Characterization and determinants of long-term immune recovery under suppressive antiretroviral therapy.

OBJECTIVE:We developed a robust characterization of immune recovery trajectories in people living with HIV (PWH) on antiretroviral treatment (ART) and relate our findings to epidemiological risk factors and bacterial pneumonia. METHODS:Using data from the Swiss HIV Cohort Study and the Zurich Primary HIV Infection Cohort Study (n = 5907), we analyzed the long-term trajectories of CD4 cell and CD8 cell counts and their ratio in PWH on ART for at least eight years by fitting nonlinear mixed effects models. The determinants of long-term immune recovery were investigated using generalized additive models. In addition, prediction accuracy of the modeled trajectories and their impact on the fit of a model for bacterial pneumonia was assessed. RESULTS:Overall, our population showed good immune recovery (median plateau [IQR]-CD4: 718 [555, 900] cells/µl, CD8: 709 [547, 893] cells/µl, CD4/CD8: 1.01 [0.76, 1.37]). The following factors were predictive of recovery: Age, sex, nadir/zenith value, pre-ART HIV-1 viral load, Hepatitis C, ethnicity, acquisition risk and timing of ART-initiation. The fitted models proved to be an accurate and efficient way of predicting future CD4+ and CD8+ cell recovery dynamics: Compared to carrying forward the last observation, mean squared errors of the fitted values were lower by 1.3% to 18.3% across outcomes. When modeling future episodes of bacterial pneumonia, using model-derived predictors improved most model fits. CONCLUSION:We described and validated a method to characterize individual immune recovery trajectories of PWH on suppressive ART. These trajectories accurately predict long-term immune recovery and the occurrence of bacterial pneumonia.