Immune checkpoint inhibitors in cancer: the increased risk of atherosclerotic cardiovascular disease events and progression of coronary artery calcium

This retrospective analysis was performed on consecutive patients with stages III or IV NSCLC at Wuhan Union Hospital from March 1, 2020, to April 30, 2022. Patients treated with ICIs therapy were defined as cases, and those treated with other antineoplastic therapies were defined as controls.

The inclusion and exclusion criteria of patients in our study were as follows. Inclusion criteria are as follows: (1) diagnosis of stages III or IV NSCLC according to the NCCN Clinical Practice Guidelines in Oncology: Non-Small Cell Lung Cancer (Version 3.2022) [25], (2) histologically or cytologically confirmed NSCLC, (3) age > 18 years, and (4) treatment with ICIs or other antineoplastic therapy for at least four consecutive cycles. Exclusion criteria are as follows: (1) combination with other primary malignant tumors, (2) NSCLC treated surgically, (3) then subsequently treated with ICIs (not when treatment started), and (4) incomplete medical records.

In addition, we conducted further screening to include patients with coronary artery calcified plaques before treatment and those who had undergone chest CT scans at 3 months (± 30 days), 6 months (± 30 days), and 12 months (± 30 days) after treatment into the imaging study. Ultimately, we included 1458 patients in the cardiovascular study, 246 of which were further enrolled in the CAC imaging study. After PSM, the two studies included 892 and 150 patients, respectively (Fig. 1).

Patient characteristics at baseline were collected from medical records and included sex, age, and body mass index, cancer characteristics (pathological types and stages), cardiovascular risk factors (hypertension, diabetes, smoking index, hyperlipidemia, history of cardiovascular disease, medications and chest radiation therapy), serum biochemical indices, ICI types (PD-1, PD-L1, and CTLA-4 antibodies), and cycles of use and survival information. In addition, we collected information regarding genetic mutations (EGFR, ALK, and KRAS).

We collected non-ECG-gated CAC data of patients at baseline and at 3, 6, and 12 months (± 30 days) after treatment from chest CT scans (Additional file 1: Fig. S1). Imaging parameters were as follows: slice thickness, 1.5 to 2.0 mm; tube current, modulated automatically; tube voltage, 120 kV; and image reconstruction, standard soft convolution kernel. Two independent radiologists (Zheng C. S. and Yang M. with 26 and 8 years of cardiovascular imaging experience, respectively) manually outlined the region of interest in four coronary arteries (the left main trunk, left anterior descending artery, circumflex, and right coronary artery) utilizing semiautomatic software (CaScoring, Syngo, Siemens Healthineers). The software automatically calculated the CAC volume and score in each coronary artery and the total (with an attenuation ≥ 130 HU and an area ≥ 1.0 mm²). CAC score adopts those described by Agatston et al. [26]. Changes in CAC volume and scores were measured to absolute progression (3, 6, 12 months data—baseline data) and relative progression \(\left(\frac{3,\;6,\;12\;\mathrm{months}\;\mathrm{data}\;-\;\mathrm{baseline}\;\mathrm{data}}{3,\;6,\;12\;\mathrm{months}}\right)\). In addition, according to the Agatston score, the CAC results were stratified as mild calcification (CAC score = 1–99), moderate calcification (CAC score = 100–400), and severe calcification (CAC score > 400) [24]. A third radiologist subjectively classified the image quality of the patient as good, moderate, or poor due to motion artifacts and excluded patients with poor image quality at any time. Furthermore, we randomly selected patients with two CT scans less than 30 days apart, and there was the excellent interscan agreement for CAC score (intraclass correlation [ICC] = 0.92; 95% confidence interval [CI], 0.86–0.96) and volume (ICC = 0.95; 95% CI, 0.91–0.97). All three readers were blinded to patients’ clinical information. Any disagreements were resolved through group discussion.

Patients underwent repeated imaging examinations during outpatient or inpatient follow-up visits before ICI or non-ICI treatment and 3, 6, and 12 months after treatment. The primary endpoint of this study was the occurrence of ASCVD events, which included myocardial infarction, coronary revascularization, and ischemic stroke. Potential events were assessed through the review of imaging, medical records for hospitalizations and outpatients, death certificates, and discharge summaries. A telephonic interview was conducted with each patient or a family member. Four experienced radiologists and cardiologists independently adjudicated each event based on standardized definitions described by Drobni et al. [14] and the 2013 ACC/AHA Guidelines on the Assessment of Cardiovascular Risk [27]. They were blinded to all other data. Any disagreements were reconfirmed through interviews with the next of kin and finalized through discussion and negotiation. Secondary endpoints included the progression of CAC volume and score after treatment. All patients were followed up until April 2023 or death.

