Natural History and Outcome of New AIDS-Related Cytomegalovirus Retinitis Diagnosed in the Era of Highly Active Antiretroviral Therapy

The incidence, natural history, and outcome of therapy for AIDS-related cytomegalovirus (CMV) retinitis were well-defined in the era before the widespread availability of highly active antiretroviral therapy (HAART), that is, up to 1996 [1, 2]. By 1997 the incidence of new CMV disease was reported to have become ∼80% lower than that in the pre-HAART era [3–5]. However, little is known from cohort or population-based studies about the natural history and outcome of CMV retinitis that is newly diagnosed and treated in patients who have already received or will promptly initiate HAART. Therefore, we reviewed cases of AIDS-related CMV retinitis that occurred within the San Francisco Public Health Department Community Health Network, which has provided primary medical care for a relatively constant number of individual patients with HIV disease (3400–600) from 1994 through 1998. During this period, all patients in the Community Health Network system who had newly diagnosed CMV retinitis were referred to the University of California at San Francisco AIDS Program CMV Patient Management Service at San Francisco General Hospital, to coordinate ophthalmologic follow-up and treatment. The clinical activity and location of CMV retinitis were determined by an experienced ophthalmologist in each case. HAART was defined as a combination of at least 3 concomitant antiretroviral agents, including 2 nucleoside reverse transcriptase inhibitor (NRTI) agents and at least 1 HIV protease inhibitor (PrI) or nonnucleoside transcriptase inhibitor agent. The number of new cases of CMV retinitis decreased from 46 in 1994 and 48 in 1995 to only 4 in 1997 (following widespread introduction of HIV PrI drug therapy in the Community Health Network system in 1996) but increased to 10 in 1998. All cases of CMV retinitis that were diagnosed before 1998 occurred in patients who never received prior HAART. In contrast, 5 of the 10 new cases diagnosed in 1998 occurred in patients who had received 7–16 months of prior HAART. The ophthalmologic characteristics, at time of diagnosis, of the 14 new cases of CMV retinitis diagnosed in 1997–1998 are summarized in table 1. Of note, 5 (36%) of these 14 patients had zone 1 disease with optic nerve involvement. Clinical characteristics and treatment of 14 patients with AIDS diagnosed with CMV retinitis in 1997–1998 within the San Franscisco Public Health Department Community Health Network. These 14 patients could be categorized into 3 groups on the basis of HAART exposure before or after the diagnosis of retinitis: group A, 5 patients who had never received HAART either before or after the retinitis diagnosis; group B, 4 patients who had not received HAART before the retinitis diagnosis but did initiate HAART a median of 2.5 months (range, 2–6 months) after the retinitis diagnosis; and group C, 5 patients who had received HAART before the retinitis diagnosis. Outcomes differed markedly between these 3 groups (see table 1 for details). Of the 5 group A patients, three either died or discontinued medical monitoring because of a moribund status within 2 months of the retinitis diagnosis, and 1 had active retinitis despite 6 weeks of parenteral anti-CMV therapy. All 4 group B patients had no subsequent retinitis reactivation over a median of 15.5 months of ophthalmologic monitoring. Of note, 3 of these patients were completely naive to anti-retroviral drugs before initiating HAART, and the fourth patient (B1) received a HAART regimen for which his previous anti-retroviral therapy probably would not have induced cross-resistant HIV. Of the 5 group C patients, 4 (C1, C2, C4, and C5) had been treated extensively with suboptimal antiretroviral regimens (e.g., 2 NRTI agents or a PrI agent plus 1 NRTI agent) for 6–46 months (median, 20 months) before initiating a HAART regimen. Once HAART was initiated, none of these 4 patients had virological control of HIV infection (i.e., plasma HIV RNA level below the level of quantitation) or a CD4+ lymphocyte response >50/µL. Two of these 4 HAART nonresponders died within 1 month of the diagnosis of CMV retinitis; the other 2 had active retinitis when last examined, despite 2–4 months of systemic anti-CMV therapy. In contrast, 1 group C patient (C3) was antiretroviral drug-naive before initiating HAART and did have virological and CD4 cell count responses to HAART initiation but then discontinued all antiretroviral therapy, after which CMV retinitis was diagnosed. At this point, the same HAART regimen was reinitiated along with iv ganciclovir. Four months later, at the time of the last follow-up, the viral load was again undetectable, the CD4 cell count had increased, and retinitis was inactive. Overall, 11 of the 14 patients with new cases of CMV retinitis that were diagnosed in 1997–1998 had follow-up ophthalmologic examinations. Retinitis healed in response to anti-CMV therapy and did not subsequently reactivate in all 5 patients who had a good immunologic and virological response to subsequent HAART. However, retinitis reactivated during follow-up despite anti-CMV therapy in 5 of the 6 patients who did not receive or had a poor response to HAART. To our knowledge, this is the first population-based report of the natural history and outcome of CMV retinitis in patients who were diagnosed with this complication of AIDS after HAART became widely available. Our most important observation is that the outcome for patients after the diagnosis of CMV retinitis was strongly associated with the subsequent initiation of a potent combination antiretroviral drug regimen to which the patient could have been expected to respond. In conclusion, CMV retinitis can occur in patients who have failed to respond to HAART. Clinical outcome of new cases of CMV retinitis may be predictable on the basis of the likely future response to HAART, which could be a relevant factor to consider in choosing anti-CMV treatment. We thank L. Joseph Wheat, W. Lawrence Drew, Gary Holland, Steven Deeks, and Douglas Jabs for their helpful advice.