Inconsistent associations of gut microbiota with cardiometabolic diseases by using paired fecal and blood metabolomics data

Published in Microbiology

Gut microbiome is regarded as the second organ or brain of humans, as it plays essential roles in host physiology and pathology through producing a myriad of metabolites1. Gut microbiota-derived metabolites serve as the intermediates in the cross-talk between gut microbiome and host health, and are potential actionable microbial targets for cardiometabolic diseases1.


Both blood and fecal samples are used in human studies to profile the gut microbial metabolites. Particularly, blood is commonly collected in large-scale epidemiological studies, which facilitates the profiling of microbial metabolites and the investigation of their associations with cardiometabolic diseases, although those microbial metabolites identified primarily based on animal-based studies may not necessarily be derived by corresponding human microbes. Nevertheless, human studies showed that gut microbiota composition was largely reflected by fecal metabolome (average explained variance: 67.7%)2, and recent studies demonstrated significant contributions of gut microbiota to blood metabolome (average explained variance: 4.6%~15%)3, 4, 5, 6. Given the markedly different contributions of gut microbiota to fecal and blood metabolome shown in previous studies, we are curious about whether the same metabolites in blood and feces have different associations with gut microbiota and cardiometabolic diseases.


Leveraging the large-scale multi-omics data in Guangzhou Nutrition and Health Study (GNHS) with matched fecal metagenomic (149 species and 214 pathways) and paired fecal and blood targeted metabolome data (132 paired metabolites), we systematically compared paired fecal and blood metabolites in their associations with gut microbiota (including taxonomic composition and microbial pathways) and cardiometabolic diseases (including type 2 diabetes (T2D), obesity, nonalcoholic fatty liver disease (NAFLD), and hypertension). Surprisingly, we found that paired fecal and blood metabolites had generally low correlations in terms of phenotypic correlations (a direct comparison of each metabolite’s values between feces and blood; correlation coefficients [mean±sd]: 0.05±0.12) and genetic correlations (the proportion of shared heritability between paired fecal and blood metabolites; correlation coefficients [mean±sd]: 0.13±0.75).


Using the machine learning pipeline to estimate the associations between taxonomic composition/microbial pathways and fecal/blood metabolites, we further found that taxonomic composition and microbial pathways were more broadly associated with fecal metabolites compared with blood metabolites. We defined well-predicted metabolites based on correlation coefficient > 0.3 and FDR < 0.05, and identified 98 well-predicted fecal metabolites and 10 well-predicted blood metabolites based on taxonomic composition. We found that most well-predicted fecal metabolites had superior predictability over their corresponding paired blood metabolites (FDR < 0.05). Interestingly for metabolites that were well-predicted in both feces and blood (8 metabolites), the models of most well-predicted fecal metabolites significantly outperformed their respective models of well-predicted blood metabolites (FDR < 0.05). The results based on microbial pathways showed consistent results. Most of our identified gut microbiota-metabolite associations were replicated in an independent validation cohort involving 103 participants.


We then explored the associations of well-predicted fecal and blood metabolites with cardiometabolic diseases, and found 12 significant associations between well-predicted fecal metabolites and cardiometabolic diseases (7 associations for T2D, 4 associations for obesity and 1 association for NAFLD). However, we did not find any significant association for the well-predicted blood metabolites.


Taken together, our study showed disparate associations with gut microbiota and cardiometabolic diseases when using fecal or blood metabolites, which suggested that sampling criteria may be a relevant factor in metabolomics-based association studies, and caution should be taken when inferring microbiome-cardiometabolic disease associations from either blood or fecal metabolome data in epidemiological studies.



  1. Krautkramer KA, Fan J, Bäckhed F. Gut microbial metabolites as multi-kingdom intermediates. Nat Rev Microbiol 19, 77-94 (2021).


  1. Zierer J, et al. The fecal metabolome as a functional readout of the gut microbiome. Nat Genet 50, 790-795 (2018).


  1. Bar N, et al. A reference map of potential determinants for the human serum metabolome. Nature 588, 135-140 (2020).


  1. Chen L, et al. Influence of the microbiome, diet and genetics on inter-individual variation in the human plasma metabolome. Nat Med, (2022).


  1. Dekkers KF, et al. An online atlas of human plasma metabolite signatures of gut microbiome composition. Nat Commun 13, 5370 (2022).


  1. Diener C, et al. Genome-microbiome interplay provides insight into the determinants of the human blood metabolome. Nat Metab, (2022).


Please sign in or register for FREE

If you are a registered user on Research Communities by Springer Nature, please sign in

Subscribe to the Topic

Life Sciences > Biological Sciences > Microbiology

Related Collections

With collections, you can get published faster and increase your visibility.

Applied Sciences

This collection highlights research and commentary in applied science. The range of topics is large, spanning all scientific disciplines, with the unifying factor being the goal to turn scientific knowledge into positive benefits for society.

Publishing Model: Open Access

Deadline: Ongoing