Clinical features and serum biomarkers in HIV immune reconstitution inflammatory syndrome after cryptococcal meningitis: a prospective cohort study

Abstract

Background: Although antiretroviral therapy (ART) improves survival in persons with cryptococcal meningitis (CM) and AIDS, ART frequently elicits HIV immune reconstitution inflammatory syndrome (IRIS), an exaggerated and frequently deadly inflammatory reaction that complicates recovery from immunodeficiency. The pathogenesis of IRIS is poorly understood and prediction of IRIS is not possible.

Methods and findings: We prospectively followed 101 ART-naïve Ugandans with AIDS and recent CM for one year after initiating ART, and used Luminex multiplex assays to compare serum cytokine levels in participants who did or did not develop IRIS. IRIS occurred in 45% of participants with recent CM on ART, including 30% with central nervous system (CNS) manifestations. The median time to CM-IRIS was 8.8 wk on ART. Overall mortality on ART was 36% with IRIS and 21% without IRIS. CM-IRIS was independently associated with death (HR = 2.3, 95% CI 1.1-5.1, p = 0.04). Patients experiencing subsequent CM-IRIS had 4-fold higher median serum cryptococcal antigen (CRAG) levels pre-ART (p = 0.006). Higher pre-ART levels of interleukin (IL)-4 and IL-17 as well as lower tumor necrosis factor (TNF)-α, granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF), and vascular endothelial growth factor (VEGF) predicted future IRIS in multivariate analyses (area under the curve [AUC] = 0.82). An algorithm based on seven pre-ART serum biomarkers was a robust tool for stratifying high (83%), moderate (48%), and low risk (23%) for IRIS in the cohort. After ART was initiated, increasing levels of C-reactive protein (CRP), D-dimer, IL-6, IL-7, IL-13, G-CSF, or IL-1RA were associated with increasing hazard of IRIS by time-to-event analysis (each p≤0.001). At the time of IRIS onset, multiple proinflammatory cytokine responses were present, including CRP and IL-6. Mortality was predicted by pre-ART increasing IL-17, decreasing GM-CSF, and CRP level >32 mg/l (highest quartile). Pre-ART CRP level >32 mg/l alone was associated with future death (OR = 8.3, 95% CI 2.7-25.6, p<0.001).

Conclusions: Pre-ART increases in Th(17) and Th(2) responses (e.g., IL-17, IL-4) and lack of proinflammatory cytokine responses (e.g., TNF-α, G-CSF, GM-CSF, VEGF) predispose individuals to subsequent IRIS, perhaps as biomarkers of immune dysfunction and poor initial clearance of CRAG. Although requiring validation, these biomarkers might be an objective tool to stratify the risk of CM-IRIS and death, and could be used clinically to guide when to start ART or use prophylactic interventions.

Figure 1. Normal Immunology of Cryptococcosis.

Left: Myeloid dendritic cells serve as APCs that recognize and phagocytose C. neoformans and generate cytokine signals of IL-1β, IL-6, and TNF-α. GXM polysaccharide capsule (i.e. CRAG) activates APCs via Toll-like receptor 4 and CD14. Center: Activated dendritic cells migrate to lymphatic tissue and present cryptococcal mannoprotein antigen to naïve T cells via MHC-II receptors. This coupled with CD28 costimulation results in Th1 differentiation. Right: Antigen-specific Th1 T cell recognition results in IFN-γ secretion causing classical activation of macrophages with up-regulation of reactive oxygen species (ROS) production; IL-6, TNF-α, IL-8 (CXCL8), CXCL10, fibroblast growth factor (FGF)-2 secretion by macrophages. T cells also secrete additional inflammatory mediators of GM-CSF and VEGF. Th2 T cell responses are not protective. Macrophages also release G-CSF which up-regulates leukotriene synthesis and anticryptococcal activity of neutrophils.

Figure 2. Cumulative survival in persons with prior cryptococcal meningitis newly initiating HIV therapy in Uganda is stratified by the occurrence of cryptococcal IRIS.

IRIS was associated with increased mortality (HR = 2.4, 95% CI 1.1–5.3, p = 0.035). Included with the controls are three suspected, but unproven cases of CM-IRIS, three unknown causes of death (two suspected pulmonary emboli). Two deaths have been excluded of persons with known virologic suppression with clinical IRIS who refused lumbar punctures to exclude alternative etiologies of their deterioration, of which one of these deaths was attributed to suicide.

Figure 3. Cumulative incidence of cryptococcal IRIS.

The data are based on a weighed predictive biomarker score composed of pre-ART serum levels of seven biomarkers—IL-4, IL-17, G-CSF, GM-CSF, CCL2 (MCP-1), TNF-α, and VEGF—calculated as follows. Probability of IRIS =  where:Probability risk is scored from 0 to 1.0 with the categorical classification being defined for low risk (0–0.40), moderate risk (0.41–0.60), and high risk (0.61–1.0). The performance as quantified by the AUC is 0.82 for correct prediction (i.e. C-statistic).

Figure 4. Individual longitudinal increases in CRP and IL-6 from baseline pre-ART, to the time point before IRIS, and at time of IRIS event.

The gray shading represents the interquartile range (25^th to 75^th percentile) of the CM cohort controls without IRIS over the first 8 wk of ART. The median time of IRIS was 8.8 wk. IL-6 and CRP are increased before the IRIS event (p<0.001) and at time of IRIS (p<0.001).

Figure 5. Survival on ART in persons with prior cryptococcosis stratified by their mortality biomarker risk score.

(A) Full predictive model (AUC = 0.84). (B) Parsimonious three-biomarker model based on IL-17 (p = 0.015), GM-CSF (p = 0.022), and categorical CRP >32 mg/l (p = 0.004) (AUC = 0.76). With a categorical cutoff point for the model with high risk of probability >0.5 to 1.0, the positive predictive value was 69% and negative predictive value 83% with 48% sensitivity and 92% specificity. (C) Survival stratified by baseline, pre-ART CRP level of<32 or ≥32 mg/l (AUC = 0.65). The three respective ROC curves are presented in Figure S4.