Dr. Kravcik raises several issues related to the design of clinical trials evaluating promising anti-CMV therapies. Although untreated CMV retinitis spreads throughout the retina, it does so relatively slowly. Patients with small peripheral CMV retinitis lesions are not at risk for immediate loss of visual acuity and do not require immediate institution of therapy. In clinical practice, initiation of therapy may be deferred for a short period for personal reasons. "Small peripheral" retinitis studies, such as the two clinical trials of cidofovir reported in Annals, enroll only patients who are at low risk for visual loss, and they use a threshold for "progression" that is sufficiently small (750 microm) that patients are not at an increased risk for loss of visual acuity. Dr. Kravcik's assertion that deferral of therapy for patients with peripheral retinitis was "irrelevant" and "of questionable prudence" is incorrect. The HPMPC Peripheral CMV Retinitis Trial provides long-term follow-up data on visual outcomes, and no differences in visual acuity outcomes were seen between persons assigned to deferred therapy and those assigned to immediate therapy. Furthermore, Dr. Kravcik's assertion that we know the natural history of CMV retinitis may be incorrect in the era of highly active antiretroviral therapy; anecdotal reports suggest that progression may be slower in some patients.

Dr. Kravcik's opinion that these studies were "unethical" was not shared by the 23 institutional review boards at the centers participating in these two trials and the two data and safety monitoring committees, all of which approved these studies. Dr. Kravcik's comparison with withholding therapy for P. carinii pneumonia or cryptococcal meningitis is spurious. The pace of these disorders is different from that of CMV retinitis, and delaying the initiation of therapy for them has adverse consequences that do not occur with peripheral CMV retinitis.

Dr. Kravcik seems to suggest that the initial study of any promising anti-CMV agent should be a comparison with established therapy. This suggestion presumes that the investigator knows that the drug is safe and effective before such a trial begins. We believe that it is unethical to treat patients with posteriorly located, newly diagnosed CMV retinitis (that is, zone 1 disease), who are at risk for immediate visual loss, with a drug that is not known to be effective; doing so would place patients at risk for severe visual loss if the drug is ineffective. Trials designed to show the equivalence of two anti-CMV agents require a sample size of approximately 240 patients for adequate power. Because patients with "small peripheral" retinitis lesions account for only one quarter of patients with CMV retinitis [1], doing an equivalence study in this population would take substantially longer and hence delay general use if the drug proves effective. Conversely, trials involving deferred therapy for peripheral retinitis require fewer patients. If the drug is effective, approval is hastened; if the drug is ineffective or toxic, fewer patients are placed at risk.

If a promising new agent is less effective than standard therapy and the trial showing this is the only one done, it is unclear whether the drug is 1) less effective but still effective and potentially of value or 2) ineffective and of no value. Given the occurrence of resistance in patients requiring long-term therapy for CMV retinitis [2], new agents, which may not be equivalent to existing agents but have different mechanisms of action, may be of great value in the long-term management of this chronic disease. Hence, U.S. Food and Drug Administration disapproval of a drug because it is less effective than another drug may deprive patients with CMV disease of important options for long-term management. When the entire picture is evaluated, trials comparing a brief period of deferral with therapy for a new agent of unproven value are not only ethical but superior.

Finally, we agree with Dr. Kravcik that a study comparing a new treatment with "standard" therapy is an important step in the development of any new anti-CMV agent. Comparative studies permit treating physicians to better understand the relative merits and the role of a new agent, and we believe that all new agents with demonstrated efficacy should be tested against standard therapy.