点击显示 收起
1 From the INSERM U525, Nancy, France (SG, GS, SV, and BH); the Faculté de Pharmacie, Nancy, France (PL); and the Centre de Médecine Préventive, Vandoeuvre-lès-Nancy, France (RG and GS)
2 Supported by grants from The Centre d'Etude et d'Information sur les Vitamines and Kellogg's PA. 3 Address reprint requests to B Herbeth, INSERM U525, Centre de Médecine Préventive, 2 rue Jacques Parisot, 54500 Vandoeuvre-lès-Nancy, France. E-mail: bernard.herbeth{at}cmp.u-nancy.fr
ABSTRACT
Background: Although numerous environmental factors are documented to influence serum retinol and -tocopherol concentrations, little is known about the genetic versus the environmental contributions to variations in these traits.
Objective: The aim of this study was to estimate additive genetic heritability and household effects for serum retinol and -tocopherol concentrations in a variance component analysis.
Design: In a sample of 387 French families, information on serum retinol and -tocopherol concentrations, usual dietary intake, lifestyle, and serum lipid profiles and related polymorphisms (apolipoprotein E, apolipoprotein C-III, apolipoprotein B, cholesteryl ester transfer protein, and lipoprotein lipase) was obtained.
Results: For serum retinolafter adjustment for sex, age, body mass index, alcohol consumption, oral contraceptive use, and serum albumin, triacylglycerol, and apolipoprotein A-I concentrationsadditive genetic effects and shared common environment contributed 30.5% and 14.2% of the total variance, respectively. For serum -tocopherol, 22.1% of the total variance was due to the additive effects of genes and 18.7% to those of household environment, after adjustment for the covariates sex, age, vitamin E intake, oral contraceptive use, and cholesterol, triacylglycerol, and apolipoprotein A-I concentrations. For both vitamins, the influence of measured polymorphisms was not significant. Moreover, heritability and household effect estimates were not significantly different between the 4 classes of relatives and did not vary significantly when families shared more meals at home.
Conclusions: The results show that serum retinol and -tocopherol concentrations are under genetic control in healthy families.
Key Words: Retinol -tocopherol family resemblance genetics household environment
INTRODUCTION
Developing statistical methods to analyze data collected from nuclear families allows investigation of the traits that underlie the strong influence of family history on the risk of various chronic diseases, such as cancer, osteoporosis, and atherosclerosis, and related cardiovascular complications. The identification of heritability should be a prerequisite for the search for quantitative trait loci affecting such traits and thereby modulating the risk of disease. Such investigations, however, must account for the fact that these intermediate traits themselves are multifactorial in nature, being influenced by both genes and environmental factors. Strong evidence indicates genetic familial influences on bone mineralization (1), body mass index (2), dyslipemia (3), glucose intolerance (3), hypertension (3), hemostatic factors (4), platelet aggregation (5), total antioxidant activity (6), and vitamin D concentrations (7). However, data regarding the genetic epidemiology of circulating indexes of retinol and -tocopherol status are lacking.
Mutations in genes encoding key proteins, such as retinol-binding protein (RBP) (8)the main carrier for vitamin Aor prealbumin (9), which is known to form a complex with RBP in plasma, have been identified. Such genetic defects induce severe biochemical vitamin A deficiency and low concentrations of plasma retinol and RBP.
Concerning vitamin E, mutations in the -tocopherol transfer protein (-TTR) gene have been detected in patients with low plasma -tocopherol and ataxia with isolated vitamin E deficiency (10, 11). -TTR is a liver protein responsible for the selective retention of -tocopherol from dietary vitamin E and for its transfer into nascent VLDL. This key phase is the major determinant of plasma -tocopherol concentrations in humans (12).
Because of the particular metabolism of these 2 fat-soluble vitamins (13, 14), genes whose products affect lipoprotein metabolism, eg, apolipoproteins (apos), enzymes, and receptors, particularly in response to dietary change, should be taken into account (15, 16). However, few studies have investigated this possibility (17). The present study aimed at estimating additive genetic heritability and household effects of serum retinol and -tocopherol concentrations in a variance component analysis, with adjustment for the influence of known environmental covariates and polymorphisms of apo E, apo C-III, and apo B; cholesteryl ester transfer protein (CETP); and lipoprotein lipase (LPL) known to affect lipid and lipoprotein metabolism.
