Association between immune checkpoint inhibitors and heart failure hospitalizations: a population-based self-controlled case series

Abstract

Abstract Background Immune checkpoint inhibitors (ICIs) have become a common treatment for many cancers. Though with proven anti-oncogenic effects, they are associated with cardiotoxicity. Specifically, some suggested that ICIs were associated with the diagnosis of heart failure (HF). However, the effects of ICIs on HF hospitalizations (HFH) have remained unexplored. Purpose To explore possible associations between of ICI use and HFH. Methods This self-controlled case series (SCCS) included patients with cancer who received any ICI within 1/1/2014-31/12/2020 in a prospectively recorded, population-based database from a developed Asian city. Those without HFH between one year before the first-ever ICI prescription (baseline period) to 31/12/2021 were excluded as SCCS only requires cases and evaluates effects in paired, pre-post comparisons. Exposure periods included contiguous ICI prescriptions and the following 180 days. Post-exposure periods were periods not described above, until death, end of follow-up (31/12/2021), or having reached three years after ICI initiation, whichever earlier; follow-up was restricted to three years to limit the potential for time-varying confounders. The study design is summarized by Figure 1A. Fixed-effects conditional Poisson regression was used to estimate incidence rate ratios (IRRs) of HFH during and after ICI exposure, compared to baseline. Multivariable adjustments were not applicable as the paired, pre-post comparisons in SCCS inherently adjust for time-invariant confounders. Results Of the 3684 identified ICI users, 19 had HF during the study period (median age 75 [interquartile range 68-84] years; 16 (84.2%) males; median follow-up 525 [274.5-967] days). HFH occurred to nine during baseline, eight during exposure, and four during post-exposure. Cohort-level HFH rates were low (4.1, 2.9, and 2.3 hospitalizations per 1000-person-years during baseline, exposure, and post-exposure, respectively). Compared to baseline, HFH rates were not different in days 1-180 (IRR 0.75 [95% confidence interval 0.27, 2.07], p = 0.574), days 181-366 (IRR 1.22 [0.31, 4.69], p = 0.777), or days ≥367 (IRR 0.37 [0.04, 3.66], p = 0.399) of exposure, nor post-exposure (IRR 1.28 [0.33, 4.98], p = 0.722). One patient had HF-related mortality; excluding her produced consistent results (IRR for exposure days 1-180: 0.75 [0.27, 2.07], p = 0.574; days 181-366: 1.30 [0.33, 5.09], p = 0.704; days ≥367: 0.43 [0.04, 4.45], p = 0.480; post-exposure: 0.91 [0.22, 3.82], p = 0.902). Meanwhile, six had follow-up <1 year; excluding them produced consistent results (IRR for exposure days 1-180: 1.22 [0.29, 5.10], p = 0.785; days 181-366: 1.66 [0.38, 7.18], p = 0.496; days ≥367: 1.23 [0.30, 5.01], p = 0.775; post-exposure: 1.60 [0.38, 6.84], p = 0.524). Conclusions ICIs are probably not associated with significantly different HFH rates in Asian patients with cancer. HFH may be rare in these ICI users.Figure 1