Pharmacomicrobiomics

Abstract

Oral medications encounter gut commensal microbes that participate directly and indirectly in drug effects through metabolism, interactions with drug metabolites, or production of substrates that compete with drugs for drug-metabolizing enzymes, consequently influencing drug pharmacokinetics. The microbiota can also affect drug efficacy or toxicity by modulating the immune system; for example, variability in response to cancer immunotherapy, such as anti-PD-1 and anti-CTLA-4 therapies, is increasingly attributed to differences in gut microbial composition and function. These conditions indicate the need and opportunity to intentionally leverage the microbiome for drug effect; as such, the study of how intra- and inter-individual differences in the microbiome affect drug response has gained a definition termed pharmacomicrobiomics. While the need is clear, clinical studies evaluating pharmacomicrobiomic interactions are challenging due to microbiome variability, multiple potential confounders, no standardization of statistical and bioinformatics methods, and the reluctance of potential clinical study participants. In this review, we make the case for pharmacomicrobiomic clinical studies; for the use of modeling and simulation to provide a quantitative framework for data integration, hypothesis testing, and translational-to-late-stage clinical predictions; and the application of real-world data to support both using a within-subject comparison approach. We argue that an integrated and cohesive approach can address the large "inherent" inter-individual variability in the microbiome, attributed to factors such as age, lifestyle choices, environmental factors, chemical and biological exposures, and disease. In summary, there are many challenges to pharmacomicrobiomics research but also enormous potential to improve the development and utilization of pharmaceutical products.

Figure 1

Pharmacomicrobiomics: The study of how intra‐ and inter‐individual differences in the microbiome affect drug response. Individual variability and environmental factors such as season, diet, chemical or biological exposures, and geographic location shape the microbiome composition. The influence of host physiological processes on microbiome composition and functions is also becoming evident. In turn, the understanding of the microbiome’s role in physiological and pathological processes is consistently being expanded. Intriguingly, the microbiome affects drugs’ bioavailability, effectiveness, and toxicity, and therefore, is a potentially promising target for pharmacological modulation. An integrated approach including clinical studies, modeling and simulation, and real‐world studies will address some of the challenges and opportunities to leverage pharmacomicrobiomics research for better care. MIDD, model‐informed drug development; RWD, real‐world data.

Figure 2

PBPK simulation results of (a) sulfinpyrazone and (b) sulfide metabolite after a 200 mg tablet administration (solid line: simulated; square: observed). (c) Simulated % percent of dose excreted in urine and cleared by gut microbiome.

Figure 3

Example sample and data collection timeline for participants in a simple drug exposure study, including a pre‐ and post‐drug treatment periods. The study is designed to test for gut microbiome effects on short‐term drug levels and/or drug response. The timeline also illustrates the large number of potential influences on the microbiome, including dietary habits, updated medical history, and concomitant drug treatment. Note that the example study includes longitudinal repeat fecal sampling to account for intra‐individual variation and also to a pre‐ and post‐drug treatment phases to characterize an individual’s microbiome in the absence of drug treatment. While some aspects need only be tested once during the study, such as germline human genomic sampling and basic demographic information, most aspects should be repeatedly sampled to assess potential confounding effects on the microbiome. [Correction added on 6 October 2025, after first online publication: Figure 3 has been replaced in this version.]