Exploratory profiles of phenols, parabens, and per- and poly-fluoroalkyl substances among NHANES study participants in association with previous cancer diagnoses

Prostate and breast cancer are the most commonly diagnosed cancers among men and women, respectively, yet established risk factors do not explain most cases. Multiple environmental and social factors likely contribute to initiation and progression, particularly for hormone-mediated cancers (breast, prostate, ovarian, endometrial/uterine, testicular, thyroid, and melanoma) whose growth depends on endogenous steroid and thyroid hormones. Identifying environmental insults that alter hormone pathways could inform prevention and mitigation strategies, including exposure-reduction interventions, regulation, and substitution with safer alternatives. Phenols, parabens, and PFAS are endocrine-disrupting chemicals (EDCs) with widespread human exposure via food packaging, personal care products, and consumer goods. They have been linked to perturbations in estrogen, thyroid hormones, and testosterone, and hormone effects are a key characteristic of carcinogenesis. While some case–control studies suggest associations of BPA and PFAS with breast cancer, relationships with other phenols and with other hormone-active cancers remain understudied. NHANES provides biomonitoring data for these chemicals and self-reported cancer diagnoses in adults, enabling exploratory cross-sectional evaluation of associations between current exposures to phenols, parabens, PFAS and prior endocrine-active cancer diagnoses. The study aims to assess these associations overall and to explore effect modification by race/ethnicity as a proxy for structural social factors.

Prior literature indicates: (1) EDCs, including phenols, parabens, and PFAS, can alter estrogen, thyroid, and androgen pathways and may be carcinogenic via hormonal mechanisms. (2) Epidemiologic evidence linking PFAS/BPA to breast cancer includes suggestive or positive associations in multiple case–control studies, but data on other phenols and other cancer types are limited. (3) For melanoma, prior large cohorts and occupational studies focusing mainly on PFOA/PFOS generally found null associations, often relying on modeled exposures rather than biomonitoring and with predominantly male cohorts and few cases. (4) For thyroid cancer, limited prior studies suggest possible links with PFAS contamination and with higher urinary parabens, though prior analyses often combined sexes and lacked biomonitoring detail. (5) For ovarian cancer, toxicological studies suggest estrogenic compounds like BPA can promote EMT, migration, and proliferation in ovarian cancer cell lines, but epidemiologic studies of phenols and ovarian cancer are lacking. (6) Racial/ethnic disparities exist in both exposures and cancer outcomes; NHANES has documented higher levels of certain phenols/parabens in non-White groups and higher levels of BPF, BP3, PFOA, PFOS in White participants, potentially shaping differential exposure–outcome associations. This study addresses these gaps by using NHANES biomonitoring across multiple chemical classes and by examining racial/ethnic effect modification.

