Data-adaptive multi-locus association testing in subjects with arbitrary genealogical relationships.

Genome-wide sequencing enables evaluation of associations between traits and combinations of variants in genes and pathways. But such evaluation requires multi-locus association tests with good power, regardless of the variant and trait characteristics. And since analyzing families may yield more power than analyzing unrelated individuals, we need multi-locus tests applicable to both related and unrelated individuals. Here we describe such tests, and we introduce SKAT-X, a new test statistic that uses genome-wide data obtained from related or unrelated subjects to optimize power for the specific data at hand. Simulations show that: a) SKAT-X performs well regardless of variant and trait characteristics; and b) for binary traits, analyzing affected relatives brings more power than analyzing unrelated individuals, consistent with previous findings for single-locus tests. We illustrate the methods by application to rare unclassified missense variants in the tumor suppressor gene BRCA2, as applied to combined data from prostate cancer families and unrelated prostate cancer cases and controls in the Multi-ethnic Cohort (MEC). The methods can be implemented using open-source code for public use as the R-package GATARS (Genetic Association Tests for Arbitrarily Related Subjects) .

Assessing the goodness of fit of personal risk models.

We describe a flexible family of tests for evaluating the goodness of fit (calibration) of a pre-specified personal risk model to the outcomes observed in a longitudinal cohort. Such evaluation involves using the risk model to assign each subject an absolute risk of developing the outcome within a given time from cohort entry and comparing subjects' assigned risks with their observed outcomes. This comparison involves several issues. For example, subjects followed only for part of the risk period have unknown outcomes. Moreover, existing tests do not reveal the reasons for poor model fit when it occurs, which can reflect misspecification of the model's hazards for the competing risks of outcome development and death. To address these issues, we extend the model-specified hazards for outcome and death, and use score statistics to test the null hypothesis that the extensions are unnecessary. Simulated cohort data applied to risk models whose outcome and mortality hazards agreed and disagreed with those generating the data show that the tests are sensitive to poor model fit, provide insight into the reasons for poor fit, and accommodate a wide range of model misspecification. We illustrate the methods by examining the calibration of two breast cancer risk models as applied to a cohort of participants in the Breast Cancer Family Registry. The methods can be implemented using the Risk Model Assessment Program, an R package freely available at http://stanford.edu/~ggong/rmap/.

Estimating gene penetrance from family data.

Family data are useful for estimating disease risk in carriers of specific genotypes of a given gene (penetrance). Penetrance is frequently estimated assuming that relatives' phenotypes are independent, given their genotypes for the gene of interest. This assumption is unrealistic when multiple shared risk factors contribute to disease risk. In this setting, the phenotypes of relatives are correlated even after adjustment for the genotypes of any one gene (residual correlation). Many methods have been proposed to address this problem, but their performance has not been evaluated systematically. In simulations we generated genotypes for a rare (frequency 0.35%) allele of moderate penetrance, and a common (frequency 15%) allele of low penetrance, and then generated correlated disease survival times using the Clayton-Oakes copula model. We ascertained families using both population and clinic designs. We then compared the estimates of several methods to the optimal ones obtained from the model used to generate the data. We found that penetrance estimates for common low-risk genotypes were more robust to model misspecification than those for rare, moderate-risk genotypes. For the latter, penetrance estimates obtained ignoring residual disease correlation had large biases. Also biased were estimates based only on families that segregate the risk allele. In contrast, a method for accommodating phenotype correlation by assuming the presence of genetic heterogeneity performed nearly optimally, even when the survival data were coded as binary outcomes. We conclude that penetrance estimates that accommodate residual phenotype correlation (even only approximately) outperform those that ignore it, and that coding censored survival outcomes as binary does not substantially increase the mean-square error of the estimates, provided the censoring is not extensive.

Performance of prediction models for BRCA mutation carriage in three racial/ethnic groups: findings from the Northern California Breast Cancer Family Registry.

PURPOSE: Patients with early-onset breast and/or ovarian cancer frequently wish to know if they inherited a mutation in one of the cancer susceptibility genes, BRCA1 or BRCA2. Accurate carrier prediction models are needed to target costly testing. Two widely used models, BRCAPRO and BOADICEA, were developed using data from non-Hispanic Whites (NHW), but their accuracies have not been evaluated in other racial/ethnic populations. METHODS: We evaluated the BRCAPRO and BOADICEA models in a population-based series of African American, Hispanic, and NHW breast cancer patients tested for BRCA1 and BRCA2 mutations. We assessed model calibration by evaluating observed versus predicted mutations and attribute diagrams, and model discrimination using areas under the receiver operating characteristic curves. RESULTS: Both models were well-calibrated within each racial/ethnic group, with some exceptions. BOADICEA overpredicted mutations in African Americans and older NHWs, and BRCAPRO underpredicted in Hispanics. In all racial/ethnic groups, the models overpredicted in cases whose personal and family histories indicated >80% probability of carriage. The two models showed similar discrimination in each racial/ethnic group, discriminating least well in Hispanics. For example, BRCAPRO's areas under the receiver operating characteristic curves were 83% (95% confidence interval, 63-93%) for NHWs, compared with 74% (59-85%) for African Americans and 58% (45-70%) for Hispanics. CONCLUSIONS: The poor performance of the model for Hispanics may be due to model misspecification in this racial/ethnic group. However, it may also reflect racial/ethnic differences in the distributions of personal and family histories among breast cancer cases in the Northern California population.

