Abstract
<title>Abstract</title> <p>Type 2 diabetes is a heterogeneous disease in which impaired insulin secretion, insulin resistance, altered lipid handling, and tissue-specific stress responses contribute in different proportions across individuals and populations. This heterogeneity complicates the development of broadly effective interventions and may contribute to variable responses to nicotinamide adenine dinucleotide precursors such as nicotinamide mononucleotide and nicotinamide riboside. The present study examined whether pathways linked to NMN-associated transcriptional rescue in aging tissues also carry genetic associations with type 2 diabetes across diverse ancestry groups. We analyzed S-PrediXcan TWAS results from five type 2 diabetes analysis groups—African American, East Asian, European, Hispanic, and South Asian—using 10 NMN-informed KEGG gene sets. The pipeline combined tissue-level Z-scores using a Stouffer framework, evaluated competitive enrichment by permutation, estimated bootstrap confidence intervals, and performed gene-level leave-one-out influence analysis. Recurrent genes were ranked according to their ancestry breadth, pathway breadth, and influence on the aggregate Stouffer statistic. All 10 gene sets reached the prespecified absolute Stouffer Z threshold in at least one analysis group. The strongest individual signal was a negative fatty-acid-degradation association in the European group, with Stouffer Z = −9.70 and permutation p = 0.0077. Autophagy in the African American group was the only other gene-set-by-ancestry result that met the joint absolute-Z and permutation criteria. Stage 3 analysis identified 548 unique leave-one-out-influential genes, including 32 genes observed across at least four ancestry groups. MAP2K1 and MAPK3 were the most recurrent multi-pathway drivers, whereas CPT2 was especially notable because it was directionally concordant across four groups and had also been identified in the NMN-responsive gene analysis. The collapsed top-30 driver module was protective-leaning in the European and Hispanic groups but risk-leaning in the East Asian group. These findings provide a hypothesis-generating bridge between NMN-responsive metabolic programs and human type 2 diabetes genetics. They support ancestry-aware, pathway-stratified evaluation of NAD+ precursors and prioritize CPT2, MAP2K1, MAPK3, and selected trafficking genes for functional validation and biomarker development.</p>