Familial Aggregation Of Disease Research Articles

Towards more comprehensive nationwide familial aggregation studies in Denmark: The Danish Civil Registration System versus the lite Danish Multi-Generation Register.

Linking information on family members in the Danish Civil Registration System (CRS) with information in Danish national registers provides unique possibilities for research on familial aggregation of diseases, health patterns, social factors and demography. However, the CRS is limited in the number of generations that it can identify. To allow more complete familial linkages, we introduce the lite Danish Multi-Generation Register (lite MGR) and the future full Danish MGR that is currently being developed. We generated the lite MGR by linking the current version of the CRS with historical versions stored by the Danish National Archives in the early 1970s, which contain familial links not saved in the current CRS. We describe and compare the completeness of familial links in the lite MGR and the current version of the CRS. We also describe planned procedures for generating the full MGR by linking the current CRS with scanned archived records from Parish Registers. Among people born in Denmark in 1960 or later, the current CRS contains information on both parents. However, it has limited parental information for people born earlier. Among the 732,232 people born in Denmark during 1950-1959, 444,084 (60.65%) had information on both parents in the CRS. In the lite MGR, it was 560,594 (76.56%). The lite MGR offers more complete information on familial relationships than the current CRS. The lite and full MGR will offer an infrastructure tying together existing research infrastructures, registers and biobanks, raising their joint research value to an unparalleled level.

P901 A genetic analysis of familial aggregation in Inflammatory Bowel Disease multiplex families

Abstract Background Relatives of patients with inflammatory bowel disease (IBD) have a higher risk to develop IBD than the general population. In some families, even a remarkably high prevalence of disease is found. Here, we aim to determine the role of the known IBD risk loci in familial aggregation of disease, as well as to study if there are common variants specific to familial IBD. Methods Imputed immunochip genotypes of 55 families [157 Crohn’s disease (CD), 29 ulcerative colitis (UC) and 138 unaffected relatives] that have at least three affected, first-degree relatives were used. A sporadic dataset (1705 CD, 917 UC, 873 controls) was used for comparison. For each individual, a polygenic risk score (PRS) was calculated with PRSice based on the IBD summary statistics of de Lange et al (2017), a p-value threshold ≤0.01 and MAF≥0.01. The mean PRS of all relatives of one family is denoted the familial PRS. Family-based association analyses were performed with SAIGE. The significance threshold for association used was p<1e-4. Results As a group, unaffected relatives have a lower PRS than affected relatives (p=1.36e-5, β(se)=-0.60(0.14)), but still higher than unrelated controls (p=2.07e-4, β(se)=0.39(0.11))(fig 1). Twenty-four families have a familial PRS that is higher than the mean PRS of sporadic cases, and seven families have a familial PRS that is lower than the mean PRS of unrelated controls, the low PRS families (fig 2A). In most families, the affected members have a higher mean PRS than the unaffected members (fig 2B). This relationship however is sometimes reversed, especially in the low PRS families. SAIGE found nine variants to be associated with familial IBD (fig 3A, table 1). Two independently associated variants, rs2241130 in IL1RL2 (p=1.19 e-5, β(se)=1.21(0.28)) and rs144641193 in IL1RL1 (p=8.23e-5, β(se)=1.29(0.33)) reside in a known IBD locus (IL18RAP). Some other identified variants were also located in genes previously implicated in IBD, either through GWAS [rs72781786 in PRKCQ (p=2.93e-5, β(se)= -0.95(0.23))], or via functional studies (rs2242601 in EPHA1 and rs2272766 in CTSB). A restricted analysis of the low PRS families identified only one significant variant: rs11579543, located between RPE65 and DEPDC1 (p=6.12e-5, β(se)=-2.71(0.68))(fig 3B, table 1). Conclusion Our analysis indicates that common risk variants play an important role in familial aggregation of disease in many multiplex families. Some families however have a very low polygenic risk, indicating shared environmental factors might be relatively more important, or genetic factors not captured by the score, e.g. rare variants, may have contributed. Also, a few loci are specifically implicated in familial IBD, some related to the immune system, e.g. CLECL1 and CD69.

Contribution of Common Genetic Variants to Familial Aggregation of Disease and Implications for Sequencing Studies.

