Gut Microbiome Explains 6% of Triglyceride Variance in 893 Adults

This cross-sectional analysis of 893 adults in the LifeLines-DEEP cohort, published in Circulation Research, used 16S rRNA sequencing to quantify how much of the variation in blood lipids and body mass index is explained by gut microbiome composition. Independent of age, sex, and host genetics, the microbiome explained 6% of triglyceride variance and roughly 4% each of HDL cholesterol and BMI variance -- effects on the same order of magnitude as known genetic risk loci. The trial is one of the largest population-level investigations of the gut-cardiovascular axis and is frequently cited as the epidemiological proof that resident gut bacteria modulate human lipid biology.

Why This Study Matters

By 2015, animal experiments had already shown that gut bacteria can shift cholesterol and triglyceride levels in mice. What the field lacked was a clean population-scale signal: was the same thing happening in free-living humans, and how much of the variability in blood lipids -- after controlling for diet, age, sex, and genetics -- could actually be attributed to microbes? Without that number, "the microbiome influences cardiovascular risk" stayed at the level of mechanism rather than measurable contribution.

This study set out to put a number on it. The researchers analysed 893 participants from LifeLines-DEEP, a deep-phenotyped subset of the larger LifeLines cohort in the Northern Netherlands. They sequenced stool samples, profiled blood lipids and BMI, then ran cross-validation models to estimate how much variance the microbiome explained on top of host genetics and demographics.

The result was the first robust human estimate of the lipid-microbiome relationship at scale -- and it was large enough to matter. The 6% of triglyceride variance the microbiome accounted for is comparable in magnitude to the variance explained by polygenic risk scores from genome-wide association studies, which the field had spent more than a decade building. The implication: gut bacteria belong in the same conversation as genetics when discussing what shapes blood lipids.

How It Was Designed

The basics are in the study design bar above: 893 adults, cross-sectional, 16S rRNA sequencing of stool, LifeLines-DEEP cohort at University Medical Center Groningen. A few design choices deserve attention.

LifeLines-DEEP was built for exactly this kind of multi-omics work. Each participant had genome-wide genotyping, full lipid panels, anthropometric data, food-frequency questionnaires, and stool collection -- all on the same individuals. That meant the researchers could adjust for genetic risk directly rather than estimating it, and they could partition variance attributable to microbes vs. host genetics rather than confounding the two.

The analysis used cross-validated variance explained rather than in-sample R-squared. This is an important methodological choice: in-sample fits inflate when you have thousands of microbial features and a few hundred subjects. Cross-validation -- training on one subset, testing on another -- gives an honest estimate of how much signal would generalise outside the discovery sample. The 6% triglyceride figure is cross-validated, not the (much larger) in-sample number.

The researchers also explicitly tested whether microbiome associations were independent of established cardiovascular risk factors. Models were adjusted for age, sex, BMI (when BMI wasn't the outcome), and 31 host single-nucleotide polymorphisms known to influence lipid levels. The bacterial associations held after all of these adjustments, which is what licenses the "independent of host genetics" framing.

What They Found

The microbiome explained a measurable, replicable share of variance in body mass index and two of the four major lipid fractions. The headline numbers from the cross-validation analysis:

Trait	Variance Explained by Microbiome	Direction	What It Measures
Triglycerides	6.0%	Significant association	Blood triglyceride concentration
HDL cholesterol	4.0%	Significant association	High-density lipoprotein
Body mass index	4.5%	Significant association	Weight relative to height
LDL cholesterol	Not significant	No detectable signal	Low-density lipoprotein
Total cholesterol	Not significant	No detectable signal	Total blood cholesterol

Variance-explained values are from cross-validated models adjusted for age, sex, and host genetic risk. "Not significant" means the microbiome contribution was indistinguishable from zero.

Beyond the global variance estimates, the researchers identified 34 specific bacterial taxa whose abundances were associated with BMI or blood lipids after correction for age and sex. Most of those associations were novel. When the microbiome features were added to a risk model that already included host genetics and standard covariates, the full model explained up to 25.9% of HDL variance -- significantly better than the same model without the microbiome.

Reading the Results

The findings group into three patterns worth pulling apart.

