On different approximations to multilocus identity-by-descent calculations and the resulting power of variance component-based linkage analysis

An empirical comparison between three different methods for estimation of pair-wise identity-by-descent (IBD) sharing at marker loci was conducted in order to quantify the resulting differences in power and localization precision in variance components-based linkage analysis. On the examined simulated, error-free data set, it was found that an increase in accuracy of allele sharing calculation resulted in an increase in power to detect linkage. Linkage analysis based on approximate multi-marker IBD matrices computed by a Markov chain Monte Carlo approach was much more powerful than linkage analysis based on exact single-marker IBD probabilities. A "multiple two-point" approximation to true "multipoint" IBD computation was found to be roughly intermediate in power. Both multi-marker approaches were similar to each other in accuracy of localization of the quantitative trait locus and far superior to the single-marker approach. The overall conclusions of this study with respect to power are expected to also hold for different data structures and situations, even though the degree of superiority of one approach over another depends on the specific circumstances. It should be kept in mind, however, that an increase in computational accuracy is expected to go hand in hand with a decrease in robustness to various sources of errors.

Quantitative trait loci on chromosomes 2p, 4p, and 13q influence bone mineral density of the forearm and hip in Mexican Americans

We performed a genome scan using BMD data of the forearm and hip on 664 individuals in 29 Mexican-American families. We obtained evidence for QTL on chromosome 4p, affecting forearm BMD overall, and on chromosomes 2p and 13q, affecting hip BMD in men.

INTRODUCTION

The San Antonio Family Osteoporosis Study (SAFOS) was designed to identify genes and environmental factors that influence bone mineral density (BMD) using data from large Mexican-American families.

MATERIALS AND METHODS

We performed a genome-wide linkage analysis using 416 highly polymorphic microsatellite markers spaced approximately 9.5 cM apart to locate and identify quantitative trait loci (QTL) that affect BMD of the forearm and hip. Multipoint variance components linkage analyses were done using data on all 664 subjects, as well as two subgroups of 259 men and 261 premenopausal women, from 29 families for which genotypic and phenotypic data were available.

RESULTS

We obtained significant evidence for a QTL affecting forearm (radius midpoint) BMD in men and women combined on chromosome 4p near D4S2639 (maximum LOD = 4.33, genomic p = 0.006) and suggestive evidence for a QTL on chromosome 12q near locus D12S2070 (maximum conditional LOD = 2.35). We found suggestive evidence for a QTL influencing trochanter BMD on chromosome 6 (maximum LOD = 2.27), but no evidence for QTL affecting the femoral neck in men and women combined. In men, we obtained evidence for QTL affecting neck and trochanter BMD on chromosomes 2p near D2S1780 (maximum LOD = 3.98, genomic p = 0.013) and 13q near D13S788 (maximum LOD = 3.46, genomic p = 0.039), respectively. We found no evidence for QTL affecting forearm or hip BMD in premenopausal women.

CONCLUSION

These results provide strong evidence that a QTL on chromosome 4p affects radius BMD in Mexican-American men and women, as well as evidence that QTL on chromosomes 2p and 13q affect hip BMD in men. Our results are consistent with some reports in humans and mice. J Bone Miner Res 2003;18:2245-2252

Admixture effects in the traditional linkage analysis of admixed families

INTRODUCTION

Families of admixed ancestry are routinely excluded from traditional linkage analysis (LOD score) or are analyzed as deriving from a homogeneous population of the proband's ethnicity. Using traditional linkage analysis methods to analyze these families is complicated due to the admixture of different disease rates and allele frequencies that occurs. The presence of admixture violates the key assumptions of Hardy-Weinberg equilibrium (HWE) and linkage equilibrium (LE) invoked in the current methods of linkage analysis. If one or both of these assumptions are violated, incorrect inference for linkage could result.

DESIGN AND METHODS

We propose a pooling procedure to correct for inflated estimates of the recombination fraction that can result from admixture when performing traditional linkage analysis. Data were simulated with 30 families per each of 200 replications for a dominant, highly selective, linked disease locus model in order to conduct further testing for allelic LE and HWE, while using an allele pooling procedure to account for allele frequency differences between the 2 populations. The differences in allele frequencies between the populations for the polymorphic loci were 0.05, 0.10 and 0.14.