The normal distribution of variables was assessed by the Kolmogorov–Smirnov test. Continuous variables were presented as mean (SD) or median (IQR) and categorical variables as counts (percentages). Continuous variables were compared using the t-test or Wilcoxon rank-sum test, whereas categorical variables were compared using the Pearson χ² test or Fisher’s exact test. Cumulative incidence curves were constructed using the Kaplan–Meier method, and differences were evaluated by the log-rank test. To determine risk factors for ASCVD events, we performed univariable and multivariable regression analyses using Cox proportional hazards model. Except for cases with special instructions, all parameters with P < 0.10 in the univariable analysis were included in the multivariable model, and hazard ratios (HR) and 95% CIs were calculated. We also conducted exploratory subgroup analyses based on covariates of clinical interest. HRs with 95% CIs were reported within each subgroup. Imaging data were analyzed at baseline and 3, 6, and 12 months after treatment. The progression of the CAC volume and score between the ICI and non-ICI groups was compared using the Wilcoxon rank-sum test, and the progression of the CAC score grade between the two groups was compared using the chi-square test. Three multiple linear regression models were used to evaluate the association between ICIs and CAC and reported as the correlation coefficient (beta) adjusted for different covariates. Spearman rank correlation was used for correlation scatter plots of cycles of ICI versus the progression of the CAC volume and score. Propensity score matching (PSM) was performed in a 1:1 ratio based on all baseline characteristics to reduce potential confounding factors with the caliper value of 0.05. A P-value < 0.05 was considered statistically significant. Statistical analyses were performed using SPSS (version 26.0; IBM Corp) and R (version 4.3.0; R Foundation) statistical software.

The ASCVD event study involved a total of 1458 patients, including 487 (33.4%) patients who received ICI therapy and 971 (66.6%) patients who did not. After PSM, there were 446 patients in each group. The baseline demographic characteristics of the patients are shown in Table 1, and the baseline laboratory values are presented in Additional file 1: Table S1. Overall, the proportion of male participants in the ICI group was higher than that in the non-ICI group (87.5% vs. 59.7%, P < 0.001). The ICI group had higher proportions of patients aged < 65 years (46.0% vs. 38.6%, P = 0.007) and patients with smoking index > 400 (41.7% vs. 19.8%, P < 0.001). In the ICI group, the proportions of adenocarcinoma and squamous cell carcinoma were comparable, whereas, in the non-ICI group, the most dominant pathological type was adenocarcinoma (44.6% and 48.3% vs. 76.4% and 18.0%, respectively; P < 0.001). Patients in the non-ICI group had more advanced tumor clinical stages (stage IV, 81.9% vs. 71.7%, P < 0.001). After PSM, there were no statistically significant differences in the baseline factors among the two groups. In addition, in the ICI group, PD-1 antibodies were the most commonly used antibodies, followed by PD-L1 antibody. All treatments exceeded four cycles.

