Currently, irAEs are believed to reflect an exaggerated host immune reaction (Naqash et al., 2019). The role of both auto-reactive T- and B-cells, together with antibody-mediated processes, and a decreased T-regulatory phenotype, has been investigated (Petrelli et al., 2018; Esfahani et al., 2020a). The diversification and sub-compartmental expansion of lymphocytes induced by ICIs in secondary lymphoid organs, peripheral tissues and blood has been hypothesized to promote the development of autoreactive T cells and B cells (Esfahani et al., 2020a). Emerging data on patients developing irAEs support a cross-reactivity of T-cell clones with antigens and/or epitopes shared between tumor cells and healthy tissue (Berner et al., 2019; Rapisuwon et al., 2019). Another main feature may be the loss of naive T cells and the accumulation of overactive memory T-cells invading peripheral organs, triggering inflammatory damage (Waldman et al., 2020). Modulation of the regulatory T-cells (Treg) stability may also contribute to the balance between antitumor immunity and irAEs, since this subset of T-cells is involved in the exhaustion of the immune response. Targeting Treg-derived EZH2 or NRP1 genes, both of which are highly expressed in tumor-infiltrating Tregs, has been demonstrated to bolster the production of proinflammatory cytokines, hence producing antitumor immunity without autoimmune consequences (Overacre-Delgoffe et al., 2017; Wang et al., 2018a). Depletion of FoxP3+ Treg also seemed to worsen liver toxicity in murine models (Bartkowiak et al., 2018). Of note, ongoing disease responses despite ICI discontinuation due to irAEs have been described. The potential cross-reactivity of ICI-induced persistent CD8⁺ T effector memory subsets against tumor cells and normal tissues may be a plausible explanation for the durable benefit gained with ICIs. Moreover, anti-PD-1 antibodies are known to have a T-cells receptor occupancy of >2 months, and a half-life that spans 3–4 weeks, a steady state concentration being achieved within 19 weeks (Naqash et al., 2019). However, the varying populations of initial priming T-cells targeted by ICIs may influence the severity and time to onset of irAEs (Milling et al., 2017). A trend to higher frequencies of CD8⁺ T cells in both newly detected clones and increased preexisting clonotypes, as compared to CD4⁺ T cells, has also been reported in patients developing irAEs (Oh et al., 2017). The depletion of Treg at tumor sites and increase of peripheral CD8⁺ T effector cells have been hypothesized to drive irAEs, induced by CTLA-4 blocking antibodies (Milling et al., 2017). In particular, anti-CTLA-4 antibodies of the IgG2a isotype seemed to be capable of engaging Fcγ receptors expressed by tumor-associated macrophages within the tumor microenvironment in mouse models, hence to mediate an antitumor activity via selective elimination of intratumoral T cell populations, particularly Tregs (Bulliard et al., 2013; Simpson et al., 2013). Instead, other isotypes expanded both Treg and effector T-cells at the periphery (Selby et al., 2013). Overall, Treg depletion may lead to increased lymphocytes and leukocytes infiltration into the mucosal linings of organs including the lungs and gastrointestinal tract (Milling et al., 2017).

The role of B cells in the pathogenesis of irAEs is less well characterized than that of T-cells. Firstly, humoral immunity may play a role in some forms of irAEs (Young et al., 2018). Pathogenic autoantibodies, in turn, may be generated as a result of nascent autoreactive T cells and B cells that are able to escape central tolerance (Esfahani et al., 2020a). Supporting this concept, autoantibodies have been reported in some cases of ICI-induced type 1 diabetes, thyroiditis and hypothyroidism, and rheumatic disorders, but whether these antibodies represent a pre-existing immune substrate or are the consequence of ICI-induced self-tolerance alterations remains unknown (Esfahani et al., 2019). Antibodies to islet cell antigens and glutamic acid decardoxylase-65 were also reported to be associated with ICI-induced diabetes (Weinmann and Pisetsky, 2019). In PD-1 knockout mice, there is early evidence of the development of a lupus-like disease, with glomerulonephritis and renal deposition of IgG3 and C3, inflammatory arthritis in the majority of cases, and autoimmune dilated cardiomyopathy with severe heart failure (Nishimura et al., 1999; Nishimura et al., 2001). Conversely, autoantibody positivity for islet-associated antigens has been found in less than 50% of cases at diagnosis, versus 90% in conventional type 1 diabetes. Similarly, rheumatological irAEs have been associated with negative rheumatoid factor and cyclic citrullinated protein antibodies, and frequently low antinuclear antibodies at presentation. On the contrary, thyroid antibodies have been observed in cases of anti-PD-1-induced thyroid irAEs in non–small cell lung cancer (Stamatouli et al., 2018; Kurimoto et al., 2020). Nevertheless, the development of antinuclear antibodies or a lupus-like disease is not reported as a prominent feature of irAEs in the available literature. It is possible that a distinct pattern of B cell expression occurs after CTLA-4 inhibition (Weinmann and Pisetsky, 2019). In a series of 127 patients treated with ICIs by de Moel et al.,19.2% of them, who were antibody negative before treatment, developed new autoantibodies, more frequently anti-TPO (thyroperoxidase) and anti-TG (thyroglobulin), but the relationship between the emergence of autoantibodies and irAEs was not statistically significant. irAEs were reported in 78.9% of autoantibody-positive patients, compared to 57.5% of patients who were autoantibody negative before starting ICI therapy; thyroid dysfunction was observed in 44.4% of patients who exhibited anti-thyroid antibodies during anti-CTLA-4 therapy and then received anti-PD-1 ICI (de Moel et al., 2019).

