Microplastics are making mice fat. What about humans?

The 5 groups: group 1, clean water; group 2, 5 µm particles at 0.1 µg/ml; group 3, 5 µm at 1.0 µg/ml; group 4, 0.5 µm at 0.1 µg/ml; group 5, 0.5 µm at 1.0 µg/ml. Each was 2 cages of 5 male C57BL/6 mice (n = 10), 12 weeks old on purchase, with 1 week to settle. The authors estimated intake at 0.36 or 3.6 µg of particles a day, assuming 3.6 ml of water a day from metabolic-cage data. For the 0.5 µm particles at 1.0 µg/ml they state more than 100,000 particles per ml.^[1]

Weight: groups 3, 4 and 5 gained more than control at week 3; groups 3 and 5 at weeks 9 and 12; Figure 2A also marks group 3 at week 6. Groups 2 and 4 did not differ from control at week 12. Body fat was scanned once, at week 9: groups 3, 4 and 5 higher. The abstract says the weight gain came "with a corresponding increase in body fat" as early as 3 weeks; fat was not scanned at week 3.^[1]

Eating and moving: in a 24-hour metabolic chamber, group 5 against control, dark phase (6 PM to 6 AM), mean (SE): food 76.1 (6.8) against 90.3 (2.8) g/kg; water 90.8 (8.9) against 107.9 (4.7) ml/kg; total activity 698.3 (42.4) against 677.0 (38.4). None of these differed significantly.^[1]

Blood sugar: groups 2, 4 and 5 higher at week 3; only group 5 at week 12; the paper notes levels "varied and actually decreased" in later weeks, controls included. Glucose tolerance test (all groups, week 9) and insulin tolerance test (group 5 against control): no differences. Fasting insulin and HOMA-IR, a standard score for insulin resistance (when the body stops responding well to insulin): group 5 higher. Gut bacteria, group 5 against control, 20 stool samples at 12 weeks: lower beta diversity (PERMANOVA q = 0.028) and a relative rise of Firmicutes over Bacteroidetes; the authors note that other studies link a higher Firmicutes to Bacteroidetes ratio to obesity in mice and in people. The insulin chart's axis reads µg/ml, which is almost certainly a labeling error, so no insulin value is given here.^[1]

Blood fats: total cholesterol, LDL and triglycerides did not change; HDL rose in group 5 only, 43.7 (SE 2.0) against 35.7 (SE 2.2) mg/dL. Blood cells: eosinophils and basophils rose with the 5 µm particles only. Aorta, tested outside the body with the 0.5 µm groups combined: 1.75-fold more sensitive to phenylephrine (EC50 61.6 ± 10.7 against 35.2 ± 5.5 nM; n = 5 against 10), which the authors call a mild sensitization, with no clear loss of function in the vessel lining.^[1]

Fat around blood vessels, group 5 against control, 4 mice each, pooled RNA array: 5 genes up at least 1.5-fold and 14 down, with Dkk1 up 4.234-fold and Gata3 down to 0.051. The follow-up tests on single samples carry no p-values in Table 2. The authors read the pattern as consistent with more fat-cell formation and a shift toward storage fat.^[1]

Main experiment: 3 groups of 10 male C57BL/6 mice (2 cages of 5), 12 weeks old on purchase, 1 week to settle, then clean water or 0.5 µm or 5 µm particles at 1.0 µg/ml for 13 weeks; water intake 3 to 5 ml a day per mouse.^[2]

Weight: the Results report a "potentiated weight gain" with the 0.5 µm particles after 13 weeks, "primarily due to adipose tissue expansion", say the main 2022 weight result was "repeated in the current treatment", and state that the 5 µm particles did not produce a significant extra gain. The Abstract says the plastic-water mice gained extra weight without naming a size; this article follows the Results. The weights are only in Supplementary Figure 1.^[2]

Fat tissue: F4/80-positive macrophages rose in epididymal fat with 0.5 µm particles (p < .05; n = 8 to 10), not with 5 µm; the M1 rise was not significant (p = .06); M2 macrophages and T cells did not differ; adiponectin, resistin, leptin and PAI-1 did not differ (n = 10); fat genes showed minor changes in belly fat and a mild rise of Mcp-1 and TNFα in fat around vessels (Table 1, no p-values). The authors conclude the fat "did not appear to be dysfunctional" at this time point.^[2]

