Parkinson's and the Microbiome Health Score Debate

April 24, 2026 ·

Thaena Inc.

Two major 2026 papers just landed on the gut-Parkinson's question, and they pull the field in opposite directions. One claims your microbiome can reveal neurodegeneration risk before a single tremor shows up. The other shows that swapping the microbiome temporarily improves motor symptoms in newly diagnosed patients. Read together, they force a much harder question into the open: what actually counts as a healthy microbiome, and why is no single test giving us a clean answer?

Lit Review Friday · Learn Something with Thaena · Published 2026 · Reading time: ~18 minutes

Two Papers Just Landed on the Gut-Parkinson's Question

Parkinson's disease is a progressive neurodegenerative disorder defined clinically by motor symptoms: tremor, rigidity, slow movement. By the time any of those are visible, a catastrophic amount of damage has already occurred in the brain. The dopamine-producing neurons in the substantia nigra have been massively depleted. Finding a biomarker that can flag the disease during the prodromal phase, meaning the silent years before tremors begin, is one of the most important unsolved problems in neurology.

For a decade now, the gut has been a prime suspect. The enteric nervous system is densely innervated and communicates directly with the brain through the vagus nerve. Misfolded alpha-synuclein, the protein aggregate that defines Parkinson's pathology, appears in gut tissue in some patients years before it accumulates in the brain. Severe constipation is one of the most reliable non-motor prodromal symptoms. And decades of population data confirm that the gut microbiome of diagnosed Parkinson's patients looks structurally different from healthy controls.

Two papers published in 2026 pushed the conversation forward in very different directions. One is a predictive-diagnostic paper that tried to build an early warning tool from a stool sample. The other is a therapeutic paper that tested whether swapping the microbiome could actually move the needle on motor function. We are going to walk through both, then unpack the deep question that both papers ultimately depend on: what does a healthy microbiome even look like, and how do we know when we are looking at one?

Paper One: Menozzi et al. 2026 in Nature Medicine

Menozzi and colleagues asked whether the gut microbiome carries a detectable signature of Parkinson's disease in people who do not yet have the clinical diagnosis. The design leaned heavily on genetics. The team recruited three groups: 271 patients with a confirmed Parkinson's diagnosis, 150 healthy controls, and the crown jewel, 43 people carrying a mutation in the GBA1 gene who had no clinical diagnosis of Parkinson's.

GBA1 encodes glucocerebrosidase, a lysosomal enzyme that helps cells clear out misfolded proteins and other molecular waste. Mutations in GBA1 are the strongest known monogenic risk factor for Parkinson's disease. A carrier's lifetime risk can jump by up to thirty-fold compared to the general population. But only about twenty percent of GBA1 carriers ever develop the disease. That incomplete penetrance is the biological mystery the paper set out to illuminate. If the microbiome is the environmental second hit that pushes some carriers into disease, we might be able to see an intermediate signature emerging in carriers who are on the path toward conversion.

What the whole-genome sequencing showed

The team used shotgun metagenomics, which sequences all the DNA in a stool sample rather than just one marker gene. This gave species-level resolution and access to functional gene content, which matters enormously for what comes next. In the overt Parkinson's cohort, they found 176 species that were significantly altered compared to healthy controls. The pattern aligned with prior literature: depletion of butyrate-producing bacteria like Roseburia intestinalis and Faecalibacterium prausnitzii, alongside enrichment of Bifidobacterium and several opportunistic taxa. An inflamed ecosystem starved of short-chain fatty acid signaling.

Then they asked the critical question: did those same 176 disease-associated species show any shift in the 43 healthy GBA1 carriers? They used a coherence analysis based on Cliff's delta, which measures the direction of the shift rather than its magnitude. Out of 176 species, 142 were moving in the same direction in the GBA1 carriers as they were in the diagnosed patients. The authors framed this as an "intermediate" microbial signature, a partially drifted ecosystem sitting somewhere between healthy and diseased.