SUBJECTS AND METHODS
Subjects and study design
This work is part of the Stanislas Family Study, a 10-y longitudinal follow-up study conducted since 1994 in 1006 families selected at the Center for Preventive Medicine of Vandoeuvre-lès-Nancy (France) who were free of chronic or acute disease that could influence nutritional status (18). In this article, we present data from the first examination (19941995), which was obtained from a random subsample of 383 families composed of 2 parents aged 2859 y and at least one child between 6 and 24 y of age (n = 1487). All subjects underwent a complete medical examination, including weight and height measurements. Data on alcohol consumption, smoking status, and drug use (especially oral contraceptives, lipid-lowering agents, and vitamin supplements) were collected through validated questionnaires under the supervision of trained nurses. The research protocol was approved by the "Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale de Lorraine" and each subject gave written informed consent.
Dietary record
Dietary intake was assessed with a 3-d dietary record (19), which was completed during 2 weekdays and a 1 weekend day assigned at random for each family. All subjects received guidelines from a dietitian on the procedures for completing the dietary record and measuring food portions. For young children, the 3-d diary was filled in by the mother and the child together. One week later, the 3-d record was checked and completed by the dietitian with the use of colored photographs of foods, each with 3 different portion sizes. Macronutrient and micronutrient intakes were estimated with an updated computerized version of the Répertoire Général des Aliments (20). Vitamin A intakes were expressed in retinol activity equivalents and estimated as preformed retinol + ß-carotene equivalents/12; ß-carotene equivalents were the sum of ß-carotene and one-half of the amounts of -carotene, -carotene, and cryptoxanthin (21). Vitamin E intakes, expressed in -tocopherol equivalents, were calculated from measurements of -tocopherol and other compounds with vitamin E activity by using recommended conversion factors (20): -tocopherol + ß-tocopherol/2.5 + -tocopherol/10 + -tocopherol/100 + -tocotrienol/3.3 + -tocotrienol/20 + -tocotrienol/100.
In addition, families had to report the number of meals shared at home with all other relatives during the week. An index of shared meals was computed for each family as the average number of meals shared at home by 2 persons in the family in a week and was used as an index of similarity in statistical analyses. We showed previously, as in other studies (22), that associations of parent and child diet intakes tend to be stronger when the family eats more meals at home (19).
Because of the design of the Stanislas Family Study, the 3-d food consumption diary was filled during the week after the blood sampling. The 3-d diary used in the Stanislas Family Study is considered to be a dietary instrument that estimates habitual diet with a relatively good repeatability; this several-day gap should not induce imprecision in the estimation of the relations between the vitamin intakes and their concentrations in serum. Within this short period of time, changes in dietary habits are minimized. Moreover, the amounts of retinol and -tocopherol in the circulating body pool are highly regulated and independent of recent diet intake.
Biological measurements
Blood samples were collected after the subjects fasted overnight. Serum albumin, cholesterol, and triacylglycerol concentrations were measured by using commercially available kits (Merck, Darmstadt, Germany) on AU5021 apparatus (Olympus; Merck). Serum apo A-I and B were determined by immunonephelometry with a Behring Nephelometer Analyser and Behring reagents (Rueil-Malmaison, France). Frozen aliquots of serum were stored in the bio bank of the Centre de Médecine Préventive (Vandoeuvre les Nancy, France).
A reversed-phase HPLC method was used for the simultaneous determination of serum retinol and -tocopherol as adapted from Rudy et al (23). Analytes were isolated by liquid-liquid extraction, concentrated by evaporation, and then chromatographed on an C18, 15 cm x 4.6 mm ODS 5-µm column (Beckman Instruments, Gagny, France), with the use of a water/methanolethyl acetate/isopropanol solvent system. Retinol and -tocopherol are spectrophotometrically detected at 325 and 292 nm, respectively. Standards (retinol and -tocopherol) and their respective internal standards (retinol acetate and -tocopherol acetate) were purchased from Fluka Chemie AG (Buchs, Switzerland) and solvents (HPLC grade) from Merck. Between-run variability was controlled by using a frozen control serum that was measured in duplicate within each of the 45 runs of analysis. The interassay CVs were 8.1% for retinol and 10.7% for -tocopherol.
