The PubMed search yielded 156 articles, of which 32 were review or perspective articles, 2 were published before 2017, 81 did not report adverse effects of CAR-T therapy, and 30 did not investigate cardiotoxic effects of CAR-T therapy (Fig. 1). Eleven articles were found in the PubMed search after exclusion. In addition, 2 articles were found outside of the PubMed search that discussed cardiotoxic effects of CAR-T therapy.

PRIMSA Selection Flowchart

A total of 13 articles were included in our study. In this review, we described and discussed individual papers in alphabetic orders, followed by summary findings, and future perspectives.

Alvi et al. [12] performed a retrospective cohort study composed of 137 patients, with a median age of 62 (IQR: 54–70). 54 patients developed CRS grade 2 or higher, while 6 patients developed CRS grade 3 or higher. 17 adverse cardiac effects were noted, which included 6 cardiovascular deaths, 6 decompensated HF, and 5 new-onset arrhythmias. All 17 patients who experienced adverse cardiac events had CRS grade 2 or higher. All 6 patients who experienced cardiovascular death has BNP levels greater than 3,000 pg/mL. Of the 29 patients who had pre and post echocardiographic data, 8 patients experienced a reduction in left ventricular ejection fraction (LVEF). Additionally, 53 patients also have pre and post troponin levels taken, of which 29 had an increase in troponin levels after CAR-T infusion. It was noted that patients with increased troponin levels were older (64 ± 9.5 years) in comparison to patients who did not experience elevated troponin levels (58 ± 8.7 years) (p = 0.02). Additionally, certain cardiac risk factors were higher in the elevated troponin cohort, such as diabetes, hypertension, hyperlipidemia, atrial fibrillation/flutter, and CAD, although insignificant (p > 0.05). 16 adverse cardiac events occurred in the elevated troponin cohort, compared to one adverse cardiac event in patients who did not experience elevated troponin. Lastly, CRS grade 2 or higher was associated with elevated troponin levels (p < 0.001) as 83% of the elevated troponin cohort experienced CRS grade 2 or higher, compared to only 33% in the non-elevated troponin cohort.

Brammer et al. [13] performed a retrospective study composed of 90 patients, with an average age of 61 ± 10.9 years. Eighty patients developed CRS, of which 44 patients developed CRS grade 2 or higher. Common symptoms of patients with CRS included hypotension (79 patients) and fever. 17 adverse cardiac events were reported, including 11 patients with arrhythmia, 2 patients with myocarditis, 1 patient with heart failure, and no cardiovascular deaths.

Burstein et al. [14] performed a retrospective study composed of 98 patients, with a median age of 10 (Range: 2–27). 24 patients experienced hypotension-requiring inotropic support. In this cohort, no cardiovascular deaths were observed, how one patient suffered from a cardiac arrest that occurred 2 months after CAR-T infusion. This study noted how of the 24 patients that experienced hypotension-requiring inotropic support, 10 patients presented with abnormal echocardiograms, all of which revealed systolic dysfunction. 13 patients had serum cardiac biomarker data available, of which 12 patients presented with abnormal BNP levels. Evaluation of risk factors for patients with hypotension-requiring inotropic support revealed that no significant differences in age, sex, and race were found between patient cohorts who did or did not experience hypotension-requiring inotropic support. However, patients with baseline characteristics of reduced LVEF (p = 0.019) or diastolic dysfunction (p = 0.021) were observed to have increased rates of hypotension-requiring inotropic support.

Fitzgerald et al. [15] performed a retrospective cohort study composed of 39 patients, with a median age of 11 (Range: 5–22) In total 36 patients developed CRS, with 34 patients developing CRS grade 2 or higher. Common symptoms of CRS included prolonged high fevers, tachycardia, and myalgias. In total, 14 patients had an adverse cardiac event; 13 patients had had profound fluid-refractory vasodilatory shock and 1 patient had cardiomyopathy. This singular patient was also observed to have decreased LVEF values and systolic dysfunction. Organ dysfunction was also noted in 25 patients, which was most commonly hepatic dysfunction. In addition, respiratory failures were also studied. Of the 39 cohort, 8 patients had respiratory failure; of which, 5 patients were later diagnosed with pediatric acute respiratory distress syndrome (PARDS).

