OUP user menu

Diagnosing familial hypercholesterolaemia: the relevance of genetic testing

Emily S. van Aalst-Cohen , Angelique C.M. Jansen , Michael W.T. Tanck , Joep C. Defesche , Mieke D. Trip , Peter J. Lansberg , Anton F.H. Stalenhoef , John J.P. Kastelein
DOI: http://dx.doi.org/10.1093/eurheartj/ehl113 2240-2246 First published online: 6 July 2006

Abstract

Aims We assembled a cohort of patients with familial hypercholesterolaemia (FH) for both basic and clinical research. We used a set of established diagnostic criteria to define FH. Some put forward that a definite diagnosis of FH is made when a mutation in the LDL-receptor (LDLR) gene is identified. We therefore set out to determine in these patients whether patients with a DNA diagnosis would differ significantly from those diagnosed clinically.

Methods and results We randomly selected 4000 hypercholesterolaemic patients from the Dutch Lipid Clinic network database. Phenotypical data were acquired by reviewing medical records. After review of medical records, 2400 patients could be defined as having FH. An LDLR mutation was identified in 52.3% of these patients. Patients with and without an LDLR mutation demonstrated different clinical and laboratory characteristics. Low-density lipoprotein cholesterol was higher in patients with an LDLR mutation, whereas triglycerides were higher in patients without an LDLR mutation. The phenotypic differences between the groups remained even after stratification for the presence or absence of tendon xanthomas.

Conclusion Despite the use of stringent clinical criteria to define FH patients, two cohorts could be identified within our study population, namely those patients with and those without an LDLR mutation. Our findings suggest that among those without an LDLR mutation, patients with other causes of dyslipidaemia may be present. These observations underline the relevance of genetic testing in FH for clinical practice, for screening purposes, and for research involving these patients.

  • Familial hypercholesterolaemia
  • Genetic testing
  • Diagnostic methods
See page 2150 for the editorial comment on this article (doi:10.1093/eurheartj/ehl172)

Introduction

Heterozygous familial hypercholesterolaemia (FH) is a common (1:500) and inherited autosomal dominant disorder of lipoprotein metabolism. Clinically, FH is characterized by elevated serum levels of total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-C) and the presence of tendon xanthomas, xanthelasmata, and/or an arcus cornealis. In most patients, there is also excessive deposition of cholesterol in the arterial wall, leading to accelerated atherosclerosis and premature cardiovascular disease (CVD). Typically, approximately 45% of male and 20% of female patients suffer from coronary artery disease (CAD) by the age of 50.1

The clinical diagnosis of FH is based on personal and family history, physical examination, and laboratory findings. The underlying molecular defect of FH consists of mutations in the gene coding for the LDL-receptor protein, detection of which provides the only unequivocal diagnosis.2 Therefore, there is a need for accurate clinical diagnostic criteria as long as a genetic diagnosis has not been made. We assembled a large cohort of FH patients for both clinical and basic research. We used a set of established clinical diagnostic criteria to define FH.1,36 Subsequently, the clinical and biochemical features of this FH cohort, stratified by the presence of LDL-receptor (LDLR) mutations as well as the presence of tendon xanthomas were compared.

Methods

Study design and study population

The present investigation was a retrospective, multi-centre, cohort study. The study population has been described elsewhere.7 Briefly, lipid clinics in the Netherlands submit DNA samples from clinically suspected FH patients to a central laboratory for LDL-receptor (LDLR) mutation analysis. This laboratory, located at the Academic Medical Centre of the University of Amsterdam, manages a DNA-bank database, which currently contains over 9300 samples.8 We randomly selected 4000 hypercholesterolaemic patients with the aid of a computer programme (Microsoft Excel) from this database. These patients had been referred from 27 Lipid Clinics throughout the Netherlands.