The leap the paper took, and the backlash

The paper's real ambition was not documenting the overt disease signature, which is well-trodden ground. The ambition was leveraging those findings into a predictive tool. They built the PDMS-16 score, a diagnostic metric composed of 16 specific bacterial species selected for their ability to separate groups in the data. They then applied the score to their 150 healthy controls, none of whom carried GBA1 mutations, and reported that a subset of these otherwise-healthy individuals showed microbiome patterns resembling Parkinson's disease. These high-scoring healthy individuals also tended to report more constipation, altered eating habits, and low mood. The paper's core provocation: maybe we can screen for Parkinson's risk from a stool sample, even in people without known genetic predisposition.

That is the headline that caught press attention. It is also the claim that William DePaolo, a prominent microbiome scientist, took apart publicly within days of publication.

The DePaolo Critique: Four Reasons to Slow Down

DePaolo's commentary is worth reading in full because it is a master class in distinguishing what a paper actually proves from what its abstract claims. He is explicit that the Menozzi study is not a bad paper. It uses rigorous shotgun sequencing, leverages a genuinely interesting at-risk cohort, and validates the overt disease signature across three continents. The issue is the gap between what the data show and what the narrative promises.

The PDMS-16 score caught DePaolo's sharpest attention. The score was derived entirely within a single dataset. The authors themselves acknowledged in their peer review response that PDMS-16 has "no immediate clinical usefulness" and is intended as a proof-of-concept for future development. But that nuance rarely survives the press release. DePaolo's warning is blunt: a microbiome signature derived from overfitting a cross-sectional cohort is not a validated biomarker, and treating it as one risks terrifying healthy people into believing they are on the path to neurodegeneration based on statistical artifacts.

The honest reading of Menozzi et al. is this: Parkinson's disease has reproducible microbial associations. Some of those associations appear in directionally similar, lower-magnitude form in non-manifesting GBA1 carriers and a subset of healthy controls. These findings justify decade-long longitudinal studies to test whether microbiome profiles can actually predict conversion. Nothing in the paper proves they can today.

Paper Two: Zhang et al. 2026 in Signal Transduction and Targeted Therapy

If Menozzi asked whether the microbiome reflects Parkinson's, Zhang asked whether changing the microbiome can change the disease. The answer matters, because it maps directly onto the cause-versus-consequence question DePaolo raised. If microbes are just rubberneckers, replacing them should do nothing for symptoms. If microbes are actively participating in the disease loop, replacement should move the needle.

Zhang and colleagues designed a randomized, double-blind, placebo-controlled Phase 2 trial of repeated donor fecal microbiota transplantation in drug-naïve Parkinson's disease patients. Seventy-two patients were enrolled and randomized one-to-one to receive donor FMT (dFMT) or autologous FMT (the patient's own stool, serving as placebo). Sixty-six completed the trial. Each FMT cycle consisted of seven days of administration per four-week cycle, with 200 mL on days one through three and 50 mL on days four through seven.

Why "drug-naïve" is the most important word in the trial

Levodopa is the cornerstone of Parkinson's symptom management. It is also known to directly alter microbial composition and metabolism. Most microbiome-and-Parkinson's studies enroll patients who have been on levodopa for years, which means their stool samples reflect both the disease and its treatment, inextricably mixed. Zhang's team specifically recruited patients who had been newly diagnosed but had not yet started any dopaminergic medication. That single design choice isolates the disease from the treatment and eliminates the most significant confounder in the Parkinson's microbiome literature.

What the trial showed

A 3.8-point improvement on the UPDRS motor scale in drug-naïve patients is clinically meaningful. For context, the minimal clinically important difference on this scale is often cited as around 2.5 to 3.0 points. The effect was real while treatment was active, and the mechanistic correlates line up with what you would predict if the intervention were genuinely biological: reduced colonic alpha-synuclein aggregation, restored barrier integrity, and shifts in dopamine-related metabolites in the stool.

What Zhang proves, and what it doesn't

Zhang does not prove that microbes cause Parkinson's disease in the first place. The trial enrolled people who already had the diagnosis. What it proves is that the microbiome is an active participant in the symptom presentation of established disease, not an inert bystander. Replacing the flora measurably improved motor function and GI symptoms, and the effect faded when replacement stopped. That is incompatible with a pure passenger model.

