Immune Responses in Checkpoint Inhibitor–Associated Cardiovascular Immune–Related Adverse Events

Abstract

Although immune checkpoint inhibitors have improved cancer outcomes, their use can trigger autoimmune toxicities affecting essentially any organ system. In the cardiovascular system, these immune-related adverse events can manifest as myocarditis, pericarditis, accelerated atherosclerosis, vasculitis, arrhythmias, and other conditions. These cardiovascular toxicities are driven by the loss of peripheral tolerance, but the underlying mechanisms remain incompletely understood. Recent murine and human studies, particularly in immune checkpoint inhibitor–associated atherosclerosis and myocarditis, have begun to reveal the cellular and molecular mediators. Disinhibited T cell function is a central driver, but other immune and nonimmune cell types have also been implicated. Multiple inflammatory and chemokine receptor signaling pathways are activated, offering potential targets for intervention. The growing mechanistic understanding of these cardiovascular toxicities has begun to inform new diagnostic and therapeutic strategies, though additional prospective trials are needed.

Central Illustration

Highlights

• •ICIs can cause a variety of cardiovascular toxicities through disinhibited T cell function.
• •CD8⁺ T cell actions and IFN-γ signaling are shared mediators of ICI-associated atherosclerosis and myocarditis.
• •Mechanistically distinguishing beneficial antitumor immune responses from cardiovascular toxicities may generate new therapeutic approaches.

Immunologic mechanisms are central mediators of many cardiovascular conditions. This recognition is longstanding; the role of inflammation in atherosclerosis was first proposed in the mid-19th century by the pathologist Rudolph Virchow.¹ The rapidly growing use of immune checkpoint inhibitors (ICIs) over the last 15 years has further highlighted the interplay between the immune and cardiovascular systems. ICIs are monoclonal antibodies that bind and inhibit the signaling of immune checkpoint molecules (such as PD-1, PD-L1, PD-L2, CTLA-4, or LAG3), which are key regulators of T cell activation and function.² This abrogation of inhibitory signals leads to increased T cell targeting of cancer cells and ultimately improved cancer outcomes. However, the broad disinhibition of T cells represents a loss of a key mechanism of peripheral self-tolerance, leading to an array of autoimmune side effects, termed immune-related adverse events (irAEs), that can affect essentially every organ system.³ These irAEs can affect the cardiovascular system, leading to myocarditis, pericarditis, accelerated atherosclerosis and vascular events, vasculitis, and arrhythmias, among others.⁴ Treatment usually relies on withholding ICI therapy (at least temporarily, but sometimes permanently) and administering immunosuppressive medications, though there are currently no rigorous prospective trial data to guide management.

The path to understanding the mechanistic basis for cardiovascular irAEs has been challenging, but steady progress has been made. The first detailed report of myocarditis associated with ICI therapy in 2016 noted increased tissue accumulation, activation, and clonal expansion of T cells in cardiac and skeletal muscle, highlighting the role of these cells in disease pathogenesis.⁵ Additional work in the subsequent decade has further characterized T cell responses and highlighted the roles of other immune and nonimmune cell populations.6, 7, 8 Pivotal roles for key inflammatory and chemokine receptor signaling pathways have also been demonstrated.9, 10, 11, 12, 13 Yet, despite these advancements, key questions remain: What mechanisms determine whether a patient will develop a cardiovascular irAE? What signaling pathways orchestrate the immune responses? What are the antigens driving the T cell responses? Answering these and other questions is necessary to fill the substantial gaps in our understanding and support efforts to improve diagnostic and therapeutic approaches.

This review discusses the current understanding of the mechanisms underlying cardiovascular irAEs. Focusing on atherosclerosis and myocarditis, the best-studied irAEs and those with the greatest clinical impact in cardio-oncology, we highlight recent insights into molecular and cellular mediators. Special attention is paid to the relationship of cardiovascular irAE biology to antitumor immunity. Finally, we consider how emerging findings can translate into new diagnostic strategies and therapeutic interventions.

The Immune Response in ICI-Associated Atherosclerosis and Acute Vascular Events

Landmark pharmacovigilance analyses first signaled that ICI cardiovascular toxicity extended beyond myocarditis.¹⁴ In a retrospective single-center cohort, ICI exposure was associated with an approximately threefold higher rate of atherosclerotic cardiovascular events and, in a 40-patient imaging substudy, progression of total thoracic aortic plaque volume increased from 2.1% to 6.7% per year after ICI initiation, an effect attenuated by concurrent statins or corticosteroids.¹⁵ This increased vascular risk has since been broadly reproduced in lung cancer and melanoma cohorts, aortic plaque-volume imaging studies, and propensity-matched analyses of arterial thromboembolic events.16, 17, 18 The central question this section addresses is mechanistic: why should agents that disinhibit antitumor T cells accelerate atherosclerotic disease that is itself fundamentally T cell–driven?

Immune checkpoints and inflammation in atherosclerosis

The unifying insight across laboratories is that PD-1, LAG3, and CTLA-4 are not merely mediators of tumor tolerance, but rather are established negative regulators of the adaptive immune response within the artery wall. Atherosclerosis is a chronic inflammatory disease in which CD4⁺ T helper 1 cells and cytotoxic CD8⁺ T cells recognize plaque antigens (oxidized low-density lipooprotein–derived epitopes), secrete proinflammatory interferon γ (IFN-γ) and tumor necrosis factor (TNF), and license macrophage activation, whereas FOXP3⁺ regulatory T cells (Tregs) restrain this response.¹⁹ The same coinhibitory receptors that tumors exploit normally limit intraplaque T cell activity, and genetic loss-of-function studies support this role. The PD-1 axis was implicated early: compared with Ldlr^−/− (atherosclerosis-prone) control mice, Ldlr^−/−Pdcd1^−/− mice (lacking Pdcd1, which encodes PD-1) developed larger plaques with more abundant CD4⁺ and CD8⁺ T cells and macrophages (Figure 1), whereas administration of a blocking anti-PD-1 antibody increased lesional inflammation.²⁰ PD-L1/PD-L2 deficiency on an Ldlr^−/− background produced a roughly 2- to 3-fold increase in aortic plaque burden with marked CD4⁺ and CD8⁺ T cell and macrophage accumulation and higher serum TNF-α.²¹ Additionally, despite an expansion of Tregs, Ldlr^−/−Pdcd1^−/− mice displayed exaggerated systemic CD4⁺ and CD8⁺ activation and massive lesional T cell infiltration in larger plaques, indicating that effector disinhibition outpaces regulatory containment.²² CD8⁺ T cells are themselves proatherogenic, promoting monopoiesis and lesional macrophage accumulation.²³

Figure 1.

Immune Checkpoints Restrain Atherosclerosis Across Preclinical Models and Patients

(A) Altered immune checkpoint signaling impacts atherosclerotic lesion size across mouse models. All point estimates shown reflect statistically significant changes; rows without point estimates indicate nonsignificant changes in lesion size. Several short-term antibody blockade regimens leave plaque size unchanged despite increasing plaque inflammation. Cd273 and Cd274 encode the PD-1 ligands PD-L2 and PD-L1, respectively. (B) Diminished immune checkpoint signaling increases the numbers of intraplaque effector and regulatory T cells.¹¹⁴ Conversely, CTLA-4 overexpression lowers plaque CD4⁺ T cell content. Shown are fold changes in intraplaque cell number for each indicated population relative to controls in the context of the indicated models. (C) Shown are annual rates of aortic plaque progression in clinical cohorts; yellow circles denote comparator populations or pre–immune checkpoint inhibitor (ICI) values, as applicable, and purple circles denote ICI-treated values. Throughout the figure, numbers in square brackets denote relevant references.