DNA polymorphism determination
Genomic DNA was extracted from peripheral blood samples by the salting out method (24). The genotypes of apo E Cys112Arg and Arg118Cys, apo B Thr71Ile, apo C-III C(482)T, apo C-III C1100T, CETP Ile405Val, and LPL Ser447term were determined with a multiplex assay that was described previously by Cheng et al (25).
Statistical analysis
Individual statistical analyses were performed by using the SAS software package version 8.01 (SAS Institute Inc, Cary, NC). Triacylglycerol concentrations were log10-transformed in the analyses to improve normality. A chi-square test was performed to determine whether genotype frequencies were in Hardy-Weinberg equilibrium. Before individual and familial statistical analyses, serum retinol and -tocopherol concentrations were adjusted for the effect of between-run variation. Briefly, concentrations were regressed on mean values of the frozen pool of serum measured in duplicate in each series of dosage; the variable used was the sum of residual + crude mean of the overall sample.
For continuous variables, an analysis of variance was performed for characteristic differences between the 4 groups of relatives. When the analysis of variance was significant, a Tukey-Kramer test was used to detect which groups were statistically different from each other. The significance of differences among the groups for the categorical variables was analyzed by using the chi-square test or the Fisher's exact test when cells had expected counts of <5.
In the overall sample, stepwise multiple regression analysis was carried out to select significant covariates (P 0.05) among lifestyle factors, diet intake, related biological analytes, and genetic variants. Then, regression coefficients were computed for the overall sample and for the 4 sex-by-generation groups (fathers, mothers, sons, and daughters). In addition, the significance of differences in regression coefficients between fathers, mothers, sons, and daughters was assessed by testing interaction terms between each covariate and the 4 groups in the overall sample. Because individuals within a family are not independent, statistical analyses were based on the estimating equation technique by using the SAS GENMOD procedure with a repeated statement.
Intrafamilial correlations were estimated by using maximum likelihood techniques (26) with and without adjustment for covariates. This statistical program allowed adjustment for covariates within models, simultaneously and separately for fathers, mothers, sons, and daughters. The significance of various familial correlations or sex and generation differences in correlations was tested by using the log-likelihood ratio test. Correlations were computed under 4 sets of hypotheses: sex effects on correlations for parents and children (general model), sex effects only for children (submodel 1), sex effects only for parents (submodel 2), and no sex effects at all (submodel 3).
Variance component analysis was applied to assess the relative contributions of genetic factors, common household factors, and individual specific environment in family aggregation of serum retinol and -tocopherol concentrations. The variable used to estimate the variance component was adjusted for significant covariatesseparately for fathers, mothers, sons, and daughtersand was standardized to zero mean and unit variance within each sex-by-generation group. The analysis was conducted by using a multivariate normal model for pedigree analysis as described by Lange et al (27, 28) with FISHER software, which also performed tests of goodness-of-fit of the underlying multinormal distribution. The general model assumed that the studied trait was the result of the sum of 3 independent random components: a polygenic component (G) representing additive genetic factors, household factors common to individuals within a family (H), and unmeasured environmental factors particular to an individual, including measurement error (E). These 3 components were assumed to be normally distributed with a mean equal to 0 and a variance equal to 2G, 2H, and 2E, respectively.
The hypothesis of no polygenic component or no household effect was checked by comparing a model including 2Gr, 2Hr, and 2E with a model including only 2Hr and 2Er or 2Gr and 2Er, respectively. In addition, possible effects of covariates on these variance components were tested, such as age, sex, and frequency of sharing meals (families were categorized in 2 groups of similar size: <12 (n = 199) or 12 (n = 184) meals shared/wk.
