Immune related adverse events after immune check point inhibitors: understanding the intersection with autoimmunity

Summary

Immune checkpoint inhibitor therapies act through blockade of inhibitory molecules involved in the regulation of T cells, thus releasing tumor specific T cells to destroy their tumor targets. However, immune checkpoint inhibitors (ICI) can also lead to a breach in self-tolerance resulting in immune-related adverse events (irAEs) that include tissue-specific autoimmunity. This review addresses the question of whether the mechanisms that drive ICI-induced irAEs are shared or distinct with those driving spontaneous autoimmunity, focusing on ICI-induced diabetes, ICI-induced arthritis and ICI-induced thyroiditis due to the wealth of knowledge about the development of autoimmunity in type 1 diabetes, rheumatoid arthritis and Hashimoto’s thyroiditis. It reviews current knowledge about role of genetics and autoantibodies in the development of ICI-induced irAEs and presents new studies utilizing single cell omics approaches to identify T cell signatures associated with ICI-induced irAEs. Collectively, these studies indicate that there are similarities and differences between ICI-induced irAEs and autoimmune disease and that studying them in parallel will provide important insight into the mechanisms critical for maintaining immune tolerance.

Introduction

The human immune system is carefully regulated to allow it to protect the body from pathogens and tumors while maintaining tolerance to self. Multiple molecular mechanisms participate in this balancing act, among them are a series of cell surface molecules found on T cells that are stimulatory or inhibitory. Inhibitory molecules blunt the response of T cells to stimulation and promote a state of exhaustion leading to ineffective responses upon recognition of self or benign foreign molecules. T regulatory cells (Tregs) utilize these molecules to suppress effector responses. The realization that tumors utilize these molecular interactions to evade immune defenses resulted in a revolution in cancer therapy, immune checkpoint inhibitors (ICI). However, with this new approach to treating cancer and selective blockade of the immune system’s regulatory mechanisms came a series of immune-related adverse events (irAEs), characterized by targeted inflammation in otherwise healthy tissues ranging from rashes, arthritis, endocrinopathies, enteropathy, and pneumonitis^1,2. These ICI-induced irAEs may target a single organ/tissue or can be systemic in nature, raising the question of why some individuals develop irAEs while others do not and what determines the type of irAE that an individual will develop. Studies thus far suggest that the underlying cancer is not a strong indicator of the type of irAE but the specific ICI treatment does appear to be important^1,2. Notably, irAEs have also been associated with therapeutic benefit in some studies^1,2, suggesting that the irAE and the therapeutic benefit that result from checkpoint inhibitor therapy may depend on the inherent qualities of the immune response of an individual that influences the frequency and functional phenotype of tumor specific and autoreactive T cells that respond to ICI.

Since ICI-induced irAEs represent a failure of immune tolerance, how ICI-irAEs relate to spontaneous autoimmunity and whether predicting and treating these irAEs can be informed by our understanding of autoimmune diseases has been an area of study since their introduction into the clinic. Understanding the similarities and differences between ICI-induced irAEs and spontaneous autoimmunity has the potential to shed light on the mechanisms that drive these diseases and identify ways to intervene. In this review, we will address the question of whether the mechanisms that drive ICI-induced irAEs are shared or distinct with those driving spontaneous autoimmunity, focusing on ICI-induced diabetes, ICI-induced arthritis and ICI-induced thyroiditis, which have an incidence of 0.2–1.9%³, 5%⁴ and 2–15%%³. We have chosen to focus on these irAEs based on the wealth of knowledge about their spontaneous autoimmune disease counterparts, type 1 diabetes (T1D), rheumatoid arthritis (RA) and Hashimoto’s thyroiditis, where the contribution of genetics and the immune alterations involved in the break in immune tolerance are well-defined for each of these autoimmune diseases.

