Due to the emerging use of ICIs for the treatment of cancer, the cumulated number of irAEs is increasing exponentially. irAEs have been observed in almost all organ system, including the gastrointestinal, cutaneous, pulmonary, neurologic, genitourinary, cardiovascular, and integumentary systems, with the median onset of 2–16 weeks from the commencement of therapy (Weber et al., 2015). Moreover, irAEs are graded using the Common Terminology Criteria for Adverse Events (CTCAE) and range from grade 1 to grade 5, which refers to mild, moderate, severe, life-threatening or fatal events (National Cancer Institute, 2019) (Table 1). Onset of irAEs ranged from a few days of ICI initiation to over 1 year after completion of therapy (Yoest, 2017; Parakh et al., 2018), and the risk of first-onset irAEs was threefold lower in the time period between 4 weeks and the end of treatment than during the first 4 weeks of treatment (Kanjanapan et al., 2019). For instance, the dermatological toxicities were observed at approximately 2–3 weeks, gastrointestinal and hepatic toxicities at 6–7 weeks, and endocrinologic events at 9 weeks after treatment initiation of ipilimumab (Weber et al., 2012). Noteworthily, irAEs presumably persisted long after the cessation of treatment, and it was possible that irAEs might last for many years after treatment. A recent case indicated that autoimmune hepatitis occurred 8 months after discontinuation of nivolumab in a patient with metastatic melanoma (Parakh et al., 2018). More importantly, the development of characteristic irAEs was positively associated with improved patient outcome (Maher et al., 2019; Eggermont et al., 2020; Shankar et al., 2020). A multicenter study reported an increase in overall survival in patients with related irAEs compared with those with no related irAEs (Maher et al., 2019). Likewise, another multicenter study demonstrated that there was a positive association between multisystem irAEs and improved survival in patients with advanced non-small-cell lung cancer (NSCLC) treated with ICIs (Shankar et al., 2020). Eggermont et al. revealed that patients with stage III melanoma who developed irAEs had a longer recurrence-free survival (RFS) (Eggermont et al., 2020). Given the immune mechanisms of anti-PD-1/PD-L1 antibodies, it was reasonable to associate the occurrence of autoimmune events with improved prognosis, as activation of the immune system can lead to tumor responses and autoimmunity. However, current evidence suggested that the early or specific irAEs were correlated with worse survival in ICI-treated patients with NSCLC. The reasonable explanation might be the difference of adjudication and attribution of irAEs, as well as immortal-time bias (Dall'Olio et al., 2020).

