Cardiovascular events after cancer immunotherapy as oncologic emergencies: Analyses of 610 head and neck cancer patients treated with immune checkpoint inhibitors

Cardio‐oncology and emergency medicine are closely collaborative, as many cardiac events in cancer patients require evaluation and treatment in the emergency department (ED). Immune checkpoint inhibitors (ICIs) have become a common treatment for patients with head and neck cancer (HNC). However, the immune‐related adverse events (irAEs) from ICIs can be clinically significant.


| INTRODUCTION
The Emergency Department (ED) is a frequent site of urgent and emergency care for millions of cancer patients.As of January 2019, there were an estimated 16.9 million cancer survivors in the United States (US), and this number is projected to increase to 22.2 million by 2030. 1 Cancer treatment is typically delivered on an outpatient basis; therefore, the ED is a frequent site of care for acute manifestations of complications from cancer and treatment-related toxicities.Cardio-oncology and emergency medicine are closely collaborative.3][4] A recent analysis of a Nationwide Emergency Department Sample in the United States showed that almost 30 million ED visits were made by patients with cancer for the period January 2006 to December 2012. 5Congestive heart failure, cardiac dysrhythmias, and chronic obstructive pulmonary disease were among the top 10 reasons for ED visits by cancer patients for both cancer-specific and all visits.Of these visits, adult cancer-related ED visits resulted in inpatient admissions significantly more frequently (59.7%) than non-cancer-related visits (16.3%). 5ead and neck cancer (HNC) is a common malignancy, with an estimated 54 000 new cases in the United States in 2022. 6While a majority of HNC patients present with locoregionally advanced disease, an estimated 30% of patients present with, or will develop, recurrent or metastatic (R/M) disease. 7The prognosis for patients with R/M HNC is poor, with a median overall survival of 10-13 months. 8,9In 2016, the Food and Drug Administration approved the use of immune-checkpoint inhibitors (ICI) for R/M HNC. 10 ICIs are monoclonal antibodies that target immune checkpoints which prevent immune escape by cancer cells and promote the destruction of cancer cells through the activation of the immune system.Specifically, two programmed cell death protein 1 (PD-1) inhibitors, pembrolizumab and nivolumab, were approved for patients with R/M HNC resistant to platinum chemotherapy. 11In 2019, pembrolizumab either as a single agent or combined with chemotherapy was also adopted as the standard first-line therapy for patients with R/M HNC. 10,11n HNC clinical trials with ICI as monotherapy, the incidence of treatment-related adverse events (trAEs) of any grade ranged from 58% to 64%.][14] About 13%-17% of patients had grade 3-4 (severe) trAEs with pneumonitis, diarrhea, fatigue, and an increase in transaminases being the most common.Many of the trAEs are immune-mediated, and these immune-related adverse events (irAEs), while rare can be fatal.The rate of fatal irAEs was less than 1%, with pneumonitis as the most common cause. 15ecently, studies suggested potential cardiotoxicities associated with ICI therapy. 16,17A retrospective singlecenter matched cohort study of 672 patients treated with ICIs between 2010 and 2020 assessed the incidence of major adverse cardiovascular events (MACE). 18MACE was a composite of acute coronary syndrome, heart failure, and stroke.The results showed that as many as 10.3% of patients developed MACE.In 252 lung cancer patients treated with ICI between 2015 and 2018, Chitturi and colleagues found that MACE occurred in 13.3% of ICI patients, with a median time to event of 51 days. 19nother single institution study of 378 patients receiving ICI treatment between 2014 and 2020 (median follow-up of 50.5 months) showed the incidence of cardiovascular events, defined as electrocardiographic alterations (such as prolonged QT, bundle branch or atrioventricular blockades or atrial fibrillation), congestive heart failure, pulmonary embolism, acute coronary syndrome, pericarditis, and myocarditis, may be as high as 16.7%.Using the data from 2011 to 2018 from the largest US insurance claims database of 2 687 301 patients with International Classification of Diseases (ICD 9/10) codes for any cancer, Jain and colleagues found that among patients who received ICI therapy, the incidence proportion of cardiovascular events were as follow: stroke (4.6%), heart failure (3.5%), atrial fibrillation (2.1%), conduction disorders (1.5%), myocardial infarction (0.9%), myocarditis (0.05%), vasculitis (0.05%), and pericarditis (0.2%). 20Bar and colleagues retrospectively reviewed 1215 non-small cell lung cancer patients treated with ICI and found that approximately 1% of the patients developed a myocardial infarction or stroke within the first 6 months after initiation of ICI therapy. 21hile these studies have begun to show that cancer patients treated with ICIs are at risk for cardiovascular irAEs that may be fatal and potentially require emergency care, few studies have characterized the spectrum of CVD-related irAEs presenting to the ED.Filling this knowledge gap will raise awareness and facilitate early detection, resulting in prompt intervention and favorable outcomes.In this study, we characterized the spectrum of CVD in a large sample of HNC patients who received ICI.Using institutional databases, we evaluated the frequency, timing, and emergency care utilization of CVD in ICI-treated R/M HNC patients.

