Available online at www.sciencedirect.com

(2009) 899–903

Clinical Biochemistry 42

Genetic diagnosis of familial hypercholesterolemia using aDNA-array based platform

Rodrigo Alonso a, Joep C. Defesche b,⁎, Diego Tejedor c, Sergio Castillo d, Marianne Stef c,Nelva Mata e, Pilar Gomez-Enterria f, Ceferino Martinez-Faedo f, Lluis Forga g, Pedro Mata a

a Lipid Clinic, Fundación Jiménez Díaz, Madrid, Spainb Department of Vascular Medicine, room G1-105, Academic Medical Center at the University of Amsterdam, P.O. Box 22 660,

NL – 1100 DD Amsterdam, The Netherlandsc Progenika Biopharma SA, Derio, Spain

d Genetic Diagnosis Laboratory, Lacer SA, Barcelona, Spaine Department of Epidemiology, Institute of Public Health, Madrid Regional Authority for Health and Consumer Affairs, Spain

f Department of Endocrinology, Hospital Central de Asturias, Oviedo, Spaing Department of Endocrinology, Hospital de Navarra, Pamplona, Spain

Received 21 November 2008; received in revised form 23 January 2009; accepted 24 January 2009Available online 6 February 2009

Abstract

Objectives: The aim of this study was to validate the Lipochip® genetic diagnostic platform by assessing effectiveness, sensitivity, specificityand costs for the identification of patients with familial hypercholesterolemia (FH) in Spain. This platform includes the use of a DNA micro array,the detection of large gene rearrangements and the complete resequencing of the low-density lipoprotein receptor gene.

Design and methods: DNA samples of patients with clinically diagnosed FH were analyzed for mutations by application of the Lipochip®

platform. Results obtained were confirmed by DNA sequencing and MLPA analysis by two other, independent laboratories.Results: Of 808 patients tested, Lipochip® detected a mutation in 66% of the cases and of these 78% were detected by the micro array. A

specificity of 99.5% at a sensitivity of 99.8% was reached. A positive test result could be reported within 22 days after start of analysis. The totalaverage screening costs of $ 350 per case were significantly lower compared to other existing screening programs.

Conclusion: Lipochip® provides a reliable, fast and cheap alternative for the genetic testing of patients with clinically diagnosed FH.© 2009 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.

Keywords: Familial hypercholesterolemia; Low-density lipoprotein receptor; DNA micro array; Premature atherosclerosis; Genetic screening

Introduction

Familial hypercholesterolemia (FH, OMIM 143890) is aninherited disorder of lipoprotein metabolism, transmitted in anautosomal dominant manner and clinically characterized byelevated levels of total and low-density lipoprotein (LDL)cholesterol in the circulation, the presence of tendon xanthomasand premature atherosclerosis. The genetic basis of FH is a largearray of mutations in the LDL-receptor (LDLR) gene (OMIM606945) (http://www.ucl.ac.uk/ldlr), resulting in a lack of

⁎ Corresponding author. Fax: +31 20 566 9232.E-mail address: j.defesche@amc.uva.nl (J.C. Defesche).

0009-9120/$ - see front matter © 2009 The Canadian Society of Clinical Chemistsdoi:10.1016/j.clinbiochem.2009.01.017

functional receptors for LDL on the liver cell surface, givingrise to increased plasma LDL levels [1]. Similarly, mutations inthe gene coding for apolipoprotein B (APOB; OMIM 107730)also reduce LDL clearance, resulting in the disorder FamilialDefective Apolipoprotein B (FDB, OMIM 144010), which isclinically indistinguishable from FH [2,3]. Recently, a thirdputative FH locus located on chromosome 1, encoding proproteinconvertase subtilisin/kexin 9 (PCSK9; OMIM 607786), wasidentified [4]. Certain mutations in PCSK9 lead to hypercholes-terolemia, while other mutations can cause hypocholesterolemia,a situation comparable to that of APOB [5,6].

The clinical phenotype caused by mutations in LDLR,APOB or PCSK9 and characterized by elevated levels of plasma

. Published by Elsevier Inc. All rights reserved.

