Immune checkpoint inhibitor (ICI) therapy has changed the landscape of managing multiple cancer types in the last decade. They work by altering the immune regulatory pathways and thereby promoting cell-mediated destruction of the tumor cells. Several agents targeting Cytotoxic T-Lymphocyte Antigen 4 (CTLA4), programmed cell death-1 (PD-1), and Programmed death-ligand 1 (PD-L1) have been approved by the FDA (Food and Drug Administration) for multiple cancer types, and the indications for utilization is constantly expanding (1). ICIs widely used in the current clinical practice are PD-1 inhibitors (pembrolizumab, nivolumab, and cemiplimab), PD-L1 inhibitors (durvalumab, avelumab, and atezolizumab), and CTLA4 inhibitors (ipilimumab and tremelimumab). In addition, newer targets that are under investigation include LAG3 (lymphocyte activation gene-3), TIGIT (T cell immunoglobulin and ITIM domain), TIM3 (transmembrane immunoglobulin and mucin domain 3, GITR (glucocorticoid-induced tumor necrosis factor receptor [TNFR]–related), and OX40 targeting the lymphoid pathway, NKG2A, targeting the natural killer cell pathway and a bifunctional fusion protein targeting TGF-β and PD-L1 to name a few (2–10).

T cells and cancer cells have a complex regulatory mechanism. Typically, presenting major histocompatibility complex (MHC) peptide to the T cell results in cytoskeletal reorganization, downstream pathways resulting in genetic alteration, CD28 mediated co-stimulation, and an immunological synapse (11). Alternatively, several co-inhibitors such as PD-1 and CTLA – 4 on T cells prevent T cell stimulation (11). PD-L1 (B7-H1) is expressed on the surface of many cell types, including tumor cells, with PD-L2 predominantly on hematopoietic cells. During malignancy, chronic antigen presentation results in peripheral T cell exhaustion from the PD-1 and PD-L1 interaction despite the co-stimulation with CD28 (12, 13). CTLA4 has increased affinity and competes with CD28, interacting with CD80 and CD86 on antigen-presenting cells, resulting in decreased release of cytokines and cytotoxic enzymes (14). Apart from checkpoint inhibition, the addition of agents targeting tyrosine kinase inhibitors and vascular endothelial growth factors is being considered as there is evolving evidence that such combinations have more durable responses (15)

Immune-related adverse events (irAE) refer to a set of side-effects in the patients receiving ICIs similar to autoimmune responses (16). The irAE reporting is standardized with Common Terminology Criteria for Adverse Events (CTCAE) grading (17). A large meta-analysis reported all-grade incidence of irAE is about 83% with CTLA4 inhibitors, 72% with PD-1 inhibitors, and 60% PD-L1 inhibitors (18). CTLA4 inhibitors predominantly cause irAE by reinitiating exhausted T effector cells in the tumor microenvironment leading to the production of autoantigens apart from neo antigens, destroying normal tissue (19, 20). Disturbance of immune tolerance in peripheral tissues and T follicular helper cell-mediated generation of aberrant B cells mediated humoral autoimmunity are proposed mechanisms with PD1/PD-L1 inhibitors (19, 20).

The irAE profile in an individual depends on the organ that exhibits autoimmune-like activity. It may vary with the class of the ICI agent used (PD-1/PD-L1 vs. CTLA4). Compared to PD-1/PD-L1 inhibitors, CTLA4-inhibitors frequently cause colitis (irAE-GI), hypophysitis, and dermatitis (irAE-skin), and less frequently, cause pneumonitis (irAE-lung), hypothyroidism, and skeletal symptoms such as myalgias and arthralgias (21, 22). A much better understanding of the pathways involved with irAE incidence is needed to prevent them (20). It is essential to shed light on this aspect of the care given the encouraging results with ICI in the current clinical practice.

