Can a Microbiome Test Predict Parkinson's Disease?

A 2026 paper in Nature Medicine proposed that a stool sample might flag Parkinson's disease before the first tremor. The claim is more interesting, and considerably more fragile, than the headline suggests. Working out why turns into a lesson about what a microbiome test can actually measure.

Lit Review Friday · Can a Microbiome Test Predict Parkinson's Disease? · Published 2026 · 21 minute read

Start With What a Microbiome Test Actually Is

Sequencing a stool sample is a census. It tells you which species are present and roughly how many of each. That is genuinely useful. It also cannot tell you what a single one of them did today.

The census comes with a headline number on it, called alpha diversity and usually reported as a Shannon index. All it really does is count how many different kinds of bug are in the room, and how evenly they are spread. Two decades of microbiome research have trained everyone to read that number in one direction. Low diversity tracks with disease. Higher is better.

That direction is broadly right, and the reason is not flattering to the last forty years. Most people in industrialized food environments are microbially depleted: low fiber, heavy antibiotic exposure, preservatives, and a general erosion of the inputs a diverse community needs. Against that baseline, more diversity usually does mean a healthier person.

Then there is the asterisk.

The Diversity Paradox

Severe constipation produces some of the highest diversity scores in the literature.

Henrik Roager and colleagues worked out why (Nature Microbiology, 2016). When material moves slowly through the colon, the fast fiber fermenters chew through the available carbohydrate early and then sit there with nothing left to eat. Slower organisms take over the time that is left. What you end up with is a crowded, genuinely diverse community that has switched from fermenting fiber to fermenting protein.

That switch changes the output. Instead of butyrate and the other short-chain fatty acids that colonocytes run on, the community produces ammonia, branched-chain fatty acids and phenolic compounds. Diversity went up. The chemistry got worse.

Doris Vandeputte and colleagues found the same relationship from the other side (Gut, 2016), reporting that stool consistency, which is a practical proxy for transit time, is negatively correlated with species richness and tracks strongly with overall community configuration. Their conclusion was blunt: stool consistency assessment is crucially important in microbiome association studies. It is also, as we will see, inconsistently done.

So What Would a Better Measurement Look Like?

Several groups are trying to build one. Three approaches are competing, and they disagree about almost everything except the answer.

GMWI2 pooled roughly eight thousand whole-genome-sequenced samples across twenty-six countries to produce a disease-agnostic wellness score. The ZOE Microbiome Health Ranking, published in 2025, built a leaderboard tied directly to cardiometabolic markers, and reported something that should give everyone pause: twenty-two of its top fifty beneficial species had never been formally named. The HACK index moved past counting species entirely and started scoring the functional genes the community carries.

Three philosophies, three methods, and they converge on the same organisms: Faecalibacterium prausnitzii, Roseburia intestinalis and Eubacterium rectale. The butyrate makers.

Which sounds like a solved problem, and is not, for two reasons.

The first is that measuring the producer does not guarantee the product. You can have the factory standing, machinery intact, staff on the payroll, and if there is no fiber arriving the line is idle. Butyrate output collapses while the sequencing report still reports a healthy roster.

The second is that these organisms feed each other. I like to think about it as a food web rather than a list. Roseburia does not eat the fiber you swallow. It eats the acetate that Bifidobacterium makes first. Take the Bifidobacterium out and Roseburia starves, even though nothing touched it directly.

Now the Provocative Part

Elisa Menozzi and colleagues, publishing in Nature Medicine in 2026, asked whether the gut microbiome could flag Parkinson's disease before any motor symptom appeared.

Their design had a genuinely clever centerpiece. They sequenced 271 people with diagnosed Parkinson's, 150 healthy controls, and forty-three people carrying a variant in the GBA1 gene. That variant raises Parkinson's risk up to thirty-fold, and yet only about twenty percent of carriers ever develop the disease. That eighty percent gap is where a second hit would have to live, and the researchers went looking for it in the gut.

Whole-genome shotgun sequencing gave them a signature of established Parkinson's: depleted butyrate producers, enriched Bifidobacterium, an ecosystem that reads as inflamed and underfed. Then they asked whether that signature appeared, faintly, in the healthy gene carriers. Their coherence analysis said yes, with 142 of 176 disease-associated species drifting in the same direction. They proposed a sixteen-species score, PDMS-16, as a possible early screening tool, and reported similar alterations in three independent cohorts from the United States, Korea and Turkey.

