A glyphosate test looks like the start of an answer. It is not. The result reports what you ate yesterday, it carries no reference range, and nothing in it describes the bacterial community this compound was designed to act on.
Chapter 22 of The Angry Gut, The Antibiotic on Your Salad, lays out the mechanism. What follows is the measurement problem underneath it: which numbers exist, which compartment they came from, and what terrain work is left when no panel can score an exposure.
What a urine glyphosate test is reporting
The population figure is the strongest fact in the file. Across 2,310 urine specimens in a nationally representative American survey, glyphosate was detectable in 81.2% of people aged six and over, at a median of 0.392 micrograms per liter. Levels ran significantly lower in those who had gone more than eight hours without eating.
Fasting moves the tail, not only the middle. Landing above the 95th percentile was 1.94 times more likely in people who had fasted eight hours or less, interval 1.06 to 3.54. Children carry more of it than their parents do. Read that as evidence about the route, not a body burden: skipping breakfast is no cleanse, it is proof the exposure arrives on a plate and leaves fast.
The compartment nobody sampled
The rodent experiment everyone cites deserves its full arithmetic, because the arithmetic is the finding. The animals drank it for ninety days at 0.5, 50 and 175 mg/kg of body weight per day. The lowest arm is the whole European daily allowance, and that allowance sits roughly 33 times above the maximum measured chronic dietary intake. So a dose regulators accept means the legal ceiling, not the supermarket.
What moved was modest. The species count inside each animal did not differ between groups, though beta diversity, which compares animals with each other, separated treated from control, and the most abundant bacteria never budged. The organisms that shifted were minor residents: Shinella zoogleoides, Acinetobacter johnsonii and Eggerthellaceae, with Akkermansia muciniphila up only in the animals given the bottled product rather than the molecule.
The metabolites carry the measurement lesson. Shikimic acid and 3-dehydroshikimic acid accumulated in the cecum, which is substrate backing up behind a blocked enzyme. In blood, the metabolites that shifted pointed to nicotinamide, branched-chain amino acid, methionine, cysteine and taurine metabolism, read by the investigators as an oxidative stress response. A routine serum panel would have called every one of those animals normal.
The same laboratory ran a second rodent study that year on a low-dose mixture of six pesticides, glyphosate among them, and the bacteria did not change. What moved sat underneath, in the liver: 257 genes with altered expression and 4,255 differentially methylated CpG sites. No chemical can be pulled back out of a six-chemical mixture, so that is no clean null, but the endpoint I care about most is the one that stayed still.
A gene list is not an activity report
Two computational surveys asked whether the target enzyme sits in your flora and came back opposed. One sorted the enzyme into sensitive and resistant forms and put 54% of core gut species in the sensitive column. The other read expression rather than presence: among 44 subspecies reference genomes covering 72% of total assigned microbial abundance across 2,144 human fecal metagenomes, 9 carried the resistant form, the pathway was largely quiet in settled adult guts, and some Proteobacteria may even degrade the compound, which would make part of your flora a sink rather than a target.
The classification tool states its own reach: it can sort sequences from nearly 90% of eukaryotes and more than 80% of prokaryotes. Classifying is not killing. This is the trap inside every stool panel sold as a microbiome report, because the census is not the activity, and a species list of any length says nothing about what that population is doing with your lunch. It is why a low stool short-chain fatty acid result is not a deficiency.
What the simulator showed, including the part nobody quotes
A human-microbiota simulator seeded with an infant community was run at 100 mg/L. Fermentation shifted toward lactate and acetate and away from butyrate, propionate and valerate. Overall fermentation activity rose and acidified the vessel, mostly under the bottled product, and ammonium production reflecting proteolytic breakdown rose with it.
Then the result that almost never travels with the citation. Lactobacilli grew, stimulated by that acidification, which is the opposite of the claim popular writing repeats most. I print it because it is evidence against the side I argue for. Hold the tier honestly as well: a plastic reactor at an antimicrobial screening concentration, orders of magnitude above what arrives with food, cannot set a dose. What it does show is where repair has traction, because what moved was fermentation, and fermentation is the part of this system you can feed. That is the workable end of a low-grade inflammatory terrain.
