicity as Well as Efficacy
The intestinal microbiome adds another dimension to the biology of irAEs. Microbial communities influence intestinal barrier function, innate immune activation, T-cell differentiation and metabolite production, making the gut a particularly plausible site where microbial ecology could modify checkpoint-induced inflammation.
Studies have associated microbial composition with the development of gastrointestinal and other immune toxicities, although no microbiome signature is sufficiently validated to guide routine clinical decisions. Importantly, associations between particular microbial taxa and toxicity have not been fully consistent across cohorts.
The more interesting long-term possibility is therapeutic. If microbial states contribute to susceptibility to intestinal inflammation, microbiome-directed strategies might eventually modify the environment in which colitis develops rather than suppressing activated immune cells only after toxicity is established.
At present this remains investigational. Nevertheless, it illustrates the broader transition occurring in irAE research from treating established inflammation toward understanding why particular patients and particular organs become vulnerable in the first place.
Do irAEs Mean Immunotherapy Is Working?
Numerous observational studies have reported associations between development of certain irAEs and favorable cancer outcomes. This has generated the attractive hypothesis that toxicity can sometimes reflect particularly effective immune activation.
The relationship is considerably more complicated. Patients who remain on checkpoint therapy longer have more time both to respond and to develop toxicity, creating exposure and immortal-time biases. Different irAEs may also have different relationships with antitumor immunity, and the biological mechanisms responsible for endocrine toxicity may differ substantially from those driving inflammatory colitis or myocarditis.
The clinical conclusion should therefore not be that toxicity is desirable or that treatment should be withheld to preserve an immune response. Severe irAEs can cause permanent organ dysfunction and can be fatal.
The more interesting scientific question is whether antitumor immunity and autoimmunity share an initial phase of checkpoint-driven immune activation but subsequently diverge into different antigen specificities, T-cell clones and tissue environments. If that divergence can be mapped precisely, it may eventually become possible to suppress the pathogenic branch while preserving the tumor-reactive branch.
Rechallenge Is Not a Simple Yes-or-No Decision
Once an irAE resolves, another difficult question emerges: should checkpoint therapy be restarted? The answer depends on the affected organ, severity of the original toxicity, degree of recovery, previous benefit from immunotherapy, current cancer status and availability of alternative treatments.
An updated VigiBase analysis published in BMJ Open in 2024 examined 1,016 informative irAE cases following ICI rechallenge. Recurrence of the same irAE was reported in 323 cases, corresponding to an overall recurrence rate of 31.8%. Recurrence was not uniform across organs: it was reported in 50% of nephritis cases, 44% of skin irAEs and 39% of colitis cases. Combination checkpoint therapy was also associated with a higher reporting odds of recurrence than anti-PD-(L)1 monotherapy.
These figures should not be interpreted as precise individual probabilities. VigiBase is a pharmacovigilance database and is affected by reporting bias, incomplete clinical information and lack of standardized toxicity grading. The nephritis estimate, for example, had a wide 95% confidence interval of 25% to 75%.
Nevertheless, the findings demonstrate that a previous irAE does not automatically mean that the same toxicity will recur after rechallenge. At the same time, the absence of universal recurrence does not make restarting immunotherapy appropriate after every adverse event. Severe myocarditis, life-threatening neurological toxicity and certain severe dermatological reactions require a substantially more cautious approach than many lower-grade and reversible toxicities.
Rechallenge should therefore be viewed as an individualized risk-benefit decision rather than a universal rule. The relevant question is not simply whether an irAE occurred, but what organ was affected, how severe and reversible the injury was, how much benefit the patient derived from immunotherapy and what the consequences would be if the same toxicity returned.
Can We Treat the Organ Without Suppressing the Entire Immune System?
Many of the developments in irAE management converge on this question.
Vedolizumab provides one model by selectively interfering with intestinal lymphocyte trafficking. Hormone replacement in established endocrinopathies provides another, because physiological function can sometimes be restored without broadly suppressing the immune system. Abatacept in myocarditis represents a more direct attempt to interrupt a defined T-cell activation pathway, while biomarker research is beginning to identify immune populations associated with tissue-specific injury.
These approaches are biologically different, but they share a common objective: replace uniform systemic immunosuppression with treatment matched to the mechanism and location of toxicity whenever clinically possible.
This does not mean that organ-selective therapy will replace corticosteroids across all irAEs. Some toxicities are systemic, rapidly progressive or poorly understood, and broad immune suppression remains necessary. In other settings, waiting for a highly specific intervention could be dangerous.
The more realistic future is a layered strategy. Corticosteroids will remain essential when rapid control is required, but early biological characterization could identify patients who need targeted second-line treatment, patients who can transition quickly to steroid-sparing therapy and patients whose toxicity can be managed without substantial systemic immunosuppression at all.
From Toxicity Management to Immune Precision
The traditional approach to irAEs has largely followed a common sequence: identify the affected organ, grade toxicity, interrupt checkpoint therapy when necessary, administer corticosteroids and escalate immunosuppression if the patient fails to improve.
That framework remains clinically important, but it treats many biologically distinct inflammatory syndromes through a relatively small number of therapeutic pathways.
A more precise model would incorporate several layers of information: the affected tissue, severity and reversibility of injury, dominant immune-cell populations, inflammatory pathways, circulating biomarkers and potentially the relationship between the toxicity-associated immune response and the patient’s tumor-reactive immunity.
Different toxicities could then lead to different interventions. Gut-selective trafficking blockade might be prioritized for appropriate gastrointestinal inflammation, costimulatory pathway modulation for severe myocarditis, hormone replacement for established endocrine destruction and other targeted strategies as the underlying biology becomes clearer.
The objective is not simply to develop more drugs for irAEs. It is to determine which component of the immune response needs to be suppressed and which component should be preserved.
