CMV’s Real World: When Two Prophylaxis Studies Declare Opposite Findings, Clinicians And Their Patients Are Lost in the Quandary

The most important CMV data to come out of the recent ICAAC conference was the late-breaking presentation of CPCRA 023, a randomized, placebo-controlled trial of oral ganciclovir for prophylaxis of CMV in HIV-infected individuals. Anticipation of this presentation was great, since NIAID’s CPCRA had sent out the executive summary a week earlier, concluding that “the CPCRA CMV study does not support the use of oral ganciclovir as prophylaxis for symptomatic CMV retinal or gastrointestinal disease.” This seemed to fly in the face of a similar study, Syntex 1654, which was stopped early (after 20 months) by the DSMB because it showed a reduction of CMV retinitis by approximately 50%. (Incidentally, the authors of “TAG Does ICAAC” incorrectly reported that the CMV retinitis endpoints in CPCRA 023 were not confirmed by ophthalmologists; they were.)

CPCRA 023, a 26-month study, showed no statistically significant difference between oral ganciclovir (3 g/day) and placebo. But while the trial designs for CPCRA 023 and Syntex 1654 were superficially similar, there were significant differences. The only true similarity was the fact that they used a 2:1 allocation of study drug vs. placebo; the study designs in all other terms were different. Syntex 1654 required patients to be examined by an ophthalmologist every two months, while CPCRA 023 required patients to be examined by an ophthalmologist only after visual symptoms arose. Thus, 1654 was more likely to detect CMV retinitis before it became clinically apparent, and 023 only afterwards. 023 was attempting to replicate a more real world scenario, equating “usual care” with ophthalmologic exams by experienced ophthalmologists only after visual symptoms occur. Could such a difference in study design account for the discordant results seen in the studies? Many researchers say “yes.” Thus discussions regarding 023 and 1654 do not center on whether or not oral ganciclovir is effective for prophylaxis, but on issues of trial design, CMV patient management and treatment of asymptomatic versus symptomatic CMV disease.

Many are concerned with 023’s decision not to require baseline screening for retinitis if patients did not complain of symptoms. If 023 patients did complain of symptoms, their primary care providers would refer them to an ophthalmologist for a dilated funduscopic exam in order to confirm a diagnosis of CMV retinitis. Some critics of 023 are troubled by the fact that all patients were not screened at baseline asking, “How do we know that some patients did not already have asymptomatic CMV retinitis when they entered the study?”

Some clinicians are concerned about the 023 study design and its belief that 023 study sites were providing “usual care” in the “real world.” In fact, one clinician interviewed was particularly troubled by the study design, saying that “[the 023 study designers] are pretending when they say it is ‘real world’ because how they manage their patients is not how I manage patients in the ‘real world.’ The appropriate thing in the real world is to have patients with very low CD4 counts see an ophthalmologist for frequent screenings.”

Thus the 023 study–and its discrepancies with 1654–leave us with a number of questions underlying how little is still known about the natural history of CMV end-organ disease in people with AIDS:

1. How important is treating patients with asymptomatic CMV disease?
2. Should all patients with low CD4 counts see an ophthalmologist for routine screening for detection of asymptomatic disease?
3. Can all patients afford regular visits to an ophthalmologist?
4. Can all patients with low CD4 counts afford oral ganciclovir for CMV prophylaxis?

One of the principal reasons some clinicians are apprehensive about using oral ganciclovir for routine prophylaxis of CMV disease in patients with CD4 counts < 50 cells/mm3 is the possibility of the emergence of ganciclovir resistant virus. In a separate presentation by UCSF’s Larry Drew (co-chair of the 1654 study), only one of 18 CMV isolates tested (mean ganciclovir exposure of 10.1 months; range: 4.4-18.6 months) was found to be resistant to ganciclovir (IC50 > 12 mcM). This data confers that the estimated prevalence of resistance to oral ganciclovir is less than 1%, and appears to dovetail well with Drew’s 1991 analysis of patients treated with intravenous ganciclovir in which he observed ganciclovir resistant CMV in 7.6% of patients tested. (The numbers in this intravenous treatment study are higher due to the fact that patients had active CMV disease compared to those who were taking ganciclovir merely as prophylaxis.) So rather than the emergence of drug-resistant CMV, failures on ganciclovir therapy may instead reflect the oral formulation’s low bio-availability, high pre-treatment CMV levels in the body and inadequate penetration into the eyes. Clearly though, more research is called for.