The management of hypercholesterolemia (HC) in children relies on their potential level of risk for premature cardiovascular disease (CVD) in adulthood (1). Among children with HC, those with dominant inheritance have a much higher risk for premature CVD and reduced life expectancy compared to sporadic or polygenic forms of HC (2). Familial HC (FH), the most prevalent form of dominantly inherited HC, is characterized by permanently elevated low-density lipoprotein cholesterol (LDLC) and premature CVD and death (3). The disease is caused by a variety of LDL receptor (LDLR) gene mutations. Homozygous FH is rare (1/1 million) and is easily recognized early in life by the presence planar xanthoma and silent or manifest atherosclerosis before age 10. By contrast, heterozygous FH (hFH) has an estimated general prevalence of 1 in 500 (4), thereby representing one of the most frequent human monogenic conditions, and has a discrete clinical presentation in childhood. The only constant and detectable sign after birth is isolated elevation of plasma LDLC (type IIa HC), but with variable levels influenced by genetic makeup, such as apolipoprotein E gene (APOE) polymorphism (5,6). Although clinically asymptomatic, children with hFH experience progressive atherosclerosis, arterial wall thickening on noninvasive imaging, and functional or biological signs of endothelial dysfunction (7,8). Recent evidence that early lesions or arterial wall alterations may be reversed by lipid-lowering therapy fosters therapeutic intervention in children with hFH (9,10). As a consequence, identification of hFH during childhood, at the time of the silent and reversible phase of the disease, seems to be of first importance.
Definite diagnosis of hFH may be obtained through genetic testing and identification of causative DNA mutations (11). However, genetic testing remains costly, requires delays for implementing complex investigations, and is usually not accessible in current practice. In adults, 3 clinical scoring systems have been implemented to identify patients with hFH as part of screening programs of asymptomatic persons at high risk for CVD (12–14). All have proved useful, cheap, and simple for case finding in common practice and as selection criteria for referral to genetic testing or clinical trials (15). However, the scoring systems presently used in adults with hFH rely mainly on detection of clinical signs of HC (eg, tendinous xanthoma) or manifestations of coronary heart disease, all signs usually absent in children (16). The purpose of the present study was to elaborate and validate in a group of mostly prepubertal children with HC a simple and accurate clinical scoring system for the diagnosis of children with hFH based on family history and biological signs.
PATIENTS AND METHODS
From a cohort of dyslipidemic patients regularly monitored at the gastroenterology and nutrition clinic of the Children's Hospital Armand Trousseau in Paris, 100 children with type IIa HC were retrospectively selected. Inclusion criteria were age older than 2 years and younger than 12 years at diagnosis, member of unrelated families, fasting plasma LDLC above 130 mg/dL or plasma total cholesterol (TC) above 240 mg/dL before any dietary or drug therapy (both corresponding to the 95th percentile in French children (17)), and result of a complete genetic testing.
Exclusion criteria were secondary HC due to obesity, defined as body mass index above age- and sex-specific 95th percentile (18); hormonal, hepatic, or renal dysfunction or diabetes likely to modify lipoprotein metabolism; plasma triglycerides above 150 mg/dL in children or their parents; and a homozygous or compound heterozygous genotype for an LDLR, or apolipoprotein B (APOB) or proprotein convertase subtilisin-like kesin type 9 (PCSK9) mutation after genetic testing.
Clinical Features in Child and Family
Data were collected from the child's clinical record and from a family questionnaire. Data collected from the child's record at time of diagnosis were as follows: sex; age, weight and height; presence of extravascular deposits (xanthelasma, xanthoma); complete fasting plasma lipid profile free of any dietary or drug intervention; whether HC was discovered fortuitously or through a family study. Data collected from the family questionnaire were as follows: presence of HC and maximal TC level in first- and second-degree relatives; premature CVD defined as coronary heart disease, stroke, or peripheral arterial disease before age 55 in men and age 65 in women; major CVD risk factors in parents (high blood pressure, current or past smokers, diabetes, overweight); lipid-lowering therapy, and type of medication, if any, in 1 or both parents.
