During the transition from a healthy state to cardiometabolic disease, patients become heavily medicated, which leads to an increasingly aberrant gut microbiome and serum metabolome, and complicates biomarker discovery1,2,3,4,5. Here, through integrated multi-omics analyses of 2,173 European residents from the MetaCardis cohort, we show that the explanatory power of drugs for the variability in both host and gut microbiome features exceeds that of disease. We quantify inferred effects of single medications, their combinations as well as additive effects, and show that the latter shift the metabolome and microbiome towards a healthier state, exemplified in synergistic reduction in serum atherogenic lipoproteins by statins combined with aspirin, or enrichment of intestinal Roseburia by diuretic agents combined with beta-blockers. Several antibiotics exhibit a quantitative relationship between the number of courses prescribed and progression towards a microbiome state that is associated with the severity of cardiometabolic disease. We also report a relationship between cardiometabolic drug dosage, improvement in clinical markers and microbiome composition, supporting direct drug effects. Taken together, our computational framework and resulting resources enable the disentanglement of the effects of drugs and disease on host and microbiome features in multimedicated individuals. Furthermore, the robust signatures identified using our framework provide new hypotheses for drug–host–microbiome interactions in cardiometabolic disease.
This is a preview of subscription content, access via your institution
Open Access articles citing this article.
Genome Medicine Open Access 05 January 2023
Nature Biotechnology Open Access 02 January 2023
Nature Aging Open Access 14 October 2022
Subscribe to Nature+
Get immediate online access to Nature and 55 other Nature journal
Subscribe to Journal
Get full journal access for 1 year
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Tax calculation will be finalised during checkout.
Get time limited or full article access on ReadCube.
All prices are NET prices.
The source data for the figures are provided at Zenodo (https://doi.org/10.5281/zenodo.4728981). Raw shotgun sequencing data that support the findings of this study have been deposited at the ENA under accession codes PRJEB41311, PRJEB38742 and PRJEB37249 with public access. Raw spectra for metabolomics have been deposited in the MassIVE database under the accession codes MSV000088043 (UPLC–MS/MS) and MSV000088042 (GC–MS). The metadata on disease groups and drug intake are provided in Supplementary Tables 1–3. The demographic, clinical and phenotype metadata, and processed microbiome and metabolome data for French, German and Danish participants are available at Zenodo (https://doi.org/10.5281/zenodo.4674360).
The new drug-aware univariate biomarker testing pipeline is available as an R package (metadeconfoundR; Birkner et al., manuscript in preparation) at Github (https://github.com/TillBirkner/metadeconfoundR) and at Zenodo (https://doi.org/10.5281/zenodo.4721078). The latest version (0.1.8) of this package was used to generate the data shown in this publication. The code used for multivariate analysis based on the VpThemAll package is available at Zenodo (https://doi.org/10.5281/zenodo.4719526). The phenotype and drug intake metadata, processed microbiome, and metabolome data and code resources are available for download at Zenodo (https://doi.org/10.5281/zenodo.4674360). The code for reproducing the figures is provided at Zenodo (https://doi.org/10.5281/zenodo.4728981).
Sinha, R. et al. Assessment of variation in microbial community amplicon sequencing by the Microbiome Quality Control (MBQC) project consortium. Nat. Biotechnol. 35, 1077–1086 (2017).
Costea, P. I. et al. Towards standards for human fecal sample processing in metagenomic studies. Nat. Biotechnol. 35, 1069–1076 (2017).
Rothschild, D. et al. Environment dominates over host genetics in shaping human gut microbiota. Nature 555, 210–215 (2018).
Schmidt, T. S. B., Raes, J. & Bork, P. The human gut microbiome: from association to modulation. Cell 172, 1198–1215 (2018).
Vujkovic-Cvijin, I. et al. Host variables confound gut microbiota studies of human disease. Nature 587, 448–454 (2020).
Tsuda, A. et al. Influence of proton-pump inhibitors on the luminal microbiota in the gastrointestinal tract. Clin. Transl. Gastroenterol. 6, e89 (2015).