Comparison of nested models was based on the likelihood ratio criteria. Eventually, the best parsimonious model was selected. The contribution (as a percentage) of the 3 componentsadditive genetic factors (heritability), household factors, and residual environmentalto residual phenotypic variance (after adjustment for covariates) was deduced.
RESULTS
Descriptive characteristics and daily intakes of energy, macronutrients, and vitamins E and A for the 4 sex-by-generation groups are summarized in Table 1. Vitamin A and vitamin E intakes were significantly higher in fathers than in daughters; the mother's and son's intakes were intermediate. When expressed per kJ, the nutritional densities of vitamin A were significantly higher in parents than in children. For vitamin E, nutritional density was significantly higher in mothers than in the 3 other groups.
View this table:
TABLE 1. Descriptive characteristics according to the 4 sex-by-generation groups: lifestyle factors and diet intake1
Serum retinol and -tocopherol concentrations, related biological analytes, and allelic frequencies of selected polymorphisms are presented in Table 2. The apo E, apo B, apo C-III, CETP, and LPL polymorphism distribution did not significantly deviate from Hardy-Weinberg equilibrium. Retinol concentrations in fathers were significantly higher than in children; mothers had intermediate concentrations. -Tocopherol concentrations were significantly different between the 4 groups: fathers had the highest concentrations and sons the lowest. Others characteristics of Tables 1 and 2 were in accordance with the values found in apparently healthy individuals of the same age and sex.
View this table:
TABLE 2. Descriptive characteristics according to the 4 sex-by-generation groups: serum vitamin concentrations, related biological covariates, and genetic variants1
Crude and adjusted mean serum retinol and -tocopherol concentrations are shown according to sex and age classes (59, 1012.4, 12.514, 1517.4, 17.519, 2024, 2534, 3539, 4044, 4549, and 5059 y) in Figures 1 and 2. Values were adjusted for the covariates in Tables 3 and 4.
FIGURE 1.. Mean (±SEM) serum retinol concentrations according to age and sex groups. Adjusted values were adjusted for significant predictors identified in multiple regression analysis (BMI, alcohol consumption, oral contraceptive use, and albumin, triacylglycerol, and apolipoprotein A-I concentrations). For sons, the age classes [n (%)] were 59 y (27, 7.9%), 1012.4 y (63, 18.5%), 12.514 y (120, 35.2%), 1517.4 y (71, 20.8%), 17.519 y (36, 10.6%), and 2024 y (21, 7.0%). For fathers, the age classes [n (%)] were 2534 y (15, 3.9%), 3539 y (94, 24.6%),4044 y (171, 44.6%), 4549 y (79, 20.6%), and 5059 y (24, 6.3%). For daughters, the age classes [n (%)] were 59 y (38, 10.0%), 1012.4 y (77, 20.2%), 12.514 y (94, 24.7%), 1517.4 y (77, 20.3%), 17.519 y (47, 12.4%), and 2024 y (47, 12.4%). For mothers, the age classes [n (%)] were 2534 y (31, 8.1%), 3539 y (150, 39.2%),4044 y (128, 33.4%), 4549 y (62, 16.2%), and 5059 y (12, 3.1%). Two-way ANOVA by age and sex on crude values: P (age) 0.001, P (sex) 0.001, and P (age x sex interaction) 0.001. Two-way ANOVA by age and sex on adjusted values: P (age) 0.001, P (sex) 0.001, and P (age x sex interaction) 0.001.
FIGURE 2.. Mean (±SEM) serum -tocopherol concentrations according to age and sex groups. Adjusted values were adjusted for significant predictors identified in multiple regression analysis (vitamin E intake, vitamin E supplementation, oral contraceptive use, and cholesterol, triacylglycerol, and apolipoprotein A-I concentrations). For the number of persons in each class, see legend for Figure 1. Two-way ANOVA by age and sex on crude values: P (age) 0.001, P (sex) = 0.809, and P (age x sex interaction) 0.001. Two-way ANOVA by age and sex on adjusted values: P (age) 0.001, P (sex) = 0.020, and P (age x sex interaction) = 0.231.
