PFAS: transport-protein binding, hormone metabolism, and developmental vulnerability

PFOS, PFOA, PFHxS, and shorter-chain replacement PFAS continue to be detected at low μg/kg or ng/L levels in targeted surveys of seafood, drinking water, eggs, and composite foods, mirroring the PFAS profile commonly observed in serum biomonitoring [3, 22, 26]. In non-occupational populations, diet and drinking water are the dominant routes of PFAS exposure [3, 27], making PFAS a suitable model for pathway-based assessment of food-chain mixtures.

PFAS are among the best-characterized contaminant groups for endocrine relevance because both mechanistic and epidemiologic data support thyroid-axis perturbation. Mechanistically, PFAS can bind thyroid-hormone transport proteins (e.g., transthyretin) and compete with T4 [28, 29], interfere with thyroid hormone synthesis via thyroid peroxidase (TPO) inhibition [29–31], and increase hepatic clearance through enzyme induction and enhanced glucuronidation/biliary excretion [29, 30]. Consistent with these mechanisms, Du et al. synthesized evidence linking PFAS mixtures with shifts in thyroid hormones across pregnant women, adolescents, and adults [32].

PFAS exposure in the general population occurs as mixtures. Populations are hardly ever single-contaminant cohorts of either PFOS or PFOA; most are mixed groups of PFOS, PFOA, PFHxS, PFNA, and occasionally of shorter-chain replacements [23]. Summarized cohorts reported low single-digit ng/mL median serum concentrations—similar to national biomonitoring levels—supporting relevance to general-population exposure [32, 33]. This is relevant to developmental windows: Adequate maternal thyroid hormone supply during pregnancy is critical for fetal brain development [25, 34], and maternal hypothyroxinemia has been associated with lower offspring cognitive scores and increased risk of ADHD in some cohorts [35]. Accordingly, even modest PFAS-associated shifts toward higher TSH and lower free T4 during early pregnancy may be clinically meaningful at the population level.

Similarly, recent mixture studies using BKMR or WQS have reported that adults—particularly women—with higher combined PFAS exposure are more likely to exhibit metabolic syndrome, central adiposity, and dyslipidemia [33], although some analyses (including all-sex models) report null or inverse associations. Such heterogeneity is common in endocrine outcomes and may reflect effect modification by sex, adiposity, and life stage. These patterns are more consistent with chronic, low-dose pathway perturbation than with metabolic inertness of PFAS.

Bisphenol analogues and phthalates: receptor activity and steroidogenic inhibition

The use of food-contact materials continues to be a viable cause of bisphenol and phthalate exposure despite restrictions imposed on BPA in certain products [20, 36]. BPA, BPS, BPF, and phthalate plasticizers in food simulants are still detected in migration tests of can linings, polycarbonate replacements [8, 20, 37], and some flexible packaging; these findings align with human biomonitoring trends in which BPA has declined while BPS/BPF detection has increased. This convergence of occurrence in food-contact settings and internal exposure supports joint consideration of these compounds as a class. Although BPS and BPF have substituted bisphenol A due to endocrine issues, the receptor-binding motif was not eliminated by structural replacement. A 2015 systematic review in the Environmental Health Perspectives by Rochester and Bolden [38] revealed that BPS and BPF bind to estrogen receptors ERα and ERβ, stimulate estrogen-responsive reporter systems, and in some instances are weak androgen receptor antagonists, with potencies comparable to BPA in some assays [39]. Subsequent studies also have found interactions of certain bisphenol analogs with thyroid hormone receptors, suggesting that nuclear-receptor interference in this family is not limited to the estrogen–androgen axis [37, 40]. The fact that the structure does not correspond to safety is straightforward: Phenolic rings and hydroxyl groups that cause BPA to be a good estradiol mimic are still present in BPS and BPF [41]; alteration of the bridging moiety may change kinetics and lipophilicity, without necessarily erasing the pharmacophore.

