Adverse Cardiovascular and Pulmonary Events Associated with Chimeric Antigen Receptor T-cell Therapy

Abstract

Background:

Pivotal trials of chimeric antigen receptor T-cell (CAR-T) have identified common toxicities but may have been underpowered to detect cardiovascular and pulmonary adverse events (CPAEs).

Objectives:

To investigate CPAEs associated with commercial CD19-directed CAR-T therapy.

Methods:

In this retrospective, pharmacovigilance study we used the FDA adverse event reporting system to identify CPAEs associated with axicabtagene-ciloleucel and tisagenlecleucel. We evaluated disproportionate reporting by the reporting odds ratio (ROR) and the lower bound of the Information Component (IC) 95% credibility interval (IC₀₂₅>0 is deemed significant). Significant associations were further adjusted to age and sex (adj.ROR).

Results:

We identified CAR-T reports of 2,657 patients, including 546 (20.5%) CPAEs. CPAEs overlapped with cytokine release syndrome in 68.3% (373/546) of the reports. Compared to the full database, CAR-T was associated with over-reporting of tachyarrhythmias (n=74[2.8%], adj.ROR=2.78 [95% CI, 2.21–3.51]), cardiomyopathy (n=69[2.6%], adj.ROR=3.51 [2.42–5.09]), pleural disorders (n=46[1.7%], adj.ROR=3.91 [2.92–5.23]), and pericardial diseases (n=11[0.4%], adj.ROR=2.26 [1.25–4.09], all IC₀₂₅>0). Venous-thromboembolic events (VTEs) were associated only with axicabtagene-ciloleucel therapy (n=28[1.6%], adj.ROR= 1.80 [1.24–2.62], IC₀₂₅>0). Atrial fibrillation (n=55) was the leading tachyarrhythmia, followed by ventricular arrythmias (n=14). Tachyarrhythmias and VTEs were reported more often following axicabtagene-ciloleucel than tisagenlecleucel in an age and sex-adjusted model (adj.ROR=1.82 [1.04–3.18] and adj.ROR=2.86 [1.18–6.93], respectively). Finally, the fatality rate of CPAEs was 30.9%.

Conclusions:

In this largest post-marketing study to date, we identified an association between CAR-T and various CPAEs, including tachyarrhythmias, cardiomyopathy, pericardial and pleural disorders, and VTEs. Our findings should be considered in the multi-disciplinary assessment for and monitoring of CAR-T therapy recipients.

Condensed abstract:

Clinical trials of chimeric antigen receptor T-cell (CAR-T) may have been underpowered to detect cardiovascular and pulmonary adverse events (CPAEs). Using the FDA adverse events reporting system, we identified safety reports of 2,657 patients, including 546 (20.5%) CPAEs. CAR-T therapy was associated with various CPAEs, including tachyarrhythmias, cardiomyopathy, pericardial and pleural disorders, and venous thromboembolism (VTE). CPAEs overlapped with cytokine release syndrome in 68.3% of the reports and carried a considerable fatality rate (30.9%). Furthermore, tachyarrhythmias and VTE were reported more frequently with axicabtagene-ciloleucel than tisagenlecleucel. Our findings should be considered in the care of CAR-T recipients.

Introduction

Chimeric antigen receptor T cells (CAR-T) are genetically engineered T-cells programmed to identify tumor antigens and induce a cytotoxic immune response. This form of immunotherapy has transformed cancer care over the past decade.(1) Axicabtagene-ciloleucel and tisagenlecleucel are CD19-directed CAR-T products. In relapsed or refractory (r/r) diffuse large B-cell lymphoma (DLBCL), response rates and 1-year survival with both products exceed 60% and 50%, respectively.(2, 3) Traditional therapy has yielded a substantially lower response rate (26%) and poor median overall survival (6.3 months).(4) Similarly, tisagenlecleucel resulted in an unprecedented remission rate and 1-year overall survival of 81% and 76%, respectively, in children and young adults with r/r B-cell acute lymphoblastic leukemia (B-ALL).(5) Median survival with alternative FDA-approved agents ranges from 3 to 8 months.(6) These remarkable results led to FDA-approval of axicabtagene-ciloleucel and tisagenlecleucel for r/r DLBCL in adults and tisagenlecleucel for r/r B-ALL in patients up to 25 years of age. Indications for CAR-T are expanding with promising results in indolent and mantle cell lymphoma,(7, 8) as well as multiple myeloma.(9) Therefore, use of CAR-T therapy is expected to increase rapidly in the coming years,(10) exposing a wide range of populations to the treatment.

