Objective: To determine criteria for the diagnosis of cytomegalovirus (CMV) colitis and to analyse stages of the course and prognosis of CMV colonic involvement in HIV-1-infected patients.
Design: Prospective search for CMV colonic involvement with systematic biopsies to search for CMV intranuclear inclusion bodies and for CMV culture. The evolution of CMV colonic involvement was estimated using further coloscopies and autopsy.
Setting: Infectious diseases department in a tertiary referral teaching hospital in Paris, France.
Participants: Fifty-five consecutive patients with HIV-1 infection, who had not previously received anti-CMV drugs, and who had at least one coloscopy performed.
Results: According to initial coloscopy, colitis, either ulcerative or inflammatory, was found in nine (16%) out of the 55 patients, CMV intranuclear inclusions were present in the colon of four (7%) patients, and colonic cultures were positive for CMV in 15 (27%) patients. The results of the initial coloscopy showed a positive correlation between endoscopic colitis (either ulcerative or inflammatory), CMV inclusions and positive CMV culture from colonic biopsies. The absence of endoscopic ulcerative lesions had a 98% (49 out of 50) negative predictive value for recording CMV inclusions in the colon (95% confidence interval, 89-100). CMV inclusions were recorded in three out of five ulcerative colitis. Male homosexuality, HIV-1 infection stages IVB, C1, D or E, according to the Centers for Disease Control and Prevention classification, CD4 lymphocyte count < 200 x 10(6)/l and CMV viraemia also correlated positively with CMV colonic involvement. During the observation period (mean, 7.3 months), the estimated incidence of CMV colitis according to coloscopic studies was 13%. Deterioration in condition was the most frequent spontaneous evolution of CMV colonic infection, whereas anti-CMV treatment resulted in an improvement. Ulcerative lesions occurred earlier in patients with colonic CMV inclusions or positive colonic CMV culture than in patients without CMV colonic involvement at the initial coloscopy. CMV colitis occurred late in the course of HIV-1 infection, on average 4 months before death. The presence of CMV inclusions was an indicator of poor prognosis with earlier occurrence of CMV viraemia and retinitis and no survival after 9 months.
Conclusions: These results confirm that the colon is a target organ for CMV in HIV-1-infected patients. Coloscopy should be used to diagnose CMV colitis, because of the close correlation between endoscopic and histological data (i.e., intranuclear inclusions). This combination allows us to propose an evolutive staging of CMV colonic involvement and provide stratification criteria to assess the efficacy of anti-CMV drugs.