Inheritance pattern of familial hypercholesterolemia and markers of cardiovascular risk.

Studies in children and adults have resulted in conflicting evidence in the quest for the answer to the hypothesis that offspring from hypercholesterolemic mothers might have an increased cardiovascular risk. Previous studies might have suffered from limitations such as cohort size and clinical sampling bias. We therefore explored this hypothesis in large cohorts of both subjects with familial hypercholesterolemia (FH) and unaffected siblings in a wide age range. In three cohorts (cohort 1: n = 1,988, aged 0-18 years; cohort 2: n = 300, 8-30 years; cohort 3: n = 369, 18-60 years), we measured lipid and lipoproteins as well as carotid intima-media thickness (c-IMT) in offspring from FH mothers versus FH fathers. For LDL cholesterol, triglycerides (TGs), and c-IMT, we performed a pooled analysis. No significant differences could be observed in c-IMT, lipid, or lipoprotein levels from offspring of FH mothers versus FH fathers. Pooled analyses showed no significant differences for either LDL cholesterol [mean difference 0.02 (-0.06,0.11) mmol/l, P = 0.60], TGs [mean difference 0.07 (0.00,0.14) mmol/l, P = 0.08], or c-IMT [mean difference -0.00 (-0.01,0.01) mm, P = 0.86]. Our data do not support the hypothesis that cardiovascular risk markers are different between offspring from FH mothers and FH fathers.


Study population
For the current study, participants of different Dutch cohorts were included. The fi rst cohort consisted of children with FH and their unaffected siblings aged 0-18 years, who consecutively visited the AMC Pediatric Lipid Clinic between 1989 and July 2012 (cohort 1, n = 3,010). The second cohort comprised children with FH (aged 8-18 years) who participated in a randomized placebo-controlled trial to assess the effi cacy and safety of pravastatin, and their healthy siblings in the same age range (cohort 2a, n = 309) ( 16,17 ). Both the FH patients and their siblings were followed up after 10 years, and they constituted cohort 2b (n = 277). The third cohort consisted of participants of a crosssectional study in which subjects aged 18-60 years were recruited from the database of the screening organization for FH in The Netherlands, within 18 months after genetic screening (cohort 3, n = 440) ( 18 ). The study was approved by the Medical Ethical Committee of the Academic Medical Center, Amsterdam, The Netherlands. All participants of cohorts 2a, 2b, and 3 gave informed consent.
The diagnosis of FH in all patient cohorts was based on the presence of a documented pathogenic LDL receptor (LDLR) or apoB mutation. For all mutations, functionality has been established by cosegregation analysis of pedigree data. Siblings were included if they had a documented absence of the known family mutation. FH patients and unaffected siblings were then divided into two groups: one with subjects whose mother had FH and one with subjects whose father had FH. Patients with homozygous FH or compound heterozygous FH, and subjects for whom the mode of inheritance of FH was unknown were excluded from the analyses. Information about demographic characteristics, classical risk factors, and fasting lipid levels was obtained from the patient's medical records, or obtained from patients at their fi rst visit at the AMC Pediatric Lipid Clinic.

Lipid and lipoprotein levels
Lipid and lipoprotein levels were determined in fasting patients in all cohorts. Plasma total cholesterol (TC), HDL-cholesterol (HDL-C), and triglyceride (TG) levels were measured by standard methods, and LDL-C levels were calculated using the Friedewald formula. Plasma apoB-100 and apoA-I were measured by standard methods.