Dietary advice usually provided for cholesterol lowering in a pediatric setting was given to parents and children by the same single professional dietitian, allowing standardization of the test diet for score elaboration. The diet, compliant with National Cholesterol Education Program and European Society of Paediatric Gastroenterology, Hepatology, and Nutrition recommendations (19,20), consisted of a balanced diet with normal daily energy intake for age, total fat intake limited to 35% of total daily energy intake, reduced saturated fat to 10% of total energy intake, proportionally increased monounsaturated and polyunsaturated fat intake, and cholesterol intake <300 mg/day. Recommendations were adapted to age, and a written document was given to the family. Response to the test diet was assessed after at least 3 to 6 months of continuous dietary intervention by 12-hour fasting lipid testing and recording by the dietitian of changes in child's dietary habits: mode of cooking (ie, added fat), snacking, delicatessen food, industrial food or meals (ie, hidden fat), type of dairy products (ie, saturated fat), number of servings of fruits, whole-grain cereals, vegetables per day.
Study Population Samples
The population selected for score elaboration (“elaboration sample”) by means of the above-mentioned criteria consisted of 50 children who were carriers of a heterozygous LDLR gene mutation and 50 children who were noncarriers of an LDLR, APOB, or PCSK9 mutation. An independent sample of 38 other children (“validation sample”) was selected from the same cohort with the same criteria to test predictive performances of the score.
Informed consent was collected from each study participant or legal representative. The study was compliant with current French regulations for genetic testing for medical purposes and was approved by the local institutional review board for clinical research in humans (Hôpital Saint Antoine, Paris) according to the principles of the Declaration of Helsinki.
Plasma Lipids and Lipoproteins
Venous blood was collected after an overnight fast. Plasma TC, triglycerides, and high-density lipoprotein cholesterol (HDLC) were assayed enzymatically in certified laboratories complying with National Quality Controls. Plasma LDLC was estimated by the formula of Friedewald (21).
DNA was extracted from blood leukocytes by a phase exchange method (Puregene, Gentra systems, Minneapolis, MN). Genomic sequences of LDLR (National Center for Biotechnology Information [NCBI] accession no. NM_000527.2), APOB (NCBI accession# NM_000384.2), and PCSK9 (NCBI accession# NM_174936.2), all shown to be proven as disease-causing genes, were amplified by polymerase chain reaction and sequenced as previously described (22,23). Proximal promoter, exonic, and periexonic regions were sequenced by the Big-Dye terminator cycle-sequencing protocol on a 16-capillary DNA sequencer (ABI 3130, Applied Biosystems, France). Electronic DNA sequence tracks were read independently by use of Seqscape software (version 3.2, Applied Biosystems, France) for mutation detection and Gensearch software (version 3.34, Phenosystems, Belgium) for mutation functionality qualification against HUGO criteria and locus-specific databases (www.ucl.ac.uk/fh;www.umd.necker.fr;http://database.jojogenetics.nl). Large LDLR gene rearrangements were investigated by Southern blotting or MLPA (Multiple Ligation-dependent Probe Amplification, MRC-Holland). APOE polymorphism (alleles E2, E3, and E4) was assessed by allele-specific hybridization of polymerase chain reaction products on nitrocellulose strips (InnoGenetics, Belgium).
The Fisher exact test was used for qualitative variables and the Mann-Whitney test for quantitative variables. A logistic regression model with backward selection was used to identify independent predictors of FH. A score was then constructed that combined the independent predictors. The diagnostic value of the score was assessed by the area under the receiver-operating characteristics curve (AUROC) (24). Because the apparent AUROC of a predictive model is overestimated when simply determined on the original sample of study participants that is used to construct the model, we used a nonparametric bootstrap procedure for calculating an internally validated estimate of AUROC (25). The predictive performance of our score was also validated in an independent group of 38 children (“validation sample”). Finally, we built a simple 2-way entry table describing the predicted probabilities of FH for given combinations of predictors defined by their quartiles. Data analysis was done with SAS software (version 9.1, SAS Institute, Cary, NC).