900 R. Alonso et al. / Clinical Biochemistry 42 (2009) 899–903

LDL-cholesterol is nowadays referred to as AutosomalDominant Hypercholesterolemia (ADH, OMIM 603776).

The prevalence of heterozygous ADH is about one in 500individuals in most Western countries and the diagnosis isusually made on the basis of clinical signs [1]. But since thephysical stigmata develop later in life, establishing the diagnosisin younger patients is often difficult. Genetic analysis, i.e. thedemonstration of a causative mutation provides the onlyunequivocal diagnosis. Maximum health benefit can beobtained in ADH if patients are identified at young age andtreatment is initiated early as possible [7]. For identification ofpatients with ADH systematic screening for causative mutationsin clinically diagnosed patients is therefore warranted.

The overall detection rate of ADH varies considerably fromless than 10% to 50% of the total number of patients estimatedto be present in the population. This depends on the clinicalcriteria used for the diagnosis, the genetic heterogeneity of thepopulation, the different analytical methods employed and theimplementation of active screening [8–11].

A survey that was carried out in Spain between 1999 and2002, showed that the molecular genetic basis of FH is veryheterogeneous and more than 116 different functional mutationswere described initially [12]. Based on that situation, adiagnostic tool in the form of a DNA-array was developedand has been used since 2004 by the Spanish Health System forimproving the genetic diagnosis of FH [8,13]. An initial studyshowed that the DNA-array was able to detect mutations in 50%patients with a clinical diagnosis of FH. This percentageincreased to 70% when cases negative on the DNA-array wereanalyzed for large gene rearrangements and new pointmutations by resequencing the complete LDLR [8]. To date,more than 220 different mutations have been identified in theSpanish population, and the DNA-array has been expanded todetect 191 different functional point mutations in LDLR and 4different mutations in APOB.

A screening genetic program for FH was started in Spain in2004 using a platform that we refer to as Lipochip® that includesthe use of the DNA-array, the detection of large gene rear-rangements (large insertions and deletions) and the completeresequencing of LDLR.

The aim of this study is to assess the effectiveness, sensitivityand specificity of this genetic diagnostic platform for theidentification of patients with inherited hypercholesterolemia inSpain.

Materials and methods

Patients

This study was conducted in 808 unrelated cases withclinical diagnosis of FH after the application of the Dutch LipidClinic Network criteria which are based on the presence of thetypical physical symptoms as xanthomata, xanthelasmata andarcus cornealis, family history, the occurrence of cardiovasculardisease and LDL-cholesterol levels [14,15]. Patients included inthis study had a score of 6 points or more, making a clinicaldiagnosis of FH probable or certain. In 11 Spanish regions that

participated in the screening program for FH blood samples ofpatients with FH were collected. All participants signed aspecific informed consent and the ethical committee of eachregion approved the study.

DNA analysis

Blood samples were obtained in the morning and were sentto a central laboratory for genetic testing (Progenika BiopharmaS.A in Derio, Vizcaya, Spain). DNA was obtained using acommercial kit (QiAmp Blood DNA Mini Kit, Qiagen,Germany) as previously described [12]. In the first step,samples were analyzed using the DNA-array able to detect 191different point mutations in LDLR and 4 different mutations inAPOB (Supplementary Table) following a protocol previouslydescribed [8].

Samples in which the DNA-array failed to detect mutationswere subsequently analyzed for large deletions or insertionsusing an adapted quantitative multiplex PCR methodology(QMFSP) [16]. Finally, samples negative for mutations in thefirst two procedures, were analyzed by automated sequencing ofthe LDLR on an MegaBace 1000 apparatus (General Electric,Fairfield, CT) as described before [17]. To test the specificity ofthe DNA-array system, 125 different DNA-array positivesamples were randomly selected and sequenced automaticallyusing the procedure described above. Sensitivity of the DNA-array is determined by the number of mutations detected bysequencing LDLR in the samples in which the DNA-array failedto detect mutations.

In order to evaluate reproducibility of the genetic diagnosticplatform a blind analysis of FH samples from Spain and TheNetherlands was performed. Fifty three DNA samples of DutchFH patients with different LDLR point mutations were analyzedwith the DNA-array tool. On the other hand, 43 samples fromSpanish FH patients positive for large deletions and insertions,were analyzed in the Dutch laboratory by the multiplex ligation-dependent probe amplification (MLPA) technique [18].