For CTLA-4, the data are concordant: constitutive T cell CTLA-4 overexpression in Apoe^−/− mice reduced lesion size and plaque macrophage and CD4⁺ content,²⁴ whereas antibody-mediated CTLA-4 blockade in Ldlr^−/− mice drove T cell–dependent endothelial inflammation and larger plaques.²⁵ The therapeutic mirror image strengthens causal inference because pharmacologically augmenting the same brakes is atheroprotective. Abatacept is a CTLA-4-Fc fusion protein that binds CD80/CD86 on antigen-presenting cells and thereby blocks T cell costimulation, leading to T cell anergy; its administration reduced accelerated atherosclerosis by roughly 78% in hyperlipidemic ApoE3∗Leiden mice while curbing CD4⁺ T cell activation and reducing IFN-γ levels.²⁶ Correspondingly, an agonistic PD-1 antibody in Ldlr^−/− mice reduced T cell activation and IFN-γ–producing CD4⁺ cells and reduced plaque size by roughly 25%.²⁷ Thus, across genetic deletion, antibody blockade, and pharmacological agonism, immune checkpoints behave as atheroprotective brakes.

The costimulatory arm of the immune synapse is a powerful lever, and perhaps a more tractable one. CD28–CD80/86 and CD40–CD40L signaling are established drivers of atherogenesis, and blocking the downstream CD40–TRAF6 interaction in macrophages reduces atherosclerotic burden by roughly 80% in Apoe^−/− mice, chiefly by curtailing monocyte recruitment.²⁸ Critically, it does so without impairing antibody production or antigen-dependent T cell proliferation. Such selectivity makes this approach attractive in ICI-treated patients, in whom systemic immunosuppression could blunt the antitumor response. Confining the blockade to macrophages—for example, by packaging inhibitors in nanoparticles that are phagocytosed or by using bispecific antibodies—offers a way to restrain vascular inflammation while leaving antitumor immunity intact.

Additional work has characterized the mechanistic basis for ICI effects on atherosclerosis. Isolated CTLA-4 blockade in Ldlr^−/− mice produced a 2-fold increase in aortic arch lesion area alongside a larger necrotic core, reduced collagen and smooth muscle content, and drove a shift toward lymphoid-dominated inflammation.²⁵ Combined anti-PD-1/anti-CTLA-4 treatment in Ldlr^−/− mice left plaque size unchanged but remodeled its composition; lesional CD8⁺ T cells increased 2.7-fold, the necrotic core enlarged 3.9-fold, and endothelial vascular cell adhesion molecule 1 and intercellular adhesion molecule 1 increased 2.2-fold and 1.6-fold, respectively. The discordant findings are mechanistically important: isolated CTLA-4 blockade increased lesion size, whereas short-term combined blockade left it unchanged and instead remodeled plaque composition, indicating that whether ICIs enlarge plaques depends on the checkpoint targeted and the duration of exposure. Additionally, short-term combined anti-PD-1/anti-CTLA-4 treatment of Apoe^−/− mice did not increase ¹⁸F fluorodeoxyglucose (¹⁸F-FDG) uptake (a myeloid/macrophage-driven signal) in the aorta. An important caveat when interpreting these studies is that Ldlr^−/− and Apoe^−/− mice differ in lipoprotein profile, lesion size, and disease chronicity, and are not interchangeable models. Collectively, these findings suggest that the proximate effect of ICI treatment is lymphoid, T cell–driven plaque destabilization rather than acute expansion of bulk plaque or systemic myeloid inflammation.

In contrast, genetic models lacking PD-1 signaling demonstrated increased plaque size.^21,22 Additionally, CCR2–targeted positron emission tomography (PET) imaging in anti-PD-1–treated Apoe^−/− and Ldlr^−/− mice demonstrated accelerated, IFN-γ–dependent plaque inflammation with a prominent CCR2⁺ monocyte/macrophage component.¹³ Of note, an integrated single-cell classification spanning 12 mouse and 11 human datasets defined CCR2^int major histocompatibility complex class II⁺ inflammatory macrophages as a distinct subset conserved between species and separate from resident and TREM2^hi foamy macrophages.²⁹ The CCR2 signal therefore may highlight a defined recruited macrophage population rather than bulk myeloid burden. The different results across model systems are likely reconciled by the varying experimental contexts rather than true biological contradiction. Chronic genetic absence of an immune checkpoint from birth permits cumulative lesion growth, whereas short-term antibody blockade in established disease affects pre-existing plaque. Divergent results between these paradigms are therefore not necessarily contradictory. The apparent myeloid–lymphoid tension dissolves if IFN-γ from disinhibited T cells is the upstream event that secondarily recruits and activates CCR2⁺ macrophages. The common denominator is an IFN-γ–centered, T cell–initiated program.

The effect of short-term ICI treatment on vascular wall inflammation in humans remains unclear. In patients with melanoma after 6 weeks of ICI therapy, ¹⁸F-FDG uptake was not increased relative to baseline in the thoracic aorta, carotid arteries, spleen, or bone marrow. A directly contrasting clinical observation, however, comes from serial ¹⁸F-FDG PET in ICI-treated patients, in which arterial wall uptake increased significantly after therapy, particularly in noncalcified and mildly calcified segments, without any increase in splenic or bone marrow signal.³⁰ A controlled longitudinal study reinforced this finding, showing that arterial ¹⁸F-FDG uptake increased by 2.5% annually in ICI recipients compared with only 0.8% in those who did not receive ICI therapy (P = 0.001). The study authors nonetheless cautioned that the increase was modest relative to interindividual variability and questioned whether ¹⁸F-FDG uptake—a predominantly myeloid signal—can capture an ICI-induced, lymphocyte-driven process.³¹ Whether ICIs measurably increase human arterial FDG uptake therefore remains contested, likely reflecting differences in patient age, baseline plaque, imaging timing, and segment selection.

Mechanistic refinements: which T cells, and how

Recent work has begun to identify the cellular substrate of ICI action in the artery. PD-1⁺ T cells in murine atherosclerotic aortae are predominantly progenitor exhausted, rather than terminally exhausted; they retain proliferative capacity and produce IFN-γ in vivo.³² Meanwhile, circulating PD-1⁺ T cells in humans are also associated with subclinical coronary atherosclerosis.³² Single-cell profiling of human plaques independently established that the lesional T cell compartment is dominated by activated CD4⁺ and CD8⁺ populations that frequently express exhaustion markers, providing the human counterpart to these murine findings.³³ Single-cell RNA sequencing and mass cytometry atlases of mouse aortic leukocytes place these T cells within the wider lesional immune landscape.^34,35 Paired single-cell T cell receptor (TCR) sequencing of human plaque and blood adds an important qualification: effector CD4⁺ T cells undergo plaque-restricted clonal expansion and express markers of recent antigen encounter, indicating local, antigen-driven stimulation rather than passive recruitment.³⁶ This mirrors the progenitor-exhausted populations that mediate ICI responses in tumors and offers a unifying cellular target linking antitumor efficacy to vascular toxicity.