A recent study reported that the incidence of all grade irAEs from ICI monotherapy was as high as 90% (Pauken et al., 2019). irAEs were first identified after the introduction of anti-CTLA therapy in clinical practice (Wolchok et al., 2010; June et al., 2017). Previous studies indicated an overall incidence over 70% with anti-CTLA-4 and 27%–78% with anti-PD-1/anti-PD-L1 agents (Topalian et al., 2012; Kumar et al., 2017; Maughan et al., 2017; Elkrief et al., 2019; Wang et al., 2019). Among the diverse irAEs, cutaneous toxicities including rash, pruritus, and vitiligo were the most frequent (Weber et al., 2012). The overall incidence of cutaneous toxicities stemmed from CTLA-4 blockade ranged from 37% to 70% for all-grade and 1%–3% for grade 3 or higher cutaneous toxicities (Hodi et al., 2010; Eggermont et al., 2015). Whereas, cutaneous toxicities were less frequently reported with anti–PD-1 agents (17%–37%). Of note, ICIs could result in dose-dependent toxicities. Motzer et al. found that the incidence was 75%, 67% and 78% for all grade irAEs in patients with metastatic renal cell carcinoma receiving nivolumab 0.3, 2, and 10 mg/kg, respectively, and 5%, 17% and 13% for grade 3 or 4 irAEs (Motzer et al., 2015). Furthermore, a phase IIIb/IV study involving 360 advanced melanoma patients determined that the incidence of grade 3 or 4 irAEs was 33.9% with nivolumab 3 mg/kg plus ipilimumab 1 mg/kg versus 48.3% with nivolumab 1 mg/kg plus ipilimumab 3 mg/kg, while the incidence of all grade irAEs was 85.6% versus 93.8% (Lebbé et al., 2019). Anti-PD-1 therapy was associated with a lower rate of irAEs compared with anti-CTLA-4 therapy; however, other studies discovered that there were several treated patients (ranging from 7% to 24%) with severe or even fatal (grade 2 or higher) irAEs (Baumeister et al., 2016; Marrone et al., 2016). Moreover, a recent clinical study with metastatic melanoma patients revealed that 14% of patients experienced severe (grade 3 or 4) toxicities with PD-1/PD-L1 inhibitors alone (Vesely and Chen, 2020). The major fatal irAEs included cardiotoxicity, neurotoxicity and interstitial pneumonia, with the incidence of as high as 45% (Wang D. Y. et al., 2018). A recent review demonstrated that myocarditis, a type of fatal irAEs, was reported in up to 3% of patients treated with anti-CTLA-4 but less than 1% of patients receiving PD-L1 inhibitors (Dougan and Pietropaolo, 2020). Another study found high neurotoxicity in advanced melanoma patients after nivolumab, ipilimumab, or combination treatment with an overall rate of 2.8% and in combination treatment with a rate of 14% (Spain et al., 2017). Furthermore, a recent study highlighted ICIs-related stevens-Johnson syndrome/toxic epidermal necrolysis (SJS/TEN) events as fatal irAEs (Ma et al., 2021). For these fatal irAEs, the mortality rate of ICI-related SJS was 19.9%, and the mortality rate of ICI-related TEN was 61.6% (Zhu et al., 2021). Besides, patients with advanced melanoma (stage III or IV) could experience long-term toxicities after termination of anti-PD-1, and up to 43.2% of patients developed chronic irAEs (defined as irAEs persisting for at least 12 weeks after anti–PD-1 therapy cessation) (Patrinely et al., 2021).

The clinical features of irAEs are relatively obscure with subtle imaging changes and thus are difficult to determine, especially in the early stage. Therefore, it is imperative to understand and explore predictive risk factors or potential biomarkers for the occurrence of irAEs (Table 2).

Previous studies utilized a CTLA-4 knockout mice model and found that the mice developed T cell lymphoproliferation and T cell-mediated autoimmunity (Wing et al., 2008; Klocke et al., 2016). In accordance with these data in mice, individualsx with CTLA-4-related genetic disorders also had regulatory T (Treg) cell defects and autoimmune infiltration (Lo et al., 2016). Consistent with these results, several recent researches suggested that activation or reactivation of T cells was thought to be a dominant factor in the development of ICI-related irAEs (June et al., 2017; Dougan et al., 2021a). Recently, Takahashi et al. found that CD8⁺ lymphocyte infiltration was significantly greater in irAE colitis than that in ulcerative colitis (Takahashi et al., 2023). Further research confirmed that activated cytotoxic CD8⁺ T cells accelerated anti-PD-1 antibody-induced psoriasis-like dermatitis via IL-6 (Tanaka et al., 2020). Typically, Treg cells downregulate immune responses by inhibiting effector T cell proliferation and cytokine release to regulate self-tolerance. The functional homeostasis between Treg cells and type 17 T helper (Th17) cells served as a prominent factor in irAEs associated with ICIs (Knochelmann et al., 2018). Enhanced Th17 cell responses could contribute to the production of proinflammatory cytokines such as IL-17A, IL-21 and IL-22, consequently participating in the pathogenesis of autoimmune diseases and irAEs (Noack and Miossec, 2014; Tarhini et al., 2015).

Interestingly, a comprehensive review highlighted that cross-reactivity between anti-tumor T cells and similar antigens on healthy cells might also underlie the pathogenesis of some irAEs (Michot et al., 2016). For instance, vitiligo appeared to be more frequent in patients with melanoma, which suggested cross-reactivity between T cells and tumor antigens and melanocytes (Sandigursky and Mor, 2018). Due to the low selectivity among the tumor-reactive T cell subjects, cross-reactivity had also been suggested for ICI-related myocarditis \(Cheng and Loscalzo, 2017). Johnson et al. reported that two patients developed fatal myocarditis after treatment with ipilimumab and nivolumab (Johnson et al., 2016). Possible mechanism for this toxicity was that the tumor, skeletal muscle, and heart shared the common high-frequency T-cell receptors, which led to T-cell-driven drug reaction. In recent years, Berner et al. identified nine shared antigens between tumor tissue and skin, which was able to stimulate CD8⁺ and CD4⁺ T cells in vitro, highlighting a potential mechanism of ICI-mediated autoimmune toxic effects (Berner et al., 2019).

