Relative Contribution of Pharmacokinetics and Immune Signatures to Clinical Outcomes in Patients With HIV-associated Cryptococcal Meningitis

Abstract

Background: Host immune responses to HIV-associated cryptococcal meningitis are critical in disease outcome. Their interaction with antifungal drug exposure is poorly understood. This study explored associations between immune biomarkers, antifungal drug exposure, and clinical outcomes in HIV-associated cryptococcal meningitis.

Methods: We analyzed serial plasma and cerebrospinal fluid immune biomarkers from 64 participants recruited from the AMBITION-cm trial. We estimated individual-level exposure to amphotericin B, flucytosine, and fluconazole. Associations between immune biomarkers, pharmacokinetic parameters, and clinical outcomes were evaluated.

Results: An inflammatory cerebrospinal fluid response, characterized by coordination between tumor necrosis factor-α, granulocyte colony-stimulating factor, and interleukin-7 signaling, was linked to low fungal burden, low intracranial pressure, and survival. However, the value of specific immune biomarkers did not predict EFA or mortality. Exposure to amphotericin B was significantly associated with EFA.

Conclusions: Favorable clinical outcomes from HIV-associated cryptococcal meningitis are associated with coordinated inflammatory and cytotoxic responses in the central nervous system. Antifungal drug exposure was the dominant predictor of EFA.

Keywords: amphotericin B; cryptococcal meningitis; immunomodulation; pharmacodynamics; pharmacokinetics.

Figure 1.

Baseline immune biomarker concentrations. Baseline soluble biomarker concentrations in A, plasma and B, CSF in patients with HIV-associated cryptococcal meningitis. Boxes indicate median and interquartile range. Whiskers represent 10th and 90th percentiles. Abbreviation: CSF, cerebrospinal fluid.

Figure 2.

Change in biomarker values over time in A) plasma and B) cerebrospinal fluid. Gray lines represent individual patient biomarker values over time; black line is the mean across the patient population.

Figure 2.

Change in biomarker values over time in A) plasma and B) cerebrospinal fluid. Gray lines represent individual patient biomarker values over time; black line is the mean across the patient population.

Figure 3.

Principal component analysis. A, Plasma, day 1: PCs 1, 2, and 3 accounted for 38.7%, 15.2%, and 8.7% of the variance, respectively. The majority of soluble biomarkers in the dataset contribute relatively homogeneously to the variance in PC1. PC2 shows strongly negative loading scores in IL-12p70, IL-2, and VEGF, with PC3 predominated by a strongly positive loading score in IL-17A. B, CSF, day 1: PCs 1, 2, and 3 accounted for 57.0%, 12.3%, and 8.5% of the variance, respectively. Again, there is relatively homogeneous contribution to variance in PC1. In PC2 is overwhelmingly characterized by positive loadings in G-CSF, IL-7, and TNF-α. PC3, in contrast, is predominantly defined by a negative loading score in RANTES, as well as a moderately positive score in TNF-α. C, Plasma, slope values: PCs 1, 2, and 3 accounted for 37.1%, 21.6%, and 8.8% of the variance, respectively. PC1 was again characterized by relatively homogeneous, moderately positive loading scores across the majority of biomarkers. PC2 comprised negative loading scores in IL-2, IL-5, IL-12p40, and IL-12p70. PC3 was characterized by strong positive loading scores in the slope values for IL-6 and IL-7, and negative scores in the slope values for IP-10 and RANTES. D, CSF, slope values: PCs 1, 2, and 3 accounted for 45.0%, 12.3%, and 9.3% of the variance, respectively. PC1 was characterized by moderately positive loading scores across the majority of biomarkers. PC2 was characterized by relatively strong contributions from GM-CSF and RANTES, with a strong negative score for IL-2. PC3 comprised strong positive loading scores for G-CSF, IL-7, and TNF-α. Abbreviations: GM-CSF, granulocyte-macrophage colony-stimulating factor; IL, interleukin; PC, principal component; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor.

Figure 4.

Network analyses showing the associations among immune biomarkers in plasma at baseline. Network plots showing associations between baseline levels of immune biomarkers in plasma in patients categorized according to clinical outcomes. Patients with baseline fungal burden greater than or equal to 6 log₁₀ colony forming units (CFU) per mL cerebrospinal fluid (CSF); patients with baseline fungal burden less than 6 log₁₀ CFU per mL CSF; patients with lumbar opening pressure greater than or equal to 30 cm H₂O; patients with lumbar opening pressure less than 30 cm H₂O; patients with early fungicidal activity (EFA) slower than −0.2 log₁₀ CFU/mL/day; patients with EFA faster than or equal to −0.2 log₁₀ CFU/mL/day; patients who died by 10 wk into the study; patients who survived to 10 wk. Color scheme is purely for presentation.

Figure 5.

Network analyses showing the associations among immune biomarkers in cerebrospinal fluid (CSF) at baseline. Network plots showing associations between baseline levels of immune biomarkers in CSF in patients categorized according to clinical outcomes. Patients with baseline fungal burden greater than or equal to 6 log₁₀ colony forming units (CFU) per mL CSF; patients with baseline fungal burden less than 6 log₁₀ CFU per mL CSF; patients with lumbar opening pressure greater than or equal to 30 cm H₂O; patients with lumbar opening pressure less than 30 cm H₂O; patients with early fungicidal activity (EFA) slower than −0.2 log₁₀ CFU/mL/day; patients with EFA faster than or equal to −0.2 log₁₀ CFU/mL/day; patients who died by 10 wk into the study; patients who survived to 10 wk. Colour scheme is purely for presentation.