The downstream consequences map onto the classical determinants of plaque vulnerability: IFN-γ impairs smooth muscle collagen synthesis and promotes matrix metalloproteinase–mediated cap thinning; cytotoxic CD8⁺ effectors (granzyme/perforin), together with IFN-γ, drive macrophage and smooth muscle apoptosis, enlarging the necrotic core; and TNF/IFN-γ–mediated endothelial activation sustains leukocyte recruitment.⁶ The net structural phenotype—a larger lipid-necrotic core beneath a thinned, inflamed fibrous cap—is precisely the high-risk morphology that predisposes to rupture.^37,38

From accelerated atherosclerosis to acute vascular events

The clinical correlate of this biology is 2-fold. Chronically, ICIs accelerate plaque progression, as captured by aortic plaque-volume imaging.^15,16 Acutely, the same destabilizing inflammatory milieu can precipitate atherothrombotic events: cohort data associated ICI therapy with excess myocardial infarction, ischemic stroke, and other arterial thromboembolic events, with relative risks ranging from roughly 1.5-fold in meta-analyses to 3-fold in single-center cohorts.^15,38 Several non–mutually exclusive mechanisms have been proposed for the acute phenotype: rupture or erosion of an ICI-destabilized plaque; checkpoint blockade–driven endothelial dysfunction and a prothrombotic, inflammatory state; and, in some cases, immune-mediated vasculitis. The evidence for the last of these is least secure. Defective vessel wall PD-L1 with unopposed PD-1⁺ T cell activation is well documented in large-vessel vasculitides such as giant cell arteritis, but the same mechanism has not been demonstrated directly in ICI-associated vasculitis.³⁸ Pharmacovigilance data have independently confirmed overreporting of ICI-associated vasculitis, notably temporal arteritis arising early in therapy.¹⁴ Distinguishing accelerated atherothrombosis from primary vasculitis is clinically consequential because management diverges. It is equally important that cancer itself confers a prothrombotic state that can promote vascular events, so separating the ICI-specific contribution from tumor- and chemotherapy-related risk remains methodologically demanding.

Open questions and therapeutic implications

Key questions remain unresolved. First, the relative primacy of lymphoid vs myeloid drivers and whether the divergent arterial FDG findings reflect differences in timing, model, or imaging sensitivity remain unsettled.^30,37 Second, further data are needed to clarify whether ICIs principally increase plaque size, alter plaque composition and vulnerability, or both, and whether these depend on ICI treatment duration or on pre-existing atherosclerotic disease. Third, the benefits of aggressive cardiovascular risk factor optimization and now proposed management frameworks for ICI-treated patients remain unclear despite preclinical PD-1 agonism data and attenuation of plaque progression by statins and corticosteroids.^15,17,27 The causal contribution of inflammation to atherothrombosis and the viability of cytokine-directed therapy were established by the CANTOS (Canakinumab Anti-Inflammatory Thrombosis Outcome Study) trial, which showed reduced recurrent cardiovascular events among patients treated with the anti-interleukin-1β (IL-1β) antibody canakinumab.³⁹ This result also provides a template for IFN-γ axis–directed or IL-1β–directed strategies that might decouple vascular toxicity from antitumor efficacy.

Overall, a mechanistically coherent picture has emerged across laboratories: the checkpoints restrain an IFN-γ–driven, CD8-weighted intraplaque immune response, and their blockade unleashes it. Notably, cholesterol metabolism may emerge as a shared axis that shapes the response to checkpoint blockade and, through lipid-lowering therapy, offers a point of therapeutic leverage, even as the quantitative and contextual details continue to be refined.³⁸

Mouse Models and Mechanisms of ICI Myocarditis

Multiple mouse models have been created that reconstitute the immune landscape, cardiovascular remodeling, and overall clinical presentation of ICI myocarditis through genetic and pharmacological methods. Mouse models of ICI myocarditis can illuminate new therapeutic targets and disease mechanisms. In humans, approximately 1% of patients treated with ICI therapy develop ICI myocarditis, whereas incidence across murine models can be as high as 100% in genetic models and 20% to 90% across various pharmacological models.⁶ Despite the complications and methodological differences of genetic and pharmacological approaches (Figure 2), murine ICI myocarditis models generally recapitulate the clinical hallmarks of human ICI myocarditis. In this section, we summarize the most common murine models of ICI myocarditis, elaborating on their methodologies, fidelity to disease, and underlying molecular and cellular mechanisms.

Figure 2.

Shared and Distinct Features of Mouse Models of ICI Myocarditis

Certain features are shared by genetic and pharmacologic mouse models of immune checkpoint inhibitor (ICI) myocarditis, whereas others vary. Both model systems have yielded insights into the mechanistic basis of the disease. irAE = immune-related adverse event; ICIM = immune checkpoint inhibitor myocarditis.

Genetic models of ICI myocarditis

Mouse models of ICI myocarditis can be divided into genetic and pharmacological models as well as into tumor-bearing models. Whereas genetic models directly target genes encoding immune checkpoints, pharmacological models directly treat mice with different concentrations of monoclonal antibodies against PD-1, PD-L1, LAG3, and/or CTLA-4 (Table 1). Genetic models of ICI myocarditis involve global knockout of genes encoding immune checkpoint proteins, such as CTLA-4 (encoded by Ctla4), PD-1 (Pdcd1), PD-L1 (Cd274), or LAG3 (Lag3).⁴⁰ Importantly, the phenotype of ICI myocarditis in mice is heavily influenced by both the target genes and the genetic background. Because individual or paired deletions of Ctla4, Pdcd1, Cd274, and/or Lag3 across different mouse strains produce varied disease presentations, researchers should tailor mechanistic questions to the specific genetic model.

Table 1.

Genetic and Pharmacological Mouse Models of ICI Myocarditis

Mouse Strain/Genotype	Treatment	Phenotype	Reference(s)
Genetic models
BALB/c Pdcd1−/−	—	DCM without myocarditis	44
MRL Pdcd1−/−	—	70% mortality from ICI myocarditis by 10 wk. Accompanied by massive cardiac infiltration of CD8+ and CD4+ T cells, atrial thrombosis, and pleural effusions.	11,41,42
MRL Faslpr, Cd274−/−	—	Majority of mice develop lupus-like autoimmunity, pneumonitis, DCM, and myocarditis with cardiac infiltration of CD8+ T cells, leading to heart failure.	45
MRL Cd274−/−	—	Majority of mice develop lupus-like autoimmunity, pneumonitis, DCM, and myocarditis with cardiac infiltration of CD8+ T cells, leading to heart failure.	45
MRL Faslpr, Pdcd1−/−	—	A total of 20% of mice develop mild myocarditis accompanied by reduced cardiac CD8+ T cell infiltration and minor cardiac damage.	43
C57BL/6 Pdcd1−/−	—	Lupus-like glomerulonephritis alongside arthritis without myocarditis	41
C57BL/6 Ctla4−/−	—	Spontaneous lymphocytic infiltration in the heart, lungs, bone marrow, liver, and pancreas by 2 wk of age, early death resulting from myocardial failure.	48
C57BL/6 Ctla4+/−, Pdcd1−/−	—	Cardiac CD8+ T cell infiltration, arrhythmias, clonal expansion of α-myosin–specific CD8+ T cells with lethal myocarditis in 50% of mice.	48,61
C57BL/6 Pdcd1−/−Lag3−/−	—	Cardiac CD8+ T cell infiltration, arrhythmias, and increased troponin I/T and myocarditis, with 80% of mice dying prematurely by 10 wk.	12
Pharmacological models
C57BL/6	200 μg anti-PD-1 3 times a week	Mild myocarditis accompanied by minor systolic dysfunction but without CD8+ T cell infiltration.	50
C57BL/6	400 μg anti-PD-1 every other day	Acute myocarditis with CD8 T cell infiltration, systolic dysfunction, cardiomyocyte damage, and arrhythmia.	52
MRL	400 μg anti-PD-1 + 400 μg anti-CTLA-4 twice weekly	A total of 90% of mice develop myocarditis with CD8 T cell and macrophage cardiac infiltration alongside increased serum troponin.	11
A/J	150 μg anti-PD-1 every 3 d	Multiorgan inflammation with immune infiltration in the heart, lungs, colon, and pancreas. A total of 20% of mice develop myocarditis.	53

DCM = dilated cardiomyopathy; ICI = immune checkpoint inhibitor; IL = interleukin.

