GLP‐1RAs and cardiovascular risk in HS: concerns about follow‐up time and effect estimates

Dear Editor,

We read a paper by Chou et al. ¹ with great interest. The reported association between GLP‐1 receptor agonists (GLP‐1RAs) use and lower cardiovascular risk in hidradenitis suppurativa (HS) is clinically meaningful and directionally consistent with broader evidence from cardiometabolic literature and from other chronic inflammatory dermatoses. ² GLP‐1RAs has also been linked to improved cardiometabolic profiles and a reduction in major adverse cardiovascular events in other inflammatory dermatoses. ³ , ⁴ At the same time, we were struck by the unexpectedly low absolute event proportions reported at long‐term horizons (20‐year observation), which appear substantially lower than would be anticipated for HS populations given their elevated cardiovascular risk burden. ⁵ To contextualize the unexpectedly low absolute proportions, it is important to consider established epidemiology in HS populations. Prior epidemiological studies report crude MACE incidence of 4.0–6.6 per 1000 person‐years in HS cohorts, ⁶ , ⁷ which implies ~10‐year cumulative MACE risk of 4%–6.6% in low‐risk HS and 8%–20% in those with metabolic comorbidities. The proportions reported at the study's ‘20‐year’ horizon appear substantially lower than expected. This discrepancy underscores the absolute necessity of transparent follow‐up reporting and time‐to‐event methodology.

As active users of the TriNetX platform, we attempted to reproduce the cohort selection and outcome analyses; however, replication is inherently challenging without a detailed, step‐by‐step methodological description (including exclusion logic, index definitions, washouts, exposure windows and censoring rules), which we recommend be provided in the Supporting Information.

The first factor to consider is the timing of GLP‐1RA availability within the clinical landscape. The most currently utilized agents were introduced into clinical practice within the past decade, with semaglutide receiving regulatory approval between 2017 and 2018, followed by dual incretin therapies. ⁸ This limited timeframe suggests a paucity of long‐term observational data for these agents in exposed populations. In our parallel healthcare system cohort utilizing TriNetX, only approximately 4% of patients exposed to GLP‐1RAs received exenatide, while about 18.5% received liraglutide. These figures indicate that the majority of exposures involve agents that have been available more recently, inherently restricting the duration of the at‐risk period in the exposed cohort relative to the unexposed cohort.

Moreover, in analogous TriNetX queries aligned with the study cohorts, extending the analysis window from 5 to 20 years does not significantly increase follow‐up duration. HS patients exposed to GLP‐1RA had a median follow‐up of around 573 days across all time horizons. The unexposed group, on the other hand, had a longer follow‐up, with a median duration of around 1091 days. This indicates that actual observation periods are much shorter than the labelled 5, 10 or 20 years, potentially underestimating event rates in the exposed cohort. The significant, almost 90% difference in follow‐up duration between the treated and untreated groups could potentially increase event counts and, importantly, bias effect estimates. Therefore, relying exclusively on odds ratios derived from raw event counts—without adjusting for person‐time or employing rate‐based metrics—poses a risk of misrepresenting the actual effect sizes. For accurate hazard assessment, time‐to‐event models or rate‐based methods should replace simple odds ratios based on distant horizon labels.

These findings require clarification of the methodology to confirm that the absolute cardiovascular risks in HS are not underestimated and that readers accurately interpret the reported proportions, considering the actual follow‐up duration and the timing of GLP‐1RAs' availability.

FUNDING INFORMATION

The authors have nothing to report.

CONFLICT OF INTEREST STATEMENT

No conflict of interest in the current manuscript.

ETHICAL APPROVAL

Not applicable, as no ethical approval was necessary for this methodological commentary.

ETHICS STATEMENT

This work adheres to ICMJE and COPE ethical standards. It did not involve any individual patient data, human subjects or animal research.

DATA AVAILABILITY STATEMENT

Data sharing is not applicable to this article as no new data were created or analysed in this study.