PloS one 2016 8 411(8) e0159727 doi 10.1371/journal.pone.0159727
Cryptococcal meningitis (CM) is a leading cause of HIV-associated mortality. In clinical trials evaluating treatments for CM, biomarkers of early fungicidal activity (EFA) in cerebrospinal fluid (CSF) have been proposed as candidate surrogate endpoints for all- cause mortality (ACM). However, there has been no systematic evaluation of the group-level or trial-level evidence for EFA as a candidate surrogate endpoint for ACM.
We conducted a systematic review of randomized trials in treatment of CM to evaluate available evidence for EFA measured as culture negativity at 2 weeks/10 weeks and slope of EFA as candidate surrogate endpoints for ACM. We performed sensitivity analysis on superiority trials and high quality trials as determined by Cochrane measures of trial bias.
Twenty-seven trials including 2854 patients met inclusion criteria. Mean ACM was 15.8% at 2 weeks and 27.0% at 10 weeks with no overall significant difference between test and control groups. There was a statistically significant group-level correlation between average EFA and ACM at 10 weeks but not at 2 weeks. There was also no statistically significant group-level correlation between CFU culture negativity at 2weeks/10weeks or average EFA slope at 10 weeks. A statistically significant trial-level correlation was identified between EFA slope and ACM at 2 weeks, but is likely misleading, as there was no treatment effect on ACM.
Mortality remains high in short time periods in CM clinical trials. Using published data and Institute of Medicine criteria, evidence for use of EFA as a surrogate endpoint for ACM is insufficient and could provide misleading results from clinical trials. ACM should be used as a primary endpoint evaluating treatments for cryptococcal meningitis.