How Do You Know It Is Plastic?
For research on microplastics and nanoplastics in people, the question “were particles found?” is only the beginning. This chapter focuses on the measurement chain that turns blood, stool, semen, airway fluid, or another biological sample into a reported MNP result. Understanding that chain lets you ask the more useful questions: what was actually measured, what could have been missed, and whether two numbers are comparable.
A measurement is a chain, not a single scan
A human sample does not arrive at a machine carrying a label that says “plastic.” Researchers must prepare it, separate or expose material within it, identify a polymer signal, and decide how to express the result. Each step can add information, remove information, or introduce uncertainty. The reported result is therefore not just a property of the particles. It is also a property of the method used to look for them.
The first choice is what kind of question the study wants to answer. Is it asking which polymers are present? How many particles are present? How much polymer mass is present? What sizes and shapes do the particles have? Where in a tissue are they located? These questions overlap, but no single technique answers all of them equally well.
Spectroscopic methods identify materials through how they interact with light. Raman spectroscopy measures inelastic scattering and produces a molecular “fingerprint.” Fourier-transform infrared spectroscopy, or FTIR, detects infrared absorption associated with chemical bonds. Laser-directed infrared spectroscopy, or LDIR, uses a quantum-cascade laser to scan larger areas and combine chemical information with automated images. In the human review, Raman was used in eight of the included studies, LDIR in seven, and FTIR in three. Microscopy was often added to characterize particle size, shape, colour, or position.
These methods can preserve a particle as a particle. They may provide a count, an approximate size distribution, and sometimes shape or location alongside a polymer identification. That makes them useful when the biological question involves individual particles or their position in tissue. But their sensitivity is not unlimited. Raman signals can be weak, and biological pigments or other materials can create fluorescence that obscures the molecular signal. FTIR generally performs best for particles above roughly 20 micrometres, and LDIR has a similar practical limitation in the review’s comparison. Smaller particles can therefore be undercounted or missed, not necessarily because they are absent, but because the method cannot see them reliably in that sample.
Pyrolysis–gas chromatography/mass spectrometry, usually shortened to Py-GC/MS, takes a different route. Instead of inspecting intact particles, it thermally decomposes the sample. The resulting chemical products are separated and identified, allowing the method to estimate the mass of a polymer. This is a strength when the question is how much polymer is present regardless of particle size. It is also a fundamental trade-off: because the sample is destroyed, the method does not preserve the original particle count, size, shape, or location. In the human review, Py-GC/MS was the most frequently used approach, appearing in nine studies.
That difference explains why two valid methods can produce different-looking answers from related material. A spectroscopic method might report a number of particles above its detection limit. A microscopy-assisted result might add size and shape. Py-GC/MS might report polymer mass while saying nothing about how that mass was divided among particles. None of those outputs is automatically the “real” quantity and the others a mistake. They are answers to different measurement questions.
A simple example makes the distinction concrete. Imagine two samples with the same total polymer mass. One contains a few larger particles; the other contains many smaller ones. A particle-count method could distinguish them, while a mass method could report similar totals. The reverse problem also occurs: two samples with a similar particle count could contain very different total mass if their particles differ in size. A result expressed in particles per gram is not interchangeable with one expressed as polymer mass per gram, even if both use the same biological material.
The difference matters especially for nanoplastics. Smaller particles may have different surface area, cellular uptake, and ability to cross biological barriers than larger microplastics. Yet the human review notes that current in vivo studies rarely distinguish nano- from micro-scale particles because the analytical task is difficult in complex biological material. A study that identifies a polymer but cannot resolve its smallest particles has not necessarily failed; it has answered a narrower question than a headline about “all plastic particles” might imply.
To interpret any result, it helps to separate four questions that are often compressed into one sentence. Identity asks whether the signal matches a plastic polymer rather than merely resembling one. Visibility asks which particles the method could see: a size range, a signal strength, and a biological matrix all matter. Quantity asks whether the output is a count, an estimated mass, or a direct mass measurement. Context asks where the material was found and how the sample was expressed—for example, per gram of tissue, per volume of fluid, or in a localized region. A result can be strong on one layer and weak on another.
This separation also clarifies what “characterization” adds. Identifying polyethylene in a sample answers a chemical question. Showing that it occurs as particles of a particular approximate size and shape answers a physical question. Showing that those particles sit in a vascular lesion or reproductive tissue answers a location question. None of these observations automatically answers the biological-effect question. They become more informative when a study connects them to a measured outcome, but the connection still depends on the study design and the reliability of each earlier step.
Visibility is especially easy to mistake for abundance. If an infrared method works mainly on larger particles, a sample containing many smaller particles may appear to have a low burden. If a pyrolytic method captures polymer mass from particles of many sizes, it may report material that a particle-imaging method could not count. The two results can both be accurate within their measurement windows. The apparent disagreement may describe different windows onto the same exposure rather than a conflict about whether plastic exists.