Prevalence of pathogenic BRCA1 mutation carriers in 5 US racial/ethnic groups.

CONTEXT: Information on the prevalence of pathogenic BRCA1 mutation carriers in racial/ethnic minority populations is limited. OBJECTIVE: To estimate BRCA1 carrier prevalence in Hispanic, African American, and Asian American female breast cancer patients compared with non-Hispanic white patients with and without Ashkenazi Jewish ancestry. DESIGN, SETTING, AND PARTICIPANTS: We estimated race/ethnicity-specific prevalence of BRCA1 in a population-based, multiethnic series of female breast cancer patients younger than 65 years at diagnosis who were enrolled at the Northern California site of the Breast Cancer Family Registry during the period 1996-2005. Race/ethnicity and religious ancestry were based on self-report. Weighted estimates of prevalence and 95% confidence intervals (CIs) were based on Horvitz-Thompson estimating equations. MAIN OUTCOME MEASURE: Estimates of BRCA1 prevalence. RESULTS: Estimates of BRCA1 prevalence were 3.5% (95% CI, 2.1%-5.8%) in Hispanic patients (n = 393), 1.3% (95% CI, 0.6%-2.6%) in African American patients (n = 341), and 0.5% (95% CI, 0.1%-2.0%) in Asian American patients (n = 444), compared with 8.3% (95% CI, 3.1%-20.1%) in Ashkenazi Jewish patients (n = 41) and 2.2% (95% CI, 0.7%-6.9%) in other non-Hispanic white patients (n = 508). Prevalence was particularly high in young (<35 years) African American patients (5/30 patients [16.7%]; 95% CI, 7.1%-34.3%). 185delAG was the most common mutation in Hispanics, found in 5 of 21 carriers (24%). CONCLUSIONS: Among African American, Asian American, and Hispanic patients in the Northern California Breast Cancer Family Registry, the prevalence of BRCA1 mutation carriers was highest in Hispanics and lowest in Asian Americans. The higher carrier prevalence in Hispanics may reflect the presence of unrecognized Jewish ancestry in this population.

Prevalence of BRCA1 mutation carriers among U.S. non-Hispanic Whites.

Data from several countries indicate that 1% to 2% of Ashkenazi Jews carry a pathogenic ancestral mutation of the tumor suppressor gene BRCA1. However, the prevalence of BRCA1 mutations among non-Ashkenazi Whites is uncertain. We estimated mutation carrier prevalence in U.S. non-Hispanic Whites, specific for Ashkenazi status, using data from two population-based series of San Francisco Bay Area patients with invasive cancers of the breast or ovary, and data on breast and ovarian cancer risks in Ashkenazi and non-Ashkenazi carriers. Assuming that 90% of the BRCA1 mutations were detected, we estimate a carrier prevalence of 0.24% (95% confidence interval, 0.15-0.39%) in non-Ashkenazi Whites, and 1.2% (95% confidence interval, 0.5-2.6%) in Ashkenazim. When combined with U.S. White census counts, these prevalence estimates suggest that approximately 550,513 U.S. Whites (506,206 non-Ashkenazim and 44,307 Ashkenazim) carry germ line BRCA1 mutations. These estimates may be useful in guiding resource allocation for genetic testing and genetic counseling and in planning preventive interventions.

Breast cancer risk for noncarriers of family-specific BRCA1 and BRCA2 mutations: findings from the Breast Cancer Family Registry.

PURPOSE: Women with germline BRCA1 and BRCA2 mutations have five- to 20-fold increased risks of developing breast and ovarian cancer. A recent study claimed that women testing negative for their family-specific BRCA1 or BRCA2 mutation (noncarriers) have a five-fold increased risk of breast cancer. We estimated breast cancer risks for noncarriers by using a population-based sample of patients with breast cancer and their female first-degree relatives (FDRs). PATIENTS AND METHODS: Patients were women with breast cancer and their FDRs enrolled in the population-based component of the Breast Cancer Family Registry; patients with breast cancer were tested for BRCA1 and BRCA2 mutations, as were FDRs of identified mutation carriers. We used segregation analysis to fit a model that accommodates familial correlation in breast cancer risk due to unobserved shared risk factors. RESULTS: We studied 3,047 families; 160 had BRCA1 and 132 had BRCA2 mutations. There was no evidence of increased breast cancer risk for noncarriers of identified mutations compared with FDRs from families without BRCA1 or BRCA2 mutations: relative risk was 0.39 (95% CI, 0.04 to 3.81). Residual breast cancer correlation within families was strong, suggesting substantial risk heterogeneity in women without BRCA1 or BRCA2 mutations, with some 3.4% of them accounting for roughly one third of breast cancer cases. CONCLUSION: These results support the practice of advising noncarriers that they do not have any increase in breast cancer risk attributable to the family-specific BRCA1 or BRCA2 mutation.

