Does the LDL receptor play a role in the risk of developing type 2 diabetes?

Abstract

Over the last 5 years, major randomized clinical trials and meta-analyses of these trials have indicated that statin therapy, which inhibits 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR), is associatedwith amodestly higher risk of developing diabetes in a dose-dependent fashion.1,2 Statins act by reducing the synthesis of cholesterol, which then drives an increase in hepatic low-density lipoprotein (LDL) receptor expression, the key step leading to a reduction in circulatingLDLcholesterol levels. Further supportive data for this relationship have emerged from genetic studies, typically less prone to confounding than observational studies,whichhave linkedgeneticallydetermined lower LDL cholesterol to higher diabetes risk.3,4 In one such study, individualswith lower circulatingLDLcholesterol due tovariants in theHMGCR gene, a proxy for statin therapy, displayed higher glucose levels, higher weight, and increased diabetes risk,3 consistentwithdata fromrandomizedclinical trials.However, the exact mechanism by which any such effect is mediated remains unclear. Heterozygous familial hypercholesterolemia is an autosomal dominant condition now thought to affect as many as 1 in every 250 to 300 people.5,6 Familial hypercholesterolemia exposes affected individuals to lifelong higher circulating LDL cholesterol levels and, as a consequence, markedly elevated cardiovascular risk. Genetic testing for familial hypercholesterolemia typically focuses on mutations that can occur in 3 genes: most commonly the LDL receptor (LDLR) gene, less frequently the apolipoprotein B (APOB) gene, and, very infrequently, the proprotein convertase subtilisin/kexin type 9 (PCSK9) gene. Such mutations result in a reduction in the expression of LDL receptors or in their capacity to transport LDL cholesterol into the cell, the opposite of what occurs during statin treatment. More than a thousand different familial hypercholesterolemiamutationshavebeen identified, and thevalue inusing a genetic test to identify a patient, or index case, with such a mutation is that familymembers can then be investigated for the samemutation, so-calledcascade testing.Apioneeringnational program for identifying patients with familial hypercholesterolemia with genetic testing and cascade testing affected families has been followed in the Netherlands for the last 20 years.7 This program has provided much of the evidence on which treatment of familial hypercholesterolemia, including in children, is based. The familial hypercholesterolemia database now includes information on more than 63 000individualsgenetically testedfor theconditionofwhom 7%were tested due to being possible familial hypercholesterolemia index cases and 93% were cascade tested due to having an affected family member. Anecdotal observations have suggested that patients with familial hypercholesterolemia seldom have type 2 diabetes. Until now, however, the only relevant published data were from a small case-control comparison of ageand sex-matched patients (N = 204) with or without familial hypercholesterolemia and undergoing coronary angiography. The prevalence of diabetes among those with familial hypercholesterolemia was only a third of that observed among individuals without.8 In this issue of JAMA, Besseling and colleagues9 report cross-sectional dataon theprevalenceofdiabetes in individualswith a familial hypercholesterolemiamutation (n = 25 137) and individuals without (n = 38 183) in the Netherlands. The authors found that individuals with familial hypercholesterolemiamutationswere far less likely to report a history of diabetes than those without in unadjusted analyses (prevalence, 1.75%withmutation vs 2.93%without mutation; odds ratio [OR], 0.62 [95% CI, 0.55-0.69]). When adjusted for potential confounders includingage,bodymass index, andstatin use, the prevalencewas 1.44% (OR, 0.49 [95%CI, 0.41-0.58]). These findings remained robust in comparisons stratified according to age, statin use, smoking status, and index or cascade status. Furthermore, therewas evidenceof adose-dependent association, such that persons with an LDLR mutation had the lowest diabetes prevalence (1.63%), whereas those having APOB mutations were at intermediate diabetes risk (prevalence, 2.42%), findings that are consistent with the more severe phenotype andhigher LDL cholesterol observed inLDLR mutation carriers. Similarly, individuals with LDLR negative mutations had lower diabetes prevalence (1.41%) than LDLR defective mutation carriers (1.80%). The authors acknowledge several important weaknesses in the study, such as its cross-sectional design, the use of self-reported diabetes, survival bias, and the likelihood that some unmeasured confounders were present, and their conclusions are appropriately cautious. Nonetheless, no better study currently exists to address this question, and the fact that this massive cohort is dominated by cascaded family members, divided into groups with familial hypercholesterolemia and without based on mutation status, which is randomly allocated at birth, should limit the influence of confounders compared with conventional observational Related article page 1029 Opinion