Efficient use of longitudinal CD4 counts and viral load measures in survival analysis

CD4 counts and viral loads are dynamic quantities that change with time in HIV-infected persons. Commonly used single summary measures, such as viral load set point or early CD4 count, do not explicitly account for changes in viral load or CD4 counts or other features of the overall time course of these measures. However, the efficient use of all repeated measurements within each subject is often a challenge made more difficult by sparse and irregular sampling over time. Here, we illustrate how functional principal component (FPC) analysis provides an effective statistical approach for exploiting the patterns in CD4 count and viral load data over time. We demonstrate the method by using data from Kenyan women who acquired HIV-1 during follow-up in a cohort that practices high-risk activities and were subsequently followed up prospectively from early infection. The FPC scores for each woman obtained using this method served as informative summary statistics for the CD4 count and viral load trajectories. Similar to baseline CD4 count or viral set point, the first FPC score can be interpreted as a single-value summary measure of an individual's overall CD4 count or viral load. However, unlike most single-value summaries of CD4 count or viral load trajectories, the first FPC score summarizes the dynamics of these quantities and is seen to reveal specific features of the trajectories associated with mortality in this cohort. Moreover, the FPC scores are shown to be a more powerful prognostic factor than other common summaries when used in survival analysis.