To evaluate the relative effectiveness and safety of foscarnet versus ganciclovir for the treatment of cytomegalovirus (CMV) retinitis in people with AIDS; to evaluate the relative effect on survival of the use of these two anti-CMV agents in the treatment of CMV retinitis; to compare the relative benefits of immediate treatment with foscarnet or ganciclovir versus deferral of treatment for CMV retinitis limited to less than 25 percent of zones 2 and 3.

CMV retinitis is a common opportunistic infection in patients with AIDS. Ganciclovir is currently the only drug approved for treatment of CMV retinitis in immunocompromised patients. Ganciclovir suppresses CMV infections, and relapse occurs in virtually all AIDS patients when ganciclovir is discontinued. Because of their similar hematologic (blood) toxicities, the simultaneous use of ganciclovir and zidovudine (AZT) is not recommended. More recently the drug foscarnet has become available for investigational use. Studies so far indicate that remission of CMV retinitis occurs in 36 to 77 percent of patients, and that relapse occurs in virtually all patients when the drug is discontinued. The relative effectiveness of foscarnet compared with ganciclovir for the immediate control of CMV infections is unknown. Further, the long-term effects of foscarnet or ganciclovir on CMV retinitis, survival, and morbidity are unknown. There is also no definitive information on the relative effectiveness and safety of deferred versus immediate treatment for CMV retinitis confined to zones 2 and 3.

CMV retinitis is a common opportunistic infection in patients with AIDS. Ganciclovir is currently the only drug approved for treatment of CMV retinitis in immunocompromised patients. Ganciclovir suppresses CMV infections, and relapse occurs in virtually all AIDS patients when ganciclovir is discontinued. Because of their similar hematologic (blood) toxicities, the simultaneous use of ganciclovir and zidovudine (AZT) is not recommended. More recently the drug foscarnet has become available for investigational use. Studies so far indicate that remission of CMV retinitis occurs in 36 to 77 percent of patients, and that relapse occurs in virtually all patients when the drug is discontinued. The relative effectiveness of foscarnet compared with ganciclovir for the immediate control of CMV infections is unknown. Further, the long-term effects of foscarnet or ganciclovir on CMV retinitis, survival, and morbidity are unknown. There is also no definitive information on the relative effectiveness and safety of deferred versus immediate treatment for CMV retinitis confined to zones 2 and 3.

Patients are assigned to one of two groups: (1) Patients with any retinitis in zone 1 or patients with retinitis involving 25 percent or more of zones 2 and 3; and (2) Patients in whom retinitis is confined to less than 25 percent of zones 2 and 3 of the retina. Half the patients in group 1 get immediate treatment with ganciclovir; the other half receive immediate treatment with foscarnet. Patients in group 2 are treated with foscarnet or ganciclovir either immediately or treatment is deferred. If patients in group 2 have strong preferences regarding when therapy is instituted, they may elect immediate treatment or deferral of treatment.

Inclusion Criteria

Concurrent Medication:

Allowed:

• Topical anti-Herpesvirus agents.
• Zidovudine (AZT) for patients in deferral or foscarnet treatment groups:
• 100 mg every 4 hours. For patients on ganciclovir:
• 100 mg every 8 hours.
• Dideoxyinosine (ddI) and other antiretroviral available via expanded access programs, investigational triazoles, granulocyte-macrophage colony-stimulating factor, and erythropoietin to treat marrow toxicity. The use of other investigational drugs will be considered on a drug by drug basis.
• It is not recommended that patients receiving ganciclovir take AZT simultaneously. If AZT is prescribed for patients taking ganciclovir, it should be prescribed at reduced doses and discontinued if hematologic toxicity develops.

Patients must have:

• Diagnosis of AIDS by CDC criteria or a documented HIV infection.
• Cytomegalovirus (CMV) retinitis that does not require surgical intervention diagnosed in one or both eyes by a SOCA-certified ophthalmologist.
• The means available for compliance with follow-up visits (including a caregiver if necessary).
• Must consent to study or consent of parent or guardian if less than 18 years of age.
• Willingness to take reduced dose of zidovudine (AZT) if dictated by treatment assignment.
• Willingness to discontinue other systemic treatments for Herpesvirus infections while receiving foscarnet or ganciclovir.

Prior Medication:

Allowed:

• Zidovudine (AZT).

Exclusion Criteria

Co-existing Condition:

Patients with the following conditions or symptoms are excluded:

• Sufficient media opacities to preclude fundus photographs in both eyes.

Concurrent Medication:

Excluded:

• Other systemic treatments for Herpesvirus infections.
• Other anti-cytomegalovirus therapy.
• Excluded with foscarnet:
• Parenteral pentamidine, amphotericin B, or aminoglycosides.
• Use of marrow toxic agents with ganciclovir and nephrotoxic agents with foscarnet is discouraged, and alternative treatment should be used whenever possible.

Patients with the following are excluded:

• Sufficient media opacities to preclude fundus photographs in both eyes.
• Known or suspected allergy to one of the study medications.

Prior Medication:

Excluded:

• Foscarnet or ganciclovir used previously to treat cytomegalovirus (CMV) retinitis.
• Excluded within 14 days of study entry:
• CMV hyperimmunoglobulin or other anti-CMV agents.
• Excluded within the past 28 days:
• Anti-CMV therapy.

Active intravenous drug or alcohol abuse, sufficient in the investigator's opinion to prevent adequate compliance with study therapy and follow-up.