| Study setting
This is a retrospective cohort study.R/M HNC patients who received ICI between April 1, 2016 and December 31, 2020 were reviewed.This study was conducted at The University of Texas M. D. Anderson Cancer Center, Houston, TX in accordance with ethical principles for medical research involving human subjects and adherence to a clinical research protocol approved by the MD Anderson Institutional Review Board (protocol 2022-0359).

| Study population and data collection
All consecutive cancer patients who were treated at The University of Texas M. D. Anderson Cancer Center with ICI between April 1, 2016 and December 31, 2020 were identified by querying the institution's pharmacy database.Patients with a diagnosis of HNC and aged ≥18 years at the time of ICI initiation were included.Patients who had a second malignancy in addition to their HNC were excluded.

| Study variables
Demographic characteristics, ED visits, hospital admissions and observations, diagnostic ICD-10 codes, cancerrelated clinicopathological data, and vital status data were collected from the institution's data warehouse, tumor registry database, and the electronic medical records.

| CVD-related diagnosis
The billing database was queried for all CVD-related diagnoses (ICD-10, see Table S1, Supporting Information) from the patient's initial presentation to M. D. Anderson up to January 1, 2022.Using the date of initiation of ICI therapy as the index date, all CVD-related diagnoses documented in the billing database before ICI initiation were considered as pre-existing conditions.New cases of CVD-related diagnoses after ICI initiation were considered as potential CVD-irAEs.CVD-related diagnoses reported within 48 h of an ED presentation were considered as ED related, otherwise, it was considered non-ED related.Other related comorbidities included hypertension, peripheral vascular disease, diabetes, and dyslipidemia.

| Statistical analysis
The main characteristics of the final cohort were summarized using descriptive statistics.We reported continuous variables as medians and interquartile ranges or means and standard deviations, as appropriate, whereas categorical variables were reported as counts and percentages.The chi-square test was used to compare frequencies for the categorical variables, while Welch's t test or Wilcoxon-Mann-Whitney test (if the normality assumption was not met) were used to compare continuous variables.We analyzed the reporting of CVD events using a competing risk model, with death as a competing event. 22umulative incidence functions, which measure the subdistribution of the cardiac event were estimated for nivolumab and pembrolizumab.A competing risk analysis of patients with no baseline heart disease who were treated with pembrolizumab was used to investigate a doseresponse relationship with the occurrence of CVD, reporting the subdistribution hazard ratio (SHR) and its 95% confidence interval (95%CI).All statistical analyses were performed using R software (version 4.1.2,The R Foundation; http://www.r-project.org).

