To study the prognostic role of cotton-wool spots on HIV infection, a prospective cohort study was set up between September 1990 and November 1992. The 135 HIV-infected patients included in the cohort were those hospitalized in an Infectious Diseases Department and submitted to a systematic retinography, whatever the stage of HIV infection. Clinical examination, CD4+ lymphocyte count, and retinography were systematically performed at inclusion. Patients with active Cytomegalovirus (CMV) or toxoplasmic retinitis were excluded from the study. Overall, 41 subjects were diagnosed with cotton-wool spots (spots+) and 94 without (spots-). At inclusion, the patients spots+ were at a more advanced stage of immunosuppression than the patients spots- : the proportion of AIDS patients was, respectively 78 % vs 60 % (P = 0,04) and their median CD4+ count was 28/mm(3) vs 70/mm(3) (P = 0,03), The mean follow-up of the cohort was 13 months. Adjustment was made according to delay between HIV infection diagnosis and inclusion, age, clinical stage, antiretroviral therapy and CD4+ at the time of retinography. Cotton-wool spots appeared to be an independent prognostic factor in the onset of CMV retinitis (adjusted relative risk (RR) = 5.1 with a 95 % Confidence Interval [CI] : 1.3-20.4) and of opportunistic cerebral pathology (RR = 2.6 - IC = 1.1-6.4). In conclusion, because of the increased risk of CMV retinitis in patients spots+, a regular ophthalmologic surveillance (fundoscopy, retinography) is mandatory in HIV-infected patients, whatever their immunosuppression level may be. The presence of cotton-wool spots justifies a more frequent retinal examination and should be taken into consideration for inception of primary prophylaxis of CMV retinitis.
