The study protocol was approved by the medical Ethical Review Boards of the Tianjin Medical University Second Hospital (number: KY2018K090). The study was registered with the Chinese Clinical Trial Registry (registration number: ChiCTR18000162216). All participants provided informed consent. A total of 55 consecutive patients hospitalized in the Urology department and Oncology Department of Tianjin Medical University Second Hospital from June 2018 to October 2020 who were naive to ICI treatment were enrolled in the study.

The inclusion criteria included the following: (1) 18–90 years old, (2) cancer confirmed by histopathological results, and (3) normal liver and renal function.

The exclusion criteria were patients with existing (1) heart failure, (2) myocardial infarction, (3) myocarditis, (4) cardiomyopathy, (5) severe valvular diseases, (6) patients who were pregnant or were lactating, (7) patients with poor echocardiography image qualities, and (8) those with less than 6 months survival.

Laboratory findings included cardiac enzymes [troponin, creatine kinase (CK), and creatine kinase MB (CKMB)], blood routine test (hemoglobin and platelet levels), liver function [albumin, alanine transaminase (ALT), aspartate aminotransferase (AST), and total bilirubin], renal function (urea, creatinine, and uric acid), glucose and thyroid function [triiodothyronine (T3), tetraiodothyronine (T4), and thyroid-stimulating hormone (TSH)], and clinical history including gender, age, height, weight, blood pressure, heart rate, hypertension, diabetes mellitus, heart diseases, smoking, and alcohol intake history, which were recorded at baseline. Transthoracic echocardiograms were performed before immunotherapy, and pre-ICI and after ICI administration.

The echocardiographic images were performed in the left recumbent position of the patients. Echocardiographic measurements were based on current guidelines for assessing cardiovascular structure and function. All patients underwent transthoracic echocardiography using a Philips IE33 ultrasound system equipped with an X5-1 (2.5–3.5 MHz) probe. Images of at least five cycles of sinus rhythm were digitally stored in the original Digital Imaging and Communications in Medicine format for offline analysis. The images were recorded in the following standard views: parasternal long-axis view, parasternal short-axis view, apical four-chamber view, apical three-chamber view, and apical two-chamber view.

Left atrial diameter (LAD), left ventricular end-diastolic diameter (LVEDD), left ventricular end-systolic diameter (LVESD), left ventricle posterior wall thickness (LVPWT), and interventricular septum thickness (IVST) were measured in the parasternal long-axis view. The right ventricular area change rate (RVFAC) was calculated via the maximum and minimum areas of the right ventricle [(RVAmax–RVAmin)/RVAmax]. The tricuspid annular plane systolic excursion (TAPSE) was measured in the M mode of the four-chamber view. Mitral inflow velocities and mitral annulus velocities were obtained by pulsed and tissue Doppler echocardiography, respectively. Left atrial volume (LAV) was measured in three-dimensional echocardiography.

The QLAB 10.8 software was used to analyze the images offline. The left ventricle was divided into 17 myocardial segments, and the GLS was calculated as the average longitudinal strain at the end of the systole. The left ventricular global circumference strain (GCS) was measured in the parasternal short-axis views (SAX B, SAX M, and SAX A) according to the software instructions. The right ventricular longitudinal strain and free-wall longitudinal strain were measured accordingly at AP4 view. Left ventricular end-systolic volume (LVESV), left ventricular end-diastolic volume (LVEDV), and 3D LVEF were calculated in three-dimensional echocardiography imaging of a four-chamber view.

Cardiac toxicity was defined as LVEF less than 53% or decreasing more than 10% in LVEF or 15% in GLS when compared with the baseline value (6, 8). In addition, cardiac toxicity was also defined as a >5% decrease in LVEF along with the presence of heart failure symptoms, troponin increased extending normal range combined with the patient having the sign of myocarditis, cardiovascular death, cardiogenic shock, and cardiac arrest. TAPSE less than 16 mm represented right ventricular systolic dysfunction (9). Subclinical cardiac toxicity was defined as early changes in left ventricular and right ventricular mechanics after ICI treatment, but clinical symptoms of ICI-associated were absent or happened later. The cutoff follow-up time was 220 days. According to reports, the cutoff of LVGLS, RVGLS, and RVFWLS was –18, –17.3, and –19.5%, respectively (10–12).