| RESULTS
A total of 610 adult R/M HNC patients with a median age of 64 years were included in the study (Figure 1 and Table 1).There were more men (n = 493; 81%) than women, and the cohort was predominantly white (79%).As many as 26% of patients had pre-existing hypertension.Dyslipidemia, type 2 diabetes mellitus, and peripheral vascular disease were reported in 15.4%, 8.5%, and 4.1% of patients, respectively.The median for Charlson Comorbidity Index (CCI) was 6 (interquartile range = 6-7), and the mean body mass index was 25 (standard deviation = 6.8).Most patients received only one ICI (n = 558), with pembrolizumab (n = 373) and nivolumab (n = 91) being the most widely prescribed.Concurrent cytotoxic chemotherapy or targeted chemotherapy were ordered in 20.0% and 8.5% of the patients, respectively.Almost one third of the patients (35.1%) had previous or concurrent radiotherapy.
Using ICD-10 codes, we found that as many as 25.7% of HNC (n = 157) patients had at least one pre-existing CVD-related diagnosis (Table S2) and 55% (n = 86) of these patients had at least one CVD-related diagnosis after ICI.Of the remaining 453 patients without preexisting CVD, 35.5% (n = 143; p < 0.05) had at least one CVD-related diagnosis after ICI initiation (Table S3).
Table 2 (Panel A) shows a total of 255 CVD-related diagnoses were made after initiation of ICI among the 453 patients without pre-existing CVD-related diagnoses.The top 5 CVD-related diagnoses were tachyarrhythmias (n = 91); other ischemic heart diseases (n = 42); other heart diseases (n = 19); valvular heart disease (n = 19) and conduction disorders (n = 18).Myocarditis and cardiac arrest were documented in six and four patients, respectively.As many as 45% (n = 115) of these diagnoses were ED-related, with a higher proportion of patients receiving a diagnosis of myocarditis and tachyarrhythmias in the ED (Table 2, Panel B).Of the 110 ED visits, 76% were admitted, 13% were observed in an observation unit, and 11% were discharged.We also found that 28.6% of those discharged revisited the ED within 72 h from the time of discharge.
Since cardiovascular events cannot occur after a patient has died, the cumulative incidence for selected CV-related diagnosis was adjusted for overall mortality using competing risk analysis.Figure S1 shows there were no statistically significant differences in the cumulative incidence of CV-related diagnoses for pembrolizumab and nivolumab.
Table S4 shows a comparison of time to diagnosis of CV-related diagnosis between pembrolizumab and nivolumab.The time to diagnosis of myocarditis (pembrolizumab: 5.7 ± 2.2 months (median ± interquartile range),  In the univariate competing risk sub-analysis of patients with no baseline heart disease and were treated with pembrolizumab, we found that patients who had a high (400 mg or more) cumulative dose of pembrolizumab had an increased risk of CVD compared to patients with a low cumulative dose, however, the association was not significantly different (SHR [95%CI]: 2.15 [0.85, 5.42]; p = 0.110).Higher body mass index and concurrent chemotherapy were found to be significantly associated with CVD (SHR [95%CI]: 1.05 [1.00, 1.10]; p = 0.033 and SHR [95%CI]: 1.72 [1.04, 2.83]; p = 0.033, respectively (Table S5).The same trend was observed in the multivariable analysis for the high versus low cumulative dose (SHR [95%CI]: 1.92 [0.76, 4.89]; p = 0.170) (Table S6).