CAR-T cells are administered as a single infusion after a short course of lymphodepleting chemotherapy, typically the combination of cyclophosphamide and fludarabine. Lymphodepletion produces a favorable immune environment that supports CAR-T cells expansion and activity.(11) Cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) are common reversible CAR-T toxicities. CRS incidence ranges between 60% to 90%, depending on the CAR-T product, tumor burden, and kinetics of T-cell expansion. (2, 3, 5, 12, 13) While CRS and ICANs have been well described, clinical trials may have been underpowered to detect infrequent toxicities.(14) Early data suggest that commercially available CAR-T have cardiovascular and pulmonary adverse events (CPAEs) rates exceeding 10%.(15–20) Alvi et al. reported that among patients experiencing high-grade CRS, 28% experienced a decrease in ejection fraction.(16) Ganatra et al. observed new or worsening cardiomyopathy in 10% of the CAR-T recipients.(17) Single-center studies have also raised concerns regarding arrhythmias and venous-thromboembolic events (VTEs).(18, 20) Since CAR-T is a new therapeutic class, there is a clinical need to understand the spectrum of its toxicities. Furthermore, given the availability of several FDA-approved CD-19-directed CAR-T products and others in the pipeline, the choice of therapy may be informed by the product’s safety profile. Clinical trials include selected populations and may be underpowered to detect CPAEs. Therefore, post-marketing surveillance data from large data repositories may help identify these adverse events (AEs) and inform clinicians about their risks. Here, we aimed to characterize CAR-T cardiovascular and pulmonary toxicities in the real-world setting using the FDA Adverse Event Reporting System (FAERS) database.

Methods

Data sources and study design

This is an observational, retrospective pharmacovigilance study using the FAERS database, a global repository of post-marketing safety reports. The FAERS draws upon voluntary reports of health care professionals and consumers, and reports of manufacturers, mandated by regulation. The database was screened for reports containing any of the following terms for drug trade or generic names between 07–01-2014 and 06–30-2020: “axicabtagene”, “ciloleucel”, “yescarta”, “axi-cel”, “KTE-C19”, “tisagenlecleucel”, “kymriah”, “tisa-cel”, “CTL019”. Only reports citing these as the primary suspects for an AE were included. We identified duplicate reports and retained only the latest case version. This study focuses on the adult population. Therefore, we included patients aged 16 years or older. Tisagenlecleucel is the only FDA-approved CAR-T product for B-ALL and is approved for children and young adults up to the age of 25 years. Thus, B-ALL patients older than 25 years were excluded. Since the FAERS is a publicly available and anonymized database, institutional review board approval and informed consent were waived.

To identify patients with cardiovascular risk factors within FAERS, reported concomitant drugs and their indications were extracted. We defined as high cardiovascular risk CAR-T recipients receiving therapy for any of the following indications: hypertension, dyslipidemia, diabetes, arrhythmias, ischemic heart disease, heart failure, valvular disorders, and cardiomyopathies.

AEs in the FAERS are coded using preferred terms of the Medical Dictionary for Regulatory Activities (MedDRA) classification. Due to substantial overlap between different cardiovascular and pulmonary disease symptoms, we grouped MedDRA terms related to the same medical condition (Online Table 1) based primarily on Standardized MedDRA Queries (SMQs).(21) Sinus tachycardia reports were excluded from the tachyarrhythmias category, and atrial flutter reports were included in the atrial fibrillation (AF) sub-category. We defined a category of severe cardiovascular AEs, including tachyarrhythmias, cardiomyopathy, and cardiogenic shock. Additional information for each report in the database includes administrative information (country, reporter occupation, and reporting year), patient characteristics (age and sex), drug administration (indication, dosage, and route of administration), AE occurrence date, and outcome.

To estimate the total number of patients receiving CAR-T therapy in the US, we used the cellular therapy registry of the Center for International Blood and Marrow Transplant Research (CIBMTR), which is estimated to capture 65% of commercial CAR-T products in the United States.(22)

Statistical analysis

We conducted a disproportionality analysis to detect increased reporting of drug-AE combinations in comparison to the full database. To identify significant over-reporting of CPAEs, we used the reporting odds ratio (ROR) and the lower bound of the information component 95% credibility interval (IC₀₂₅), which are validated methods for detecting safety signals in post-marketing passive-surveillance databases.(23–25) These measures evaluate whether a drug-AE combination is disproportionately over-reported relative to background incidence across the entire FAERS database. ROR is a frequentist measure of association, derived from a contingency table in which the exposure is the use of the drug, and the outcome is the AE of interest. A lower bound of the ROR 95% confidence interval (CI) greater than 1 is the accepted threshold for statistical signal detection, similarly to the odds ratio. IC is a Bayesian measure that also indicates whether a drug-AE combination has been reported higher-than-expected (Online Table 2). A positive IC₀₂₅ value is being used for statistical signal detection, and has been shown to reduce false positives even with a small number of cases.(23–25)

We calculated the adjusted ROR (adj.ROR) using a logistic regression model with the full database as the comparator group. The regression model included age, sex, and CAR-T type as covariates and each CPAE as an outcome. Missing age data were imputed separately for non-Hodgkin lymphoma (NHL) and B-ALL as the median within each group. For each CPAE, we evaluated the significance of a nonlinear age term and potential interactions, including significant terms in the final model. Indication for treatment (NHL vs. B-ALL) was not added to the model due to multicollinearity with CAR-T product, as indicated by high variance inflation factor values. Therefore, we performed a sensitivity analysis including only NHL patients to mitigate potential confounding by indication.

