CROI 2016 Abstract eBook

Abstract Listing

Poster Abstracts

473

Predictors of CD4 Count Recovery in the CIPRA HT-001 Trial of Early ART Sean E. Collins 1 ; Michaela Calnan 2 ; Marc Antoine Jean Juste 3 ; Patrice Severe 3 ; Rode Secours 3 ; Daphne Bernard 3 ; Ashita Batavia 4 ; DanielW. Fitzgerald 4 ; Serena P. Koenig 5 ; JeanW. Pape 3 1 Stanford Univ, Stanford, CA, USA; 2 Analysis Group, Boston, MA, USA; 3 GHESKIO, Port-au-Prince, Haiti; 4 Weill Cornell Med Coll, New York, NY, USA; 5 Brigham and Women’s Hosp, Harvard Med Sch, Boston, MA, USA Background: Early antiretroviral therapy (ART) in the CIPRA HT-001 study improved survival and decreased tuberculosis (TB) risk. After an interim analysis, all participants were offered ART regardless of CD4 count, thus the impact of ART timing on CD4 recovery is underestimated by intention to treat analyses. Methods: From August 2005 to July 2008, 816 treatment-naïve HIV-infected Haitians with a CD4 count between 200 - 350 cells/mm 3 were randomized to immediate ART or to defer ART until CD4 <200 or AIDS illness occurred. In June 2009, all participants were offered ART regardless of CD4. Data were collected until August 2014. We examined median CD4 counts from the time of ART initiation and used generalized estimating equations (GEE) to assess independent predictors of CD4 recovery. An on-ART model incorporated viral suppression, TB co-infection and adherence. We estimated the time to a CD4 increase of 100, 200 and 300 and the time to an absolute CD4 >350 or >500 using Kaplan-Meier analysis stratified by gender and by CD4 at ART start. Results: 763 of 816 participants (94%) started ART and are included in the analysis, 408 in the immediate treatment arm and 355/408 (87%) in the deferred arm. In the deferred arm 185/355 (52%) started ART with a CD4 <200. Median CD4 counts over time were higher in those starting ART with CD4 ≥200 compared to <200, but with overlapping interquartile ranges. In GEE models, a higher CD4 at ART start, female gender and time on ART were independent predictors of CD4 recovery (p<.001). When we controlled for CD4 at ART start, the randomization armwas not an independent predictor of CD4 recovery or the rate of increase. In the on-ART model, a pre-ART viral load >100,000 copies/mL and viral suppression after 1 year of ART were associated with a greater CD4 recovery (p=.006 and p<.001), and active TB while on ART with a lower CD4 (p=.03). Controlling for adherence to clinic visits and drugs did not alter these associations. Participants with CD4 <200 had a faster rise in CD4 by 100 and 200 cells/mm 3 compared to those who started with CD4 ≥200 (log-rank p=0.02 and .01) but fewer reached an absolute CD4 count >500 (80% vs 90% p<.001). Men were slower than women to increase by 200 or 300 CD4 cells/ mm3 (p=.03 and .003) and fewer men reached an absolute CD4 count >500 (83% vs 91% p= .01). Conclusions: A higher CD4 count at ART start, female gender, viral suppression and time on ART are independent predictors of CD4 recovery. Active TB is associated with impaired CD4 recovery on ART.

Poster Abstracts

183

CROI 2016