Read alongside the DePaolo critique, the synthesis is nuanced. DePaolo is right that a cross-sectional snapshot cannot prove microbes drove the original disease. But the gut-brain relationship in Parkinson's is clearly bi-directional. The failing autonomic nervous system slows gut motility. The slow gut changes the microbial ecosystem. The altered ecosystem produces different metabolites that feed back onto neuroinflammation and alpha-synuclein dynamics. Zhang shows this loop is real and at least temporarily interruptible. Menozzi shows the ecosystem state carries information, even if what we can do with that information remains unclear.

Both papers, and the debate between them, ultimately lean on a question the field has not yet resolved. What does a healthy microbiome actually look like?

So What Actually Counts as a Healthy Microbiome?

This is the question underneath every microbiome test, every probiotic pitch, and every conclusion drawn from a paper like Menozzi's. If you cannot define a healthy microbiome with precision, then comparing a "Parkinson's-like" signature against a "healthy" baseline is comparing one moving target to another. The honest answer is that the field is in the middle of overhauling its definition of gut health, and every major method we have comes with sharp limitations.

The tech leap that reset the conversation: 16S to whole-genome shotgun

For two decades, most microbiome research relied on 16S rRNA gene sequencing. The 16S gene is found in all bacteria, encodes part of the ribosome, and is conserved enough to serve as a universal barcode. You amplify it by PCR, sequence it, and match the results to reference databases. It is cheap, fast, and gave us the first detailed maps of which major bacterial lineages inhabit the human gut.

But 16S is essentially looking through a peephole. It tells you roughly who is in the room at the genus level, sometimes species. It tells you almost nothing about what they are doing. You cannot see which metabolic pathways they carry, which antibiotic resistance genes they hold, which toxins they can produce, or how they communicate with their neighbors. For a long time that was the best we had.

Whole-genome shotgun (WGS) metagenomics changed the resolution entirely. Instead of targeting one gene, you take all the DNA in a stool sample, fragment it, and sequence the fragments indiscriminately. Computational algorithms then reassemble those fragments into contiguous stretches and match them against species-level references. You get strain-level resolution when the data is good enough. You also get the full functional gene content, which means you can read which enzymes each species carries, which metabolic pathways they can run, and which virulence or resistance genes they harbor. WGS even lets researchers assemble entirely new genomes directly from stool, revealing species that have never been cultured in a lab. The ZOE study published in 2025 in Nature identified 22 beneficial gut species that were completely new to science, simply by running WGS at scale.

This is the tool that all three papers in today's discussion used. It is the reason they can say anything meaningful at the species and strain level. It is also the reason the old rules of gut health are getting rewritten.

The alpha-diversity paradox

The most durable piece of microbiome folk wisdom is that higher diversity equals better gut health. It is the rule repeated in every nutrition magazine, every probiotic marketing email, and most clinical reports. The concept is borrowed from macroecology and is typically measured as alpha-diversity: a combination of species richness (how many different species are present) and evenness (how equally abundant they are). The Shannon index is the standard summary statistic.

The logic seemed bulletproof. People with inflammatory bowel disease have markedly low diversity. Severe C. difficile infection collapses diversity. Isolated hunter-gatherer populations have much higher diversity than industrialized Western populations. Case closed, more is better.

Except it is not. High alpha-diversity is strongly associated with severe constipation. When colon transit time slows and stool remains in the gut for days on end, the ecosystem fundamentally reorganizes. Fast-growing fiber fermenters run out of substrate and die off. Slow-growing species specialized for extracting every residual calorie take over. The total species count can actually rise, because the stagnant environment accommodates more microbial lodgers. But what those bacteria are producing shifts catastrophically. With fermentable fiber exhausted, they turn to putrefying proteins, generating ammonia, branched-chain fatty acids, and phenolic compounds that are toxic to the colonic epithelium.

A 2024 meta-analysis in ISME J concluded explicitly that "diversity alone does not reliably indicate the healthiness of an individual." Disease contexts in which high diversity is not protective, or is actually associated with worse outcomes, include bacterial vaginosis, HIV/AIDS, irritable bowel syndrome, and several neurological conditions. For Parkinson's disease specifically, alpha-diversity has been repeatedly shown to be a non-significant marker. Which is part of why Menozzi et al. went past diversity metrics to build their PDMS-16 score on specific species identity.

