Nonconcussive head impacts sustained during American football were associated with short-term and seasonal changes in gut microbiome composition in a small prospective cohort study.
The authors described the findings as the first reported evidence linking nonconcussive head impacts—rather than diagnosed concussion or mild traumatic brain injury (TBI) specifically—with gut microbiome alterations. However, the study included only six analyzed athletes, lacked a control group, and reported statistical power below 3% for mixed-effects modeling, limiting confidence in the findings and precluding causal inference.
Researchers at Colgate University monitored head acceleration events, fecal microbiome composition, physical activity, and clinical and behavioral variables across a football season in NCAA Division I athletes. Of 19 enrolled participants, 6 were ultimately included in the final analysis after 3 exclusions for incomplete head-impact data and 10 for insufficient fecal sampling. From those 6 participants, 20 additional fecal samples were then excluded for antibiotic exposure or duplicate samples.
The final cohort consisted entirely of male participants who were White or Caucasian and aged 21 or 22 years. Investigators analyzed 226 fecal samples using 16S rRNA sequencing.
Head impacts were measured using the Riddell InSite helmet-based monitoring system. Investigators derived a cumulative "impact load" score from grouped helmet-force categories ranging from 15 to 19 g to 63 g or greater.
Acute Microbiome Associations
To examine short-term associations, investigators identified 13 four-day windows across the 6 participants in which a participant sustained an impact load score at or above the study's 75th-percentile threshold—a score of 70 or greater—followed by 3 days without additional impacts at that threshold. The authors noted that this cutoff was selected for analytical tractability rather than physiologic validation.
In this subset analysis, Bray-Curtis dissimilarity—a measure of how much microbial composition differed from each participant's preseason baseline—was significantly higher at 48 to 72 hours and 72 to 96 hours after substantial impact exposure compared with the 0 to 24-hour period.
The investigators separately performed mixed-effects linear modeling across the broader dataset while adjusting for 15 measured clinical and behavioral covariates, including sleep, stress, illness, medication use, caffeine intake, and physical activity.
In those models, impact load during the preceding 48 to 72 hours was nominally associated with increased Bray-Curtis dissimilarity, but the association did not remain significant after Benjamini-Hochberg correction for multiple comparisons. Time across the season, player load, and pre-workout energy drink consumption did remain significant after correction.
Taxa-Level Findings
Taxon-level analyses found that higher prior impact load was associated with lower relative abundances of Coriobacteriales, Prevotellaceae, and Prevotella and higher relative abundances of Ruminococcus and, marginally, Verrucomicrobiales.
After correction for multiple testing, only the Prevotellaceae and Prevotella associations remained significant.
Sensitivity analyses using centered log-ratio transformation showed directional consistency for several findings, although some associations weakened, suggesting susceptibility to compositional bias.
The clinical implications of these microbiome shifts remain unclear. However, the authors noted that prior TBI literature has linked reductions in Prevotellaceae to decreased butyrate production, impaired blood-brain barrier restoration, and pro-inflammatory signaling. Increases in Ruminococcus have also previously been associated with inflammatory metabolite production in gastrointestinal disease states. The authors emphasized that these mechanistic interpretations remain speculative and that microbiome responses likely vary substantially between individuals.
Seasonal Changes and Potential Confounders
Investigators also evaluated microbiome composition longitudinally using the first three, middle three, and last three available fecal samples from each participant.
Bray-Curtis dissimilarity increased significantly from early to late in the collection period, while Faith's Phylogenetic Diversity did not significantly change.
The authors cautioned that the seasonal microbiome shifts could not be attributed specifically to repeated head impacts because numerous other factors may have contributed, including exercise intensity, NSAID use, sleep, stress, dietary variation, supplement use, and changes in living or dining circumstances.
Several nonimpact variables—including physical activity intensity and preworkout energy drink consumption—showed stronger and more statistically robust associations with microbiome dissimilarity than impact load itself.
Important Limitations
The study's limitations were substantial.
In addition to the very small final cohort and absence of a control group, the study experienced a high attrition rate, with only 6 of 19 enrolled athletes included in the final analysis. Three were excluded for missing head-impact data, and 10 more for submitting fewer than 15 fecal samples, raising the possibility of selection bias related to participant compliance or season-related burden.
Many covariates were self-reported, and missing survey data required statistical imputation.
The authors also acknowledged that the mixed-effects models had statistical power below 3% at the 95% confidence level, increasing uncertainty around nominally significant findings and raising the possibility of false-positive associations related to small-sample variability.
The cohort also excluded female athletes despite known sex differences in TBI responses.
Bottom Line
The study provides preliminary observational evidence suggesting that nonconcussive head impacts may be associated with short-term gut microbiome changes in collegiate football players. However, the findings were inconsistent across analytical approaches, several associations did not survive correction for multiple testing, and the study was severely underpowered.
The authors called for larger, controlled longitudinal studies incorporating inflammatory biomarkers, cognitive testing, dietary monitoring, and more diverse participant populations.
The authors reported no competing interests.
Source: PLOS One