Immune checkpoint inhibitors (ICIs) mainly act on the activation of T cells to fight against tumor cells. There are currently FDA-approved drugs that target cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4), programmed cell death protein 1 (PD-1), and programmed cell death ligand 1 (PD-L1). Although ICI-related adverse effects are overall less severe than chemotherapy, there are potentially lethal adverse effects, such as cardiotoxicity, neuromuscular toxicity, and pulmonary toxicity.

The first case of fatal ICI-related myocarditis was reported by Läubli et al. in 2015 (1), in a 73-year-old patient with metastatic melanoma who developed ICI-associated myocarditis after receiving pembrolizumab, leading to acute heart failure. Two cases of fulminant myocarditis were reported in 2016. Both patients became symptomatic within 2 weeks of administration and aggravated rapidly, leading to death within a short period despite the admission of high-dose glucocorticoid and supportive care. These two cases of severe cardiotoxicity warned clinicians of the security of ICIs (2).

Cardiotoxicity associated with ICIs is broadly classified into myocarditis, pericarditis, and arrhythmia. Other clinical manifestations, such as hypertension, Takotsubo-like syndrome, myocardial ischemia, and myocardial infarction, are less reported (3, 4). In addition, some studies have roughly divided them into inflammatory (including myocarditis, pericarditis, etc.) and non-inflammatory (arrhythmia, myocardial infarction, etc.).

Early studies showed that the incidence of ICI-related cardiotoxicity is less than 1% (5). However, as Ganatra et al. (6) stated, the increasing application of ICIs in real-world and the diverse presentation of cardiotoxicity suggest a higher incidence of cardiotoxicity than previously reported. In addition, subclinical cases and increased awareness of toxicity may also be the contributing factors. Among these classifications, myocarditis maintained the highest incidence and fatality rate, as well as the worst prognosis. However, the presentations of cardiotoxicity in real-world were not separated. It is not rare for myocarditis to occur simultaneously with arrhythmia or pericarditis. Studies have found that approximately 19% of patients with pathologically defined cardiotoxicity developed arrhythmias (7). Two patients with fulminant myocarditis reported in 2016 also developed complete atrioventricular conduction block (2). The pathophysiological manifestation suggested the infiltrated T cells and macrophages in the conduction system. As reported, atrial fibrillation, ventricular arrhythmias, and conduction disturbances were detected in 17–30% of patients with ICI-related cardiotoxicity, of whom 3–13% had mere arrhythmias without myocarditis (8). Although the mechanism is unclear, the investigators concluded that there may be underlying undiagnosed myocarditis (8). There are limited studies on ICI-related pericardial disease. In addition, the incidence is unclear due to the lack of cardiovascular monitoring during ICI clinical trials but maybe more common than previously recognized. In an evaluation based on VigiBase, Salem et al. included 95 cases of pericardial disease associated with ICI, of which 60% were severe and 21% were associated with death (7). Unlike myocarditis, the incidence of pericarditis did not increase significantly with combination immunotherapy. In addition, the study found that more than half of the cases with pericardial diseases were reported in patients with lung cancer (7). Based on the preclinical and clinical studies, several possible influencing factors of ICI-related pericardial involvement were proposed. Given the disproportionately high incidence of pericardial disease in patients with lung cancer and synergy between radiotherapy and immunotherapy is considered a possible driver.

In the previous studies, autoimmune disease, prior heart disease, and combined ICI therapy can be the probable risk factors. Results are mostly derived from databases based on overall irAEs with few real-world studies based on cardiotoxicity. Therefore, we conducted a real-world retrospective study based on the whole immunotherapy population to explore the risk factors for ICI-related cardiotoxicity.

We found that the incidence of cardiotoxicity was 7.0%, which was higher than the previously reported. By collecting the cardiovascular risk factors of the enrolled patients, we found that diabetes was an independent risk factor for the occurrence of cardiotoxicity. This conclusion can be seen in the retrospective study of Mahmood et al. (11). Through a multicenter retrospective study of 35 patients, they found that myocarditis was more common in patients with diabetes mellitus, especially in patients receiving a combination of ICI therapy. Although they ultimately failed to prove it an independent risk factor, their research suggested a certain direction. In our analysis, an exact association of diabetes with the development of cardiotoxicity was found. However, we did not find an association between other cardiovascular risk factors, such as hypertension and the occurrence of cardiotoxicity.