Forslund, K. et al. Disentangling type 2 diabetes and metformin treatment signatures in the human gut microbiota. Nature 528, 262–266 (2015).
Maier, L. et al. Extensive impact of non-antibiotic drugs on human gut bacteria. Nature 555, 623–628 (2018).
Le Bastard, Q. et al. Systematic review: human gut dysbiosis induced by non-antibiotic prescription medications. Aliment. Pharmacol. Ther. 47, 332–345 (2018).
Vich Vila, A. et al. Impact of commonly used drugs on the composition and metabolic function of the gut microbiota. Nat. Commun. 11, 362 (2020).
Vieira-Silva, S. et al. Statin therapy is associated with lower prevalence of gut microbiota dysbiosis. Nature 581, 310–315 (2020).
Falony, G. et al. Population-level analysis of gut microbiome variation. Science 352, 560–564 (2016).
Jackson, M. A. et al. Gut microbiota associations with common diseases and prescription medications in a population-based cohort. Nat. Commun. 9, 2655 (2018).
Conlon, M. A. & Bird, A. R. The impact of diet and lifestyle on gut microbiota and human health. Nutrients 7, 17–44 (2014).
Blaser, M. J. Antibiotic use and its consequences for the normal microbiome. Science 352, 544–545 (2016).
Imhann, F. et al. Proton pump inhibitors affect the gut microbiome. Gut 65, 740–748 (2016).
Imhann, F. et al. The influence of proton pump inhibitors and other commonly used medication on the gut microbiota. Gut Microbes 8, 351–358 (2017).
Wu, H. et al. Metformin alters the gut microbiome of individuals with treatment-naive type 2 diabetes, contributing to the therapeutic effects of the drug. Nat. Med. 23, 850–858 (2017).
Koeth, R. A. et al. γ-Butyrobetaine is a proatherogenic intermediate in gut microbial metabolism of l-carnitine to TMAO. Cell Metab. 20, 799–812 (2014).
Kopp, E. & Ghosh, S. Inhibition of NF-κB by sodium salicylate and aspirin. Science 265, 956–959 (1994).
Davalli, A. M., Perego, C. & Folli, F. B. The potential role of glutamate in the current diabetes epidemic. Acta Diabetol. 49, 167–183 (2012).
Schmidt, T. S. et al. Extensive transmission of microbes along the gastrointestinal tract. eLife 8, e42693 (2019).
Shimazu, T. et al. Suppression of oxidative stress by β-hydroxybutyrate, an endogenous histone deacetylase inhibitor. Science 339, 211–214 (2012).
Shah, B. R. & Hux, J. E. Quantifying the risk of infectious diseases for people with diabetes. Diabetes Care 26, 510–513 (2003).
Korpela, K. & Vos, W. M. de. Antibiotic use in childhood alters the gut microbiota and predisposes to overweight. Microbial Cell 3, 296–298 (2016).
Le Chatelier, E. et al. Richness of human gut microbiome correlates with metabolic markers. Nature 500, 541–546 (2013).
Arumugam, M. et al. Enterotypes of the human gut microbiome. Nature 473, 174–180 (2011).
Costea, P. I. et al. Enterotypes in the landscape of gut microbial community composition. Nat. Microbiol. 3, 8–16 (2018).
Cameron, A. R. et al. Anti-inflammatory effects of metformin irrespective of diabetes status. Circ. Res. 119, 652–665 (2016).
Dandona, P. et al. Increased plasma concentration of macrophage migration inhibitory factor (MIF) and MIF mRNA in mononuclear cells in the obese and the suppressive action of metformin. J. Clin. Endocrinol. Metab. 89, 5043–5047 (2004).
Branca, F., Nikogosian, H. & Lobstein, T. The Challenge of Obesity in the WHO European Region and the Strategies for Response. Summary, https://www.euro.who.int/en/publications/abstracts/challenge-of-obesity-in-the-who-european-region-and-the-strategies-for-response-the.-summary (WHO, 2007).
