Strong molecular regulation of PARK7 with limited causal evidence for lung squamous cell carcinoma susceptibility: an integrative reanalysis

Background

PARK7 is a redox-responsive cancer-related gene. Whether this regulation is accompanied by multiplicity-corrected evidence for lung squamous cell carcinoma (LUSC) susceptibility or B-lineage tissue localisation remains uncertain. We reassessed PARK7 using genetic, molecular and tissue resources.

Methods

We applied Benjamini–Hochberg correction to a fixed universe of 16,597 OneK1K gene×cell-type hypotheses in six documented screens based on five public LUSC summary-statistic products whilst accounting for cohort overlap. Linkage-disequilibrium-aware principal-component inverse-variance weighting/generalised method of moments and external ImmuNexUT correlated-instrument generalised least squares were used to estimate ratio-scale associations rather than calibrated causal effects per expression unit. Colocalisation assessed regional sharing; exploratory SuSiE characterised exposure-side signals. The complete PARK7 cis-pQTL region from the China Kadoorie Biobank was compared with four LUSC outcomes. Visium, single-cell, GeoMx, TCGA and archived single-marker immunohistochemistry provided tissue context.

Results

PARK7 did not pass BH-FDR in any of the six fixed gene×cell-type screens; its best exploratory GCST004750 gene-level aggregation BH q was 0.505. LD-aware estimates were non-significant before and after perfect-LD-group collapse (89 variants: OR 1.183, P=0.085; 47 representatives: OR 1.181, P=0.082). ImmuNexUT identified strong rs226249 cis-eQTLs in five B-cell subsets from one mixed-disease Japanese cohort (P=3.73×10⁻ ⁵ ⁷ to 2.24×10⁻ ⁴² ). Only the Naive B subset provided two candidates from distinct deposited conditional ranks. Correlated-instrument estimates were compatible with the null across four LUSC products (P=0.743–0.996). None of 20 eQTL–LUSC colocalisations met the analysis-defined confirmation criterion (maximum default PP4=0.0579; maximum across the prior grid=0.235). The China Kadoorie Biobank identified a strong Olink cis-pQTL at rs551550334 (P=5.1×10⁻ ²² ; effect-allele frequency=0.9933; MAF≈0.0067), but the sentinel and high-linkage-disequilibrium proxies were absent from the predominantly European LUSC intersections. Transcript analyses did not consistently support PARK7 enrichment in B cells over epithelial/tumour or myeloid contexts. The 15-specimen immunohistochemistry series showed PARK7 immunoreactivity (mean H-score 198.67, SD 32.21) but did not permit B-cell lineage assignment.

Conclusions

External QTL datasets identified strong PARK7 cis-regulation for peripheral B-cell expression and circulating protein levels in an East Asian population. The available analyses did not establish an effect on LUSC susceptibility or B-cell-specific protein localisation. Adequately powered independent LUSC genetic evidence and direct multiplex and perturbation experiments are required before PARK7 can be considered a causal LUSC susceptibility gene or therapeutic target, or before a B-cell-specific PARK7 mechanism can be proposed.