RESULTS AND CONCLUSIONS

Our pooling procedure does not eradicate all disequilibrium, because those replicates in which the disequilibrium exists are no longer affected by the disequilibrium in terms of maximization for linkage. Furthermore, our pooling procedure was able to exclude uninformative families or families far removed from HWE and/or LE that their LOD scores were unreliable.

The effect of pedigree complexity on quantitative trait linkage analysis

Due to the computational difficulties of performing linkage analysis on large complex pedigrees, most investigators resort to simplifying such pedigrees by some ad hoc strategy. In this paper, we suggest an analytical method to compare the power of various pedigree simplification schemes by using the asymptotic distribution of the likelihood-ratio statistic. We applied the method to the large Hutterine pedigree. Our results indicate that the breaking and reduction of inbreeding loops can greatly diminish the power to localize quantitative trait loci. We also present an efficient Monte Carlo method for estimating identity-by-descent allele sharing in large complex pedigrees. This method is used to facilitate a linkage analysis of serum IgE levels in the Hutterites without simplifying the pedigree.

GAW12: simulated genome scan, sequence, and family data for a common disease

The Genetic Analysis Workshop (GAW) 12 simulated data involves a common disease defined by imposing a threshold on a quantitative liability distribution. Associated with the disease are five quantitative risk factors, a quantitative environmental exposure, and a dichotomous environmental variable. Age at disease onset and household membership were also simulated. Genotype data, including 2,855 microsatellites on 22 autosomes, were simulated for 1,497 individuals in 23 families. Phenotype data and sequence data for seven candidate genes were provided for 1,000 of these individuals who were "living" and available for study. Data were simulated for 50 replicate samples in each of two populations, a general population and a population isolate formed from a small group of founders.

Factors of insulin resistance syndrome--related phenotypes are linked to genetic locations on chromosomes 6 and 7 in nondiabetic mexican-americans

Insulin resistance syndrome (IRS)-related phenotypes, such as hyperinsulinemia, obesity-related traits, impaired glucose tolerance, dyslipidemia, and hypertension, tend to cluster into factors. We attempted to identify loci influencing the factors of IRS-related phenotypes using phenotypic data from 261 nondiabetic subjects distributed across 27 low-income Mexican-American extended families. Principal component factor analyses were performed using eight IRS-related phenotypes: fasting glucose (FG), fasting specific insulin (FSI), BMI, systolic blood pressure (SBP), diastolic blood pressure (DBP), HDL cholesterol, ln triglycerides (ln TGs), and leptin (LEP). The factor analysis yielded three factors: factor 1 (BMI, LEP, and FSI), factor 2 (DBP and SBP), and factor 3 (HDL and ln TG). We conducted multipoint variance components linkage analyses on these factors with the program SOLAR using a 10--15 cM map. We found significant evidence for linkage of factor 1 to two regions on chromosome 6 near markers D6S403 (logarithm of odds [LOD] = 4.2) and D6S264 (LOD = 4.9). We also found strong evidence for linkage of factor 3 to a genetic location on chromosome 7 between markers D7S479 and D7S471 (LOD = 3.2). In conclusion, we found substantial evidence for susceptibility loci on chromosomes 6 and 7 that appear to influence the factors representing the IRS-related phenotypes in Mexican-Americans.

Linkage of high-density lipoprotein-cholesterol concentrations to a locus on chromosome 9p in Mexican Americans

High-density lipoproteins (HDLs) are anti-atherogenic lipoproteins that have a major role in transporting cholesterol from peripheral tissues to the liver, where it is removed. Epidemiologic studies have shown that low levels of high-density lipoprotein-cholesterol (HDL-C) are associated with an increased incidence of coronary heart disease and an increased mortality rate, indicating a protective role of high concentrations of HDL-C against atherogenesis and the development of coronary heart disease. HDL-C level is influenced by several genetic and nongenetic factors. Nongenetic factors include smoking, which has been shown to decrease the HDL-C level. Exercise and alcohol have been shown to increase HDL-C levels. Decreased HDL-C is often associated with other coronary heart disease risk factors such as obesity, hyperinsulinemia and insulin resistance, hypertriglyceridemia and hypertension. Although several genes have been identified for rare forms of dyslipidemia, the genes accounting for major variation in HDL-C levels have yet to be identified. Using a multipoint variance components linkage approach, we found strong evidence of linkage (lod score=3.4; P=0.00004) of a quantitative trait locus (QTL) for HDL-C level to a genetic location between markers D9S925 and D9S741 on chromosome 9p in Mexican Americans. A replication study in an independent set of Mexican American families confirmed the existence of a QTL on chromosome 9p.

