What the Human Studies Actually Show
This chapter maps what living-human studies have reported about microplastics and nanoplastics in the body: where particles were sampled, which health findings appeared alongside them, and how the study design limits the conclusion. The practical goal is to read a body-site result as evidence with a denominator, comparator, and uncertainty—not as a population-wide diagnosis.
A wide map built mostly from snapshots
A recent systematic review searched five major databases and included 25 original studies that quantified MNPs in biological samples from living people. The studies covered five broad areas: cardiovascular, reproductive, gastrointestinal, respiratory, and ocular research. That range matters. It shows that researchers are finding ways to investigate MNP exposure in real human bodies rather than relying only on environmental samples or laboratory models.
It does not mean that the same amount of evidence exists for every organ, or that the studies estimate how much illness MNPs cause in the general population. The evidence map is wide, but most of its points are small and observed at one moment.
The design mix makes that clear. Fourteen studies were cross-sectional: they measured exposure and outcome around the same time. Nine were case–control studies, comparing people with a condition or clinical feature with a comparison group. One was quasi-experimental, and one was a prospective cohort. A cross-sectional study can show that a higher measured burden and a health difference occur together. A case–control study can show that the two groups differ. Neither design, by itself, establishes that exposure came first. The prospective study has more temporal information, but one longitudinal study cannot resolve every issue shared by the field.
The participant numbers also need to be read in context. The review grouped five cardiovascular studies with 454 participants, seven reproductive studies with 327, nine gastrointestinal studies with 537, three respiratory studies with 171, and one ocular study with 49. These figures describe the studies included in each research area; they are not a single representative sample of the population, and the groups should not be treated as if they were necessarily independent of one another. Some participants were healthy volunteers. Others were already patients, including people with acute coronary syndrome, pregnancy complications, airway disease, or ocular conditions.
Ten of the 25 studies lacked control groups. Many were conducted at one or two centres, and some recruited volunteers. The research was also concentrated geographically, especially in China. Those features do not invalidate the observations. They do affect how far the findings can travel. A result from a patient group in one centre may identify an important signal without telling us whether the same association occurs in people with different diets, occupations, healthcare access, genetic backgrounds, or environmental exposures.
The design is therefore part of the result. If a paper reports a higher MNP burden in people with a disease, the reader needs to know whether the study followed participants before the disease appeared, compared them with suitable controls, measured the relevant tissue, and used an exposure method that could see the particles of interest. Without that context, “higher” sounds more conclusive than it is.
The review’s selection flow also contains internal inconsistencies in its published text. The Results section and the reproduced flow-diagram caption give different numbers for the screened records, and the caption gives a different search year from the Methods. The table also contains a publication date outside the stated search window. These discrepancies do not remove the stable conclusion that the review included 25 human studies and found a predominantly observational, heterogeneous evidence base. They are a reminder not to build a strong argument on a fragile administrative count when the substantive limitation is already clear from the study designs.
The design categories also describe different kinds of uncertainty. A cross-sectional study can be useful for discovering a pattern: perhaps people with a higher measured burden also have a different biomarker. But because the exposure and outcome are observed together, the study cannot show whether the burden preceded the biomarker, whether the health state changed what was retained or excreted, or whether a third factor influenced both. A case–control study adds a comparison between people with and without a condition, but the choice of participants and the condition itself can shape what exposure is observed. If the groups differ in occupation, diet, treatment, or healthcare setting, those differences may travel with the MNP measurement.
A prospective cohort improves the temporal question by measuring people before an outcome is recorded. That is a meaningful advance, not a guarantee of causation. Exposure still has to be measured well, relevant confounders still need attention, and one cohort may not represent other populations. A quasi-experimental study can examine a change in exposure or circumstance, but it is not automatically equivalent to a randomized experiment. The review’s design mix therefore tells us why the human evidence is valuable and why it remains early: the studies are close to real life, but most were not built to isolate one cause from a changing environment.
The review’s risk-of-bias assessment makes the same point in another language. Of the 25 studies, 11 were judged low risk overall, eight had some concerns, and six were judged high risk. The most frequent problems involved confounding, exposure measurement, and selective reporting. “Low risk” does not mean that a study found a low health risk; it means the reviewers saw fewer problems in the assessed study methods. “High risk” does not prove that the association is false. It means that important alternative explanations or measurement weaknesses make the estimate less secure. Reading the labels this way prevents a common error: turning methodological quality into a direct health verdict.
The denominators matter at every level. The review’s cardiovascular total of 454 participants does not mean that 454 randomly selected adults were tested for future heart disease. Some were patients with vascular disease or acute coronary syndromes, while others served as controls. The reproductive total combines distinct samples and outcomes, from semen and urine to placenta and amniotic fluid. A number attached to a tissue sample cannot be used as though it were a blood concentration, and a group of patients cannot be treated as though it were a general-population survey. Before asking whether a result is large, ask “large compared with whom, measured where, and using what outcome?”
This is also why detection in many compartments should not be mistaken for proof that MNPs have spread evenly through every body or carry the same risk everywhere. Researchers choose samples that are available, clinically relevant, or capable of answering a particular question. A positive result in an atheroma shows material in that lesion; a result in stool shows material being excreted or passing through the digestive tract; a result in semen or placenta concerns a particular reproductive compartment. Those observations can be connected into a research hypothesis about distribution, but they are not interchangeable measurements of whole-body burden.
The appropriate conclusion is consequently more specific than either “MNPs are harmless” or “MNPs are causing widespread disease.” The review shows that human researchers have identified exposure signals and investigated health relationships in several systems. It also shows that the strength of a claim changes with the design, comparator, sample, endpoint, and method. Keeping those pieces attached is what turns a catalogue of findings into evidence that can be judged.
