Host transcriptional response to TB preventive therapy differentiates two sub-groups of IGRA-positive individuals

Abstract

We hypothesised that individuals with immunological sensitisation to Mycobacterium tuberculosis (Mtb), conventionally regarded as evidence of latent tuberculosis infection (LTBI), would demonstrate binary responses to preventive therapy (PT), reflecting the differential immunological consequences of the sterilisation of viable infection in those with active Mtb infection versus no Mtb killing in those who did not harbour viable bacilli. We investigated longitudinal whole blood transcriptional profile responses to PT of Interferon gamma release assay (IGRA)-positive tuberculosis contacts and IGRA-negative, tuberculosis-unexposed controls. Longitudinal unsupervised clustering analysis with a subset of 474 most variable genes in antigen-stimulated blood separated the IGRA-positive participants into two distinct subgroups, one of which clustered with the IGRA-negative controls. 117 probes were differentially expressed over time between the two cluster groups, many of them associated with immunological pathways important in mycobacterial control. We contend that the differential host RNA response reflects lack of Mtb viability in the group that clustered with the IGRA-negative unexposed controls, and Mtb viability in the group (1/3 of IGRA-positives) that clustered away. Gene expression patterns in the blood of IGRA-positive individuals emerging during the course of PT, which reflect Mtb viability, could have major implications in the identification of risk of progression, treatment stratification and biomarker development.

Keywords: Latent tuberculosis infection; Preventive therapy; Transcriptome.

Figure 1

Study overview, showing patient numbers and exclusions.

Figure 2

Volcano plots showing genes significantly differentially expressed between IGRA+ and IGRA- individuals. Plots are shown for TB1-stimulated samples at Visit (V) 1 [A] and V2 [B] and TB2-stimulated samples at V1 [C] and V2 [D]. Genes overexpressed in IGRA+s with log2Foldchange (LFC) > 1 and Benjamini-Hochberg adjusted p value ≤ 0.05 are shown in red. Genes underexpressed in IGRA+ individuals with LFC <-1 and BH adjusted p value ≤ 0.05 are shown in blue. Genes with LFC >2.7 and < −1.7 are annotated with their gene symbols. Dotted line denotes the significance cut-off (BH adjusted p value ≤ 0.05).

Figure 3

Longitudinal differential gene expression analysis between patient cluster groups 1 and 2 in TB2-stimulated whole blood samples. With 1° of freedom, 117/474 transcripts were SDE over time and between Cluster groups 1 and 2 (BH corrected p value < 0.05). The coefficients obtained were used to group together significant genes with similar longitudinal expression patterns. MaSigPro identified 9 gene groups. Plots of gene expression against time for these gene groups are shown for patient Cluster groups 1 (green) and 2 (blue). Lines join the median expression values of the gene groups at each timepoint. The gene symbols are listed for each gene group.

Fig. 4

Longitudinal changes in monocyte: lymphocyte ratio through preventive therapy in IGRA+ subgroups A and B. Cibersortx was used to estimate the abundance of monocytes and lymphocytes in the TB2-stimulated whole blood samples at each visit, and the monocyte: lymphocyte ratio was calculated. (A) Boxplots showing the Monocyte: Lymphocyte ratios at Visits 1, 2 and 3 for IGRA- healthy controls and IGRA+groups A and B. NS denotes p > 0.05, * denotes p ≤ 0.05. Scatterplots showing the change in Monocyte: lymphocyte ratio over the time-course of the study period for (B) IGRA+subgroup A and (C) IGRA+subgroup B, where Visit 1 is 0 weeks, Visit 2 is 2 weeks and Visit 3 is 13 weeks, with 90% confidence intervals shown.