Evaluating shared gene prioritization for type 2 diabetes and FinnGen R10 chronic periodontitis

To assess whether selected genome-wide association studies support shared gene-level prioritization between type 2 diabetes (T2D) and periodontitis.

European DIAMANTE T2D and FinnGen R10 chronic periodontitis summary statistics were analyzed using a custom whole-gene test accounting for linkage disequilibrium, with trait-specific Bonferroni correction. Published Stage III/IV Grade C periodontitis gene results were compared descriptively. Secondary analyses used multi-ancestry DIAMANTE statistics to construct minimum-P window labels, with a relaxed label selected after threshold inspection. Annotation-based elastic-net models were evaluated in repeated nested cross-validation before and after excluding 13 predictors dependent on locus-to-gene scores. Negative matching retained dependence on these scores.

Among 19,280 eligible T2D genes and 19,378 periodontitis genes, 683 T2D genes crossed the corrected threshold (353 using point probabilities and 330 using upper bounds); none crossed for periodontitis. No gene met both cutoffs among 19,275 commonly eligible genes. Three published periodontitis entries crossed the comparison cutoff (0.05/19,496), representing FCER1G and two SIGLEC5 identifiers; none met the primary T2D cutoff. The relaxed window label yielded 50 positive genes, evaluated with 250 matched negatives per panel. Across ten elastic-net configurations, mean pooled area under the receiver operating characteristic curve was 0.527 (range, 0.452-0.583) with full predictors and 0.482 (range, 0.437-0.528) after predictor exclusion; mean pooled average precision was 0.205 and 0.165, respectively. Genome-wide scoring was not performed.

The primary analysis did not establish shared gene-level associations. Secondary models did not justify genome-wide scoring of unlabeled genes. The findings do not exclude shared genetic susceptibility.