The map is best understood as a chain. A design determines what comparison is possible. The sample determines which part of the body is represented. The analytical method determines what burden can be seen. The outcome determines whether the result is a biomarker, symptom, anatomical feature, or clinical event. Only then can the association be interpreted. A gap at any link narrows the claim; it does not automatically erase the observation.
Signals by organ system are not interchangeable
The review describes the most developed human signals in cardiovascular and reproductive research, while gastrointestinal, respiratory, and ocular findings remain more emerging. That does not create a simple league table of harms. It reflects differences in the number of studies, the available samples, the outcomes measured, and the quality of comparisons.
Cardiovascular findings. Five studies examined blood, thrombi, atheromas, or related vascular material. The reported patterns included higher MNP burdens alongside inflammatory markers, altered coagulation measures, greater vascular complexity or stenosis, and adverse cardiac events. One line of evidence found particles embedded in thrombi, while other studies associated burden in atherosclerotic plaques with later myocardial infarction, stroke, or death. These observations make cardiovascular effects an important question, but they do not show that the particles initiated the vascular disease. A damaged or inflamed lesion might be more likely to trap circulating particles. The disease, its treatment, or a shared exposure might also influence the measured burden and the clinical outcome.
The outcome matters too. An inflammatory marker is not the same as a heart attack, and a complex vascular score is not the same as a confirmed event. A study can report a statistically clear relationship with a surrogate or intermediate measure without establishing a change in a patient-relevant endpoint. The cardiovascular results therefore raise concern and motivate follow-up; they do not yet provide a general probability that a detected burden will cause an event.
Reproductive findings. Seven studies examined the genitourinary system or reproductive tissues. In male samples, researchers reported PS, PVC, PE, and PP in semen and urine, with associations between particular detected materials and reduced sperm count, concentration, or motility. In female reproductive research, MNPs were reported in urine, placenta, amniotic fluid, uterine fibroids, myometrium, cervical-cancer tissue, and prostate-related samples. Some studies linked higher placental burden with impaired fetal growth or measured particles in amniotic fluid alongside gestational age.
The reproductive evidence contains an important counterexample to a one-directional story. One study found no difference in urinary MNP levels between people with endometriosis and controls. That finding does not establish that MNPs have no role in endometriosis; the study did not measure the particles inside the ectopic lesions, and the relevant exposure may not be represented by urine. At the same time, a positive finding in placenta or tumour-associated tissue does not establish that particles caused impaired growth or cancer. Tumours and other abnormal tissues can have altered blood supply, cell activity, and uptake pathways that passively accumulate circulating material.
Gastrointestinal findings. Nine studies examined fecal or gastric material and related outcomes. The review reports associations with inflammatory bowel conditions, metabolic-associated steatotic liver disease, obesity-related measures, and changes in the gut microbiome. These findings are biologically interesting because the digestive tract is an obvious interface with ingested material. But they are also particularly exposed to confounding. Seven of the nine gastrointestinal studies did not control for diet, body mass, and exercise, all of which can influence gut health and may also track food packaging, bottled-water use, or other exposure patterns.
A microbiome difference is not automatically a disease, and a disease-associated microbiome is not automatically evidence that MNPs caused it. The direction of influence can run more than one way: diet may alter both the microbiome and the amount of plastic encountered; a gastrointestinal condition may change transit and excretion; or another environmental exposure may affect all three. The studies identify relationships worth testing with better controls and repeated measurements.
Respiratory findings. Three studies reported MNPs in airway-related samples and associations with allergic rhinitis, chronic rhinosinusitis, nasal microbiome changes, or interstitial lung disease. These results fit inhalation as a plausible route, but the review notes that higher MNP load was not accompanied by a clear dose–response relationship with symptom severity in the upper-airway studies. In one interstitial-lung-disease study, particles were reported in bronchoalveolar lavage fluid but not in blood. That may point to a local airway process, or it may reflect the technical difficulty of detecting particles in blood. It cannot, on its own, establish a systemic or pulmonary causal pathway.
Ocular findings. The most preliminary signal comes from a single study of 49 patients with existing ocular conditions. It reported MNPs in vitreous humour and a positive relationship between burden, intraocular pressure, and aqueous opacities. Those measurements are relevant to eye health, but the participants were not a healthy population followed from exposure to disease. Their pre-existing conditions could affect both the local environment and visual outcomes. One study can open a research question; it cannot establish that MNP exposure causes glaucoma or vision loss.
Across all five areas, the same rule applies: keep the sample, measurement, comparator, and outcome together. “MNPs were found in atherosclerotic tissue” is a localization claim. “Higher burden was associated with inflammatory markers” is an association claim. “MNPs cause cardiovascular events” is a causal claim that requires additional evidence. Similar wording can conceal very different evidentiary steps.
The review did not perform a meta-analysis because the studies were too heterogeneous. They used different samples, populations, analytical methods, particle definitions, health outcomes, and study designs. Combining them into one number could create the appearance of precision without solving the underlying comparability problem. Heterogeneity does not mean that every result is unreliable. It means the sources do not justify treating unlike findings as measurements of one common effect.
That is the most useful conclusion from the human map. MNP research has moved beyond a single compartment: particles and polymer signals have been reported in multiple human samples, and several studies have found concerning associations. But a broad map of signals is not the same as a causal estimate of disease burden. Reproductive findings make that distinction especially vivid because human tissue observations sit beside a much larger literature on possible biological mechanisms. The next question is what those mechanisms add—and what they still cannot prove about people.