Study design and population: Cross-sectional analysis using NHANES data from 2005–2018 among non-institutionalized, nationally representative U.S. adults. Inclusion: participants ≥20 years with complete covariates (age, serum cotinine, poverty-income ratio, race, education, BMI, and urinary creatinine for phenol/paraben analyses). Two separate analytical datasets were created due to non-overlapping subsamples for biomonitoring: PFAS (N=16,696; men 8,010; women 8,686) and phenols/parabens (N=10,428; men 5,084; women 5,344; not measured in 2017–2018). Analyses were sex-specific to assess sex-specific cancers. Exposure assessment: Serum PFAS (7 chemicals): PFHS, MPAH (2-(N-methyl-PFOSA) acetic acid), PFDE, PFNA, PFUA measured 2005–2018; PFOA and PFOS measured in all cycles except 2013–2014. Quantified via HPLC-turbo ion spray ionization-tandem MS. Urinary phenols/parabens (12 chemicals): measured 2005–2016 for BPA, BP3, triclosan (TCS), MPB, EPB, PPB, BPB; measured 2013–2016 for BPF, BPS, triclocarban (TCC), 2,4-dichlorophenol (DCP24), 2,5-dichlorophenol (DCP25). Quantified via online SPE coupled to HPLC-MS/MS. Biomarkers were right-skewed and natural log-transformed. Values <LOD imputed as LOD/√2. BPB, EPB, TCC had >50% <LOD and were modeled categorically: <LOD (reference), detectable < median, and ≥ median of detectable values. Outcomes: Self-reported physician-diagnosed cancers from the NHANES medical conditions questionnaire among adults ≥20 years. Cancer types: breast, ovarian, uterine (endometrial), prostate, testicular, thyroid, and melanoma. A combined reproductive cancer variable was created: breast/ovarian/uterine for women; prostate/testicular for men. Covariates: Age at survey, ln-serum cotinine, poverty-income ratio, race, education, BMI, and NHANES cycle indicator; phenol/paraben models additionally adjusted for ln-urinary creatinine. Sunscreen use considered for melanoma models but not included due to insufficient overlap. Statistical analysis: Logistic regression estimated odds of previous cancer diagnosis per interquartile range (IQR) increase in biomarker concentration (or categorical contrasts for chemicals with >50% <LOD). Bivariate checks informed confounder adjustment. Effect modification by race/ethnicity: separate models for non-Hispanic Black, Mexican American, and other Hispanic participants compared with White participants, using interaction terms with p-int<0.05 denoting significant differences. Sensitivity analysis: survey sampling weights applied following an approach prioritizing the smallest biomarker subsample to assess robustness and generalizability to the U.S. population. All analyses in R 4.0.4. Sample sizes and case counts: PFAS subset cancer cases among men: prostate 199, testicular 8, thyroid 7, melanoma 52; among women: breast 178, ovarian 35, uterine 51, thyroid 28, melanoma 39. Phenols subset cancer cases among men: prostate 104, testicular 7, thyroid 3, melanoma 20; among women: breast 114, ovarian 20, uterine 37, thyroid 9, melanoma 27.

- PFAS and previous cancer (women): Higher PFDE, PFNA, and PFUA were associated with increased odds of previous melanoma diagnosis in women: PFDE OR 2.07 (95% CI: 1.25, 3.43); PFNA OR 1.72 (1.09, 2.73); PFUA OR 1.76 (1.07, 2.89). PFNA was positively associated with previous uterine cancer: OR 1.55 (1.03, 2.34). PFUA showed a marginal positive association with previous ovarian cancer: OR 1.61 (1.00, 2.59). - Phenols/parabens and previous cancer (women): Ovarian cancer showed increased odds with higher DCP25 OR 2.80 (1.08, 7.27), BPA OR 1.93 (1.11, 3.35), and marginally with BP3 OR 1.76 (1.00, 3.09). Reproductive cancers (combined) in women were associated with DCP25 OR 1.61 (1.13, 2.29) and DCP24 OR 1.42 (1.06, 1.90). Uterine cancer showed an inverse association with EPB: OR 0.31 (0.12, 0.85). Melanoma in women was associated with BP3 OR 1.81 (1.10, 2.96), DCP25 OR 2.41 (1.22, 4.76), and DCP24 OR 1.85 (1.05, 3.26). - Men: No significant associations between PFAS biomarkers and previous cancers were observed. There was a marginal positive association between PPB and previous prostate cancer: OR 1.35 (1.00, 1.83). - Effect modification by race/ethnicity: Several associations differed by race. Examples include stronger associations in White women than Black women for ovarian cancer with PFOS (White OR 4.34 [1.24, 15.1] vs Black OR 0.75 [0.31, 1.80]; p-int=0.010) and PFDE (White OR 2.56 [1.27, 5.16] vs Black OR 0.87 [0.36, 2.09]; p-int=0.051), and for uterine cancer with PFDE (White OR 3.08 [1.53, 6.21] vs Black OR 0.42 [0.14, 1.27]; p-int=0.002), PFNA (White OR 2.36 [1.28, 4.37] vs Black OR 0.85 [0.36, 1.98]; p-int=0.043), PFUA (White OR 3.37 [1.89, 6.04] vs Black OR 0.41 [0.13, 1.25]; p-int=0.001). Other patterns included higher odds of previous breast cancer with MPAH in Mexican American women (OR 2.46 [1.19, 5.09]; vs White OR 1.03 [0.74, 1.45]; p-int=0.026) and with BP3 in other Hispanic women (OR 3.03 [1.22, 7.50]; vs White OR 0.94 [0.67, 1.31]; p-int=0.017). Among men, White men had higher odds of previous prostate cancer with increases in BP3 (White OR 1.42 [1.07, 1.89] vs Black OR 0.70 [0.41, 1.21]; p-int=0.022) and BPF (White OR 1.40 [0.78, 2.54] vs Black OR 0.33 [0.10, 1.12]; p-int=0.034). - Sensitivity analyses with survey weights generally supported the main findings and clarified that generalizable estimates may differ when associations are present primarily in oversampled minority groups. - Descriptive characteristics: Median age ~49 years; median PIR ~2; median BMI ~27 kg/m²; serum cotinine higher in men; racial distribution approximately 40% non-Hispanic White, 20% non-Hispanic Black, 15% Mexican American, 10% other Hispanic, 15% other.