The effect of diabetes on the development of ICI-related cardiotoxicity may be related to the long-term chronic inflammation in patients with diabetes. Till now, diabetes has been regarded as a chronic inflammatory disease. The high-glucose environment of diabetes significantly increases the cytokines, such as interleukin 4(IL-4), interleukin 5(IL-5), interleukin 6(IL-6), interleukin 13(IL-13), and tumor necrosis factorα (TNF-α). These cytokines maintained the balance of the autoimmune microenvironment. The chronic pathological state of immune imbalance caused by diabetes can promote many diseases (12–14). In addition, the oxidative stress produced by diabetes can also produce many inflammatory cytokines, such as TNF-α, IL-6, and transforming growth factor β (14). The cytokine pathway has been confirmed to be closely related to irAEs in many studies (15). In addition, diabetes affects the number and activity of T cells, B cells, and NK cells in peripheral blood, and the result of immune imbalance may also play an important role (14). Moreover, whether diabetes, as an autoimmune disease, also plays an important role in irAEs like other types of autoimmune diseases remains to be further studied.

We also collected baseline inflammatory indicators, such as IL-6, C-reactive protein (CRP), and lactic dehydrogenase (LDH), of all patients and calculated NLR, PLR, and LMR. We found that lower baseline LMR was an independent risk factor for the development of cardiotoxicity. In previous studies, the correlation between peripheral lymphocyte levels and adverse reactions has already been investigated. In a study of risk factors for cardiotoxicity based on more than 4,000 immunotherapy patients (16), the authors identified the association between low baseline peripheral lymphocyte levels and cardiotoxicity through a machine learning method. In a retrospective study of NSCLC, an association of LMR with irAEs was also found (17). Activated T lymphocytes in patients with tumors after using PD-1/PD-L1 antibodies can not only attack tumor cells, but also cause irAEs. According to the autopsy result of patients with cardiotoxicity as previously reported, lymphocytic infiltration was widely seen in the diseased tissue (2). Therefore, lymphocytes may play a key role in the response of irAEs. However, no direct correlation between peripheral lymphocyte count (or percentage) and the occurrence of cardiotoxicity was found in our cohort. Moreover, these indicators are often dynamic during tumor treatment. Therefore, how does their changing act on adverse reactions? What role do lymphocytes play in it? They deserve further discussion.

In terms of gender, ICI was used in more men than women in our study, possibly due to the higher tumor prevalence in male patients. In this study, male patients received ICI proportionally more often than female patients, but no significant association between gender and the development of myocarditis was found. One study found higher occurrence rate of ICI associated cardiovascular with men (18). Several additional studies had the same finding (7, 11, 19). However, another research found a different result that women were at higher risk of ICI-related myocarditis. These studies suggested the relationship of gender and the occurrence of ICI-related myocarditis. In addition, we found a higher incidence of ICI-related cardiotoxicity in patients with respiratory neoplasms than in other types of neoplasms. Salem et al. (7) also found a higher rate of ICI-related pericardial disease among patients with lung cancer. Basic research has found upregulation of PD-L1 in models of myocardial injury induced by ischemiareperfusion and hypothermia, which may be a cytokine-mediated mechanism of cardiac protection. Besides, radiation therapy in patients with lung cancer may present with potential exposure to cardiac antigens. Immune activation thereby conferring them a higher incidence of cardiotoxicity. As for the correlation between comorbid autoimmune diseases and cardiotoxicity, there were numerous reports of thymoma patients with ICI-related cardiotoxicity. Toi et al. (20) retrospectively recruited 137 patients with positive autoimmune antibody receiving ICI therapy. They found that positive autoimmune antibody was significantly associated with a higher occurrence of irAEs and better clinical benefit in NSCLC. In addition, another study also suggested a relationship between autoimmune antibodies and irAEs (20). Therefore, we are more cautious about ICI initiation in patients with autoimmune diseases in real-world. Therefore, patients with active autoimmune disease were excluded from our study. Because the treatment of active autoimmune diseases is contrary to tumor treatment, they can affect the judgment to response of tumor treatment and drug-related adverse reactions.

