HFE C282Y homozygotes have reduced LDL cholesterol: the Atherosclerosis Risk in Communities (ARIC) Study

Recent studies have raised questions about the long-term health risks for individuals with mutations in the HFE gene, although previous studies may have been plagued by selection bias or lack of population-based comparison groups. We examined cardiovascular disease risk factors, iron and liver biomarkers, and morbidity and mortality associated with the C282Y and H63D variants of HFE in the ARIC study, a population-based cohort of nearly 16,000 U.S. white and black men who were 45–64 years old at baseline. Subjects were followed for an average of 15 years for death, incident coronary heart disease, stroke, and heart failure, and 8 years for incident diabetes. The prevalence of C282Y homozygosity was 0.42% (45/10,800) in whites, similar to other North American populationbased studies. C282Y homozygotes had statistically significantly lower mean LDL cholesterol and fibrinogen and higher mean levels of iron (ferritin, transferrin saturation) and liver biomarkers (alanine aminotransferase, Hepascore) compared to HFE wild-type subjects. Rates of all-cause mortality, cardiovascular disease, and diabetes were similar across HFE genotypes. These prospective, population-based data indicate higher serum iron indices and possible mild liver dysfunction or disease in some C282Y homozygotes, but provide little evidence that HFE C282Y or H63D mutations are related to all-cause mortality, cardiovascular disease, or diabetes. Reduced LDL in C282Y homozygotes may be due to effect of excess iron on cholesterol metabolism and lipoprotein formation in the liver. Introduction Hereditary hemochromatosis is a genetic disorder that may lead to excessive iron accumulation in various body tissues. Shortly after the discovery in 1996 that homozygosity for the 845 G>A [C282Y] mutation in the hemochromatosis gene (HFE) was the principal cause of hereditary hemochromatosis in individuals of northern European descent [1], it was assumed that most C282Y homozygotes would eventually develop excessive body iron deposition and clinically overt disease, most notably in the liver, heart, pancreas, joints, or pituitary [2]. However, large screening studies from Norway [3] and southern California [4] showed that Correspondence and reprints: James S. Pankow, Division of Epidemiology and Community Health, University of Minnesota, 1300 South Second Street, Suite 300, Minneapolis, MN 55454, Phone: 612-624-2883, Fax: 612-624-0315, E-mail: [email protected]. Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. NIH Public Access Author Manuscript Transl Res. Author manuscript; available in PMC 2009 July 1. Published in final edited form as: Transl Res. 2008 July ; 152(1): 3–10. doi:10.1016/j.trsl.2008.05.005. N IH PA Athor M anscript N IH PA Athor M anscript N IH PA Athor M anscript the prevalence of clinical disease in target organs was similar in C282Y homozygotes compared to subjects without the mutation, and none of the 23 homozygotes in another population-based developed clinical signs or symptoms suggesting hereditary hemochromatosis during 25 years of follow-up [5]. Some of these studies have been criticized for a failure to ascertain subclinical disease, such as liver fibrosis [6]. Further concerns have been raised about the possibility of selection bias, as individuals who already had clinical manifestations were unlikely to be screened or may have died prematurely [7], although at least in some studies, evaluation of Hardy-Weinberg equilibrium found no deficit in the anticipated number of C282Y homozygotes [8]. Although it is unlikely that the vast majority of C282Y homozygotes eventually develop serious morbidity and early mortality, as was suggested less than a decade ago [2], the clinical significance of HFE mutations remains uncertain [9]. The U.S. Preventive Services Task Force recently concluded that there is “insufficient evidence to confidently project the impact of, or estimate the benefit from, widespread or high-risk genetic screening for hereditary hemochromatosis [9].” Because health benefits of screening for a disease depend largely on the natural history, determining the incidence of clinically significant morbidity or mortality related to HFE mutations is highly relevant. In the present report, we evaluated morbidity and mortality associated with the C282Y and H63D mutations of HFE in a U.S. cohort ascertained using carefully designed and executed population-based sampling techniques. Methods Study population In 1987–89 the Atherosclerosis Risk in Communities (ARIC) Study enrolled 15,792 subjects ages 45–64 years in four U.S. locations: Forsyth County, NC; Jackson, MS; seven northwestern suburbs of Minneapolis, MN; and Washington County, MD [10]. Black residents were oversampled as part of the area-probability sampling of Forsyth County, NC, while enrollment at the Jackson, MS, site was restricted to black residents. Eligible residents were invited to participate in a baseline clinical examination (visit 1) and three subsequent follow-up examinations approximately three years apart (visits 2–4). For the present analysis, we excluded 48 participants who were not white or black, 43 who did not provide consent for genetic studies, and 1,216 with missing data for either the C282Y or H63D variants, leaving 14,485 subjects (10,800 whites and 3685 blacks) available for analysis. The research was carried out according to the principles of the Declaration of Helsinki. Institutional review boards at each study institution provided approval and all participants gave informed consent. Baseline data collection At baseline, interviewers collected information on menopausal status, smoking status, and previous diagnoses, including cancer [10]. Fasting lipids, glucose, insulin, and fibrinogen were measured at a central laboratory using methods described previously [11–13], and hematology was performed at local laboratories. Prevalent coronary heart disease was defined as a selfreported history of a physician-diagnosed heart attack, evidence of an old myocardial infarction (MI) by electrocardiogram based on the Minnesota codes, or reported history of coronary revascularization. Prevalent stroke was defined as a reported stroke diagnosed by a physician. Prevalent diabetes was defined as reported physician diagnosis, use of anti-diabetes medications, fasting (≥ 8hr post prandial) serum glucose ≥ 7 mmol/L (140 mg/dL), or nonfasting glucose of ≥ 11.1 mmol/L (200 mg/dL). Prevalent heart failure was defined as reported current intake of heart failure medication at visit 1 or evidence of manifest heart failure defined by the Gothenburg criteria stage 3 [14,15]. Carotid intimal-medial thickness was measured by B-mode ultrasound [16,17] and the mean intimal-medial thickness across three segments Pankow et al. Page 2 Transl Res. Author manuscript; available in PMC 2009 July 1. N IH PA Athor M anscript N IH PA Athor M anscript N IH PA Athor M anscript (internal carotid, bifurcation, common carotid) was used in analysis. History of fatigue was assessed at visit 2 by the question “Do you often feel tired?” History of arthritis was assessed at visit 4 by the question “Has a doctor ever told you that you have arthritis?” HFE variants were typed at the ARIC DNA Laboratory, University of Texas Health Sciences Center at Houston using a standard Taqman method. All C282Y homozgyotes (C282Y/C282Y) were verified at CLIA-certified Clinical Laboratories of the University of Minnesota Medical Center, Fairview using a restriction endonuclease method [1]. Iron and liver biomarkers Because of limited resources, iron and liver biomarkers were measured only in C282Y homozygotes and a comparison group of 48 wild-type subjects randomly selected from white ARIC controls in the Longitudinal Investigation of Thromboembolism Etiology study [18], frequency matched to C282Y homozygotes by age group (45–54, 55–64, or 65–74 years) at the time of venipuncture. When available, serum from the most recent full cohort exam (visit 4, 1996–98) was used; serum from visit 2 (n=8) or visit 1 (n=3) was used for a few subjects, and was unavailable for all visits in one C282Y homozygote. Transferrin saturation was measured using automated spectrophotometric measurement of iron and unsaturated iron binding capacity on a Roche/Hitachi 911 analyzer using commercially available reagents (Roche Diagnostics/Boehringer Mannheim Corp., Indianapolis, IN). Ferritin, alanine aminotransferase, γ-glutamyltransferase, and total bilirubin were measured using the same equipment and commercially available assays from Roche Diagnostics. Alpha-2macroglobulin was assayed using the Immage analyzer and commercially available reagents (Beckman-Coulter, Inc., Brea, CA), and hyaluronic acid was measured using an ELISA assay kit (Corgenix Inc., Westminster, CO). Coefficients of variation for all assays were less than 5.4% except ferritin and unsaturated iron binding capacity, which were 9.9% and 8.4%, respectively. To assess liver fibrosis, we used a predictive model (Hepascore) based on bilirubin, γ-glutamyltransferase, hyaluronic acid, α2-macroglobulin, age, and sex previously validated in subjects with chronic hepatitis C infection [19]. Morbidity and mortality follow-up Liver disease was assessed at visit 3 by the question “Has a doctor ever said you had cirrhosis or another chronic liver disease?” and searches of underlying and contributing causes of death (ICD-9 570–573 or 155; ICD-10 K7 or C22) and hospital discharge diagnoses (ICD-9 CM 570–573 or 155). Clinical diagnosis of hemochromatosis was ascertained as a discharge diagnosis of hemochromatosis (ICD-9 CM 275.0). All-cause mortality was ascertained by reviews of death certificates, annual follow-up interviews, hospital charts, and other means. Details on quality assurance for ascertainment and classification of coronary heart disease (CHD) and stroke events have been published elsewhere [20–22]. Incident CHD events were defined as a validated definite or probable hospitalized MI, definite CHD