Results

Efficacy of DNA-array and Lipochip® platform

In 808 clinically diagnosed FH patients a mutation wasdetected in 537 cases (66%) either in LDLR (521 cases) or inAPOB (16 cases) (Fig. 1) after application of the completegenetic diagnostic procedure. The DNA-array was able to detectmutations in 419 cases (52% of the total cases and 78% of all themutation carriers; 403 mutations in LDLR and 16 in APOB). Alarge deletion or insertion was found to be the cause of FH in 77of the remaining 389 negative cases (10% of all patientsanalyzed and 15% of all mutation carriers). Of the remaining312 patients the complete LDLR was sequenced and only 41patients were found to carry one of 18 functional mutationsnewly detected by resequencing (5% of all patients analyzedand 8% of all mutation carriers).

The Lipochip® genetic diagnostic platform was able to detect120 different point mutations and 9 different large deletions or

Fig. 1. Lay out of the Lipochip® platform and results obtained.

901R. Alonso et al. / Clinical Biochemistry 42 (2009) 899–903

insertions were newly characterized. The seven most prevalentmutations (point mutations and large rearrangements) representalmost 43% of this series and are shown in Table 1.

The prevalence of FDBwas 3% in the mutation positive casesand 2% in the total group of clinically diagnosed FH patients. InAPOB, only the R3500Q-mutation was found in this series.

The average time to complete the Lipochip® platform was68 days (range 10–93 days). The DNA-array took an average of15 days including the DNA extraction and analysis. The largerearrangement detection added another 7 days to the analysistime and the sequencing of LDLR another 45 days. This entailshowever that the majority of mutations is detected within 15 to22 days after start of the analysis.

Blind analysis of Dutch and Spanish samples

DNA samples of 53 Dutch FH patients with characterizedmutations known to be present on the DNA-array wereanalyzed. The DNA-array correctly detected mutations in 51cases. However, one false positive and one false negative resultwere observed. The false negative case was the failure to detectthe C646Y mutation in one sample, which was detected by thesequencing step. The false positive case was the detection of therare S174N mutation in a sample that in fact contained theN804K (c.2475 CNG) mutation.

A total of 43 different samples, containing several insertionsand deletions in LDLR characterized by an adapted QMFSPassay, were also analyzed by MLPA. The mutations detected by

Table 1Most frequent LDLR mutations in the Spanish population screening program for Fa

Mutation n (%)

313+1GNC 71 (13.9)Deletion of exons 17 and 18 40 (7.8)V408M 34 (6.6)Deletion of exon 4 to 6 26 (5.1)Q12X 20 (3.9)Q427X 20 (3.9)W-18X 18 (3.5)

both techniques were identical in 42 samples, while the QMFSPassay detected a deletion between exon 12 to exon 15 in onesample where MLPA-analysis resulted in a deletion betweenexon 13 and 15. Resequencing of LDLR in 125 samples thattested positive for a mutation on the DNA-array yieldedidentical results.

Specificity and sensitivity

The sensitivity of the DNA-micro array technique was99.8%: 441 correct negative calls of a total of 442 real negativecalls. The specificity was 99.5%: 1 false positive was found in atotal of 178 true positive calls.

Discussion

A definitive and unequivocal diagnosis of FH is possibleonce a functional LDLRmutation has been identified. A definitediagnosis of FH has important clinical implications, becausepreventive measures and lipid-lowering therapy should beinitiated early in life to prevent the development of cardiovas-cular disease in this population. The information about specificmolecular defects that causes FH is also relevant because someLDLR gene mutations are associated with a high cardiovascularrisk and several LDLR defects have been associated with theresponse to statin therapy [7,10,19,20].

During the last decades genetic diagnosis of FH has beenestablished by a number of molecular biology techniques.Single-strand conformation polymorphism (SSCP) analysis hasa sensitivity between 75 and 85%, indicating that mutationshave been missed due to technical reasons in 15 to 25% ofpatients with clinical diagnosis of FH [21]. Recently, resequen-cing of the whole gene has been assessed as the most accuratetechnique for the detection of point mutations causing FH [22].