The Critique, and Why It Lands

William DePaolo published a detailed critique of the paper, and it is worth walking through because the reasoning generalizes well beyond this one study.

His first point is about coherence versus magnitude. One hundred and forty-two species drifting in the same direction sounds overwhelming. But direction and size are separate questions. If the stock market falls five thousand points and your checking account drops a nickel, you have both moved downward, and only one of you has a problem. Most of those 142 species were moving very slightly.

The rest of his critique is structural. The study is cross-sectional, a single snapshot, which cannot demonstrate progression or predict who converts. Forty-three gene carriers is a small subgroup carrying a large share of the paper's claim. Testing that many species invites multiple-comparison problems. And PDMS-16 is unvalidated, which is exactly what a proof-of-concept score is supposed to be, right up until someone builds a product on it.

There is a broader version of this problem. Reviewing sixteen case-control microbiome studies in Parkinson's, Boertien and colleagues (2019) counted more than a hundred differentially abundant taxa reported across the literature, some replicated, many contradictory, and concluded that methodological variation could not be separated from biology.

What Makes Anyone Think the Gut Is Involved at All?

It would be easy to read the last two sections as a dismissal. It is not one. The reason this field exists is that several independent lines of evidence point at the gut, and some of them are considerably harder to explain away than a signature study.

The bowel changes first, and it changes early

In 1971, the Honolulu Heart Program began following 6,790 Japanese-American men aged 51 to 75. Twenty-four years later, Robert Abbott and colleagues reported (Neurology, 2001) that men having fewer than one bowel movement a day had 2.7 times the risk of a Parkinson's diagnosis compared to men having one a day, adjusted for smoking, coffee, laxative use, physical activity and produce intake.

A 2016 meta-analysis of nine studies covering 741,593 people put the pooled odds ratio at 2.27, and when the analysis was restricted to constipation documented more than ten years before diagnosis, it barely moved: 2.13, with no heterogeneity between studies (Adams-Carr et al.). Work from the Rochester Epidemiology Project found the association holding at twenty or more years before motor onset (Savica et al., 2009).

Pathology can travel the nerve

Braak's hypothesis, proposed in the early 2000s, held that Parkinson's pathology might begin in the gut and ascend the vagus nerve to the brainstem. For years it was a compelling story without a demonstration.

Sangjune Kim and colleagues supplied one (Neuron, 2019). Alpha-synuclein fibrils injected into the duodenal and pyloric muscle layers of mice ascended to the dorsal motor nucleus, then the locus coeruleus, then the amygdala and substantia nigra, producing dopaminergic loss and both motor and non-motor deficits. Truncal vagotomy prevented it. So did knocking out alpha-synuclein. The route is real, at least in a mouse.

And the microbes themselves seem to matter

The strongest experimental result in this field came earlier. Timothy Sampson and colleagues (Cell, 2016) worked with mice engineered to overproduce alpha-synuclein. Raised germ-free, those mice had reduced motor deficits, less microglial activation and less alpha-synuclein aggregation than conventionally raised littermates. Antibiotics given in adulthood improved things. Recolonization brought the deficits back.

Then they colonized germ-free mice with stool from people. Mice receiving microbiota from Parkinson's patients showed worse motor impairment than mice receiving microbiota from healthy donors.

In this model the gut bacteria were not along for the ride. The phenotype needed them to show up at all. Now the caveats, and they are big ones. This is a mouse genetically loaded to overproduce the protein in question, which is a long way from idiopathic Parkinson's disease, and the human-stool arm rests on a small number of donors. As far as we can find, no independent laboratory has replicated that colonization result. That absence is the single largest weakness in the mechanistic case and it should not be smoothed over.

The Result That Should Make Everyone Slow Down

There is a detail inside the Sampson paper that gets quoted far less often than the rest of it, and it is the one worth dwelling on.

When the researchers gave short-chain fatty acids orally to germ-free alpha-synuclein mice, the SCFAs promoted neuroinflammation and motor symptoms.