Alberti, K. G. M. M., Zimmet, P. & Shaw, J. The metabolic syndrome–a new worldwide definition. Lancet 366, 1059–1062 (2005).
Petersmann, A. et al. Definition, classification and diagnosis of diabetes mellitus. Exp. Clin. Endocrinol. Diabetes 127, S1-S7 (2019).
Whelton, P. K. et al. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on clinical practice guidelines. Hypertension 71, 1269–1324 (2018).
Yancy, C. W. et al. 2017 ACC/AHA/HFSA focused update of the 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology/American Heart Association Task Force on clinical practice guidelines and the Heart Failure Society of America. J. Card. Fail. 23, 628–651 (2017).
Verger, E. O. et al. Dietary assessment in the MetaCardis study: development and relative validity of an online food frequency questionnaire. J. Acad. Nutr. Diet. 117, 878–888 (2017).
Criscuolo, A. & Brisse, S. AlienTrimmer: a tool to quickly and accurately trim off multiple short contaminant sequences from high-throughput sequencing reads. Genomics 102, 500–506 (2013).
Li, J. et al. An integrated catalog of reference genes in the human gut microbiome. Nat. Biotechnol. 32, 834–841 (2014).
Cotillard, A. et al. Dietary intervention impact on gut microbial gene richness. Nature 500, 585–588 (2013).
Prifti, E. & Le Chatelier, E. MetaOMineR. A quantitative metagenomics data analyses pipeline v.hal-02800484 (2014).
Nielsen, H. B. et al. Identification and assembly of genomes and genetic elements in complex metagenomic samples without using reference genomes. Nat. Biotechnol. 32, 822–828 (2014).
Sunagawa, S. et al. Metagenomic species profiling using universal phylogenetic marker genes. Nat. Methods 10, 1196–1199 (2013).
Kultima, J. R. et al. MOCAT2: a metagenomic assembly, annotation and profiling framework. Bioinformatics 32, 2520–2523 (2016).
Coelho, L. P. et al. NG-meta-profiler: fast processing of metagenomes using NGLess, a domain-specific language. Microbiome 7, 84 (2019).
Prest, E. I., Hammes, F., Kötzsch, S., van Loosdrecht, M. C. M. & Vrouwenvelder, J. S. Monitoring microbiological changes in drinking water systems using a fast and reproducible flow cytometric method. Water Res. 47, 7131–7142 (2013).
Vandeputte, D. et al. Quantitative microbiome profiling links gut community variation to microbial load. Nature 551, 507–511 (2017).
Katz, M. H. Multivariable Analysis. A Practical Guide for Clinicians and Public Health Researchers (Cambridge Univ. Press, 2011).
Costea, P. I. et al. metaSNV: a tool for metagenomic strain level analysis. PLoS ONE 12, e0182392 (2017).
Hahsler, M., Gruen, B., Hornik, K. arules–a computational environment for mining association rules and frequent item sets. J. Stat. Softw. 14, 1–25. (2005).
Imai, K., Keele, L. & Tingley, D. A general approach to causal mediation analysis. Psychol. Methods 15, 309–334 (2010).
Seabold, S. & Perktold J. statsmodels: econometric and statistical modeling with Python. In Proc. 9th Python in Science Conference (2010).
Palleja, A. et al. Recovery of gut microbiota of healthy adults following antibiotic exposure. Nat. Microbiol. 3, 1255–1265 (2018).