Genetic absence of PD-1 yields varied results depending on the strain of mouse used. Murphy Roths large (MRL) mice with Pdcd1 deletion develop severe myocarditis with 67% disease penetrance by 4 to 6 weeks of age and clinical features similar to human ICI myocarditis, including elevated serum troponin levels, cardiomyopathy, pericardial effusions, and conduction abnormalities.^11,41,42 Using single-cell multiomic approaches, MRL Pdcd1^−/− mice showed increased numbers of intracardiac Cd8⁺ T cells expressing Cxcr3 and macrophages expressing Ccr2; the latter population was further characterized by expression of Cxcl9 and Cxcl10, which serve as ligands for CXCR3.¹¹ A mechanistic role for CXCR3⁺CD8⁺ T cells was supported by their ability to induce cardiomyocyte apoptosis in an in vitro coculture assay and by the diminished myocarditis severity observed after treatment with CXCR3-blocking antibodies. Furthermore, Cxcr3⁺Cd8⁺ cardiac T cells showed increased clonal expansion, indicating an antigen-driven response. Collectively, these findings implicate a model in which intracardiac CCR2⁺ macrophages secrete CXCL9 and CXCL10, which recruit and retain CXCR3⁺ CD8⁺ T cells that then clonally expand and cause direct myocardial damage.

In contrast to MRL mice, neither C57BL/6 Pdcd1^−/− nor BALB/c Pdcd1^−/− mice develop myocarditis. However, C57BL/6 Pdcd1^−/− mice developed lupus-like glomerulonephritis and arthritis, whereas BALB/c Pdcd1^−/− mice developed dilated cardiomyopathy and dies prematurely.^43,44 By contrast, C57BL/6 Pdcd1^−/−Lag3^−/− mice develop severe myocarditis and die prematurely.¹² Notably, monoallelic Lag3 loss is not sufficient to induce ICI myocarditis but must occur in either Ctla4^+/− or Pdcd1^−/− mice to recapitulate the hallmarks of the disease. Pdcd1^−/−Lag3^−/− mice demonstrated increased numbers of intracardiac T cells expressing CXCR6, which binds to CXCL16.¹² Consistent with this finding, CXCL16⁺ macrophages were also enriched in these hearts. Abrogation of CXCR6 signaling with antibodies led to marked improvements in histologic severity, arrhythmia burden, and animal survival, supporting a mechanistic contribution of CXCR6 to myocarditis pathogenesis in this model.

Targeted deletion of PD-L1, encoded by Cd274, can also result in a myocarditis-like phenotype. MRL Fas^lpr mice possess a lymphoproliferation mutation in Fas that causes markedly increased T cell and B cell proliferation, ultimately leading to lupus-like autoimmune dysfunction. Homozygous Cd274 deletion in MRL Fas^lpr mice causes fulminant myocarditis with pneumonitis and subsequently leads to dilated cardiomyopathy and heart failure.⁴⁵ Cd274 deficiency in wild-type MRL mice also results in a lethal myocarditis phenotype, suggesting that disease onset is not entirely dependent on the Fas^lpr mutation.^45,46 MRL Cd274^−/− and MRL Fas^lpr Cd274^−/− mice both demonstrate 100% penetrance of myocarditis by age 2 months, exceeding the 70% penetrance seen in MRL Pdcd1^−/− mice.^11,41,45 MRL Fas^lpr Pdcd1^−/− mice develop myocarditis less consistently, with only 20% developing disease.⁴²

In contrast to Pdcd1^−/− mice, a complete homozygous knockout of Ctla4 has not been used to model ICI myocarditis because Ctla4^−/− mice on either C57BL/6 or BALB/c background die at 2 weeks of age from fulminant multiorgan autoimmune disease.⁴⁷ Of note, Ctla4^+/− mice have not been reported to develop systemic lymphoproliferation unless crossed with another genetic knockout associated with autoimmunity; Ctla4^+/− Pdcd1^−/− mice often develop multiple concurrent irAEs, including myocarditis.⁴⁸ Using Ctla4^+/−Pdcd1^−/− mice, Ma et al⁴⁹ found features similar to those in the MRL Pdcd1^−/− model: increased intracardiac CD8⁺ T cells and CCR2⁺ macrophages, with the latter expressing high levels of the chemokines Cxcl9 and Cxcl10. This study implicated IFN-γ signaling because IFN-γ blockade blunted macrophage expansion and attenuated disease. Cell-cell communication analysis predicted a bidirectional feedback loop that largely replicated the model suggested by data from MRL Pdcd1^−/− mice, though with the addition of CD8⁺ T cell–derived IFN-γ utilizing JAK1/2-dependent signaling in CCR2⁺ macrophages to induce CXCL9/CXCL10 secretion.

Pharmacological models of ICI myocarditis

Direct treatment of mice with antibodies that block immune checkpoint molecules has yielded models that recapitulate the origin and progression of human ICI myocarditis. Although these models all rely on direct ICI treatment, their outcomes vary by mouse strain. Treating C57BL/6 mice with anti-PD-1 and/or anti-CTLA-4 antibodies can produce a range of cardiotoxic effects depending on concentration and dosing of ICIs, with about 33% of treated mice developing some form of myocarditis. Administration of 200 μg of anti-PD-1 antibody 3 times a week resulted in mild myocarditis phenotype, accompanied by mild left ventricular systolic dysfunction but lacking significant CD8⁺ T cell infiltration.⁵⁰ Increasing the dose to 250 μg or 400 μg every other day resulted in more substantial CD8⁺ T cell infiltration concurrent with decreased LVEF.^51,52 Taken together, these studies demonstrate the significant role that dosage and ICI concentration have on the initiation of myocarditis.

A/J mice also develop myocarditis in response to anti-PD-1 antibody, with cardiac infiltration of CD8⁺ and CD4⁺ T cells alongside proinflammatory monocytes, macrophages, and natural killer cells.⁵³ Interestingly, these mice displayed preserved cardiac function despite histopathological confirmation of ICI myocarditis. Whereas anti-PD-1 treatment alone resulted in 20% incidence, the addition of anti-CTLA-4 or IL-12 increased the incidence to 35% and 50%, respectively, further supporting the influence of treatment on myocarditis incidence.

A new question emerging from mechanistic studies and targeted therapy development is whether tumors should be incorporated into murine models to more accurately mimic patients with ICI myocarditis. Consequently, the field has begun developing tumor-bearing mouse models to identify and test clinical targets that are cardioprotective without disrupting the antitumor effects of ICIs; mice models of melanoma, breast cancer, and colorectal cancer across several genetic backgrounds have been used.^40,46 ICI myocarditis can be induced in these models as in non–tumor-bearing mice, with the added advantage of simultaneously assessing ICI myocarditis biology and antitumor immunity.

Human ICI Myocarditis Mechanisms and Immune Responses

ICI myocarditis histology across the severity spectrum

The foundational clinicopathological description of ICI myocarditis by Johnson et al⁵ reported on 2 fatal cases of fulminant myocarditis in patients with melanoma receiving combination ipilimumab/nivolumab therapy. Postmortem immunohistochemistry revealed a dense lymphohistiocytic infiltrate composed predominantly of CD8⁺ and CD4⁺ T cells and CD68⁺ macrophages. Injured myocytes expressed PD-L1, a finding that has been recapitulated in several larger cohort studies of ICI myocarditis.54, 55, 56, 57, 58 Increased myocardial PD-L1 expression in response to injury has also been described in non-ICI myocarditis, including acute cellular allograft rejection and viral or autoimmune myocarditis, suggesting a conserved mechanism of defense against immune-mediated myocyte destruction.^56,59,60 A subset of PD-1⁺ T cells in ICI myocarditis were granzyme B⁺, providing early biopsy-level evidence that cardiac-infiltrating T cells are cytotoxically primed in situ, a discovery that has been supported by transcriptomic studies.^54,61,62 These early observations established a now reproducible histopathological template characterized by a T cell–rich and macrophage-rich infiltrate, PD-L1 upregulation on injured cardiomyocytes, and evidence of in situ T cell activation (Figure 3). In a retrospective analysis of 113 endomyocardial biopsy specimens from patients with ICI myocarditis, fewer than 10% had detectable viral pathogens, suggesting that ICI myocarditis is a distinct, primarily autoimmune disease, rather than a variant of viral myocarditis.⁶³

Figure 3.

