Predicting AIDS onset for individual patients.

Today’s physician may be confronted with a modern dilemma-the need to apply accumulating research information regarding markers predictiveof acquired iminunodeficiency syndrome (AIDS) to an individual patient who is infected with human immunodeficiency virus (HIV). Every physician would liketo have precise estimates ofrisk to guide the choiceoftherapy and to allow him or her to concentrateon the treatment ofthe patient and on the psychosocial complications of HW infection. Many studies have shown that AIDS risk is inversely related to CD4+ lymphocyte count (1-8). This fact partly justifies current recommendations to use prophylaxis with aerosolized pentamidine for Pneumocystis carinii pneumonia (9) in persons with CD4+ counts <200 ce!ls/L and to begin treatment with zidovudine (AZT) to reduce AIDS incidence in persons with counts <500 ce!ls/j.L (10-13). Following up early interest in serological markers (14-16), recent work has shown that AIDS risk for many (5, 6, 17-20) but not all (8) cohorts can be predicted more precisely by measuring not only CD4+ lymphocyte count but also the concentration of /32-microglobulin, neopterin, or interferon in serum. Improved risk models not only are helpful for managing individual patients, but also can be used to form homogeneous strata to increase the power of clinical trials. The work of Reibnegger et al. (21), reported in the March issue of Clinical Chemistry, not only showed that neopterin adds useful prognostic information to CD4+ counts but also presented a model that predicts absolute risk. Reibnegger et al. (21) reported the development of a statistical model for predicting individualized probabilities of developing AIDS over a 4.5-year follow-up period,based on initialCD4+ lymphocyte counts and neopterin concentrations in a cohort of 68 HIV-infected homosexual men. Their statistical model implies that the probability of disease in time interval t (t = 1,2,. 9, corresponding to successive half-year intervals), conditional on being at risk at the beginning of interval t, is