Secondly, aberrations in various B-cell subsets have been observed (Dubrot et al., 2010). Heterozygous germline mutations in the gene encoding CTLA-4 have been associated with a decreased absolute B cell count, switched memory B cell counts, and the expansion of the subset CD21_lo memory B cells, where low levels of CD21 are usually responsible for B cell exhaustion and chronic antigen exposure in humans, who are thus more likely to develop irAEs (Rakhmanov et al., 2009; Schwab et al., 2018; Esfahani et al., 2020a). Such findings, together with a restricted expression of PD-1 and increased IFNɣ signaling in CD21_lo B cells themselves, were also observed after combined immune checkpoint blockade (Kuehn et al., 2014). Combination ICI-therapy has also been shown to be associated with a significant decrease in the number of circulating B cells after treatment, but increased expression of immunoglobulin genes and number of plasmablasts, together with plasma levels of the chemokine CXCL13, a marker of germinal centre activation in humans. Such early changes in B cells seemed to be correlated with higher rates of grade 3 or higher irAEs at 6 months after immune checkpoint inhibition (Weinmann and Pisetsky, 2019), as well as an earlier time to onset and severity of immune toxicity, irrespective of the site. However, it is not known whether such reduced proportions of B cells may reflect an effect on antibody production (Das et al., 2018; Young et al., 2018; Weinmann and Pisetsky, 2019).

There is a growing interest in T-cell receptor (TCR) repertoire profiling of T lymphocytes in tissues and/or peripheral blood, as a possible means of predicting the onset and severity of irAEs, and/or the therapeutic response to ICIs (Hofman, 2019). For instance, patients treated with anti-CTLA-4 antibody for metastatic castration-resistant prostate cancer who developed irAEs showed a significant early diversification of the CD4⁺ and CD8⁺ T-cell repertoires, and the generation of self-reactive clonotypes within 2 weeks of treatment, whereas non-irAE patients did not (Esfahani et al., 2020a). The development of irAEs was associated with an increased number of new TCR clones and preexisting clonotypes following ipilimumab plus Granulocyte-macrophage colony-stimulating factor (GM-CSF) treatment, as compared with non-irAE patients (Oh et al., 2017). Pre-clinical evidence suggested that polyclonal T-cell expansion may cause an increased accumulation of mononuclear, cytotoxic cells in the portal areas of the liver, thus contributing to hepatocytes apoptosis and subsequent hepatitis (Niu et al., 2007; Dubrot et al., 2010). A recent study showed that immune toxicities elicited by CTLA-4 blockade were associated with early diversification of the T cell repertoire (Oh et al., 2017). In addition, alterations to the T-cell repertoire in response to ICIs have been reported to be correlated with both the therapeutic response and the severity of irAEs (Young et al., 2018). A different total number of productive T cell repertoire V b sequences in the complementarity determining region (CDR3) in patients with and without irAEs has also been reported, but a concomitant expansion of particular clones in the peripheral blood was not observed (Weinmann and Pisetsky, 2019).

