Objective: Governments, clinicians and drug-licensing bodies have adopted changes in CD4 cell counts and HIV-1 RNA levels as evidence of effectiveness for new therapeutic interventions. We aimed to determine the strength of the association between the magnitude of the effect of changes in CD4 cell count and HIV-1 RNA and progression to AIDS or death in the highly active antiretroviral therapy (HAART) era.
Methods: We identified all randomized clinical trials (RCTs) evaluating the effect of HAART on both clinical and surrogate endpoints (1994 to September 2006). We performed a meta-regression and weighted linear regression. We additionally estimated potential RCT sample sizes that would be required to assess the effectiveness of new interventions in terms of clinical endpoints.
Results: We included data from 178 RCTs. We were unable to demonstrate a strong relationship at any time-point. Specifically, this was the case when CD4 T-cell change and clinical outcomes were examined at week 24 [coefficient -0.01, 95% confidence interval (CI) -0.03 to 0.001, P=0.54], week 48 (coefficient -0.01, 95% CI -0.02 to 0.001, P=0.83) and week 96 (coefficient 0.00, 95% CI -0.03 to 0.04, P=0.76). This was also the case when viral load was examined as a surrogate marker. Given the small number of clinical events occurring in new interventional RCTs, any RCT aiming to evaluate clinical endpoints within these time-points would require an exceptionally large sample size.
Conclusions: Our findings indicate that, within short-term clinical trial settings, it is not possible to estimate the proportion of treatment effect associated with surrogate endpoints.