Answers are grounded in this paper and cite their sources.
This editorial highlights that familial hypercholesterolemia is more common than historically estimated (1:217) and emphasizes the importance of phenotypic screening using DLCN criteria over genetic testing alone.
This editorial refers to ‘Mutations causative of familial hypercholesterolaemia: screening of 98 098 individuals from the Copenhagen General Population Study estimated a prevalence of 1 in 217’†, by M. Benn et al., on page 1384. Familial hypercholesterolaemia (FH) has been described extensively ever since the first descriptions of the clinical hallmarks of the disease by Muller in 1938.1 Today, rarely a week passes without a new publication about FH (see Figure 1) and this has been increasing steadily over recent years. This might reflect not only our greater understanding of FH and the public health challenge it poses, but also the fact that a large number of novel lipid-lowering therapies are currently under development. FH is a genetic disorder characterized by high plasma levels of LDL-cholesterol (LDL-C) from birth onwards,2 which results in accelerated atherogenesis and clinical manifestations of premature vascular disease. This autosomal co-dominant condition is caused by mutations in the genes encoding the LDL receptor, apolipoprotein B100 (apoB), or proprotein convertase subtilisin/kexin-9 (PCSK9). The historical prevalence of FH was estimated to be in the order of 1 in 500, but more recent data3 suggest that it is between 1:200 and 1:250. Similar numbers were found in other populations, while using different study methods.4,5 A considerably higher prevalence was observed in populations with a founder effect such as French Canadians.6 Number of publications on familial hypercholesterolaemia per year. Currently the detection rates of FH vary from 70% in The Netherlands to <5% in most countries.7 It is noteworthy that these numbers are based on a prevalence of 1:500, and, as a consequence, detection rates are even worse when using the more recent assumption of approximately half that, i.e. a prevalence of 1:250. Much of this variation in detection stems from whether to screen and, if so, how. Unsurprisingly as FH is a disease of elevated cholesterol, premature vascular disease, and is monogenic, these three components feature heavily in the three most commonly used ‘scoring systems’. The MEDPED criteria are largely used in North America and employ age-specific LDL-C values, whereas the Simon Broome criteria use age-specific LDL-C levels, family history, and clinical manifestations (xanthomas) as well as genetic data to assist diagnosis. The most detailed of all methods is the Dutch Lipid Clinic Network (DLCN) criteria arising from data from the extensive screening programme, and the DLCN criteria are recommended in order to establish the clinical diagnosis of FH.7 However, whether one algorithm is more suitable than another has largely not been assessed. It is of critical importance to diagnose FH uniformally and to establish the prevalence of FH, both in light of the currently set up screening programmes and for reasons of possible reimbursement of novel agents, such as PCSK9 inhibitors, which are potentially restricted to FH patients. The study of Benn and co-workers, reported in this issue of the journal,9 adds to our insights into the prevalence of FH and also provides guidance about the LDL-C cut-off value that could be used in order to identify a molecular defect. The authors also evaluated which of the three clinical risk scores is best for the prediction of the presence of a pathogenic mutation. The authors genotyped the three most prevalent mutations in LDLR and one mutation in APOB in a cohort comprising almost 100 000 individuals of Danish descent. The participants in this Copenhagen General Population Study were between 20 and 100 years of age and were enrolled since 2003 without any specific selection criteria and so the population may reflect survival biases. Participants with potential secondary causes of elevated LDL-C levels were excluded. In a previous study it was found that the four mutations account for ∼38% of all FH-causing mutations in the Danish population and, based on this number, the prevalence of FH was calculated to be 1:217. This observation is in line with results from earlier studies showing a prevalence of FH between 1:200 and 1:250. The metric used to calculate the prevalence is, however, based on the relative distribution of the four genotypes among 142 patients with molecular defined FH. In the authors' supplementary table 2, it is shown that 13% of the 142 patients with definite FH carry an APOB R3500Q mutation, while 15, 9, and 1% are heterozygous for the W66G, W23X, and W556S mutation, respectively. In the current study, however, an APOB mutation was found in 64% of the mutation carriers, which is higher compared with previous reports and strikingly higher compared with the cohort that was used for the prevalence estimation. This has a huge impact on the overall outcome. Moreover, it should be emphasized that these APOB mutation carriers were largely identified in the groups classified by the DLCN clinical criteria as ‘unlikely FH’ or ‘possible FH’ (105 of the total of 111 APOB mutation carriers) and no carriers were found among the 25 participants that were classified as ‘definite FH’. This clearly shows the relatively weak effect of APOB R3500Q mutations on LDL-C levels, compared with the LDLR mutations, an observation that was previously described.10 These data highlight the dangers or relying on genetic diagnosis alone rather than the phenotype as discussed below. The present analyses also provide a comparison of the three commonly used criteria to identify FH clinically. Of these, Benn and colleagues observe that the DLCN criteria provided the highest odds of a pathogenic mutation, as high as 439 [95% confidence interval (CI) 170–1138] for definite FH, 90 (95% CI 53–152) for probable FH, and 18 (95% CI 13–25) for ‘possible’ FH vs. ‘unlikely’ FH. In contrast, these figures were lower using MEDPED criteria: 40 (95% CI 28–58) for probable vs. unlikely FH, and 27 (95% CI 20–36) for ‘possible’ vs. ‘unlikely’ FH using the Simon Broome criteria. These findings probably reflect the deeper phenotyping involved in the DLCN criteria and are likely to apply to other populations. Hence, from a practical