Read the regulator’s answer in full
The European peer review did address the microbiome question, and both camps misquote it. Its position is that standardized guidance and harmonized criteria for assessing a microbiome do not exist, so no definitive conclusions can be drawn from the published studies, and that the mammalian toxicity dataset is trusted to cover any health effect mediated that way. The same review raised the acute reference dose threefold to 1.5 mg/kg of body weight and identified no critical areas of concern.
Strip out the politics and one sentence survives: no validated method exists for this endpoint, which is a fact about how the class is regulated rather than a verdict about your gut.
The one human endpoint, and what it is attached to
A single analysis pairs a measured internal dose in ordinary people with a physiological outcome. Urinary glyphosate tracked with a liver fibrosis index at a coefficient of 0.09, interval 0.06 to 0.12, as the leading contributor in a mixture of non-persistent pesticides. Handle it with tweezers. A chlorpyrifos breakdown product produced the same point estimate in the same analysis, a spot urine measures the last day of intake while fibrosis takes years, and the fibrosis was inferred from serum indices rather than tissue. That is a correlation which should start a study, not end one, and the sort of signal a proper metabolic audit exists to put in context.
What to repair when there is nothing to measure
So the field offers a mechanism measured in animals, a population exposure figure, a rodent cecum with substrate stacked behind an enzyme, and no test to order. Know that before anyone sells you a panel for it.
What remains is terrain, and terrain has numbers. Feed the guild that ferments fiber, since fermentation is what the reactor data disturbed. Change the source of the food you eat raw, since the exposure comes with meals and clears fast, which makes sourcing a better lever than any cleanse. And name the economic driver, because it explains the silence better than a villain does: the compound is cheap and effective, the assay that cleared it was designed before anyone thought a flora worth measuring, and no budget pays for the compartment where the effect turns up.
Then the history question that outranks any assay. Ask what stripped your bacterial community down to bare ground: courses of antibiotics, an abdominal operation, an inpatient stay. A quiet pathway in a settled gut says nothing about a gut rebuilding itself, and rebuilding is when a community has to synthesize its own amino acids. Every study and every number sits in the Chapter 22 Deep Dive, which spells out what each finding shows and what it does not. Nothing here is a reason to start or stop a medication. Next comes a fluid your body makes on a schedule to keep the upper gut thin, and what the number that sends a gallbladder to surgery is made of.
Frequently asked questions
Is a urine glyphosate test worth ordering?
Not for a clinical decision. It reflects the last day or two of meals, it has no target range, and no study links a person’s level to their gut community or to a treatable finding. Fasting alone moves it, which tells you how unstable it is. A test earns its place when the result changes what happens next. That is the standard every marker here has to clear.
Does a glyphosate detox protocol work?
No protocol has been tested against a measured outcome in people, so anything sold as one is selling a story. The exposure is not banked for years the way the word detox implies; it comes in with food and leaves quickly. What has mechanism behind it is changing the source of raw produce and rebuilding fermentation capacity. The community itself is the repairable target.
Can a stool microbiome test show pesticide damage?
It cannot. A stool panel names organisms; it does not report what they are doing, and no reference data exist for this exposure at all. Two computational surveys of the same question reached opposite conclusions, and neither one measured a living community. Additive research shows the same gap between a species list and a behavior. Emulsifiers are where that gap gets expensive.
Could this be why my liver markers are drifting up?
The one human analysis linking urinary glyphosate to a liver fibrosis index is cross-sectional, rests on a single urine sample, and shares its point estimate with an unrelated insecticide metabolite. Drifting liver markers have causes that are measurable, repeatable and far more common, and those deserve the first look. A normal liver enzyme is not the same as a healthy liver.