Key steps in the loss of tolerance and development of autoimmunity

Autoimmune diseases are a broad category of diseases that share the common feature of immune responses that target “self” leading to inflammation and tissue injury that may be localized as is the case for T1D, where insulin-producing pancreatic beta cells are targeted, or systemic such as RA where inflammation extends beyond the joints. Multiple pathogenic mechanisms drive autoimmune diseases and develop due to a combination of genetic risk, environmental factors, and immune experience ultimately resulting in a loss of tolerance to self⁵ (Figure 1). The immune system is designed to protect an individual from foreign pathogens and tumors, but also to avoid inflammation that will cause injury and autoimmunity. This process involves multiple tolerance checkpoints, including the deletion of autoreactive T and B cells during their development, Tregs that actively suppress autoreactive effector T cells and the development of T cell anergy and exhaustion in the context of chronic stimulation⁵ (Figure 2A). Autoimmune diseases occur when one or more of these checkpoints are breached. Specifically, high affinity autoreactive T cells escape thymic selection, T cell development is skewed toward pathogenic lineages (Th1, Th17, Tfh, Tph) promoted by a proinflammatory cytokine milieu or cellular interactions, and there is a failure to control pathogenic T cells by either Tregs or co-inhibitory signals that blunt T cell activation and promote T cell exhaustion^5,6. Each of these failures may be due to genetic or environmental factors and typically require a combination of factors and tolerance failures over time to develop autoimmunity.

Figure 1. Factors promoting the development of autoimmunity.

Autoimmune diseases develop over time and due to a combination of factors. Individuals are born with genetic risk, environmental factors are thought to trigger the loss of tolerance and expansion of autoreactive T and B cells until finally due to additional triggers or immune experience inflammation leads to clinical disease.

Figure 2. Mechanisms involved in the maintenance and loss of immune tolerance.

(A) Mechanisms maintaining self-tolerance; autoreactivity is regulated at multiple points in the development of B and T cells. Central or deletional tolerance, B cell help provided by Tfh, blunted of responses to activation and suppression by Treg. Co-inhibitory receptor involvement is shown by gold stars in the diagram. (B) Impact of immune checkpoint inhibitor therapy on self-tolerance. Regulatory pathways inhibited by ICI therapy are shown by an “X”. Created using BioRender.com

ICI therapy and tolerance checkpoints

ICI therapy is designed to overcome these tolerance checkpoints through blockade of the co-inhibitory molecules CTLA-4, PD-1, PD-L1 and more recently LAG-3 (Figure 2B)⁷. Yet the development of irAEs due to these therapies is not universal or uniform in type. Understanding how genetics and the ways in which tolerance is lost in autoimmunity may shed light on the questions of why and how irAEs develop in some individuals and how we can prevent or treat these adverse events to enhance the treatment of patients with cancer. However, before we review the current literature addressing these questions, it is important to note that there are several substantial challenges in the field. First, immune variation between individuals is high in humans. Second, the ICI treatment landscape continues to rapidly change with the advent of new drugs and new tumor targets making development of study cohorts difficult and data interpretation complex. Additionally, it remains to be determined how prior interventions such as chemotherapy influence the immune response to ICI therapy. Third, to date, there are only a handful of longitudinal studies with samples collected before and after ICI therapy that directly compare to spontaneous autoimmunity thus it is unclear whether irAEs are associated with pre-existing preclinical autoimmunity. Last, there are also obstacles for lab-based studies including ICI treatment related-anemia and lymphopenia reducing both the volume and quality of the sample available for study. Another limitation is the lack of good murine models for ICI-induced irAEs.

The contribution of genetics to ICI-induced irAEs

The link between genes and autoimmunity is clear. Rare single gene mutations that result in autoimmunity⁸ have highlighted the importance of the tolerance checkpoints described above. However, the investigation of genetic risk in autoimmunity was initiated over 40 years ago based on the observation that autoimmune diseases frequently occurred within families. For example, first-degree relatives of individuals with T1D have 15 times greater risk than the general population^9,10. Studies of autoimmune disease in monozygotic twins also indicate strong genetic contribution but also highlight the importance of additional factors¹¹. Notably, in the setting of ICI-induced irAEs, the risk of irAEs is increased in individuals with pre-existing autoimmunity or a family history of autoimmunity indicating a link to the genetic risk factors that contribute to spontaneous autoimmunity in the development of irAEs^12,13.