The characteristics and family history of 100 unrelated HC children, either carriers or noncarriers of a heterozygous LDLR mutation, are described in Table 1. Most children were prepubertal; 83% were boys (age <10) and 82% were girls (age <9). They were white, and a great majority were of European ancestry (95%). They resided in an urban area (>15 million individuals). There was no difference between both groups for age, sex ratio, body mass index, or triglyceridemia (70 ± 20 mg/dL vs 70 ± 30 mg/dL, NS). There was no within-group difference in plasma lipid parameters by age, sex, or body mass index.
At the time of diagnosis, despite similar dietary habits in both groups, children with a DNA mutation had significantly higher plasma levels of TC (28%) and LDLC (47%) than did nonmutation carriers. Differences in cholesterol levels persisted between both groups after lipid-lowering diet (TC 37% and LDLC 58% higher in carriers), despite similar dietary responses (ΔLDLC). There was no difference by time (ie, <6 months vs >6 months, NS) in postdiet changes in plasma lipid parameters, despite a wide range in time to postdiet clinical examination and lipid testing (median 7 months, range 4–12 months). Moreover, only 1 mutation carrier had LDLC <130 mg/dL after the dietary changes, whereas in 32% of noncarriers the LDLC returned to normal after dietary changes (P < 0.001). Plasma HDLC was lower in mutation carriers before and after dietary changes than compared to nonmutation carriers.
Several components of family history also differed between both groups. Heterozygous HF children were more frequently identified through family screening. Although parental age and frequency of major CVD risk factors was similar in both groups, maximal plasma TC was higher and statins were more frequently used in parents of hHF children. The prevalence of CVD was low and similar in parents from both groups. However, a higher prevalence of premature CVD and death was observed in the grandparents of children with a genetic mutation.
Genetic Characteristics of HC Children
There was no difference in APOE allelic frequency distribution as a function of mutation carrier status: E2 = 0.04; E3 = 0.84; E4 = 0.12 (carrier) versus E2 = 0.02, E3 = 0.82, E4 = 0.16 (noncarrier). There were no within-group or between-group differences in plasma lipids before or after dietary changes as a function of APOE allele carrier status.
Among the 50 children who were carriers of a heterozygous LDLR mutation, 25 were carriers of a missense mutation and were classified as defective mutation carriers; the remainder were carriers of nonsense (n = 7), small insertions, or deletions resulting in a frameshift (n = 10) or splice site mutations (n = 8), all classified as negative mutation carriers. Defective mutation carriers had lower TC and LDLC at baseline (TC 308 ± 51 vs 351 ± 56 mg/dL, P = 0.007; LDLC 242 ± 49 vs 284 ± 52 mg/dL, P = 0.005) and after dietary changes continued for more than 3 months (TC 273 ± 47 vs 321 ± 56 mg/dL, P = 0.002; LDLC 202 ± 51 vs 247 ± 53 mg/dL, P = 0.004), compared with negative mutation carriers. They did not differ otherwise by other personal signs or components of family history.
Clinical Score for Identification of hFH in Children
Of 60 distinct clinicobiological items from the child's record or the family questionnaire, multivariate logistic regression analysis identified only 3 independent factors strongly associated with hFH: LDLC at diagnosis before dietary intervention (LDLb in mg/dL): odds ratio (OR) 1.034 (95% confidence interval [CI] 1.015–1.054), P < 0.001; LDLC after dietary intervention (LDLd in mg/dL): OR 1.030 (95% CI 1.011–1.05), P = 0.002; and statin therapy in the HC parent (yes vs no) OR 6.2 (95% CI 1.4–28.3), P = 0.018. In other words, mutation carrier risk increased by 35% to 40% per 10 mg/dL increase of LDLC and was further increased by parental statin usage.
These 3 independent hFH predictors were incorporated into the scoring function. The score was calculated as follows: −12.8 + 0.034 × LDLb (mg/dL) + 0.03 × LDLd (mg/dL) + 1.8 if statin therapy in HC parent. The diagnosis precision of the score assessed by the AUROC on the elaboration sample was 0.94 (SD 0.02) 95% CI (0.91–0.98) (Fig. 1). When applied to the independent validation sample, the AUROC for the score was 0.90 (SD 0.05) 95% CI (0.79–1.00).
