The paradox of SGLT2 inhibitors in heart failure: Caution with drug target Mendelian randomization

Mendelian randomization (MR) is a well-established technique in epidemiological research. Despite its utility, several challenges in drug target MR applications remain underexplored. This study investigates the association between sodium-glucose cotransporter-2 inhibitors (SGLT2i) and heart failure (HF), employing various selections of instrumental variables (IVs) for SLC5A2 and emphasizing the necessity of positive controls. Initial summary data-based MR analysis showed that increased expression of SLC5A2 (equivalent to a one standard deviation [SD] increase) was significantly associated with Type 2 diabetes (T2D) (OR: 0.76, p = 0.038). The association with HF was not statistically significant (odds ratio [OR]: 1.30, p = 0.107). Subsequently, inverse variance weighted (IVW)-MR analysis demonstrated the association between SLC5A2-mediated HbA1c (one SD increase) and T2D (OR: 1.40, p = 0.196) and HF (OR: 0.61, p = 0.026). These IVs potentially indicated the invalid of positive control. We further constructed IVs based on the mRNA expression of SLC5A2 (p < 0.001, primarily in human blood) and assessed the association of each variant with HbA1c. IVW-MR analysis revealed significant associations between SLC5A2-mediated HbA1c (one SD increase) and T2D (OR: 2.37, p = 2.30E-7) and HF (OR: 0.45, p < 0.001). Finally, we constructed IVs from the most significant SNPs within all tissues, yielding significant associations for both T2D (OR: 3.05, p = 2.30E-7) and HF (OR: 0.44, p = 0.003). Therefore, we advocate for researchers conducting MR analysis of drug targets to report their instrumental variables, comprehend the pharmacological mechanisms involved, present positive control results and cautiously interpret their conclusions.