Measure the terrain, not the residue
When an exposure has no test and no reference range, the work moves to the markers that have both. A metabolic and inflammatory workup shows whether your terrain can rebuild a bacterial community, and what is standing in the way.
Questions? Call (314) 295-3000 or text (314) 886-5902.
Sources
- European Food Safety Authority (EFSA), Alvarez, F., Arena, M., Auteri, D., Binaglia, M., Castoldi, A. F., Chiusolo, A., Crivellente, F., Egsmose, M., Fait, G., Ferilli, F., Gouliarmou, V., Herrero Nogareda, L., Ippolito, A., Istace, F., Jarrah, S., Kardassi, D., Kienzler, A., Lanzoni, A., … Villamar-Bouza, L. (2023). Peer review of the pesticide risk assessment of the active substance glyphosate. EFSA Journal, 21(7), e8164. https://doi.org/10.2903/j.efsa.2023.8164
- Leino, L., Tall, T., Helander, M., Saloniemi, I., Saikkonen, K., Ruuskanen, S., & Puigbò, P. (2021). Classification of the glyphosate target enzyme (5-enolpyruvylshikimate-3-phosphate synthase) for assessing sensitivity of organisms to the herbicide. Journal of Hazardous Materials, 408, 124556. https://doi.org/10.1016/j.jhazmat.2020.124556
- Mesnage, R., & Antoniou, M. N. (2020). Computational modelling provides insight into the effects of glyphosate on the shikimate pathway in the human gut microbiome. Current Research in Toxicology, 1, 25-33. https://doi.org/10.1016/j.crtox.2020.04.001
- Mesnage, R., Calatayud, M., Duysburgh, C., Marzorati, M., & Antoniou, M. N. (2022). Alterations in infant gut microbiome composition and metabolism after exposure to glyphosate and Roundup and/or a spore-based formulation using the SHIME technology. Gut Microbiome (Cambridge, England), 3, e6. https://doi.org/10.1017/gmb.2022.5
- Mesnage, R., Teixeira, M., Mandrioli, D., Falcioni, L., Ibragim, M., Ducarmon, Q. R., Zwittink, R. D., Amiel, C., Panoff, J.-M., Bourne, E., Savage, E., Mein, C. A., Belpoggi, F., & Antoniou, M. N. (2021a). Multi-omics phenotyping of the gut-liver axis reveals metabolic perturbations from a low-dose pesticide mixture in rats. Communications Biology, 4(1), 471. https://doi.org/10.1038/s42003-021-01990-w
- Mesnage, R., Teixeira, M., Mandrioli, D., Falcioni, L., Ducarmon, Q. R., Zwittink, R. D., Mazzacuva, F., Caldwell, A., Halket, J., Amiel, C., Panoff, J.-M., Belpoggi, F., & Antoniou, M. N. (2021). Use of shotgun metagenomics and metabolomics to evaluate the impact of glyphosate or Roundup MON 52276 on the gut microbiota and serum metabolome of Sprague-Dawley rats. Environmental Health Perspectives, 129(1), 17005. https://doi.org/10.1289/EHP6990
- Ospina, M., Schütze, A., Morales-Agudelo, P., Vidal, M., Wong, L.-Y., & Calafat, A. M. (2022). Exposure to glyphosate in the United States: Data from the 2013-2014 National Health and Nutrition Examination Survey. Environment International, 170, 107620. https://doi.org/10.1016/j.envint.2022.107620
- Shu, S., Li, Y., Yu, X., Chen, X., Abdullah, U., & Yu, Y. (2025). Association between mixed exposure of non-persistent pesticides and liver fibrosis in the general US population: NHANES 2013-2016. Ecotoxicology and Environmental Safety, 290, 117776. https://doi.org/10.1016/j.ecoenv.2025.117776
- Walsh, L., Hill, C., & Ross, R. P. (2023). Impact of glyphosate (Roundup) on the composition and functionality of the gut microbiome. Gut Microbes, 15(2), 2263935. https://doi.org/10.1080/19490976.2023.2263935