Abstract Body

Background: Randomised trials (RCTs) from low-income countries have shown that early ART among antiretroviral-naïve HIV-1–infected patients presenting with cryptococcal meningitis (CM) is associated with higher mortality than delayed ART. There is limited information on the impact of timing of ART on short-term mortality in patients with CM cared for in high-income settings.

Methods: Data on ART-naive patients diagnosed with CM between 1998 and 2009 were combined from cohorts contributing to the COHERE, NA-ACCORD and CNICS collaborations. Follow-up time was calculated from date of CM diagnosis to the first of death, last follow-up or 6 months post-diagnosis. We mimicked an RCT comparing regime A (“start ART within 14 days of CM diagnosis”) with regime B (“defer ART until 14 -56 days after CM diagnosis”). Marginal structural models were used to compare the effects of these regimes on all-cause mortality. Models were adjusted for gender, age, mode of transmission, CD4, viral load (VL), AIDS (other than CM), year of diagnosis and whether data were European/North American.

Results:
Of 235 patients (84% male) from 28 cohorts across Europe and N. America with full covariate data, 42 (18.0%) died within 6 months. Median age at CM diagnosis was 38 years (IQR 34-44). Of 150 patients (64%) who started ART, 18 (12%) died within 6 months. 7/62 (11%) patients died among those who started ART within 2 weeks of CM diagnosis, compared with 11/88 (12%) among those who started ART after 2 weeks. The graph shows crude survival over time, according to when the patient started ART. In data mimicking an RCT, there were 15 deaths (33.3 years follow-up) in regime A and 26 deaths (47.8 years) in regime B. The crude and adjusted hazard ratios comparing deferred with early treatment were 1.29 (95% CI 0.68-2.43) and 1.30 (0.66-2.55).

Conclusions: We found little evidence that early ART was associated with higher mortality after CM than deferred ART, although confidence intervals were wide. Although we adjusted for potential confounding factors, confounding and selection bias may not be fully adjusted for; we aim to address this limitation by ascertaining additional information on treatment of CM after diagnosis. Mortality among patients cared for in high income settings was clearly lower than reported in the RCTs conducted in low-income countries.


Kaplan-Meier survival estimates according to time of starting ART