LPA genetic variation is associated with type 2 diabetes after adjustment for lipoprotein(a) concentration: results of the Aragon Workers’ health study

Abstract Background Epidemiological and Mendelian randomization studies have reported an inverse relationship between lipoprotein(a) [Lp(a)] concentration and type 2 diabetes (T2D), although findings have not been uniform across studies. We hypothesized that genetic variation at the LPA locus may contribute to T2D development beyond circulating Lp(a) concentration alone, and therefore aimed to investigate the association between common LPA variation and incident T2D. Methods We analysed 5058 male participants from the Aragon Workers’ Health Study with approximately 15 years of follow-up. T2D was defined as the presence of antidiabetic treatment or HbA1c ≥ 6.5%. Common LPA single-nucleotide variants (SNVs) with a minor allele frequency ≥ 1% were selected. Several weighted LPA genetic scores were constructed, with the 15-SNV score defined as the primary LPA genetic score and alternative and more restricted scores evaluated in sensitivity analyses. Associations of single LPA variants and LPA genetic scores with incident T2D were assessed using multivariable logistic regression models adjusted for age, body mass index, ln-transformed Lp(a), and a T2D polygenic risk score. False discovery rate correction was applied to account for multiple testing. Robustness was assessed using leave-one-out analyses and Cox proportional hazards models. Results Among 4744 participants free of T2D at baseline, 578 (12.2%) developed incident T2D. Baseline Lp(a) concentrations were lower in participants who developed incident T2D than in those who remained free of T2D (12.6 [5.0–32.5] versus 17.0 [6.1–40.0] mg/dL; p < 0.001). No individual LPA SNV remained significantly associated with incident T2D. In contrast, the primary 15-SNV LPA genetic score was associated with incident T2D after adjustment for measured baseline Lp(a) concentration (odds ratio per one-standard-deviation increase 1.17, 95% confidence interval 1.07–1.29; p = 0.001). The 13-SNV, 12-SNV, 10-coding/splice, and 2-SNV raising-Lp(a) scores were also significantly associated with incident T2D after false discovery rate correction. The association with the 15-SNV LPA genetic score was supported by leave-one-out and Cox analyses. Conclusions A 15-SNV LPA genetic score was associated with incident T2D after adjustment for measured baseline Lp(a) concentration and a T2D polygenic risk score. These findings suggest that common LPA variation represented by the 15-SNV genetic score may contribute to T2D risk through mechanisms not fully captured by circulating Lp(a) concentration alone.