Cardiotoxicity associated with immune checkpoint inhibitors: A retrospective analysis of patients at an academic tertiary care center

Background Immune checkpoint inhibitors (ICIs) are a novel class of anticancer agents that have demonstrated clinical response for both solid and hematological malignancies. ICIs are associated with development of immune-related adverse events including cardiotoxicity. We estimated the incidence of newly diagnosed cardiovascular disease ICI-related cardiotoxicity in patients treated with ICIs at a large, tertiary care center. Methods All patients with a cancer diagnosis who received any ICI treatment in the University of Florida’s Integrated Data Repository from 2011-2017 were included. Cardiovascular disease was dened as a new ICD diagnosis code for cardiomyopathy, heart failure, arrhythmia, heart block, pericardial disease, or myocarditis after initiation of ICI treatment. Results Of 102,701 patients with a diagnosis of malignancy, 424 patients received at least one ICI. Sixty-two (14.6%) patients were diagnosed with at least one new cardiovascular disease after initiation of ICI therapy. Of the 374 patients receiving one ICI, 21 (5.6%) developed heart failure. Of the 49 patients who received two ICIs sequentially, three (6.1%) developed heart failure and/or cardiomyopathy. Incident cardiovascular disease was diagnosed at a median of 63 days after initial ICI exposure. One patient developed myocarditis 28 days after receiving nivolumab. Mortality in ICI treated patients with a concomitant diagnosis of incident cardiovascular disease was higher compared to those who did not (66.1% vs. 41.4%, odds ratio = 2.77, 1.55-4.95, p = 0.0006). Conclusions This study suggests a high incidence of newly diagnosed cardiovascular disease after the initiation of ICI therapy. (GLS) and post-treatment developed ICI-induced cardiotoxicity. These data were obtained using TomTec® software (Chicago, IL). Troponin, N-terminal pro-brain natriuretic peptide (NT-pro BNP), and brain natriuretic peptide (BNP) levels were included when available within the time frame of this study. Statistical analysis: Demographic and medical history information of those with and without a diagnosis of attributable cardiotoxicity were compared using Student’s unpaired t-test for continuous variables and chi-square test for categorical variables as appropriate. The percentages of patients with cardiotoxicity following exposure to each drug were estimated. Multivariable logistic regression analysis was performed to estimate the odds ratios (ORs) and 95% condence intervals (CIs) for mortality adjusting for demographics, comorbidities, and ICIs. Covariates with univariate p-value of < 0.2 were considered in the multivariable logistic regression and variables with p < 0.05 were retained in the model. All analyses were performed in SAS v. 9.4 (Cary, NC). This study was approved by the University of Florida Institutional Review Board (IRB) (IRB# 201702876).


Abstract
Background Immune checkpoint inhibitors (ICIs) are a novel class of anticancer agents that have demonstrated clinical response for both solid and hematological malignancies. ICIs are associated with development of immune-related adverse events including cardiotoxicity. We estimated the incidence of newly diagnosed cardiovascular disease ICI-related cardiotoxicity in patients treated with ICIs at a large, tertiary care center. Methods All patients with a cancer diagnosis who received any ICI treatment in the University of Florida's Integrated Data Repository from 2011-2017 were included. Cardiovascular disease was de ned as a new ICD diagnosis code for cardiomyopathy, heart failure, arrhythmia, heart block, pericardial disease, or myocarditis after initiation of ICI treatment. Results Of 102,701 patients with a diagnosis of malignancy, 424 patients received at least one ICI. Sixty-two (14.6%) patients were diagnosed with at least one new cardiovascular disease after initiation of ICI therapy. Of the 374 patients receiving one ICI, 21 (5.6%) developed heart failure. Of the 49 patients who received two ICIs sequentially, three (6.1%) developed heart failure and/or cardiomyopathy. Incident cardiovascular disease was diagnosed at a median of 63 days after initial ICI exposure. One patient developed myocarditis 28 days after receiving nivolumab. Mortality in ICI treated patients with a concomitant diagnosis of incident cardiovascular disease was higher compared to those who did not (66.1% vs. 41.4%, odds ratio = 2.77, 1.55-4.95, p = 0.0006). Conclusions This study suggests a high incidence of newly diagnosed cardiovascular disease after the initiation of ICI therapy.