Phenotypical data were acquired by reviewing these patients' medical records by a trained team of data collectors.9 The inclusion and exclusion criteria for participation in the study are outlined in Table 1. The FH diagnostic criteria were based on criteria used in the USA (the MedPed criteria6), the UK (the Simon Broome Register criteria10), and the Netherlands (the Dutch Lipid Clinic Network criteria5). Contrary to the criteria we applied in our study, these sets distinguish ‘possible’ from ‘definite’ FH patients. Specifically, the Dutch Lipid Clinic Network uses a numeric score: the higher the score, the more probable the diagnosis of FH. If a mutation has not been identified, the presence of tendon xanthomas is required for a definite diagnosis of FH in the MedPed and Simon Broome Register criteria. In the Dutch scoring system, the presence of xanthomas is not required but increases the score substantially.

View this table:
Table 1

Inclusion and exclusion criteria

Inclusion criteria
Males and females
Age 18 years and older
FH diagnostic criteria
Presence of a documented LDL-receptor mutation or
LDL-cholesterol level above the 95th percentile for sex and age
In combination with at least one of the following
 The presence of typical tendon xanthomas in the patient or in a first degree relative
 An LDL-cholesterol level above the 95th percentile for age and sex in a first degree relative
 Proven CAD in the patient or in a first degree relative under the age of 60
Exclusion criteria
Secondary causes of hypercholesterolemia such as renal, liver, or thyroid disease
Hypercholesterolemia due to other genetic defects, such as familial defective apolipoprotein B

Additional data were collected from the patients' medical records on classical risk factors for CVD, family history of CVD, medication use, physical examinations, laboratory parameters, and extensive information on CVD.7 All patients gave informed consent and the Ethics Institutional Review Board of each participating hospital approved the protocol.

Laboratory analysis

All laboratory parameters were measured in fasting blood samples. Lipid levels, as stated in the medical records, were determined after at least 6 weeks of withdrawal of any lipid-lowering medication.7 Mutations in the LDLR gene were assessed by methods described previously.11 In short, all patients were analysed by PCR for the 14 most common mutations found in the Netherlands.11 If no mutation was found, the promoter and all coding regions of the LDLR gene were analysed by denaturing gradient gel electrophoresis and aberrant bands were sequenced. In addition, long-range PCR and Southern blotting techniques were used to identify large deletions and insertions. However, since these techniques were time-consuming and not very discriminating, a different approach was used in a later stage: the promoter region and all exons (including exon-intron boundaries) of the LDLR gene were sequenced directly and major LDLR gene rearrangements were identified by the multiplex ligation-dependent probe amplification technique.

Furthermore, exons 26 and 29 as ligand-binding domains of the APOB gene were sequenced in all patients.

Sample size

The sample size of 2400 is appropriate to detect a minimal difference of 0.2 standard deviations (continuous variables, t-test) or 10% (categorical variables, χ2-test) between groups with 90% power and a two-sided alpha of 0.001.

Statistical methods

Comparison of subgroups for continuous and categorical variables was performed using linear and logistic regression, respectively, with or without adjustment for possible confounders: age, gender, smoking, alcohol use, concomitant medication use, and body mass index (BMI). For all models, the generalized estimating equations method was used to correct for possible correlations between patients due to family relations. Total outpatient clinical follow-up was calculated for each individual as the difference between age at end of observation and age at first visit. For statistical testing, total clinical follow-up was rank transformed, whereas triglycerides and glucoses were loge-transformed. To correct for multiple testing, two-sided P-values <0.001 were regarded as statistically significant. All statistical analyses were performed using the SAS system (SAS 9.1; Cary, North Carolina). For the purpose of this study, a representative random sample of 199 patients was sequenced from the patients in whom an LDLR mutation was not yet found (denoted as the LDL-R minus group).

Results

After thorough review of the 4000 medical records and application of the FH diagnostic criteria, 2400 patients were defined as having FH (Figure 1). An LDLR mutation was identified in 52.3% of these patients (denoted as the LDL-R plus group). In the remaining 47.8% (denoted as the LDL-R minus group), an LDLR mutation has not yet been found, but at least the 14 most prevalent Dutch LDLR gene mutations can be excluded11. Further sequencing of the coding regions is currently being performed. In the random sample of 199 patients from the LDL-R minus group, an LDLR mutation was found by sequencing in 40 (20%) of these patients (data not shown). Individuals with inherited hypercholesterolaemia due to familial defective apolipoprotein B (FDB), caused by mutations in the apolipoprotein B gene, were excluded from this study.