Probability Classes for Simple Use of the Score
For practical use, probabilities were derived from the score (P = 1/(1+exp(−score)). Table charts were built to predict the individual probability for an HC child to have hFH, from observed values (within corresponding quartiles) of LDLC at diagnosis, and after more than 3 months of using a test diet and from statin usage in an HC parent (Fig. 2). The children were further classified into 4 classes of predicted probabilities: definite hFH (P ≥ 75%), probable hFH (50≤P≤74%), possible hFH (25≤P≤49%), and improbable hFH (P < 25%). In the elaboration sample, 90% children with a disease-causing mutation (45/50) were in hFH classes (definite or probable or possible hFH), whereas about 3 of 4 of children without such mutations (37/50) were in the improbable hFH class (Fig. 3). This picture was confirmed in the validation sample, inasmuch as 22 of 25 mutation-carrier children (88%) were classified as having hFH by the score (either definite, probable, or possible FH) and 77% noncarriers were classified as having improbable hFH. The genotypic distribution in mutation carriers was similar in the validation sample (11 defective mutation and 14 negative mutation carriers). However, 10% children of the elaboration sample and 12% children of the validation sample with a causative DNA mutation were predicted to have improbable hFH and were obviously misclassified. It is noteworthy, however, that the high discriminative power of the clinical scoring was indicated by low frequencies of mutation carriers or noncarriers in the intermediate classes (probable or possible FH).
Justification for Early Diagnosis and Specific Preventive Care in Children With hFH
There is now consistent evidence that FH patients are at high risk for premature CVD and death, compared the risk in the general population. Pioneering work by Goldstein et al (3) showed that 50% of men and 20% of women with hFH will experience a first major cardiovascular event before age 50. Data from the Simon Broome Register Group in the United Kingdom (2) confirmed in 1185 hFH patients followed up for 16 years that the relative risk of premature coronary death was multiplied by 125 in women and 48 in men before age 40, compared with the general population. As much as 9% of premature major coronary events may result from hFH (26). In hFH, plasma LDLC is already at twice the normal levels as early as the first month of life, and it remains fixed at these levels throughout lifetime. Pathological studies have shown that fatty streaks and fibrous plaques are detectable in the aorta and coronary arteries in children with HC as young as 2 years old, the extent being strongly associated with plasma LDLC level (27). Signs of silent atherosclerosis have been reported in children with hFH by noninvasive imaging studies (28). Endothelial dysfunction and increased intima–media thickness of the arterial wall have been shown to be relevant predictors of future cardiovascular events. That these alterations may be reversed by statin therapy in children further justifies early and targeted intervention in children with hFH as part of primary prevention of CVD.
Usefulness of a Simple Diagnostic Score for hHF in Children
At present, simple tools are lacking in current practice to precisely identify children with hHF at high risk for premature CVD, from children with polygenic HC in whom risk may be lower or less precisely defined. Owing to the rarity of clinical signs, the diagnosis of hFH in children with HC currently relies on family history and cascade testing in families with FH (16,29,30). This strategy may not apply to split families living in urbanized societies and to adoptees. To identify as many high-risk HC children as possible, the present score was established from logistic regression analysis of more than 60 potential clinical and biological parameters obtained from the patients' clinical records and from family questionnaires.
Three independent clinical and biological signs usually collected in current pediatric practice were identified in HC children. Indeed, these parameters closely reflect the molecular mechanisms of the disease: altered LDL clearance from plasma through deficient LDL receptor-mediated endocytosis (accounting for high fasting LDLC at diagnosis and for its poor sensitivity to a diet focused on cholesterol and saturated fat reduction), and dominant inheritance of permanently high LDLC (accounting for statin usage in an HC parent). By contrast, other classic predictors of FH were not found (eg, higher rate of premature CVD in parents) or were found but with a lower predictive value, such as low symptom prevalence (absence of xanthoma) in a small study sample. This could reflect contemporary medical practice in a Western urbanized population (eg, fasting plasma LDLC measurements in child and statin usage in parents) in which appropriate screening and early treatment may be proposed early in life. Thus, by preventing parental generation from early CVD events, a classic hallmark of the natural history of FH appeared with low frequency and discriminative power in our study population. Indeed, premature CVD was still obvious in the grandparents' generation (ie, higher rate of CVD in FH grandparents) who were not offered efficient cholesterol-lowering therapies earlier on. Eventually, the highest and most independent predictors were incorporated into a scoring function, further internally validated in the score elaboration sample by the robust method of AUROC estimation. Scoring algorithm was validated on an independent sample from the same cohort of HC children.