The highly polymorphic HLA locus on chromosome 6 has the strongest genetic association with autoimmunity¹⁴ and includes HLA Class I and Class II alleles, which are central players in CD8 and CD4 T cell activation, respectively. The HLA Class II receptor is a heterodimer composed of α and β chains expressed on the surface of antigen presenting cells. Its function is to bind peptide antigens within a binding groove on its surface which are then presented to the T cell receptor (TCR) on CD4 T cells resulting in T cell activation. Thus, HLA Class II plays a central role in determining T cell selection in the thymus and in directing the specificity of the T cell responses in the periphery. The association of a specific HLA Class II allele(s) with autoimmune diseases has suggested that they influence the tissue specificity of that disease. Examples of these associations include T1D, where HLA DRB1*0401, *0301 and DQ*0302 carry the strongest association¹⁵, RA where a set of HLA DRB1 alleles with a shared binding motif (DRB1*0401,0404, 0405, 0408 0101, 1001) are strongly associated with seropositive RA¹⁶ and HLA DR3 and DR4 have been linked to autoimmune thyroid disease¹⁷.

The tissue specificity seen in irAEs has raised the possibility that HLA type may participate in the risk for inflammation in specific tissue sites. Importantly, the association of HLA type with irAEs is likely influenced by the geographic population, the type of irAE and potentially the cancer type. Studies that looked broadly at HLA across ICI-treated subjects with irAEs have not found global association with HLA but have found associations in the HLA Class I and Class II regions for specific irAEs^18,19. More informative are the studies that investigate specific irAEs. Cappelli et al.²⁰, examined class I and II HLA alleles associated with classic forms of inflammatory arthritis in patients with ICI-induced arthritis and observed an increase in the frequency of shared epitope alleles in ICI-induced arthritis similar to that seen in RA patients of European descent compared to control subjects. In ICI-induced diabetes, Stamatouli et al.²¹, found an increase in HLA-DR3 similar to that seen in T1D and a predominance of HLA-DR4, notably to an even greater degree than in a Caucasian T1D cohort and this has been confirmed in other studies^22–24. Notably, there is no evidence that alleles known to be protective in T1D are decreased^22,23. The associations of ICI-induced arthritis and ICI-induced diabetes with HLA DR4 and their occurrence in the context of anti-PD-1/PDL-1 therapy raise the possibility of a pathogenic link between HLA DR4 and PD-1 in the development of inflammation targeting these sites.

Studies of genetic risk have expanded with the ability to evaluate the entire genome through single nucleotide polymorphism (SNP)-based analysis or DNA sequencing. These studies have demonstrated the complexity of genetic risk in autoimmunity, with >100 genes associated with any one disease, and each variant contributing modestly to risk overall²⁵. Importantly, many risk genes are shared across autoimmune diseases^25,26. Broad studies of genetic risk factors for irAEs have been less extensively studied. A recent genome-wide association study (GWAS) by Groha et al.²⁷, using a large patient population inclusive of multiple cancer types and ICI therapies was performed to identify germline genetic variants associated with development of irAEs. This study characterized irAEs based on severity and identified three loci that reached genome-wide significance, including one near the gene encoding interleukin-7 (IL-7), a cytokine important in T cell survival; IL-7 blockade has been shown to impede PD-1 expression, impair effector T cells and reverse autoimmune disease²⁸. Groha et al.²⁷, then replicated the association between this IL7 variant and irAEs in three independent studies and demonstrated that patients carrying the IL7 variant had in increased lymphocyte stability, which could allow development of autoimmunity in this setting. Interestingly, in a separate study by Taylor et al.²⁹, ICI treated patients carrying this IL7 variant had increased IL7 expression on B cells at baseline prior to treatment and this phenotype was independently associated with irAEs. Further evidence for the role of genetic risk of autoimmunity in ICI-induced irAEs is provided by Khan et al.³⁰, who developed a polygenic risk score (PRS) utilizing a GWAS of Europeans with hypothyroidism or vitiligo and then applied it to a cohort of patients treated with anti-PD-L1 ICI therapy, finding that it could predict risk of ICI-induced irAEs and longer overall survival; note this was specific for Europeans and was not seen in non-European cohorts. Importantly the genes from the PRS that were retained in modeling of hypothyroidism irAE risk include BCL6 the transcription factor required for Tfh cell lineage commitment³¹, PTPN22 R620W variant associated with multiple autoimmune diseases involved in both B and T cell activation²⁶, and CTLA4 and CD69 both involved in T cell activation^7,32. This study demonstrated a strong link between the genetic risk of autoimmunity and the development of irAEs in the context of thyroiditis and further suggests that this genetic risk background may provide a survival benefit.