A total of 24 ASCVD events (4.9%) occurred in the ICI group during entire period of follow-up (median follow-up 23.1 months), including 12 myocardial infarctions (2.5%), 6 coronary revascularizations (1.2%), and 7 ischemic strokes (1.4%). A total of 14 ASCVD events (1.4%) occurred in the non-ICI group, including 6 myocardial infarctions (0.6%), 4 coronary revascularizations (0.4%), and 4 ischemic strokes (0.4%). After PSM, 24 (5.4%) and 5 (1.1%) ASCVD events occurred in the ICI and non-ICI groups, respectively. The event rates for the total ASCVD and each endpoint increased after ICIs therapy, both before and after PSM. The cumulative incidence of the total and individual component endpoints in the ICI and non-ICI groups is depicted by Kaplan–Meier curves in Additional file 1: Fig. S2 and Fig. 2. In univariable analysis, smoking index, history of cardiovascular disease, neutrophil-to-lymphocyte ratio, and aspirin use were identified as potential predictors, where ICI therapy was associated with a 3.6-fold increase in the total risk of ASCVD events before PSM (HR, 3.6 [95% CI, 1.8–6.9]; P < 0.001). These covariates were further incorporated into the multivariable Cox regression model, showing that the correlation between ICI therapy and ASCVD events was attenuated but still significant (HR, 3.0 [95% CI, 1.5–6.0]; P = 0.002) (Additional file 1: Table S2). After PSM, ICIs, smoking index, and history of cardiovascular disease were identified as potential risk factors to include into multivariable analysis. Finally, ICIs (HR, 5.1 [95% Cl, 1.9–13.4]; P < 0.001) and history of cardiovascular disease (HR, 2.8 [95% Cl, 1.8–6.7]; P = 0.020) were independent risk factors associated with higher risk of ASCVD events (Table 2). We performed independent Cox regression analysis on lipids and cycles of ICI; however, there were no positive findings (Additional file 1: Table S3, Table S4, Table S5). Additionally, EGFR was identified as a potentially protective factor for ASCVD events in univariable analysis (HR, 0.3 [95% Cl, 0.1–0.9]; P = 0.035). We did not find an association between ALK and KRAS with ASCVD events (Additional file 1: Table S6). Increasing risk of ASCVD events with ICI therapy was consistently observed in the subgroups based on baseline characteristics both before and after PSM (Additional file 1: Fig. S3, Fig. S4). Compared with males, the risk of ASCVD events caused by ICIs was relatively higher in females before PSM (HR, 12.6 [95% CI, 3.0–53.1] vs. 2.4 [95% CI, 1.1–5.0]; P = 0.050) (Additional file 1: Fig. S3). However, after PSM, this difference did not reach statistical significance (HR, 5.3 [95% CI, 0.6–45.5] vs. 5.1 [95% CI, 1.7–14.9]; P = 0.979) (Additional file 1: Fig. S4). We compared the survival status of patients with and without ASCVD events and found that patients without ASCVD events had higher survival rates in half a year, 1 year, and 2 years, although this did not reach statistical significance (Additional file 1: Table S7).

A total of 246 patients (113 with ICI therapy and 133 without) were further enrolled in the CAC imaging study cohort. After PSM, 150 patients with balanced baseline characteristics were enrolled in the two groups at a 1:1 ratio. Their baseline characteristics (Additional file 1: Table S8, Table S9) were similar to those in the cardiovascular study cohort. All these patients had calcified coronary artery plaques before treatment, and baseline CAC volume and score did not differ between the two groups both before and after PSM.

The patients underwent chest CT examinations 3, 6, and 12 months after treatment, and their CAC data at the four time points are summarized in Additional file 1: Table S10 and Table S11. Overall, we observed that the ICI group had significantly worse CAC volume and score progression than the non-ICI group did. Bean plots presents the variances in CAC progression in the two groups at the four time points before (Additional file 1: Fig. S5) and after PSM (Fig. 3). Specifically, there was no difference in CAC volume and score progression between the two groups at 3 months after treatment, but differences in volume and score progression emerged at 6 months, and absolute volume progression (29.2 mm³ vs. 10.1 mm³; P < 0.001 before PSM; 22.1 mm³ vs. 8.3 mm³; P = 0.013 after PSM), absolute score progression (34.3 vs. 12.8; P < 0.001 before PSM; 26.6 vs. 12.4; P = 0.002 after PSM), relative volume progression (27.4% vs. 20.1%; P = 0.014 before PSM; 30.4% vs. 19.7%; P = 0.004 after PSM), and relative score progression (30.6% vs. 21.2%; P = 0.016 before PSM; 30.6% vs. 19.4%; P = 0.003 after PSM) of CAC showed significant difference at 12 months after treatment (Additional file 1: Table S12, Table 3). Each coronary artery segment was analyzed independently. Their CAC progression trend was similar to that of the total population, with left anterior descending artery being the most prominent (Additional file 1: Fig. S6A). A similar trend appeared after PSM (Additional file 1: Fig. S6B). In addition, we quantified the CAC score at each time point as the grade representing calcification severity, according to the description of Budoff et al. [24] (Additional file 1: Table S13). The CAC score grade progression was significantly more severe in the ICI group than in the non-ICI group and showed statistically significant difference at 12 months (16.8% vs. 6.8%; P = 0.013 before PSM; 20.0% vs. 8.0%; P = 0.032 after PSM) (Additional file 1: Table S14, Fig. S7). In the multiple linear regression analysis, ICI therapy was a significant independent predictor of absolute volume progression, absolute score progression, relative volume progression, and relative score progression of CAC at 12 months across all models (absolute volume progression before PSM was marginally significant in model 2), both before and after PSM (Additional file 1: Table S15, Table 4). We performed Spearman correlation analysis on ICI cycles and CAC progression. However, an association between these two variables was not established (Additional file 1: Fig. S8).