Background: The START study demonstrated a 57% risk reduction in serious AIDS, serious non-AIDS events or death (primary endpoint) with immediate versus deferred ART. Most events occurred at CD4+ counts >500 cells/mm3. We investigated the extent to which treatment differences in CD4+, CD8+ and plasma HIV RNA (VL) explained the clinical benefit of immediate ART.
Methods: ART-naïve HIV+ participants with CD4+ >500 cells mm3 were randomised to immediate (n=2326) or deferred ART (n=2359) with assessments every 4 months for CD4+, CD8+ counts/percent, and VL. We estimated average biomarker levels using longitudinal mixed models, and hazard ratios (HR, Immediate/Deferred ART) using Cox models with and without adjustment for most recent (latest) biomarker levels.
Results: Over mean follow-up of 3 years, 138 participants experienced primary events. Average levels of CD4+, CD8+ and VL were 194 cells/mm3 higher, 189 cells/mm3 lower, and 1.2 log10 copies/mL lower, respectively, with immediate compared to deferred ART. Latest CD4+ count was modestly associated with the primary endpoint in the deferred arm ((HR=0.87 per 100 cells higher; 95% CI: (0.79, 0.97), p=0.01) but not in the immediate arm (HR 0.98 (0.88, 1.11) p=0.78). In contrast, higher VL, higher CD8+, CD8+% and lower CD4+/CD8+ ratio were strong predictors of the primary endpoint regardless of treatment group. The association with CD8+ count and CD4+/CD8+ ratio remained strong after adjusting for VL (data not shown). As shown in the table, treatment differences in CD4+%, CD8+%, or CD4+/CD8+ ratio during follow-up predicted less than half the observed risk reduction conferred by immediate ART.

Proportion of treatment benefit (immediate vs deferred ART) that could be explained by treatment differences in laboratory markers
[Proportion of treatment benefit (immediate vs deferred ART) that could be explained by treatment differences in laboratory markers]

Conclusions: In this population, latest CD8+ count and percent and CD4+/CD8+ ratio are better predictors of clinical outcome than latest CD4+ count. The reduction in incidence of the primary endpoint conferred by Immediate ART could be explained mostly by the reduction in VL and, to a lesser extent, by increase in log CD4/CD8 ratio.