The blood-gut mirror: Wilmanski 2019

If staring at the stool gives us this tangled, paradoxical picture, could we sidestep the problem by measuring the gut indirectly, through the blood? That is the question Wilmanski and colleagues asked in a 2019 paper in Nature Biotechnology. They enrolled 399 people in a wellness program and performed untargeted metabolomics on their blood plasma, measuring over a thousand small molecules.

The first finding was a failure of conventional diagnostics. Seventy-seven standard clinical laboratory tests (liver enzymes, lipid panels, blood counts) and 263 plasma proteins (mostly inflammatory markers) were incapable of predicting gut microbial diversity. The standard medical toolkit was effectively blind to what was happening in the gut.

The second finding was a triumph of untargeted metabolomics paired with machine learning. The team used a lasso regression, which aggressively penalizes variables that do not carry real signal and shrinks their contribution toward zero. Out of 659 small molecules detected in plasma, the lasso model isolated just 40 metabolites that together explained 45 percent of the variance in gut alpha-diversity. That is a staggering level of predictive power for a measurement taken in a completely different body compartment.

The identity of those 40 molecules is what makes the paper conceptually important. Thirteen of them were explicitly of microbial origin. Many of the rest were what the authors called "human-microbial co-metabolites," molecules whose production requires both human and microbial enzymatic steps. Bile acids are the classic example. The liver secretes primary bile acids. Gut bacteria chemically modify them into secondary bile acids by cleaving off specific amino acids. Those secondary bile acids are reabsorbed into circulation, where they act as signaling hormones affecting metabolism. Hippurate is another example. Plant polyphenols from diet are fermented by gut bacteria into benzoic acid, which is absorbed, conjugated with glycine in the liver, and excreted as hippurate. If the microbial step goes missing, the blood signature changes.

Eleven of the 40 metabolites proved so robust that they survived every statistical stress test the team threw at them, including validation in an independent cohort of 540 people. The blood genuinely does carry a readable reflection of what the gut microbiome is doing.

With one massive caveat. In people with class two or class three obesity (BMI 35 or higher), the 40-metabolite signature lost its predictive power. The host metabolism was so disrupted at baseline that the microbial signal got drowned out. When the host is inflamed and shouting, the microbes whispering in the background become unhearable. The blood mirror shatters under severe systemic metabolic stress.

This is the foundational multi-omics result for the field. It proves the gut behaves like an endocrine organ, with its outputs detectable in circulation. It also proves the measurement is context-dependent, vulnerable to host state in ways that mirror the problems with stool-based diversity metrics.

The index wars: GMWI2, ZOE, and HACK

Given the failure of diversity as a singular metric and the host-state dependence of blood metabolomics, the race is on to build a unified whole-genome-based microbiome health index. Three major contenders currently dominate the conversation, each representing a different philosophy.

GMWI2 (Gut Microbiome Wellness Index 2), published in 2024, is a brute-force statistical approach. The team pulled 8,069 whole-genome sequenced stool samples from 54 published studies, spanning 26 countries across six continents. They built a log-ratio model comparing the abundance of "health-prevalent" species (the ones consistently enriched in healthy individuals across datasets) against "health-scarce" species (consistently enriched in diseased individuals), integrated with diversity metrics. The resulting score is explicitly disease-agnostic. It does not diagnose a specific condition. It asks how far your ecosystem has drifted from the baseline of human wellness. Cross-validated balanced accuracy against clinically diagnosed vs. non-diagnosed individuals is around 80 percent. A respectable B-minus for diagnostic performance, and a meaningful improvement over any single diversity metric.

ZOE microbiome health ranking, published in 2025 in Nature, took a different philosophical path. Rather than comparing healthy to diseased, the ZOE team tethered microbial composition directly to cardiometabolic clinical markers (blood sugar response, triglyceride clearance, visceral fat) across 34,000 samples from the US and UK. They evaluated 661 species and ranked the top 50 most reliably associated with favorable metabolic markers and the top 50 most reliably associated with unfavorable ones. The score is a 0-to-1 ranking. Low is good. It is dynamic, responsive to dietary intervention, and tied to measurable outcomes. The critique is reductionism. Ranking individual species as golf scorecards ignores that bacteria live in cross-feeding networks and a "bad" bug in isolation might be a critical intermediate for a "good" bug in context.