The study identified cross-sectional associations between current biomarker concentrations of PFAS, phenols, and parabens and prior diagnoses of hormonally active cancers, particularly among women. Notably, PFDE, PFNA, and PFUA were associated with previous melanoma in women but not men, underscoring potential sex-specific mechanisms. The authors discuss biologic plausibility via estrogenic pathways, immune and oxidative stress differences, and the presence of estrogen receptors in melanoma, suggesting endocrine disruption could contribute to melanoma risk, consistent with sexually dimorphic melanoma epidemiology. For ovarian cancer, associations with BPA, BP3, and dichlorophenols align with toxicological evidence that estrogenic compounds can promote EMT, migration, and proliferation in ovarian cancer cells. The study also highlights racial/ethnic disparities in exposure–outcome associations, reflecting documented differences in biomarker levels across racial/ethnic groups and broader environmental injustice contexts (e.g., product use patterns and PFAS drinking water contamination). Some PFAS–cancer associations were stronger among White women, whereas certain phenol/paraben associations were stronger among non-White groups. These findings point to the necessity of evaluating structural determinants of exposure, potential differences in metabolism, and the importance of subgroup-specific analyses. Sensitivity analyses with survey weights emphasize considering analytic goals: generalizability to the U.S. population versus targeted inference within oversampled minority groups. Overall, the findings suggest estrogen-dependent mechanisms could be relevant for melanoma and ovarian cancer, and that cancer survivors may have elevated body burdens of EDCs, potentially interacting with hormone therapies. The exploratory results support prioritizing prospective studies with biomonitoring to elucidate temporality, mechanisms, and disparities, and may inform risk assessment and prevention strategies.

This exploratory NHANES analysis found multiple associations between current exposures to PFAS, phenols, and parabens and prior diagnoses of hormonally active cancers. Among women, PFDE, PFNA, and PFUA were associated with previous melanoma, and several phenols/parabens (BPA, BP3, DCP24, DCP25) were associated with previous ovarian cancer; EPB showed an inverse association with uterine cancer. Associations varied by race/ethnicity, underscoring exposure and outcome disparities. The study provides the first epidemiological evidence linking phenol exposures with previous cancer diagnoses and is the first NHANES analysis to examine racial/ethnic disparities in these associations. Future research should use prospective designs to assess temporality and mechanisms—particularly estrogen-related pathways in melanoma and ovarian cancer—and incorporate environmental justice perspectives to guide targeted interventions and policy. These findings can help prioritize chemicals for surveillance and risk assessment in communities facing environmental contamination risks.

- Cross-sectional design precludes causal inference and raises potential for reverse causation (exposures measured after cancer diagnoses; behavioral or treatment-related changes may affect biomarker levels). - Lack of information on time since cancer diagnosis limits interpretation relative to exposure window and disease course. - Potential exposure misclassification: single biomarker measurements may not reflect historical exposures; differences in metabolism and elimination not fully captured. - Possible outcome misclassification due to self-reported cancer diagnoses (though accuracy is high for some cancers); inability to validate against registries here. - Residual confounding from unmeasured factors (e.g., family history, surgical history such as oophorectomy/thyroidectomy, other lifestyle factors); correlations among covariates and biomarkers may bias estimates. - Multiple comparisons without correction increase the chance of false positives; results are exploratory and hypothesis-generating. - Limited case numbers for some cancers, especially in subgroup analyses, reduce precision and may yield unstable estimates. - Survey weighting considerations: Weighted analyses aim for population generalizability but can attenuate associations present primarily in oversampled groups; choice of weighting affects interpretability based on study goals.