A major locus for fasting insulin concentrations and insulin resistance on chromosome 6q with strong pleiotropic effects on obesity-related phenotypes in nondiabetic Mexican Americans

Insulin resistance and hyperinsulinemia are strong correlates of obesity and type 2 diabetes, but little is known about their genetic determinants. Using data on nondiabetics from Mexican American families and a multipoint linkage approach, we scanned the genome and identified a major locus near marker D6S403 for fasting "true" insulin levels (LOD score 4.1, empirical P<.0001), which do not crossreact with insulin precursors. Insulin resistance, as assessed by the homeostasis model using fasting glucose and specific insulin (FSI) values, was also strongly linked (LOD score 3.5, empirical P<.0001) with this region. Two other regions across the genome were found to be suggestively linked to FSI: a location on chromosome 2q, near marker D2S141, and another location on chromosome 6q, near marker D6S264. Since several insulin-resistance syndrome (IRS)-related phenotypes were mapped independently to the regions on chromosome 6q, we conducted bivariate multipoint linkage analyses to map the correlated IRS phenotypes. These analyses implicated the same chromosomal region near marker D6S403 (6q22-q23) as harboring a major gene with strong pleiotropic effects on obesity and on lipid measures, including leptin concentrations (e.g., LOD(eq) for traits-specific insulin and leptin was 4.7). A positional candidate gene for insulin resistance in this chromosomal region is the plasma cell-membrane glycoprotein PC-1 (6q22-q23). The genetic location on chromosome 6q, near marker D6S264 (6q25.2-q26), was also identified by the bivariate analysis as exerting significant pleiotropic influences on IRS-related phenotypes (e.g., LOD(eq) for traits-specific insulin and leptin was 4.1). This chromosomal region harbors positional candidate genes, such as the insulin-like growth factor 2 receptor (IGF2R, 6q26) and acetyl-CoA acetyltransferase 2 (ACAT2, 6q25.3-q26). In sum, we found substantial evidence for susceptibility loci on chromosome 6q that influence insulin concentrations and other IRS-related phenotypes in Mexican Americans.

Bivariate quantitative trait linkage analysis: pleiotropy versus co-incident linkages

Power to detect linkage and localization of a major gene were compared in univariate and bivariate variance components linkage analysis of three related quantitative traits in general pedigrees. Although both methods demonstrated adequate power to detect loci of moderate effect, bivariate analysis improved both power and localization for correlated quantitative traits mapping to the same chromosomal region, regardless of whether co-localization was the result of pleiotropy. Additionally, a test of pleiotropy versus co-incident linkage was shown to have adequate power and a low error rate.

GAW10: simulated family data for a common oligogenic disease with quantitative risk factors

GAW10 Problem 2 involves a simulated common disease defined by imposing a threshold, T, on a quantitative trait, Q1. Every individual with a value of Q1 > or = T (where T = 40) is defined as affected. Also thought to be associated with the disease as intervening variables are four other quantitative traits (Q2, Q3, Q4, and Q5) and an environmental factor (EF). Each individual has genotypes for 367 highly polymorphic markers on 10 chromosomes. The tasks for GAW10 were to characterize the genetic and environmental contribution to Q1 (and the disease), Q2, Q3, Q4, and Q5, and to localize any single genes that have detectable effects. Multiple replicates provided an opportunity to address questions of power and false positive rates.

Exploiting pleiotropy to map genes for oligogenic phenotypes using extended pedigree data

We investigated the utility of two approaches for exploiting pleiotropy to search for genes influencing related traits. To do this we first assessed the genetic correlations among a set of five closely related quantitative traits (Q1, Q2, Q3, Q4, Q5). We then used the genetic correlations among these five traits both to remove the common genetic effects of the four remaining traits, thereby identifying the unique genetic contribution to each trait, and to extract a synthetic phenotype which exploits the shared genetic information (pleiotropy) among these five traits. After obtaining these conditional traits, we then searched for evidence of quantitative trait loci (QTLs) (using variance component linkage) influencing the unique residual genetic component for each trait as well as those influencing the expression of the synthetic traits. From this work, we conclude that the removal of the common genetic effects of other traits in a group may be of greater utility when the majority of the pleiotropy initially detected between traits is attributable to the shared additive effects of polygenes, rather than to those of major loci. By contrast, decomposition of the genetic covariance matrix to its principal components is a greater utility when the majority of pleiotropy is attributable to major loci.