Immune Responses and Pathogenic Mechanisms in Human ICI Myocarditis

(A) Typical features for each tier of clinical severity in immune checkpoint inhibitor (ICI) myocarditis are shown; some patients may have features from more than one tier. (B) The cardiac immune microenvironment of ICI myocarditis. Clonally expanded CD8⁺ T effector memory cell re-expressing CD45RA (TEMRA) cells engage PD-L1⁺ cardiomyocytes through granzyme B–mediated cytotoxicity. In human ICI myocarditis tissue, a CXCL9⁺CXCL10⁺ macrophage population and CXCR3⁺ and CXCR6⁺ T cells have been identified. The depicted interferon γ (IFN-γ)–CXCL9/CXCL10–CXCR3 and CXCL16–CXCR6 signaling relationships are based on mechanisms established in mouse models. Conventional dendritic cells (cDCs) and inflammatory fibroblasts upregulate antigen-presentation machinery, potentially sustaining local T cell activation. (C) The antigen-specificity landscape in ICI myocarditis. α-Myosin heavy chain is an established cardiac autoantigen, whereas the contribution of tumor-shared antigens remains uncertain. Ag = antigen; LVEF = left ventricular ejection fraction; MHC = major histocompatibility complex; TCGA = The Cancer Genome Atlas; TCR = T cell receptor; ULN = upper limit of normal.

ICI myocarditis does not present as a single clinicopathological entity. Champion and Stone⁵⁶ demonstrated that ICI myocarditis occurs across a histological severity spectrum, with high-grade cases showing dense CD8⁺ T cell and macrophage infiltrates and low-grade cases displaying comparatively sparse inflammation. High-grade cases also displayed more frequent and abundant PD-L1⁺ macrophages and myocytes. Palaskas et al⁵⁸ formalized a 3-tier grading system (grades 0-2) from a retrospective analysis of ICI myocarditis endomyocardial biopsy specimens, correlating histological grade with clinical outcomes. They further demonstrated that a subset of patients with low-grade myocarditis could continue ICI therapy without immunosuppression, which has important implications for rechallenge decision making.^58,64 This framework was recently extended by evidence that myocardial PD-L1 expression on biopsy correlates with adverse prognosis, positioning it as a potential prognostic biomarker and mechanistically linking greater inflammation with disease severity.⁵⁵

Most early descriptions of ICI myocarditis involved severe, often fulminant cases associated with electrical abnormalities; however, ICI myocarditis can also present as “smoldering” disease, characterized by biopsy-confirmed myocardial inflammation with troponin elevation but preserved hemodynamic function and the absence of life-threatening arrhythmias.⁶⁵ Smoldering myocarditis can present after ICI discontinuation, unlike severe cases that usually occur within 2 months of ICI initiation.⁶⁶ Histologically, fulminant and smoldering ICI myocarditis share the same fundamental histological substrate: an infiltrate dominated by CD8⁺ T cells and CD68⁺ macrophages; however, a small case series suggested less cardiomyocyte death on histology in smoldering myocarditis.^58,66

It is possible that intact regulatory mechanisms, such as residual checkpoint signaling, Treg responses, or higher activation thresholds, constrain disease progression in a subset of patients. Smoldering disease could also reflect differences in the magnitude and self-amplifying capacity of the cardiac immune response. In fulminant disease, the T cell–macrophage IFN-γ amplification circuit described by Ma et al⁴⁹ appears to operate at full capacity. In smoldering disease, this amplification cascade may either fail to reach a self-sustaining threshold (possibly because of lower antigen load, lower T cell avidity, or more intact regulatory mechanisms) or become engaged but never crosses the threshold of hemodynamic compromise before resolution or clinical detection and treatment. Whether these represent distinct immunological states or simply differences in timing and antigen exposure remains unresolved.

T cell–centric pathogenesis

Four complementary lines of human evidence converge on clonal, cytotoxic CD8⁺ T cells as the dominant pathogenic effectors in ICI myocarditis. Zhu et al⁴¹ performed the first comprehensive single-cell immunophenotyping of peripheral blood from patients with ICI myocarditis, identifying a clonally expanded CD8⁺CD45RA⁺ TEMRA (T effector memory cells re-expressing CD45RA) population with an upregulated cytotoxic program and elevated CCL4/CCL5 chemokine expression consistent with active cardiac homing.⁴¹ Longitudinal profiling in that study showed that these clones transitioned from an activated/cytotoxic phenotype to an exhausted phenotype after glucocorticoid treatment, providing a cellular correlate of therapeutic response.

Second, Axelrod et al⁶¹ used single-cell RNA sequencing and TCR sequencing of cardiac immune infiltrates from the Ctla4^+/−Pdcd1^−/− ICI myocarditis mouse model to establish clonal CD8⁺ T cell dominance. Human data complemented this finding, showing that peripheral blood T cells stimulated with alpha myosin heavy chain (α-MHC) shared TCR clonotypes with diseased cardiac and skeletal muscle infiltrates.⁶¹ This provided patient-level evidence of antigen-specific, cardiac-homing CD8⁺ T cells. Other independent studies have demonstrated that cardiac myosin-specific T cells are present in naïve A/J mice and are expanded and activated by anti-PD-1 monotherapy alone, corroborating the cardiac myosin autoantigen model in a second independent preclinical system without requiring tumor inoculation or exogenous immunization.⁵³

Third, a single-cell RNA sequencing analysis confirmed that cytotoxic CD8⁺ T cells are enriched in human myocarditis cardiac tissue.⁶² This study identified a circulating, cycling CD8⁺ T cell population bearing cardiac-expanded TCR clones as a novel peripheral biomarker of fatal disease outcome. Finally, CXCR3⁺ and CXCR6⁺ T cells are present in cardiac tissue from patients with ICI myocarditis, consistent with the roles of these receptors in T cell signaling established in mouse models.^11,12,49

Non–T cell mediators: Macrophages, dendritic cells, and fibroblasts

Whereas CD8⁺ T cells are the dominant effectors, the ICI myocarditis cardiac microenvironment is a multicellular ecosystem in which macrophages, dendritic cells, and fibroblasts also play active roles. Blum et al⁶² identified enrichment of activated macrophages, conventional dendritic cells, and a transcriptionally distinct fibroblast subset in cardiac tissue from patients with ICI myocarditis, with proinflammatory transcriptional programs upregulated across all enriched nonlymphoid cell types. Notably, multiple nonimmune cell subsets upregulated antigen-presentation machinery, as may be expected in an IFN-rich microenvironment, suggesting that local nonprofessional antigen presentation within the myocardium may help sustain T cell activation in a manner resembling tertiary lymphoid tissue.