Ganatra et al. [16] performed a retrospective cohort study composed of 187 patients, with a median age of 63 (Range: 19–80). 12 patients developed new or worsening cardiomyopathy and experienced a decrease in median LVEF from 58 to 37% after CAR-T infusion. Of these 12 patients, 6 experienced heart failure. 13 patients developed new or worsening arrhythmia, and 86 patients developed CRS grade 2 or higher. 4 patients died in the hospital.

Korell et al. [17] performed a prospective cohort study with 137 patients, with a median age of 60 (Range: 20–83). 25 patients experienced hypotension, 5 patients developed new atrial fibrillation. 75 patients developed CRS, of which 5 patients developed CRS grade 3 or higher. 17 patients developed cardiac decompensation. In terms of echocardiographic findings, 8 patients had abnormal echocardiogram changes, 5 of which were related to atrial fibrillation and 3 of which were unspecific. Additionally, 4 patients had a drop in LVEF measures after CAR-T infusion, of which one patient experienced a decrease of LVEF greater than 10%, while three patients experienced a decrease of LVEF below 50%. Both CRS grade 3 (p < 0.001) or higher and ICANS (p = 0.012) were identified as a risk factor for survival.

Lee et al. [18] performed an observational study on 90 patients, with a median age of 68. A total of 11 adverse cardiac events were reported, 10 patients developed atrial fibrillation, while 1 patient developed cardiomyopathy, which was later reported as acute myocarditis evidenced on cardiac MRI. This patient also presented with abnormal echocardiogram findings and reduced LVEF, which was defined as either a decrease of LVEF greater than 10% or a decrease of LVEF below 50%. Cardiac comorbidities were not found to be significantly different between cardiac events and non-cardiac events groups. Patients in the cardiac event group were statistically older, had higher baseline creatinine, and a larger indexed left atrium volume. 31 patients developed CRS grade 2 or above, of which 6 belonged to the cardiac events group and 25 belonged to the non-cardiac events group. Survival outcomes between the cardiac events group and no-cardiac events group was not significantly different. Of the 88 patients with cardiac biomarker data, there was no difference in Day 5 troponin levels between the cardiac events group and non-cardiac events group. However, all patients in the cardiac events group had BNP levels (125.0 ng/dL) larger (p = 0.019) than the no-cardiac events group (63.0 ng/dL). In this study, abnormal troponin levels were defined as greater than 0.03 ng/mL, and abnormal BNP levels were defined as greater than 100 pg/mL.

Lefebvre et al. (2020) [19] performed a retrospective study composed of 145 patients, with a median age of 60 (IQR: 50–66). 31 patients experienced adverse cardiac events. 2 patients died due to cardiac causes. 21 patients experienced heart failure and 11 patients experienced atrial fibrillation. 2 events of other arrhythmias (supraventricular tachycardia and non-sustained ventricular tachycardia) and 2 events of ACS was noted as well. 104 patients experienced CRS grade 2 or higher. It was also noted that CRS grade 3 and 4 were independently associated with adverse cardiac events. Lastly, it was noted that 22 patients had grade 1 diastolic dysfunction, 1 patient had grade 2 diastolic dysfunction, 3 patients had grade 2 diastolic dysfunction, and 21 patients had indeterminate diastolic function.

Lefebvre et al. (2023) [20] performed a prospective observational study composed of 44 patients, with an average age of 58 (± 11). 11 patients experienced CRS grade 2 or higher. No significant differences in clinical or echocardiographic baseline characteristics were noted in patients who did or did not develop CRS. No significant differences in LVEF were noted as well. In the 23 patients who developed CRS of any grade, 14 patients had elevated BNP levels compared to their baseline characteristics (362 pg/mL vs 63 pg/mL). One patient developed HF and one patient developed atrial fibrillation.

Maude et al. [3] performed a single-center phase 1–2a study composed of 75 patients, with a median age of 11 (Range: 3–23). It was noted that 71 patients had a minimum of one adverse event believed to be due to CAR-T infusion. The most common symptoms of patients after infusion included CRS, pyrexia, decreased appetite, febrile neutropenia, and headache. 35 patients developed CRS grade 3 or higher. Additionally, this study recorded 30 patients experiencing neurologic events, which included encephalopathy, confusional state, delirium, tremor, agitation, somnolence, and seizure. Most of these neurologic events occurred following CRS. Lastly, 13 patients developed hypotension after CAR-T infusion.