The action of phthalates is a complementary action, equally well-described. On exposure, high-molecular-weight phthalates like DEHP, DBP, BBzP, or DiNP are quickly hydrolyzed to monoesters, and these monoesters inhibit important steroidogenic enzymes, including 3β-hydroxysteroid dehydrogenase and 17β-hydroxysteroid dehydrogenase [42, 43]. They also repress the expression of steroidogenic acute regulatory (StAR) protein, which regulates the transport of cholesterol into the mitochondria, a rate-limiting step in the metabolism of steroid hormones [32, 44]. The human fetal testis is particularly susceptible during the androgen window (or gestational weeks 8–14), during which androgens define the distance between the sexes and masculinization of the reproductive tract [45]; disrupting androgen production at this stage has long-term effects. Several phthalates do the same steroidogenic pathway with varying strengths; that is just the type of thing where component-based, pathway-centered grouping is more realistic than treating each phthalate as an independent entity. In 2024, an adult literature review demonstrated that increased urinary phthalate metabolites are also linked to reduced serum testosterone in men, by about 5–15% across four levels of exposure, individually modest, but consistent with environmental disruption of endocrine systems [46].

Micro- and nano plastics: size, polymer, and dual ways

MNPs are not as homogeneous as PFAS or bisphenols, which is why they should be described more anatomically. Food-production and food-packaging processes are now a recognized source, as the abrasion of food-processing equipment, cutting boards, and plastic wrapping may liberate micro-scale food particles into directly-consumable ready-to-eat food [47], and multiple recent surveys have reported polymer fragments in table salt, bottled water, and seafood that are meant to be consumed directly [48, 49]. Microplastics (1–5 μm) are retained in the gut lumen or taken up by the Peyer patches, where they can cause local inflammation and change barrier integrity primarily due to mechanical interactions, reactive oxygen species (ROS) generation, and cytokine release [50, 51]. Nanoplastics (less than 1 μm) are able to traverse epithelial barriers, be distributed in the circulation, and secondary organs [52]; animal experiments have shown distribution to the liver, spleen, and even the placenta [53]. Polymer type also matters. The surface chemistries, sorption properties, and (probably) inflammatory potential of PE, PVC, polystyrene (PS), and polypropylene (PP) differ, and the PS particles used in experimental studies are not necessarily representative of all food-relevant polymers.

The post-2024 NEJM article contamination debate is noteworthy in that it unveiled the point of the procedure most susceptible to compromise [9]. The polymers can be detected by pyrolysis-GC/MS, although laboratory air includes airborne plastic particles, surgical drapes, tubing, and gloves may release microplastics, and handling of samples may introduce particles unless processed in very clean conditions [54, 55]. The May 2024 letters pointed to procedural blanks that were required, polymer-specific controls, and clean-room conditions. However, the authors responded by using further controls, and most importantly, contamination should be random, not necessarily concentrated in the plaques of patients who subsequently developed poor clinical outcomes [9]. The fact that the clinical association went on despite these concerns makes the signal worthy of integration, and further refinement of methodology is still needed. Methodological uncertainty remains a key limitation in MNP detection because contamination control, procedural blanks, polymer-specific controls, and reporting of detection limits can materially influence measured tissue burdens and cross-study comparability [56]. Standardized clean-lab workflows and inter-laboratory comparisons are therefore essential for translating occurrence signals into interpretable exposure metrics.

In terms of mechanism, the two ways that MNPs can place stress on pathways are both applicable to PDL. Particle-induced pathway includes downregulation of tight junction proteins, including ZO-1 and occludin, ROS production and inflammasome activation, including NLRP3, resulting in a leakier gut and low-grade systemic inflammation [57]. The chemical-carrier pathway takes advantage of the hydrophobic surface of plastic particles [58], which may absorb PFAS, bisphenols, and polycyclic aromatic hydrocarbons in the environment or in the food-production chain. These adsorbed chemicals may be released on an acidic or surfactant-rich portion of the gastrointestinal tract, effectively delivering a focused burst of EDCs to the epithelium. Correspondingly, assuming particles can reach plaques as the NEJM study indicates, they can presumably reach or at least penetrate other tissues, and animal studies already demonstrate liver and placenta accumulation [9]. Both pathways contribute to the barrier-inflammation axis already strained by PFAS and endocrine-active plastics. These two mechanistic routes by which micro and nanoplastics damage the gut barrier and amplify endocrine-disrupting co-exposures are illustrated in Figure 2.