This study confirms that the Lipochip® genetic diagnosticplatform is a useful tool for the genetic diagnosis of familialhypercholesterolemia. The DNA-array was able to detect 78%of all mutation carriers and 52% of the total cases analyzed withhigh sensitivity and specificity in a shorter period of time and atlower costs compared to DNA sequencing. In 15% of patients acomplementary method was required to detect LDLR generearrangements (10%) and point mutations that were not inclu-ded in the DNA-array (5%).

Discrete sensitivity, total costs and the duration to completethe whole gene analysis have been the most importantarguments against resequencing to achieve a general screening

milial Hypercholesterolemia.

Location Nucleotide change

Intron 3 g to c at c.313+1Exons 17 and 18 2 kb deletion of cytoplasmic domainExon 9 G to A at c.1285Exon 4 to 6 5 kb deletion in ligand binding domainExon 2 C to T at c.97Exon 9 C to T at c.1342Exon 1 G to A at c.12

902 R. Alonso et al. / Clinical Biochemistry 42 (2009) 899–903

program for the detection of FH in large and genetic hetero-geneous populations. However, the development of new potentscreening tools such as DNA-arrays, the improvement oftraditional molecular biology techniques and the use of highthroughput analytical systems and ICT technology, has broughtgeneral genetic screening in reach. Lipochip®, the first world-wide commercial platform for the diagnosis of FH is a goodexample of this evolution. Based on micro array technology,Lipochip® includes three different complementing molecularassays that allow an accurate and reliable genetic diagnosis ofFH. Firstly, a DNA-array now comprising 195 different defectsin LDLR and APOB detects the majority of FH-causingmutations in Spain. Secondly, an adapted QMFSP assay detectslarge gene rearrangements and finally, the complete resequen-cing of LDLR yields mutations not included on the DNA-array.Since the sample preparation for analysis in the DNA-array isbased on multiplex PCR, detection primers for newlycharacterized mutations can be added without having to reca-librate or revalidate the micro array.

Since Lipochip® was launched during the second quarter of2004, more than 750 samples have been already analyzed.Positive detection rate was 65%, although this percentage mayrise to 74% when considering only patients with definitiveclinical diagnosis (data not shown). These detection rates arecomparable or higher than those observed in other studies[8–11]. In most of these studies, different clinical criteria for theselection of patients were used, making comparison of resultsdifficult, although the application of improved clinical criteriafor the diagnosis of FH, published during the course of thepresent study, may facilitate such a comparison [23–25].

The occurrence of one false negative call in all samplesanalyzed can be explained by the intrinsic design of the DNA-array and the analysis software. When the value of a signal isintermediate between a negative and a positive signal, theanalysis software scores this as a negative signal. So, in cases ofintermediate signal strength the DNA-array preferentiallyproduces false-negative rather than false positive results,because a false-negative sample will be identified by theresequencing step, which is not the case for a false-positivesample.

The failure to detect the N804K mutation was due to the factbase pair change in the DNA sample containing this mutationwas c.2475CNG, while the detection oligonucleotides designedto detect N804K contained the base pair change c.2475CNA.Both these changes in codon 804, AACNAAG in the patient'sDNA sample and AACNAAA in the detection oligonucleo-tides, lead to the amino acid change of asparagine to lysine(N to K). The mistyping of the N804K mutation by the S174Nmutation was caused by a high similarity in the sequences of thedetection oligonucleotides for these variants on the array andthis required redesign of the mutation- and wild-type-specificoligonucleotides. In spite of this, the DNA-array method stillreached a sensitivity of 99.5%, much higher than the onesdemanded for in vitro diagnostic assays in a recent Europeancommission decision (2002/364/CE).

In addition to our validation studies, the diagnostic platformwas also validated by a separate laboratory by re-analysis of 100

randomly selected samples that were tested positive as well asnegative. In 750 different samples analyzed with the commer-cial version of Lipochip®, disagreement has not been found sofar. The results obtained demonstrate that the Lipochip®

platform is an appropriate tool for the genetic diagnosis ofFH. Specificity and sensibility are favorable compared tospecifications described for other molecular screening techni-ques used for the diagnosis of other human conditions [26–28].