Scoring systems were proposed for adult case finding with FH in the general population (13–16). The MedPed scoring system was mathematically derived from observations of 5 large UTAH FH pedigrees, whereas the Simon Broome Register and Dutch Lipid Clinics scoring systems were intuitively designed and validated a posteriori in different populations. The US-MedPed scoring uses TC and LDLC cut points to identify people with FH and is more likely adapted to adults. When the MedPed LDLC cut point of 200 mg/dL in youth (age <18 years) was applied to the present population of prepubertal HC children, sensitivity was high (92%), but a significant number of children were misclassified (36%). This overlap has been repeatedly described in adults. A recent study in Austria found that conventional laboratory methods would miss at least 21% of FH children with LDLR mutations (31). In keeping with previous observations for adult scoring systems, the present scoring system led us to a similar misclassification rate in children (36%). Remarkably, however, the false-negative rate (12%) was lower with our scoring system, indicating that most misclassified children were negative mutation carriers (false positives). This inherent imprecision of any clinical screening strategy for FH, recently underlined in a meta-analysis (32), emphasizes the relevance of genetic testing for definite diagnosis of FH and screening purposes in affected families (11,15,30).
Genetic testing remains costly, however, and is restricted to specialized centers, making this approach impracticable for decision making at the time of pediatric consultation. This is the reason why we used molecular genetics to elaborate a diagnosis scoring system in a population of children with type IIa HC at a time in their lives when the clinical and biological phenotype may be closer to its genetic determinants than it will be later in life. The clinical and biological signs in children with hFH were similar to those previously reported for this age range in other populations of white background (3,7,29,32), and they were found similarly influenced by the type of defective allele at the LDLR locus (5,6).
Inasmuch as the present study was restricted to a monocentric pediatric population, further validation in a multicentric pediatric or family practice setting, enlarged to broader age ranges or to other populations, is needed to determine whether the present scoring is more widely applicable. Experience from adult FH patients, however, suggests that indeed such studies could further consolidate the initial findings (15). To remain close to usual practice, dietary recommendations and monitoring were not restrictive or strictly standardized, and delays for testing diet efficiency were variable, which may limit the confidence given to postdiet LDLC, a key component of the score. One may also argue that statin usage in a parent may not be easily applicable in certain populations, compared with fasting lipid testing in parents. However, objectives of a diagnosis scoring system are mainly pragmatic. Considering that 2 of 3 adult scoring systems for FH were intuitively defined, we chose a more robust statistical approach to identify the best indicators of the disease in children with HC in the setting of present common medical practice. Nevertheless, care was taken not to be excessively restrictive in defining clinical criteria, family history, or dietary recommendations or their monitoring, to allow broader use of the score, even by nonspecialists or specialized centers.
Another limitation is that score elaboration was restricted to a FH population with LDLR mutations only. In mixed populations, LDLR mutations may cause up to 75% of cases in which HC is dominantly inherited (4). However, dominant HC may result from other known or unknown monogenic causes (33,34). Mutations within the LDLR-binding domain of APOB are a cause of familial defective APOB-100, a disease closely resembling hFH in children. However, familial defective APOB-100 usually does not exceed 5% of genetic causes of inherited HC (35). More recently, missense mutations causing a gain-of-function and a phenotype of hFH were identified on PCSK9, a regulator of liver LDLR activity and of LDL metabolism (34). However, PCSK9 disease-causing mutations remain a rare cause of FH (23). Because our aim was to build a scoring system from a homogenous group of hFH children, we excluded these cases from the score elaboration study sample to avoid potential genotypic biases.