The IBM SPSS statistics 20.0 software (IBM Corporation, Armonk, NY, USA) was used for statistical analysis. All continuous variables were expressed as means ± standard deviation and compared by paired t-test. Categorical data were presented as numbers (%) and compared through the chi-square test. The data were analyzed using two-tailed tests, and P < 0.05 indicated that the difference was statistically significant. Subgroup analysis was performed. Cox regression was employed for survival analysis. ICI-related cardiotoxicity occurred time and survival cumulative percentage was analyzed through the Kaplan–Meier curve, and the P-value was calculated by log-rank test. The receiver operator characteristic (ROC) curve was done for analyzing LVGLS.

In total, 55 patients met the criteria of this study. Among them, 40 (72.3%) were male patients with a mean age of 62, 17 patients had hypertension, nine patients had diabetes mellitus, five patients had a history of cardiovascular diseases, and 23 patients had a smoking history. Their cardiac enzymes (troponin, CK, and CKMB), routine blood results (hemoglobin and platelet), and liver and renal function (albumin, ALT, AST, total bilirubin, urea, creatinine, and uric acid) were within normal ranges. The baseline clinical data of the patients are shown in Table 1. All patients had a normal left ventricular ejection fraction of more than 53% at the baseline. The most common indications for ICIs were urinary system malignancies and lung cancer (87.3%) (Table 1). Patients were more likely to have received single ICI therapy and later combination ICI therapy with chemotherapy, and clinical laboratory results are shown in Table 2. Compared with the pre-ICI data, all the aforementioned clinical laboratory results were similar to the post-ICI data, except for the AST parameter, though AST was in the normal range either before or after ICI treatment.

Compared with the pre-ICI data, RVFAC [(52.16 ± 12.18)% vs. (48.62 ± 9.84)%, P = 0.064] had no significant change after ICI treatment. mPAP (16.77 ± 8.71 vs. 22.03 ± 9.08, P = 0.0001), IVST (9.15 ± 1.86 vs. 9.74 ± 1.95, P = 0.004), LVEDD (46.03 ± 4.06 vs. 47.27 ± 3.59, P = 0.01), LVEDS (25.47 ± 3.86 vs. 27.39 ± 4.34, P = 0.002), LVPWT (8.75 ± 1.41 vs. 9.39 ± 1.41, P = 0.0001), E/e’ ratio (10.09 ± 2.51 vs. 11.04 ± 2.45, P = 0.011), LAD (35.09 ± 4.93 vs. 35.87 ± 4.63, P = 0.038), and LAV (30.40 ± 10.17 vs. 34.79 ± 12.32, P = 0.005) were significantly increased and LV3DEF [(60.89 ± 11.77)% vs. (57.00 ± 8.18)%, P = 0.022] was significantly decreased, while they were in normal range either before or after ICI treatment. Compared with the baseline data, the LVEF, LVGLS, LVGCS, TAPSE, RVGLS, and RVFWLS were significantly worsened after ICI treatment [(64.49 ± 4.36)% vs. (60.75 ± 4.43)%, P = 0.0001; (–18.63 ± 2.53)% vs. (–17.35 ± 2.58)%, P = 0.000; (–28.83 ± 5.09)% vs. (–24.98 ± 4.51)%, P = 0.0001; 18.29 ± 6.23 vs. 14.57 ± 3.81, P = 0.0001; (–18.45 ± 4.65)% vs. (–14.98 ± 3.85)%, P = 0.0001; and (–25.49 ± 6.42)% vs. (–21.33 ± 6.00)%, P = 0.001, respectively] (Figure 1). LVEF, LVGCS, and RVFWLS were in the normal range either before or after ICI treatment. LVGLS (–17.57 ± 2.65, P = 0.001), TAPSE (14.57 ± 3.81, P = 0.0001), and RVGLS (–14.98 ± 3.85%, P = 0.0001) were below their normal range individually after ICI immunotherapy. Echocardiography results are shown in Table 3. Subgroup analysis of LVGLS showed similar results in age, gender, and drug subgroups (Table 4).

Fifty-five patients enrolled in the present study were followed up for 220 days. Six (10.9%) patients died, 15 (27.3%) patients had more than 15% LVGLS decrease, 10 (18.2%) patients had cardiac toxicity (LVEF reduced by 10% and less than 53%), and four (7.3%) patients developed interstitial pneumonia. The KM curve analysis showed that the standard of LVGLS decreased by more than 15% and was more sensitive than the cardiac toxicity events to assess ICI-related subclinical cardiac toxicity, but there was no statistical significance (log-rank P = 0.205) (Figure 2). Kaplan–Meier curve analysis demonstrated that patients with decreased LVEF, LVGLS, RVGLS, and TAPSE tended to have higher rates of long-term mortality, although statistical significance was not reached (Figure 2). Cox regression analysis clarified two factors significantly related to the survival rates, including the single or combined treatment and LVGLS decreased or preserved (Table 5). The ROC curve for analyzing the LVGLS showed that the area under the ROC curve was 0.693, and the cutoff value of ΔLVGLS was –13% (Figure 3).