When employing the Lipochip® diagnostic platform insubjects with the clinical diagnosis of FH, other considerationsare also of importance. Firstly the throughput time, i.e. thewaiting time for screenees to obtain the test results, is with anaverage of 67 days acceptable, although most of the positiveresults can be reported within 15 to 22 days. Secondly, withaverage total screening costs of $ 350, cost-effectiveness isfavorable compared to other, existing screening programs, alsoin view of the health benefits obtained without causingpsychological or social damage [26–29]. After geneticidentification, most newly identified patients start with statintreatment, gaining on average 3.3 years of life [30].

In conclusion, all aspects surrounding large scale geneticcascade screening for FH are becoming so favorable thatintegration of such a preventive program in the national healthsystem should be considered.

Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.clinbiochem.2009.01.017.

References

[1] Goldstein JL, Hobbs HH, Brown MS. Familial hypercholesterolemia. In:Scriver CR, Beaudet AL, Sly WS, Valle D, editors. The metabolic and mole-cular basis of inherited disease.NewYork:McGraw-Hill; 2001. p. 2863–913.

[2] Boren J, Ekstrom U, Agren B, Nilsson-Ehle P, Innerarity TL. Themolecular mechanism for the genetic disorder familial defective apolipo-protein B100. J Biol Chem 2001;276:9214–8.

[3] Defesche JC, Pricker KL, Hayden MR, Van der Ende BE, Kastelein JJ.Familial defective apolipoprotein B100 is clinically indistinguishable fromfamilial hypercholesterolemia. Arch Intern Med 1993;153:2349–56.

[4] Abifadel M, Varret M, Rabes JP, et al. Mutations in PCSK9 causeautosomal dominant hypercholesterolemia. Nat Genet 2003;34:154–6.

[5] Kotowski IK, Pertsemlidis A, Luke A, et al. A spectrum of PCSK9 allelescontributes to plasma levels of low-density lipoprotein cholesterol. Am JHum Genet 2006;78:410–22.

[6] Fouchier SW, Sankatsing RR, Peter J, et al. High frequency of APOB genemutations causing Familial Hypobetalipoproteinaemia in patients of Dutchand Spanish descent. J Med Genet 2005;42:23–6.

[7] Koeijvoets KC, Rodenburg J, Hutten BA, Wiegman A, Kastelein JJP,Sijbrands EJ. Low-density lipoprotein receptor genotype and response topravastatin in children with familial hypercholesterolemia: substudy of anintima-media thickness trial. Circulation 2005;112:3168–73.

[8] Tejedor D, Castillo S, Mozas P, et al, Spanish FH Group. Reliable low-density DNA array based on allele-specific probes for detection of 118mutations causing familial hypercholesterolemia. Clin Chem 2005;51:1137–44.

[9] Damgaard D, Larsen ML, Nissen PH, et al. The relationship of moleculargenetic to clinical diagnosis of familial hypercholesterolemia in a Danishpopulation. Atherosclerosis 2005;180:155–60.

[10] Humphries SE, Cranston T, Allen M, et al. Mutational analysis in UKpatients with a clinical diagnosis of familial hypercholesterolemia:

903R. Alonso et al. / Clinical Biochemistry 42 (2009) 899–903

relationship with plasma lipid traits, heart disease risk and utility in relativetracing. J Mol Med 2006;84:203–14.

[11] Van Aalst-Cohen E, Jansen A, Tanck M, et al. Diagnosing familialhypercholesterolemia: the relevance of genetic testing. Eur Heart J 2006;27:2240–6.

[12] Mozas P, Castillo S, Tejedor D, et al. Molecular characterization of familialhypercholesterolemia in Spain: identification of 39 novel and 77 recurrentmutations in LDLr. Hum Mutat 2004;24:187.

[13] Tejedor D, Castillo S, Mozas P, et al. Comparison of DNA array platformvs DNA sequencing as genetic diagnosis tools for familial hypercholester-olemia. Clin Chem 2006;52:1971–2.

[14] Mata P, Alonso R, Castillo S, Pocovi M.MEDPED and the Spanish FamilialHypercholesterolemia Foundation. Atherosclerosis, Suppl 2002;2:9–11.