Such findings suggest that irAEs may derive from a mobilization of large amounts of T cells, some of which are autoreactive to multiple antigens and gradually emerge after checkpoint blockade (Weinmann and Pisetsky, 2019). However, despite the detected early increase in TCR diversity in irAE patients, the actual toxicities usually manifested with variable kinetics: the skin was typically involved earliest, followed by the onset of colitis (after 1–3 administrations), and lastly of hepatitis and endocrinopathies (Oh et al., 2017; Hofman, 2019). This means that TCR repertoire diversification may represent only a part of the pathogenic mechanisms underlying toxicity, including possible additional repertoire changes that occur nearer in time to the actual onset of irAEs.

There is evidence of gastrointestinal side effects in cancer patients treated with anti-CTLA-4 antibody, associated with an increased expression of the biliary glycoprotein CECAM-1 and the cell surface glycoprotein CD177, which is directly correlated with neutrophil activation (Friedlander et al., 2018). Abnormal CTLA-4 recycling and subsequent lysosomal degradation have also been shown to contribute to toxicities, and possibly to a reduced drug effectiveness (Waldman et al., 2020).

Eosinophils have been reported to be related to the occurrence of irAEs, but seemed to have no impact on survival during treatment. Functional links between Interleukin (IL)-17 and eosinophils have been hypothesized, since eosinophils do not only act as effector cells but also drive the IL-23/IL-17 axis and contribute to the recruitment of inflammatory cells (Schindler et al., 2014). Indeed, increases in eosinophil counts and release of the pro-inflammatory cytokine IL-17 have been identified as predictive biomarkers of toxicity of ICIs, regardless of the affected organ (Waldman et al., 2020).

There is also evidence of monocyte-derived macrophages having a critical role in the PD-1/PD-L1 inhibitor-induced autoimmune diabetes. Hu and colleagues demonstrated that the high levels of IFN-γ produced by exhausted CD8⁺ lymphocytes promote the recruitment of monocyte-derived macrophages in mice. Such highly activated monocyte-derived macrophages acquire cytocidal activity against pancreatic islet β-cells, hence driving diabetogenesis, while their temporal depletion dramatically protected mice from β-cell damage. (Hu et al., 2020). Autopsy-examined patients treated with anti-CTLA-4 antibodies alone or in combination with PD-1/PD-L1 inhibitors confirmed predominant T-cell infiltrates in irAEs, but also a consistent inflammatory infiltration by type 2 macrophages (M2) and Treg depletion. Such findings further suggested that the progressive organ damage mediated by M2 and depletion of T-regs play an important role in the pathogenesis of irAEs (Mihic-Probst et al., 2020). The activation of peripheral monocytes has also proven to be involved in the pathogenesis of ICI-induced hepatitis. A recent analysis of phenotypic and transcriptional profiling of peripheral immune subsets of patients treated with ICIs and developing hepatitis showed the presence of activated monocytes and enhanced effector CD8⁺ T cells, compared to those without hepatitis and healthy controls. To support this, CD163⁺/CCR2⁺ macrophages and CD8⁺ T cells were found in liver inflammation infiltrate (Gudd et al., 2021).

Moreover, off-target effects of ICIs on non-haematopoietic cell lines bearing the target immune checkpoint ligand have been described in a preclinical model. In this regard, the CTLA-4 receptor is also expressed on hypothalamic and pituitary tissues, so that hypophysitis is particularly common with CTLA-4 antibodies and combination therapy, but such immune-related endocrine side effect has also been observed during anti-PD-1 use, although less frequently (Albarel et al., 2019; Esfahani et al., 2020a).

The collateral damage to non-transformed tissues distributed in and around the tumor microenvironment induced by adaptive and innate anti-tumor immune responses is presumed to facilitate the host autoimmunity. An interesting model speculated on molecular mimicry leading to antigenic cross-presentation, or the release of self-antigens, which favor the activation of autoreactive T or B cells, although no evidence of specific, self-antigen- directed adaptive immune responses is available in literature (Rojas et al., 2018). Cells of both the adaptive and the innate immune systems produce IL-17, which represents a cross-link between immune and non-immune cellular activities. This pathway inhibits the production of Treg, thus reversing anti-cancer immune-response exhaustion (Esfahani et al., 2020a). The balance between IL-12 and IL-23 has been identified as an important regulator of carcinogenesis and autoimmunity: IL-12 promotes T helper (Th) 1 cells-mediated innate and adaptive immune responses, while IL-23 has a key role in Th17-mediated adaptive immunity (via IL-17 production) (Ngiow et al., 2013).