Neelapu et al. [21] performed a multicenter, phase 2 trial composed of 101 patients, with a median age of 58 (Range: 23–76). It was noted that all 101 patients experienced an adverse events grade 3 or higher. Of these, the most common effects were neutropenia, anemia, and thrombocytopenia. 53 patients experienced CRS grade 2 or higher, of which the most common symptoms included pyrexia, hypoxia, and hypotension (9 patients). 1 patient with CRS experienced sudden cardiac arrest. No cardiovascular deaths were reported in this study. 39 patients also were diagnosed with tachycardia.

Shalabi et al. [22] performed a retrospective study composed of 52 patients, with a median age of 13 (Range: 4–30). 37 patients developed CRS, of which 23 patients developed CRS grade 2 or higher. Of patients who developed CRS, 6 patients also developed cardiac dysfunction. 36 patients also developed tachycardia, and 9 patients developed hypotension—requiring vasopressor support. In addition, one patient suffered from cardiac arrest. Between the cardiac event group and no cardiac event group, there was no significant difference in baseline LVEF values (p = 0.59). However, of the 13 patients with echocardiographic data, 4 patients experienced both decreased LVEF values and increased troponin levels. Abnormal LVEF values were defined as a greater than 10% absolute decrease in LVEF. Lastly, of the 7 patients with BNP data, 6 patients reported with abnormal BNP levels.

Shouval et al. [23] performed a retrospective cohort study with 236 patients, with a median age of 65 (IQR: 16). 23 patients experienced cardiac arrhythmias, of which 17 patients experienced atrial fibrillation. 108 patients developed CRS grade 2 or higher. In terms of echocardiographic findings, 12 patients had echocardiograms taken within 90 days of an arrhythmic event. Of those, 3 patients experienced decreased LVEF, which was defined as a decrease in LVEF of at least 10% from baseline to a value below 53%. In addition, of patients who experienced an arrhythmic event, 2/14 had elevated cardiac troponin levels and 6/14 had elevated BNP levels.

We organized summary findings as reports on cardiac events (mostly from ECG), echocardiograms, cardiac blood biomarkers associated with CAR-T cardiotoxicity. We also assessed risk factors and confounding factors associated with CAR-T cardiotoxicity.

Cardiac event summary findings are shown in Table 1. Nine studies [12, 13, 16,17,18,19,20,21, 23] were in adult (N = 1167) and 4 [3, 14, 15, 22] in pediatric (N = 264) patients, with a combined total of 1,431 patients. Patients had predominately DLBCL, PMBCL, ALL, and MCL cancers. All patients were treated in facilities within the US, except for one study center in Israel [21]. Most studies were performed between 2016 to 2022 but a few were performed in 2010 (CAR-T was approved by the FDA in 2017 [24]). Most follow-up time was less than or about a year post therapy with some reported findings during or soon after CAR-T therapy.

Of the adult patients, 60.0% were male and 89.0% were white. Roughly half of all adult patients (42.6%) had CRS grade 2 or higher. The most prevalent cardiotoxic effects were hypotension-requiring inotropic or vasopressor support (34.5%) and tachycardia (12.2%). Note that hypotension and tachycardia are part of the CRS grade definition. The remaining cardiac event findings included heart failure/decompensation (6.3%), atrial fibrillation (5.9%), new or worsening cardiomyopathy (4.7%), arrhythmia (5.8%), myocarditis (1.7%), cardiac arrest (1.7%), and cardiovascular death (1.5%).

Of the pediatric patients, 58.7% were male and 75.2% were white. Pediatric patients had higher incidence of tachycardia (69.2% vs 12.2%) but lower incidence of hypotension (20.4% vs 34.5%) compared to adults. The physiological differences between adult pediatric patients contributed to the differences in tachycardia and hypotension, namely, pediatric patients are more susceptible to tachycardia and less susceptible to hypotension. Pediatric patients showed similar prevalence of CRS grade ≥ 2 (43.9% vs 42.6%), new or worsening cardiomyopathy (2.6% vs 4.7%), arrhythmia (0.0% vs 5.8%), cardiac arrest (1.3% vs 1.7%), and cardiovascular death (0.0% vs 1.5%). Studies on pediatric patients are sparse, with only 4 studies and 264 patients to date.

ECG summary findings are shown in Table 2. Seven studies [12, 16,17,18,19,20, 23] were in adult (N = 976) and 3 [14, 15, 22] in pediatric (N = 189) patients, with a combined total of 1,165 patients. Patients had predominately DLBCL, PMBCL, ALL, and MCL cancers. Most follow-up time was less than or about a year post therapy with some reported findings during or soon after CAR-T therapy.