As proposed previously, CAR-T products (axicabtagene-ciloleucel and tisagenlecleucel) were compared to one another by a disproportionality analysis, with tisagenlecleucel as reference.(26, 27) We calculated unadjusted and adjusted RORs, but not IC_025, which has been validated only for comparison versus the full database.(25)

Between-group comparisons of continuous and categorical variables were performed using the Mann–Whitney U and Chi-square tests, respectively. The Pearson coefficient measured correlations between AEs. The binomial distribution was used to calculate proportions CI. All tests were two-sided with significance defined as p-values< 0.05. P-values and 95% confidence intervals presented in this study have not been adjusted for multiplicity, and therefore inferences drawn from these statistics may not be reproducible. Data processing and statistical analysis were performed in R statistical software version 3.6.0.

Results

Characteristics of CAR-T reports in the FAERS database

Among 7,549,882 eligible patients in the FAERS database, 2,657 patients treated with an FDA-approved CAR-T were identified between 07–01-2014 and 06–30-2020 (axicabtagene-ciloleucel, n=1,732 [65.2%]; tisagenlecleucel, n=925 [34.8%]) (Online Figure 1). The number of CAR-T recipients in the FAERS has been consistently increasing over time (n=61 [≤2017], 663 [2018], 811 [2019], p<0.001), corresponding with increased activity in the US (Online Figure 2). Health professionals reported most cases (89.7%) (Table 1). NHL was the leading diagnosis (83.7%); 14.4% of patients had B-ALL and were only present in the tisagenlecleucel group. The median age was higher with axicabtagene-ciloleucel compared to tisagenlecleucel (61 [interquartile range (IQR): 53–68] vs. 56 years [23–67], p<0.001). However, among NHL patients only (Online Table 3), the median age was similar (p=0.15) between products.

Table 1:

Demographic and clinical characteristics of CAR-T reports in the FAERS database.

	CAR-T (n= 2657)	Axicabtagene (n= 1732)	Tisagenlecleucel (n= 925)	P.value
Reporting region
Americasa	2299/2656 (86.6)	1517/1713 (87.6)	782/925 (84.5)	0.12
Europe	313/2656 (11.8)	207/1731 (12.0)	106/925 (11.5)	> 0.99
Australia	25/2656 (0.9)	1/1731 (0.1)	24/925 (2.6)	< 0.001
Asia	19/2656 (0.7)	6/1731 (0.3)	13/925 (1.4)	0.02
Reporter
Health professional	2286/2548 (89.7)	1537/1634 (94.1)	749/914 (81.9)	< 0.001
Consumer/Lawyer	262/2548 (10.3)	97/1634 (5.9)	165/914 (18.1)
Event year
≤2017	61/1719 (3.5)	35/1343 (2.6)	26/376 (6.9)	< 0.001
2018	663/1719 (38.6)	552/1343 (41.1)	111/376 (29.5)	< 0.001
2019	811/1719 (47.2)	614/1343 (45.7)	197/376 (52.4)	0.10
2020b	184/1719 (10.7)	142/1343 (10.6)	42/376 (11.2)	> 0.99
Age, years
Median (IQR)	60 (47–68) n=1823	61 (53–68) n=1240	56 (23–67) n=583	< 0.001
Sex
Female	815/2136 (38.2)	529/1391 (38.0)	286/745 (38.4)	0.91
Male	1321/2136 (61.8)	862/1391 (62.0)	459/745 (61.2)
Indication
NHL	1777/2122 (83.7)	1308/1314 (99.5)	469/808 (58.0)	< 0.001
ALL	306/2122 (14.4)	0/1314 (0.0)	306/808 (37.9)
Other	39/2122 (1.8)	6/1314 (0.5)	33/808 (4.1)
Time to AE onset, days
Median (IQR)	3 (1–6) n=1272	3 (1–6) n=972	3 (1–10) n=300	0.02
Min-maxc	0–270	0–270	0–253

Abbreviations: ALL- Acute lymphoblastic leukemia; NHL- Non-Hodgkin lymphoma.

Values are n (%) unless otherwise specified. P-values were calculated for the comparison of axicabtagene-ciloleucel and tisagenlecleucel, with the Bonferroni correction for multiple comparisons.

^a.Among CAR-T recipients from the Americas, 2,277 were from the US (axicabtagene-ciloleucel, n=1,503; tisagenlecleucel, n=774).

^b.Only reports until June 2020 were included.

^c.Value of 0 indicating the AE occurred at the starting day of the treatment.

Of the 2,657 CAR-T recipients safety reports, 1,457 (54.8%) and 546 (20.5%) reported CRS and CPAEs, respectively (Online Figure 1). Concurrent CRS was reported in 68.3% (373/546) of CPAEs reports (Figure 1A), ranging from 50.7% to 100% in specific CPAEs (Figure 1B). Correlations between selected CPAEs are presented in a chord diagram (Figure 1C). All CPAEs were positively correlated with CRS. Weaker correlations were observed between hypotension and other CPAEs, except pericardial diseases.

Figure 1: Overlap of CAR-T-reported cardiovascular and pulmonary adverse events (CPAEs).

(A) Overlap between CRS, cardiovascular, and pulmonary adverse events. (B) Overlap between each CPAE category with CRS. (C) Overlap between CPAEs that were over-reported with CAR-T treatment in the disproportionality analysis. Arc widths represent the strength of the overlap as calculated by the Pearson correlation.

Abbreviations: CRS- cytokine release syndrome; CVAEs- cardiovascular adverse events; IHD- ischemic heart diseases; ILD- interstitial lung diseases; TdP- torsade de pointes; VTE- venous thromboembolism.