View this table:
TABLE 3. Predictors of serum retinol concentrations in multiple regression analysis in the whole sample and in the 4 groups separately1
View this table:
TABLE 4. Predictors of serum -tocopherol concentrations in multiple regression analysis in the whole sample and in the 4 groups separately1
Crude values for serum retinol concentrations increased with age in both boys and girls from 10 to 24 y of age (P 0.001); the difference between sexes was not significant. In men and women, serum retinol concentrations did not vary significantly with age. Between 25 and 59 y, serum retinol concentrations were significantly higher in males than in females (P 0.001). After adjustment for covariates, age-related variations were of lower range.
From 10 to 14 y, crude values for -tocopherol concentrations significantly decreased and then increased from 15 to 59 y in both males and females (P 0.001). Before the age of 24 y, females had higher mean values than did males (P 0.05); after the age of 24 y the opposite was observed (P 0.01). As for serum retinol concentrations, age-related variations of adjusted -tocopherol concentrations were weaker and the interaction of age and sex was not significant.
Predictors of serum retinol concentrations in parents and offspring are presented in Table 3. The proportion of phenotypic variance accounted for by the measured covariates ranged from 11.3% (in fathers) to 48.8% (in daughters). In the overall sample, females had significantly lower concentrations than did males, and oral contraceptive use, alcohol consumption, and serum albumin and apo AI concentrations were significantly and positively related to serum retinol concentrations; interactions between each covariate and the 4 subgroups were not statistically significant. On the other hand, statistically significant interactions were observed for age (positive regression coefficients being significant only in offspring and significantly higher in sons than in daughters), serum triacylglycerol (positive regression coefficients being significant in the 4 subgroups and significantly lower in sons than in parents and daughters), and BMI (regression coefficients being significant only in fathers and offspring).
Predictors of serum -tocopherol concentrations in parents and offspring are presented in Table 4. The proportion of phenotypic variance accounted for by the measured covariates ranged from 40.4% (in sons) to 52.7% (in fathers). In the overall sample, females had significantly higher concentrations than did males, and age, vitamin E intake, and cholesterol and apo A-I concentrations were significantly and positively associated with serum -tocopherol concentrations; interactions between each covariate and the 4 subgroups were not statistically significant. On the other hand, statistically significant interactions were observed for serum triacylglycerol (positive regression coefficients being significant in the 4 subgroups and significantly higher in parents than in children), vitamin E supplement use (positive regression coefficients being significant only in fathers), and oral contraceptive use (negative regression coefficient being significant only in mothers).
The polymorphisms of apo E, apo C-III, apo B, CETP, and LPL were not significantly related to serum retinol concentrations when either crude or adjusted values for the covariates listed in Table 3 were used (data not shown). Apo E 4 and apo C-III 1100C alleles were significantly associated with crude values of serum -tocopherol concentrations (P 0.01), whereas a significant inverse association was noted for the apo E 2 allele (P 0.01). However, after adjustment for the covariates listed in Table 4, no polymorphism was significantly related to serum -tocopherol concentrations (data not shown).
The patterns of family correlations for serum retinol and tocopherol concentrations are given in Tables 5 and 6 for crude and adjusted values. For both serum retinol and -tocopherol concentrations, the general model that took into account covariates was the most parsimonious. When adjusted values were used, the 2 variables showed significant familial correlations by rejecting the hypothesis that no familial resemblance exists within the families (all P 0.001; data not shown). Significant correlations (P 0.01and P 0.001) were found for all the various pairs of relatives, except for son-son serum -tocopherol correlations.
View this table:
TABLE 5. Estimates of familial correlations for serum retinol concentrations1
View this table:
TABLE 6. Estimates of familial correlations for serum -tocopherol concentration1
The hypothesis of no difference in correlations for father-offspring and mother-offspring (FS = MS and FD = MD), but with different correlations according to child sex, was tested with submodel 1. Submodel 2 assumed no effect of child sex on family correlations (FS = FD, MS = MD and SS = SD = DD). Submodel 3 hypothesized no effect of sex on family correlations (FS = MS = MD = FD and SS = SD = DD). For both serum retinol and -tocopherol concentrations, the most adequate parsimonious model was submodel 3. Spouse, parent-offspring, and offspring-offspring correlation coefficients were 0.129, 0.289, and 0.368 for serum retinol, respectively, and 0.198, 0.307, and 0.243 for serum -tocopherol, respectively.
