Antibiotic treatment for Tuberculosis induces a profound dysbiosis of the microbiome that persists long after therapy is completed

Abstract

Mycobacterium tuberculosis, the cause of Tuberculosis (TB), infects one third of the world's population and causes substantial mortality worldwide. In its shortest format, treatment of TB requires six months of multidrug therapy with a mixture of broad spectrum and mycobacterial specific antibiotics, and treatment of multidrug resistant TB is longer. The widespread use of this regimen makes this one of the largest exposures of humans to antimicrobials, yet the effects of TB treatment on intestinal microbiome composition and long-term stability are unknown. We compared the microbiome composition, assessed by both 16S rDNA and metagenomic DNA sequencing, of TB cases during antimycobacterial treatment and following cure by 6 months of antibiotics. TB treatment does not perturb overall diversity, but nonetheless dramatically depletes multiple immunologically significant commensal bacteria. The microbiomic perturbation of TB therapy can persist for at least 1.2 years, indicating that the effects of TB treatment are long lasting. These results demonstrate that TB treatment has dramatic effects on the intestinal microbiome and highlight unexpected durable consequences of treatment for the world's most common infection on human ecology.

Figure 1

(a) Shannon diversity index measured for all groups used in this study, based on 16S rDNA sequencing data. The LTBI (treatment) group indicates subjects who are the age-matched controls for the treatment group, and the LTBI (cured) group indicates the age-matched controls for the cured group. (b) Raw number of observed OTUs clustered at 97% similarity for the indicated groups.

Figure 2

HRZE treatment perturbs the taxonomic structure of the microbiome. (a) NMDS ordination of HRZE treated subjects (treatment, purple) or LTBI controls (blue) based on 16S rDNA sequencing (b) Family taxonomic distribution of the intestinal microbiota from subjects with LTBI and subjects with TB on treatment. (c) Heatmap of the top 50 most abundant taxa generated with DESeq2 showing unsupervised clustering of TB cases on treatment vs. LTBI controls. Age and sex are also shown but were not accounted for the in DESeq model. Genus and species names are based on OTU identification (Supplementary Table 2) and therefore names may be redundant, but represent different 16S-based OTUs. (d) Taxonomic abundance profiling comparing treatment vs LTBI participants using LeFSe to determine differentially abundant Genera. Box and whisker plots of differentially abundant genera are shown based on the DESeq normalized data. Plots show the first and third quartiles of the abundance data, the line represents the median, and the whiskers show 1.5 times the value of the interquartile range.

Figure 3

Taxonomic and biochemical microbiomic perturbation induced by HRZE. (a) NMDS ordination plot on metagenomic taxonomy data demonstrating microbiomic differences between healthy individuals and subjects on HRZE treatment. For this comparison, the healthy group consists of LTBI and Mtb uninfected subjects. (b) Comparative abundance plots between healthy Haitian individuals and cases on HRZE treatment showing the most abundant species. (c) Unsupervised hierarchical clustering of significantly altered taxa from species-level metagenomic data. (d) Abundance of significantly different KEGG modules between healthy volunteers and cases on treatment.

Figure 4

TB treatment induces a lasting alteration in microbiome structure. (a) DPCoA ordination plot of cured cases compared to LTBI controls based on 16S rDNA sequencing. (b) Family level taxonomic distribution of the intestinal microbiota from subjects with LTBI or who are cured. (c) Heatmap of the 40 most abundant taxa generated with DESeq2 showing unsupervised clustering of cured vs. LTBI subjects. Age and sex are also shown but were not accounted for the in DESeq model. The number of days that each patient has been off treatment is also shown. Genus and species names are based on OTU identification (Supplementary Table 3) and therefore names may be redundant, but represent different 16S-based OTUs. (d) Taxonomic abundance profiling comparing cured vs LTBI subjects. Taxa are significant from LeFSe (p < 0.05 and LDA cutoff >3.0).

Figure 5

TB treatment induces a lasting alteration in microbiome structure and function. (a) DCA ordination plot on metagenomic taxonomy data in healthy (combined Mtb-uninfected and LTBI community controls) and cured individuals. (b) Comparative abundance plots between healthy Haitian individuals and cured subjects showing the top 40 most abundant species between the two groups. (c) Unsupervised hierarchical clustering of significantly altered taxa. (d) Abundance of significantly different KEGG modules between healthy and cured subjects.