CAR-T therapy-associated CPAEs

We identified significantly increased reporting of the following CPAEs compared to the full database (Table 2, Figure 2, Online Table 4): hypotension (n=286 [10.8%], adj.ROR=13.43 [95% CI, 10.78–16.73]), cardiomyopathy (n=69 [2.6%], adj.ROR=3.51 [2.42–5.09]), tachyarrhythmias (n=74 [2.8%], adj.ROR= 2.78 [2.21–3.51]), pericardial diseases (n=11 [0.4%]), adj.ROR=2.26 [1.25–4.09]), cardiogenic shock (n=49 [1.8%], adj.ROR=1.99 (1.50–2.64)], respiratory failure (n=71 [2.7%], adj.ROR=6.01 [4.28–8.44]), pleural disorders (n=46 [1.7%], adj.ROR=3.91 [2.92–5.23]). All over-reporting signals also corresponded to an IC₀₂₅>0 (Online Table 5). Within the tachyarrhythmia category, both AF and ventricular arrhythmias were significantly over-reported (adj.ROR = 3.83 [95% CI, 2.93–5.00], adj.ROR= 2.87 [1.70–4.86], both IC₀₂₅>0). The composite outcome of severe cardiovascular AEs had 3-fold increased reporting compared to the database as a whole. Stratification by CAR-T product revealed a significant over-reporting signal of VTE following axicabtagene-ciloleucel therapy (n=28 [1.62%], adj.ROR=1.80 [95% CI, 1.24–2.62]). Overall, results of the unadjusted and adjusted disproportionality analysis in the entire population were consistent in a sensitivity analysis restricted to NHL patients (Online Tables 6–7), except to pericardial disease.

Table 2:

Disproportionality analysis of CAR-T associated cardiovascular and pulmonary adverse events as compared to the full database

Adverse event	CAR-T (n=2657a)		Axicabtagene (n = 1732a)		Tisagenlecleucel (n = 925a)
	n (%)	ROR (95% CI)	n (%)	ROR (95% CI)	n (%)	ROR (95% CI)
CRS	1457 (54.84)	7797 (7087–8578)	1023 (59.06)	6767 (6077–7535)	434 (46.92)	3034 (2649–3476)
Severe CVAEb	162 (6.10)	2.18 (1.86–2.55)	114 (6.58)	2.36 (1.95–2.86)	48 (5.19)	1.83 (1.37–2.45)
Hypotension	286 (10.76)	11.68 (10.33–13.21)	179 (10.33)	11.15 (9.55–13.02)	107 (11.57)	12.64 (10.33–15.47)
Tachyarrhythmia	74 (2.79)	2.78 (2.21–3.51)	58 (3.35)	3.37 (2.59–4.37)	16 (1.73)	1.71 (1.04–2.80)
AF/AFLc	55 (2.07)	3.63 (2.78–4.74)	42 (2.42)	4.26 (3.14–5.79)	13 (1.41)	2.44 (1.41–4.22)
Vent. arrhythmiasc	14 (0.53)	3.50 (2.07–5.93)	12 (0.69)	4.62 (2.62–8.15)	2 (0.22)	1.43 (0.36–5.74)
Pleural disorders	46 (1.73)	3.99 (2.98–5.34)	32 (1.85)	4.26 (3–6.04)	14 (1.51)	3.48 (2.05–5.89)
Respiratory failure	71 (2.67)	3.31 (2.62–4.19)	41 (2.37)	2.92 (2.14–3.99)	30 (3.24)	4.04 (2.81–5.82)
Cardiogenic shock	49 (1.84)	2.41 (1.82–3.2)	33 (1.91)	2.5 (1.77–3.52)	16 (1.73)	2.26 (1.38–3.71)
Cardiomyopathy	69 (2.60)	1.99 (1.57–2.53)	45 (2.60)	1.99 (1.48–2.68)	24 (2.59)	1.99 (1.33–2.98)
Pericardial diseases	11 (0.41)	2.43 (1.34–4.4)	5 (0.29)	1.69 (0.7–4.07)	6 (0.65)	3.82 (1.71–8.52)
VTE	34 (1.28)	1.46 (1.04–2.05)	28 (1.62)	1.86 (1.28–2.7)	6 (0.65)	0.74 (0.33–1.64)
Cerebral ischemia	36 (1.35)	1.05 (0.76–1.46)	26 (1.50)	1.17 (0.79–1.72)	10 (1.08)	0.84 (0.45–1.56)
ILD	14 (0.53)	0.99 (0.58–1.67)	7 (0.40)	0.76 (0.36–1.59)	7 (0.76)	1.42 (0.68–2.99)
Hemorrhage	87 (3.27)	0.56 (0.45–0.7)	46 (2.66)	0.45 (0.34–0.61)	41 (4.43)	0.77 (0.56–1.05)
Cerebral hemorrhage	14 (0.53)	0.95 (0.56–1.61)	7 (0.40)	0.73 (0.35–1.53)	7 (0.76)	1.37 (0.65–2.89)
QT prolongation/TdP	5 (0.19)	0.93 (0.39–2.25)	3 (0.17)	0.86 (0.28–2.67)	2 (0.22)	1.07 (0.27–4.3)
Hypertension	18 (0.68)	0.35 (0.22–0.56)	9 (0.52)	0.27 (0.14–0.52)	9 (0.97)	0.51 (0.27–0.99)
IHD	12 (0.45)	0.28 (0.16–0.49)	5 (0.29)	0.18 (0.07–0.42)	7 (0.76)	0.47 (0.22–0.98)
ATEd	4 (0.15)	0.69 (0.26–1.85)	1 (0.06)	0.27 (0.04–1.88)	3 (0.32)	1.49 (0.48–4.64)
Vasculitis	2 (0.08)	0.46 (0.11–1.84)	1 (0.06)	0.35 (0.05–2.5)	1 (0.11)	0.66 (0.09–4.69)
Bradyarrhythmias	2 (0.08)	0.33 (0.08–1.32)	2 (0.12)	0.51 (0.13–2.02)	0 (0.00)	-
Dyslipidemia	1 (0.04)	0.09 (0.01–0.65)	1 (0.06)	0.14 (0.02–1.00)	0 (0.00)	-
Myocarditis	2 (0.08)	1.52 (0.38–6.07)	0 (0.00)	-	2 (0.22)	4.37 (1.09–17.49)

Abbreviation: AF- atrial fibrillation; AFL- atrial flutter; ATE- arterial thromboembolism. CRS- cytokine release syndrome; CVAE- cardiovascular adverse event; IHD- ischemic heart disease; ILD- interstitial lung disease; TdP- torsade de pointes; Vent.- ventricular; VTE- venous thromboembolism.

Values in bold are statistically significant, defined as a lower bound of the ROR 95% CI greater than 1 and a positive lower bound of the Information Component 95% CI (IC₀₂₅> 0). IC₀₂₅ values are provided in supplemental eTable 3 and were concordant with ROR values in terms of significance, apart from tisagenlecleucel-associated tachyarrhythmias IC_{0 2 5}, which was not significant.

^a.The total number of unique patients (a patient can have more than one report).

^b.Severe cardiovascular adverse events category includes cardiomyopathy, tachyarrhythmias and cardiogenic shock.

^c.Sub-group of tachyarrhythmias group. Every individual in the database may report more than one AE, thus parent groups are not equal to the arithmetic sum of sub-groups.

^d.Arterial embolic events category does not include cases of cardiac or cerebral ischemia, which were included in designated category.

Figure 2: Adjusted RORs of CAR-T associated cardiovascular and pulmonary adverse events.

RORs and the 95% CI limits of CAR-T associated CPAEs compared to the complete database, adjusted for sex and age by a logistic regression model. Colors represent CAR-T type, and error bars represent the 95% confidence interval (CI). A lower limit of the ROR 95% confidence interval above 1 is considered significant.

* Reports of atrial fibrillation/flutter and ventricular arrhythmias were included in the parent category, tachyarrhythmias.

Abbreviations: CPAEs- cardiovascular and pulmonary adverse events; CVAE- cardiovascular adverse event; ROR- reporting odds ratio; VTE- venous thromboembolic event.

When comparing CAR-T products, axicabtagene-ciloleucel was associated with higher reporting of CRS (n=1,023 [59.1%] vs n=434 [46.9%], adj.ROR=1.40 [95% CI, 1.04−1.90]), tachyarrhythmias (n=58 [3.4%] vs n=16 [1.7%], adj.ROR=1.82 [1.04−3.18]), and VTE (n=28 [1.6%] vs n=6 [0.7%], adj.ROR=2.86 [1.18−6.93]) (Figure 3). Reporting of composite severe cardiovascular AEs was similar between the groups.

Figure 3: Cardiovascular and pulmonary adverse events with axicabtagene-ciloleucel compared to tisagenlecleucel.

Comparison of axicabtagene-ciloleucel and tisagenlecleucel CPAEs reporting by disproportionality analysis. ROR and adjusted ROR (to sex and age) were calculated with tisagenlecleucel as the comparator group.

Abbreviations: Axi- axicabtagene-ciloleucel; Tisa- tisagenlecleucel; ROR- reporting odds ratio; Adj.- adjusted; CPAEs- cardiovascular and pulmonary adverse events; CRS- cytokine release syndrome; CVAEs- cardiovascular adverse events; VTE- venous thromboembolism.

Clinical characteristics of CPAEs

We investigated specific terms frequently occurring in selected CPAE categories (Figure 4A). Within cardiomyopathy reports, the cardiorenal syndrome was common (n=24 [34.8%]). Predominantly reported tachyarrhythmias included AF (n=55 [74.3%]) and ventricular arrhythmias (n=14 [18.9%]: VT [n=10], VF [n=4]). Of note, ventricular arrhythmias were only associated with axicabtagene-ciloleucel therapy (Table 2, Figure 2). Pleural and pericardial disorders were primarily attributed to non-malignant effusions (n=41 [89.1%], n=11 [100%], respectively). The VTE category included 19 (55.9%) reports of deep vein thrombosis and 12 (35.3%) pulmonary embolism reports. The median time from infusion to CPAE ranged between 0 to 2 days, depending on the AE. Notably, the vast majority of CPAEs were reported within the first 10 days following infusion. However, later peaks, before day 30, were reported for VTEs and cardiogenic shock (Figure 4B).