Performance of BRCA1/2 mutation prediction models in Asian Americans.

PURPOSE: There are established differences in breast cancer epidemiology between Asian and white individuals, but little is known about hereditary breast cancer in Asian populations. Although increasing numbers of Asian individuals are clinically tested for BRCA1/2 mutations, it is not known whether computer models that predict mutations work accurately in Asian individuals. We compared the performance in Asian and white individuals of two widely used BRCA1/2 mutation prediction models, BRCAPRO and Myriad II. PATIENTS AND METHODS: We evaluated BRCAPRO and Myriad II in 200 Asian individuals and a matched control group of 200 white individuals who were tested for BRCA1/2 mutations at four cancer genetics clinics, by comparing numbers of observed versus predicted mutation carriers and by evaluating area under the receiver operating characteristic curve (AUC) for each model. RESULTS: BRCAPRO and Myriad II accurately predicted the number of white BRCA1/2 mutation carriers (25 observed v 24 predicted by BRCAPRO; 25 predicted by Myriad II, P > or = .69), but underpredicted Asian carriers by two-fold (49 observed v 25 predicted by BRCAPRO; 26 predicted by Myriad II; P < or = 3 x 10(-7)). For BRCAPRO, this racial difference reflects substantial underprediction of Asian BRCA2 mutation carriers (26 observed v 4 predicted; P = 1 x 10(-30)); for Myriad II, separate mutation predictions were not available. For both models, AUCs were nonsignificantly lower in Asian than white individuals, suggesting less accurate discrimination between Asian carriers and noncarriers. CONCLUSION: Both BRCAPRO and Myriad II underestimated the proportion of BRCA1/2 mutation carriers, and discriminated carriers from noncarriers less well, in Asian compared with white individuals.

Optimal designs for estimating penetrance of rare mutations of a disease-susceptibility gene.

Many clinical decisions require accurate estimates of disease risks associated with mutations of known disease-susceptibility genes. Such risk estimation is difficult when the mutations are rare. We used computer simulations to compare the performance of estimates obtained from two types of designs based on family data. In the first (clinic-based designs), families are ascertained because they meet certain criteria concerning multiple disease occurrences among family members. In the second (population-based designs), families are sampled through a population-based registry of affected individuals called probands, with oversampling of probands whose families are more likely to segregate mutations. We generated family structures, genotypes, and phenotypes using models that reflect the frequencies and penetrances of mutations of the BRCA1/2 genes. We studied the effects of risk heterogeneity due to unmeasured, shared risk factors by including risk variation due to unmeasured genotypes of another gene. The simulations were chosen to mimic the ascertainment and selection processes commonly used in the two types of designs. We found that penetrance estimates from both designs are nearly unbiased in the absence of unmeasured shared risk factors, but are biased upward in the presence of such factors. The bias increases with increasing variation in risks across genotypes of the second gene. However, it is small compared to the standard error of the estimates. Standard errors from population-based designs are roughly twice those from clinic-based designs with the same number of families. Using the root-mean-square error as a measure of performance, we found that in all instances, the clinic-based designs gave more accurate estimates than did the population-based designs with the same numbers of families. Rough variance calculations suggest that clinic-based designs give more accurate estimates because they include more identified mutation carriers.

Segregation analysis of prostate cancer in 1,719 white, African-American and Asian-American families in the United States and Canada.

UNLABELLED: Some data suggest that brothers of prostate cancer patients have higher disease risk than their fathers, supporting an X-linked or recessive mode of inheritance. However, higher observed frequencies in brothers than fathers may merely reflect the strong temporal changes in US incidence rates. OBJECTIVES: (a) to evaluate the fit of X-linked, recessive, and dominant modes of inheritance to prostate cancer incidence, specific for calendar year, age, and race, in population-based samples of US and Canadian families; and (b) to evaluate a simple multifactorial model for familial aggregation of prostate cancer due to shared low-penetrance variants of many genes or shared lifestyle factors. METHODS: The data consist of reported prostate cancer incidence in first-degree relatives of 1,719 white, African-American, and Asian-American men with and without prostate cancer at ages <70 years. Model parameters were estimated by maximizing a pseudo-likelihood function of the data, and goodness of model fit was assessed by evaluating discrepancies between observed and expected numbers of pairs of relatives with prostate cancer. RESULTS: After adjusting for temporal trends in prostate cancer incidence rates we found that the X-linked model fit poorly. underpredicting the observed number of affected father-son pairs. This also was true of the recessive model, although the evidence for poor fit did not achieve statistical significance. In contrast, the dominant model provided adequate fit to the data. In this model the race-specific penetrance estimates for carriers of deleterious genotypes were similar among African-Americans and whites, but lower among Asian-Americans: risk by age 80 years for carriers born in 1900 was estimated as 75.3% for African-Americans and whites, and 44.4% for Asian-Americans. None of the Mendelian models fit the data better than did the simple multifactorial model. CONCLUSIONS: The good fit of the multifactorial model suggests that multiple genes, each having low penetrance, may be responsible for most inherited prostate cancer susceptibility, and that the contribution of rare highly penetrant mutations is small.