Not that I'm holding my breath for CETP inhibitors but the cited 4% reduction in MI risk was calculated from the association between MI and SNPs and observed in genetic variant (modelled allele) near the CETP gene. CETP inhibitors might have a stronger effect on LDL (and on MI risk) than the predicted genetic variant plus they also reduce triglycerides and perhaps have an effect on ApoA-I, lpLPA2 and such.