Quantitative genetic analyses were performed for serum retinol concentrations after adjustment for the covariates previously described in Table 3. The components of variance attributable to additive genetic effects, shared household effects, and residual environmental factors (including assay imprecision) are shown in Table 7. The full model 1, including the 3 components, cannot be reduced to a simpler model (submodels M2, M3, and M4 were rejected by log-likelihood tests). The proportion of phenotypic variability accounted for by household was smaller (about half) than that accounted for by genetic effects: 30.5% compared with 14.2%. Component attributable to residual environment factors represented more than half of the total variance (54.8%). More complex models hypothesizing effects of sex and generation on the 3 variance components (models 5 to 7) or an effect of the number of meals shared together at home (model 8) did not significantly improve the log-likelihood function.
View this table:
TABLE 7. Variance components of serum retinol concentrations1
Quantitative genetic analyses for serum -tocopherol concentrations after adjustment for the covariates described in Table 4 are shown in Table 8. As for serum retinol concentrations, the model giving the best description of the variance decomposition included the 3 components (model 1); heritability was lower and close to that attributable to household effects (22.1% compared with 18.7%), and the component attributable to residual environmental factors represented more than half of the total variance (59.1%). As for serum retinol concentrations, taking into account the relative level and the number of meals taken together in 2 class (<12 or 12 meals shared per week) did not improve significantly the likelihood function (models 5 to 8).
View this table:
TABLE 8. Variance components of serum -tocopherol concentrations1
DISCUSSION
No data are available regarding the inheritance of serum concentrations of retinol and -tocopherol, apart from results about genetic mutations of key enzymes or proteins involved in the regulation of these 2 fat-soluble vitamins in serum (8, 11). In our study, for both serum retinol and -tocopherol concentrations, spouse, parent-offspring, and sibling correlation coefficients were significant with no significant difference between fathers and mothers or between sons and daughters. Effects of lifestyle-related covariates previously reported in the literature (2932) did not substantially modify these family correlations. In addition, none of the candidate lipid-related polymorphisms were correlated with serum retinol and -tocopherol concentrations.
After the covariates were accounted for, our results documented the importance of the genetic effect: 30.5% and 22.1% of the total phenotypic variability for retinol and -tocopherol, respectively. In addition, for both vitamins, household effects were significantly different from zero: 14.2% for retinol and 18.7% for -tocopherol. Moreover, heritability and household effect estimates were not significantly different in the 4 classes of relatives (fathers, mothers, sons, and daughters) and when the family eats more meals at home.
Because our study was done on a random subsample of 383 families of the overall Stanislas population, conclusions drawn from this subsample should be valid for families living in the east of France with similar characteristics. Comparisons of our results with those of studies to come should take into account the characteristics of these French nuclear families. In fact, estimates of effects attributable to genes and to shared households depend on numerous factors: study design, type of related individuals included in the sample population (nuclear families or twins), inclusion of subjects with extreme values, distribution of environmental factors (eg, the range of vitamin intakes from foods or supplements), the covariates included in the model, and expression of variance components (proportion of the total variance or of residual variance after adjustment for covariates). In addition, variance component analysis requires strict assumptions, such as additive effects of multiple genes, environment factors, family environment, and cultural transmission (33). The assumptions of these models may have a marked effect on their results, particularly the tendency to overestimate heritability because of 3 main issues: 1) genotypic variance may include shared environmental variance that has not been removed by design or analysis, 2) estimates could greatly differ across populations according to the distribution of environmental and genetic factors, and 3) the assumption of independence between genotype and environment is likely to be violated when covariation and interaction are present.
