Long-term patterns in CD4 response are determined by an interaction between baseline CD4 cell count, viral load, and time: The Asia Pacific HIV Observational Database (APHOD)

Abstract

Background: Random effects models were used to explore how the shape of CD4 cell count responses after commencing combination antiretroviral therapy (cART) develop over time and, in particular, the role of baseline and follow-up covariates.

Methods: Patients in Asia Pacific HIV Observational Database who first commenced cART after January 1, 1997, and who had a baseline CD4 cell count and viral load measure and at least 1 follow-up measure between 6 and 24 months, were included. CD4 cell counts were determined at every 6-month period after the commencement of cART for up to 6 years.

Results: A total of 1638 patients fulfilled the inclusion criteria with a median follow-up time of 58 months. Lower post-cART mean CD4 cell counts were found to be associated with increasing age (P < 0.001), pre-cART hepatitis C coinfection (P = 0.038), prior AIDS (P = 0.019), baseline viral load < or equal to 100,000 copies per milliliter (P < 0.001), and the Asia Pacific region compared with Australia (P = 0.005). A highly significant 3-way interaction between the effects of time, baseline CD4 cell count, and post-cART viral burden (P < 0.0001) was demonstrated. Higher long-term mean CD4 cell counts were associated with lower baseline CD4 cell count and consistently undetectable viral loads. Among patients with consistently detectable viral load, CD4 cell counts seemed to converge for all baseline CD4 levels.

Conclusions: Our analysis suggest that the long-term shape of post-cART CD4 cell count changes depends only on a 3-way interaction between baseline CD4 cell count, viral load response, and time.

Figure 1

Observed mean CD4 cell counts (cell/μL) over time for TAHOD and AHOD.

Figure 2

Predicted mean CD4 cell counts (cell/μL) according to post-cART viral burden levels (a) 0 (b) 0.5 (c) 1 and baseline CD4 cell count levels 200, 350, 500 cells/μl for subjects with the following set predictor characteristics; 45 years old, AHOD cohort, no pre-cART hepatitis C coinfection, pre-cART viral load less than 100,000 copies/ml, no pre-cART AIDS related illness and with homosexual contact as the reported mode of HIV exposure.