Read that sentence again. Short-chain fatty acids are the molecules this whole field, us included, treats as the good news. And here they made things worse. So it is worth being really careful about what that result does and does not show, because at face value it contradicts a great deal of other work.

We are not resolving that tension here, and it would be dishonest to try. The takeaway is narrower and it holds up better: a molecule does not carry its meaning around with it. It does something in a context. Change the context and the same molecule can do the opposite.

How Would Gut Bacteria Reach the Brain?

Two routes have real evidence behind them, and both are worth naming precisely, because the precision is where the honesty lives.

The nerve. Kaelberer, Bohórquez and colleagues (Science, 2018) described neuropod cells. The cool thing about these is that they are gut lining cells that form an actual synapse with a nerve, which gives the gut a fast direct line to the brainstem instead of a slow hormonal one. Kim's 2019 work then showed that alpha-synuclein pathology can travel that anatomical route in mice.

The chemistry. Veit Rothhammer and colleagues (Nature Medicine, 2016) showed that dietary tryptophan is converted by gut bacteria into molecules that act on the aryl hydrocarbon receptor on astrocytes, limiting central nervous system inflammation. Supplementing indole, indoxyl-3-sulfate, indole-3-propionic acid or indole-3-aldehyde restored that control in mice, and people with multiple sclerosis showed reduced circulating levels of those agonists. Their 2018 follow-up in Nature extended the pathway through microglia.

Indole-3-propionic acid should look familiar. It is the molecule Episode 33 was entirely about: your own cells cannot make it, only certain gut bacteria complete the final step, and it turns up in tissues throughout the body. This is that same chemistry arriving in the brain.

Two honest caveats. Rothhammer's disease model is multiple sclerosis, not Parkinson's, and nobody has demonstrated that pathway operating in a Parkinson's model. And while neuropod cells are real, there is no published demonstration of neuropod signaling in a Parkinson's or alpha-synuclein context. What exists is the anatomical conduit and evidence that pathology can use it.

The Confound Nobody Adjusts For

Now put the diversity paradox from the beginning of this piece next to the epidemiology.

A person with prodromal Parkinson's has, on the evidence above, been constipated for a decade or more before diagnosis. And constipation by itself reshapes the gut microbiome in a specific, stereotyped direction: richness up, Akkermansia up, fermentation shifted from carbohydrate toward protein.

So a substantial part of what gets reported as the Parkinson's microbiome could be the slow-transit microbiome, measured in people who happen to have Parkinson's.

This is not speculation dropped in from a different literature. Mihai Cirstea and colleagues (Movement Disorders, 2020) sequenced 197 people with Parkinson's and 103 controls and found that the taxonomic shifts and the elevated proteolytic metabolites, p-cresol and phenylacetylglutamine, were strongly associated with stool consistency and constipation within the patient group. The authors read that as mechanism. The data is equally consistent with slowed transit producing the signature.

Here is the part that matters, and it cuts both ways. This does not show the microbiome is irrelevant to Parkinson's. It shows the measurement is contaminated, which means the observational signature studies cannot settle the question in either direction. Something else has to.

What Happens When You Actually Change the Microbiome?

Intervening is the only way past a correlation deadlock, and here the record is thin and genuinely mixed.

The largest and best-blinded trial is negative. Filip Scheperjans and colleagues (JAMA Neurology, 2024) randomized 47 people to colonoscopic fecal microbiota transplantation or placebo. The difference on the primary endpoint was 0.97 points, with a confidence interval running from minus 5.10 to plus 7.03 and a p-value of 0.75. Gastrointestinal adverse events were considerably more frequent in the treated group, 53 percent versus 7 percent, and the placebo group did better on several secondary measures.

Three smaller trials point the other way to varying degrees. Arnout Bruggeman and colleagues (2024) reported a modest motor improvement in 46 patients. Work from China in 54 patients reported gains in autonomic and gastrointestinal symptoms (Cheng et al., 2023). A twelve-person pilot found motility improved and motor benefit faded (DuPont et al., 2023). A 2026 trial of repeated donor transplantation in drug-naive patients reported motor and constipation improvements alongside reduced Escherichia-Shigella and decreased colonic alpha-synuclein aggregation (Zhang et al.).