perspective, the present data suggest that the DLCN criteria probably should be used as the basic screening clinical tool rather than MEDPED or Simon Broome criteria. The definite answer about the prediction by means of the algorithms is not provided by this study, since some crucial aspects of the different models were not available in this study (i.e. data about xanthomas). Another important question is when is molecular screening likely to yield a molecular defect? Unsurprisingly, LDL-C concentrations were found to discriminate between carriers and non-carriers in the current study and while using receiver operating characteristic (ROC) curves the optimal threshold across all ages combined was 4.4 mmol/L, and the area under the curve (AUC) was the largest in individuals below 40 years of age. This observation is in line with the finding that high LDL-C levels in young individuals (especially in children) are a stronger predictor for the presence of a mutation compared with similar LDL-C levels in older people. In the latter, secondary causes are more likely to be a causal factor. In a previous study, Huijgen and co-workers used the LDL-C level at the 90th percentile for age and sex in their analysis performed in a cohort of >5000 mutation carriers and >15 000 family controls. The sensitivity of the 90th percentile was 91.3% while focusing on pathogenic mutations, whereas a sensitivity of 68.5% was observed when analysing the data for all mutations (including non-pathogenic mutations).11 This clearly refers to the observed large overlap of LDL-C levels among carriers and non-carriers, especially in those with relatively mild mutations. It is important to emphasize that the vascular consequences of FH are driven by the LDL-C level, and not by the genotype. As such, the clinical hallmarks are widely considered to be of larger clinical relevance than the genotype per se. The consensus panel highlights that the decision to start lipid-lowering therapy depends not on the genotype but on the phenotype, i.e. even in the absence of a known mutation an elevated LDL-C warrants lipid modification therapy.7 A putative mutation might not be found in up to 40% of patients meeting the clinical criteria of FH and, in these cases, an additive effect of multiple common genetic variants, each with a modest effect on LDL-C,8 or a novel monogenic defect might be the causal factor.12 Conversely among those discovered to have a gene mutation whose LDL-C levels are below the recommended levels for starting therapy, a surveillance approach is advocated.7 In this regard, what is the role of genetics? Knowledge about the genotype has benefits for patient engagement in some cases such as compliance with medication, lifestyle, etc., but also for screening family members after identification of the index case (cascade screening). Typically genotyping the index case costs more as the mutation is unknown, but can be narrowed down in relatives once the initial genetic defect is identified. Information about who is likely to have one of the major causal mutations for FH and establishing which criteria best identify this or at a global level which LDL-C cut-off point is most useful to guide referrals has potential benefits. This, however, may vary from population to population, and it is not immediately clear whether in other racial groups the figure of 4.4 mmol/L holds. For instance, in lipid clinics in Western Australia, an LDL-C of 5.9 mmol/L predicts a pathogenic mutation in index cases with maximal sensitivity and specificity, and this number is even higher for instance in Brazil at 6.5 mmol/L (G.F. Watts, personal communication). The most unbiased assessment of the appropriate LDL-C threshold to use for screening probably comes from the DLCN criteria which advocate a threshold of 5.9 mmo/L. Given the limitations of the data pertaining to APOB in the present study, it does not make sense to suggest that the threshold for genetic screening be reduced to 4.4 mmol/L in other countries. The definition of a threshold above which genotyping should be performed is driven by decision-making about the balance between the higher cost of screening that comes with a lower treshold (genotyping in a large cohort, resulting in a large negative screening outcome) and the lower cost of screening once a higer threshold is chosen (at a price of not diagnosing FH patients without very high LDL-C levels). In their study, Benn and colleagues also addressed the role of the presence of a molecular defect on cardiovascular disease (CVD) risk, and the adjusted odds ratio was 11 and 4.7 for carriers of LDLR and APOB mutations, respectively. It should be noted that carriers on lipid-lowering therapy were compared with non-carriers off medication, and this might have resulted in an inflation of the effect of the mutations. In previous studies, odds ratios for cardiovascular disease have been described to vary between 1.8 and 16.4,13 The diversity in risk is likely to be related to the definition of FH, the control population analysed, and the medication used by cases and controls. The striking consistency in all reports, however, is the finding of the sharply increased risk. That observation per se should not only result in greater awareness among the medical community, but should also encourage active screening and subsequent adequate CVD risk reduction in FH patients as suggested by the EAS FH Studies Collaboration.14 Conflict of interest: K.K.R. received honoraria for consultancy, ad boards, and Steering Committees for Amgen, Aegerion, Abbivie, Medcon, Pfizer, AZ, Kowa, Sanofi, Regeneron, MSD, ISIS pharma, Alogorithm, Lilly, Novo Nordisk, Novartis, Takeda, and BI. He received research support from Pfizer, Sanofi, Amgen, Rgeneron, and MSD. G.K.H. or his institution received honoraria for consultancy, ad boards, and/or conduct of clinical trials from: AMGEN, Aegerion, Pfizer, Astra Zeneca, Sanofi, Regeneron, KOWA, Ionis pharmaceuticals, and Cerenis. He received research support from Aegerion, AMGEN, Sanofi Astra Zeneca, and Synageva.
No takes yet. Share an insight, caveat, or question.
Ray et al. (2016) conducted an editorial in Familial hypercholesterolaemia (n=98,098). LDLR and APOB mutations vs. Non-carriers was evaluated on Cardiovascular disease (CVD) risk for LDLR mutation carriers (OR 11).
Synapse has enriched 10 related papers. Consider them for comparative context: