Table 1.
Genetic variant | Gene | Effect allele/non-effect allele | Effect allele frequency | Effect (95% CI) | p value |
---|---|---|---|---|---|
SGLT2 (primary) | |||||
rs1232538 | SLC5A2 | G/T | 0.73 | −0.014 (−0.009 to −0.019) | 4.0 × 10−8 |
rs28675289 | SLC5A2 | T/C | 0.04 | −0.038 (−0.027 to −0.049) | 1.5 × 10−11 |
rs28692853 | SLC5A2 | A/C | 0.50 | −0.015 (−0.010 to −0.019) | 2.8 × 10−10 |
rs45625038 | SLC5A2 | C/T | 0.97 | −0.041 (−0.028 to −0.055) | 1.2 × 10−9 |
rs55766044 | SLC5A2 | C/T | 0.72 | −0.018 (−0.013 to −0.023) | 3.9 × 10−12 |
rs557720784 | SLC5A2 | C/T | 0.95 | −0.026 (−0.016 to −0.037) | 6.1 × 10−7 |
rs8050500 | SLC5A2 | C/T | 0.45 | −0.027 (−0.022 to −0.031) | 1.2 × 10−30 |
SGLT2 (stringent) | |||||
rs9930811 | SLC5A2 | G/A | 0.37 | −0.016 (−0.021 to −0.012) | 8.7 × 10−12 |
rs35445454 | SLC5A2 | T/C | 0.34 | −0.013 (−0.018 to −0.008) | 1.2 × 10−8 |
Notation: two sets of instruments proxying SGLT2 inhibition using different instrument selection processes are listed here. For the main analysis, primary instruments selected genetic variants that were robustly associated with HbA1c (p < 1 × 10−6) in the SLC5A2 region. Stringent instruments selected genetic variants that were associated with both expression of SLC5A2 gene and HbA1c levels and showed colocalization evidence between the two (colocalization probability > 0.7) in the SLC5A2 region, which were used in the main analysis. Two pairs of primary and stringent instruments were in moderate LD (r2 between rs9930811 and rs8050500 = 0.56, r2 between rs35445454 and rs1232538 = 0.23), which suggested that the two different selection processes picked two shared genetic signals as instruments in this region.