Hypothesis

County-level PFAS concentrations in public water systems (EPA UCMR3/UCMR5 data) predict thyroid cancer incidence (NCI SEER registries) with a 5-7 year latency period, and this association is modified by county-level autoimmune thyroid disease prevalence — a critical confounder no existing study has controlled for.

What Exists

A 2023 ecological study (PMC10537801) was the first to match UCMR3 drinking water PFAS data to county-level cancer registries, finding significant positive correlations for PFOA (r=0.031, p=0.043) and PFNA (r=0.058, p≤0.001) with thyroid cancer across 1,634 U.S. counties. The C8 Science Panel concluded a "probable link" between PFOA in drinking water and thyroid disease. EPA screening found PFOS and PFHxS inhibit the human sodium/iodide symporter (hNIS), the gatekeeper for thyroid iodine uptake (EPA assessment).

The Gap

The existing ecological study acknowledged critical limitations: no confounder adjustment, short latency between monitoring and cancer period, and no individual exposure data. No study has:

1. Built a dose-response model incorporating latency periods (thyroid cancer takes 5-15 years to develop)
2. Tested whether county-level TPO-antibody positivity (autoimmune thyroid disease prevalence) modifies the PFAS-cancer relationship — despite strong biological rationale from individual-level studies showing PFAS effects are amplified in TPO-Ab+ individuals (PMC7851056)
3. Used the newly available UCMR5 data (2023-2025) which tests for 29 PFAS compounds vs. UCMR3's 6

Testable Predictions

1. Latency-adjusted models win: Poisson regression of county-level thyroid cancer incidence on UCMR3 PFAS concentrations should show stronger associations (lower AIC) when PFAS exposure is lagged 5-7 years vs. concurrent exposure, controlling for county demographics (age, sex, race, income, healthcare access via ACS)
2. Effect modification by autoimmunity: Counties in the top quartile of thyroid autoimmune disease prevalence (estimable from Medicare claims for ICD-10 E06.3) should show 2-3x stronger PFAS-thyroid cancer associations than bottom-quartile counties
3. Chemical specificity: Among UCMR5-tested PFAS, compounds with demonstrated hNIS inhibition (PFOS, PFHxS) should show stronger thyroid cancer associations than PFAS without hNIS activity, providing a mechanistic fingerprint
4. Dose-response threshold: The association should exhibit a nonlinear threshold consistent with EPA's proposed PFAS MCLs (4 ppt for PFOA/PFOS)

Falsification

If latency-adjusted Poisson models show no significant association (IRR 95% CI crossing 1.0) between UCMR3 PFAS concentrations and county-level thyroid cancer incidence across all available SEER counties (n > 500) after adjusting for demographic confounders, the hypothesis is falsified.

Data Sources (All Free)

• EPA UCMR3/UCMR5: Public water system PFAS monitoring (epa.gov/dwucmr)
• NCI SEER: Cancer incidence by county (seer.cancer.gov)
• Census ACS: County demographics for confounder adjustment
• CMS Medicare claims: ICD-10 E06.3 prevalence for autoimmunity effect modification

Significance

This converts correlational PFAS-cancer epidemiology into a testable causal framework by introducing latency modeling and a biologically motivated effect modifier (autoimmune susceptibility). If the autoimmune interaction term is significant, it implies PFAS screening should prioritize communities with high thyroid autoimmune burden — a targetable public health intervention.