Interindividual variability in low density lipoprotein (LDL) cholesterol (LDL-C) response during treatment with statins is well documented but poorly understood. To investigate potential metabolic and genetic determinants of statin responsiveness, 19 patients with refractory heterozygous familial hypercholesterolemia were sequentially treated with placebo, atorvastatin (10 mg/d), bile acid sequestrant, and the 2 combined, each for 4 weeks. Levels of LDL-C, mevalonic acid (MVA), 7-alpha-OH-4-cholesten-3-one, and leukocyte LDL receptor and hydroxymethylglutaryl coenzyme A reductase mRNA were determined after each treatment period. Atorvastatin (10 mg/d) reduced LDL-C by an overall mean of 32.5%. Above-average responders (LDL-C -39.5%) had higher basal MVA levels (34.4+/-6.1 micromol/L) than did below-average responders (LDL-C -23.6%, P<0.02; basal MVA 26.3+/-6.1 micromol/L, P<0.01). Fewer good responders compared with the poor responders had an apolipoprotein E4 allele (3 of 11 versus 6 of 8, respectively; P<0.05). There were no baseline differences between them in 7-alpha-OH-4-cholesten-3-one, hydroxymethylglutaryl coenzyme A reductase mRNA, or LDL receptor mRNA, but the latter increased in the good responders on combination therapy (P<0.05). Severe mutations were not more common in poor than in good responders. We conclude that poor responders to statins have a low basal rate of cholesterol synthesis that may be secondary to a genetically determined increase in cholesterol absorption, possibly mediated by apolipoprotein E4. If so, statin responsiveness could be enhanced by reducing dietary cholesterol intake or inhibiting absorption.