Purpose: To characterize the natural course and clinical predictors of cytomegalovirus (CMV) retinitis in human immunodeficiency virus (HIV)-infected patients after initiation of highly active antiretroviral therapy (HAART).
Methods: Retrospective analysis of 53 HIV-positive patients (73 eyes with CMV retinitis) treated with and without HAART. All participants continued to take anti-CMV therapy. Survival analysis was used to characterize the natural course of CMV retinitis. Proportional hazards analysis was performed to assess the correlation of the nine potential clinical predictors (baseline CD4 count, post-HAART CD4 count, post-HAART rise in CD4 count, baseline weight, post-HAART rise in weight, post-HAART percentage rise in weight, log of baseline HIV viral load, log of minimum post-HAART HIV viral load, and post-HAART log unit reduction in HIV viral load) with the duration of CMV retinitis remission.
Results: Patients receiving HAART had a median CMV retinitis remission duration of 574 days (95% confidence interval, 336-NA) whereas those not receiving HAART had a median remission duration of 80.5 days (95% confidence interval, 28-NA;P < 0.001). Within the HAART-treated population, the minimal viral load reached after HAART was the only clinical predictor to demonstrate significance (P = 0.0075). Several other clinical predictors demonstrated borderline significance; however, this was most likely due to the high correlation of these variables with the minimum post-HAART viral load. A potential secondary clinical predictor identified was the post-HAART rise in CD4 count (P = 0.085).
Conclusion: With the introduction of HAART, HIV-infected patients have much longer remission durations from recurrent CMV retinitis. The minimum HIV viral load level reached after the initiation of HAART treatment appears to be more important than other clinical variables in the prediction of favorable CMV retinitis remission status. Furthermore, a rise in CD4 T-lymphocyte count by itself appears to be a less significant clinical predictor but may be useful in combination with the HIV viral load data. Selective discontinuation of anti-CMV therapy may be considered in patients with a favorable set of clinical predictors.