In this review, the available evidence on the risk factors contributing to irAE and biomarkers that predict the irAE occurrence was reviewed in a pattern that can be adopted in daily clinical practice. The patient’s baseline characteristics that increase the possibility of irAE before administering the first dose of ICI were classified into patient-specific risk factors and measurable entities as biomarkers. The basic history-taking sequence steps for risk factors were followed, starting with demographics (age, race, and gender), past medical history, and medication history. Few physical exam findings and social habits were covered in demographics. The tumor-specific and agent-specific risk factors were included to personalize it for a patient and the choice of the agent. The biomarkers with predictive value incorporated range from simple blood counts to complex microbiome studies. The idea is to provide helpful information for a treating physician to predict the incidence of irAEs.

The published studies that evaluated risk factors or biomarkers for irAE were historically conducted in populations that are specific to a primary tumor (example or, e.g., irAE in non-small cell lung cancer or NSCLC); ICI-agent (e.g., pembrolizumab in breast cancer, NSCLC, urothelial cancers or UC, and melanoma or MM); organ system involving irAE (e.g., pneumonitis in multiple cancers treated with PD-1 or PD-L1 or CTLA4 inhibitors); one primary and one agent (e.g., nivolumab in NSCLC); combinations (e.g., colitis in NSCLC treated with nivolumab or with PD-1 or PD-L1 or CTLA4 inhibitors in lung cancer and MM) (23–28). Therefore, it is difficult to generalize the results to an entire class of ICIs or a particular primary tumor, and the conclusions of the literature must be interpreted cautiously. It is often the combination of one or more risk factors and biomarkers that can predispose to more adverse events than one single factor/biomarker.

Risk factors refer to a variable associated with an increased risk of disease or infection (29). In the context of irAEs, the risk factors were grouped into three broad categories, a) patient-specific, based on demographics, social history, past medical history, and medication history; b) tumor-specific, based on the primary tumor (organ of primary-specific or histology specific); c) agent-specific, based on the ICI used. The risk factors are summarized in Figure 1 and Table 1 below. The risk was measured with odds ratio or in the majority of the studies. In Table 1, the available OR is indicated in the parenthesis beside the risk factor. Statistically significant (p<0.05) risk factors were mentioned in Table 1. The possible mechanisms involved are discussed in a Supplementary Table S1.

This section discusses the risk factors for irAE in the patient even before initiating ICI and further divides them into three more sets, demographics and social history, medical history, and medication history.

The severity and profiles of irAEs seem to depend on histology for the primary tumor (21, 64). Head-head studies are rare to identify them. When PD-1 inhibitors are used, MM patients tend to have a) all-grade irAE-GI (colitis and diarrhea) and irAE-skin (rash and pruritis) than NSCLC patients; b) more irAE-GI (diarrhea), irAE-skin, irAE-endocrine (hypothyroidism), and irAE-musculoskeletal (arthralgia) than RCC patients; c) fewer irAE-lung (pneumonitis and dyspnea) compared to NSCLC and RCC patients d) irAE incidence is later than lung cancer patients (5.2 vs. 2.1 months) (21, 34). In myeloid malignancies such as relapsed/refractory acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS) with ICI, high-grade events (≥ grade 3) usually involve the skin (15%) and liver (11%) (64).

A meta-analysis published in 2020 with 11 trials reported the fatal adverse events (FAE), not necessarily irAE, associated with pembrolizumab, were highest among breast cancer patients (3.2-1%), followed by NSCLC (2%), UC (0.8), and melanoma (0.2) (25). Combining chemotherapy with ICI worsened the risk of fatal adverse events (7% to 0.7%), but there was no significant difference between ICI and chemotherapy groups. Infectious complications were the most common cause of FAEs, followed by cardiac toxicity and pneumonitis. In another study, non-lung cancer patients such as MM and RCC were at a lower risk of hospitalization from irAEs (72).. Differences in the tumor microenvironment and microbiome composition among various cancers can explain the variations of irAE from a single ICI agent. There is also a significant disparity in utilization/pre-treatment of other treatment modalities such as chemotherapy and radiation in breast and lung cancer patients compared to patients receiving predominantly targeted therapy among MM and RCC, altering the irAE profile.

