Familial hypercholesterolemia (FH) is caused by defects in the low-density lipoprotein receptor (LDLR) gene, which give rise to a well-characterized clinical phenotype [1]. The prevalence of heterozygous FH is approximately 1 in 500 among the general population in most countries worldwide. The prevalence of homozygotes is generally about 1 in 1 million [1]. Despite its strong genetic background, FH shows great variability in phenotypic expression in terms of lipid profile, frequency of xanthomas, and onset and severity of cardiovascular disease (CVD) [1]. It has been shown that at least 50% of males and 20% of females with FH who fail to receive effective treatment suffer from coronary events by the age of 50 years [2].

The plasma levels of low-density lipoprotein cholesterol (LDL-C) in FH heterozygotes are lower and more dependent on other genetic and environmental factors than those in FH homozygotes. Although the nature of the molecular defect has some impact on the severity of hypercholesterolemia, FH heterozygotes with the same LDLR mutation can have widely different plasma levels of LDL-C. In all patients with FH caused by LDLR defects, phenotype severity seems to depend more on environmental factors than on the type of mutation [3]. However the clinical diagnosis is not always evident, especially in young patients without physical stigmata, and molecular diagnostic techniques would be helpful in these cases [4].

The LDLR gene locus is located on chromosome 19p13.1-13.3 and spans 45 kb, with 18 exons and 17 intervening introns, encoding a mature protein of 839 amino acids. Mutations can occur in the promoter, or in the introns or exons. The majority of variants fall within the ligand-binding (40%) or epidermal growth factor precursor-like (47%) domains, with the highest frequency of mutations reported in exon 4 (20%) [5]. This high frequency can be explained by a selection bias; i.e., individuals with functional mutations in this region may be overrepresented in the lipid clinic populations surveyed for FH screening [6]. However, when the number of mutations is normalized to base pairs, the mutation rate is similar in most exons, except for the promoter region and exons 16-18 [5].

A total of 1,066 individual validated LDLR variants from FH patients are now listed on the University College London (UCL) database [5]. The LDLR FH database http://www.ucl.ac.uk/ldl underlines the vast molecular heterogeneity of this disorder [5]. The current study therefore aimed to describe the molecular spectrum of FH in Malaysian subjects (by screening the most likely affected exons), and to identify and correlate the LDLR gene mutations with the clinical manifestations of FH.

Screening of all clinically diagnosed cases with FH identified 29 LDLR variants in 117 of 154 subjects included in the study (76.0%). Six of the variants were detected in the control group, with a non-significant difference in frequency between FH subjects and controls (c.190+58C>T, c.1060+7 T>C, c.1060+10 G>C, p.Pro539Pro, c.1705+56C>T and p.Arg744Arg). All of the variants were reported to occur in a heterozygous state, except for c.2232A>G; (p.Arg744Arg) in exon 15 and c.1060+7 T>C in intron 7, which were homozygous.

The most frequent variant was the c.1060+7 T>C, with a frequency of 11.7% (18/154), followed by the c.1411A>G substitution in exon 10, resulting in p.Arg471Gly variant, which was detected at a frequency of 11.0% (17/154).

Eight variants were previously unreported while the other 21 variants have been previously described in public databases http://www.hgmd.cf.ac.uk; http://www.ucl.ac.uk/ldlr, and http://www.ncbi.nlm.nih.gov/Genbank (Table 2).

There was considerable diversity in the types of variants, as eight LDLR gene mutations were reported (1 frameshift, 1 large rearrangement, 5 missense and 1 splice site).

One novel frameshift mutation was reported, p.Trp666ProfsX45 (Figure 1), resulting from the deletion of 17 nucleotides (c.1996_2012del 17). A novel silent variant was reported in one FH subject (p.Asp100Asp). Two novel missense variants were also reported, p.Asp139His and p.Arg471Gly.

A duplication of exon 12 was found to co-segregate with hypercholesterolemia in two family members. A typical result of duplication is shown in the electrophoregram in Figure 2A and 2B, which shows the peak height for each of the test and control probes. The 37 labeled peaks included probes for the LDLR gene promoter, its 18 exons, 16 reference probes, and 2 probes for the upstream and downstream genes in chromosome 19. The peak height that represents the exon 12 probe was at 8,084 relative fluorescence units (RFU), which was about 1.7 times that of the control sample (4,768 RFU), suggesting duplication. This result was confirmed by second multiplex ligation-dependent probe amplification (MLPA).

The c.190+4A>T splice site mutation was reported in one FH subject.

Three previously reported silent variants, p.Ile398 Ile, p.Asn591Asn, and p.Val653Val, were found in the FH subjects but not in the control group, while one silent variant (p.Pro539Pro) was reported in both cases and controls. One novel variant in intron 8 (c.1186+41T>A) and two previously unreported variants in intron 11 (c.1705+117T>G and c.1705+112C>G) were also noticed.