Figure 1

Study subjects profile.

The two cohorts demonstrated significantly different clinical and laboratory profiles (Table 2). LDL-C was higher in LDL-R plus patients, whereas triglycerides were higher in LDL-R minus patients. The LDL-R minus patients were older, more often males and had, on average, higher values of BMI, blood pressure, and glycaemia, in addition to a higher prevalence of smoking and a positive family history for premature CVD. The differences in lipid levels (including HDL-cholesterol), BMI and diastolic blood pressure remained significant after adjustment for age and gender. Differences in lipid levels remained significant also after adjustment for smoking, alcohol use, concomitant beta-blocker use, and BMI.

View this table:
Table 2

Clinical characteristics of FH patients with and without a known LDL-receptor mutation

LDL-R plus n=1255LDL-R minus n=1145P-value
Demographics
 Male gender (%)45.8 (575/680)52.8 (605/540)<0.001
 Age at first visit (year)42.1 (±12.6)47.6 (±12.2)<0.001
 Age at end of observation48.0 (±13.4)51.7 (±12.7)<0.001
 Total outpatient clinic follow-up (years)4.3 [2.0–8.5]3.1 [1.2–6.1]<0.001
Risk factors
 Smoking, ever (%)68.7 (787/359)79.5 (811/209)<0.001
 Hypertension7.8 (97/1146)11.7 (133/1000)0.001
 Diabetes mellitus (%)5.0 (63/1192)6.6 (75/1070)0.114
 First-degree family history of premature CVD (%)56.4 (596/460)65.5 (664/350)<0.001
Physical examination
 BMI (kg/m2)24.7 (±3.4)25.6 (±3.6)<0.001
 Systolic blood pressure (mmHg)133 (±19)137 (±20)<0.001
 Diastolic blood pressure (mmHg)81 (±10)83 (±10)<0.001
 Tendon xanthomas (%)41.8 (489/682)40.2 (419/624)0.462
Laboratory parameters
 TC (mmol/L)10.25 (±2.13)8.80 (±1.54)<0.001
 LDL cholesterol (mmol/L)8.18 (±2.05)6.61 (±1.47)<0.001
 HDL cholesterol (mmol/L)1.19 (±0.35)1.23 (±0.36)0.003
 Triglycerides (mmol/L)1.39 [0.98–2.03]1.71 [1.24–2.35]<0.001
 Glucose (mmol/L)4.9 [4.5–5.3]5.0 [4.6–5.5]<0.001
  • Values are given as mean levels (±SD), except where given as percentages (frequencies). Total outpatient clinic follow-up, triglycerides, and glucose are given as median [interquartile range]. LDL-R+/− indicates LDL receptor known/unknown.

Subsequently, both the LDL-R plus and the LDL-R minus groups were divided into two further groups, namely those characterized by the presence or absence of tendon xanthomas (Figure 1 and Table 3). LDL-R plus patients in whom tendon xanthomas are present are characterized by significantly increased mean TC and LDL-C when compared to LDL-R plus patients without tendon xanthomas (11.03 vs. 9.70 mmol/L and 8.94 vs. 7.66 mmol/L, respectively, both P<0.001) (3). Moreover, these patients are characterized by a higher prevalence of CAD (29.5 vs. 17.8% (data not shown). In the LDL-R minus group, patients with tendon xanthomas were also characterized by significantly increased TC and LDL-C levels (9.34 vs. 8.43 mmol/L and 7.20 vs. 6.25 mmol/L, respectively) (Table 3).