| DISCUSSION
ICI has significantly improved the overall survival of patients with R/M HNC, and although overall well tolerated, IRAEs can occur and may lead to long-term or permanent toxicities/morbidities that affect the quality of life of HNC patients.We found that one in four of our patients had pre-existing CVD-comorbidity, which is consistent with other studies of cancer patients. 23,24After excluding these patients from our cohort, we found that a third of our patients received a diagnosis of potential CVD-irAE after receiving ICI therapy.
To our knowledge, this is the first study to evaluate potential CVD-irAEs in a large sample of HNC patients.In our previous systematic review, 25 we found only 11 published studies (6 case reports/series, 4 clinical trials, and 1 prospective study) on irAEs in HNC patients.We found from case reports/case series a total of eight patients with serious irAEs.Enterocolitis, pneumonitis, 26 autoimmune encephalitis, 27,28 autoimmune hepatitis, 27,29 oral mucositis, 30 Stevens-Johnson syndrome-related conjunctivitis, 30 lichenoid dermatitis, 31 psoriasis, 26 and arthritis 26 were reported.From the pooled 791 patients who participated in clinical trials of PD-1 inhibitors, 11,14,15,32 we identified 46 trAEs, with the most frequent potential irAEs involving the endocrine, cutaneous, and gastrointestinal systems.To date, studies of potential CVD-irAEs in HNC patients are from case reports of myocarditis. 33,34atients with myocarditis and tachyarrhythmias mainly presented to the ED for care.This is not surprising since the symptomatology of these conditions predominantly have acute or subacute onset with changes in vital signs that can be life-threatening.Consistent with other studies, we found that myocarditis presented relatively early after the use of ICIs.While rare, myocarditis is associated with a nearly 50% mortality rate and serious cardiovascular sequelae in up to 46% of patients. 35,36Of the patients who presented to the ED, 76% were admitted, 13% were observed in an observation unit, and 11% were discharged.We also found that 28.6% of those discharged presented to the ED within 72 h from the time of discharge.The high rates of hospitalization and ED revisit attest to the high acuity and potentially serious nature of CV-irAEs.At our institution, patients present to the clinic for routine follow-up, and when new CV symptoms or abnormal vital signs are noted, they are sent to the ED for a full diagnostic workup and management.For the other CVD diagnoses, the cardiology clinic provides comprehensive cardiac care to patients who have co-existing heart disease and who develop cardiotoxicities.
We did not find significant differences in the timing and frequency of occurrence of CV-irAES between pembrolizumab and nivolumab.This is consistent with data from clinical trials. 36Moreover, a recent study of 736 melanoma patients treated with nivolumab or pembrolizumab confirmed no statistically significant differences in toxicity between the two drugs. 37embrolizumab monotherapy has a more favorable safety/toxicity profile than pembrolizumab plus chemotherapy owing to the added toxicity profiles of individual agents. 7s expected, we found a higher percentage of CVDrelated diagnoses after ICI among those with pre-existing CVD.8][19][20] Several mechanisms by which ICIs cause CV-irAEs have been proposed: (1) antigens common to tumor cells and cardiomyocytes may lead to cross-reactivity 38,39 ; (2) ICIs promote the formation of autoantibodies (e.g., autoantibodies reacting to cardiac troponin I 40,41 ; (3) ICIs can lead to increased levels of pro-inflammatory cytokines, some of which can upregulate PD-L1 on cardiomyocytes to protect from immune-mediated cardiac injury; (4) disruption of PD-1/ PD-L1 signaling may promote the development of cardiotoxicity 41,42 ; (5) blocking PD1/PDL1 signaling will inhibit Treg cells, and disrupting the immune homeostasis between cytotoxic T cells and Treg cells in normal myocardial tissue can lead to cardiotoxicity [43][44][45] ; (6) theoretically, the anti-PD-L1 monoclonal antibodies binding to PD-L1 at the surface in cardiomyocytes may mediate killing of cardiomyocytes by NK cells through antibodydependent cell-mediated cytotoxicity 46,47 ; and (7) ICIs may accelerate atherosclerosis; blockading CTLA-4 increases T cells abundance in vascular plaques and exacerbates atherosclerosis in a mouse model. 48Abe et al. found that ICIs can induce senescence-associated secretory phenotype (SASP) to aggravate atherosclerosis. 49,50ost of these mechanisms will lead to clinical manifestation in the short term (days to months) but atherosclerosis may lead to delayed CVD and CV events that occur years later.There is evidence to suggest that ICIs may contribute to plaques progression and coronary events by altering the type of inflammation in atherosclerotic plaques.For example, a matched cohort study including 2842 patient pairs explored whether ICIs are associated with atherosclerotic cardiovascular events and showed that patients receiving ICIs have a 3-fold increased risk of cardiovascular events. 51ICIs are a recent advance in the treatment of HNC, the extent of morbidity and early mortality due to accelerated ICI-induced accelerated atherosclerosis is yet to be fully examined.
There are limitations to our study.First, this is a retrospective, single-center study which may limit the generalizability of our findings.The retrospective study design may have led to the misclassification of potential CVD-irAEs.We did not collect data on diagnostic tests or potential biomarkers.Patients may have visited other EDs and are not a part of our sample.We were also not able to assess CVD-related diagnosis for mono versus combination therapy due to a small sample size.Using ICD-10 codes, while valid, may not be reliable since its use is primarily for billing purposes.Further, our retrospective cohort study only consisted of patients who were treated with ICIs for failing first-line or second-line cancer treatment, lacking a comparison group.However, with the current shift in using ICI to be a first line of treatment for some patients, a future study comparing the risk of CVD in HNC patients treated with ICI to similar patients who were not treated with ICI, can better define the risk of CVD that is attributable to the use of ICI.We also have a small number of patients to enable sub-group analyses for specific CVD diagnoses.Prospective studies with larger cohorts of patients are needed to address causality.Nonetheless, the use of ICI in HNC is still expanding and early and delayed manifestations of CV irAEs are yet to be defined in HNC survivors.Our study is the first to characterize the spectrum of potential CV-irAEs in HNC patients on ICI therapy and CV-related ED visits.Current clinical guidelines addressing cancer treatment-related cardiotoxicity are broad and are primarily based on anthracycline and anti-HER2 agent toxicities.While recent clinical guidelines for diagnosis, grading, and management of irAEs include CV-irAEs, data are lacking to support more comprehensive coverage of all types of CV-irAEs.Given the heterogeneity among HNC patients with respect to cardiotoxicity risk, the timing of onset, clinical presentation, and outcome, efforts for personalized risk stratification, early detection, prevention, monitoring, and management of CV-irAEs during treatment and survivorship are of utmost importance.

F I G U R E 1
Study flow diagram showing the exclusion and eligibility criteria to identify the final cohort for analysis [Color figure can be viewed at wileyonlinelibrary.com]T A B L E 1 Demographics and clinical characteristics for head and neck patients treated with immune checkpoint inhibitor (n = 610) Characteristic N (%) T A B L E 2 New CVD-related diagnoses (Panel A) and ED versus non-ED presentation (Panel B) a Patients may have more than one reported.