Molecular genetic diagnosis is currently the most specific method for identifying patients with FH. Although numerous LDLR gene mutations have been identified in FH patients, genetic data for the Malaysian population are rare [4, 8]. The present study cohort of 154 patients with clinical features of FH was relatively young (with a mean age of 44.6 years), and already had a high prevalence of CVD (68.2%). The average LDL-C level was classified as high, according to National Cholesterol Education Program (NCEP ATP III) classification [9]. This, together with the high prevalence of CVD, were consistent with the recognized association of elevated LDL-C levels with a high risk of atherosclerotic disease [10].

The presence of gene sequence variants in the promoter region and in exons 2-15 of the LDLR gene was investigated, and LDLR variants were identified in 117 out of 154 patients (76.0%). These comprised 8 previously unreported and 21 previously reported LDLR sequence variations. Eight pathogenic mutations were reported, including one intronic (c.190+4A>T).

One subset was relatively common, such as a variant in exon 10 (p.Arg471Gly), which was found in 17 patients. The donor splice site mutation (c.190+4A>T) was the commonest among Filipino FH [11], while L547V mutation was predominant among Japanese FH [12]. The frequent detection of a deleterious mutation may be the result of consanguinity, recurrent mutational events, genetic drift, or multiple introduction of the mutation into a population [13].

The overall mutation rate of 42.2% (65/154), which was higher than that reported in previous studies of Malaysian FH (26%) [4] or Filipino FH (20%) patients [11], but lower than that reported in European populations (52%) [14].

Only the p.Cys255Ser, p.Asn591Asn, and p.Val653Val variants detected in the current study have previously been reported among Malaysian FH subjects [8]; the remainders were newly identified in this study population. This may be attributed to the high sensitivity of DHPLC for detecting gene sequence variants, and the restricted inclusion criteria.

Five missense mutations that resulted in altered amino acids (p.GluA101Lys, p.Asp139His, p.Glu201Lys, p.Cys255Ser and p.Asp700Glu) were predicted to exert pathogenic effects based on the in silico analyses.

p.Glu101Lys was predicted to be pathogenic by Alamut™software with change from medium size and acidic (Glutamate) to large size and basic (Lysine). This residue occurs in the second disulfide-rich repeat in the binding domain of the receptor protein, and affects processing and intracellular transport of the newly synthesized protein. It was suggested that the normal formation of disulfide bonds in the second repeat may be impeded in the mutant protein [15].

p.Asp700Glu was predicted to be pathogenic by Alamut™software with change in the 3 D structure of the protein. It was reported among Spanish FH subjects [16].

p.Asp139His is a missense mutation, reported for the first time in this study, located in the ligand binding domain and was predicted to be ''probably damaging'' by PolyPhen due defect in the ligand binding site.

p.Glu201Lys and p.Cys255Ser are previously reported mutations among Russian patients [17] and Malaysian patients [8], respectively and were predicted to be ''probably damaging'' by PolyPhen.

It has been reported that a significant proportion (approximately 18.9%) of the gene sequence variants observed in patients with FH have no effect on the protein coding sequence [18]. These variations are often located in the 5' splice-donor site or the 3' splice-acceptor site of an intron, and are predicted to result in either exon skipping or retention of an intron in the mRNA, although this has not always been confirmed experimentally. One such variant, c.190+4A>T, has been reported in this study as well as in the Netherlands [19] and the Philippines [11]. We reported the possibility that it may cause splicing abnormalities through an in-silico analysis. However, the true effect should be experimentally clarified using splicing analysis.

For c.1060+10G>C variant in intron 7, it was reported in French and Greek FH patients [20, 21]. In the present study this variant was reported among FH subjects and control groups with a non significant difference in the occurrence between both groups which may indicate that this variant may be a polymorphism.

c.1359-30C>T is detected in unrelated FH individuals in Denmark [22]. Webb et al. [23] failed to detect any effect of this polymorphism on plasma lipid concentrations in their population. p.Ile398 Ile, p.Pro539Pro, and p.Cys27Cys are previously reported among European [24], Russian [25] and Chinese FH patients [26], respectively, with unknown clinical associations with the disease.

p.Trp666ProfsX45 was identified in four unrelated patients within this study, which could be attributed to a founder effect. The extent or nature of the deletion in the DNA was not clear, and the amplified mutant fragment was therefore cloned and sequenced. The predicted translation products of the alleles carrying this mutation resulted in the creation of premature stop codon downstream of the deletion.

We identified a large genomic rearrangement (c.1706_1845dup;p.Asp616IlefsX96), a duplication of exon 12, in the LDLR gene which have not previously been detected among Malaysian FH patients [8, 27]. It is interesting to note that while large deletion generally represents 85% of large rearrangements in the LDLR gene [5], our finding showed the predominance of a large duplication and the absence of large deletions. Large duplications seem to be much more prevalent than large deletions among the Malaysian population, while large deletions are more common than large duplications in North European Caucasians [28]. Therefore, the prevalence of large duplications/rearrangements in FH patients indicates that MLPA should be included in the diagnostic service for dominant hypercholesterolemia.