A major susceptibility locus influencing plasma triglyceride concentrations is located on chromosome 15q in Mexican Americans

Although several genetic forms of rare or syndromic hypertriglyceridemia have been reported, little is known about the specific chromosomal regions across the genome harboring susceptibility genes for common forms of hypertriglyceridemia. Therefore, we conducted a genomewide scan for susceptibility genes influencing plasma triglyceride (TG) levels in a Mexican American population. We used both phenotypic and genotypic data from 418 individuals distributed across 27 low-income, extended Mexican American families. For the analyses, TG values were log transformed (ln TG). We used a variance-components technique to conduct multipoint linkage analyses for localizing susceptibility genes that determine variation in TG levels. We used an approximately 10-15-cM map, which was made on the basis of information from 295 microsatellite markers. After accounting for the effects of sex and sex-specific age terms, we found significant evidence for linkage (LOD = 3.88) of ln TG levels to a genetic location between the markers GABRB3 and D15S165 on chromosome 15q. This putative locus explains 39.7+/-7% (P=.000012) of total phenotypic variation in ln TG levels. Suggestive evidence was found for linkage of ln TG levels to two different locations on chromosome 7, which are approximately 85 cM apart from each other. Also, there is some evidence for linkage of high-density lipoprotein cholesterol concentrations to a genetic location near one of the regions on chromosome 7. In conclusion, we found strong evidence for linkage of ln TG levels to a genetic location on chromosome 15q in a Mexican American population, which is prone to disease conditions such as type 2 diabetes and the insulin-resistance syndrome that are associated with hypertriglyceridemia. This putative locus appears to have a major influence on ln TG variation.

An empirical test of the significance of an observed quantitative trait locus effect that preserves additive genetic variation

We propose a constrained permutation test that assesses the significance of an observed quantitative trait locus effect against a background of genetic and environmental variation. Permutations of phenotypes are not selected at random, but rather are chosen in a manner that attempts to maintain the additive genetic variability in phenotypes. Such a constraint maintains the nonindependence among observations under the null hypothesis of no linkage. The empirical distribution of the lod scores calculated using permuted phenotypes is compared to that obtained using phenotypes simulated from the assumed underlying multivariate normal model. We make comparisons of univariate analyses for both a quantitative phenotype that appears consistent with a multivariate normal model and a quantitative phenotype containing pronounced outliers. An example of a bivariate analysis is also presented.

Quantitative trait locus detection using combined linkage/disequilibrium analysis

We describe an extension of the variance component linkage method that augments identity-by-descent information from relatives with identity-by-state information from unrelated individuals, exploiting disequilibrium to facilitate fine mapping of quantitative trait loci. An advantage of the combined linkage/disequilibrium test is that it detects association only in the presence of linkage and is not biased by population stratification.

Linkage of type 2 diabetes mellitus and of age at onset to a genetic location on chromosome 10q in Mexican Americans

Since little is known about chromosomal locations harboring type 2 diabetes-susceptibility genes, we conducted a genomewide scan for such genes in a Mexican American population. We used data from 27 low-income extended Mexican American pedigrees consisting of 440 individuals for whom genotypic data are available for 379 markers. We used a variance-components technique to conduct multipoint linkage analyses for two phenotypes: type 2 diabetes (a discrete trait) and age at onset of diabetes (a truncated quantitative trait). For the multipoint analyses, a subset of 295 markers was selected on the basis of optimal spacing and informativeness. We found significant evidence that a susceptibility locus near the marker D10S587 on chromosome 10q influences age at onset of diabetes (LOD score 3.75) and is also linked with type 2 diabetes itself (LOD score 2.88). This susceptibility locus explains 63.8%+/-9.9% (P=. 000016) of the total phenotypic variation in age at onset of diabetes and 65.7%+/-10.9% (P=.000135) of the total variation in liability to type 2 diabetes. Weaker evidence was found for linkage of diabetes and of age at onset to regions on chromosomes 3p, 4q, and 9p. In conclusion, our strongest evidence for linkage to both age at onset of diabetes and type 2 diabetes itself in the Mexican American population was for a region on chromosome 10q.