Recent reports have expanded the effect of ICI therapy on the cardiovascular system to include an increase in cardiac dysfunction without myocarditis, arrhythmias, venous thromboembolic disease, accelerated atherosclerosis, and atherosclerosis-related cardiovascular events (13). The risk factors associated with ICI-related cardiotoxicity include male sex, sleep apnea, higher BMI, history of radiation, and low neutrophil-to-lymphocyte ratios (14, 15). However, single ICI did not increase the risk of cardiotoxicity compared with chemotherapy and combination immunotherapy did not increase the risk of cardiotoxicity compared with single ICI (16). The association between immune checkpoint inhibitor therapy and an increase in cardiovascular events is not only limited to events occurring within the first few weeks after starting therapy but also can include events that occur months to years after therapy (13). The median time from treatment initiation to clinical manifestation of ICI-related events was 30 days for myocarditis, 30 days for pericardial disorders, and 55 days for vasculitis (17). The current gold standard for myocarditis diagnosis is histological findings of endomyocardial lymphocyte infiltration with myocyte necrosis (18). Transthoracic echocardiogram (TTE) and elevated cardiac biomarkers, including serum troponin, CK-MB, total CK, and natriuretic peptide levels are important tests in the evaluation of suspected myocarditis (19). TTE can also detect other cardiac manifestations, including pericardial effusion or pericardial thickening and intra-cardiac thrombi (19). However, there were no standard examinations to detect ICI-associated cardiac dysfunction. Our study found that LVGLS, RVGLS, and TAPSE were more sensitive indices for the early detection of ICI-related subclinical myocardial toxicity. Given the involvement of immune checkpoints in many aspects of cardiovascular homeostasis, the mechanisms are probably multiple and interrelated (20). Therefore, the change in myocardial mechanics during PD1/PDL1 immunotherapy requires continuous monitoring. For the early recognition of ICI-related cardiac events, a basic assessment of cardiac status, including ECG, echocardiography, and troponin, should be considered before starting ICI therapy (21).

Previous studies found that the cutoff value of RVGLS and RVFWLS was –17.3 and –19.5%, respectively (10, 11). In pathological situations, right ventricular remodeling can occur when ventricular dilatation is produced by pressure overload (22). Because the interventricular septum is shared by the left ventricle and right ventricle, the RVGLS is influenced by both left and right ventricular function. It is reported that RVFWLS was the prior method to assess the right ventricle function of patients with pulmonary arterial hypertension (23). A slight change in right systolic function in patients with tumors could be identified through TAPSE (24). TAPSE less than 16 mm represents right ventricular systolic dysfunction (9). The study showed that compared with the baseline, TAPSE was significantly decreased post-ICIs. Kaplan–Meier curve analysis presented patients with reduced TAPSE had a higher incidence of death. The result coincidentally proved the important role of TAPSE to evaluate right ventricular systolic function.

Left and right ventricular toxicity can occur simultaneously (25). LVGLS and RVGLS were important indicators to assess the function of both chambers and can provide an early indicator of prognosis in patients with cancer. Our research demonstrated that LVGLS and RVGLS after ICI immunotherapy were significantly reduced than the baseline and lower than their normal range. Our results also proved that LV and RV functions were injured simultaneously after the administration of ICIs. Among the 55 patients, 15 patients were diagnosed with subclinical cardiac toxicity, according to the definition of LVGLS, reduced by more than 15%. Adverse cardiovascular events occurred in nine patients. The Kaplan–Meier curve analysis demonstrated that the criteria of LVGLS reduced by more than 15% were more sensitive than cardiac toxicity events to predict subclinical cardiac toxicity. LVEF is a load-dependent index, and preload, afterload, and myocardial contractility can all affect the LVEF value. All of these effects may lead to errors that affect the accuracy and sensitivity of LVEF (4, 12, 26).

Our early findings found no obvious increase in cTnI or NT-proBNP during follow-up, except in two cases. Laboratory parameters, including cTnI or NT-proBNP, were not optimum guidance for cardiotoxicity. Lena et al. (27) reported in EcoR Registry that the number of patients with increased cTnI or NT-proBNP was higher in patients with chemotherapy-induced cardiomyopathy than in patients without cardiotoxicity.