Autoantibodies and the development of irAEs

Many autoimmune diseases are characterized by the presence of autoantibodies. These autoantibodies may be targeted to the affected organ in the disease, as is the case for T1D where autoantibodies directed to pancreatic beta cell antigens (GAD65, zinc transporter 8, insulin, islet antigen 2) are present³³, and in Hashimoto’s thyroiditis where autoantibodies against thyroid peroxidase and/ or thyroglobulin are present³⁴. In contrast, other autoimmune diseases have autoantibodies directed toward more ubiquitous antigens such as rheumatoid factor (RF) and anti-citrullinated protein antibodies (ACPA) in patients with RA³⁵ and anti-nuclear autoantibodies (ANA) in systemic lupus erythematosus³⁶. Notably, many autoantibodies precede the development of autoimmunity^35,37, which has led investigators to ask whether individuals with irAEs have autoantibodies prior to treatment and if this could be used to predict irAEs prior to therapy. It has become clear from this body of work that pre-existing autoantibodies are predictive of organ-specific irAEs at least in the case of ICI-induced thyroiditis^38–40. This suggests that ICI therapy may accelerate progression to autoimmunity or overcome remaining tolerance mechanisms in individuals poised to develop autoimmunity. However, it is important to note that autoantibodies are not seen in all organ-specific irAEs; for example, islet autoantibodies have only been detected in approximately 50% of the patients with ICI-induced diabetes^21,22,41,42. Moreover, it is unclear whether these autoantibodies are present prior to ICI initiation. Future longitudinal studies will be very helpful in understanding whether patients with pre-existing autoantibodies might be at higher risk of developing ICI-DM.

Unlike rheumatoid factor (RF) and anti-cyclic citrullinated peptide (anti-CCP) positivity being very common in RA, most cases of ICI-inflammatory arthritis tend to be seronegative. Specifically, in their systematic literature review of case reports and case series, Ghosh et al.⁴³, found that only 9% of cases of ICI-related inflammatory arthritis were seropositive. In spontaneous Hashimoto’s thyroiditis, anti-thyroglobulin and thyroperoxidase antibodies are seen in ~90 – 95% of the cases but they are observed in only approximately 18–78% of cases of ICI-thyroiditis^3,34.

More broadly the question of whether autoantibodies overall predispose to irAEs in general has yielded conflicting results. Toi et al.⁴⁴, found that the presence of pre-treatment anti-nuclear autoantibodies (ANA), RF or anti-thyroid autoantibodies predicted irAEs overall, independent of organ type, suggesting that the presence of autoantibodies may be a harbinger of a predisposition toward irAEs. Yet there is also an observation that the development of autoantibodies after ICI treatment may portend more severe irAEs, this has been described in most depth in the context of ANA seroconversion⁴⁵. Lastly, there is also evidence that ICI influence the B cell compartment and that these changes may promote ICI-induced irAEs⁴⁶. Das et al.⁴⁷, observed a reduction in circulating B cells in patients with melanoma after ICI therapy and that this was associated with greater risk of ICI-induced irAEs as well as irAE severity. More recently, Gatto et al.⁴⁸, reported that the frequency of transitional B cells is higher in patients that develop ICI-induced arthritis compared to those who do not. Collectively, these studies indicate that the induction of irAEs may be seen in individuals who already have a break in B cell tolerance as demonstrated by pre-treatment autoantibodies, but that ICI therapy may also promote loss of B cell tolerance, potentially through the promotion of the Tfh-B cell axis, resulting in autoantibodies that contribute to disease or are evidence of the evolution of autoreactivity due to ICI therapy.