DNA analysis
In all cohorts, genomic DNA was prepared from 5 ml whole blood on an AutopureLS apparatus according to a protocol provided by the manufacturer (Gentra Systems; Minneapolis, MN). In patients from families with a known molecular diagnosis, mutations in LDLR, APOB, and PSCK9 were detected as previously described ( 19 ). In patients with an unknown mutation in the family, mutation identifi cation in the LDLR and APOB genes was performed by direct Sanger sequencing; identifi cation of large rearrangements in the LDLR gene was done by multiplex ligation-dependent probe technique as described previously in more detail ( 19 ). Sequence analysis was performed by direct sequencing with the Big Dye Terminator ABI Prism Kit, version 1.1 (Applied Biosystems; Foster City, CA). Products of sequence reactions were run on a Genetic Analyzer 3730 (Applied Biosystems), and sequence data were analyzed by the use of the Sequencer package (GeneCodes Co; Ann Arbor, MI).

c-IMT
Carotid ultrasound measurements of IMT were performed in the subjects of cohort 2 (a and b) and in cohort 3. These measurements were performed according to a standardized and validated methodology as described in detail before ( 20 ). Ultrasound measurements of cohort 2a were performed by one experienced sonographer using the Acuson 128XP/10v (Acuson Corp., Mountain View, CA) ultrasound instrument equipped with a 5-10 MHz L7 transducer. In cohorts 2b and 3, ultrasound measurements on all participants were made by one (cohort 2b) or two (cohort 3) experienced sonographer(s) using the Acuson Sequoia 512 instrument (Siemens AG; Malvern, PA and Erlangen, Germany) equipped with an 85 Mhz linear array transducer. Still images were saved as DICOM fi les. One certifi ed image analyst analyzed these images. Both the sonographers and the image analyst were blinded to clinical genetic data and inheritance pattern. Mean c-IMT was defi ned as the mean IMT of the right and left common carotid, the carotid bulb, and the internal carotid far wall segments.

Statistical analysis
The different cohorts were analyzed separately. Differences in variables between the "mother" and "father" group were evaluated for the subjects with and without FH separately, as well as for the cohort as a whole. Differences in variables with a continuous or a dichotomous distribution between groups were evaluated using linear or logistic regression analyses, respectively. By means of multivariable regression analysis, we adjusted for potential confounders. The analyses were performed using the generalized estimating equation method to adjust for correlations within families.
Variables with a skewed distribution were log-transformed before statistical analyses. A P value <0.05 was considered statistically signifi cant. The analyses were performed with the SPSS package version 19 (SPSS, Inc.; Chicago, IL).
With cohort 1, 2b and 3, a pooled analysis was performed for LDL-C and log-transformed TG levels, and with cohort 2b and 3 for c-IMT. For each of the cohorts, we calculated a mean difference and 95% CI, adjusted for age, sex, body mass index (BMI), statin use, family relations, and, in the case of c-IMT, also blood pressure. A random-effects model according to the method of inverse variance was used. The forest plots were visually examined; we tested for heterogeneity of mean differences across the cohorts using the Cochran's Q test, and we measured the proportion of between-cohorts differences not attributable to chance with the I 2 statistic. A z test was performed to test the overall effect. Analyses were performed using Review Manager 5 (the Cochrane Collaboration).

General characteristics
Of the children in cohort 1 who were eligible for this study, 1,664 children had a documented pathogenic LDLR or APOB mutation, and in 324 siblings, the specifi c family mutation was absent. These 1,988 children of cohort 1 could be included in the study (see supplementary In cohort 2, of 211 children with molecularly proven FH, 89 unaffected siblings could be included at baseline (cohort 2a); and 192 subjects with molecularly proven FH and 78 siblings without FH could be traced after 10 years and were included in cohort 2b. Of the third cohort, 267 subjects with molecularly proven FH and 102 subjects without FH were included. Because the subjects of cohort 2b were the same subjects of cohort 2a (but 10 years later), only inherited FH maternally (13.4 ± 3.0 and 12.3 ± 2.7 years, respectively; P = 0.006). After a mean follow-up period of more than 10 years (cohort 2b), more statin users were seen in the group of FH patients who inherited FH paternally compared with subjects who inherited FH maternally (92.1% versus 77.1%, respectively; P = 0.004). Both the FH group and the sibling group of cohort 3 were comparable. For patients with FH in cohort 3, median (interquartile range) duration of statin use was 0 (0-7.5) months for subjects whose mother had FH and 0 (0-9.25) months for subjects whose father had FH ( P = 0.956).