Of the adult patients, 60.7% were male and 87.4% were white. Roughly half of all patients (47.7%) had CRS grade 2 or higher. The most prevalent echocardiographic changes were systolic dysfunction (9.0%) and diastolic dysfunction (17.9%). Other echocardiographic changes were categorized as abnormal echocardiogram findings (2.3%). Definitions for abnormal echocardiogram findings varied among papers.

Of the pediatric patients, 59.2% were male and 75.2% were white. Pediatric patients had higher incidence of abnormal echocardiograms (41.7% vs 2.3%) and systolic dysfunction (19.7% vs 8.4%) compared to adults. No pediatric data was available for diastolic dysfunction. Pediatric patients had slightly lower incidences of CRS grade ≥ 2 (42.9% vs 47.7%).

Biomarker findings are shown in Table 3. There were only 6 [12, 14, 18, 20, 22, 23] studies (N = 657 patients) reporting biomarkers, of which 65.6% were male, and most patients were (85.1%) white. About one-third (38.4%) of all patients had CRS grade ≥ 2. Greater than half (59.5%) of all patients had abnormal BNP levels, and roughly a quarter (26.4%) of patients had abnormal troponin levels. Definitions for abnormal BNP and troponin levels varied significantly across papers.

Only four papers statistically modeled risk factors for cardiotoxicity to our knowledge. In the pediatric population, Burstein et al. [14] found pretreatment blast percentage > 25% on bone marrow biopsy was the most predominant factor associated with increased risk for cardiac events (odds ratio, 15.5; 95% CI, 5.1 to 47.1; P < 0.001) (N = 98 patients). Patients with lower ejection fraction (p = 0.019) or diastolic dysfunction (p = 0.021) before treatment had increased rates of hypotension-requiring inotropic support. Notably, pre-existing cardiomyopathy (p = 0.062), TBI (p = 0.629), or higher anthracycline dose (p = 0.444) were not associated with hypotension-requiring inotropic support. In an adult study (n = 236), Shouval et al. [23] found that risk factors for atrial arrhythmia post CAR-T treatment were a history of atrial arrhythmia (OR = 6.80 [2.39–19.6]) and using CAR-T product with a CD28-costimulatory domain (OR = 5.17 [1.72–18.6]) in a multivariable analysis. Lefebvre et al., [19] found risk factors associated with a primary endpoint of MACE (defined as cardiovascular death, symptomatic heart failure, acute coronary syndrome, ischemic stroke and de novo cardiac arrhythmia) included prior atrial fibrillation (p = 0.035), aspirin use (p = 0.034), statin use (p = 0.025), insulin use (p = 0.005), baseline creatinine levels (p = 0.026), overall CRS grading (p < 0.001), CRS grade 2 (p = 0.029), CRS grade 3 (p = 0.002), CRS grade 4 (p < 0.001), diastolic blood pressure (p = 0.007), hemoglobin (p = 0.035), platelet count (p = 0.027), and a higher mitral E/e’ ratio (p = 0.046) using univariable Cox proportional cause-specific hazards regression. Multivariable Cox proportional cause-specific hazards regression analysis determined that baseline creatinine and Grade 3 or 4 CRS were independently associated with MACE. Alvi et al. [12] found a longer duration from the development of CRS to the administration of tocilizumab to be associated with an increased likelihood of having a positive troponin with a hazard ratio analysis (adjusted p = 0.008).

Given the small sample size and the large number of covariates, results from these statistical modeling need to be interpreted with caution.

With CAR-T therapy FDA approval in 2017 [24], it is not surprising that the sample sizes reported in the literature are small. Future studies need to further include larger cohorts, more diverse races and ethnicities to achieve broader generalizability. Future studies should include cardiac computed tomography and magnetic resonance imaging, relevant blood biomarkers, as well as other tests (such as cardiac stress test). Future studies should also include long-term monitoring of cardiac function post CAR-T treatment. The contributions of pre-existing cardiac disease and comorbidities and prior exposure to cardiotoxic therapies to CAR-T cardiotoxicity need further investigation. Statistical modeling is needed to identify risk factors of CAR-T cardiotoxicity. Identifying risk factors could help to refine CAR-T constructs to minimize off-target effects and developing cardioprotective agents for CAR-T therapy.

We did not perform meta-analysis of these cardiotoxicity data from the literature because of small sample sizes and non-uniform reporting. CAR-T also has other side effects (such as neurotoxicity) that were not reviewed in this study.