We thank the MetaCardis participants for their participation in the study, and particularly the patient associations (Alliance du Coeur and CNAO) for their input and interface; D. Bonnefont-Rousselot (Department of Metabolic Biochemistry, Pitié-Salpêtrière Hospital) for the analysis of plasma lipid profiles; and the nurses, technicians, clinical research assistants and data managers from the Clinical Investigation Platform at the Institute of Cardiometabolism and Nutrition for patient investigations, the Clinical Investigation Center (CIC) from Pitié-Salpêtrière Hospital and Human Research Center on Nutrition (CRNH Ile-de-France) as well as the university hospital of Leipzig for the investigation of healthy control individuals. Quanta Medical provided regulatory oversight of the clinical study and contributed to the processing and management of electronic data. This work was supported by the European Union’s Seventh Framework Program for research, technological development and demonstration under grant agreement HEALTH-F4-2012-305312 (METACARDIS). Part of this work was also supported by the EMBL, by the Metagenopolis grant ANR-11-DPBS-0001, by the H2020 European Research Council (ERC-AdG-669830) (to P.B.), and by grants from the Deutsche Forschungsgemeinschaft (SFB1365 to S.K.F. and L.M.; and SFB1052/3 A1 MS to M.S. (209933838)). Assistance Publique-Hôpitaux de Paris is the promoter of the clinical investigation (MetaCardis). M.-E.D. is supported by the NIHR Imperial Biomedical Research Centre and by grants from the French National Research Agency (ANR-10-LABX-46 (European Genomics Institute for Diabetes)), from the National Center for Precision Diabetic Medicine – PreciDIAB, which is jointly supported by the French National Agency for Research (ANR-18-IBHU-0001), by the European Union (FEDER), by the Hauts-de-France Regional Council (Agreement 20001891/NP0025517) and by the European Metropolis of Lille (MEL, Agreement 2019_ESR_11) and by Isite ULNE (R-002-20-TALENT-DUMAS), also jointly funded by ANR (ANR-16-IDEX-0004-ULNE), the Hauts-de-France Regional Council (20002845) and by the European Metropolis of Lille (MEL). R.J.A. is a member of the Collaboration for joint PhD degree between EMBL and Heidelberg University, Faculty of Bioscience. The Novo Nordisk Foundation Center for Basic Metabolic Research is an independent research institution at the University of Copenhagen partially funded by an unrestricted donation from the Novo Nordisk Foundation.
F.B. is shareholder in Implexion Pharma AB. K.C. is a consultant for Danone Research, LNC therapeutics and CONFO therapeutics for work that is unassociated with the present study. K.C. has held a collaborative research contract with Danone Research in the context of MetaCardis project. M.B. received lecture and/or consultancy fees from AstraZeneca, Boehringer-Ingelheim, Lilly, Novo Nordisk, Novartis and Sanofi. The other authors declare no competing interests.
Peer review information Nature thanks Peter Turnbaugh and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
Extended Data Fig. 1 A post-hoc testing approach for deconfounding univariate biomarker analysis for multiple medications and risk factors.
The schematic highlights our covariate control approach. All significant associations between putative drivers (e.g., disease D) and covariates (C1...Cn) to each measured feature (Y1...Ym) are taken. The outcome of the test is denoted with ai for a positive outcome (“yes”) and āi for a negative outcome (“no”). A significant predictor is called “confounded” and is filtered out in a post-hoc test if there is at least one covariate (e.g., drug treatment or combination) such that the predictor does not add significant predictive capacity beyond the covariate (“confounded”). If no such covariate itself passes the same test (i.e., covariates cannot in turn be shown to have predictive capacity beyond tested predictor), the predictor is considered ambiguous (“ambiguously deconfounded”). Otherwise, the predictor is considered “confidently deconfounded” (we note that “confidently deconfounded” is defined as no confounders were found among all covariates measured in our study).
Extended Data Fig. 2 Previously reported metabolic disease associations are replicated in the MetaCardis cohort under drug deconfounding, highlighting systemic inflammation, short-chain fatty acid and branched-chain amino acid mechanisms underlying insulin resistance.
Cuneiform plot marker hues and direction show sign of effect size (Cliff’s delta), intensity and size show amplitude of effect size, comparing metabolic diseased proband subsets (horizontal axis) with healthy control subject in the MetaCardis population for different microbiome, metabolome and host features (vertical axis). Bold and opaque markers show significant associations (two-sided MWU FDR < 0.1) not reducible to any significant drug or demographic confounder. Full associations are found in Supplementary Table 9; here a preselected subset is displayed reflecting previously reported risk and protective factors, validated in MetaCardis. 1H NMR features are shown with retention time in parentheses, functional modules with GMM or KEGG identifier in parenthesis, analogous for metagenomic species and mOTUs.