In addition to the regulation of T cells, both anti-CTLA-4 and anti-PD-1/PD-L1 antibodies could increase cytokine production. Anti-CTLA-4 antibody could enhance CD4⁺ and CD8⁺ T cell activation, subsequently releasing cytokines including tumor necrosis factor (TNF), interferon-γ (IFNγ) and IL-2, which contributed to further T cell proliferation and reactivation (Laurent et al., 2013; Seidel et al., 2018). Besides, Luoma et al. observed the upregulated expression of cytokines and their receptors in both T cell and myeloid cell of ICI-related colitis patients, which suggested that other innate and T cell-derived cytokines were associated with ICI-related colitis and could serve as additional treatment targets (Luoma et al., 2020). In recent years, TNF inhibitors have been successfully adopted to treat different irAEs in patients receiving ICIs, further highlighting the importance of cytokines in the pathogenesis of irAEs (Kim et al., 2017). Fundamentally, further studies are required to address the precise roles of these cytokines in irAEs and the development of T cell activation or reactivation.

Interactions between T cells and B cells can result in autoantibody production and play an important role in humoral immunity, while aberrant interactions are associated with autoimmunity (Petersone et al., 2018). Das et al. found B cell changes in patients with irAEs after the first cycle of therapy with either anti-CTLA4 or anti-PD1, or in combination, and they further identified B cell activation in cells with genomic profiles of CD21lo B cells by single-cell RNA sequencing, which implied that targeting B cells might reduce irAEs in these patients (Das et al., 2018). On the other hand, the development of autoantibodies was common with ipilimumab treatment, and autoantibody formation was associated with irAEs and antitumor efficacy (de Moel et al., 2019). Toi et al. indicated that the presence of the preexisting antibodies was independently associated with the development of irAEs in patients with NSCLC treated with anti-PD-1 (Toi et al., 2019). A recent study indicated that ICIs might exacerbate bullous pemphigoid (BP) as an irAE through the generation of anti-BP180-NC16A IgG autoantibodies (Sadik et al., 2019). Whether these autoantibodies reflect an underlying mechanism needs further elaborated experimental researches to validate.

Given that activation or reactivation of T cells is considered to be the dominant factor in the development of ICI-related irAEs, inhibition of T cell response and migration may be a tailored strategy for irAEs. Glucocorticoids are the preferred treatment for most irAEs except for endocrine irAEs and effective in mitigating symptoms. Animal models demonstrated that glucocorticoids might inhibit antitumor T cell responses and were well known to induce apoptosis of activated T cell (Pauken et al., 2019). Ambiguously, there are some disparities in findings from retrospective studies of patients’ treatment with ICIs who received glucocorticoids, though it is suggested that they overall had deleterious effects on antitumor responses, particularly when glucocorticoids were used at initiation of immunotherapy (Arbour et al., 2018; Faje et al., 2018). Moreover, there is a need to introduce steroid-sparing strategies with the occurrence of steroid-refractory irAEs, mainly including immune-related hepatitis, nephritis and pancreatitis. For instance, patients with steroid-refractory pneumonitis can be treated with mycophenolate-containing immunosuppressants (Thompson et al., 2019). Due to the significant side effects in the treatment of ICI-associated colitis with corticosteroids or immunosuppressive agents, a recent study provided novel evidence that modulation of the gut microbiome via fecal microbiota transplantation could abrogate ICI-associated colitis with a substantial reduction in CD8⁺ T-cell density (Wang Y. et al., 2018). Recent studies reported that vedolizumab, an inhibitor of α4β7 integrin that inhibited the migration of T cells into gastrointestinal mucosa, could be used instead of infliximab against immune-related colitis (Bergqvist et al., 2017; Abu-Sbeih et al., 2018), proposing α4β7 integrin as an attractive target for immunotherapy toxicities.