In addition, the present scoring system found a non-negligible number of hFH children who were obviously carriers of a causative mutation, classified as improbable hFH. A penetrance of 90% is a hallmark of FH; consequently, normolipidemia has been repeatedly reported in a small proportion of adult or pediatric patients with hFH (3,15,16,30,31,36). Not all forms of HC with apparent family clustering are monogenic, however, and may also be a cause of misclassification of non-FH patients. In any case, any best screening tool may not substitute for careful clinical evaluation of the child in the context of the child's family and environment, so that in clinical practice, misclassification may be eventually corrected.
The final aim of this score is for case finding and use as a surrogate to genetic tests in common practice. This simple tool, built from criteria closely reflecting the mechanism of disease, may help physicians to identify children with hFH and to take appropriate preventive measures for children at high risk for premature CVD.
1. McCrindle BW, Urbina EM, Dennison BA, et al
. Drug therapy of high-risk lipid abnormalities in children and adolescents: a scientific statement from the American Heart Association Atherosclerosis, Hypertension, and Obesity in Youth Committee, Council of Cardiovascular Disease in the Young, with the Council on Cardiovascular Nursing. Circulation 2007; 115:1948–1967.
2. Scientific Steering Committee on Behalf of the Simon Broome Register Group. Mortality in treated heterozygous familial hypercholesterolemia: implication for clinical management. Atherosclerosis
3. Goldstein JL, Hobbs HH, Brown MS. Familial hypercholesterolemia. In: Scriver CR, Beaudet AL, Sly WS, et al., editors. The Metabolic and Molecular Basis of Inherited Diseases. New York: McGraw-Hill; 2001. pp. 2863–2913.
4. Austin MA, Hutter CM, Zimmern RL, et al
. Genetic causes of monogenic heterozygous familial hypercholesterolemia: a HuGE prevalence review. Am J Epidemiol 2004; 160:407–420.
5. Torres AL, Moorjani S, Vohl MC, et al
. Heterozygous familial hypercholesterolemia in children: low-density lipoprotein
receptor mutational analysis and variation in the expression of plasma lipoprotein-lipid concentrations. Atherosclerosis 1996; 126:163–171.
6. Wiegman A, Sijbrands EJ, Rodenburg J, et al
. The apolipoprotein epsilon4 allele confers additional risk in children with familial hypercholesterolemia. Pediatr Res 2003; 53:1008–1012.
7. Tonstad S, Joakimsen O, Stensland-Bugge E, et al
. Risk factors
related to carotid intima-media thickness and plaque in children with familial hypercholesterolemia and control subjects. Arterioscler Thromb Vasc Biol 1996; 16:984–991.
8. Aggoun Y, Bonnet D, Sidi D, et al
. Arterial mechanical changes in children with familial hypercholesterolemia. Arterioscler Thromb Vasc Biol 2000; 20:2070–2075.
9. de Jongh S, Lilien MR, op't Roodt J, et al
. Early statin therapy restores endothelial function in children with familial hypercholesterolemia. J Am Coll Cardiol 2002; 40:2117–2121.
10. Wiegman A, Hutten BH, de Groot E, et al
. Efficacy and safety of statin therapy in children with familial hypercholesterolemia. JAMA 2004; 292:331–337.
11. van Aalst-Cohen ES, Jansen AC, Tanck MW, et al
. Diagnosing familial hypercholesterolaemia: the relevance of genetic testing. Eur Heart J 2006; 27:2240–2246.
12. Scientific Steering Committee on Behalf of the Simon Broome Register Group. Risk of fatal coronary heart disease in familial hypercholesterolaemia. BMJ
13. Williams RR, Hunt SC, Schumacher MC, et al
. Diagnosing heterozygous familial hypercholesterolemia using new practical criteria validated by molecular genetics
. Am J Cardiol 1993; 72:171–176.
14. World Health Organization Human Genetics
DoNDP. Familial Hypercholesterolemia: Report of a Second WHO Consultation. Geneva: WHO; 1999.
15. Damgaard D, Larsen ML, Nissen PH, et al
. The relationship of molecular genetic to clinical diagnosis of familial hypercholesterolemia in a Danish population. Atherosclerosis 2005; 180:155–160.