Interferons (type I IFNα, IFNβ, IFNε, and IFNω) and type II interferon (IFNɣ) also drive innate and adaptive immune responses. Interferons mostly act through the JAK–STAT signal transduction pathway. In the context of anticancer immune responses, an intact IFNɣ axis is essential for the expression of PD-1 and PD- L1, hence in regulating responses to ICIs (Esfahani et al., 2020a). Likewise, the mTOR pathway modulates both innate and adaptive immune responses (Powell et al., 2012). Combining mTOR inhibition and anti-PD-1 therapy in cancer patients requiring ICIs, with a history of solid organ transplantation, ensured allograft tolerance by abating the threshold for innate and adaptive immune response activation, while preserving the IFNγ signaling required for the efficacy of anti- PD-1 therapy (Esfahani et al., 2020a).

Integrin-mediated T cells extravasation into peripheral tissues is known to be responsible for tissue inflammation. In the cancer immunotherapy setting, the release of inflammatory cytokines from immune cells may potentially play an important role in mediating irAEs (Esfahani et al., 2020a).

IL-1 has a crucial role in autoinflammatory diseases (Sutton et al., 2019). This cytokine is predominantly involved in the innate immune response, secreted by macrophages and monocytes. Higher levels of IL-1β and lower levels of IL-1 receptor antagonist (IL-1RA) were found in myeloid immune cells of patients with solid tumors treated with ICIs who developed immune-related pneumonitis. Elevated blood levels of IL-1 have also been detected in patients with melanoma who suffered a wide spectrum of irAEs (Lim et al., 2019; Suresh et al., 2019).

Cytokines like tumor necrosis factor (TNF) α and IL-17 are known to enable tumor surveillance, but they can also mediate autoimmune reactions. In fact, the use of IL-17 monoclonal antibody for the treatment of immune-related colitis and psoriasis has been reported to reduce immune toxicities, but also induce tumor escape. Conversely, TNFα blockade has been demonstrated to uncouple the efficacy and toxicity of ICIs targeting PD-1 in the treatment of ICI-related colitis in animal models (Esfahani et al., 2020a). Elevated baseline circulating levels of TNFα have also been described as a biomarker for the subsequent development of irAEs (Head et al., 2019a). The CD8⁺ T-cell production induced by TNFα, in turn, has been demonstrated to be critical in the development of splenomegaly, lymphadenopathy, hepatomegaly, and hepatitis (Milling et al., 2017).

IFNɣ and type I interferons were reported to contribute to the expansion of blood cells and mislocalization of T-cells (Milling et al., 2017). Serum levels of IFNɣ inducible chemokines CXCL-9 and CXCL-10 have also been shown to be correlated with the development of irAEs (Khan and Gerber, 2020). Preliminary evidence suggests that IL-27 produced by anti-CD137 stimulated myeloid subsets to mediate the recruitment and activation of liver damaging T cells (Milling et al., 2017).

Increased circulating levels of IL-6 and C-reactive protein (CRP) (whose expression is directly dependent on IL-6 during ICI administration) have been recorded in patients with different tumor types developing a broad range of irAEs, and a synergistic antitumor activity has been demonstrated in mouse models combining ICIs with inhibition of the IL-6 axis (Esfahani et al., 2020a). Evidence of elevated levels of IL-6 and CRP in patients with irAEs, as compared to baseline pre-treatment levels, as well as elevations in IL-17 and IL-10 levels, indirectly supported the induction by the altered cytokine physiology of an immune dysregulation during irAEs. However, these results need to be interpreted with caution, given the lack of prospective validation to account for confounding factors (e.g., potential infection) which may contribute to cytokines dysregulation (Naqash et al., 2019).

Current analyses of irAEs lack studies on epigenetics, so the impact of genetic susceptibilities on the development of irAEs remains poorly explored. Distinct human leukocyte antigen (HLA) haplotypes, and polymorphisms in immunoregulatory genes such as CTLA-4 and PD-1 have been associated with a variety of classical autoimmune diseases, and likely play an important role in the development of irAEs (e.g., HLA-DQ8 and HLA-DR53 for lymphocytic hypophysitis), and the predominance of HLA-DR4 among patients treated with PD-1- or PD-L1-directed ICIs developing autoimmune insulin-dependent diabetes (Khan et al., 2018)). Conversely, the effect of genetic variants of CTLA-4, PD-1, and PD-L1 on the risk or severity of irAEs is still little understood (Esfahani et al., 2019).

The human microbiome has been historically defined as the set of microbes, mainly bacteria, and their genes in the human body, involved in the regulation of key metabolic pathways and immunomodulation (Joseph et al., 2020).