Extended Data Fig. 3 Previously reported drug-microbiome associations are replicated in the MetaCardis cohort for metformin and PPI.
Bar plots show the magnitude and direction of effect size (Cliff’s delta) of metformin treatment (left) and PPI treatment (right) on microbiome features. These effects are compared to the previously published data from two independent patient cohorts10. Only features with direct match on the taxonomic level were included in the comparison. Full list of associations is provided in Supplementary Table 6.
Cuneiform plot shows change in abundance of bacterial species in the gut in subjects taking/not taking PPIs (controlling for other drugs and demographic factors) in each clinical group separately, and for all subjects pooled together. Rows marked “SNV” show whether oral strain single nucleotide markers are significantly (two-sided MWU FDR < 0.1) enriched over gut strain markers in subjects taking PPIs, controlling for abundance of each species. Marker direction, colour and size denote the sign and value of Cliff’s delta standardized effect size; opaque markers are significantly altered (two-sided MWU FDR < 0.1; passing all confounder checks). Bacteria are shown if their abundance is significantly altered under PPI consumption, and there are SNPs distinguishing oral from gut strains in HMP samples. (See Supplementary Tables 5–7).
Extended Data Fig. 5 Breakdown of antibiotics association into individual features, selected features shown.
Left cuneiform plot (markers show Spearman correlation direction by shape and colour, scope by size and colour, significance (two-sided MWU FDR < 0.1, deconfounded for other drug and demographic features) by edge opacity) shows association between each feature and total number of antibiotics courses in CMD groups as well as in healthy controls. Right cuneiform shows whether the same features are significantly different (two-sided MWU FDR < 0.1) between healthy controls and CMD subjects following drug deconfounding (markers show Cliff’s delta effect size), requiring significant and deconfounded correlation with number of antibiotic courses demonstrable in at least one proband group and at least one group showing significant and deconfounded alteration compared to healthy controls. Core features include increased carriage of possible disease-associated Ruminococcus gnavus and various Clostridia species, alongside decreased carriage of commensals such as Faecalibacterium species. Full list of associations is provided in Supplementary Table 12.
For bacterial species where an effect on abundance of total antibiotics courses in MetaCardis could be demonstrated (significant at Spearman FDR < 0.1 and deconfounded), and where effect of antibiotic intervention has been tested in a recent antibiotic intervention study52, MetaCardis correlation on vertical axis vs intervention log-transformed fold change on horizontal axis are displayed. Separate markers are shown for each MetaCardis patient group within which antibiotic effect can be demonstrated. Bold markers achieve significance (FDR < 0.1) in the intervention study as well. For the majority of taxa overlapping between studies, direction of changes matches, consistent with a causal impact of antibiotics on the microbiota in MetaCardis.
Cuneiform shows normalized regression coefficients of logistic models for each 4-class enterotype as a function of antibiotics courses in the past 5 years, separately for controls and metabolic disease patient groups. All significant (two-sided Wald FDR < 0.1) models show depletion of Ruminococcus and Prevotella enterotypes, and enrichment for Bacteroides enterotypes; in the case of metabolic disease patients, this is strongest for the low cell count Bacteroides 2 enterotype.
Extended Data Fig. 8 Illustration of flow cytometry gating strategy. A fixed gating/staining approach was applied45.
