Weighing Antitumor Immunity against Life-threatening Myocarditis from Immune-Checkpoint Inhibitors

Immune-checkpoint inhibitors (ICI) targeting the regulatory receptors CTLA-4 and PD-1 and the PD-1 ligand PD-L1 have produced impressive responses in diverse malignancies, yet these therapies are also limited by a wide range of inflammatory toxicities (1). These toxicities, collectively referred to as immune-related adverse events (irAE), can affect any organ system in the body. Although most irAEs are relatively mild, severe and even life-threatening toxicities can occur (1). Systemic glucocorticoids are first-line therapy for nearly all irAEs, though they are not always effective. ICI myocarditis, which often occurs with generalized myositis, is among the most dangerous toxicities from ICI therapy, with mortality rates around 50% (1–3). Retrospective data suggest that early high-dose systemic glucocorticoids are likely critical for improving outcomes in ICI myocarditis, and small studies have indicated a potential role for abatacept in refractory disease (3, 4). Building upon this work, Salem and colleagues established a prospective study of ICI myocarditis treatment (2). The study included a total of 40 consecutive patients with confirmed ICI myocarditis who were treated at the cardio-oncology unit at the Pitié-Salpêtrière University Hospital (2). The first 10 patients received guideline-based management that involved the use of systemic high-dose glucocorticoids with consideration of escalation to various forms of secondary immunosuppression in patients who did not respond to initial therapy. The mortality rate was 60%, well within the expected range for ICI myocarditis (2). The authors used animal models and clinical observations to propose a change in the treatment protocol to improve ICI myocarditis outcomes, adding screening for respiratory muscle failure, CTLA-4–Ig (abatacept), and the JAK1 and JAK2 inhibitor ruxolitinib as up-front therapy (2, 4). The subsequent 30 patients had a 3% overall mortality, suggesting a substantial treatment benefit associated with at least one of these interventions (2). Although this study is not a randomized controlled trial, these results are striking and will likely alter standard clinic practice despite study limitations. Nevertheless, establishing prospective clinical trials to determine optimal therapy for ICI myocarditis remains an urgent need. The ongoing ATRIUM trial (NCT05335928) is a multicenter trial investigating abatacept compared with placebo as add-on therapy to high-dose systemic glucocorticoids in patients with newly diagnosed ICI myocarditis and should answer whether abatacept is a critical element of the intervention implemented by Salem and colleagues.

Beyond the important clinical findings, this work adds to the growing body of evidence largely acquired in ICI colitis that implicates IFNγ signaling in the pathogenesis of immunotherapy toxicities (5–7). ICI colitis is among the most common severe toxicities from checkpoint inhibitors, and biopsies are relatively easy to acquire. Immune phenotyping from the colons of patients with ICI colitis identified colitis-associated populations of expanded CD8⁺ and CD4⁺ T cells that were not found in healthy control colons or in the colons of patients who had received immunotherapy but did not have colitis (5). These expanded populations expressed high levels of IFNγ, and IFNγ signaling was similarly one of the key defining gene sets identified in colitis-associated T cells and myeloid cells (5). Consistent with a critical role of IFNγ in ICI colitis, treatment with the JAK1 and JAK3 inhibitor tofacitinib, which inhibits signaling downstream of the IFNγ receptor, seems to have activity in treatment-refractory ICI colitis (6). Perhaps not surprisingly, IFNγ was also upregulated in a separate analysis of ICI dermatitis, linking this cytokine to two of the most common inflammatory toxicities from checkpoint inhibitors (7). Salem and colleagues used PD-1–deficient mice that were heterozygous for the gene encoding CTLA-4 as a model for spontaneous ICI myocarditis; gene ­expression profiling in the hearts of these mice identified multiple upregulated cytokine signaling pathways, prominently including JAK1, JAK2, and STAT1, the JAK/STAT pathway activated by IFNγ (2). These preclinical findings provided part of the rationale for testing a JAK1/JAK2 inhibitor in ICI myocarditis. Coupled with the findings from ICI colitis and dermatitis, Salem and colleagues’ preclinical findings suggest that inhibition of IFNγ signaling plays a central role in the apparent efficacy of their treatment strategy.