Display full size

Figure 2. Micro- and nano plastics: dual pathway mechanisms. MNPs contribute to pathway disruption through two complementary mechanisms.

Mechanism 1 (direct particle toxicity): MNPs cause physical barrier disruption through tight junction downregulation (ZO-1, occludin), ROS generation, and NLRP3 inflammasome activation, leading to increased intestinal permeability, dysbiosis, and metabolic endotoxemia.

Mechanism 2 (chemical carrier/amplification): Hydrophobic MNP surfaces adsorb endocrine-disrupting chemicals (BPA, phthalates, PFAS) and release them in the gastrointestinal tract, amplifying EDC exposure and contributing to thyroid, reproductive, and metabolic disruption. Both mechanisms contribute to the gut barrier-inflammation component of PDL. Clinical evidence includes Marfella et al. (2024) showing MNPs in atherosclerotic plaques [9] and Jin et al. (2019) demonstrating gut dysbiosis and metabolic disorders in mice [59].

The significance of chronic disease with respect to these pathways

The importance of this convergence is that it involves three pathways: Thyroid-metabolic, nuclear receptor/steroidogenesis, and gut barrier-inflammation, which are all connected to the prevalence of common chronic diseases.

The thyroid-metabolic axis controls basal metabolic rate, lipid metabolism, and, during pregnancy, neurodevelopment of the fetus [60]. Maternal hypothyroxinemia (TSH above pregnancy-optimised ranges, free T4 at the lower end) has been linked in multiple cohorts with worse offspring cognitive scores and increased risk of ADHD [34]. Subclinical hypothyroidism in the case of adults (TSH 4-10 mIU/L and normal T4) is associated with dyslipidaemia and cardiovascular disease development [61]. Even environmentally chemical-induced, population-wide changes in TSH and free T4 of even minimal magnitude will thus shift the prevalence of metabolic syndrome in the wrong direction.

The steroidogenic and nuclear receptor impairment is directly applicable to male reproductive health [62], and it has been demonstrated that a decline in testosterone is linked to increased central adiposity, insulin resistance, and loss of bone density with age [63]. Anti-androgenic exposure during the womb, which reduces the length of the anogenital distance, is predictive of cryptorchidism, hypospadias, and perhaps reduced sperm counts in adulthood [64]. Bisphenol estrogenicity is of concern because of earlier puberty in girls and cancer sensitivity to hormones [36]; cause and effect are unconfirmed, but the mechanism is reasonable.

The disruption of gut barriers and low-grade inflammation is currently becoming a mainstream cause of non-alcoholic fatty liver disease, insulin resistance, and type 2 diabetes [65]. In these states, zonulin and LBP are increased; translocation of bacterial lipopolysaccharide through a permeable barrier results in the generation of metabolic endotoxaemia and a sustained increase in IL-6/CRP that increases cardiovascular risk [66]. When MNPs, together with emulsifiers, together with endocrine-active migrants of packaging weaken the barrier, they will increase the very process of these inflammation-metabolism interactions.