View this table:
Table 3

Clinical characteristics of FH patients with (LDL-R plus) and without (LDL-R minus) a documented LDL-receptor mutation; stratified to the presence or absence of tendon xanthomas

LDL-R plusLDL-R minus
Xanthomas present (n=489)Xanthomas absent (n=682)P-valueXanthomas present (n=419)Xanthomas absent (n=624)P-value
Demographics
 Male gender (%)44.6 (218/271)46.6 (318/364)0.51549.9 (209/210)53.2 (332/292)0.373
 Age at first visit (year)43.5 (±12.2)40.4 (±12.5)<0.00149.0 (±13.2)46.5 (±11.8)0.002
 Age at end of observation49.6 (±12.8)45.8 (±13.2)<0.00153.4 (±13.6)50.2 (±12.4)<0.001
 Total outpatient clinic follow-up (years)4.6 [2.5–8.5]3.9 [1.5–8.1]<0.0013.4 [1.3–6.7]2.7 [1.1–5.2]0.004
Risk factors
 Smoking, ever (%)69.8 (313/135)68.1 (425/199)0.58973.7 (264/94)81.7 (465/104)0.005
 Hypertension9.1 (44/441)6.7 (45/632)0.16611.3 (47/369)12.0 (74/541)0.689
 Diabetes mellitus (%)5.7 (28/461)4.0 (27/655)0.1625.7 (24/395)6.9 (43/581)0.498
 First-degree family history of premature CVD (%)56.2 (248/193)56.4 (321/248)0.95850.3 (184/182)74.0 (425/149)<0.001
Physical examination
 BMI (kg/m2)25.0 (±3.5)24.4 (±3.3)0.00225.6 (±3.8)25.6 (±3.5)0.918
 Systolic blood pressure (mmHg)134 (±19)133 (±19)0.527137 (±20)136 (±20)0.736
 Diastolic blood pressure (mmHg)81 (±11)80 (±10)0.12483 (±11)83 (±10)0.644
Laboratory parameters
 TC (mmol/L)11.03 (±2.15)9.70 (±1.95)<0.0019.34 (±1.75)8.43 (±1.27)<0.001
 LDL-C (mmol/L)8.94 (±2.07)7.66 (±1.87)<0.0017.20 (±1.75)6.25 (±1.19)<0.001
 HDL-C (mmol/L)1.18 (±0.34)1.20 (±0.36)0.4641.21 (±0.33)1.24 (±0.35)0.271
 Triglycerides (mmol/L)1.40 [1.01–2.15]1.37 [0.93–1.96]0.1141.68 [1.20–2.20]1.70 [1.28–2.39]0.237
 Glucose (mmol/L)4.90 [4.50–5.30]4.90 [4.50–5.30]0.1845.00 [4.50–5.40]5.00 [4.60–5.50]0.020
  • Values are given as mean levels (±SD), except where given as percentages (frequencies). Total outpatient clinic follow-up, triglycerides, and glucose are given as median [interquartile range].

When the LDL-R minus patients with tendon xanthomas were compared with LDL-R plus patients, total and LDL-cholesterol levels differed significantly (9.34 vs. 10.25 mmol/L and 7.20 vs. 8.19 mmol/L, respectively, both P<0.001), as well as median triglyceride levels (1.68 vs. 1.39 mmol/L; P<0.001).

Discussion

We assembled a cohort of 2400 FH patients for clinical and basic research. We used a set of established clinical diagnostic criteria to define FH.1,36 However, analysis of this cohort revealed significantly different clinical and laboratory profiles between those patients with and those without a known LDLR mutation. Moreover, further LDLR gene sequencing of a random sample of the LDL-R minus group revealed that only 20% had an LDLR mutation. Our findings suggest that in fact the LDL-R minus group comprises, partially, patients with other causes of dyslipidaemia.

FH vs. other causes of (familial) dyslipidaemia

Similar to earlier findings,12 patients in our study population who are clinically diagnosed as having FH but yet without a known mutation are characterized by lower TC, lower LDL-C, higher triglyceride and HDL-cholesterol levels, a significantly higher prevalence of hypertension, and higher glucose levels. The LDL-R minus patients possess clinical characteristics of other forms of dyslipidaemia, namely familial combined hyperlipidaemia, polygenic hyper-cholesterolaemia, familial dyslipidaemic hypertension, hyperapobetalipoproteinaemia and the dyslipidaemias associated with the metabolic syndrome.1315 FCH, the most common genetic dyslipidaemia, is a complex phenotype, which most likely comprises a group of related disorders.13 Accordingly, FCH is also known as ‘multiple-type hyperlipidaemia’. An elevated plasma apolipoprotein B (apoB) level is often used to distinguish between these disorders.1618 Unfortunately, apoB levels were not determined in our study population. Therefore, we consider the LDL-R minus group to represent a mixture of these lipid disorders, possessing similarities with ‘the’ FCH phenotype.

