Impact of cerebrospinal fluid leukocyte infiltration and activated neuroimmune mediators on survival with HIV-associated cryptococcal meningitis

Abstract

Introduction: Cryptococcal meningitis remains a prominent cause of death in persons with advanced HIV disease. CSF leukocyte infiltration predicts survival at 18 weeks; however, how CSF immune response relates to CSF leukocyte infiltration is unknown.

Methods: We enrolled 401 adults with HIV-associated cryptococcal meningitis in Uganda who received amphotericin and fluconazole induction therapy. We assessed the association of CSF leukocytes, chemokine, and cytokine responses with 18-week survival.

Results: Participants with CSF leukocytes ≥50/microliter had a higher probability of 18-week survival compared with those with ≤50 cells/microliter (68% (52/77 vs. 52% (151/292); Hazard Ratio = 1.63, 95% confidence interval 1.14-2.23; p = 0.008). Survival was also associated with higher expression of T helper (Th)-1, Th17 cytokines, and immune regulatory elements. CSF levels of Programmed Death-1 Ligand, CXCL10, and Interleukin (IL)-2 independently predicted survival. In multivariate analysis, CSF leukocytes were inversely associated with CSF fungal burden and positively associated with CSF protein and immune parameters (interferon-gamma (IFN-γ), IL-17A, tumor necrosis factor alpha (TNF)-α, and circulating CD4+ and CD8+ T cells).

Conclusion: 18-week survival after diagnosis of cryptococcal meningitis was associated with higher CSF leukocytes at baseline with greater T helper 1 (IFN-γ, IL-2 and TNF-α cytokines), T helper 17 (IL-17A cytokine) and CXCR3+ T cell (CXCL10 chemokine) responses. These results highlight the interdependent contribution of soluble and cellular immune responses in predicting survival and may support potential pathways for adjunctive immune therapy in HIV-associated cryptococcal meningitis.

Fig 1. Unadjusted association of CSF white blood cells with CSF and blood clinical features.

A (i)–levels of stratified CSF white blood cells. A (ii)–levels of CSF proteins. B (i)–levels of peripheral blood CD4⁺ T cells. B (ii)–levels of peripheral blood–CD8⁺ T cells. C–CSF fungal burden. D–peripheral white blood cells. The interlinking bars–shows two variable unpaired comparison. The flat bar shows three variables analysis of variance (ANOVA) comparisons. The error bars show median and 95% confidence intervals (CI). A-(i) both the medians and interquartile range was 4 cells/μL, (reported as <5 cells/μL). Asterisk *—show statistically significant variables reported at p-value <0.050, at 95% CI. §The flat bar shows three variables analysis using ANOVA test.

Fig 2. Unadjusted Correlation of CSF cytokines and chemokine levels with CSF leukocyte counts.

A- Th1 cytokines; A (i)—Interleukin 2, A (ii)—Interferon gamma, A (iii)—Tumor necrosis factor alpha. B—Th17 cytokine, IL-17A. C–Immune regulatory elements; (i)–Interleukin 10 (IL-10), and programmed death 1 ligand (PD-L1). D Chemokines; D (i) CXCL10/IP-10 and D (ii)–CCL11/Eotaxin. The CSF white cells; (≤50 cells/μL; n = 318), (51–200 cells/μL; n = 57) and (201–500 cells/μL; n = 26) participants. The interlinking bars–shows two variable unpaired comparison. Error bars–show median and 95% CI. The flat bar shows three variables ANOVA comparisons. Asterisks *—show statistically significant variables reported at p-value <0.050, at 95% confidence intervals.

Fig 3. Correlation of CSF white cells with 18-week survival.

A–survival by CSF white cell intervals (<5 cells/μL; n = 245), (5–20 cells/μL; n = 31), (21–50 cells/μL; n = 42), (51–100 cells/μL; n = 26), (101–200 cells/μL; n = 31), and (201–500 cells/μL; n = 26). B—18 weeks survival by CSF white cells; (≤50 cells/μL; n = 318), (51–200 cells/μL; n = 57) and (201–500 cells/μL; n = 26) participants. C– 18-week survival with CSF ≤50 cells/μL. D (i-iii)–illustrates trends in immune responses between survivors and those who died during 18-weeks of follow-up. Statistics—Mann-Whitney unpaired t-test. *—show statically significant variables. NS- not significant. Error bars–show 95% confidence intervals. p-values, p<0.050 were statistically significant.

Fig 4. Principal Component Analyses showing data clusters and variations on principal component, PC1 and PC2.

(A)—(I & ii) orthogonal Eigenvectors showing clustering and variation of the principal components between CSF fungal burden and survival with the expressed cytokine/chemokine profile. (i & iii) diagonal Eigenvectors showing clustering and variation of the principal components between CSF fungal burden and survival with CSF fungal burden and host survival and CSF white cells, CSF protein, peripheral white cells, CD4⁺ and CD8⁺ T cells. Eigenvectors projection at >5 from the center of the plane shows greater power of the principal components to predict the outcome. Also, the furthest the component to the cluster variation among all principal components. C—contribution of each principal component to the cluster variation among a subset (cytokines and chemokines) principal components.