Background
Immune checkpoint inhibitors (ICIs) have revolutionized the management of a diverse spectrum of solid and hematological malignancies previously associated with poor prognosis. Immune checkpoint blockade removes inhibitory signals of T-cell activation enabling tumor-reactive T cells to mount an effective antitumor response by overcoming regulatory mechanisms. 1 Currently, FDA-approved ICIs are inhibitors of either the cytotoxic Tcell lymphocyte-associated protein-4 (CTLA-4) or the programmed death receptor 1 (PD-1) or its ligand (PD-L1). Robust research efforts evaluating other checkpoint targets such as lymphocyte-activation gene-3 (LAG-3) and T-cell immunoglobulin and mucin-domain containing-3 (TIM-3) are ongoing.
ICIs have been reported to cause a range of immune-related adverse events (irAE), mostly involving the skin, endocrine system, liver, lungs, and gastrointestinal tract. These targeted therapies affect speci c signaling pathways that can also induce cardiotoxicity. IrAEs occur due to inhibition of immune checkpoints that boost physiological barriers against autoimmunity, leading to local and systemic autoimmune responses. 2 Fulminant myocarditis is currently the most recognized irAE, but complete heart block, conduction abnormalities, pericarditis, stress-induced cardiomyopathy, and left ventricular dysfunction have also been reported. [3][4][5] Limited data are available on the incidence of cardiotoxicity after ICI initiation, and there is scarce evidence to guide prevention, surveillance, and treatment. [6][7][8] The reported incidence of ICI-induced myocarditis ranges from 0.06% to 1.14% of patients receiving ICIs. 9,10 However, absence of systematic monitoring and coding mechanisms for cardiac events in immunotherapy trials suggest that cardiac irAEs may be under-reported. 7 Accordingly, we estimated the incidence of cardiotoxicity among patients treated with ICIs using electronic health records at a large tertiary care center.

Methods
This was an observational cohort study using data extracted from electronic health records (EHRs). Supported by the University of Florida (UF) Clinical and Translational Science Institute (CTSI), the UF Health Integrated Data Repository (IDR) is a large-scale database that collects and organizes information from across the UF-Health clinical and research enterprises, thereby including most inpatient and outpatient care services.
The IDR provides access to Health Insurance Portability and Accountability Act (HIPAA) compliant and Institutional Review Board (IRB) approved limited datasets that include demographics, medications, lab results, diagnosis, and clinical encounters. For this study, the UF IDR was queried to extract information relevant to all patients receiving anticancer drugs from 2011-2017. All patients with the International Classi cation of Disease, ninth and tenth revisions, clinical modi cation (ICD-9-CM 140-239.99) and (ICD-10-CM C00-D49) codes for malignancy were included.
The current study consisted of patients who had received at least one dose for any ICI including PD-1 inhibitors (pembrolizumab, nivolumab), PD-L1 inhibitors (atezolizumab, durvalumab, avelumab), and CTLA-4 inhibitors (ipilimumab, tremelimumab). Baseline demographic information was collected at the encounter of the rst ICI administration. Comorbidities such as history of hypertension, hyperlipidemia, diabetes, and ischemic heart disease were de ned based on the presence of International Classi cation of Diseases, 9 th revision, Clinical Modi cation (ICD-9-CM) and ICD-10-CM diagnosis codes prior to the rst ICI prescription date.
Cardiotoxicity was de ned by ICD-9-CM and ICD-10-CM codes for cardiomyopathy, heart failure, myocarditis, arrhythmia, pericardial disease and heart block (Supplemental Table 1). The case de nition for potential ICI-induced cardiotoxicity was a new diagnosis code for any of these conditions as entered by a clinician after initiation of ICI therapy in patients with no prior history of the incident cardiac condition. The performances of these computable phenotypes were previously reported. The ranges of the estimates were 79% -95% for sensitivity, 90% -98.9% for speci city, 70% -94% for positive predictive value, and of 95% -99.4% for negative predictive value, respectively. 11,12 Patients with existing diagnosis codes for cardiomyopathy, heart failure, myocarditis, arrhythmia, pericardial disease, and/or heart block before ICI initiation were considered to have preexisting disease, thus, did not meet the case de nition.
Global longitudinal strain (GLS) data were calculated from pre-and post-treatment transthoracic echocardiograms for a subset of patients who developed ICI-induced cardiotoxicity. These data were obtained using TomTec® software (Chicago, IL). Troponin, N-terminal pro-brain natriuretic peptide (NT-pro BNP), and brain natriuretic peptide (BNP) levels were included when available within the time frame of this study.
Statistical analysis: Demographic and medical history information of those with and without a diagnosis of attributable cardiotoxicity were compared using Student's unpaired t-test for continuous variables and chi-square test for categorical variables as appropriate. The percentages of patients with cardiotoxicity following exposure to each drug were estimated. Multivariable logistic regression analysis was performed to estimate the odds ratios (ORs) and 95% con dence intervals (CIs) for mortality adjusting for demographics, comorbidities, and ICIs. Covariates with univariate p-value of < 0.2 were considered in the multivariable logistic regression and variables with p < 0.05 were retained in the model. All analyses were performed in SAS v. 9.4 (Cary, NC). This study was approved by the University of Florida Institutional Review Board (IRB) (IRB# 201702876).