The duplicated exon 12 (c. c.1706_1845dup; p.Asp616IlefsX96) is predicted to cause a frameshift protein, with premature termination 95 positions downstream. Due to its nucleotide number (1706-1845 bp) that were duplicated, either deletion or duplication of exon 12 will cause frameshifted protein, (i.e. disruption of reading frame starting from the point of deletion or duplication). The mutation effect is different with that of previously reported deletion of exon 15 [29] which is a common yet mild cause of FH in Finland. Deletion of exon 15 caused only internal truncation of the protein without inducing premature termination, since the reading frame was maintained.

In silico predictions of the eight variants categorized as pathogenic may provide grounds for further experimental studies aimed at revealing mRNA abnormalities, as well as altered protein-protein interactions and loss of function.

mRNA expression studies could be performed for the c.1706_1845dup; p.Asp616IlefsX96 and c.1996_2012del;p.Trp666ProfsX45, as they were also predicted to be targets for nonsense-mediated decay.

There were significant differences in baseline LDL-C levels between the two patients groups; subjects with pathogenic mutations had a mean baseline LDL-C level of 5.2 mmol/l, compared with level of 4.9 mmol/l in patients of the non pathogenic group.

Also those with pathogenic mutations show a more severe phenotype (higher frequencies of CVD, TX, and family history of hyperlipidemia) than those with non pathogenic variants which may increase the likelihood of their pathogenicity. The exact decision about the functional implications should be done by in vitro functional study.

The presentation of TX as a diagnostic sign of FH was interesting in the current study population, as only 40.9% of FH subjects showed TX. This emphasizes the importance of DNA analysis, because no definite diagnosis of FH can be made without the identification of TX or a defined mutation.

Regarding those patients who showed definite clinical manifestations of FH but no identified LDLR sequence variations, the following possibilities exist: (1) they may have defects that are linked to the LDLR gene variant within deep intronic regions, affecting expression or splicing of the gene; (2) they may have exon defects that have not been screened; (3) the defect may lie in another gene involved in LDL-C metabolism, such as the apolipoprotein gene or the proprotein convertase subtilisin/kexin type 9 (PCSK 9) gene; or (4) it is possible that polygenic factors interacting with environmental factors may lead to a clinical diagnosis mimicking the FH phenotype [30]. It is also likely that some patients with no identified mutation do not have true monogenic autosomal dominant hypercholesterolemia. The clinical diagnostic criteria for FH are not precise, and no family studies have been carried out in patients with any detected mutations to clarify this.

There were several limitations of the present study. FH patients are generally seen by cardiologists when cardiovascular complications appear, but the disease's etiology is not always investigated. This could explain the low number of FH patients recruited in the study sample. This cohort was therefore representative of FH patients who are referred to specialists because of a severe lipid phenotype, but not of those who remain undiagnosed, who may have milder phenotypes [31]. The DHPLC method used here has been reported to show a sensitivity of 96% [32], and an additional 4% of LDLR mutations may thus have remained undetected. Furthermore, several exons with a reportedly low mutation detection rate were not examined; the detection rate may have been increased by a further 3-4% (representing the percentage of mutations in exon 1, 16,17 and 18 [5] if all exons had been included. Additionally, only 30-50 base pairs of the LDLR introns were examined by the primers used here, and mutations may exist in other regions, although few have been reported [5].

In summary, 29 LDLR gene sequence variants were detected in 76% of the analyzed patients with a clinical diagnosis of FH. Most of these variants were reported for the first time among Malaysians, with eight variants discovered in this report. Patients with pathogenic mutations had a higher rate of CVD and higher TC and LDL-C levels than those with non-pathogenic variants. These findings support the usefulness of genetic testing in FH, beyond its ability to provide an unequivocal diagnosis of the condition.

The target population included dyslipidemic patients attending the Cardiology Clinic in HUSM. One hundred and fifty-four subjects gave written, informed consent to participate in this study. The included subjects were patients who fulfilled the Simon Broome Familial Hypercholesterolemia Register diagnostic criteria for FH [2]. Patients with suspected causes of secondary hyperlipidemia were excluded from the study [33]. Demographic data, medical history, lipid-lowering therapy, physical examination for detection of the presence of TXs, xanthelasma, and arcus cornealis were recorded. Pre-treatment lipid profiles were obtained from the patients' records and family history of PCVD was obtained from all patients using a standardized form.

Eighty control subjects were recruited to detect the presence of nucleotide substitutions that could be regarded as single nucleotide polymorphisms. The control subjects were randomly chosen healthy volunteers from the staff of Universiti Sains Malaysia, who had no previous history or family history of hyperlipidemia or PCVD, no history of secondary causes of hyperlipidemia and no any clinical signs of hyperlipidemia. Ethical approval for the study was obtained from Universiti Sains Malaysia's Research and Ethics Committee.