In addition to ICI-related cardiotoxicity, we also collected ICI-related thyroid dysfunction and pneumonia. We found that both of them have a certain correlation with the occurrence of cardiotoxicity. Although the onset of ICI-related pneumonitis or abnormal thyroid function does not completely precede the onset of cardiotoxicity, we cannot establish a causal relationship between them. However, this finding suggests that there may be a link between the toxicities occurring at different times targeting different organs. The occurrence of toxicities targeting the lung or thyroid may then alert clinicians the occurrence of cardiotoxicity. Besides, myositis, hepatitis, and pneumonia may be complicated by the occurrence of cardiotoxicity. In the cardiotoxicity population, a total of 30 (41.1%) patients had concurrent toxicities targeting other organs. In a study based on the VigiBase database, serious combined ICI-related adverse reactions accounted for 42% of adverse reactions, most of which were ICI-related myocarditis and myositis (21). In a multicenter study, 32% of myositis was associated with concurrent myocarditis (22). In addition, the mortality rate is the highest in patients with combined myocarditis and myasthenia gravis. Therefore, in patients with symptoms of ICI-related myositis, they should be alert to the occurrence of myocarditis.

Previous studies have demonstrated a possible correlation between irAEs and tumor prognosis. A meta-analysis (23) explored the relationship between ICI-related adverse drug reactions and clinical benefits. It was suggested that in patients receiving ICIs, the development of irAEs was positively correlated with objective remission rate (ORR), progression-free survival (PFS), and OS, irrespective of disease site, type of ICI, and irAE. ORR was better in patients with grade 3 or higher toxicity, but OS was worse. However, in this study, we only did the analysis in overall survival, and no significant difference was found between patients with or without cardiotoxicity.

Till now, the National Comprehensive Cancer Network (NCCN) guidelines do not recommend rechallenge of ICI after grade 2 and above myocarditis. Subclinical myocarditis was recommended continue the use of ICI with close detection. In a single-center retrospective study in 2019 (24), the investigators included 93 patients with grade 2 or higher toxicity, including 43 grade 2 events, 36 grade 3 events, and 14 grade 4 events. Taking the occurrence of the second toxicity as the endpoint, the final results suggested that the time since ICI use to the occurrence of the initial irAE was related to the occurrence of the second irAE. Besides, the severity of the second irAE was not more severe than the first one. They concluded that the risk-reward ratio of anti-PD-1 or anti-PD-L1 rechallenge appears to be acceptable. However, patients with first ICI-related toxicity involving the heart were not included in this study. Conversely, there were also some studies restarting ICI after the occurrence of irAEs, resulting in recurrence of grade 5 toxicity in patients. Therefore, ICI rechallenge after irAEs is still controversial. Due to the small number of cases in our study, we cannot currently explain the safety of cardiotoxicity after restarting, because once cardiotoxicity occurs, it may be fatal drug toxicity, which is unacceptable for clinicians and patients’ families. Moreover, the timing of ICI restarting among these patients also varied, and some patients even resumed ICI 1 month after ICI withdrawal. If there are large clinical studies to confirm the feasibility of restarting immunotherapy for ICI-related cardiotoxicity, the timing of restarting ICI will also be a major focus. Furthermore, there were many patients with grade 1 cardiotoxicity who continue to use immunotherapy under strict monitoring, and a small number of patients who suffered mild cardiotoxicity for not one time, but it does not affect the progress of tumor treatment.

We retrospectively analyzed the risk factors of ICI-related cardiotoxicity in the whole population receiving ICI therapy. We found a higher incidence of ICI-related cardiotoxicity, and a high proportion of severe cases than previous reported in real-world situation. Patients with diabetes mellitus and low baseline levels of LMR have an increased incidence of cardiotoxicity, which should be closely monitored during the use of ICIs. Besides, the incidence of severe cardiotoxicity was correlated with shorter overall survival.

Despite the advantages and potential insights on this study, there are some inevitable shortcomings in our study. First of all, some potential biases are still unavoidable due to the retrospective design of this study, such as investigator bias. Second, although the overall immunotherapy sample size we included is relatively large, the number of cardiotoxicity cases is relatively small, as well as the number of patients who rechallenge ICI. Finally, due to the serious lack of some data during the study period, we could not analyze the relationship between the baseline levels of some inflammatory markers (such as LDH, CRP, and cytokines) and the occurrence of cardiotoxicity. Hence, well-designed large-scale prospective studies are urgently needed in the future.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

This study involving human participants was reviewed and approved by the Ethics Review of the First Affiliated Hospital of Xi’an Jiaotong University (no. XJTU1AF2020LSK-262) and was conducted in strict accordance with the requirements of the Declaration of Helsinki. Written informed consent was not required due to the retrospective nature of the study.

XC, TT, XL, ZR, and YY contributed to the design of the study. XC, AJ, RZ, XF, NL, JW, and XZ contributed to manuscript preparation. XC and AJ wrote the manuscript. RZ, YY, CS, and XL helped collect cases. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.