Recently, the microbiome has been reported to modulate irAEs, besides priming the immune response to ICIs. Specific microbial metabolic pathways, such as polyamine transport and vitamin B synthesis, were found to be predictive for resistance to colitis induced by CTLA-4 blockade, as well as a Bacteroidetes-rich phylotype (Esfahani et al., 2020b). The bacterial composition of the microbiome may influence differentiation of the immune lineage towards pro-inflammatory or regulatory cell subtypes, exacerbating the pro-inflammatory effects during ICI administration. In this regard, commensal Bacteroides fragilis have been shown to facilitate the development of Treg cells, producing anti-inflammatory IL-10, while these segmented filamentous bacteria promote the differentiation and generation of Th17 cells and the related pro-inflammatory pathway in the small intestine (Al-Qadami et al., 2019).

Not surprisingly, the most common irAEs involve organs with a rich content of commensal organisms, such as the skin, colon and lungs (Esfahani et al., 2020a). Accordingly, exposure to antibiotics has been reported to worsen survival outcomes, and increase the incidence and severity of irAEs (Abu-Sbeih et al., 2019; Elkrief et al., 2019). Nevertheless, studies of the effects of the microbiome on ICI efficacy and toxicity cannot ignore concomitant factors which may affect the integrity of the commensal microbiome composition, such as exposure to antibiotics or other drugs (proton-pump inhibitors, antipsychotics, antimetabolites), radiation therapy (RT), and diet (Dubin et al., 2016).

A retrospective study reported increased circulating levels of C-reactive protein (CRP) in patients with ICI-induced pituitary inflammation, hepatitis, thyroiditis and autoimmune colitis (Abolhassani et al., 2019). Since the CRP represents an inflammatory acute phase reactant, its level is susceptible to a variety of acute or chronic infections, anti-infective and anti-inflammatory drugs and autoimmune diseases, so that an increase of CRP at least twice during the 2-week interval of ICI administration, together with low levels of procalcitonin, and no evidence of infection (culture and serology) have been suggested as predictors of irAEs (Abolhassani et al., 2019).

Patients with low serum albumin showed a higher risk to develop anti-PD-1-related pneumonitis, instead (Fukihara et al., 2019). Overall, the incidence of grade 3–4 irAEs, especially pneumonitis, has been shown to be associated with increased white blood cell counts and decreased relative lymphocyte counts (Fujisawa et al., 2017). The rate of anti-PD-1 antibodies-induced toxicity was previously suggested to be related to a higher baseline and increase in absolute lymphocyte counts, and absolute eosinophil counts after ICI administration (Diehl et al., 2017). A low baseline proportion of peripheral blood CD4⁺ Tregs has been reported to be associated with anti-CTLA-4-related colitis (Shahabi et al., 2013), while eosinophils had already been found in biopsy specimens from patients with anti-CTLA-4-induced hepatitis, rash and colitis (Berman et al., 2010; Lacouture et al., 2014; Johncilla et al., 2015). Of note, the patient’s inflammatory state and a variety of non-cancer related conditions may primarily affect the blood cell count, nevertheless the additional myelosuppression eventually induced by chemotherapy and/or RT may contribute to the blood cell count impairment. (Jia et al., 2020).

Cytokines and chemokines represent other potential biomarkers for predicting irAEs. Baseline elevation of TNFα and IFN-α2 levels in melanoma patients blood samples collected prior to the start of ICI therapy has revealed to be associated with the development of higher grades of irAEs (60 and 44% of the analyzed samples, respectively), although no associations between immune markers and the number and type of ICI-induced side effects in individual patients were observed (Head et al., 2019b). Both low baseline levels, and high levels of IL-6 after ICIs have also been described to predict the occurrence of irAEs, such as Crohn’s disease (Atreya et al., 2000), psoriasiform dermatitis (Okiyama and Tanaka, 2017), and colitis (Valpione et al., 2018). The upregulation of plasma levels of CXCL9, CXCL10, CXCL11 and CXCL13 was closely related to the occurrence of irAEs, as well as significantly higher levels of CXCL2, CCL20, CXCL8 and CCL23 were found in patients with irAEs than in those without immune-related toxicities (Khan et al., 2018), but yet, no association with organ-specific irAEs has been observed. Lim et al. also showed that increased expression of colony-stimulating factor-1 (CSF-1), fractalkine, and IL families is closely related to severe irAEs (Lim et al., 2019).

