Current issues of ACP Journal Club are published in Annals of Internal Medicine


Foscarnet reduced mortality compared with ganciclovir in patients with AIDS and cytomegalovirus retinitis

ACP J Club. 1992 May-June;116:74. doi:10.7326/ACPJC-1992-116-3-074

Source Citation

Studies of Ocular Complications of AIDS Research Group, in collaboration with the AIDS Clinical Trials Group. Mortality in patients with the acquired immunodeficiency syndrome treated with either foscarnet or ganciclovir for cytomegalovirus retinitis. N Engl J Med. 1992 Jan 23;326:213-20.



To compare the efficacy of foscarnet and ganciclovir for cytomegalovirus (CMV) retinitis in patients with AIDS.


Unmasked, randomized controlled trial.


11 clinics in the United States.


234 patients aged ≥ 13 years, with AIDS, untreated CMV retinitis, a neutrophil count of > 1.00 × 109 cells/L, and a serum creatinine concentration < 180 µmol/L, were randomized. Median follow-up was 7 months.


Intravenous foscarnet (n = 107), 180 mg/kg per day, was given for 14 days, followed by a maintenance dose, 90 mg/kg per day adjusted to 120 mg/kg per day for relapse of retinitis and interrupted if serum creatinine rose above > 260 µmol/L. Intravenous ganciclovir (n = 127), 10 mg/kg per day was given for 14 days, then maintained at 5 mg/kg per day. The dose was adjusted for renal function and was interrupted if the neutrophil count fell below 0.50 × 109 cells/L or the platelet count fell below 10 × 109 cells/L. If the retinitis was not controlled or if unacceptable toxic effects occurred, the drugs could be switched. Antiretroviral drugs were given as indicated by best medical judgment.

Main outcome measures

Death, progression of retinitis, visual function, and morbidity. Analysis was by intention-to-treat.

Main results

Analysis was by intention to treat. The trial was stopped early because of excess mortality in the ganciclovir group. 65 patients (51%) assigned to ganciclovir died compared with 36 (34%) assigned to foscarnet (P = 0.007). {This absolute risk increase of 17% means that 1 additional death occured for every 6 patients who received ganciclovir (compared with foscarnet), 95% CI 3 to 21; the relative risk increase was 52%, CI 12% to 110%}*. Adjustment for use of antiretroviral drugs (66% vs 84% for the ganciclovir and foscarnet groups, respectively) did not affect the drug effect on mortality. More patients assigned to ganciclovir had a decrease in neutrophil count (41 vs 17 patients, relative risk [RR] 1.88, P < 0.02). More patients assigned to foscarnet had an increase in serum creatinine (9 vs 4 patients {RR 2.9}*, P < 0.03), and more switched treatments (39 vs 14 patients {RR 2.4}*, P = 0.001). The groups did not differ in progression of retinitis or visual deterioration.


Foscarnet reduced mortality but there was increased drug toxicity in comparison with ganciclovir in patients with AIDS and cytomegalovirus retinitis.

Sources of funding: National Institutes of Health; participating academic institutions; Astra Pharmaceutical Products, Inc.

Address for article reprint: Dr. D.A. Jabs, SOCA Chairman's Office, 550 North Broadway, Suite 70, Baltimore, MD 21205, USA.

*Numbers calculated from data in article.


The difference in mortality for patients with CMV retinitis receiving different therapies was not appreciated until this study showed a 4-month longer survival for those patients initially randomized to foscarnet.The improved survival for patients receiving foscarnet could not be fully accounted for by the concomitant use of antiretroviral agents. CD4 counts for a subgroup of patients (41%) increased for patients assigned to foscarnet but decreased in patients receiving ganciclovir. Attributing improved survival to the known antiretroviral activity of foscarnet and the synergism of the drug with zidovudine in vitro is appealing but still requires confirmation.

Should clinicians use this trial to advocate foscarnet as preferred therapy for all HIV-infected patients with CMV retinitis? Efficacy of these agents for concomitant CMV infection at other sites (gastrointestinal tract, encephalitis, or adrenalitis) may guide selection. Therapy with foscarnet is more time consuming and is more toxic than ganciclovir and foscarnet is only available intravenously; direct measurement of the effects of these drugs on quality of life would also help in selecting the appropriate agent. Physicians should provide patients with the currently available information when instituting treatment for CMV retinitis: longer survival with foscarnet but a higher incidence of adverse effects in the face of equal clinical efficacy. The treatment decision can be shared between physician and patient.

Additional strategies, including use of potent combination antiretroviral agents and cytokines, should be considered in other trials of ganciclovir and foscarnet. Data suggesting synergy between ganciclovir and foscarnet encourage further studies of combination therapy to assess efficacy, development of resistance, and toxicity (1). The morbidity of this condition deserves our best efforts in designing new clinical trials.

Anita Rachlis, MD
University of TorontoToronto, Ontario, Canada

Anita Rachlis, MD
University of Toronto
Toronto, Ontario, Canada


1. Manischewitz JF, Quinnan GV, Lane HC, Wittek AE. Antimicrob Agents Chemother. 1990;34:373-5.