Results
Of 102,701 patients with a diagnosis of malignancy, 424 patients received an ICI and their pertinent demographic and clinical characteristics are summarized in Table 1. Overall, the median age was 63 years, the majority were men, Non-Hispanic whites (63.4% and 85.6%, respectively) and 7.6% were Non-Hispanic blacks. The most frequent cancer diagnoses were lung cancer (29.7%), melanoma (17.0%), and kidney cancer (12.7%) ( Table 1). Almost half of the patients (49.5%) had hypertension, 30.2% had hyperlipidemia, 17.9% had diabetes, and 12.7% had ischemic heart disease before the initiation of ICI treatment as determined by ICD diagnosis codes (Table 1).
Forty-three patients including 9 (20.9%) with and 34 (79.1%) without a cardiotoxicity diagnosis had both pre-ICI treatment and post-ICI treatment transthoracic echocardiograms. There was no signi cant difference in mean change in ejection fraction (p=0.37) in those patients with cardiotoxicity compared to those without. Four patients also had global longitudinal (GLS) measured. Each patient demonstrated a decline in their GLS, ranging from 11% to 44% without meaningful change in ejection fraction (Supplemental Table 2).
Of the 424 ICI-treated patients, 191 (45.1%) died during the study period. The median time from ICI initiation to death was 128 days with interquartile range of 66-277 days. History of ischemic heart disease (OR: 2.11, 1.14-3.89, p = 0.017), prior use of doxorubicin (OR: 4.86, 1.31-18.11, p = 0.0184) and carboplatin use (OR: 1.86, 1.19-2.92, p = 0.0068) were also associated with higher mortality. After adjusting for the history of ischemic heart disease, and prior use of doxorubicin and carboplatin, mortality in those who developed cardiotoxicity remained higher compared to those who did not (66.1% vs. 41.4%, adjusted OR: 2.77 and 95% CI: 1.55-4.95, p = 0.0006). There was no evidence that the mortality was lower in patients treated with cardioprotective agents such as beta-blockers, angiotensin converting enzyme (ACE) inhibitors, angiotensin receptor blockers (ARBs), and statins (Table 3).

