Background
REPRIEVE demonstrated a 36% reduction in the hazard of major adverse cardiovascular (CV) events (MACE) among those randomized to pitavastatin vs placebo using a time-to-first-event (TTFE) analysis. TTFE results can be difficult to interpret with CV composite endpoints, as “less severe” events (MI and stroke) are likely to precede “more severe” events (death). Win ratio analysis is gaining traction in CV trials as an intuitive alternative that incorporates event severity, but has been criticized for dependence on follow-up time, as more severe events occur with longer follow-up. We performed a win ratio analysis of REPRIEVE and assessed trends over time.
Methods
All pairs consisting of one participant from each arm are constructed. MACE components are then compared in order of severity until it can be determined which participant did better. Here, a participant “wins” if their event/censor time is greater than their comparator’s event time. The ratio of pitavastatin to placebo wins estimates the win ratio (WR), with WR > 1 suggesting pitavastatin benefit. The WR was estimated for MACE (event tier 1: Death due to CV or undetermined causes > 2: stroke > 3: MI > 4: other CV event including transient ischemic attack, unstable angina and peripheral arterial ischemia > 5: CV procedure), and Hard MACE (1: CV death > 2: stroke > 3: MI). To investigate time-dependence, administrative censoring was performed at each of 8 follow-up years.
Results
At year 1, arms contributed equally to ordered wins on CV death and procedures (Figure). Pitavastatin had an apparent advantage on stroke and MI, while placebo had an apparent advantage on less severe CV events. As a result, the WR for Hard MACE (1.59) was larger than that of MACE (1.18). After year 1, pitavastatin steadily gained advantage over placebo on all events. Event tier contributions stabilized by approximately year 4, and the WR for MACE and Hard MACE were consistently estimated as 1.55 (95% CI: 1.20 to 1.99) and 1.65 (95% CI: 1.20 to 2.27) respectively, for years 6 through 8.
Conclusions
Time-updated win ratios suggest effects of pitavastatin are early and persistent for CV death, stroke, and MI. Less severe CV events and procedures contributed little to win effects. The reciprocal of the win ratio at 8 years (WR=1.55) is approximately the TTFE hazard ratio (HR=0.64). Our win ratio analysis agrees with REPRIEVE’s primary findings, and provides more insight into the protections of pitavastatin against severe CV events among people with HIV.
Table or Figure
