Showing that a gene that raises cholesterol is a cause of heart disease is evidence that that raised cholesterol is a suspect, but few if any causes have a single effect, changing a gene that changes ldl levels is no exception. Does familial hypercholesterolemia result in a higher particle count? Change particle size? If decreased clearance of ldl cholesterol by the liver or other organs is involved--is there some related ldl-receptor signallling that's corrupted? This may have the advantages of better randomization etc., but the variable that's being isolated isn't homocysteine or ldl cholesterol, it's the particular genetic variations. Maybe a higher probability of a generated hypothesis bearing fruit than with epidemiology, but I don't think this replaces clinical trials.