Early inflammation as a footprint of increased mortality risk in infants living with HIV from three African countries

Abstract

In this work our aim was to identify early biomarkers in plasma samples associated with mortality in children with perinatal HIV treated early in life, to potentially inform early intervention targeting this vulnerable group. 20/215 children (9.3%) with perinatal HIV, enrolled within 3 months of age died prematurely within the first year of the study, despite early ART initiation. Using a propensity score, we selected 40 alive study participants having similar clinical and virological records compared to the deceased group. 13 HIV unexposed (HU) healthy children were additionally used as controls. Baseline plasma samples were analyzed using a targeted proteomic approach, and to assess pathogen-associated and damage-associated molecular patterns (PAMPs, DAMPs) levels. Data from deceased participants were compared to both control groups, with multivariate logistic regression models used to evaluate the association between mortality and plasma proteins. We developed a machine learning model to predict mortality risk, finding that IL-6 and CXCL11 not only were higher in deceased children than Matched-children with HIV (p < 0.001 and p = 0.0034) but also predictive of mortality (accuracy of 77%); levels of PAMPs were higher in deceased children (p = 0.0016). Thus, measuring early inflammatory biomarkers, particularly IL-6, could help mortality risk prediction and potentially guide targeted interventions.

Keywords: HIV infection; IL-6; Inflammatory biomarkers; PAMPs and DAMPs; Pediatric population; Predictive model.

Fig. 1

A Study design depicting the numbers of study participants enrolled within the EARTH study and of the HU group, with a list of Deceased-HIV + data used to generate the propensity score used to select the Matched-HIV + . B Mortality reasons and their HIV progression relationship in the Deceased-HIV group. Created with BioRender.com.

Fig. 2

A Principal Component Analysis (PCA) was performed using the prcomp R function to evaluate the distribution of participants according to the data variance of inflammatory plasma markers (Olink T96) data variance. The top 15 proteins contributing to clusters both in PC1 and PC2 components are listed on the side of PCA. B Proteins differentially produced between Deceased-HIV + and Matched-HIV + (ANOVA) and ordered by the rank values calculated as -log10(Adjusted p-value)*log2(Fold Change). Positive values indicate up-regulated proteins. C Violin plots showing the NPX distribution and significance of the 3 proteins with the highest rank values. Statistical analysis was performed using ANOVA. D Univariate logistic regression analysis of the 3 proteins with the highest rank values. The curve shows how the probability of classification between the two groups (Deceased-HIV + in red and Matched-HIV + in orange) varies as protein levels change. E Partial regression plot showing the correlation between the multivariate regression analysis classification probability and the protein levels. Linear regression analysis was used to quantify this relationship.

Fig. 3

Pathway enrichment analysis in the KEGG databases using the R package “enrichr” v3.2 to investigate the biological role of the A proteins differentially produced between the 3 groups (ANOVA analysis) and B proteins associated with mortality through multivariate logistic regression analysis. Only pathways with an adjusted p-value < 0.01 were shown. The size of each bubble corresponds to the number of differentially produced proteins that enrich that pathway. The color gradient is proportional to the log10 adjusted p-value. C Analysis of protein–protein interaction networks showing up-regulated (yellow) proteins and their function [target (cycle) or effector (square)] in Deceased-HIV + versus Matched-HIV + . The relationships between proteins are sourced from the Omnipath database using the OmnipathR v3.2.8 R package. D Violin plots illustrating PAMPs and DAMPs distributions and differences. E Spearman correlation heatmap between PAMPs and DAMPs values and the levels of the 9 proteins in common between the ANOVA and the logistic regression analyses. The red color gradient indicates positive correlations, while the “x” highlights significant results (p-value < 0.05). F Scatter plots to focus on the strongest relationships between DAMPs and proteins observed in the heatmap in panel B. These associations were measured by linear regression (R2) and Spearman correlation (rho).

Fig. 4

Cartoon that illustrates the stepwise workflow in constructing the mortality predictive model. Created with BioRender.com.