Prevalence of high CSF fungal burden and its estimation using CSF glucose among HIV-infected adults with cryptococcal meningitis admitted to Masaka and Entebbe hospitals

Abstract

Globally, cryptococcal meningitis (CM) was responsible for 19% of AIDS-related deaths and of these, sub-Saharan Africa contributed 75% in 2020. It is the leading central nervous system (CNS) fungal infection worldwide. In Uganda, CM is associated with a 28-day mortality of 40%. There is need to improve survival of patients with CM through proper clinical management and follow- up care. Recommended follow-up cerebrospinal fluid (CSF) cultures in CM patients are not feasible in developing countries including Uganda because of resource limitations. A fungal burden of >100,000 CFU/mL at cryptococcal meningitis diagnosis is associated with increased mortality. We lack good quality epidemiological data regarding the prevalence of high CSF fungal burden among HIV-infected patients with CM yet this particular category of patients is at a higher risk of death. There is a need to find simpler, readily available, faster, and cheaper tests that can be used to estimate CSF fungal burden for monitoring CM treatment response. CSF glucose measurement is cheap, gives immediate results, and can be determined at the bedside. However, CSF glucose had not been evaluated for estimating CSF fungal burden in HIV-infected adults with CM. Objectives To determine the prevalence of high CSF fungal burden and accuracy of CSF glucose in estimating CSF fungal burden in HIV-infected adults with CM. Methods This was a retrospective serial cross-sectional study that utilized data collected as part of the CryptoDex trial, which was a randomized, phase III clinical trial in HIV-infected adults with CM. The primary aim of the trial was to investigate the effect of dexamethasone as adjunctive therapy

for CM on 10-week survival of HIV-infected adult patients. After receiving the necessary regulatory approvals, 212 records of former CryptoDex participants who had at least 1 lumbar puncture done were analyzed to determine prevalence of high CSF fungal burden and accuracy of using CSF glucose in estimating CSF fungal burden using non-parametric ROC regression techniques while adjusting for covariates. Results. The prevalence of high fungal burden (>100,000 CFU/mL) at baseline (day 1) was 27.5% (95% CI 21.6, 34.2). This prevalence was not significantly different between males and females (P- value = 0.420) or those on ART and those not on ART (P-value = 0.580). Those with higher CD4 (CD4 ≥100 cells per µL) had a significantly lower prevalence of 4.4% compared to 31.6% in those with CD4 <100 cells per µL (P-value = 0.004). Those with very low CD4 (CD4 < 50 cells per µL) had a very high prevalence of 34.72% and this was significantly higher than 13.04% in those with CD4 ≥ 50 cells per µL (P-value = 0.005). The AUC seen in this study at baseline was 0.42 (95% CI 0.27 – 0.58) implying that CFS-glucose test is worse than random chance in estimating CSF fungal burden. The study established that the best cut off of blood/CSF glucose ratio at baseline was 2.2, resulting in sensitivity of only 26.44% (95% CI 19.31 – 33.57), specificity of 70.00% (95% CI 62.59 – 77.41) and overall accuracy of 58.04% (95% CI 53.25 – 62.40).

It was further established that model 2 (at day 3) had worse sensitivity of 16.13% (95% CI 08.07 – 24.19), better specificity of 87.76% (95% CI 80.57 – 94.94) and better overall accuracy of 74.79% (95% CI 71.98 – 76.24) with better AUC of 0.54 (95% CI 0.34 – 0.75) though this is not statistically different from 0.5 and therefore at this point the test is still non-useful.

Conclusions. More than a quarter of patients had high CSF fungal burden (>100,000 CFU/mL) especially those with CD4 cell counts below 50 cells per µL. CSF glucose did not estimate CSF fungal burden. We recommend aggressive management of CM among HIV-infected patients, particularly those with CD4 cell counts below 50 cells per µL.