Suboptimal immune recovery during antiretroviral therapy with sustained HIV suppression in sub-Saharan Africa

OBJECTIVE: To assess incidence, determinants and clinical consequences of suboptimal immune recovery in HIV-1 infected adults in sub-Saharan Africa with sustained viral suppression on antiretroviral therapy (ART).DESIGN: Multicountry prospective cohort.METHODS: Suboptimal immune recovery was defined as proportions of participants who failed to attain clinically relevant CD4 cell count thresholds (>200, >350 and >500 cells/μl) despite sustained viral suppression on continuous first-line ART. Participants were censored at the earliest of death, loss to follow-up, last viral load less than 50 copies/ml, or database closure. Determinants of immune recovery were assessed using multivariable Cox regression. We estimated incidence rates of AIDS, pulmonary tuberculosis and all-cause mortality for CD4 strata.RESULTS: One thousand, five hundred and ninety-two participants were included; 60% were women, median age was 37 years (IQR 31-43) and median pre-ART CD4 cell count was 147 cells/μl (IQR 76-215). After 6 years of ART, suboptimal immune recovery at CD4 cell counts less than 200 cells/μl, less than 350 cells/μl, and less than 500 cells/μl occurred in 7, 27, and 57% of participants, respectively. Compared with participants with CD4 cell count greater than 500 cells/μl, on-ART incidence rates were 12.5, 4.1, 0.9 times higher for AIDS and 16.9, 3.5, and 2.3 times higher for pulmonary tuberculosis in participants with CD4 cell count less than 200, 200-349, and 350-499 cells/μl, respectively. All-cause mortality was highest in participants with CD4 cell count less than 200 cells/μl, and comparable across the higher CD4 strata. Older age, male sex, and lower pre-ART CD4 cell count were strongly associated with suboptimal immune recovery.CONCLUSION: These findings warrant close clinical and laboratory monitoring until adequate immune reconstitution is achieved and support early ART initiation before decline of CD4 cell count.