T cell signatures in ICI-induced irAEs

Although it is well established that T cells play a critical role in the pathogenesis of autoimmune disease, less is known about their role in ICI-induced autoimmunity. Evidence that T cells are involved in irAEs comes from case reports showing T cells infiltration in the primary irAE organ^49–53. For example, T cells have been found in synovium during ICI-induced arthritis^49,50 and the thyroid during ICI-induced thyroiditis^51,52. T cells have also been observed around the pancreatic islet patients with ICI-induced diabetes^54,55. Although suggestive that these T cells are contributing to the development of autoimmunity in irAEs, more work is needed in larger cohorts with longitudinal sampling and in-depth phenotypic characterization to determine if these are autoreactive T cells. This work has already begun, but it is challenging given the relative rarity of ICI-induced autoimmunity, the continually changing landscape of ICI therapy with development of new drugs and expansion to new tumor types, and the range in the timing of irAE onset, which can occur soon after the start of ICI therapy but also after therapy has stopped. Here, we highlight three recent longitudinal studies demonstrating how single cell omics is advancing our understanding of irAE-induced autoimmunity and providing insight into potential T cell biomarkers for predicting who will develop these irAEs.

The first study by Lozano et al.⁵⁶, examined ICI-induced irAEs in 71 patients with melanoma integrating mass cytometry (CyTOF) with single cell RNA-sequencing (scRNA-seq) and single cell TCR sequencing (scTCR-seq). ICI-induced irAEs were stratified by severity and did not exclude any type of irAE. This study found that the abundance of activated memory CD4 T cells and TCR diversity in peripheral blood at baseline prior to ICI treatment was associated with the development of severe irAEs. Notably, the authors then compared this baseline transcriptomic signature for risk of severe irAEs to publicly available signatures of autoimmune disease, specifically systemic lupus erythematosus and inflammatory bowel disease, and found that activated memory CD4 T cells were also associated with autoimmune disease. The authors conclude that patients who develop severe irAEs may have latent or preclinical autoimmunity.

The next study by Kim et al.⁵⁰, focused on a single irAE, specifically ICI-induced arthritis in 20 patients and integrated flow cytometry, single cell RNA sequencing (scRNA-seq), single cell T cell receptor sequencing (scTCR-seq), multiplex cytokine assays and in vitro functional assays. Tumor types include melanoma, renal cell carcinoma, non small cell lung cancer and neuroendocrine, and a variety of ICI were used for treatment. This study was also noteworthy because it included sampling at the primary tissue involved in the irAE with 8 patients having paired peripheral blood and synovial fluid samples. Key findings include the presence of CX3CR1^hi CD8 T cells and CX3CR^hi CD8 T in blood and synovial fluid respectively, suggesting a role of CD8 T effector cells in ICI-induced arthritis. They also identified an enhanced Th17 cell signature in ICI-induced arthritis after combined anti-CTLA-4 and anti-PD-1 therapy. This study identifies immune characteristics both shared and distinct from RA and highlights the role of CD8 T cells in ICI-induced arthritis, a feature observed by others⁵⁷.

In the last study to be highlighted Bukhari et al.⁵⁸, used scRNA-seq and CITE-seq to identify T cell signatures in 24 ICI-treated patients; 15 of these patients developed an irAE. The predominant tumor type was non small cell lung cancer and almost all patients were treated with anti-PD-1. There were a total of four patients with ICI-induced arthritis, four patients with ICI-induced thyroiditis, and seven patients with ICI-induced pneumonitis. Interestingly, and in contrast to Lozano et al.⁵⁶, they found that the type of irAE was dependent on the baseline T cell signature prior to ICI therapy. Compared to patients who did not develop an irAE, patients with ICI-induced arthritis had lower levels of CD8 central memory T cells at baseline, patients with ICI-induced thyroiditis had more CD4 Th17 cells at baseline and patients with ICI-induced pneumonitis had more CD4 Th2 cells at baseline. A similar finding was recently reported by Kim et al.⁵⁹, who determined through T cell profiling that severe ICI-induced irAEs clustered into four distinct subtypes. This finding that the T cell lineage driving tissue inflammation differs based on tissue type has implications for the risk of developing an irAE but also therapeutic implications. Identification and characterization of tissue-specific T cells driving irAEs could confirm and extend our understanding of these differences. Moreover, studies investigating the role of autoreactive T cells are still limited in the setting of irAEs and a head-to-head comparison of ICI-induced autoimmunity and autoimmune disease is needed to directly address the question about whether the role of T cells in the pathogenesis of ICI-induced autoimmunity is the same as spontaneous autoimmunity.

