Evidence indicates that commonly prescribed medications can modify the gut microbiome in the short term, but the magnitude and persistence of long-term effects are less well characterized. Understanding how drugs shape microbial communities is important for gut-related diseases such as colorectal cancer (CRC) and for the development and interpretation of microbiome-based CRC screening biomarkers.
The study aimed to examine associations between prescription drug use, gut microbial composition, and screening-detected colorectal lesions. A secondary aim was to determine whether exposure to specific drugs and the timing of that exposure influence microbial candidate biomarkers for CRC screening.
Investigators analyzed a cohort of 1,034 fecal immunochemical test (FIT)-positive screening participants aged 55–77 years. Gut metagenomic profiles were combined with clinicopathologic, demographic, and lifestyle data. Prescription histories covering 12 years were obtained from the national prescription registry and used to classify current, past, and chronic drug use. The team tested associations for 83 drugs at the ATC5 level against microbial diversity metrics and individual bacterial species.
Of the 83 drugs tested, 38 showed associations with microbial diversity measures and with specific bacterial species. Across all species profiled, 168 species (32.9% of detected species) were associated with at least one drug. These results indicate a substantial proportion of the gut microbiome—nearly one third of species—varied in relation to prescription drug exposures recorded in the registry.
Among the 168 drug-associated species, seven species have earlier been associated with colorectal neoplasia in the same cohort. This overlap suggests that medication use could influence microbial signals used as CRC biomarkers, potentially confounding biomarker performance or interpretation if drug exposure is not accounted for in biomarker development and validation.
The authors highlight six putative CRC biomarker species that showed positive associations with particular medications:
Escherichia coli was positively associated with metformin use.
Clostridium symbiosum was positively associated with a codeine–paracetamol combination (a commonly used painkiller formulation).
Eggerthella lenta was positively associated with clindamycin exposure (an antibiotic).
Flavonifractor plautii was positively associated with dicloxacillin, clindamycin, and the codeine–paracetamol combination.
Streptococcus parasanguinis and Streptococcus salivarius were positively associated with pantoprazole and esomeprazole, respectively; both are proton pump inhibitors (PPIs).
In addition, an unclassified Actinobacteria species (GGB34797 SGB14322) was negatively associated with clindamycin exposure.
These drug–microbe links indicate that commonly used drug classes—painkillers, metformin, antibiotics, and proton pump inhibitors—induce measurable shifts in microbial composition that overlap with species implicated in colorectal neoplasia.
DNA sequencing data from the CRCbiome study are deposited in Federated EGA under accession EGAS50000000170; the submitted FASTQ files exclude human-mapping reads in accordance with participant consent. R scripts used for analysis are available at the project’s GitHub repository (https://github.com/Rounge-lab/DrugBugCRC). The BCSN and CRCbiome studies received approval from the Regional Committee for Medical Research Ethics in South-East Norway (approval numbers reported in the source). The BCSN is also registered at ClinicalTrials.gov (NCT01538550). The authors report that participant consent procedures and data protection requirements (including GDPR constraints) were followed.
In this FIT-positive screening cohort of 1,034 participants, multiple commonly prescribed drugs were associated with alterations in gut microbiome diversity and with numerous bacterial species. Notably, several drug-associated species coincide with microbial candidates previously linked to colorectal neoplasia, suggesting medication use may influence microbiome-based CRC biomarker signals. The authors conclude that widely used medications such as metformin, antibiotics, painkillers, and proton pump inhibitors produce measurable microbiome changes that could affect CRC screening biomarker candidates.
Further details on effect sizes, temporal patterns of exposure, and the potential implications for biomarker specificity were not reported in the provided source text.