Figure 4: Clinical characteristics of CAR-T associated cardiovascular and pulmonary adverse events.

(A) Number of cases of the most frequent CPAES and corresponding fatality rate (i.e., number of reported deaths / number of reported adverse events). The primary cause of each reported death may be another co-reported adverse event of the same individual. Any individual may report more than one outcome. (B) Days from CAR-T infusion to adverse event onset.

Abbreviations: AF- atrial fibrillation; ARDS- acute respiratory distress syndrome; ARF- acute respiratory failure; CHF- congestive Heart failure; CPAEs- cardiovascular and pulmonary adverse events; CRS- cytokine release syndrome; DVT- deep vein thrombosis; LVF- left ventricular failure; NOS- not otherwise specified; PE- pulmonary embolism; PEA- pulseless electrical activity; SVT- supraventricular tachycardia; VF- ventricular fibrillation; VT- ventricular tachycardia; VTE- venous thromboembolism.

The fatality rate following CPAEs (i.e., number of reported deaths / number of reported AEs) was 30.9%, higher than the fatality rate of CRS (17.4%). In a sensitivity analysis restricted to patients without CRS, the fatality rate of CPAEs was (26.3%) (Online Table 8). Among selected CPAEs, which were over-reported with CAR-T, cardiogenic shock and respiratory failure were associated with the highest fatality rates (80.9% and 66.2%, respectively). Pericardial diseases, VTEs, tachyarrhythmias, cardiomyopathy, and pleural disorders had fatality rates between 27.3% and 40.5% (Figure 4A).

CAR-T related CPAEs were reported more often in patients at high cardiovascular risk than the rest of the cohort (n=79 [38.0%] vs n=467 [19.1%], respectively, p<0.001) (Online Table 9). Nonetheless, CRS reporting was comparable between the groups (n=125 [60.1%] vs n=1332 [54.4%], respectively, p=0.13,). Notably, axicabtagene-ciloleucel recipients were over-represented in high cardiovascular risk group compared to the entire cohort (n=162 [77.9%] vs n=1570 [64.1%], respectively, p<0.001).

Discussion

CAR-T is a rapidly evolving class of anti-tumor therapy, prompting a critical need to characterize the breadth of its potential toxicity. To our knowledge, we report the most extensive clinical characterization of post-marketing CPAEs of commercially available CD19-directed CAR-T therapy. Several CPAEs, including tachyarrhythmias, cardiomyopathy, VTE, pleural and pericardial disease, are significantly over-reported in this post-marketing surveillance study (Central illustration). Importantly, CPAEs can occur with and without CRS and are associated with considerable fatality. Moreover, CRS, tachyarrhythmias, and VTEs are over-reported with axicabtagene-ciloleucel compared to tisagenlecleucel. Overall, our results suggest that cardiovascular and pulmonary toxicity might affect a significant proportion of CAR-T recipients and may be under-appreciated in the literature to date.

Central Illustration: Cardiovascular and pulmonary toxicities of CAR-T therapy in the FAERS.

Pivotal trials of chimeric antigen receptor T-cell (CAR-T) may have been underpowered to detect cardiovascular and pulmonary adverse events (CPAEs). In post-marketing surveillance, CAR-T is associated with tachyarrhythmias, cardiomyopathy, pericardial and pleural disorders, and venous thromboembolism. Tachyarrhythmias and Venous thromboembolic events (VTEs) were reported more often following axicabtagene-ciloleucel than tisagenlecleucel. All CPAEs were positively correlated with CRS, while weaker correlations were noted with hypotension. The observed fatality rates of CPAEs ranged between 27% to 41%. CAR-T-associated CPAEs may be under-represented in the literature and should be considered in patients’ care.

In line with the pivotal trials and real-world experience,(2, 3, 5, 12) CRS is the most commonly reported AE among CAR-T recipients in the FAERS database. The greater likelihood of CRS reporting with axicabtagene-ciloleucel compared to tisagenlecleucel is expected given its more rapid T-cell expansion.(10) CPAEs and CRS had considerable overlap, consistent with previous retrospective studies.(16, 19, 28) In an observational study of 137 patients, Alvi et al. reported 17 cardiovascular events, which occurred exclusively in the setting of grade ≥ 2 CRS.(16) While the pathophysiology of CRS-related cardiac dysfunction remains unclear, myocardial dysfunction secondary to elevated interleukin-6 (IL-6) levels has been postulated to play a role akin to CRS induced by the coronavirus disease 2019 and other infectious and inflammatory states.(29–31) Therefore, there is a strong biologic rationale for treatment with IL-6 inhibitors such as tocilizumab. Indeed, retrospective data support a lower risk of cardiovascular AEs with earlier administration of tocilizumab for CRS.(16) Importantly, CPAEs were independent of CRS in 31.7% of the reports, suggesting other off-target effects of CAR-T or lymphodepletion chemotherapy (32).