Tier 1: pathway-specific biomarkers

In the case of the thyroid axis, free T4 is more informative than that of total T4 since only the unbound hormone is bioavailable and since chemicals that bind can alter total, but not the free. A TSH of between 0.4–4.0 mIU/L is termed normal, but there is persistence above, say, 2.5 mIU/L, which may signify subclinical pressure, particularly during pregnancy. Free T3 is an indicator of the peripheral deiodinase activity, which is subject to certain chemicals [67]. In steroidogenesis, precursor/product ratios are useful: A high ratio of androgens produced to testosterone may indicate 17β-HSD inhibition [68], and LC–MS/MS panels can measure multiple steroids simultaneously. To the gut barrier, zonulin is an indication of tight-junction control, LBP is the intestinal injury of bacterial products, and intestinal FABP2 is the injury of enterocytes [69]. These biomarkers are already present and measurable, with literature associating them with metabolic and liver disease.

Tier 2: functional receptor/enzyme tests

Mammalian cell lines (e.g., HEK293) transfected with ER, AR, or TR and a luciferase reporter can be subjected to food extracts, water concentrates, or serum [70]. In the case of receptor agonists or antagonists in the sample, transcription is altered, and the light production is measured in 96-well plates. To further make these assays more realistic, they can include phthalates and other chemicals that require metabolic activation, and they can be spiked with a liver S9 fraction to simulate bioactivation, as would occur in vivo. This represents true endocrine activity as compared to the testing of parent compounds.

Tier 1 and Tier 2, together, inform us that the pathway of the person is in motion, and that the exposure matrix that the person is exposed to has pathway-active agents. That is the core of PDL. An operational schematic of how Tier 1 biomarkers and Tier 2 functional assays combine into a PDL score is shown in Figure 3.

Display full size

Figure 3. Pathway Disruption Load (PDL) two-tier measurement architecture. Operational framework for assessing cumulative pathway-level disruption.

Conceptual workflow and pseudo-equation (for reproducibility without over-quantification)

For each pathway p (thyroid; nuclear receptor/steroidogenesis; gut barrier–inflammation), PDL can be represented as: The following pseudo-equation is illustrative and not intended as a finalized scoring algorithm; the weights (w₁, w₂) are hypothetical placeholders that await empirical calibration from future mixture toxicology and cohort reanalysis studies.

PDL_p = w₁·Z(Tier-1 biomarker deviation)_p + w₂·A(Tier-2 functional activity)_p

where Z(·) denotes standardized deviation within a reference population distribution (e.g., percentile or z-score) and A(·) summarizes net receptor/enzyme activity measured in relevant matrices (e.g., TR antagonism; ER/AR activity; TPO inhibition). In early implementations, Tier-1 and Tier-2 components should be reported side-by-side in an ordinal or percentile-based format (e.g., low/moderate/high relative to the reference cohort distribution) before any pathway-specific weighting (w₁, w₂) has been empirically calibrated. These weights are explicitly hypothetical placeholders at this stage; numerical PDL scoring is not supported by current datasets and should not be inferred from the formulation above.

Operationally, the workflow is:

(i) Define the pathway panel (Tier 1 biomarkers + Tier 2 assays), (ii) standardize Tier 1 values to population distributions, (iii) quantify Tier 2 net activity in relevant matrices, and (iv) integrate results into a pathway-level interpretation of relative load (e.g., low/moderate/high within-cohort distribution).

Tier 1 comprises pathway-specific biomarkers measured in human samples: thyroid axis (TSH, free T4, and free T3), steroidogenesis (testosterone, estradiol, DHEA-S, precursor/product ratios via LC-MS/MS), and gut barrier integrity (zonulin, LBP, intestinal FABP2). Tier 2 comprises functional receptor/enzyme assays applied to exposure matrices (food extracts, water concentrates, serum): receptor assays (ER, AR, TR agonist/antagonist activity), enzyme inhibition (TPO, 17β-HSD, aromatase), and total mixture activity capturing unknown co-migrants. Integration of Tier 1 biological response and Tier 2 functional burden yields a pathway-level PDL score (PDL_p = w₁·Z(Tier-1 deviation)_p + w₂·A(Tier-2 activity)_p), classified as low, moderate, or high relative to reference population distributions.

