Causal association between gut microbiota and intrahepatic cholestasis of pregnancy: mendelian randomization study.

BACKGROUND

Previous observational cohort studies have shown that the composition of the gut microbiota is related to the risk of intrahepatic cholestasis of pregnancy (ICP), although it is unclear if the association is causative. This study used Mendelian randomization (MR) to systematically examine whether the gut microbiota was causally linked to ICP.

METHODS

We obtained the genome-wide association study (GWAS) summary statistics of gut microbiota and ICP from published GWASs. Maximum likelihood (ML), MR-Egger regression, weighted median, inverse variance weighted (IVW), and weighted model were used to investigate the causal association between gut microbiota and ICP. We further conducted a series of sensitivity analyses to confirm the robustness of the primary results of the MR analyses. Reverse MR analysis was performed on the bacterial taxa that were reported to be causally linked to ICP risk in forwarding MR analysis to evaluate the possibility of reverse causation.

RESULTS

MR analysis revealed that phylum Tenericutes (OR: 1.670, 95%CI: 1.073-2.598, P = 0.023), class Bacteroidia (OR: 1.644, 95%CI: 1.031-2.622, P = 0.037), class Mollicutes (OR: 1.670, 95%CI: 1.073-2.598, P = 0.023), and order Bacteroidales (OR: 1.644, 95%CI: 1.031-2.622, P = 0.037), and were positively associated with the risk of ICP. And we identified that the relative abundance of genus Dialister (OR: 0.562, 95%CI: 0.323-0.977, P = 0.041), genus Erysipelatoclostridium (OR: 0.695, 95%CI: 0.490-0.987, P = 0.042), genus Eubacterium (brachy group) (OR: 0.661, 95%CI: 0.497-0.880, P = 0.005), genus Eubacterium (hallii group) (OR: 0.664, 95%CI: 0.451-0.977, P = 0.037), genus Holdemania (OR: 0.590, 95%CI: 0.414-0.840, P = 0.003), genus Ruminococcus (torques group) (OR: 0.448, 95%CI: 0.235-0.854, P = 0.015), and genus Veillonella (OR: 0.513, 95%CI: 0.294-0.893, P = 0.018) were related to a lower risk of ICP. Additional sensitivity analyses confirmed the robustness of the association between specific gut microbiota composition and ICP. No evidence of reverse causality from ICP to identified bacterial taxa was found in the findings of the reverse MR analyses.

CONCLUSIONS

Under MR assumptions, our findings propose new evidence of the relationship between gut microbiota and ICP risk. Our results show that the gut microbiota may be useful target of intervention for ICP.