The macrophage compartment has been specifically dissected by Ma et al,⁴⁹ who identified an inflammatory CCR2⁺ monocyte-derived macrophage subpopulation highly expressing Cxcl9 and Cxcl10 transcripts in the Ctla4^+/−Pdcd1^−/− mouse model. A cognate CXCL9⁺CXCL10⁺ macrophage population was observed in human ICI myocarditis biopsy specimens and was enriched compared with other forms of heart failure and myocarditis.^11,49 Available data suggest that the humoral immune response plays a comparatively minor role in ICI myocarditis. B cells are largely absent from endomyocardial biopsy and autopsy specimens from patients with ICI myocarditis, and deposition of complement components is observed only in necrotic myocytes in high-grade cases, arguing against antibody-mediated cardiomyocyte destruction as a primary or at least an early mechanism.^5,56,58

The antigen specificity debate

Initial clinical evidence that cardiac autoimmunity might be triggered by tumor-expressed antigens came from bulk TCR sequencing of two fatal ICI myocarditis cases, which revealed clonal T cell populations shared across myocardium, skeletal muscle, and tumor tissue.⁵ Bulk RNA sequencing of those same specimens identified the expression of multiple striated muscle transcripts, including MYH6 (encoding α-MHC), in the tumors, providing the initial basis for the tumor-shared antigen hypothesis. Later work by Axelrod et al⁶¹ identified 3 major histocompatibility complex class I–restricted TCRs from the cardiac infiltrate of Ctla4^+/−Pdcd1^−/− ICI myocarditis mice that responded to α-MHC peptides. Peripheral T cells from 3 ICI myocarditis patients expanded in response to α-MHC peptide stimulation, and 1 clone present in both blood and the heart was confirmed to recognize α-MHC_443-451. Intriguingly, this peptide sequence is identical in the α-MHC and β-MHC isoforms. This study also revealed that MYH6 transcripts were detectable in approximately 70% of melanoma specimens, corroborating a mechanistic bridge between tumor antigen priming and cardiac autoimmunity.

In a subsequent study of 8 patients with ICI myocarditis, Blum et al⁶² failed to detect any heart-enriched TCR clones recognizing α-MHC, troponin I, or troponin T. Methodological differences may explain this discrepancy. The α-MHC–specific T cells in the Axelrod et al⁶¹ study were not the most clonally abundant population in either the mouse model or the human data. In contrast, the Blum et al study focused on the most clonally expanded TCRs within the cardiac infiltrate. α-MHC–reactive T cells may participate in disease pathogenesis as part of a broader polyclonal effector response without being quantitatively dominant component.

The question of whether tumor-expressed antigens contribute to ICI myocarditis remains open. In the initial bulk TCR sequencing studies, all TCRs were considered when searching for overlapping clones across heart and tumor.⁵ This strategy cannot distinguish antigen-driven from bystander-expanded clones. Blum et al⁶² refined this approach by excluding peritumoral TCR clones, finding only 5 clones enriched in both heart and tumor, with the most expanded cardiac clones absent from paired tumor tissue. However, T cells primed in tumor-draining lymph nodes must traffic through peritumoral tissue en route to the tumor mass, meaning that this exclusion strategy may inadvertently remove those clones actively cycling between periphery and tumor. Further, a modest but nonzero overlap also remained after filtering, which could reflect an immunological trigger preceding epitope spreading rather than coincidence.

Definitive resolution will require direct experimental testing in controlled preclinical models. A tractable approach would be to enforce MYH6 expression in mouse tumors and determine whether this precipitates or accelerates myocarditis in the context of checkpoint blockade. In humans, existing data are limited to transcript-level MYH6 detection in bulk RNA sequencing of melanoma specimens and smaller institutional cohorts. Whether α-MHC protein is expressed across a broader range of tumor types, which cells within tumors express it, and whether expression is sufficient to prime T cells that cross-react with cardiac antigens remain unknown. Resolving these questions would substantially strengthen or weaken the tumor-shared antigen model.

Finally, it is worth acknowledging that α-MHC is almost certainly not the only relevant autoantigen in ICI myocarditis. The failure to identify additional candidates may reflect a fundamental limitation of how antigen discovery has been approached, rather than a true absence of targets. Potential targets include untested conventional cardiac autoantigens or post-translationally modified self-peptides (including citrullinated, adducted, carbamylated, or phosphorylated peptides) that generate cryptic neoepitopes absent from standard peptide libraries.67, 68, 69 Unbiased antigen discovery will be required to map the full antigenic landscape of ICI myocarditis and determine the true breadth of what the immune system is recognizing in this disease. Defining the full repertoire of autoantigens may allow the development of antigen-specific immunotherapies that selectively suppress the pathogenic cardiac immune response without the broad immunosuppression associated with current corticosteroid- and abatacept-based regimens, which risk compromising antitumor T cell responses. Antigen-specific tolerance strategies, such as tolerogenic nanoparticles or regulatory T cell induction targeting validated cardiac autoantigens, would represent a more precise therapeutic approach than systemic immunosuppression and could enable effective treatment of ICI myocarditis without compromising the oncologic responses to ICI therapy.

Shared vs Distinct Mechanisms of Cardiovascular irAEs and Antitumor Immunity

Association of cardiovascular irAEs and clinical antitumor outcomes

Multiple studies across tumor types have reported improved progression-free survival and overall survival in patients who develop irAEs.70, 71, 72 This observation suggests at least partly shared immunological mechanisms of efficacy and toxicity. A central challenge in the field is to identify distinct mechanisms that will open avenues to decouple autoimmune toxicity from antitumor efficacy. Many irAEs, including myocarditis, are treated with systemic corticosteroids to reduce immune-mediated inflammation.^73,74 However, systemic corticosteroids can compromise clinically important antitumor responses. In a recent retrospective analysis of nearly 40,000 patients, corticosteroid use was consistently associated with reduced overall survival, with stronger associations when corticosteroids were initiated earlier after ICI therapy, given at higher doses, and continued for longer durations.⁷⁵ This work highlights the importance of developing therapeutic strategies that selectively target toxicity without compromising the antitumor response.

Shared mechanisms: IFN-γ and CXCR6/CXCL16 signaling

IFN-γ signaling is essential for antitumor immunity following ICI therapy but, as previously noted, is also central to the cardiac immune response in ICI myocarditis.^11,49,76 Therapeutic approaches targeting this immune response have been studied, including abatacept and ruxolitinib, the latter of which is a JAK1/2 inhibitor that disrupts cytokine signaling.^48,62,77,78 Given that IFN-γ signaling is required early for antitumor T cell priming but may be dispensable for ongoing effector function, delayed ruxolitinib has been investigated as a means to block toxicity without impairing antitumor efficacy. Whereas concurrent JAK inhibition with PD-1 blockade has been found to suppress antitumor immunity, recent studies have shown that delayed JAK inhibition may instead enhance ICI activity.^79,80 Mechanistically, delayed JAK inhibition is thought to enhance ICI activity by promoting expansion of progenitor CD8 T cells, which have greater proliferative potential in response to ICIs than their terminally exhausted counterparts. Whether this concept also applies to ICI myocarditis, given its dependence on JAK signaling, remains to be determined. It could be hypothesized that ruxolitinib administered at the time of myocarditis diagnosis (ie, sequential therapy) may mitigate toxicity without abrogating ICI efficacy, whereas earlier “pre-emptive” JAK inhibition may not preserve ICI efficacy; further work is needed to test this concept.

Several studies have implicated the CXCL16-CXCR6 axis in protective antitumor immunity.81, 82, 83, 84 CXCR6 positions cytotoxic T cells to receive critical survival signals in the tumor microenvironment.⁸⁵ It will be important to determine whether blocking this axis at different stages of disease progression can potentially mitigate ICI myocarditis without significantly compromising the antitumor response.