In this pharmacovigilance study, we describe the extent, timing, and characteristics of CAR-T-related cardiovascular toxicities. Within the FAERS database, tachyarrhythmias and cardiomyopathy were the cardiovascular AEs with the highest RORs other than hypotension. Corresponding fatality rates were substantial, 36.5% and 36.9%, underscoring their importance. While there are no uniform criteria for CAR-T-related toxicity, considering hypotension as a cardiovascular AE is debatable since it could be an epiphenomenon of CRS and capillary leak. Likewise, sinus tachycardia is a frequent CRS component.(31) Therefore, we excluded its related terms from the tachyarrhythmias category. Tachyarrhythmias were still over-reported, highlighting their clinical implications beyond merely a CRS manifestation. The leading tachyarrhythmias were AF (74.3%) and ventricular arrhythmias (18.9%). Our findings accord with a retrospective study of 145 adult patients treated with commercial and investigational CAR-T; 14 out of 41 (34.1%) major adverse cardiovascular events were attributed to arrhythmia.(19) In another observational analysis, new-onset arrhythmia was observed in 9.1% of patients with grade≥ 2 CRS (n=5/55: 3 AF, 2 supraventricular tachyarrhythmias).(16) Overall, there is a correlation between CRS and tachyarrhythmia, with a possible contribution of catecholamine surge to both.(33) Thus, increased tachycardia reporting with axicabtagene-ciloleucel compared to tisagenlecleucel in the FAERS might be secondary to higher CRS rates in the former. The early timing of CPAE reports and the overlap with CRS timing further support the potential mechanistic link between CRS and CPAEs. Vigilant monitoring of vital signs and ECG is appropriate in patients treated with CAR-T, especially among those who develop CRS.(30)

Cardiomyopathy is emerging as a frequent CAR-T-related AE. Alvi et al. observed troponin elevation, reflecting myocyte damage and necrosis, in 29/54 (53.7%) adult CAR-T recipients.(16) In a pediatric and young adults cohort, 6/7 of patients had an elevated pro-B-type natriuretic peptide (pro-BNP), a biomarker of myocardial stretch injury, at CRS onset.(28) In another study, of 116 CAR-T recipients evaluated for cardiotoxicity by serial echocardiograms, mainly in the setting of grade ≥2 CRS, 10.3% developed new or worsening cardiomyopathy, with a decline in median LVEF from 58% to 37%.(17) Since troponin measurement and echocardiographic assessment are often event-driven, the true incidence of CAR-T-related cardiomyopathy is likely underestimated. Our findings support this concern, as cardiomyopathy was significantly over-reported in the FAERS. Cardiomyopathy is of clinical significance given the persistent cardiac dysfunction in up to 50% of the cases(17) and the considerable associated fatality rate in the FAERS. Moreover, the extent of cardiorenal syndrome reporting (n=24, 34.8% of the cardiomyopathy reports) suggests potential multiorgan involvement.(31) Therefore, efforts to identify patients at risk and to implement early aggressive treatment of high-grade CRS are warranted.

The current study demonstrates an association between pericardial diseases (mostly non-malignant pericardial effusion) and CAR-T. Pericardial diseases were not observed in clinical trials, and a single case of pericardial effusion was reported in a cohort of 60 DLBCL patients treated with a CD19-directed CAR-T.(20) We report the largest series of pericardial disorder cases (n=11), consistent with a recent analysis of the WHO’s VigiBase.(34) Pleural disorders, especially non-malignant pleural effusion, are also over-reported in CAR-T recipients. Fluid retention in CAR-T recipients may partly explain these potential complications.(5, 13) It may result from CRS and capillary leak, although granular data on volume overload manifestations are lacking. Nonetheless, reports of both pericardial and pleural disorders were independent of CRS in approximately 35%, and there was no apparent correlation with cardiomyopathy, suggesting mechanisms other than fluid overload. Pericardial and pleural disease have also been reported with immune checkpoint inhibitors (ICIs), and ongoing studies combine ICIs with CAR-T to increase anti-tumor efficacy.(10) The current findings suggest that caution should be exerted due to the potential for serosal inflammation.

Active hematological malignancy, hospitalization, central venous catheters, and CRS may each contribute to CAR-T recipients’ thrombogenic state. We describe an association between CAR-T therapy and VTE and provide the largest number of such reports to date (n=34). In a recent retrospective study of adults treated with CD19-directed CAR-T, 16 of 148 patients (10.8%) developed VTEs in the first 100 days after infusion;(18) as expected, VTEs were more common among patients with CRS. CRS and VTE were also overlapped markedly in our analysis, and the short time of onset (median 1-day post-infusion) supports CRS as a potent driver. The excess of VTE reports with axicabtagene-ciloleucel compared to tisagenlecleucel may therefore be related to the former’s higher propensity for inducing CRS. In addition, 12 (35.3%) of the VTE reports were pulmonary embolism, emphasizing their clinical importance. Since effective measures to prevent VTEs are available, future efforts should aim to better characterize the burden of VTE in CAR-T recipients and explore strategies for prevention.