The denominator matters as well. A value reported per gram of tissue describes a concentration relative to the portion analyzed; it is not automatically the total burden in the organ or body. A value per litre of fluid describes a different sample geometry. Tissue heterogeneity can make the sampled portion important, while fluid measurements may reflect what is circulating or being excreted at that moment. Without the denominator and sample context, a number cannot be translated into a person-wide internal dose.
Using more than one technique can narrow uncertainty, but it does not erase it. Spectroscopy paired with microscopy can support both polymer identity and particle-level description. Py-GC/MS can add a mass estimate that is less dependent on the size of each particle. If the methods disagree, the disagreement can reveal a preparation or matrix problem; if they agree, confidence in the analytical observation may improve. Neither outcome alone proves that the particles caused a health change. Measurement establishes the object of study so that the causal question can be investigated honestly.
The measurement chain also explains why a negative result needs careful wording. “Not detected” can mean that the target was absent, that its concentration was below the method’s limit of detection, that the relevant size range was inaccessible, or that preparation damaged or obscured it. Without knowing the method and its controls, absence of a signal is not automatically evidence that no MNPs were present.
The same sample can yield different kinds of “how much”
Measurement quality begins before the instrument is switched on. Plastic can be introduced or altered during collection and processing, so the human review identifies contamination prevention, standardized protocols, whole-sample blank controls, reference materials, recovery experiments, detection and quantification limits, and interlaboratory validation as part of its analytical framework. These are not administrative extras. They determine whether a small signal is likely to reflect the sample or the process used to handle it.
A blank is a control sample processed without the biological material of interest. If a similar polymer appears in the blank, researchers have evidence that the laboratory process itself may contribute to the result. A recovery experiment asks how much material can be found after a known amount is added to a sample. It tests whether digestion and processing lose particular polymers or particle sizes. Reference materials and interlaboratory comparisons ask whether different laboratories can obtain reasonably consistent results from a known or shared material. Together, these checks make a detection more credible and make disagreement easier to diagnose.
Preparation can change the material being measured. Biological samples often need digestion to remove proteins, cells, fats, or other matrix components. The review warns that acid-based digestion can degrade some particles, including polyamide, and discolour others. That creates a choice: a harsher treatment may remove more biological interference but also alter the target. A less aggressive treatment may preserve particles but leave more material that interferes with identification. The correct balance can differ by tissue and polymer.
The matrix—the biological material surrounding the particles—creates another problem. In Py-GC/MS, lipids and other biomolecules can produce chemical fragments that resemble products from some plastics. In spectroscopy, pigments and tissue residues can obscure a spectrum. A method can be highly capable in a clean reference sample and less decisive in blood, stool, or tissue. This is why a detection limit is not a universal property of an instrument. It depends on the sample, preparation, polymer, signal, and criteria used to call a match.
There is also a difference between a detection limit and a quantification limit. The first concerns whether a signal can be distinguished from background at all. The second concerns whether the amount can be estimated with acceptable reliability. A study may be able to say that a polymer signal is present while being unable to provide a precise burden. That distinction should appear in the claim. “Detected” and “measured accurately” are not synonyms.
Spectral libraries add a further selection effect. Researchers compare an observed signal with reference patterns for known polymers. The review notes that libraries are more comprehensive for common materials such as polyethylene, polypropylene, polyvinyl chloride, polyethylene terephthalate, and polystyrene. Those polymers are genuinely widespread, but they may also be easier to identify than less-studied materials. A list dominated by familiar polymers can therefore reflect both environmental prevalence and the unequal visibility of different polymers to the method.
Localization brings its own interpretive value. Knowing that particles are in a tissue is different from knowing that they are in a particular compartment, lesion, or cell-adjacent region. Microscopy paired with spectroscopy can help describe where a particle sits and what it looks like. That can support a hypothesis about exposure or biological interaction. It still does not show what the particle did there. A particle can be trapped, transported, or deposited without having caused the surrounding pathology.
The same caution applies when comparing studies. Suppose one study uses Raman to count particles above its effective size range, while another uses Py-GC/MS to estimate total polymer mass after a different digestion protocol. A larger reported value in the second study may reflect smaller particles that the first method could not see, a different way of expressing quantity, or a difference in sample preparation. It may reflect a real exposure difference, too. The number alone cannot tell us which explanation is correct.
This is why method descriptions belong next to health claims. If a paper reports that higher MNP burden is associated with an outcome, readers need to know whether burden means particle count, estimated mass, direct polymer mass, or another proxy. They also need to know whether the study could separate microplastics from nanoplastics and whether the comparison used the same analytical approach across participants. Without that information, the association may be directionally interesting but difficult to compare or reproduce.
The measurement problem does not make human MNP findings meaningless. It sets their boundary. The studies can establish that particular polymer signals were found under particular analytical conditions. They can sometimes characterize burden, size, shape, or location. They can reveal patterns that deserve health investigation. But a measured number is not a complete internal dose, and a non-detection is not automatically proof of absence.
That boundary is essential for the next question. Once researchers have turned different human samples into method-dependent MNP results, what health patterns appear alongside them? The answer requires looking across study designs and body systems while keeping the measurement conditions attached to every association.