The heritability estimate for serum retinol concentrations was higher than the shared household component: 30.5% compared with 14.2%, respectively. The lower household effect is in accordance with the fact that the amount of retinol in the circulating body pool is highly regulated and is essentially homeostatically controlled when liver stores are adequate (34) in populations in which vitamin A status was optimum. Except for supplement use, serum retinol concentrations were not related to the intake of either preformed vitamin A or provitamin A (29, 35, 36). In our study, family correlations for serum retinol concentrations were not altered when values were adjusted, especially for diet intake, and variance components did not vary significantly when families shared more meals at home. Heritability (additive genetic factors) could be due to numerous systems, proteins, and enzymes that are involved in retinol metabolism: 1) biliary acid synthesis, micelle formation, and solubilization of ingested fat, sterols, and fat-soluble vitamins (37, 38); 2) cleavage of provitamin A carotenoid to retinal by the cytosolic enzyme ß-carotene-15,15-dioxygenase (39); 3) retinol and retinal binding by cellular RBP type 2 and conversion of retinal into retinol by retinal reductase (40); 4) reesterification of retinol by lecithin:retinol acyltransferase (41, 42); and 5) carrying of retinol from its storage site in the liver to peripherical target tissues involving the RBP-TTR complex (8, 9, 43, 44). Concerning RBP, such major genetic defects (eg, 2 mutations in different exons) induce a severe biochemical vitamin A deficiency with low concentrations of plasma retinol and RBP, but a single mutation does not result in pathological symptoms (8).
Contrary to serum retinol concentrations, heritability and current household effects on serum -tocopherol concentrations were of similar magnitude: 22.1% and 18.7%, respectively. Moreover, family resemblance of serum -tocopherol concentration was not altered when covariates were taken into account and household component of variance was not improved when families shared more meals at home. In agreement, correlations between dietary and plasma or serum -tocopherol concentrations were low and consistent with those found in the literature (45, 46). Low correlations may be due, at least in part, to the relatively narrow range of vitamin E intake from foods found in this study, and genetic differences in absorption and metabolism may contribute to the poor correlation between dietary and plasma -tocopherol. The contribution of current household may result from shared household behavior not taken into account in our analysis. Although significant, the heritability of serum -tocopherol concentrations was weaker than for serum retinol and may have been due to numerous systems. The first candidate is -tocopherol transfer protein (-TTP), a liver protein implicated in the selective retention of -tocopherol from dietary vitamin E and for its transfer into nascent VLDL. This key phase is the major determinant of plasma -tocopherol concentrations in humans (12). Mutations in the -TTP gene have been detected in patients with low serum -tocopherol concentrations and ataxia with isolated vitamin E deficiency (11, 4750). In addition, any abnormality in chylomicron and VLDL synthesis could alter the transport of fat-soluble vitamins, eg, a defect of microsomal triacylglycerol transfer protein (5153).
In summary, our investigation showed for the first time that serum retinol and -tocopherol concentrations are under genetic control in healthy French families. Future studies are warranted to identify the key genetic variants that regulate the serum concentrations of these 2 vitamins in the physiologic state. Polymorphisms of genes related to proteins and enzymes involved in absorption, transport, and mobilization of retinol and -tocopherol could contribute to similarities in vitamin concentrations within families.
ACKNOWLEDGMENTS
We are deeply grateful for the cooperation of the families participating in the Stanislas Family Study. We acknowledge the management, reception, preclinical, laboratory, and medical staff of the Center for Preventive Medicine of Vandoeuvre-lès-Nancy (France). We especially thank Sylvie Péchiné for the collection of food intake data; Maryvonne Chaussard and Chantal Lafaurie for family recruitment; Sylvie Michel, Véronique Michaud, Line Grandcolas, and Dominique Aguillon for technical assistance with the biochemical assays; and Edith Lecomte for coordinating the field work. We also thank David Trégouët (INSERM U 525, Paris) for providing the program package of familial correlation computation.
SG and BH designed this specific study, performed the statistical analysis, and wrote the manuscript. PL was responsible for the laboratory analyses and helped write the manuscript. RG contributed to the statistical analysis, the interpretation of the data, and the writing of the manuscript. GS and SV are the principal investigators of the Stanislas Family Study and helped write the manuscript. None of the authors had any financial or personal conflict of interest.
REFERENCES