Benefits of enhanced infection prophylaxis at antiretroviral therapy initiation by cryptococcal antigen status.

OBJECTIVES: To assess baseline prevalence of cryptococcal antigen (CrAg) positivity; and its contribution to reductions in all-cause mortality, deaths from cryptococcus and unknown causes, and new cryptococcal disease in the REALITY trial. DESIGN: Retrospective CrAg testing of baseline and week-4 plasma samples in all 1805 African adults/children with CD4+ cell count less than 100 cells/µl starting antiretroviral therapy who were randomized to receive 12-week enhanced-prophylaxis (fluconazole 100 mg/day, azithromycin, isoniazid, cotrimoxazole) vs. standard-prophylaxis (cotrimoxazole). METHODS: Proportional hazards models were used to estimate the relative impact of enhanced-prophylaxis vs. standard-cotrimoxazole on all, cryptococcal and unknown deaths, and new cryptococcal disease, through 24 weeks, by baseline CrAg positivity. RESULTS: Excluding 24 (1.4%) participants with active/prior cryptococcal disease at enrolment (all treated for cryptococcal disease), 133/1781 (7.5%) participants were CrAg-positive. By 24 weeks, 105 standard-cotrimoxazole vs. 78 enhanced-prophylaxis participants died. Of nine standard-cotrimoxazole and three enhanced-prophylaxis cryptococcal deaths, seven and two, respectively, were CrAg-positive at baseline. Among deaths of unknown cause, only 1/46 standard-cotrimoxazole and 1/28 enhanced-prophylaxis were CrAg-positive at baseline. There was no evidence that relative reductions in new cryptococcal disease associated with enhanced-prophylaxis varied between baseline CrAg-positives [hazard-ratio = 0.36 (95% confidence interval 0.13-0.98), incidence 19.5 vs. 56.5/100 person-years] and CrAg-negatives [hazard-ratio = 0.33 (0.03-3.14), incidence 0.3 vs. 0.9/100 person-years; Pheterogeneity = 0.95]; nor for all deaths, cryptococcal deaths or unknown deaths (Pheterogeneity > 0.3). CONCLUSION: Relative reductions in cryptococcal disease/death did not depend on CrAg status. Deaths of unknown cause were unlikely to be cryptococcus-related; plausibly azithromycin contributed to their reduction. Findings support including 100 mg fluconazole in an enhanced-prophylaxis package at antiretroviral therapy initiation where CrAg screening is unavailable/impractical.