Individualized Transcriptional Resolution of Complicated Malaria in a Colombian Study

Abstract

To evaluate whether recovery from complicated malaria follows a common trajectory in terms of immunological mechanism or, rather, is highly individualized for each patient, we performed longitudinal gene expression profiling of whole blood. RNA sequencing (RNAseq) was performed on blood samples obtained from eight patients on four consecutive days between hospital admission and discharge. Six patients were infected with Plasmodium falciparum, and two with Plasmodium vivax; one patient was a pregnant woman infected with P. falciparum, who was hospitalized for several weeks. The characterization of blood transcript modules (BTM) and blood informative transcripts (BIT) revealed that patients' responses showed little commonality, being dominated by the balance of gene activity relating to lymphocyte function, inflammation, and interferon responses specific to each patient. Only weak correlations with specific complicated malaria symptoms such as jaundice, thrombocytopenia, or anemia were observed. The differential expression of individual genes, including transcripts derived from the human leukocyte antigen (HLA) complex, generally reflected differences in the underlying immune processes. Although the results of this pilot study do not point to any single process that might provide a target for complicated malaria treatment or prevention or personalized medical strategies, larger patient series and more extensive blood sampling may allow the classification of patients according to their type of response in order to develop novel therapeutic approaches.

Keywords: Plasmodium; RNAseq; gene expression; longitudinal profiling; transcriptomics.

Figure 1

Two-way hierarchical clustering of blood transcript modules (BTM) and blood informative transcripts (BIT) by individual and time. Each row represents one patient on the day indicated by the last number of the label, and each column is the principal component (PC1) score for a BTM, with co-clustered BIT PC1 highlighted. The red-to-blue scale indicates positive to negative scores. The figure highlights the tendency of the samples to cluster by individual and provides an overview of the axes of variation shared by particular patients.

Figure 2

Lymphocyte activity inference in each individual. Each plot (A–C) shows how a PC1 based on natural killer (NK) cell, T-helper cell, or T-cytotoxic cell specific transcripts correlates with Axis T and varies by individual, indicated by different colors. Graphs (D–F) plot the residuals of the linear regressions by individual, highlighting patient-specific lymphocyte activity. p-values and adjusted R² estimates were calculated through an analysis of variance (ANOVA).

Figure 3

Standardized change in BIT by individual and time. Each box shows the difference between the sample PC1 score and the sample mean for the indicated individual from Day 1 to Day 4 of sampling. Red implies strong elevation, and blue strong reduction in the expression of genes in the Axis.

Figure 4

Two-way hierarchical clustering of differentially expressed gene (DEG) by individual and time. Each row represents one patient on the day indicated by the last number of the label, and each column is the standardized expression of a DEG in ANOVA contrasting each individual with the remaining sample. The figure highlights clusters of genes that tend to be patient-specific, but shows (like the BTM analysis) that these are to some extent shared by pairs of patients. The colored bar at base indicates clusters discussed in the text.

Figure 5

Two-way hierarchical clustering of HLA complex gene expression by individual and time. Each row represents one patient on the day indicated by the last number of the label, and each column is the standardized expression of the indicated HLA gene or lymphocyte activity estimate. The figure highlights three clusters of co-expression, namely, HLA class I to the left and two HLA class II clusters in the middle and right of the grid.