The apparent efficacy of JAK inhibitors capable of blocking IFNγ signaling in at least two irAEs suggests that this may be a more general treatment strategy for refractory ICI toxicities. These medications are easy to administer and act rapidly. For severe, refractory toxicities, JAK inhibitors may well save lives, as is suggested by Salem and colleagues’ findings (2); however, widespread adoption of JAK inhibitor therapy for irAEs comes with considerable risk. IFNγ signaling is one of the clearest signals of effective antitumor immunity following ICI therapy. This has been demonstrated in mouse models, in correlates of patients who respond to ICI therapy, and in analyses of ICI resistance mechanisms, where loss-of-function mutations in JAKs downstream of IFNγ have been found in patients with tumor outgrowth after initial ICI response (8). That Salem and colleagues did not observe a worsening of tumor outcomes is not surprising (2). The study included multiple tumor types with variable treatment histories (2). Without a matched control group, assessing even a relatively large impact on antitumor responses would be impossible. In addition, ICI myocarditis has been associated with strong antitumor responses (2). Patients without residual tumor cells or who have lost all but a small fraction of relatively quiescent cells may not require further antitumor responses to achieve a durable remission. Yet this is hardly the rule for most patients on ICI therapy, or most patients who develop an ICI toxicity. Long-lasting partial responses represent a large fraction of the favorable clinical outcomes from ICI therapy, and the loss of a crucial element of antitumor immunity could jeopardize those responses (8). In the absence of high-quality, randomized, controlled studies that are sufficiently large enough to provide meaningful clinical outcomes, these basic science studies should outweigh retrospective and uncontrolled prospective clinical data, and JAK inhibitors should be reserved for life-threatening toxicities (8).

Although the data are not as clear as for IFNγ signaling, ongoing CD28 signaling appears to be important in the mechanism of action of both CTLA-4 and PD-(L)1 blockade (1, 9). These signals may be delivered directly in the tumor microenvironment through activated antigen-presenting cells such as cDC1s. Recruitment of naive T cells to broaden anti­tumor responses may also be important. Both of these functions of CD28 would be inhibited by abatacept, particularly if the drug levels are titrated to high receptor occupancy, potentially interfering with optimal antitumor responses (2). For immediately life-threatening toxicities such as ICI myocarditis, the risk of interfering with antitumor immunity is outweighed by the urgent need to control inflammation in the heart. Few toxicities require this degree of rapid response (1).

That ICI colitis, dermatitis, and myocarditis all appear to involve similar cytokine signaling pathways suggests that these toxicities may also have similar cellular origins (Fig. 1). ICI colitis appears to result from activation of colonic CD8⁺ resident memory T cells (TRM; refs. 1, 5). These cells are a normal component of the colonic mucosa and are found throughout the barrier organs, including the lung and skin. The evidence that TRMs are involved in the inflammation in ICI colitis comes from several observations. The CD8⁺ T cells that are expanded in ICI colitis also express the integrin CD103, a marker for resident memory T cells (1, 5, 6). In addition, clonotype tracking using single-cell RNA sequencing found that colitis-associated CD8⁺ T cells frequently share TCRs with cells in the TRM pool, indicating a common origin for these cells (1, 5). These findings are supported by RNA velocity calculations that found evidence that CD8⁺ TRMs were indeed differentiating into the expanded effector populations found in ICI colitis (1, 5). Do TRMs then cause ICI myocarditis as well? A healthy heart does not contain a population of physiologic TRMs the way that the colon and skin do (1). Yet perhaps the patients who develop ICI myocarditis represent a minority of patients who do have TRMs within the myocardium. A recent report using the same mouse model as Salem and colleagues, as well as samples from humans with ICI myocarditis, found that the self-protein α-myosin was recognized by cardiac-infiltrating T cells in ICI myocarditis (10). Potentially in patients who go on to develop ICI myocarditis, these T cells are present prior to the onset of inflammation, held in check by PD-1 and or CTLA-4. In contrast to the TRMs at barrier organs, which likely recognize microbial and environmental targets, T-cell recognition of a widely expressed self-protein like α-myosin should occur rarely because of the relative efficiency of negative selection in the thymus (1). Even when these T cells escape negative selection, peripheral tolerance mechanisms should remain active, though perhaps this tolerance is broken by prior cardiac infections or other inflammatory stimuli.

As we characterize the cells responsible for ICI myocarditis and their targets in further detail, we may eventually be able to detect them in the peripheral circulation. This could allow for better prediction of ICI myocarditis risk and even earlier intervention. Determining how to balance optimal first-line therapy for ICI myocarditis with optimal antitumor immunity will require both large-scale clinical trials and new basic science efforts aimed at more completely characterizing these related immune responses.