Bad company? The pericardium microbiome in people investigated for tuberculous pericarditis in an HIV-prevalent setting

Abstract

Background: The site-of-disease microbiome and predicted metagenome were evaluated in a cross-sectional study involving people with presumptive tuberculous pericarditis. We also explored the interaction between C-reactive protein (CRP) and the microbiome.

Methods: People with effusions requiring diagnostic pericardiocentesis (n=139) provided pericardial fluid for sequencing and blood for CRP measurement.

Results: Pericardial fluid microbiota differed in β-diversity among people with definite (dTB, n=91), probable (pTB, n=25), and non- (nTB, n=23) tuberculous pericarditis. dTBs were Mycobacterium-, Lacticigenium-, and Kocuria-enriched vs. nTBs. HIV-positive dTBs were Mycobacterium-, Bifidobacterium-, Methylobacterium-, and Leptothrix-enriched vs. HIV-negative dTBs. HIV-positive dTBs on ART were Mycobacterium- and Bifidobacterium-depleted vs. those not on ART. dTBs exhibited enrichment in short-chain fatty acid (SCFA) and mycobacterial metabolism pathways vs. nTBs. Additional non-pericardial involvement (pulmonary infiltrates) was associated with Mycobacterium-enrichment and Streptococcus-depletion. Mycobacterium reads were in 34 % (31/91) of dTBs, 8 % (2/25) of pTBs and 17 % (4/23) nTBs. People with CRP above (vs. below) the median value had different β-diversity (Pseudomonas-depleted). No correlation was found between enriched taxa in dTBs and CRP.

Conclusions: Pericardial fluid microbial composition varies by tuberculosis status, HIV (and ART) status and dTBs are enriched in SCFA-associated taxa. The clinical significance, including mycobacterial reads in nTBs and pTBs, requires evaluation.

Keywords: Extrapulmonary tuberculosis; HIV; Microbiome; Pericarditis.

Fig. 1.. dTBs have a distinct microbiome enriched in Mycobacterium and other potential opportunistic pathogens.

(A) Although α-diversity, which measures species diversity, was similar across the three groups (based on Kruskal-Wallis test), (B) β-diversity, which measures differences in microbial composition, differed significantly (based on PERMANOVA analysis) between pericardial fluid microbiota of pTBs (n=25) vs. dTBs (n=91) and pTBs vs. nTBs (n=23). (C) dTBs vs. nTBs were Mycobacterium-, Lacticigenium- and Kocuria-enriched, and Streptococcus-depleted. (D) dTBs vs. pTBs were Mycobacterium-, Lacticigenium-, and Caulobacter-enriched and Streptococcus-depleted. (E) pTBs vs. nTBs were Novosphingobium- and Streptococcus-enriched (red dotted line represents adjusted p-value=0.2; circle size represents relative abundance; taxa identified as possible contaminants are shown in grey). (F) Genus-level average relative abundances of top taxa differentiating between TB groups (p-values based on Kruskal-Wallis test). PERMANOVA, permutational multivariate analysis of variance; dTB: definite tuberculous pericarditis; nTB: non-tuberculous pericarditis; PF: pericardial fluid; pTB: probable tuberculous pericarditis; TB: tuberculosis; TBP: tuberculous pericarditis.

Fig. 2.. Bifidobacterium enrichment is associated with Mycobacterium and HIV positivity in TBP.

In PLHIV (n=73), (A) α-diversity did not differ by TB status (Mann-Whitney test), (B) however, β-diversity did (based on PERMANOVA), (C) where HIV-positive dTBs were Mycobacterium-, Bifidobacterium-, and Leptothrix-enriched, and Streptococcus-depleted compared to HIV-positive pTBs. In people without HIV (n=64), (D) α-diversity (Kruskal-Wallis test) and (E) β-diversity did not differ by TB status, and (F) HIV-negative dTBs were Mycobacterium, Caulobacter, and Weeksella-enriched, and Streptococcus-depleted compared to HIV-negative pTBs (red dotted line represents adjusted p-value=0.2; circle size represents relative abundance; taxa identified as possible contaminants are shown in grey). PERMANOVA, permutational multivariate analysis of variance; dTB: definite tuberculous pericarditis; HIV+: HIV-positive; HIV−: HIV-negative; nTB: non-tuberculous pericarditis; PF: pericardial fluid; pTB: probable tuberculous pericarditis; TB: tuberculosis; TBP: tuberculous pericarditis.

Fig. 3.

Microbiome composition variations in PLHIV and dTBs are influenced by CD4 count and ART. (A) Mycobacterium is depleted in people with a high CD4 count in all PLHIV and (B) dTBs. (C) PLHIV on ART were enriched in Streptococcus and depleted in Mycobacterium and Bifidobacterium. (D) dTBs on ART were depleted in Mycobacterium and Bifidobacterium whereas (red dotted line represents adjusted p-value=0.2; circle size represents relative abundance; taxa identified as possible contaminants are shown in grey). ART: antiretroviral therapy; dTB: definite tuberculous pericarditis; PLHIV: people living with HIV.

Fig. 4.. Associations between pericardial fluid microbiota and imaging, clinical characteristics and SCFA metabolic pathways.

Pulmonary infiltrates and pleural effusions are associated with specific taxa in the pericardial fluid microbiome. Compared to presumptive TBPs without pulmonary infiltrates (n=108), (A) those with a pulmonary infiltrate (n=31) were Mycobacterium-enriched and Lacticigenium- and Streptococcus-depleted, whilst (B) compared to presumptive TBPs without pleural effusions (n=51), those with pleural effusions (n=88) were Streptococcus- and Pseudomonas-enriched, and Brachymonas-depleted. Pericardial fluid in dTBs (n=91) are enriched in metabolic pathways associated with the production of SCFAs, aromatic compound degradation, and mycobacterial metabolism vs. nTBs (n=23). Red dotted line represents adjusted p-value=0.05; circle size represents relative abundance; pathways of interest are bolded. TBP: tuberculous pericarditis dTB: definite tuberculous pericarditis; nTB: non-tuberculous pericarditis; SCFA: short chain fatty acids.

Fig. 5.

Microbial composition differs in presumptive TBP people with high vs low CRP. A) CRP levels are similar in dTBs (n=72) vs. pTBs (n=20) vs. nTBs (n=20) using the Kruskal-Wallis test. B) Mycobacterial reads do not correlate with CRP levels; r_s represents the Spearman’s Rho coefficient for the correlation of mycobacterial reads with CRP. People with high CRP levels have similar C) Mycobacterium reads and D) α-diversity in their pericardial fluid compared to people with low CRP levels (Mann-Whitney test; solid circles: dTB; empty squares: pTB; empty circles: nTB). E) β-diversity differs between people with high vs low CRP. F) Pseudomonas is depleted in people with high vs low CRP. Discriminatory taxa appear above the threshold (red dotted line, FDR=0.2); circle size represents relative abundance. CI, Confidence interval; CRP, C-reactive protein; dTB: definite-TBP; FDR, false discovery rate; nTBP, non-TBP; PERMANOVA, permutational multivariate analysis of variance; pTB: probable-TBP; TBP: tuberculous pericarditis.