Lipids and lipoproteins
In cohort 1, no differences in lipid or lipoprotein levels between children whose father had FH compared with those of cohort 2b were included in the pooled analysis. So the total number of subjects included in the pooled analysis of lipid levels was 2,962, and for the pooled analysis of c-IMT, the total number was 661. Mutation distribution of the different cohorts is presented in supplementary  Tables I-III. General characteristics of the different cohorts are presented in Table 1 . In cohort 1, FH subjects whose father had FH were comparable with those whose mother had FH. In the sibling group, there were differences in smoking status between subjects whose father had FH and subjects whose mother had FH (0% versus 4.8%, respectively; P = 0.005) and in diastolic blood pressure (68.5 ± 5.0 mmHg versus 60.2 ± 7.9 mmHg, P = 0.001). In cohort 2a, subjects who inherited FH paternally were slightly older compared with subjects who adjusted P = 0.058]. In the sibling group, none of the differences in lipid levels reached statistical signifi cance ( Table 2 ). In the third cohort, a statistically signifi cant elevation in TG levels in subjects who inherited FH maternally could be observed [0.80 (0.57-1.18) mmol/l vs. 0.70 (0.48-1.00) mmol/l; P = 0.013; adjusted P = 0.002], but there was no difference in the sibling group ( Table 2 ).
In the pooled analysis of both the FH and sibling groups of cohort 1, 2b, and 3 (n = 2,962, Fig. 1 ) , no difference could be observed in LDL-C or TG levels [mean difference LDL-C of 0.02 ( Ϫ 0.06, 0.11) mmol/l; P = 0.60; mean difference TG of 0.07 ( Ϫ 0.00, 0.14] mmol/l; P = 0.08]. A subanalysis with defective LDLR mutations, defi cient LDLR mutations, and apoB mutations showed no signifi cant differences between subjects who inherited FH paternally or maternally in any of the mutation groups, except for subjects of cohort 1 with an apoB mutation (4.48 ± 0.59 with FH mother vs. 4.12 ± 0.59 with FH father, P = 0.005) (see supplementary Table IVA-C). children whose mother had FH became evident, nor in the FH group neither in the sibling group ( Table 2 ).
In the FH group of cohort 2b, elevated LDL-C levels were observed in subjects who inherited FH maternally, but the difference lost statistical signifi cance when adjusted for age, gender, BMI, statin use and family relations   Fig. 1. Forest plots of differences in levels of LDL-C, TG, and c-IMT between subjects whose mother had FH and subjects whose father had FH.

Carotid IMT
In none of the cohorts, could a difference in c-IMT between subjects whose mother had FH and subjects whose father had FH be observed, or a trend toward a thicker c-IMT in subjects whose mother had FH ( Table 2 ). In the pooled analysis (n = 661), the mean difference was Ϫ 0.00 ( Ϫ 0.01, 0.01) mm, P = 0.86 (Fig. 1) .