Four small trials, pointing in different directions, and the one designed most carefully found nothing. Across all of them, gastrointestinal symptoms improve more reliably than motor symptoms do. That pattern is itself informative, and it is not the pattern you would expect if fixing the microbiome fixed the disease.

What Actually Helps People With Parkinson's Right Now?

It would be convenient for a microbiome company if the answer here were nutrition. It is not, or at least not primarily, and saying so is more useful than the alternative.

The best-evidenced lifestyle intervention in Parkinson's disease is high-intensity aerobic exercise. The Park-in-Shape trial (Lancet Neurology, 2019) randomized 130 people in a double-blind design and found a 4.2-point advantage on off-state motor scores at six months. The SPARX trial (JAMA Neurology, 2018) found high-intensity treadmill training met its non-futility threshold in 128 patients. There is no evidence that either works through the gut.

The nutrition and fiber evidence is real and considerably earlier-stage. Dietary pattern adherence has been associated with later age of onset, up to 17.4 years in women in one analysis, but that study was cross-sectional and relied on dietary recall (Metcalfe-Roach et al., 2021). Two prebiotic fiber trials, one using a fiber bar and one using resistant starch, raised short-chain fatty acids, moved inflammatory markers and reduced non-motor symptom burden. Both were open-label with no placebo control, and the authors of the first explicitly framed it as proof of concept to justify a controlled trial.

The one high-quality microbiome-directed trial in this space is instructive. Ai Huey Tan and colleagues (Neurology, 2021) randomized 72 patients to a multi-strain probiotic and found a clear benefit: 1.3 additional spontaneous bowel movements per week versus placebo, with better stool consistency and quality of life. Fecal calprotectin, an inflammation marker, did not change. It improved the bowel and it did not touch anything neurological.

So the defensible version of the claim is narrower than the appealing one: several interventions that meaningfully help people with Parkinson's act on the gut, and the microbiome is a plausible but unproven mediator for some of them.

Where That Leaves It

I went into this episode believing the microbes were probably causing a meaningful share of Parkinson's disease. I came out less sure, and more interested.

What holds up: something in the gut changes years before diagnosis, and it changes reproducibly. There is a demonstrated anatomical route from gut to brainstem. Microbial molecules genuinely regulate the brain's immune cells. And in at least one animal model, gut bacteria are not incidental to the disease process, they are required for it.

What does not hold up: that any of this has been shown in humans, that the observational signature studies can distinguish cause from consequence while transit time is confounding the measurement, or that a stool test can currently tell an individual anything about their risk.

And one more thing sharpens the whole picture. Collin Challis and colleagues (Nature Neuroscience, 2020) repeated the gut-to-brain seeding experiment in mice of different ages. The pathology progressed to the midbrain and produced motor deficits only in aged mice. In young animals it went nowhere. Gut seeding alone was not sufficient. The host had to have aged first.

Which connects this episode directly to the one before it. Episode 5 looked at what thirty years of dietary data say about aging well, and Parkinson's disease appears on the list of eleven conditions those researchers used to define healthy aging. If aging is the gate that lets a gut signal reach the brain, then the diet, the transit time and the neurology are not three subjects. They are one, viewed from three distances.

The Honest Limitations

Almost every strong causal result described here is in rodents, and most of those are in animals that were either germ-free or genetically primed to overproduce alpha-synuclein. Neither is a normal animal.

The human data is overwhelmingly cross-sectional, which means reverse causation remains unresolved across the entire field. Every observational finding above is equally compatible with early Parkinson's pathology causing the constipation and the microbial shift, rather than the other way around. Nothing currently distinguishes those two accounts.

The Honolulu cohort was men only, of one ethnicity, in one place. The interventional trials are small. And the appendix story is a standing reminder of how a striking, widely reported finding can evaporate when someone pools the data.

Frequently Asked Questions

Does constipation mean I will get Parkinson's disease?

Almost certainly not. Chronic constipation is associated with roughly twice the odds of a later diagnosis across large populations (Adams-Carr et al., 2016), but the underlying rates are very low, and constipation is extremely common. The association is real at population scale and tells an individual person very little.

Can a microbiome test predict neurological disease?

No test is validated for that today. The score proposed in the 2026 Nature Medicine study was explicitly a proof of concept, built from a single snapshot in time, and its critics pointed out that a snapshot cannot show who goes on to develop disease. Be skeptical of anything sold as a predictive stool test.