Characterization of a composite gradient gel for the electrophoretic separation of lipoproteins

We describe a protocol for making a new type of gradient gel, the Composite gradient gel, that was designed to resolve plasma lipoproteins using nondenaturing gradient gel electrophoresis. The new gel format allows analysis both of high density lipoproteins (HDLs) and low density lipoproteins (LDLs) on the same gel. The gel gave highly repeatable (r2 = 0.999) size estimates. We compared lipoprotein phenotypes determined from the new gradient gel with those obtained using specialized HDL and LDL gradient gels. The comparisons indicated that the Composite gel gave lipoprotein particle size estimates for HDLs and LDLs that were virtually identical to those obtained, respectively, from the specialized HDL and LDL gradient gels. We measured median diameters, which reflect the distributions of absorbance, for LDLs and for HDLs and found that the Composite gel gave lipoprotein size distributions that were virtually identical to those measured using the specialized LDL and HDL gels. Finally, comparison of fractional absorbance for six lipoprotein size intervals obtained from the Composite and specialized gels revealed a close correlation (r2 = 0.828). Thus, it appears that both LDL and HDL size phenotypes may be evaluated simultaneously using a single gradient gel format.

A major quantitative trait locus determining serum leptin levels and fat mass is located on human chromosome 2

Obesity is a major predisposing factor for the development of several chronic diseases including non-insulin dependent diabetes mellitus (NIDDM) and coronary heart disease (CHD). Leptin is a serum protein which is secreted by adipocytes and thought to play a role in the regulation of body fat. Leptin levels in humans have been found to be highly correlated with an individual's total adiposity. We performed a genome-wide scan and conducted multipoint linkage analysis using a general pedigree-based variance component approach to identify genes with measurable effects on quantitative variation in leptin levels in Mexican Americans. A microsatellite polymorphism, D2S1788, mapped to chromosome 2p21 (approximately 74 cM from the tip of the short arm) and showed strong evidence of linkage with serum leptin levels with a lod score of 4.95 (P = 9 x 10(-7)). This locus accounted for 47% of the variation in serum leptin levels, with a residual additive genetic component contributing an additional 24%. This region contains several potential candidate genes for obesity, including glucokinase regulatory protein (GCKR) and pro-opiomelanocortin (POMC). Our results show strong evidence of linkage of this region of chromosome 2 with serum leptin levels and indicate that this region could contain an important human obesity gene.

Genetic control of apolipoprotein A-I distribution among HDL subclasses

We conducted genetic analyses to determine the components of variation for size distributions of apolipoprotein (apo) A-I among human plasma lipoproteins resolved on the basis of size. Analyses used data for 717 individuals in 26 pedigrees. Apo A-I distributions among lipoprotein size classes were measured by nondenaturing gradient gel electrophoresis (GGE) and immunoblotting procedures. Curves were fitted to apo A-I absorbance profiles to estimate fractional absorbance in each of five high-density lipoprotein (HDL) subclasses. Multivariate regression analyses revealed several covariates (sex, age, diabetes, and apo A-I concentrations) that were significantly associated with variation in one or more HDL subclasses. Female gender and elevated apo A-I concentrations were associated with increases in proportion of apo A-I in larger HDLs, while increasing age and diabetes were associated with decreases. The analyses showed significant heritabilities. h2, for each variable representing the different HDL subclasses. Both genetic and nongenetic effects on apo A-I size distributions were generally exerted across the range of lipoprotein sizes, as suggested by high genetic and environmental correlations between HDL subclass variables. Decomposition of total overall variance showed that unidentified environmental factors accounted for 48% of variation in apo A-I size distribution, while genetic factors explained about 36% and the identified covariates explained the remaining 16%. When considered separately, apo A-I concentration explained only 5% of the total variation in apo A-I size distribution, indicating that apo A-I concentration is a poor predictor of apo A-I size distribution. In summary, the data suggest that there are significant genetic and environmental effects on apo A-I size distribution in humans, and that they are general metabolic effects rather than effects on specific HDL subclasses.