Both blank and sample solutions were stained with SYBR Green I. a. FL1-A/FL3-A acquisition plot of a blank sample (0.85% w/v physiological solution) with gate boundaries indicated. A threshold value of 2000 was applied on the FL1 channel. b. Secondary gating was performed on the FSC-A/SSC-A channels to further discriminate between debris/background and microbial events. c, d./ FL1-A/FL3-A count acquisition of a faecal sample with secondary gating on FSC-A/SSC-A channels based on blank analyses. Total counts were defined as events registered in the FL1-A/FL3-A gating area excluding debris/background events observed in the FSC-A/SSC-A R1 gate. The flow rate was set at 14 microliters per minute and the acquisition rate did not exceed 10,000 events per second. Each panel reflects the events registered during a 30 s acquisition period. Cell counts were determined in duplicate starting from a single biological sample.
Supplementary Text (including the Supplementary Results) and Supplementary References.
Supplementary Tables 1–4 include information on the cohort characteristics and drug intake with source metadata. Source metadata for the MetaCardis cohort include disease group, gender, age, study centre, BMI (kg m−2), alternative healthy eating index, diet diversity score and dietary approaches to stop hypertension, physical activity and manual work levels, and smoking status. Source data include information on the drug intake, drug combinations, dosage and antibiotic use analysed in the MetaCardis cohort.
Multivariate breakdown of variance. The table includes multivariate breakdown of variance for each feature space by each predictor, a summary of all interaction terms in the models and the results of the confounder analysis for all patient groups analysed in the MetaCardis cohort.
Features of microbiome, host and metabolome impacted by different drug groups and drug compounds. Results of drug group (or drug compound according to the ATC classification) assessment for its impact on host and microbiome features for each patient group. The compound comparison with ref. 8 tab shows microbiome features that are negatively impacted by the drug treatment (for the ATC-level compounds) in at least one patient group, and bacterial species of which the growth was inhibited by the same drug in the in vitro experiment.
The number of microbiome and metabolome features impacted by different drug groups. Summary statistics of features impacted by different drugs, separated by host and microbiome features, and defined as either drug effects (drug effect discordant with the disease effect) or severity markers (drug effect concordant with the disease effect).
Features of the microbiome, host and metabolome impacted by different drug combinations. Analysis of the effect of drug combinations, assessed for impact on host and microbiome falling within different measurement categories in each patient group.
Features of microbiome, host and metabolome impacted by patient group/clinical indication. Analysis of the patient groups, contrasted against healthy control individuals, assessed for impact on host and microbiome features falling within different measurement categories.
Mediation analysis of host and microbiome features for drug intake, dosage and combinations. Mediation analysis using a regression model of drug effect on each host feature mediated through a microbiome feature or vice versa.
Driver analysis of antibiotic effects on the gut microbiota. Results of the principal coordinate analysis of microbiome composition (cell-count-adjusted mOTU abundance at the species level) performed on antibiotics-naive individuals with T2D as well as on healthy individuals.
Features of the microbiome, host and metabolome impacted by the number of antibiotic courses during the study period. The results of additive antibiotic exposure effect assessment for its impact on host and microbiome features for each patient group.
Features of the microbiome, host and metabolome impacted by different drug dosages. The results of drug dosage assessment for its impact on host and microbiome features falling within different measurement categories for each patient group. Both dosage-confirmed (the effect was identified both from drug intake status and relative drug dosage analysis) and dosage-unique (the effect was revealed only by relative dosage analysis) effects are shown.
Significant impacts on enterotype distribution based on disease status and medication variables. The table includes clinical status (patient versus healthy control comparisons), CMD or antibiotic drug status or dosage, or intake of drug combinations, shown for each enterotype versus the other three enterotypes in the four-enterotype classification.
About this article
Cite this article
Forslund, S.K., Chakaroun, R., Zimmermann-Kogadeeva, M. et al. Combinatorial, additive and dose-dependent drug–microbiome associations. Nature 600, 500–505 (2021). https://doi.org/10.1038/s41586-021-04177-9
This article is cited by
Genome Medicine (2023)
Nature Reviews Nephrology (2023)
Current Atherosclerosis Reports (2023)
Nature Biotechnology (2023)
Integrating and formatting biomedical data as pre-calculated knowledge graph embeddings in the Bioteque
Nature Communications (2022)