HACK (Health-Associated Core Keystone) index, published in Cell Reports in 2025, represents a conceptual leap. Rather than counting species, HACK scores species on three integrated properties: prevalence and ecological connectedness in non-diseased microbiomes, longitudinal stability over time, and negative association with disease. Built from over 45,000 gut microbiomes across 127 studies and 42 countries, HACK identified 18 species-level keystone taxa whose association strengths sit in the top 30th percentile across all three dimensions. Crucially, HACK scores functional gene carriage, not just taxonomic identity. It looks for the metabolic machinery each species actually carries, particularly the genes required to synthesize short-chain fatty acids and other beneficial metabolites.

Three different methods, three different philosophies, one remarkable point of convergence: they all identify the same small set of keystone species as central to gut health. Faecalibacterium prausnitzii, Roseburia intestinalis, and Eubacterium rectale sit at the top of every list. These are the primary butyrate producers in the human colon.

Butyrate itself is the short-chain fatty acid that connects this story to everything else. It is the primary energy source for colonocytes, the cells that line the colon. It maintains tight-junction integrity, keeping the gut barrier closed. It suppresses inflammatory cytokines. When butyrate producers are depleted, the host is almost universally in a state of disease, including inflammatory bowel disease, colorectal cancer, type 2 diabetes, and neurological decline. This is exactly the pattern Menozzi et al. documented in the Parkinson's signature.

The butyrate disconnect

Here is where commercial microbiome testing gets into genuine trouble. The logic seems clean: if butyrate producers are the VIPs and high levels on your WGS test indicate health, then a high Faecalibacterium abundance should mean you are in good shape. The factory is there. The workers are clocked in.

Measuring the producer does not guarantee the product. This is the butyrate disconnect, and it has three distinct failure modes.

First, having the genetic blueprint for a factory does not mean the machinery is switched on. WGS metagenomics detects which genes are present, not which genes are actively transcribed at the moment of sampling. A bacterium can carry butyrate synthesis pathways and simply not be running them.

Second, there are multiple biochemical pathways to make butyrate. The main route uses the butyryl-CoA:acetate CoA-transferase pathway. Others use the butyrate kinase pathway. Different pathways activate under different environmental conditions, and WGS does not distinguish which is engaged in a given gut.

Third, and most often the clinically relevant failure point, is substrate availability. Butyrate producers need specific fermentable fibers as raw material. If the host's diet is dominated by ultra-processed foods low in complex carbohydrates, the factory sits idle regardless of how abundant the producers are. You can have plenty of Faecalibacterium and no butyrate output, because there is nothing to ferment.

The ecology goes a layer deeper through cross-feeding. Many butyrate producers, especially Roseburia, are secondary scavengers. They do not directly ferment the fiber you swallow. They rely on primary fermenters like Bifidobacterium to break complex carbohydrates down into acetate, which Roseburia then converts into butyrate. If an antibiotic or a diet change wipes out Bifidobacterium, Roseburia starves even though the antibiotic never touched it directly. The community is a food web, not a spreadsheet.

This is why a direct-to-consumer microbiome test showing "high" keystone species is not a meaningful statement about metabolic output. It is a statement about potential. Actual butyrate production depends on diet, transit time, cross-feeding integrity, and redox state of the gut environment, none of which a sequencing readout captures.

The context trap: why the same bug means different things

Even the consensus on keystone taxa breaks down once you leave industrialized populations. Two well-studied examples make this concrete.

Prevotella copri is a bacterial chameleon. In industrialized Western cohorts eating processed, high-fat, low-fiber diets, high abundance of P. copri is a red flag. It correlates with insulin resistance, type 2 diabetes, and rheumatoid arthritis. A Western doctor seeing high P. copri on a patient's microbiome test reasonably interprets it as evidence of inflammation-associated dysbiosis. But in rural, non-industrialized populations consuming high-fiber agrarian diets, high P. copri is a marker of vibrant metabolic health. Same species. Different host environment. Opposite clinical meaning. The bacterium did not change. Its substrate did.