Both prominent causes of high LDL-C and early CVD in adults, patients with FCH tend to have lower LDL-C levels, lower HDL-cholesterol levels, and higher triglyceride levels than their FH counterparts.13,19,20 LDL-C levels found in the LDL-R minus group are indeed in the range generally measured in FCH patients. HDL-cholesterol levels found in the LDL-R minus group are higher than generally measured in FCH patients. The slightly elevated triglyceride levels observed in the LDL-R minus group are somewhat lower than in most FCH patients, in whom these values tend to be higher.16 However, a characteristic feature of FCH is its variability in lipid phenotype expression among patients and even in the individual patient over time, suggesting a strong influence of environmental factors.21

FCH is often associated with obesity, diabetes mellitus and hypertension.22 Higher BMI values and higher values of glycaemia were indeed measured in the LDL-R minus group, indicating a tendency towards obesity and hyperinsulinaemia.

Patients without a known LDLR mutation were older, more often males, had a higher prevalence of a positive family history for premature CVD, were more often smokers and suffered from CVD. This was due to the selection criteria used for the diagnosis of FH. If the patient did not have an LDLR mutation, the patient was included on the basis of other criteria, most often proven CAD in the patient or in a first degree relative under the age of 60 years (data not shown). Patients with CVD were older, more often males and smokers (data not shown).

Diagnostic criteria

The phenotypic heterogeneity detected in our study cohort challenges the value of the current clinical criteria for diagnosing FH. Criteria used to identify individuals with FH include a combination of clinical characteristics, personal and family history of early CVD and biochemical parameters. In the present study, we used a set of established clinical diagnostic criteria to identify patients with FH.1,36 The primary diagnostic criterion is the detection of an increased plasma LDL-C. However, due to overlap in plasma lipid levels between FH heterozygotes and the general population, the cut-off values of LDL-C are difficult to define.23 Arbitrary definition of cut-off values for LDL-C leads to a compromise in sensitivity and specificity of this marker as a diagnostic criterion. Moreover, in one important report where stringent cut-off values were used, misdiagnosis occurred in 18% of mutation carriers and non-carriers on the basis of plasma cholesterol alone.8 Analysis of the diagnostic value of LDL-cholesterol concentrations has revealed that the best available cut-off point to diagnose the disorder is the age- and sex-specific 90th percentile.8 In our set of criteria, we used the age- and sex-specific 95th percentile, as obtained from the general Dutch population, to further reduce the inclusion of false positive FH patients.

An important distinguishing feature of FH is the presence of tendon xanthomas in the patient.1,2 Surprisingly, in our study population, the prevalence of tendon xanthomas did not differ between the LDL-R plus and LDL-R minus group. However, when we further subdivided these groups, we could demonstrate that FH patients, defined by DNA diagnosis and the presence of tendon xanthomas, had more severely elevated TC and LDL-C levels and were characterized by a higher prevalence of CVD.

It is of course possible that the current study has misclassified patients who have yet to be genetically diagnosed with FH. Nevertheless, we would expect the misclassification of true FH cases to have reduced the differences observed between the two groups. The presence of tendon xanthomas as recorded in the medical record is therefore subject to speculation. According to the Simon Broome Register criteria and the MedPed criteria, in which the presence of tendon xanthomas is the key feature of ‘definite’ FH, we included ‘possible’ FH patients in our study, which may have contributed to the heterogeneity of the study population. One could conclude that we therefore should have restricted inclusion of patients to those with tendon xanthomas if a mutation had not been identified. However, when the LDL-R minus group was subdivided into those patients with and without xanthomas, the phenotypic differences between the LDL-R plus and LDL-R minus groups remained. A possible explanation for this is the false-positive palpation of tendon xanthomas. Objective palpation (i.e. without knowing a patient's LDL-C concentrations) is not likely in an outpatient clinic setting.