Distinct mechanisms: B cells, autoantibodies, and complement

As noted previously, B cells are not the primary driver of ICI myocarditis but may contribute through circulating autoantibodies and complement activation. In one study of patients with overlapping ICI myocarditis and myositis, high-titer anti-muscle serum autoantibodies were identified in 55% of patients with confirmed irAEs, with C1q⁺MRC1⁺ myeloid cells expressing FcγRIIIa and C4d deposition identified in tissue, potentially implicating antibody-dependent complement activation as a mechanism of injury.⁹ Other studies have similarly linked anti-acetylcholine receptor seropositivity with the risk of myocarditis.⁸⁶ This finding was consistent with an earlier cardiomyopathy model involving PD-1 knockout, in which a circulating IgG autoantibody to cardiac troponin I was identified.^44,87

The direct roles of B cells and antibody production in tumors appear to differ from those in ICI myocarditis. In tumors, tertiary lymphoid structures (TLSs) are putative local sites of B cell maturation. The presence and maturation state of TLSs are associated with ICI responses across tumor types; TLS presence is associated with better responses, and mature TLSs predict efficacy independently of PD-L1 expression.88, 89, 90 Mechanistically, TLS-associated B cells license CD8 T cell responses via CD40/CD40L. Despite the presence of autoantibodies, classical TLSs have not yet been reported in ICI myocarditis cardiac tissue.⁶² Whereas tumor and cardiac immune responses may be separable, cardiac and skeletal muscle immune responses may not be. Myositis or myasthenia-like presentations occur in 43% to 75% of ICI myocarditis cases, and respiratory muscle involvement has been shown to account for approximately 70% of deaths.^77,91 This suggests that therapies targeting circulating antibodies or complement biology—such as plasmapheresis or complement-directed therapy—may preferentially address cardiac humoral toxicity without directly impacting the antitumor response.

Distinct mechanisms: myeloid cells

Myeloid cells represent a further opportunity for decoupling, including IL-1 blockade and anti-CXCR6–directed therapies. IL-1β⁺TNF⁺ proinflammatory myeloid cells are significantly enriched in cardiac tissue in myocarditis, correlate with major adverse cardiovascular events, and trend toward shorter overall survival.⁹ In a high-fat diet orthotopic pancreatic ductal adenocarcinoma mouse model treated with chemotherapy plus dual immune checkpoint blockade, anti-IL-1β or anti-IL-1 receptor blockade simultaneously rescued cardiac toxicity and improved antitumor efficacy, with serum IL-1β the only cytokine selectively elevated among the 10 profiled.⁹² This was a single obesity-specific model, and further work is needed to determine if IL-1β blockade improves outcomes in patients with ICI myocarditis. There is an epidemiological precedent for targeting IL-1β: in a substudy of the CANTOS trial, canakinumab reduced incident lung cancer in a dose-dependent manner, providing a signal linking IL-1β with both cardiovascular biology and tumor incidence suppression.⁹³ However, inflammatory pathways are often highly redundant, with some overlap between the effects of IL-1β and the CXCR3–IFN-γ axis. Future work is needed to determine whether blocking the IL-1β pathway will be sufficient to dampen the inflammation driving myocarditis or whether additional agents targeting other pathways, including IFN-γ, will be necessary in these patients.

As noted earlier, CCR2⁺ macrophages have a role in ICI myocarditis biology based on murine models.^11,49 In contrast to ICI myocarditis, monocyte-derived myeloid cells in tumors are often associated with poor antitumor immunity and worse outcomes with ICI therapy.94, 95, 96 This may be because tumors efficiently polarize CCR2⁺ monocytes recruited from the circulation toward an immunosuppressive phenotype. Consequently, these cells actively contribute to tumor growth and ICI therapy resistance rather than promoting a proinflammatory antitumor response. Accordingly, targeting myeloid cells may help dampen ICI myocarditis without compromising antitumor immunity by reducing the pool available for recruitment into tumors. However, because myeloid cells can have both proinflammatory and immunosuppressive roles in tumors, depleting these populations in tumor models has been challenging. Strategies that alter myeloid cell polarization or function may therefore be useful for reducing toxicity without attenuating the antitumor response.

In summary, the shared IFN-γ–CXCL9/CXCL10 pathway in ICI myocarditis and antitumor immunity poses a major challenge to suppressing cardiac inflammation without compromising antitumor efficacy (Figure 4). Potential strategies for decoupling toxicity from efficacy include selective JAK inhibition after ICI priming, which may preserve antitumor efficacy while rescuing cardiovascular irAEs; B cell–targeted approaches, given that cardiac humoral toxicity mediated by circulating antibodies and complement differs mechanistically from tumor-localized TLS; and myeloid-based approaches such as anti-IL-1β or anti-CXCR6 therapy. Although antitumor efficacy and cardiovascular toxicity share important mechanisms, multiple laboratories have now identified several methods of decoupling efficacy from toxicity.

Figure 4.

Shared and Distinct Mechanisms of ICI Myocarditis and Antitumor Immunity

(A) Shared mechanisms of antitumor efficacy and immune checkpoint inhibitor (ICI) myocarditis are shown, focusing on the interferon γ (IFN-γ)–CXCL9/CXCL10 and CXCL16–CXCR6 pathways. (B) Distinct mechanisms of antitumor efficacy and myocarditis are displayed, showing tertiary lymphoid structures and immunosuppressive myeloid cells in antitumor efficacy compared with autoantibodies, C4d complement deposition, and proinflammatory myeloid cells in myocarditis. IL = interleukin; MHC-I = major histocompatibility complex class I; TCR = T cell receptor; TNF-α = tumor necrosis factor α.

Translating Mechanistic Insights Into Novel Diagnostic and Therapeutic Strategies

Impressive progress has been made over the last decade toward improving the study of ICI-mediated cardiotoxicities. Translating these mechanistic findings into new diagnostic and therapeutic strategies has faced several key challenges. First, cardiovascular irAEs are uncommon, and obtaining large numbers of human tissue samples from affected patients (eg, myocardial, pericardial, or vascular samples) is difficult.⁹⁷ Without these samples, studies have often relied on peripheral blood biomarkers or mouse models of uncertain fidelity to human ICI myocarditis, providing an incomplete view of human disease pathophysiology and limiting mechanistic insights.

Second, patients with cardiovascular irAEs often carry comorbid advanced malignancies, may experience concurrent noncardiac irAEs, and are treated with variable immunosuppressive regimens. Blood-based biomarkers of cardiac irAEs can therefore be confounded by this clinical heterogeneity, although multiple independent studies have identified common molecular and cellular mediators.^62,98 Third, rigorous prospective clinical trials of patients with cardiac irAEs have proven challenging to conduct. Eligible patients are uncommon and face substantial competing risks of cancer-related morbidity and mortality. Furthermore, they have received highly variable immunosuppressive regimens over time and across institutions, complicating comparisons. Collectively, although these challenges have impeded in developing novel diagnostic and therapeutic strategies, recent gains in mechanistic understanding have generated optimism that these findings can be translated toward new approaches.

Little is presently known about the factors that drive susceptibility to cardiovascular irAEs and which patients are at highest risk. Tools to identify high-risk patients would aid informed clinical decision making and facilitate clinical trials aimed at preventing irAEs. Risk-stratifying assays that leverage blood samples or other routinely collected clinical data could be particularly useful and easy to deploy. Very little is known about the genetic basis for these adverse events. Several reports have suggested that certain human leukocyte antigen alleles may increase risk of ICI myocarditis, but these preliminary findings await confirmation.^99,100 Clonal hematopoiesis of indeterminate potential has been associated with a higher incidence of ICI myocarditis and greater mortality among affected patients.¹⁰¹ Although it is a clear risk factor for atherosclerosis and vascular events, whether and how clonal hematopoiesis of indeterminate potential influences ICI-mediated acceleration of atherosclerosis remains unknown.¹⁰²

Elevated troponin before initiation of ICI therapy has been associated with an increased risk of subsequent ICI myocarditis; the mechanistic basis for this observation, if any, remains unclear.^103,104 As noted, a role in ICI myocarditis pathogenesis for T cells specific for α-MHC has also been suggested.⁶¹ Yet, because α-MHC-specific T cells have also been reported in the blood of healthy individuals, the utility of detecting these T cells in peripheral blood as a risk-stratifying biomarker is uncertain. The diagnostic and prognostic value for cardiovascular irAEs of pre-ICI serum concentrations of circulating factors such as cytokines and chemokines, peripheral white blood cell pool composition, and the circulating TCR repertoire remains to be firmly established.