Early identification of CAR-T recipients at risk for CPAEs could facilitate preventive interventions.(35) However, data on the role of cardiovascular risk factors in predicting major adverse cardiac events among CAR-T recipients are scarce, and patients with systolic dysfunction were excluded from the pivotal trials.(2, 3) Based on concomitant drug reporting in FAERS, we defined a population with cardiovascular risk factors. As expected, CPAEs and severe CVAEs were over-reported in this group. While our approach likely underestimates the incidence of cardiovascular comorbidities, it does highlight the need to screen CAR-T candidates for cardiovascular risk factors. Future studies should systematically address the impact of standard cardiovascular risk factors (e.g., hyperlipidemia, hypertension, diabetes, and smoking) and those unique to cancer patients (e.g., exposure to anthracycline and radiation, and tumor burden) on the outcome of CAR-T recipients.

Limitations

We analyzed real-world data using the FAERS, the largest global repository of post-marketing drug surveillance encompassing nearly 8 million patients between 2014 and 2020. However, the incidence of CPAEs following CAR-T therapy cannot be determined because the number of patients exposed to the drugs is unknown. Based on CIBMTR registry data,(22) we estimate that 3,100 patients received axicabtagene-ciloleucel or tisagenlecleucel in the US between 2016 and 2019 (Online Figure 2). Given that approximately 425 CPAEs were reported to the FAERS from the US in the corresponding period, and considering potential under-reporting, we estimate a CPAEs incidence of at least 425/3,100 (13.7% [(12.5%−15.0%]).

Reports to FAERS are voluntary. Therefore, granular data on patient, disease, and treatment features are limited. Since tisagenlecleucel is only approved for B-ALL patients up to the age of 25, results could be confounded by indication.(26) We addressed this potential bias by adjusting for age in the multivariable regression models. In addition, sensitivity analyses restricted to NHL, recapitulated findings in the complete cohort. We compared axicabtagene-ciloleucel and tisagenlecleucel using a disproportionality analysis, as previously suggested.(26, 27) However, comparing CAR-T products is limited by potential unbalanced reporting and differences between the populations.(26) Clinical trials are mandatory for drug safety evaluation and AE identification. Nevertheless, they are often under-powered to detect the full spectrum of toxicities and may not reflect real-world recipients(14). Therefore, post-marketing safety reports provide a complementary information source for identifying both common and uncommon drug AEs. Notably, pharmacovigilance studies have been instrumental in detecting rare side effects following ICIs, particularly cardiovascular toxicities.(26, 27)

Conclusions and clinical implications

In this pharmacovigilance study, commercially available CD19-directed CAR-T therapies were associated with various CPAEs and carried substantial mortality. Arrhythmias, cardiomyopathy, pericardial and pleural disorders, and VTEs are over-represented in CAR-T recipients and may be under-reported in clinical trials. Our results have implications at multiple time points in the CAR-T recipient’s journey. When screening candidates for CAR-T, risk factors for cardiac toxicity and VTE should be actively identified. Detailed cardiovascular history and physical examination should be performed before initiation of treatment. Patients with pre-existing cardiovascular disease, multiple cardiovascular risk factors, or active cardiovascular symptoms, will benefit from further risk stratification and optimization of cardiovascular status. An echocardiogram before CAR-T therapy is often performed to assess biventricular systolic function and exclude significant valvular disease, particularly among high-risk patients.(30) Preventing CPAEs is the most effective approach to improve patients’ outcomes. Since CRS and CPAEs are strongly correlated, aggressive management of CRS with early administration of IL-6 inhibitors may mitigate toxicity. Additional interventions that should be evaluated prospectively include cardio-protective drugs and VTE prevention by pharmacological and mechanical methods. Monitoring is another critical element for early identification of CPAEs; troponin surveillance may have a role as it is associated with subsequent cardiovascular events(16, 30), and echocardiographic assessment should be considered among patients developing high-grade CRS.(17) As most events were reported within the first 30 days post-CAR-T, close monitoring should be practiced during this time frame. Finally, since several CAR-T toxicities and ICIs overlap, there is a potential for increased toxicity in patients previously exposed to ICIs or those receiving both therapies concurrently. In summary, our findings of excessive cardiovascular AE reporting in CAR-T recipients should be considered in patient care and clinical trial design.

Clinical Perspectives.

Competency in Patient Care:

Post-marketing reports of chimeric antigen receptor T-cell (CAR-T) therapy indicate increased incidences of tachyarrhythmias, cardiomyopathy, pericardial and pleural disorders, and venous thromboembolism. Overlap of these adverse cardiovascular and pulmonary events with cytokine release syndrome (CRS) were associated with a high risk of fatality.

Translational Outlook:

Further research is needed to understand the mechanisms responsible for adverse cardiovascular and pulmonary events associated with CAR-T therapy and identify subgroups of patients at greater or lesser risk.

Abbreviations:

adj.ROR

Adjusted Reporting Odds Ratio

AEs

Adverse events

B-ALL

B-cell acute lymphoblastic leukemia

AF

Atrial fibrillation

CPAEs

Cardiopulmonary adverse events

CIBMTR

Center for International Blood and Marrow Transplant Research

CAR-T

Chimeric antigen receptor T-cell

CRS

Cytokine release syndrome

DLBCL

Diffuse large B cell lymphoma

FAERS

FDA Adverse Event Reporting System

ICIs

Immune checkpoint inhibitors

IC₀ ₂ ₅

Information component 95% credibility interval

NHL

Non-Hodgkin lymphoma

r/r

Relapsed or refractory

VTEs

Venous thromboembolic events