A definite diagnosis of FH is made after identification of an LDLR mutation known to cause FH. A molecular diagnosis of FH can be made in 50–80% of clinically identified cases.11,24,25 This broad detection range is due to clinical misdiagnosis, technical insensitivity, or causes of FH not related to the LDLR gene.10,26,27 A higher mutation detection rate has been reported in ‘definite’ FH patients (those with tendon xanthomas) as opposed to those with ‘possible’ FH (without tendon xanthomas).25

Unfortunately, the presence of an arcus cornealis and a positive family history for premature CAD are insensitive diagnostic markers. The presence of an arcus cornealis by the age of 50 years occurs in 50% of patients, but not until a decade later in non-FH patients.28 However, an arcus cornealis can also be observed in subjects with normal lipid levels.1 A strong family history for premature CAD contributes to a patients risk of coronary heart disease; however, only 5–10% of early familial CAD can be attributed to FH.19,29

The results of LDLR gene sequencing of the random sample of patients from the LDL-R minus group may indicate that this group partially comprises patients with other causes of autosomal dominant hypercholesterolaemia.10,26,27 Mutations in PCSK9, a candidate gene on chromosome 1 encoding a protease named NARC-1, have recently been proposed as a new cause for hypercholesterolaemia.30 At present, sequencing of the remainder of the random sample of patients for mutations in this gene is ongoing. However, due to the disappointing outcome of sequencing of the random sample, further LDLR gene sequencing of the group without an LDLR mutation does not have a high priority in our laboratory.

The sensitivity and specificity of similar existing sets of diagnostic criteria have yet to be compared, possibly leading to the creation of more decisive clinical diagnostic criteria. In the meantime, advances in the application of DNA diagnostic techniques will hopefully favour the ultimate decisive criterion, namely, the genetic test.

The relevance of genetic testing in clinical practice and in screening for disease

FH is a disorder that begins early in life. Effective treatment to reduce cardiovascular morbidity and mortality has been available since the introduction of statins. In children with FH, increased LDL-C deteriorates endothelial function at a very young age, in addition to rapidly increasing the intima-media thickness of peripheral arteries.31 Such findings support the notion of taking preventive measures when children are young instead of waiting until they reach adulthood. The long-term efficacy and safety of cholesterol-lowering medication in children was recently evaluated by Wiegman et al.32 2 years of pravastatin therapy induced a significant regression of carotid atherosclerosis in children with FH, without any adverse effects on growth, sexual maturation, hormone levels, or liver or muscle tissue. It is therefore justified to initiate statin therapy in FH children. Furthermore, genetic testing is possible at a young age, whereas clinical manifestations such as a high LDL-cholesterol and tendon xanthomas often appear at a later age. Thus, there is a need for accurate criteria for the early diagnosis of FH. In addition to the presence or absence of a mutation in the LDLR gene, differences in prognosis due to the type of LDLR mutation33,34 could also influence initiation of lipid-lowering treatment, although the differences due mutation type are of less clinical relevance compared to the presence or absence of the mutation.

Genetic testing plays a central role in screening programmes for FH which are under development in many Western countries.8,35 These programmes aim to identify and treat asymptomatic individuals who are at high risk for CVD.

The relevance of genetic testing in research

Proper definition of study subjects is essential in any type of research. Moreover, international collaboration in research is facilitated when study subjects are defined according to highly sensitive and specific criteria. In the various forms of hyperlipidaemia, hypercholesterolaemia is due to different underlying defects in lipid metabolism. Careful interpretation of the outcomes of clinical drug trials that include FH patients who have not been genetically diagnosed is therefore required.36

In summary, the present study emphasizes the relevance of genetic testing in FH for screening and treatment of FH patients and for proper definition of study subjects in research with FH patients.

Acknowledgements

This study was supported by a grant of the Netherlands Heart Foundation (98/165). J.J.P.K. is an established investigator of the Netherlands Heart Foundation (grant D039/66510). We thank all patients who participated in the study and the specialists of the participating lipid clinics throughout the Netherlands.

Conflict of interest: none declared.

References

View Abstract