Disease burden also seems to play a role in the incidence of irAE. In our study, MM and lung cancer patients with ≥2 metastatic sites had a higher risk of severe irAE (24). In a small retrospective review on NSCLC patients (N=42), more irAE were among patients with high tumor burden (defined as the sum of the unidimensional diameters of up to five target lesions) (62). A large post-marketing study with the data from FAERS with over sixteen thousand patients with irAE reported tumor mutational burden (TMB) as an important risk factor for irAE across multiple cancer types (63). A higher median number of somatic mutations per megabase of DNA was associated with a greater risk of irAEs (Pearson correlation coefficient R = 0.704; P <.001)). The metastatic liver disease does not predict high-grade liver-irAE (hepatitis), but the risk is significantly higher in the patients who receive combination therapy (compared to monotherapy) (67).

The incidence and profile of irAE may differ with the type of agent, PD-1, PDL1, CTLA4, or combination. Studies after studies proved the higher incidence and severity of irAE with CTLA4 inhibitors when used alone or combined with PD-1 or PD-L1 agents such as ipilimumab and nivolumab irrespective of the primary tumor treated (22, 65–67). No reliable studies proved that one class of non-CTLA4 ICI (PD-1 or PD-L1) is worse than others. Colitis, hypophysitis, and rash are frequently associated with CTLA4 inhibitors. Alternatively, pneumonitis, arthralgia, vitiligo, and hypothyroidism were common in patients treated with PD-1 inhibitors (21). A meta-analysis showed a heightened risk of irAEs in solid tumors when ICI is added to chemotherapy irrespective of the agent (and the tumor type) used (68). Likewise, there are no studies that compared irAE-profile among different (non-CTLA4) ICI agents. In summary, CTLA4 (alone or in combination) exposes patients to a higher risk of irAE, and it is not clear if one PD-1 or PD-L1 agent is better in terms of incidence or profile of irAE.

A biomarker is defined as a characteristic measured as an indicator of normal biological processes, pathogenic processes, or responses to an exposure or intervention (89). The biomarkers discussed in this paper (as illustrated in Table 2 below) are divided into circulating blood cell counts, cytokines, autoantibodies, serum proteins, HLA genotypes, microRNA and gene profiling, and intestinal microbiota. While the majority of the biomarker studied were irAE-organ non-specific, we indicated the biomarkers specific for irAE-GI specific and irAE-skin studies in the table.

Rechallenging patients with ICIs after an irAE is another aspect of the treatment that is understudied. Various cancer societies proposed guidelines for re-initiating ICIs (134). After symptomatic recovery, it can be done while on steroid taper doses that do not affect the efficacy of the ICI (usually ≤10mg prednisone) after careful assessment of the available treatment options (135). It is tough to study the impact of rechallenging on survival, recurrence of irAE, and incidence of new irAEs as discussed in a systematic review published this year (136). It is hard to draw any concrete conclusions about the best approach with the available data as there are many variables in play such as the primary tumor, organ-involved, and ICI-class used are some of the variable factors that concrete conclusions. In a pharmacovigilance database review, when rechallenged after discontinuing ICI for irAE ≥ grade 2, only 39% experienced another ≥ grade 2 irAE, and about 70% had the same irAE at recurrence (137). The recurrence rates were also organ system-specific; it was highest in irAE-GI and least in endocrine-irAE. Time to occurrence of the first irAE seems to help predict recurrence (of irAE), as Sminonaggio et al. reported (111). In the recurrence group, the average time for the first irAE was shorter (9 weeks) compared to the non-recurrent group (15 weeks). The ideal time to rechallenge ICIs from incidence of irAE or resolution of symptoms is unclear too (136). In irAE-colitis, fecal calprotectin and lactoferrin are good biomarkers to monitor treatment response and can aid in deciding when to rechallenge the eligible patient (123–127). It is high in active irAE-colitis and drops with endoscopic and histologic remission.