Foodomics application: linking food-matrix functional signals to biomonitoring and omics-derived biomarkers

PDL is intentionally positioned as a foodomics-relevant bridge between (i) food-matrix exposure signals and (ii) biological response signals in humans. In practical terms, Tier 2 functional assays can be applied to extracts of packaged foods, can linings, food-contact materials, and drinking-water concentrates to quantify integrated pathway activity (e.g., ER/AR/TR modulation or TPO inhibition), including unknown co-migrants that may be missed by targeted chemistry. Tier 1 then anchors these exposure-proxy signals to pathway movement in humans using biomonitoring/omics-compatible biomarkers (e.g., TSH–free T4 for thyroid; LC–MS/MS steroid panels for steroidogenesis; zonulin, LBP, and related permeability/inflammation markers for barrier circuits).

Operationally, a foodomics-informed PDL workflow can be implemented as: (i) characterize the food/exposure matrix using Tier 2 functional readouts; (ii) measure Tier 1 pathway biomarkers in cohort samples (or re-analyze existing cohorts with banked biospecimens); and (iii) integrate both layers into pathway-specific load distributions within the population (e.g., low/moderate/high relative burden). This approach makes it possible to relate food-system exposures (dietary patterns, packaging choice, and water source) to measurable pathway-level biological perturbations without requiring complete chemical identification of all contributors.

Named cohorts and realistic studies as priorities of research

Multiple cohorts in existence are suitable for reanalysis in the PDL style. The Norwegian Mother, Father and Child Cohort (MoBa) contains more than 100,000 banked pregnancy samples; many of them already have PFAS and phthalate data. By including thyroid and gut-barrier markers, we would be able to determine whether cumulative exposure is reflected in pathway movement during pregnancy, the most sensitive window [71]. The European HBM4EU project has standardized biomonitoring between various nations; the retrospective measurement of zonulin, LBP, and sex steroids on such samples would provide a cross-country perspective of the pathway stress. In the U.S., NHANES cycles, which now already include a huge amount of chemical biomonitoring, might include even a partial panel of pathway biomarkers in a sub-sample to determine national PDL distributions.

The PFOA + BPS, PFOA + DEHP, BPS + DEHP, and a ternary mixture of PFOA + BPS + DEHP at low (25th percentile), medium (50th), and high (75th) human-equivalent doses can be used as the basis of a mixture toxicology. The mouse models were developed that were able to read out levels of thyroid hormones in dams and pups, anogenital distance, gut permeability signatures, and liver histology. This would require 3 to 5 years to fully set up these studies, but would have a direct impact on PDL weighting. Harmonization of the MNP method inter-laboratory comparison with clean-room processing and standardized blanks should be a 2–3 year goal.

Regulatory application: inserting functional layers of approval

The present-day approval procedure of new food-contact compounds typically uses a logic of migration-exposure-threshold-of-concern. What is lacking is an endocrine or thyroid pathway activity check on the migrated material. A process that is aware of PDL could screen Tier 2 (ER/AR/TR, TPO inhibition) on the migrated extract. When it happens that it is negative, the approval is made as it is. A second-tier 28-day study using the pathway biomarkers (TSH, free T4, testosterone, gut-barrier markers) might be a possibility in case it is positive. In case such a study establishes pathway movement, the applicant might be asked to show that the material would not impose any significant additional burden on the concerned pathway, as compared to typical co-exposures (PFAS, bisphenols and phthalates). This is similar to the reasoning of data requirements stacking up with tonnage in REACH; in this case, it is the activity pathway of functional activity that can become accumulable. This would inform market-basket surveillance when packaged foods consistently elicit positive endocrine responses in reporter assays, regulators may track down the source material and may consider requesting reformulation.