Discussion
In this study, we observed that approximately 15% of patients receiving ICI therapy developed evidence for potential cardiotoxicity. This is likely an underestimate since a standardized surveillance was not in employed at the time of or following the ICI treatment, and many cases are likely to be "subclinical" as is the case with most CVD. The most commonly observed types of cardiotoxicity were heart failure and arrhythmia. As suggested in previous reports 13,14 , the incidence of myocarditis was very low: only one patient (0.24%) developed myocarditis. The time to myocarditis for this patient was 28 days after initiation of nivolumab, which was consistent with the reported median time to onset of 30 days by Salem and colleagues. 13 This relatively low prevalence may be related to inadequate screening, particularly since the study includes data starting from 2011, when the autoimmune side effects of ICIs were just being recognized in a clinical setting. However, the observed incidence of other manifestations of cardiotoxicity was higher than previously suggested. 13 Estimates of the incidence of ICI-induced cardiotoxicity vary substantially across reports. This might be explained, in part, by variations in case de nitions and a speci c focus on certain cardiac syndromes (e.g., myocarditis). Other case series on ICI-induced cardiotoxicity suggest that cardiomyopathy, myocarditis, and conduction abnormalities are under-reported. 15 The manufacturer of both ipilimumab and nivolumab reported myocarditis (0.09%) from detailed clinical trial safety data but other cardiovascular irAEs were later described in case reports. [16][17][18][19] In our study, we excluded patients with pre-existing cardiovascular disease from analysis, however this approach might have underestimated the rate of ICI-related cardiotoxicity in the real-world clinical setting as several studies have showed patients with baseline cardiovascular disease are more likely to develop cardiovascular toxicities from cancer therapies. 4,20,21 Several studies have also characterized cardiac irAEs and their incidence. Myocarditis was one of the rst recognized ICI-related AEs and has been the most studied of the ICI-related cardiotoxicities. 14 A multicenter registry including patients from the US, Canada, and Germany and found that the prevalence of myocarditis after ICI therapy was 1.14% with a median time of onset of 34 days, whereas another study reported a median time of 65 days from initiation of treatment. 10,22 The study of a multicenter registry by Mahmood and colleagues reported that 16 patients developed a major adverse cardiac events and 6 (38%) occurred in patients with a normal ejection fraction. 10 Pooled Food and Drug Administration (FDA) data on reported ICI-related adverse events in clinical trials suggested that the risks of cardiomyopathy, arrhythmia, myocarditis, and pericardial disease were 0.53%, 5.56%, 0.03%, and 0.7%, respectively. 23 A meta-analysis of clinical trials of PD-1 inhibitors (nivolumab and pembrolizumab) and PD-L1 inhibitors (atezolizumab, avelumab and durvalumab) for treatment of non-small cell lung cancer also reported lower cardiovascular adverse event rates (1% for cardiorespiratory arrest, 2% for heart failure, 1% for myocardial infarction, and 2% for strokes). 24 A case series of 30 patients with ICIrelated cardiotoxicity, suggested the most frequently observed cardiotoxicities were reduced ejection fraction, arrhythmias, and pericardial disease with almost 80% of patients having left ventricular systolic dysfunction. 22 The incidence of irAEs has been noted to be dose dependent after ipilimumab and pembrolizumab with greater toxicity at higher dose levels. 7 The differences in incidence of cardiac irAEs reported may be attributable to dose of ICI and future studies should provide details on ICI dosage, number of chemotherapy cycles, and their timing. Dosing of ICIs in clinical practice follows a predominantly xed-dosing strategy (nivolumab -240 mg, pembrolizumab -200 mg) and extended dosing intervals (Q4 -Q6 weeks).
Previous studies have also explored early detection of chemotherapy-induced changes in cardiac function using the echocardiographic measures of ejection fraction and/or global longitudinal strain. [25][26][27][28] In the subset of patients in our study who developed cardiotoxicity and had pre-and posttreatment left ventricular ejection fraction (LVEF) and GLS data available, a GLS decline was observed in the absence of a meaningful decrease in LVEF. This small dataset is congruent with the ndings from Awadalla et al, who demonstrated GLS decreases were lower in patients with ICI induced myocarditis compared to control patients and was associated with the development of major adverse cardiovascular events. 29 Several studies have suggested a potential role for the early initiation of cardioprotective medications including beta blockers and angiotensin system inhibitors to prevent the development of cardiotoxicity associated with anthracyclines and trastuzumab. [30][31][32] There is limited data regarding their bene ts in the setting of ICI-induced cardiotoxicities. Interestingly, in our study we found that baseline beta-blocker use was associated with increased mortality. There is no reason to consider beta-blockers themselves problematic in patients treated with ICI, rather they likely are a marker of a sicker population with more baseline cardiovascular disease and/or risk factors.
There are several limitations to our study that should be noted. First, varying de nitions of cardiotoxicity have been utilized in the literature, and currently there is no standard reference de nition. Also, this is a study of retrospectively collected clinical data from ICD codes. As such we were not able to con rm these ndings with direct evaluation of the electronic medical records themselves. The use of ICD code groups such as "arrhythmia" and "heart failure" represent a heterogenous collection of diseases thereby impacting the interpretation of the ndings. Key biomarkers and imaging (e.g., GLS, EF) data were not measured for all patients before and after therapy. Therefore, we cannot make de nitive conclusions about these ndings, and we could not de nitively attribute cardiac diagnosis to ICI-induced cardiotoxicity. Another limitation is that there was no compliance assessment of medication use and the dose and duration of ICI use were not easily captured. We only had one patient diagnosed with ICI-related myocarditis, so we could not evaluate the effect of corticosteroids 33 or CTLA-4 agonist abatacept 34 on the outcome of myocarditis patients. Finally, we cannot attribute the nding of increased mortality in patients with ICI cardiotoxicity to the cardiotoxicity as we were not able to perform competing risk analyses and there was no control group. As such, due these various limitations, the data presented should be considered hypothesis generating only and not lead to de nitive conclusions.

Conclusions
The results of this analysis suggest that the incidence of ICI-associated cardiotoxicity may be higher than previously suggested. To better address this important knowledge gap, baseline cardiac assessment may be helpful for certain high-risk individuals (e.g., receiving combination ICI therapy, rapid GLS decline, or a history of cardiac disease). Prospective studies are required to better characterize the incidence of speci c cardiotoxicities and identify risk factors as well as long-term complications.

Consent for publication
Not applicable.

Availability of data and materials
The data that support the ndings of this study are available from OneFlorida but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of OneFlorida with appropriate IRB approval.

Characteristics
All (n=424) Cardiotoxicity (n=62) No cardiotoxicity (n=362) P Values were reported as mean ± standard deviation for continuous variables and frequency (%) for categorical variables.