The current International Cardio-Oncology Society consensus definition of ICI myocarditis relies on a histopathological diagnosis or a combination of elevated troponin, magnetic resonance imaging findings, and clinical features.¹⁰⁵ The growing mechanistic understanding of ICI myocarditis has begun to inform efforts to improve diagnostic strategies. The recognition of inflammatory macrophages and fibroblasts among the myocardial inflammatory infiltrate has provided the basis for PET imaging that reports the presence of these cells.^{49,62,106,107} Multiple studies have reported elevated levels of cytokines and chemokines, including IL-6, CXCL9, CXCL10, and CXCL13, yet their utility in diagnostic evaluation remains undefined.^10,62,108 Additionally, a specific microRNA, hsa-miR-Chr8:96, has been associated with non-ICI myocarditis, but there are no data on the value of this or other microRNAs in ICI myocarditis.¹⁰⁹ Further investigation is needed to translate the growing understanding of ICI myocarditis biology into validated diagnostic tools.

As with diagnostic approaches, treatment of ICI myocarditis has evolved in recent years as guided by mechanistic understanding. The clearest examples of this are the use of abatacept and ruxolitinib.¹¹⁰ The use of abatacept for ICI myocarditis has been motivated by the recognition that full T cell activation through CD28-mediated costimulation is a critical step in the ICI myocarditis immune response.^48,111 The use of ruxolitinib in ICI myocarditis has been supported by the upregulation of various inflammatory cytokines that mediate their signaling through JAK1 and JAK2.^{10,49,62,108,112} It is important to note, however, that although abatacept and ruxolitinib have seen growing utilization in the treatment of ICI myocarditis, highly rigorous (ie, prospective, double-blinded, placebo-controlled) data demonstrating their effectiveness are not yet available.

Last, recent discoveries regarding the roles of CXCR3 and CXCR6 signaling in ICI myocarditis have raised the fascinating possibility that inhibiting the recruitment of immune cells into or localization within the heart may be an effective therapeutic strategy.^11,12,49 However, no studies of CXCR3 or CXCR6 antagonism in human ICI myocarditis have been published, and because both receptors have roles in tumor T cell responses, their inhibition could risk diminishing antitumor immunity.^85,113

Conclusions

The past decade has yielded substantial advances in understanding of the immunological mechanisms underlying ICI-mediated cardiovascular toxicities, yet significant gaps remain in translating these insights into clinical practice (Central Illustration). Key challenges—including the relatively low frequency of cardiovascular irAEs, difficulty obtaining human cardiac tissue, and confounding effects of concurrent irAEs and immunosuppressive therapies—have slowed the development of novel diagnostic and therapeutic strategies. Nevertheless, recent progress offers considerable grounds for optimism. Advances in understanding the molecular and cellular drivers of ICI myocarditis have provided a foundation for advances in PET imaging and treatment of ICI myocarditis in particular. Critical next steps will require multi-institutional collaborative efforts to validate emerging diagnostic and therapeutic strategies in prospective clinical trials and continue bridging the gap between mechanistic discovery and improved patient outcomes.

Central Illustration.

Mechanisms and Immune Responses Underlying ICI-Associated Cardiovascular irAEs

Immune checkpoint inhibitor (ICI)–associated atherosclerosis, ICI myocarditis, and antitumor responses to ICI therapy have shared and distinct mechanistic features. Ongoing efforts to understand the biology underlying cardiovascular immune-related adverse events (irAEs) may help achieve these stated diagnostic and therapeutic goals. IFN-γ = interferon γ.

Funding Support and Author Disclosures

Dr Villani was supported by a National Institutes of Health (NIH) grant (DP2CA247831), a Next Generation Scholar Award from the Broad Institute, a Howard M. Goodman Fellowship from Massachusetts General Hospital, and funding from the Manton Foundation and the Klarman Family Foundation. Dr Drobni was supported by the János Bolyai Research Scholarship of the Hungarian Academy of Sciences. Dr Neilan was supported by NIH grants (R01HL130539, R01HL137562, K24HL150238), the Michael and Kathryn Park Endowed Chair in Cardiology, an MGB Chair in Cardio-Oncology, and a Hassenfeld Scholar Award. Dr Zhu was supported by NIH grants (R01HL177581, R01HL174432, R03HL173146, K08HL161405), a Stanford Cardiovascular Institute Seed grant, and an American Heart Association (AHA) grant (25TPA1480322). Dr Balko was supported by an NIH grant (R01HL156021). Mr Fankhauser was supported by an AHA fellowship award (25PRE1375723) and NIH grants (1F30HL179763 and T32GM007347). Dr Moslehi was supported by NIH grants (R01HL155990, R01HL156021, R01HL160688, R01HL170038, and P01HL141084) and the Leducq Foundation. Dr Palaskas was supported by grants from the Cancer Prevention and Research Institute of Texas (RP200670), the NIH (1P01CA261669-01), and the Food and Drug Administration (1U01FD008717-01). Dr Siddiqui was supported by grants from ASCO-Conquer Cancer Foundation, Prostate Cancer Foundation, and the Cancer Research Institute. Dr Blum was supported by an NIH grant (1K08AI197864). Dr Zlotoff was supported by an AHA grant (24IPA1273052). Dr Reynolds has received consulting fees from Gilead, Regeneron, and SAGA Diagnostics for advisory board service; support from UpToDate; and educational speaking fees from CME Outfitters and MedScape outside the submitted work. Dr Villani has served as a consultant for Bristol Myers Squibb and Merck; and owns financial interest in 10x Genomics. Dr Neilan has received personal fees for consulting from Bristol Myers Squibb; personal fees from Genentech, GSK, Merck, Roche, and Sanofi; and grants from Bristol Myers Squibb and AstraZeneca. Dr Zhu has received consulting fees from Skribe Medical. Dr Moslehi has received consulting or advisory fees from Deciphera, Takeda, AstraZeneca, Regeneron, Bayer, Kiniksa Pharmaceuticals, Daiichi-Sankyo, BeiGene, Incyte, AskBio, Arcus Biosciences, Nektar Therapeutics, F. Hoffmann-LaRoche Ltd, Skribe Medical, Verastem, ImmunoCore, Innovent Biologics, Shattuck Labs, Sobi, Sumitomo, Repare Therapeutics, Pin Therapeutics, and Cytokinetics; and is a coinventor of a patent related to the use of abatacept in the treatment of immune checkpoint inhibitor–mediated myocarditis and has other patents pending related to diagnostics and therapeutics in inflammatory heart disease. Dr Palaskas has received consulting fees from Kiniksa Pharmaceuticals. Dr Siddiqui has received honoraria from Cancer Expert Now, Cincinnati Cancer Advisors, Curio Science, CardinalHealth, DAVA Oncology, and IDEOlogy Health; and consulting/advisory fees from Amgen, Astellas, Bayer, Merck, Pfizer, Johnson & Johnson, and SurvivorNet. Dr Blum has received consulting fees from Two River Consulting; holds equity in Kronos Bio, 76Bio, Allogene Therapeutics, and Candid Therapeutics; and has received research support from AstraZeneca and Astellas. Dr Zlotoff has received consulting fees from Bristol Myers Squibb and Replimune. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.