Abstract Body

With the potential adverse cardiovascular effects of newer antiretroviral therapy (ART), such as integrase inhibitors and weight gain and Tenofovir Alafenamide and elevated lipids, it is critical to continue monitoring trends in myocardial infarction (MI) rates by HIV status.

Subjects included people with HIV (PWH) from two health system cohorts: Massachusetts General Hospital (Partners) and Kaiser Permanente Northern California (KPNC), identified from 2005-2017 with follow-up through 2020. Subjects also included a 1:4 propensity-matched comparison group of people without HIV (PWoH) in Partners and a 1:2 matched group in KPNC. Propensity scores were informed by baseline demographics (age, race, sex, year) and baseline Framingham risk score components (total cholesterol, HDL, diabetes, systolic BP, hypertension treatment and smoking status). We assessed effect of HIV status on MI risk in two calendar eras defined by baseline year: 2005-2009 and 2010-2017. To ensure similar follow-up of events by era, we censored follow-up at the earliest of: 5 years after baseline, death, loss-to-follow-up, or administrative end of follow-up. We used Cox Proportional Hazards models to obtain adjusted HRs of effect of HIV status on MI (PWoH as reference) overall, and by cohort. We obtained two adjusted models: (1) adjusted for age and sex only; and (2) stepwise model adjusted for demographics and Framingham risk score components.

We included 10,312 PWH and 23,018 PWoH, of whom 238 had an MI. Mean age of the combined cohort was 43 years, 87% were men, mean cholesterol of 180 mg/dL, mean HDL of 44 mg/dL, 6% diabetes, mean systolic BP of 123 mmHg, 25% on treatment for hypertension, and 26% were smokers. The cohorts were similar except for more men (90% vs. 76%), fewer on treatment for hypertension (24% vs. 30%) and fewer smokers (23% vs. 40%) in KPNC vs. Partners. In the stepwise models for the overall cohort (Table), the HR for MI was 0.99 (0.67, 1.45) for years 2005-2009 and 1.85 (1.26, 2.72) for years 2010-2017, with a corresponding P-interaction of 0.039. The magnitude of HRs by calendar era was consistent across models stratified by cohort.

Data from two distinct US cohorts suggest increased MI risk among PWH in recent years. In light of known metabolic effects of newer ART regimens, continued surveillance for MIs is warranted.