Liver: normal on the microscope slide, no fat droplets or inflammation; liver weight as a share of body weight fell, from 5.01 ± 0.29% to 4.56 ± 0.22% (0.5 µm, p = .0165) and 4.33 ± 0.50% (5 µm, p = .0005); triglycerides unchanged; cholesterol rose, from 1.89 to 2.34 mg/g (0.5 µm) and 2.21 mg/g (5 µm): 24% more with the small particles (2.34 ÷ 1.89 = 1.238, rounded to 24%) and 17% more with the large particles (2.21 ÷ 1.89 = 1.169), our arithmetic from the paper's printed averages; ALT and AST unchanged; glutathione unchanged. Liver genes changed (q < 0.05, n = 10): 283 with 0.5 µm (203 up, 80 down) and 49 with 5 µm (41 up, 8 down), 5 shared. Pathway analysis flagged "FXR/RXR activation" and "LXR/RXR activation" with both sizes.^[2]

Bile acids: in the liver, 0.5 µm particles lowered TCDCA (0.57-fold, p = .02) and THDCA (0.33-fold, p < .05); 5 µm particles raised HDCA (1.93-fold, p = .0002) and THCA (2.02-fold, p = .03). In the cecum, 0.5 µm particles raised apocholic acid (2.17-fold, p = .005); 5 µm particles raised UDCA (1.87-fold, p = .03) and γ-muricholic acid (3.11-fold, p = .03).^[2]

Delphinidin test: delphinidin chloride, a natural pigment found in berries, 20 mg/kg, injected into the belly cavity twice a week for 4 weeks, in mice on 0.5 µm particle water; clean water with no injection (n = 4), particles plus dummy injection (n = 8), particles plus delphinidin (n = 8). The delphinidin mice gained less weight and carried less body fat than the dummy-injection mice. The text gives no values. Figure 4 marks delphinidin against the dummy shot for weight and fat, and delphinidin against clean water for weight. It shows no mark between clean water and particles plus dummy shot in this 4-week test, so it does not show plastic weight gain on its own. Nothing was given in food or drink. The small intestine showed no damage or inflammation.^[2]

Both studies started the same way. The team bought male mice that were 12 weeks old and gave them a week to settle in. Then it picked cages at random: 2 cages of 5 mice for each group.^{[1, 2]}

The particles came from a lab supplier (Creative Diagnostics), new and clean. Before each batch of water, the team shook the particle stock with sound waves for 30 minutes to break up clumps. It made fresh particle water every other day, and on the days in between it shook the bottles so the particles would not settle. The clean-water mice drank filtered, sterilized water.^{[1, 2]}

Every 3 weeks, each mouse went on the scale. In 2022 the team also checked fasting blood sugar every 3 weeks, scanned body fat at week 9 and collected stool to read the gut bacteria. In 2024 the fat scan came every 3 weeks, and at the end the team collected blood, belly fat, fat around blood vessels, liver, cecum and small intestine. Neither paper says whether the researchers knew which group each mouse was in.^{[1, 2]}

Zhao 2024 names its proposal in its title: a disrupted gut-liver-adipose axis. It reports enrichment of "FXR/RXR activation" and "LXR/RXR activation" in the liver with both particle sizes, and states that "The mechanism(s) driving the increased weight gain and adiposity in PS-fed mice remain unclear." It found no drop in beta-catenin and reads that as a sign that disrupted Wnt signaling is not a major cause. It names gut bacteria and their products as another candidate, which it says needs more study. Gut bacteria were measured in 2022, not in 2024.^[2]

The Beijing test of the gut link (Zhai 2024, China Agricultural University) used male mice on a high-fat diet with 5 µm polystyrene at 10 mg/L in their water. When antibiotics wiped out their gut bacteria, the plastic no longer made a difference to weight. Then the team moved stool from high-fat plastic mice into high-fat-fed mice whose own gut bacteria had been cleared with antibiotics; those mice put on more weight than mice given stool from high-fat mice on clean water. On a normal diet, plastic made no difference to weight gain.^[4]

The 2022 paper prints no weights, body-fat shares, blood-sugar or insulin values as numbers. They appear only as charts. This article prints no value read off a chart.^[1]