Clinical translation

The majority of what PDL needs is already in the standard laboratory menu; TSH and free T4 are common and cheap. LC–MS/MS sex steroid panels are becoming more accessible. PFAS, bisphenol, and phthalate panels are offered in specialized and some commercial labs, but this is cost- and insurance-limited. The only thing lacking is decision support that can take inputs such as upper-normal TSH, low-normal free T4, elevated PFAS, detectable BPS, and higher phthalate metabolites, and return a pathway-level interpretation (e.g., higher thyroid-axis disruption load), particularly for sensitive windows such as preconception and pregnancy. This could be incorporated into preconception counseling, antenatal visits, and endocrinology follow-up as a cumulative-exposure interpretation layer rather than as a diagnostic label. Although reversal at the individual level might be uncertain, at the population level, risk communication on feasible exposure-source modifications (e.g., drinking-water and food-contact pathways) may be inexpensive and consistent with the literature on developmental origins.

Practical interpretation can be structured as a simple sequence: (i) identify the pathway of concern (thyroid; steroidogenic/nuclear receptor; gut barrier–inflammation), (ii) review Tier-1 biomarkers relative to life-stage appropriate distributions, (iii) evaluate Tier-2 net activity in relevant food/water/packaging matrices when available, and (iv) integrate both layers as a pathway-level load signal to guide follow-up in research, surveillance, or preventive counseling contexts. In this way, “pathway load” functions as decision support for cumulative exposure patterns rather than as a diagnostic label.

Data gaps and future directions

MNPs require dose-response information in humans: What tissue burden causes measurable dysfunction of barrier function or systemic inflammation? Effects that are specific to polymers should be explained: Is PVC more inflammatory than PE, does PS leach more biologically active monomer, and do surface additives alter inflammatory potential? The clearance kinetics (lifelong particle accumulation or clearance or hepatobiliary and renal clearance) have not been adequately investigated.

Mixture effects must be mapped: Are PFAS-bisphenol mixtures additive, or do PFAS-bisphenol mixtures have more-than-additive TR or TPO effects? Do MNPs indirectly potentiate endocrine-disruptor action with a larger gut permeability and subsequent systemic absorption of the co-ingested EDCs? Do they encounter any antagonisms, e.g., induction of hepatic clearance of one chemical by another?

Vulnerability during life stages ought to be characterized in terms of PDL: The endocrine baselines during infancy, puberty, pregnancy, and aging are all different, and pathway loads cannot be simply compared across them. PFAS have been implicated in diminished vaccine response in addition to endocrine and barrier pathways, and some bisphenols have been found to have neuroactive effects; the question remains whether these are independent pathways of action or downstream of the described pathways.

Conclusion

Exposure of humans to new food-related contaminants is overlapping, chronic, and endocrine-mediated. PFAS binding to transport proteins triggering hormone clearance, bisphenol and phthalate substitutes retaining receptor-activity, MNPs disrupting gut barrier or adsorbing EDCs, all lead to three central pathways of metabolic, reproductive, and cardiovascular health. Experiments (e.g., Jin et al. on microplastics and gut barrier [59]), human meta-analyses (Du et al. [32] and subsequent PFAS-thyroid analyses), clinical observations (Marfella et al. [9] on MNPs in plaques), and consensus papers on the mechanisms all indicate the same direction.

Three take home messages: First, real-world food-chain exposure is pathway-convergent: PFAS, bisphenol/phthalate analogues, and MNPs independently and collectively stress the thyroid-metabolic axis, nuclear receptor and steroidogenic signaling, and gut barrier-inflammation circuits—the same pathways linked to prevalent chronic diseases.

Second, PDL offers a biologically coherent integration layer: By combining pathway-specific biomarkers (Tier 1) with functional bioassay outputs from relevant matrices (Tier 2), PDL translates chemical-agnostic mixture activity into interpretable pathway-level signals that neither single-substance assessment nor purely statistical mixture methods currently provide.

Third, foodomics and functional assays enable mixture-aware surveillance: Tier 2 assays applied to food extracts, packaging migrants, and drinking-water concentrates can capture total endocrine activity-including unknown co-migrants and link food-system exposures to measurable biological perturbations, supporting more effective and sustainable food safety decision-making.