death, unrecognized myocardial infarction defined by ARIC electrocardiogram readings, or coronary revascularization. Incident ischemic stroke was defined as validated definite or probable hospitalized embolic or thrombotic brain infarctions. Incident heart failure was defined as a hospital discharge diagnosis coded with heart failure at any position (ICD-9 CM 428) or death with heart failure coded as the underlying cause of death (ICD-9 428; ICD-10 I50) [23]. Subjects were defined as having incident diabetes if they met any of the criteria listed above for prevalent diabetes at any of the follow-up visits with date of diagnosis interpolated between visits [24]. Statistical analysis We tested the C282Y and H63D variants separately for Hardy Weinberg equilibrium using χ2 goodness of fit tests. Pairwise differences in baseline characteristics between each combined Pankow et al. Page 3 Transl Res. Author manuscript; available in PMC 2009 July 1. N IH PA Athor M anscript N IH PA Athor M anscript N IH PA Athor M anscript C282Y and H63D genotype group and a common reference group of subjects wild-type for both variants were tested by analysis of variance (continuous measures) or logistic regression (dichotomous measures) as implemented in SAS (version 9.1, SAS Institute Inc., Cary, NC). For iron and liver biomarkers, differences between C282Y homozygotes and wild-type subjects were tested by analysis of covariance, adjusting for age (matching variable) and sex. Triglycerides, insulin, ferritin, bilirubin, γ-glutamyltransferase, hyaluronic acid, and alanine aminotransferase were log-transformed before analysis. For longitudinal outcomes, we calculated person-years at risk for each participant as the time between the baseline examination date and the last date of follow-up (e.g., December 31, 2004 for all outcomes except diabetes), date of loss-to-follow-up, date of death, or date of disease diagnosis. We used the Epitab procedure in STATA (release 8, StataCorp, College Station, TX) to estimate rate ratios and exact confidence intervals. Results Genotype frequencies for C282Y and H63D did not deviate statistically from those expected under Hardy-Weinberg equilibrium in whites or blacks (p>0.05). Among whites, there were slightly more C282Y homozygotes (n=45) than expected under Hardy-Weinberg equilibrium conditions (n=43.3). Only four black subjects were either C282Y homozygotes or C282Y/ H63D compound heterozygotes (Table 1), thereby precluding detailed analyses of those genotypes. All further analyses were restricted to whites. Mean LDL cholesterol and fibrinogen levels at baseline were substantially lower among C282Y homozygotes compared to wild-type subjects (Table 2). Similar patterns were observed for LDL during follow-up: mean levels were 0.31 mmol/L (12 mg/dL; p=0.06), 0.44 mmol/L (17 mg/dL; p=0.007), and 0.47 mmol/L (18 mg/dL; p=0.003) lower in C282Y homozygotes at visits 2, 3, and 4, respectively. No adjustment was made for lipid-lowering medications in these analyses. At visit 4, self-reported use of lipid-lowering medications was nearly twice as common among wild-type subjects than C282Y homozygotes (16% vs. 9%), although this difference was not statistically significant. At baseline, mean hematocrit and hemoglobin were higher in carriers of the C282Y or H63D mutations compared to wild-type subjects, although for C282Y homozygotes differences were not statistically significant. Among C282Y homozygotes, 11 (25%) had ferritin levels greater than 1000 μg/L, while none of the 48 age-matched wild-type subjects had levels exceeding this cutpoint. Mean ferritin, transferrin saturation, and total iron were significantly higher while unsaturated and total iron binding capacities were significantly lower in C282Y homozygotes (Table 3). The Hepascore index was significantly higher in C282Y homozygotes (p=0.003), although there were only small differences for individual components of the index, including bilirubin, γglutamyltransferase, hyaluronic acid, and α2-macroglobulin. Mean length of follow-up was 15 years for all-cause mortality, CHD, stroke, and heart failure, and 8 years for diabetes. Rates of all-cause mortality, incident CHD, stroke, heart failure and diabetes were similar across HFE genotypes (Table 4). Five C282Y homozygotes (11%) had clinical evidence of liver disease during follow-up, including two subjects who died with an underlying cause of death coded as hepatocellular carcinoma and cholangiocarcinoma, respectively, two subjects with hospital discharge diagnoses indicating liver disease, and one subject who self-reported liver disease at visit 3. Among wild-type subjects, 3% had clinical evidence of liver disease. Prior to any genetic testing as part of this study, four C282Y homozygotes (9%) had a hospital discharge diagnosis of hemochromatosis during follow-up, while only one wild-type subject (0.01%) had such a diagnosis. Pankow et al. Page 4 Transl Res. Author manuscript; available in PMC 2009 July 1. N IH PA Athor M anscript N IH PA Athor M anscript N IH PA Athor M anscript