Response to Comment on Ye et al. The Association Between Circulating Lipoprotein(a) and Type 2 Diabetes: Is It Causal? Diabetes 2014;63:332–342

Onat et al. (1) questioned our conclusion of a noncausal inverse association between lipoprotein(a) [Lp(a)] levels and risk of type 2 diabetes (T2D) (2). We agree that autoimmune activation may play a role in the development of T2D, but how this may contribute to the pathogenesis of T2D is still unclear and needs further research in human and animal studies (3). In our work, we focused on examining the causality of association between Lp(a) levels and risk of T2D using a Mendelian randomization (MR) approach.

Our findings are robust, being based on a prospective analysis in a large sample (18,940 participants) with a comprehensive MR analysis that found a discordance between the observed and genetic effects, suggesting that the Lp(a)–diabetes association is not causal. In contrast to the work Onat et al. cite of differential associations by sex (4), in the European Prospective Investigation of Cancer (EPIC)-Norfolk cohort we found very similar associations between rs10455872 genotype and Lp(a) levels by sex, with no evidence of an interaction. Our MR findings are consistent with another recent large MR study that also assessed the causal association between Lp(a) levels and T2D using the same SNP (rs10455872) in the Danish population (n = 28,567 of 77,901 with complete data) (5). Our study was not set up to address issues of Lp(a) isoform size, but the Danish study raised the possibility that large isoform size rather than low Lp(a) levels might be the main driver of risk for T2D, suggesting that a causal association for large Lp(a) size cannot be excluded but finding no evidence for a causal association with Lp(a)levels (5).

We acknowledge that more generally the MR approach may pose challenges and limitations in addressing causal inference (6). However, we appropriately applied MR in our work, and our findings are in keeping with other prevailing evidence thus far suggesting that a causal association with Lp(a) levels is unlikely. Future research will help to advance this field further.