[15] Defesche J. Familial hypercholesterolemia. In: Betteridge DJ, editor.Lipids and Vascular Disease. London, Martin Dunitz Ltd; 2000. p. 65–76.

[16] Heath KE, Day IN, Humphries SE. Universal primer quantitativefluorescent multiplex (UPQFM) PCR: a method to detect major andminor rearrangements of the low-density lipoprotein receptor gene. J MedGenet 2000;37:272–80.

[17] Ju J, Kheterpal I, Scherer JR, et al. Design and synthesis of fluorescenceenergy transfer dye-labeled primers and their application for DNAsequencing and analysis. Anal Biochem 1995;231:131–40.

[18] Schouten JP, McElgunn CJ, Waaijer R, Zwijnenburg D, Diepvens F, PalsG. Relative quantification of 40 nucleic acid sequences by multiplexligation-dependent probe amplification. Nucleic Acids Res 2002;30:e57.

[19] Umans-Eckenhausen MAW, Sijbrands EJG, Kastelein JJP, Defesche JC.Low-density lipoprotein receptor gene mutations and cardiovascular risk in alarge genetic cascade screening population. Circulation 2002;106:3031–6.

[20] Heath KE, Gudnason V, Humphries SE, Seed M. The type of mutation inthe low density lipoprotein receptor gene influences the cholesterol-lowering response of the HMG-CoA reductase inhibitor simvastatin inpatients with heterozygous familial hypercholesterolaemia. Atherosclerosis1999;143:41–54.

[21] Jensen H, Jensen LG, Hansen PS, Faergeman O, Gregerson N. Highsensitivity of the single-strand conformation polymorphism method for

detecting sequence variations in the low-density lipoprotein receptor genevalidated by DNA sequencing. Clin Chem 1996;42:1140–6.

[22] Liguori R, Argiriou A, SimoneVD. A rapid method for detecting mutationsof the human LDL receptor gene by complete cDNA sequencing. Mol CellProbes 2003;17:15–20.

[23] Leren TP, Finborud TH, Manshaus TE, Ose L, Berge KE. Diagnosis offamilial hypercholesterolemia in general practice using clinical diagnosticcriteria or genetic testing as part of cascade genetic screening. CommunityGenet 2008;11:26–35.

[24] Civeira F, Ros E, Jarauta E, Plana N, Zambon D, Puzo J, Martinez deEsteban JP, Ferrando J, Zabala S, Almagro F, Gimeno JA, Masana L,Pocovi M. Comparison of genetic versus clinical diagnosis in familialhypercholesterolemia. Am J Cardiol 2008;102:1187–93.

[25] Civeira F, Jarauta E, Cenarro A, García-Otín AL, Tejedor D, Zambón D,Mallen M, Ros E, Pocoví M. Frequency of low-density lipoproteinreceptor gene mutations in patients with a clinical diagnosis of familialcombined hyperlipidemia in a clinical setting. J Am Coll Cardiol 2008;52:1546–53.

[26] Ahlquist DA, Shuber AP. Stool screening for colorectal cancer: evolutionfrom occult blood to molecular markers. Clin Chim Acta 2002;315:157–68.

[27] García-Díaz E, Castro-Fernández M, Romero-Gómez M, Vargas L,Romero J. The effectiveness of (IgG-ELISA) serology as an alternativediagnostic method for detecting Helicobacter pylori infection in patientswith gastrointestinal bleeding due to gastro-duodenal ulcer. Rev EspEnferm Dig 2002;94:725–36.

[28] Wong KY, Chuan YC, Aggarwal A, Tham L, Kong WM, Tan P.Identifying patterns of DNA for tumor diagnosis using capillaryelectrophoresis-amplified fragment length polymorphism (CE-AFLP)screening. J Bioinform Comput Biol 2004;2:569–87.

[29] Leren TP. Cascade genetic screening for familial hypercholesterolemia.Clin Genet 2004;66:483–7.

[30] Wonderling D, Umans-Eckenhausen MA, Marks D, Defesche JC,Kastelein JJ, Thorogood M. Cost-effectiveness analysis of the geneticscreening program for familial hypercholesterolemia in The Netherlands.Semin Vasc Med 2004;4:97–104.