Could gut bacteria actually cause Parkinson's?

It remains an open question rather than a settled one in either direction. In one mouse model, gut bacteria were required for the disease phenotype to appear (Sampson et al., 2016), which is stronger than correlation. That work has not been replicated independently and has not been shown in people.

Why would slow digestion change gut bacteria at all?

Because the community runs on what arrives and how long it stays. When material moves slowly, the organisms that ferment fiber quickly run out of food, slower organisms take over, and the community shifts toward fermenting protein instead. The output changes even though the census looks fuller (Roager et al., 2016).

Is high gut microbial diversity always a good sign?

No, and this is the thing I most want people to take away. Diversity tells you something useful about a population and almost nothing about one person. Some of the highest diversity scores on record come from people with severe constipation, where the community is crowded and producing compounds that irritate the colon lining rather than nourish it.

Does fecal transplant treat Parkinson's disease?

Not on current evidence. The largest double-blind trial found no benefit on its primary endpoint and more gastrointestinal side effects in the treated group (Scheperjans et al., 2024). Smaller trials have reported modest gains, mostly in bowel and autonomic symptoms rather than movement.

The Bottom Line

References

1. Menozzi E, Ren Y, Geiger M, et al. Microbiome signature of Parkinson's disease in healthy and genetically at-risk individuals. Nature Medicine. 2026;32(6):2096-2106. https://doi.org/10.1038/s41591-026-04318-5 · FREE FULL TEXT
2. Sampson TR, Debelius JW, Thron T, et al. Gut microbiota regulate motor deficits and neuroinflammation in a model of Parkinson's disease. Cell. 2016;167(6):1469-1480.e12. https://doi.org/10.1016/j.cell.2016.11.018
3. Abbott RD, Petrovitch H, White LR, et al. Frequency of bowel movements and the future risk of Parkinson's disease. Neurology. 2001;57(3):456-462. https://doi.org/10.1212/wnl.57.3.456
4. Adams-Carr KL, Bestwick JP, Shribman S, et al. Constipation preceding Parkinson's disease: a systematic review and meta-analysis. Journal of Neurology, Neurosurgery & Psychiatry. 2016;87(7):710-716. https://doi.org/10.1136/jnnp-2015-311680
5. Savica R, Carlin JM, Grossardt BR, et al. Medical records documentation of constipation preceding Parkinson disease. Neurology. 2009;73(21):1752-1758. https://doi.org/10.1212/WNL.0b013e3181c34af5
6. Roager HM, Hansen LBS, Bahl MI, et al. Colonic transit time is related to bacterial metabolism and mucosal turnover in the gut. Nature Microbiology. 2016;1:16093. https://doi.org/10.1038/nmicrobiol.2016.93
7. Vandeputte D, Falony G, Vieira-Silva S, et al. Stool consistency is strongly associated with gut microbiota richness and composition, enterotypes and bacterial growth rates. Gut. 2016;65(1):57-62. https://doi.org/10.1136/gutjnl-2015-309618 · FREE FULL TEXT
8. Cirstea MS, Yu AC, Golz E, et al. Microbiota composition and metabolism are associated with gut function in Parkinson's disease. Movement Disorders. 2020;35(7):1208-1217. https://doi.org/10.1002/mds.28052
9. Kim S, Kwon SH, Kam TI, et al. Transneuronal propagation of pathologic alpha-synuclein from the gut to the brain models Parkinson's disease. Neuron. 2019;103(4):627-641.e7. https://doi.org/10.1016/j.neuron.2019.05.035
10. Challis C, Hori A, Sampson TR, et al. Gut-seeded alpha-synuclein fibrils promote gut dysfunction and brain pathology specifically in aged mice. Nature Neuroscience. 2020;23(3):327-336. https://doi.org/10.1038/s41593-020-0589-7
11. Erny D, Hrabe de Angelis AL, Jaitin D, et al. Host microbiota constantly control maturation and function of microglia in the CNS. Nature Neuroscience. 2015;18(7):965-977. https://doi.org/10.1038/nn.4030
