Gut Bacterial Profiles Linked to DunedinPACE in 123 Adults

On July 14, 2026, researchers led by Braden P. Kunihiro published “Gut microbiome signatures associate with DNA methylation-based biological aging” in Scientific Reports. The study examined gut bacterial profiles alongside a blood-based estimate of aging pace in 123 adults.

Why the finding matters for aging research

The study found that differences in gut bacterial composition were associated with differences in DunedinPACE estimates of biological aging pace. Researchers used machine-learning models to assess whether microbial profiles helped predict those estimates. The species-level model explained 15.2% of variation in DunedinPACE scores in the study’s held-out test set. Its relevance to longevity research lies in connecting gut-community measurements with an aging biomarker, creating a specific question for follow-up rather than establishing a way to extend healthy life.

How to read the human study

This was a human observational study using paired stool and blood samples.

The researchers connected measurements from the same participants

According to coverage of the research, the participants ranged in age from 17 to 82. Stool samples supplied information about gut bacterial profiles. Blood samples supplied the DNA-methylation measurements used to calculate DunedinPACE. Pairing the samples allowed the researchers to examine how these measurements lined up within the same people.

DunedinPACE was the outcome the researchers sought to predict. It is a blood-based DNA-methylation measure intended to estimate the pace of biological aging rather than chronological age. The distinction matters when interpreting the study: its statistical target was an aging-rate estimate, not the number of birthdays a participant had reached.

This provides a useful way to read the result. Start with the outcome named in the analysis before interpreting what the microbial patterns mean. A model aimed at an aging-pace estimate answers a particular question about that measurement. Related context on the different questions aging clocks address helps place that target within the broader field.

The percentage describes a model evaluated in a group

A detailed account of the study reports that the held-out test set contained 37 participants. The researchers evaluated the species-level model against DunedinPACE scores in that group. This is the setting in which the reported explained variation was measured.

Put plainly, the analysis asked whether patterns in bacterial data helped account for differences in blood-based aging estimates across participants. The percentage belongs to that prediction exercise. The relevant pieces are the microbial inputs, the DunedinPACE target and the people in the test set. Keeping those pieces together preserves the meaning of the result.

A useful reading habit is to retain that context whenever the statistic appears in a headline or summary. “Explained variation” is a description of the model’s relationship with the measured scores. It should remain attached to the outcome and population in which the researchers evaluated it.

Two bacterial associations stood out

The researchers reported that Bifidobacterium adolescentis was associated with lower DunedinPACE estimates. Succinivibrio dextrinosolvens was associated with higher estimates in the study’s model. These names give future researchers specific candidates to investigate rather than leaving the finding at the broad level of “the microbiome.”

The direction of each association also makes the research question more precise. Follow-up work can ask whether the same bacterial patterns appear alongside the same aging measure in other participants. The authors describe the findings as hypothesis-generating, placing the species associations at the beginning of an investigation rather than its endpoint.

Why the evidence needs careful boundaries

The study’s limitations determine how far its findings can reasonably travel. They concern the design, the measurements and the proposed uses of the result.

• A simultaneous snapshot cannot establish cause and effect. Stool and blood were sampled at the same time. The analysis therefore cannot determine whether microbes influence aging-related biology or whether aging-related biology influences the microbiome. Other factors could also influence both measurements, so the observed relationship does not establish which process came first or what produced the association.
• The performance estimate comes from a small test set. The authors note that the limited number of held-out participants reduces the precision of the model-performance estimate. A test within this cohort provides an initial assessment of prediction. It does not establish how stable the reported performance would be if the analysis were repeated with a different set of people.
• Independent validation is missing. The authors report that they could not identify a compatible independent dataset for external validation. The model’s performance has therefore not been confirmed in a separate cohort. This is a distinct question from whether the model performed well in its own held-out group: independent replication would examine whether the relationship survives beyond the dataset used for this study.
• The cohort does not represent every population. Participants were drawn from Oʻahu and included substantial Native Hawaiian and Pacific Islander representation. The paper cautions that its microbial patterns may not generalize to populations with different backgrounds or diets. The appropriate boundary is to keep the finding attached to the population studied until other cohorts establish how broadly the pattern applies.
• Species assignments need confirmation. The researchers used 16S sequencing, which has limitations for species-level identification. They advise treating those assignments as hypotheses for follow-up rather than definitive identifications. This makes confirmation of the named bacteria an important part of evaluating the result, alongside checking whether the statistical associations recur.
• The signal was specific to the aging measure. The microbiome models did not meaningfully predict residuals from the Horvath, Levine and GrimAge2 clocks. The positive result therefore should not be described as consistent across all the aging measures examined. DunedinPACE needs to remain in the claim because it identifies the particular measurement for which the reported predictive signal appeared.
• Prediction is not a causal share of aging or evidence for an intervention. The explained-variation result does not mean gut bacteria control that fraction of a person’s aging. The study tested no probiotics or dietary changes. It also tested no fecal microbiota transplantation or other intervention, leaving no experimental basis here for choosing a bacterium or supplement as a longevity strategy.
• The findings are not individual diagnostic markers or lifespan forecasts. The authors explicitly say the results are not intended for individual diagnostic use. The study reports associations with a methylation-based pace measure rather than longevity outcomes. That places the finding on the research side of the gap between a biomarker association and clinical evidence, rather than establishing a stool-based assessment of an individual’s aging or future lifespan.

For readers evaluating a microbiome-aging claim, the practical approach is to identify the measured outcome and locate the population in which it was tested. Then check the validation method and compare the headline’s wording with those details. These questions help keep an interesting model result matched to the question the researchers actually asked. Personal decisions about biological-age testing or supplements are best discussed with a qualified clinician.

Why the next study must go further

The next logical step is independent replication in larger and more diverse cohorts, followed by longitudinal and intervention studies testing whether changes in microbial profiles precede or produce changes in aging-related outcomes.

How AgeAmaze helps

Treating the bacteria in this study as a longevity shopping list would overstate the evidence, and AgeAmaze avoids that jump by keeping research limitations visible. Its coverage gives research-minded adults clear explanations of biological-age findings and emerging interventions so they can distinguish an interesting association from evidence that supports a health claim.

Read the research

Sources

1. Researchers Identify Specific Gut Bacteria Linked to Faster and Slower Biological Aging
2. New Research Suggests Your Gut Bacteria May Hold the Key to Slower Aging