Akkermansia muciniphila tells a similar story. It is broadly associated with gut barrier integrity, favorable metabolic markers, and improved response to immune checkpoint inhibitor therapy in cancer patients. It is being actively developed as a probiotic and is generally framed as universally beneficial. But in specific mouse models of colorectal cancer, elevated Akkermansia following antibiotic-mediated microbiome disruption actually worsened barrier function and exacerbated inflammation. The mechanism is that Akkermansia degrades host mucin, the protective mucus layer on the gut lining. In a healthy, balanced context, this turnover is beneficial. In an already-inflamed gut with compromised defenses, overabundant Akkermansia strips protective slime faster than the host can rebuild it.

The pattern repeats across the field. Microbial identity does not determine function. The function emerges from the species plus the substrate plus the community plus the host immune tone plus the redox environment. Geography, ethnicity, diet, and host state are not noise to be statistically corrected out of a dataset. They are fundamental parameters of the equation.

Which means the honest answer to "what is a healthy microbiome?" is that it depends on who and where you are. The three major indexes converge on shared keystone taxa. But the clinical meaning of any specific result depends on substrate, transit, and host context in ways no score currently captures.

The Postbiotic Thesis

Pull all of this together and a specific therapeutic argument starts to emerge. Zhang et al. shows that replacing the microbiome in Parkinson's patients measurably improves symptoms, but that the effect fades when replacement stops. FMT works, at least temporarily, and the mechanistic correlates (alpha-synuclein reduction, barrier restoration, dopamine metabolite shifts) suggest the active therapeutic signal is not just which species are delivered but what those species produce.

FMT has significant practical limitations. It requires donor screening, regulatory oversight, repeat procedures, and infrastructure most clinics cannot provide. Probiotic capsules, the mainstream alternative, cannot replicate the benefit in most cases. The reasons trace back to the biology we just walked through: a single strain (or even a blend of three or five) dropped into a dysbiotic gut immediately encounters a hostile chemical environment, reorganizes its proteome to survive, and cannot produce the community-emergent metabolites that only appear when a full ecosystem is functioning. You cannot deliver a community by delivering fragments of one.

The postbiotic argument inverts the therapeutic target. Instead of trying to seed the gut with the right bacteria and hoping they behave properly in a potentially hostile environment, you deliver the finished metabolic output of a healthy community directly. The short-chain fatty acids, bile acids, indoles, polyamines, and antioxidant molecules that collectively represent the chemistry a functioning ecosystem produces. These are the molecules that actually signal to the host immune system, the gut barrier, and downstream tissues.

For the Parkinson's conversation specifically, the postbiotic argument sits downstream of what Zhang et al. demonstrated. We are not claiming postbiotics treat Parkinson's disease. We are saying the Zhang data points at the mechanism, gut-derived metabolic signaling, that any future microbiome-based intervention would need to restore or supplement. The mechanistic plausibility is real. The human clinical evidence for direct postbiotic intervention in neurodegenerative disease is not yet there.

The Honest Limitations

Every paper discussed today carries caveats worth naming directly.

Menozzi et al. is cross-sectional. The coherence analysis depends on directional statistics that can amplify weak distributed signals. The GBA1 cohort is 43 people. PDMS-16 was derived inside a single dataset and has not been validated prospectively. The authors themselves acknowledge the score has "no immediate clinical usefulness." The paper advances the hypothesis that a prodromal microbial signature exists. It does not prove one can be used to identify individuals progressing toward disease.

Zhang et al. is a 72-person Phase 2 trial. The motor improvement is real and clinically meaningful, but the effect faded when dFMT stopped, meaning ongoing administration would be required for sustained benefit. This is not a cure. It is a temporary modulation of symptom severity. The trial was conducted in China with a Chinese donor pool, and whether the effects generalize to other geographic and dietary contexts is unresolved. Replication in larger, multi-center trials is needed.

Wilmanski et al. demonstrated a predictive blood metabolome signature, but one that breaks down in obesity. The host-state dependency limits how universal any blood-based microbiome readout can be.

The index wars (GMWI2, ZOE, HACK) converge on keystone taxa, but geographic and dietary confounders remain incompletely corrected. No score is currently operationalized as a standalone clinical diagnostic. The convergence on butyrate producers is the most robust signal across them, but even that leaves the butyrate disconnect unresolved: the producers present are not necessarily the producers producing.