12. Rothhammer V, Mascanfroni ID, Bunse L, et al. Type I interferons and microbial metabolites of tryptophan modulate astrocyte activity and central nervous system inflammation via the aryl hydrocarbon receptor. Nature Medicine. 2016;22(6):586-597. https://doi.org/10.1038/nm.4106
13. Rothhammer V, Borucki DM, Tjon EC, et al. Microglial control of astrocytes in response to microbial metabolites. Nature. 2018;557(7707):724-728. https://doi.org/10.1038/s41586-018-0119-x
14. Kaelberer MM, Buchanan KL, Klein ME, et al. A gut-brain neural circuit for nutrient sensory transduction. Science. 2018;361(6408):eaat5236. https://doi.org/10.1126/science.aat5236
15. Scheperjans F, Levo R, Bosch B, et al. Fecal microbiota transplantation for treatment of Parkinson disease: a randomized clinical trial. JAMA Neurology. 2024;81(9):925-938. https://doi.org/10.1001/jamaneurol.2024.2305
16. Bruggeman A, Vandendriessche C, Hamerlinck H, et al. Safety and efficacy of faecal microbiota transplantation in patients with mild to moderate Parkinson's disease (GUT-PARFECT): a double-blind, placebo-controlled, randomised, phase 2 trial. EClinicalMedicine. 2024;71:102563. https://doi.org/10.1016/j.eclinm.2024.102563 · FREE FULL TEXT
17. Zhang R, Feng R, Wang J, et al. Gut microbiota modulation via repeated donor fecal transplantation improves motor and gastrointestinal symptoms in drug-naive Parkinson's disease: a randomized phase 2 trial. Signal Transduction and Targeted Therapy. 2026;11:94. https://doi.org/10.1038/s41392-026-02604-9
18. Tan AH, Lim SY, Chong KK, et al. Probiotics for constipation in Parkinson disease: a randomized placebo-controlled study. Neurology. 2021;96(5):e772-e782. https://doi.org/10.1212/WNL.0000000000010998
19. van der Kolk NM, de Vries NM, Kessels RPC, et al. Effectiveness of home-based and remotely supervised aerobic exercise in Parkinson's disease: a double-blind, randomised controlled trial. Lancet Neurology. 2019;18(11):998-1008. https://doi.org/10.1016/S1474-4422(19)30285-6
20. Schenkman M, Moore CG, Kohrt WM, et al. Effect of high-intensity treadmill exercise on motor symptoms in patients with de novo Parkinson disease: a phase 2 randomized clinical trial. JAMA Neurology. 2018;75(2):219-226. https://doi.org/10.1001/jamaneurol.2017.3517
21. Metcalfe-Roach A, Yu AC, Golz E, et al. MIND and Mediterranean diets associated with later onset of Parkinson's disease. Movement Disorders. 2021;36(4):977-984. https://doi.org/10.1002/mds.28464
22. Hall DA, Voigt RM, Cantu-Jungles TM, et al. An open label, non-randomized study assessing a prebiotic fiber intervention in a small cohort of Parkinson's disease participants. Nature Communications. 2023;14:926. https://doi.org/10.1038/s41467-023-36497-x · FREE FULL TEXT
23. Killinger BA, Madaj Z, Sikora JW, et al. The vermiform appendix impacts the risk of developing Parkinson's disease. Science Translational Medicine. 2018;10(465):eaar5280. https://doi.org/10.1126/scitranslmed.aar5280
24. Lu HT, Shen QY, Xie D, et al. Lack of association between appendectomy and Parkinson's disease: a systematic review and meta-analysis. Aging Clinical and Experimental Research. 2020;32(11):2201-2209. https://doi.org/10.1007/s40520-019-01354-9
25. Boertien JM, Pereira PAB, Aho VTE, Scheperjans F. Increasing comparability and utility of gut microbiome studies in Parkinson's disease: a systematic review. Journal of Parkinson's Disease. 2019;9(s2):S297-S312. https://doi.org/10.3233/JPD-191711 · FREE FULL TEXT
26. Wallen ZD, Demirkan A, Twa G, et al. Metagenomics of Parkinson's disease implicates the gut microbiome in multiple disease mechanisms. Nature Communications. 2022;13:6958. https://doi.org/10.1038/s41467-022-34667-x · FREE FULL TEXT

This post accompanies the Lit Review Friday episode of Learn Something with Thaena.