Selective effect on triglycerides and HDL, not LDL or total cholesterol. This is the most clinically interesting pattern in the paper. The microbiome's fingerprint shows up specifically on the lipid fractions most influenced by metabolism, hepatic fat handling, and bile acid recycling -- triglycerides and HDL -- and not on the fractions most directly controlled by hepatic LDL receptor expression and dietary saturated fat. That selectivity is biologically coherent: gut bacteria modify bile acids in the intestine and produce short-chain fatty acids that signal to the liver, both of which intersect more with triglyceride and HDL metabolism than with LDL clearance.

The magnitude is comparable to genetics. Genome-wide association studies for lipids have, after years of meta-analyses across hundreds of thousands of people, identified loci that collectively explain roughly 10-15% of variance for each lipid trait. The microbiome explaining 6% of triglyceride variance in a single 893-person cohort is, on a per-study basis, an effect size in the same range. The clinical implication is that targeting the microbiome could plausibly shift triglycerides and HDL by amounts that matter for cardiovascular risk -- not as a replacement for statins, but as an additional and independent lever.

Most of the associated taxa were new. Of the 34 bacteria linked to lipids or BMI in this analysis, the majority had not been previously connected to cardiovascular risk markers. That tells you the field was still in discovery mode in 2015 -- and that downstream mechanistic work, including the bile salt hydrolase studies that followed, was needed to translate these correlations into causation.

What Didn't Change

The microbiome had effectively no detectable contribution to LDL cholesterol or total cholesterol variance in this analysis. That is a meaningful null result. It places a ceiling on what gut bacteria are likely to do for those specific markers in a generally healthy adult population, and it sharpens the case that microbiome-targeted approaches are most plausible for triglycerides, HDL, and weight rather than LDL.

The study is also cross-sectional. It cannot establish that the bacterial composition caused the lipid patterns, only that the two co-vary at a population level. Reverse causation -- lipid levels or diet shaping the microbiome rather than the other way around -- remains possible. The cross-validated variance estimates are robust evidence of association; they are not evidence of direction.

Broader Context

This paper became one of the foundational citations for the gut-cardiovascular axis. It has been referenced extensively in subsequent reviews of how the microbiome influences metabolic disease, including the 2023 narrative review by Tarabanis and colleagues that maps the olive polyphenol literature against gut and cardiovascular endpoints. Companion work from the same LifeLines-DEEP group published alongside this paper detailed the environmental factors -- diet, medication, smoking -- that shape the human gut microbiome at population scale.

The mechanistic complement to this epidemiology is the bile salt hydrolase pathway. Certain gut bacteria, including several Lactobacillus strains, express enzymes that deconjugate bile acids in the small intestine, increasing faecal bile acid excretion and forcing the liver to use circulating cholesterol to synthesise replacements. Multiple randomized trials and preclinical studies in the years following Fu et al. tested specific probiotic strains against blood lipids, with the most robust effects on triglycerides and HDL -- the same fractions flagged by this population analysis.

What made this 2015 study influential was scale and methodology. Most prior microbiome-cardiovascular work was either small (under 100 participants) or done in animals. This was the first cohort large enough to cross-validate variance estimates against host genetics in humans, and it set a quantitative benchmark for the gut-CV axis that later trials could measure themselves against.

Related Research

Continue exploring olive oil and polyphenol science:

• L. reuteri NCIMB 30242 Lowered Cholesterol: A Bile Salt Hydrolase RCT
• L. plantarum CECT 7527/7528/7529 Lowered LDL: A 12-Week RCT
• Olive Oil and 19% Lower Mortality in 92,383 U.S. Adults Over 28 Years

Source: View the original study on PubMed

Olivea's Dosage

This study did not test an Olivea product. It is a population-level analysis of how gut bacteria covary with blood lipids in 893 adults, independent of any intervention. Olivea's current lineup -- extra virgin olive oil and the polyphenol capsule -- is built around hydroxytyrosol and the broader olive polyphenol profile rather than live probiotics. This paper helped establish the gut-cardiovascular axis evidence base that Olivea's research library continues to track; the current third-party certificate of analysis documents what is in the existing products.

We share this research for transparency. This is an independent study -- we did not fund it, design it, or conduct it.

Editorial Information

Research note. This article summarizes third-party research published in a peer-reviewed journal. Olivea did not conduct or fund the study. Findings reflect the cited paper only and do not establish efficacy of Olivea products.

.