Owing to the important roles of acute inflammatory cytokines, such as TNFα, IL-6 and IL-1β in irAEs, inhibiting these cytokines may be a promising strategy for the treatment of irAEs. Of these, inhibition of TNFα appeared to be effective for severe, refractory, immune-related colitis and inflammatory arthritis (Baddley et al., 2018; Mooradian et al., 2020; Dougan et al., 2021b). Johnson et al. suggested that for grades 3 and 4 colitis patients, the addition of infliximab (targeted TNFα) to glucocorticoids was significantly associated with a shorter time to symptom resolution than utilization of glucocorticoids alone (Johnson et al., 2018). In a mice model, TNF inhibitors concomitantly with combined CTLA-4 and PD-1 immunotherapy ameliorated colitis and improved antitumor efficacy, which provided clinically feasible strategies to dissociate efficacy and toxicity for cancer immunotherapy (Perez-Ruiz et al., 2019).

Monoclonal antibodies have also been recommended for the management of some steroid-refractory irAEs (Kim et al., 2017; Siakavellas and Bamias, 2018). Clinical improvement was noted in 80% of patients with nivolumab-associated grades 3 and 4 irAEs (predominantly pneumonitis) who received tocilizumab (targeted IL-6) (Stroud et al., 2019). Infliximab, a chimeric human-murine monoclonal antibody, was used to the treatment of steroid-refractory irAEs (Araujo et al., 2021). In recent years, a series of cases of successful response to monoclonal antibodies, such as abatacept (targeted CTLA-4) and alemtuzumab (targeted CD52), were reported in patients with immune-related myocarditis (Esfahani et al., 2019b; Salem et al., 2019). A recent case also demonstrated that secukinumab (targeted IL-17A) could effectively manage ICI-induced inflammatory arthropathy (Ma et al., 2022). Additionally, intravenous immunoglobulin comprised of pooled IgG antibodies from the serum of thousands of donors was considered as a second-line therapy for neurological and hematological irAEs (Schwab and Nimmerjahn, 2013). Since several irAEs were caused directly by autoantibodies, such as some hematological or neuromuscular irAEs, plasma exchange could be a feasible strategy against irAEs by removing the pathogenetic autoantibody from the circulation, especially in severe cases of myasthenia gravis or Guillain–Barré syndrome (Touat et al., 2017). For instance, Verma et al. showed that rituximab (targeted CD20) could be used to treat patients with immunotherapy-induced myasthenia gravis (Verma et al., 2021). Clinical trials evaluating strategies to prevent irAEs are needed to institute in the future.

Despite the benefits of tailored management for irAEs, early intervention and prophylaxis has grown in importance. A retrospective analysis of 179 patients with immune-related colitis of all grades found that compared with patients who received immunosuppressive therapy (infliximab or vedolizumab) > 10 days after onset of colitis, patients who received immunosuppression therapy early (≤10 days) were less likely to be hospitalized, experienced steroid taper failure less frequently, had a shorter course of steroid treatment and had a shorter duration of symptoms (Abu-Sbeih et al., 2019). The JAK-STAT and mTOR pathway played important roles in cellular processes and mediated downstream signaling of numerous cytokines. A previous studies indicated that excessive JAK1 signaling contributed to cancer evasion and development of autoimmunity (Chan et al., 2019), and inhibition of JAK-STAT displayed synergistic effects with ICIs to overcome treatment resistance (Lu et al., 2017). Furthermore, the addition of the mTOR inhibitor in a melanoma patient treated with PD-1 blockade could promote ongoing anti-PD-1 efficacy and reduce levels of irAE-related cytokines/chemokines (e.g., IL-5, IL-17, and CXCL10) (Esfahani et al., 2019a). There are also several prophylactic strategies through disease-specific therapy (such as vedolizumab for IBD) before commencing ICI to prevent toxicity in patients with underlying active autoimmune disease (Haanen et al., 2020b). Due to the fact that not all patients with autoimmune diseases experienced irAEs, prophylactic strategies might be considered and studied in prospective trials. Nevertheless, proactive commencement of disease-specific therapy was a reasonable strategy for patients with a high risk of new toxicity and few alternative treatment options (Haanen et al., 2020a).