Interestingly, the recurrent irAE was not as severe as the first one. Strategies advised for rechallenging and prevention are included in Table 3. Currently, there are no successful strategies for preventing irAE. American Gastroenterology Association published guidelines for managing irAe-colitis and irAE-hepatitis in March of 2021 (140). Budesonide is not beneficial for irAE-colitis prophylaxis. Similarly, there is no good evidence to use prophylactic steroids or any anti-inflammatory drugs in patients without any history of irAE. All the ICIs in the current practice have fixed doses, and there is no scope for modifying the dose as we might do while using chemotherapy.

In the current clinical practice, ICI use is growing exponentially. Limited prospective studies explored the factors contributing to the irAE. Retrospective studies were done in very specific populations, and it is difficult to draw solid conclusions and apply them to all cancers or agents or irAE types. An insight into the risk factors and biomarkers with significant predictive value for irAE incidence will enable the treating clinician to take the necessary steps to protect the patients from irAE and continue benefiting with ICI. In real-world practice, knowledge on organ-specific risk, gender-specific, comorbidity or medication-specific, primary tumor-specific, and ICI-agent-specific risk factors and biomarkers will allow physicians to personalize the necessary monitoring strategies to prevent irAE and restart ICIs post-irAE.

Patients must be trained to identify the irAE and alert the treating team early. High-risk individuals (chronic smokers, poor performance status, high BMI, and sarcopenia) must be monitored closely for irAE. Kidney injury (AKI) associated with ICI is frequent in patients with CKD. It is advisable to hold PPIs and NSAIDs or switch them to alternatives before initiating ICIs. The irAE- pneumonitis is frequent in patients with pulmonary fibrosis, asthma, and COPD. Patients with significant cardiac disease history (MI, HF) and uncontrolled hypertension may have more cardiac adverse events such as myocarditis or pericarditis. Patients receiving HER2 inhibitors, VEGF TKIs, and anthracyclines in prior lines of therapy should be monitored for cardiac irAEs.

Circulating blood counts and TSH/T4 are the only biomarkers accessible in the current clinical practice and must be used when possible. Absolute counts such as ALC, AEC, platelet count, or ratios such as NLR and PLR can predict irAE incidence. Following WBC and ALC might help monitor patients on ICI for irAEs. Prospective studies are needed to shed further light on other biomarkers such as cytokines, HLA genes, microRNAs, gene expression profiling, and intestinal microbiome need more studies before they can be adopted in daily practice.

Patients having an excellent response to ICI can be rechallenged based on the initial irAE grade and tolerability with careful assessment of risks and benefits. Resumption of the same agent after full recovery is popular, but class switching can be tried. Dropping CTLA4 inhibitors is also advised in patients with irAE on combination therapy. Close monitoring of liver chemistries (every week) for grade 1 hepatitis on ICI therapy and holding the therapy for grade 2 hepatitis is advised. Checking inflammatory markers (lactoferrin and calprotectin) for patients with diarrhea might help in the early diagnosis of colitis.

The majority of ongoing trials with ICIs are doing correlative studies by collecting patients’ blood, tumor, and microbiome. The focus of research is shifting towards incorporating genomic data (mutations or methylation) and tumor microenvironment (T-cells) from accessible sources such as blood, stool, and urine to study the factors affecting toxicity and efficacy of ICIs. Acknowledging the need to do more prospective studies in this unexplored facet of ICI therapy, these risk factors and biomarkers list can be a good starting point in clinical practice.

AC and AM contributed to the conception and design of the study. AC and AM organized the database, selected relevant articles, and wrote the first draft of the manuscript. Multiple revisions and additions to the final manuscript are contributed by LA, NJ, and CP. All authors contributed to manuscript revision, read, and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.