Alternatively, evaluation of muscle quality on CT scan has been proposed as a potential marker able to predict the development of severe irAEs. Sarcopenia and low muscle attenuation were described as independent factors significantly associated with high-grade toxicities from anti-CTLA antibody in metastatic melanoma (Daly et al., 2017).

In the near future, high tumor mutation burden might be a promising, nonspecific biomarker of irAEs in vulnerable cancer patients. In this regard, a significant positive correlation during anti-PD-1 treatment was shown in a variety of solid tumors (Bomze et al., 2019), but no link with organ-specific irAEs, or evidence about other ICIs like PD-L1 and CTLA-4 inhibitors have been found. Therefore, such findings need to be further explored in large prospective clinical studies. Importantly, higher baseline levels of soluble CTLA-4 (sCTLA-4) have been proposed as possibly predictive of immune toxicity in melanoma patients treated with anti-CTLA-4 antibody, especially gastrointestinal irAEs (Pistillo et al., 2019), although to some extent, high levels of sCTLA-4 may also reflect tumor immune escape and high tumor burden (Ward et al., 2013).

Specific disorders of the gut microbiome, such as the absence of Bacteroidetes and high levels of Firmicutes in stool samples have been reported with higher risk of developing immune-related colitis (Dubin et al., 2016; Chaput et al., 2017).

There is also evidence of some humoral factors as possible predictive biomarkers of gastrointestinal irAEs. Higher blood levels of neutrophil activation markers, mainly CD177, and CEACAM1, prior to the absolute neutrophil count have also been described in this type of patients (Shahabi et al., 2013). A retrospective study showed a close relationship between peripheral blood mRNA expression of CCL3, CCR3, IL-5, IL-8 and PTGS2 and immune-related diarrhoea, especially grade 2–4 diarrhea (Friedlander et al., 2018), while upregulation of IL-17 levels at baseline and 6 weeks after treatment with anti-CTLA-4 antibody was reported to correlate with grade 3 diarrhea and colitis in melanoma patients (Tarhini et al., 2015).

Increased levels of anti-CD74 autoantibodies after ICI therapy have been demonstrated to be notably correlated with immune-related pneumonia (Tahir et al., 2019).

Thyroid dysfunction, even destructive thyroiditis were reported to be significantly associated with high baseline levels of anti-TG and anti-TPO antibodies (Osorio et al., 2017; Kobayashi et al., 2018), as well as early increase (before 4 weeks) in blood thyroglobulin levels (Kurimoto et al., 2020). Besides, the baseline thyroid uptake of fluorodeoxyglucose 18F-FDG has shown to increase the risk of anti-PD-1-related thyroid toxicity (Yamauchi et al., 2019). All this considered, pre-treatment evaluation of anti-thyroid antibodies might help identify patients with a high risk of thyroidal irAEs and improve the clinical benefit of ICIs, supported by the possible role of functional imaging in monitoring the dynamic changes of thyroid 18F-FDG uptake.

Despite its rarity, anti-PD-1/PD-L1-related type 1 diabetes has also been associated with islet autoantibody, insulin autoantibody and islet antigen 2 antibody positivity in some cases (Stamatouli et al., 2018).

Elevated levels of anti-GNAL and anti-ITM2B autoantibodies have been reported in cancer patients developing immune-related hypophysitis (Tahir et al., 2019). Since GNAL and ITM2B are target proteins with an established role as signal transducers in the normal secretion of various pituitary hormones, such as Thyrotropin-releasing hormone (TRH) and Adrenocorticotropic hormone (ACTH) (Kilger et al., 2011; Peverelli et al., 2014), disorders of such pituitary hormone levels as a sign of impaired pituitary function may become another interesting field of study.

Circulating anti-conductive tissue autoantibodies (ACTA) were suggested as a possible predictive biomarker for ICI-related cardiotoxicity, together with the rise of troponins and BNP in some cases of myocarditis after ICI treatment (Läubli et al., 2015; Johnson et al., 2016).

Finally, a common genetic variant of HLA haplotype, HLAdrb1*11:01, has been observed in a retrospective study as related to itching, and may be a possible future biomarker predictive for skin toxicity (Hasan Ali et al., 2019), as well as HLA genotypes dominated by DR4 were described in patients with anti-PD-1/PD-L1-induced type 1 diabetes (Stamatouli et al., 2018). Such findings increase the credibility of susceptible HLA genotypes as potential markers to predict patients predisposition to autoimmune diabetes before or during ICI therapy.