Integration of genomic and pharmacokinetic data to predict clinical outcomes in HIV-associated cryptococcal meningitis

Abstract

Cryptococcal meningitis causes an estimated 112,000 global deaths per annum. Genomic and phenotypic features of the infecting strain of Cryptococcus spp. have been associated with outcomes from cryptococcal meningitis. Additionally, population-level pharmacokinetic variability is well documented in these patient cohorts. The relative contribution of these factors to clinical outcomes is unknown. Based in Malawi, we conducted a sub-study of the phase 3 Ambition-CM trial (ISRCTN72509687), collecting plasma and cerebrospinal fluid at serial time points during the first 14 days of antifungal therapy. We explored the relative contribution of pathogen genotype, drug resistance phenotype, and pharmacokinetics on clinical outcomes including lumbar opening pressure, pharmacodynamic effect, and mortality. We report remarkable genomic homogeneity among infecting strains of Cryptococcus spp., within and between patients. There was no evidence of acquisition of antifungal resistance in our isolates. Genotypic features of the infecting strain were not consistently associated with adverse or favorable clinical outcomes. However, baseline fungal burden and early fungicidal activity (EFA) were associated with mortality. The strongest predictor of EFA was the level of exposure to amphotericin B. Our analysis suggests the most effective means of improving clinical outcomes from HIV-associated cryptococcal meningitis is to optimize exposure to potent antifungal therapy.

Importance: HIV-associated cryptococcal meningitis is associated with a high burden of mortality. Research into the different strain types causing this disease has yielded inconsistent findings in terms of which strains are associated with worse clinical outcomes. Our study suggests that the exposure of patients to potent anti-cryptococcal drugs has a more significant impact on clinical outcomes than the strain type of the infecting organism. Future research should focus on optimizing drug exposure, particularly in the context of novel anticryptococcal drugs coming into clinical use.

Keywords: Cryptococcus; cryptococcal meningitis; genomics; pharmacodynamics; pharmacokinetics.

Fig 1

Cerebrospinal fluid processing strategy. Following patient recruitment and randomization, lumbar punctures were performed on days 1 (prior to treatment initiation), 7, and 14. At each time point, CSF was plated onto agar that contained either no fluconazole (F−) or fluconazole at 10 µg/mL (F+). Plates were incubated for 48 h at 30°C. Four colonies were selected from each plate and each colony was plated onto fresh agar with either no fluconazole (F−) or fluconazole at 10 µg/mL (F+), consistent with the initial plating conditions. Plates were again incubated for 48 h at 30°C. Isolates were stored at −80°C until retrieved for WGS and drug susceptibility testing.

Fig 2

Early fungicidal activity. A decline in fungal burden in 64 patients administered amphotericin-B-based antifungal therapy for HIV-associated cryptococcal meningitis. (A) Control arm; (B) single dose AmBisome arm.

Fig 3

Maximum likelihood phylogeny. Maximum likelihood phylogeny for each lineage found in Cryptococcus isolates using RAxML GTRCAT model: (A) 585 C. neoformans, (B) 64 C. gattii VGI, (C) 40 C. gattii VGIV. C. neoformans isolates annotated by clades or monophyletic groups. All isolates are annotated by the patient of origin.