The 2024 paper's 13-week weight numbers are in Supplementary Figure 1, part of a supplementary data file on the journal and PubMed Central pages. Our automated download met a bot check, which we did not try to get around, so no 2024 weight number appears here.^[2]

The 2024 paper's funding note reads "does represent official views" of the National Institutes of Health, where "does not" was almost certainly meant.^[2]

Dose, in the researchers' own words: in 2022 the most effective dose and particle size (0.5 µm particles at 1.0 µg/ml), over 12 weeks, "would be at the higher end of estimated human consumption"; in 2024 the dose is "in a range of estimated human exposures", citing two published estimates. This article does not convert a mouse dose into a human one.^{[1, 2]}

Huang 2023 (Tsinghua and Peking universities; abstract only, paywalled): mice on high concentrations lost significant weight, the lowest concentration showed little change, and relatively low concentrations made mice overweight, with more fat, a better appetite and lower activity. The abstract gives no doses, particle size, route, strain, sex, group sizes or length. Zhao 2024 cites Huang 2023 twice, for the weight gain and for gut bacteria, and not for the weight loss.^{[2, 3]}

Lu 2018 (Zhejiang University of Technology, Hangzhou; abstract only, paywalled): male mice exposed by mouth to 0.5 and 50 µm polystyrene for 5 weeks; at 1,000 µg/L (the same as 1 µg/ml) both sizes lowered body, liver and fat weight, and lowered liver triglycerides and liver cholesterol. The route (drinking water) and the number of mice (40 in all) come from the Chartres 2024 review's table.^{[5, 8]}

Okamura 2023 and Hasegawa 2024 (one research group, Kyoto Prefectural University of Medicine; full text): male C57BL/6J mice, 0.45 to 0.53 µm polystyrene in water at 1,000 µg/L for 4 weeks (normal or high-fat diet), and at 1,000 or 5,000 µg/L for 6 weeks (normal diet); no body-weight difference. In both studies each mouse got an equal portion of food (in 2023, of water too), which can hide a gain that comes from eating more.^{[6, 7]}

Zhai 2024 (China Agricultural University, Beijing; full text): male C57BL/6J mice, 6 per group, 5 µm polystyrene at 10 mg/L in drinking water for 8 weeks. On a normal diet, weight gain was similar with and without plastic. On a high-fat diet, body weight reached 175.21% ± 5.35 of the starting weight with plastic against 158.17% ± 8.25 without. Wiping out the gut bacteria removed the difference, and moving stool from high-fat plastic mice into high-fat-fed mice whose gut bacteria had been cleared with antibiotics carried the weight gain over.^[4]

People: Hong 2024 (121 young adults, grouped by how often they ate take-out food; abstract only) linked more plastic in stool with higher body mass index (odds ratio 1.264, 95% CI 1.108 to 1.441; the unit behind it is not stated). It measured each person once, so it is a snapshot, and take-out food can itself affect weight. Zhen 2026 (1,080 ten-year-olds in the Ma'anshan Birth Cohort; abstract only) found plastic in the urine of 91.1% and sex-specific links with both overweight and underweight; when the urine was collected is not stated in the abstract. Castrogiovanni 2026 (29 children in Cyprus) found no link with body mass index for age, and Schwenger 2024 (23 adults) found no difference in stool plastic between lean and obese groups. None of these can show cause, and Zhao 2024 does not discuss the studies that saw no gain or saw loss.^{[9, 10, 11, 12]}

Zhao 2022 was funded by US National Institutes of Health grants and the Jewish Heritage Fund for Excellence; the authors declared no competing interests. Zhao 2024 was funded by National Institutes of Health grants and the University of Louisville School of Medicine; the authors declared no conflicts. We found no correction, erratum, retraction or published dispute for either paper on PubMed, Europe PMC or Crossref as of 22 September 2026. PubPeer was not checked.^{[1, 2]}

This version was shortened and rewritten in plainer words, and the section on whether animal studies matter for humans now comes first. The two at-a-glance boxes became one. How the tests were run, the berry-pigment test and the numbers behind the liver-genes chart moved into this technical record or the questions above, and the studies that found no gain are set out in full under the wider evidence. Every figure that left the story is printed in these sections, from the papers' own text and tables.