Treg phenotypes in ICI-induced irAEs

Given their important role in the maintenance of peripheral tolerance, there has also been an interest in understanding how ICI therapy influences Tregs. Grigoriou et al.⁶⁰, explored whether ICI therapy reprograms peripheral Tregs during cancer treatment and thus, contribute to the development of irAEs. They compared patients with melanoma receiving ICI who developed an irAE to those who did not and found no significant difference in Treg frequency but did find differences in the Treg phenotype. RNA sequencing revealed that the Treg transcriptomic profile in patients with irAEs was enriched in inflammatory pathways (leukocyte activation, inflammatory response, cytokine response, oxidative stress response, type 1 IFN signaling and IFNγ signaling), and furthermore this profile was observed across different cancer types and similar to that seen for Tregs from patients with autoimmune disease. These differences in the Treg transcriptomic profile suggest a loss of Treg stability and transformation of Tregs with pathogenic potential due to ICI- therapy.

Additional gaps in knowledge: triggers and resolution

Environmental factors are known to contribute to the development of autoimmunity. However, identification of these factors is challenging requiring access to at-risk cohorts with longitudinal sampling before and after development of autoimmunity accompanied by in-depth data on lifestyle, diet and environmental exposures linked to each sample. Such cohorts have been developed for autoimmune diseases including T1D, RA and SLE and have made seminal contributions to our understanding of the natural history of preclinical autoimmunity as well as providing insight into the role of environmental factors. For example, studies taking advantage of samples such as those obtained by the US military have begun to shed light on the early development of autoantibodies in RA⁶¹ and the importance of Epstein-Barr virus seroconversion in multiple sclerosis^{62 63}. Exploring the role of environmental factors is less challenging in the setting of ICI therapy and development of irAEs, as samples can be obtained prior to initiation of ICI therapy and irAE onset in a much shorter timeframe, weeks to months versus years for spontaneous autoimmunity. This work has provided strong evidence that the microbiome participates in the response to ICI both therapeutically and in the development of irAEs⁶⁴. It is also an area where studying ICI-induced irAEs may help expand our understanding of the role of environment in spontaneous autoimmunity. Additionally, further investigation is also needed to determine why some irAEs resolve when treatment ends, in contrast to autoimmune diseases where autoimmunity is chronic. Understanding how inflammation targeting self can come back under control could shed light on mechanisms of regaining tolerance.

Concluding statement

Manipulating the immune response to treat cancer has been a longstanding goal. Our understanding of the checkpoints that maintain immune tolerance has resulted in the development of inhibitors of those checkpoints and an important advance in cancer therapy. This gain has come with the cost of lost tolerance to self, a hallmark of autoimmunity and the development of irAEs. Although irAEs have clinical features similar to those of many autoimmune diseases, the immunologic changes found in irAEs share only partial similarity to spontaneous autoimmune diseases affecting the same organ. Importantly, these similarities and differences shed light on the role of co-inhibitory molecules in the failure of tolerance in autoimmunity and the downstream results of their dysfunction. Moreover, the similarities with immune features of autoimmune diseases has assisted in the selection of therapies to offset irAEs and their associated pathology^65–67. In summary, studying the immune mechanisms underlying ICI-induced irAE in parallel with spontaneous autoimmunity will advance our understanding of immune tolerance and drive the development of the next generation of immune checkpoint inhibitors for the treatment of both cancer and autoimmune disease.