DISCUSSION
In this study, no difference in lipid and lipoprotein levels of offspring from FH fathers as compared with FH mothers could be demonstrated. Although in some of the sub-analyses, TG levels were higher in subjects whose mother had FH, this was not consequent in all cohorts, and in the pooled analysis, the difference in TG levels between those who inherited FH maternally and those who inherited the disease paternally did not reach statistical signifi cance. So this fi nding was probably due to chance. Also, no clear trend was shown toward a more-atherogenic lipid profi le in subjects whose mother had FH, and c-IMT was, in fact, similar in both groups.
These results are in line with two other human studies ( 8,15 ). Napoli et al. ( 8 ) showed that lipid levels in children from (non-FH) hypercholesterolemic mothers did not differ from those of children from normocholesterolemic mothers. However, these authors did show that atherogenesis was more pronounced in children of hypercholesterolemic mothers. Tonstad et al. ( 15 ) found no statistically signifi cant differences in lipid levels and c-IMT between the children who inherited FH maternally or paternally. However, it must be taken into account that both of these studies were only small and might have lacked the power to detect small differences.
Our results are in contrast with an animal study in which it was shown that apoE knockout mice exhibited higher cholesterol levels in offspring from hypercholesterolemic mothers as compared with genomically similar animals born from wild-type mothers ( 21 ). Moreover, in a human study by our research group, which was performed previously, we found that adult FH patients who inherited FH through their mothers had slight but signifi cantly increased levels of TC, LDL-C, and apoB compared with adult FH patients who inherited FH through their fathers ( 12 ).
Although the mode of analysis of our previous and current study were comparable, there were also some differences. In our previous study, only adult patients were involved, and nonaffected siblings were not included. Possibly, differences in lipid profi les as a result of inheritance pattern are more pronounced in adult patients. Because the majority of our current study are children, this could explain the confl icting results. Indeed, a trend toward higher TC, LDL-C, TG, and apoB levels in those who inherited FH through their mothers, is more pronounced within the adult subjects of our current study (cohorts 2b and 3) than within the children (cohorts 1 and 2a). The (trends toward) differences in lipid levels were comparable in the patients with and without FH, so the explanation that nonaffected siblings were included in the current study as a means to account for the discrepancies with our previous study is less likely. Furthermore, patients with a clinical diagnosis of FH were included in the previous study as well, in contrast to our current study, in which all FH patients had a molecular diagnosis and siblings had a proven absence of the family mutation. Possibly, in the previous cohorts, in some patients a more polygenic basis might have been involved in the FH phenotype. Mitochondrial genes might have contributed to increased lipid levels and are always inherited maternally. Possibly, this might have contributed to the higher lipid levels in subjects whose mother had a FH phenotype. In our current study, in which in all patients the FH phenotype was explained by a mutation in the LDLR or APOB gene, a mitochondrial role for increased lipid levels is less likely.
Some aspects of our study merit discussion. First, as our data are a result of post hoc analyses, the different cohorts were not designed to study our hypothesis. However, because in all cohorts, family history and pedigree data were questioned comprehensively as part of cardiovascular risk assessment, we feel that the details on the mode of inheritance are robust. Second, in both adult cohorts (cohorts 2b and 3), a substantial part of the FH patients were using statins. Because in most patients, detailed information about the exact treatment and dosage was absent, we could only adjust for the confounder "current statin use." If FH patients in either the paternal or maternal group were using more-effi cacious statins because of a higher untreated cholesterol, we missed this difference and may conclude that there is no difference in treated cholesterol levels, whereas a difference in untreated cholesterol might indeed exist. Although this is a serious limitation of our study, we did a sub-analysis with the patients whose exact treatment regime was known. Untreated LDL-C levels were calculated by the estimated LDL-C-lowering potency of the specifi c lipid-lowering drug and dose, as previously described by Huijgen et al. ( 22 ), and showed no difference between patients with an FH mother as compared with patients with an FH father.
In conclusion, our current data on subjects in the mainly pediatric age range do not show a more-atherogenic lipid profi le or greater c-IMT in offspring from mothers with FH than offspring from fathers with FH, although offspring from FH mothers are exposed to very high LDL-C levels during a very important period in their early life. Therefore, this study does not support the possible mechanism of epigenetic programming of metabolism during fetal development as a result of higher cholesterol exposure in utero. Regarding lipid levels, in the single human study in which a signifi cant difference was found, this difference was very small ( 12 ). Taken together, the question arises as to the clinical relevance of the inheritance pattern for FH. This is an important question, because if maternal inheritance actually does make a clinical difference for the outcome of the unborn child, this would indicate that there is a need for better regulation of lipid levels in pregnant women with FH. Because statins are contraindicated, other lipid-lowering options should then be considered. Furthermore, the progeny of these mothers would require more frequent follow-up and, if needed, lipid-lowering therapy at the time that their age and LDLc levels met current guidelines for intervention. We therefore plan to repeat our analyses in other, larger cohorts with both FH patients and nonaffected siblings, across the complete age range. The observation that differences in lipid levels as a consequence of inheritance pattern may possibly only express in adulthood is interesting and should be further explored. More importantly, clinical endpoints should be included in these analyses, to fi rmly establish or exclude clinical relevance.