The postbiotic thesis, as framed above, is supported by strong mechanistic evidence and proof-of-concept studies in C. difficile infection. It is not yet supported by human clinical trial data in neurodegenerative disease. The work to validate it is ongoing.

The Bottom Line

Two major 2026 papers on the gut-Parkinson's question arrived within weeks of each other, and they teach very different lessons. Menozzi et al. shows that sophisticated sequencing and careful cohort design can identify associations between microbial ecosystem state and clinical risk profiles. But statistical associations in a cross-sectional study are not predictions, and the PDMS-16 score is a hypothesis-generating tool, not a diagnostic. Zhang et al. shows that replacing the microbiome in drug-naïve Parkinson's patients measurably improves motor function and GI symptoms, proving the gut is an active participant in the symptom loop rather than a passive bystander.

Underneath both papers sits the question the field has not yet fully resolved: what counts as a healthy microbiome, and how do we measure it? The honest answer is that a century of microbial ecology is being rewritten in real time. Alpha-diversity is not the answer. The blood-gut mirror is context-dependent. The three major health indexes converge on shared keystone butyrate producers, but a high keystone count does not guarantee functional output. And geography, diet, and host state rewrite the rules of what any given species even means.

The gut is absolutely part of the Parkinson's story. Just probably not in the direction the hype implies. The next generation of microbiome science will live or die by whether the field commits to long-term longitudinal cohorts, multi-omic integration, and honest reporting of the limits of any single score. Until that data exists, the most responsible thing anyone working in gut health can do is name what is known, name what is not, and keep the difference between the two from getting blurred by a press release.

Stay curious. Take care of your ecosystem.

References

1. Menozzi E, et al. Microbiome signature of Parkinson's disease in healthy and genetically at-risk individuals. Nat Med. 2026.
2. Zhang R, Feng R, Wang J, et al. Gut microbiota modulation via repeated donor fecal transplantation improves motor and gastrointestinal symptoms in drug-naïve Parkinson's disease: a randomized phase 2 trial. Signal Transduct Target Ther. 2026;11:94. https://doi.org/10.1038/s41392-026-02604-9
3. Wilmanski T, Rappaport N, Earls JC, et al. Blood metabolome predicts gut microbiome α-diversity in humans. Nat Biotechnol. 2019;37(10):1217-1228. https://doi.org/10.1038/s41587-019-0233-9
4. DePaolo W. Commentary on Menozzi et al. 2026: A cross-sectional gradient is not progression. Substack. 2026. https://substack.com/@williamdepaolo/p-194933730
5. Chang Y, Gupta VK, Sung J, et al. Gut Microbiome Wellness Index 2 (GMWI2) enhances health status prediction from gut microbiome taxonomic profiles. Nat Commun. 2024.
6. Asnicar F, et al. Microbiome connections with host metabolism and habitual diet from 34,000 individuals: the ZOE PREDICT studies. Nature. 2025.
7. HACK index consortium. Health-Associated Core Keystone taxa identified across 45,424 gut microbiomes. Cell Rep. 2025.
8. McDonald JAK, Mullish BH, Pechlivanis A, et al. Inhibiting Growth of Clostridioides difficile by Restoring Valerate, Produced by the Intestinal Microbiota. Gastroenterology. 2018;155(5):1495-1507. https://doi.org/10.1053/j.gastro.2018.07.014
9. Ott SJ, Waetzig GH, Rehman A, et al. Efficacy of Sterile Fecal Filtrate Transfer for Treating Patients With Clostridium difficile Infection. Gastroenterology. 2017;152(4):799-811. https://doi.org/10.1053/j.gastro.2016.11.010
10. Rooks MG, Garrett WS. Gut microbiota, metabolites and host immunity. Nat Rev Immunol. 2016;16(6):341-352.
11. Roager HM, Licht TR. Microbial tryptophan catabolites in health and disease. Nat Commun. 2018;9(1):3294.
12. Routy B, Le Chatelier E, Derosa L, et al. Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors. Science. 2018;359(6371):91-97.

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