Effect of diabetes on lipoprotein size

The effects of diabetes on lipoprotein particle sizes were assessed using samples from 94 subjects with non-insulin-dependent diabetes mellitus. From a larger population of nondiabetic subjects who showed normal glucose tolerance, we selected an exact match in terms of age, sex, and menopausal status. We designed a protocol to make nondenaturing gradient gels for the resolution of LDL subfractions and generated two measures of LDL size: diameter of the predominant LDL species and proportion of LDL cholesterol (LDL-C) in particles larger than 25.5 nm (large LDL-C). Similarly, we made two measures of HDL size, large HDL cholesterol (HDL-C) and large HDL-apoAI, which represents the proportion of HDL-C and apoAI, respectively, occurring on particles larger than HDL-3. In pairwise comparisons, diabetes was associated with significantly (P < .004) smaller lipoprotein particles for all measures except large HDL-C. Each of the size measures was significantly and positively correlated with each of the others, suggesting that common metabolic mechanisms influence lipoprotein particle sizes across classes of lipoproteins. In addition, each of the size measures was correlated with a variety of measures of HDL and beta-lipoprotein concentrations, which included HDL-C, LDL-C, triglycerides, and apoAI, apoB, and apoE. We used stepwise regression analyses to select from the measures of lipoprotein concentrations those independently correlated with each of the lipoprotein size measures. After adjusting for these metabolic correlates of lipoprotein size measures, we found the effect of diabetes on lipoprotein size measures was no longer significant except for a modest effect (P = .027) on large HDL-apoAI.

Simulation of a common oligogenic disease with quantitative risk factors. GAW9 problem 2: the answers

For Problem 2, simulated family data were provided for a common oligogenic disease that was defined by imposing a threshold, T, on a quantitative trait, Q1. Every individual with a value of Q1 > T (where T = 87.5) was defined as affected. Participants were told that three other quantitative traits (Q2, Q3, and Q4) and an environmental factor (EF) also were thought to be associated with the disease as intervening variables, and that 12 candidate loci (C1 - C12) were postulated to influence traits associated with the disease. The tasks for GAW9 were to characterize the genetic and environmental contribution to Q1 (and the disease), Q2, Q3, and Q4, and to localize any single genes that had detectable effects.

Apolipoprotein B and a second major gene locus contribute to phenotypic variation of spontaneous hypercholesterolemia in pigs

The Lpb5 apolipoprotein B (apoB) allele occurs in pigs with spontaneous hypercholesterolemia. Low-density lipoprotein (LDL) from these pigs binds to the LDL receptor with a lower affinity and is cleared from the circulation more slowly than control pig LDL. However, the severity of hypercholesterolemia in pigs with the mutant apoB allele is highly variable. This study aimed to determine the metabolic basis for the phenotypic heterogeneity among Lpb5 pigs. Lpb5 pigs were divided into two groups: those with plasma cholesterol greater than 180 mg/dL (Lpb5.1) and those with plasma cholesterol less than 180 mg/dL (Lpb5.2). LDL from both Lpb5.1 and Lpb5.2 pigs was catabolized in vivo and in vitro at a similarly reduced rate. The difference in plasma cholesterol between the two phenotypic groups was in part due to a higher buoyant LDL production rate in Lpb5.1 pigs than in Lpb5.2 pigs. The in vivo LDL receptor status was evaluated by measuring the catabolism of LDL chemically modified to abrogate LDL receptor binding. Approximately 50% of LDL clearance in normal and Lpb5.2 pigs was via the LDL receptor; in Lpb5.1 pigs, 100% of LDL clearance was LDL receptor independent. Quantitative pedigree analysis of the segregation of the plasma cholesterol phenotype suggested that two major gene loci (the apoB locus and a second apparently unlinked locus) contribute to the determination of plasma cholesterol levels in this pig population.

Genetic analysis of apolipoprotein A-I in two dietary environments

Although of great clinical and biological importance, the role of genotype-diet interaction in lipoprotein metabolism and atherosclerosis is still poorly understood. We analyzed serum apolipoprotein A-I (apo A-I) concentrations of approximately 600 pedigreed baboons that were fed two dietary regimens: (1) a basal diet and (2) an atherogenic (high-cholesterol, saturated-fat) diet. Complex segregation analysis was performed separately for apo A-I concentrations in each dietary environment. A major locus model with a recessive allele for high levels of apo A-I and a polygenic component best fit the family data for both diets. Using bivariate segregation analysis, we showed that the major genes detected in the univariate analyses represent two distinct loci that act additively to determine apo A-I concentrations. These two loci accounted for approximately 40% of the total phenotypic variance in apo A-I levels in each dietary environment and were also responsible for 33% of the variation in apo A-I response to the atherogenic diet. Both major loci were influenced by genotype-diet interaction in which the two-locus genotypes exhibited heterogeneous responses to the atherogenic diet. Most genotypes responded to the atherogenic diet with an increase in apo A-I, but two genotypes showed a decrease that can be traced to the effect of one of the major loci. The presence of two major loci and genotype-diet interaction may be responsible for the equivocal results obtained in human pedigree studies of apo A-I.