16. Ose L, Tonstad S. The detection and management of dyslipidaemia in children and adolescents. Acta Paediatr 1995; 84:1213–1215.
17. Lambert D, Daubrosse E, Steinmetz J, et al
. Valeurs de référence du cholestérol et des triacylglycerols plasmatiques et lipoprotéiniques chez des enfants sains. Arch Fr Pediatr 1984; 41:145–150.
18. Rolland-Cachera MF, Cole TJ, Sempé M, et al
. Body mass index variation: centiles from birth to 87 years. Eur J Clin Nutr 1991; 45:13–21.
19. National Cholesterol Education Program. Report of the expert panel on blood cholesterol levels in children and adolescents. Pediatrics
20. ESPGAN Committee on Nutrition. Committee report: childhood diet and prevention of coronary heart disease. J Pediatr Gastroenterol Nutr
21. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low density lipoprotein cholesterol in plasma without use of the ultracentrifuge. Clin Chem 1972; 18:499–502.
22. Amsellem S, Briffaut D, Carrié A, et al
. Intronic mutations outside of Alu-repeat rich domains of the LDL receptor gene are a cause of familial hypercholesterolemia. Hum Genet 2002; 111:501–510.
23. Allard D, Amsellem S, Abifadel M, et al
. Novel mutations of the PCSK9 gene cause variable phenotype of autosomal dominant hypercholesterolemia. Hum Mutat 2005; 26:497.
24. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143:29–36.
25. Steyerberg EW, Harrel FE Jr, Borsboom GJ, et al
. Internal validation of predictive models: efficiency of some procedures for logistic regression analysis. J Clin Epidemiol 2001; 54:774–781.
26. Civeira F. Guidelines for the diagnosis and management of heterozygous familial hypercholesterolemia. Atherosclerosis 2004; 173:55–68.
27. Berenson GS. Childhood risk factors
predict adult risk associated with subclinical cardiovascular disease: the Bogalusa heart study. Am J Cardiol 2002; 90:3L–7L.
28. Wiegman A, De Groot E, Hutten BA, et al
. Arterial intima-media thickness in children heterozygous for familial hypercholesterolemia. Lancet 2004; 363:369–370.
29. Wiegman A, Rodenburg J, De Jongh S, et al
. Family history and cardiovascular risk in familial hypercholesterolemia: data in more than 1000 children. Circulation 2003; 107:1473–1478.
30. Hadfield SG, Humphries SE. Implementation of cascade testing for the detection of familial hypercholesterolaemia. Curr Opin Lipidol 2005; 16:428–433.
31. Widham K, Dirisamer A, Lindemayr A, et al
. Diagnosis of families with familial hypercholesterolemia and/or ApoB-100 defect by means of DNA analysis of LDL-receptor gene mutations. J Inherit Metab Dis 2007; 30:239–247.
32. Wald DS, Bestwick JP, Wald NJ. Child-parent screening for familial hypercholesterolaemia: screening strategy based on a meta-analysis. BMJ 2007; 335:599–606.
33. Innerarity TL, Mahley RW, Weisgraber KH, et al
. Familial defective apolipoprotein B-100: a mutation of apolipoprotein B that causes hypercholesterolemia. J Lipid Res 1990; 31:1337–1349.
34. Horton JD, Cohen JC, Hobbs HH. Molecular biology of PCSK9: its role in LDL metabolism. Trends Biochem Sci 2007; 32:71–77.
35. Viola S, Benlian P, Morali A, et al
. Apolipoprotein B Arg3500Gln mutation prevalence in children with known hypercholesterolemia: a French multicenter study. J Pediatr Gastroenterol Nutr 2001; 33:122–126.
36. Knoblauch H, Muller Myhsok B, Busjahn A, et al
. A cholesterol-lowering gene maps to chromosome 13q. Am J Hum Genet 2000; 66:157–166.
Keywords:© 2009 Lippincott Williams & Wilkins, Inc.
Genetics; Hyperlipidemia; Low-density lipoprotein; Low-density lipoprotein receptor mutation; Risk factors