Background:  No controlled trial assessed the efficacy of short-course HAART initiated at the time of primary HIV infection. Observational studies are prone to several biases, mainly indication bias. Using propensity analysis to control for bias in treatment assignation and prognostic imbalances, we compared the time to CD4 decline below 350/mm³ after interruption of HAART in patients who received short-course HAART vs after enrollment in patients who received no HAART at the time of primary HIV infection diagnosis.

Methods:  From 1996 to 2007, 439 primary HIV infection patients were enrolled within 2 months (median 39 days) of infection in the ongoing French ANRS PRIMO cohort. Of these, 73 patients received HAART for 6 to 24 months after enrollment (short-course group) while 149 patients received no HAART at enrollment and remained untreated for ≥3 months (median time off treatment 14 months; deferred group). Potential selection biases were adjusted for by developing a propensity score for short-course HAART through a multivariate logistic regression model, which included sex, age, symptomatic primary infection, CD4, and HIV RNA at enrollment and time between infection and enrollment. This score was then used to match patients from the 2 groups. The time to CD4 decline below 350/mm³ from baseline, i.e., from enrollment in deferred patients or from treatment interruption in short-course patients, was assessed through Kaplan-Meier curves.

Results:  From each group, 63 patients could be matched according to their propensity score. At enrollment, median CD4 count was 493/mm³ in short-course patients vs 498/mm³ in deferred patients; it reached 745/mm³ at HAART interruption in short-course patients. CD4 counts tended to decline below 350 /mm³ more rapidly in deferred patients than in short-course patients in the first 10 months of follow-up from baseline but less rapidly later on (interaction term with time, p = 0.04). This led to similar cumulated proportions of patients whose CD4 count remained >350/mm³ 36 months after baseline in the 2 groups (57% vs 58%). Sensitivity analysis considering the time to CD4 decline below 350/mm³ or to initiate/reinitiate treatment as the outcome led to similar results.

Conclusions:  Using a propensity analysis that compared patients who received short-course HAART at the time of primary HIV infection with their matched deferred counterparts, we found that short-course HAART initiated at the time of primary HIV infection provides no sustained benefit in terms of time to CD4 decline below 350/mm³.