Despite genetics being accepted as the primary cause of familial aggregation for most diseases, it is still unclear whether afflicted families are likely to share a single highly penetrant rare variant, many minimally penetrant common variants, or a combination of the two types of variants. We therefore use recent estimates of SNP heritability and the liability threshold model to estimate the proportion of afflicted families likely to carry a rare, causal variant. We then show that Polygenic Risk Scores (PRS) may be useful for identifying families likely to carry such a rare variant and therefore for prioritizing families to include in sequencing studies with that aim. Specifically, we introduce a new statistic that estimates the proportion of individuals carrying causal rare variants based on the family structure, disease pattern, and PRS of genotyped individuals. Finally, we consider data from the MelaNostrum consortium and show that, despite an estimated PRS heritability of only 0.05 for melanoma, families carrying putative causal variants had a statistically significantly lower PRS, supporting the idea that PRS prioritization may be a useful future tool. However, it will be important to evaluate whether the presence of rare mendelian variants are generally associated with the proposed test statistic or lower PRS in future and larger studies.

Commentary: Frailty and cancer.

of Twin Registries (INTR): building a platform for international collaboration. Twin Res Hum Genet 2014;17:574–77. 3. McRae AF, Powell JE, Henders AK et al. Contribution of genetic variation to transgenerational inheritance of DNA methylation. Genome Biol 2014;15:R73. 4. Hopper JL. Disease-specific prospective family study cohorts enriched for familial risk. Epidemiol Perspect Innov 2011;8:2. 5. Aalen OO. Modelling the influence of risk factors on familial aggregation of disease. Biometrics 1991;47:933–45. 6. Dowty JG, Win AK, Buchanan DD et al. Cancer risks for MLH1 and MSH2 mutation carriers. Hum Mutat 2013;34: 490–97. 7. Win AK, Hopper JL, Buchanan DD et al. Are the common genetic variants associated with colorectal cancer risk for DNA mismatch repair gene mutation carriers? Eur J Cancer 2013;49:1578–87. 8. Tomasetti C, Vogelstein B. Variation in cancer risk among tissues can be explained by the number of cell divisions. Science 2015;347:78–81. 9. Mathews JD, Forsythe AV, Brady Z et al. Cancer risk in 680,000 people exposed to computed tomography scans in childhood or adolescence: data linkage study of 11 million Australians. BMJ 2013;346:f2360. 10. Pearce DC, Pallaghy PK, McCaw JM, McVernon J, Mathews JD. Understanding mortality in the 1918-1919 influenza pandemic in England and Wales. Influenza Other Respir Viruses 2011;5:89–98.

The familial risk of autism.

Autism spectrum disorder (ASD) aggregates in families, but the individual risk and to what extent this is caused by genetic factors or shared or nonshared environmental factors remains unresolved. To provide estimates of familial aggregation and heritability of ASD. A population-based cohort including 2,049,973 Swedish children born 1982 through 2006. We identified 37,570 twin pairs, 2,642,064 full sibling pairs, 432,281 maternal and 445,531 paternal half sibling pairs, and 5,799,875 cousin pairs. Diagnoses of ASD to December 31, 2009 were ascertained. The relative recurrence risk (RRR) measures familial aggregation of disease. The RRR is the relative risk of autism in a participant with a sibling or cousin who has the diagnosis (exposed) compared with the risk in a participant with no diagnosed family member (unexposed). We calculated RRR for both ASD and autistic disorder adjusting for age, birth year, sex, parental psychiatric history, and parental age. We estimated how much of the probability of developing ASD can be related to genetic (additive and dominant) and environmental (shared and nonshared) factors. In the sample, 14,516 children were diagnosed with ASD, of whom 5689 had autistic disorder. The RRR and rate per 100,000 person-years for ASD among monozygotic twins was estimated to be 153.0 (95% CI, 56.7-412.8; rate, 6274 for exposed vs 27 for unexposed ); for dizygotic twins, 8.2 (95% CI, 3.7-18.1; rate, 805 for exposed vs 55 for unexposed); for full siblings, 10.3 (95% CI, 9.4-11.3; rate, 829 for exposed vs 49 for unexposed); for maternal half siblings, 3.3 (95% CI, 2.6-4.2; rate, 492 for exposed vs 94 for unexposed); for paternal half siblings, 2.9 (95% CI, 2.2-3.7; rate, 371 for exposed vs 85 for unexposed); and for cousins, 2.0 (95% CI, 1.8-2.2; rate, 155 for exposed vs 49 for unexposed). The RRR pattern was similar for autistic disorder but of slightly higher magnitude.We found support for a disease etiology including only additive genetic and nonshared environmental effects. The ASD heritability was estimated to be 0.50 (95% CI, 0.45-0.56) and the autistic disorder heritability was estimated to 0.54 (95% CI, 0.44-0.64). Among children born in Sweden, the individual risk of ASD and autistic disorder increased with increasing genetic relatedness. Heritability of ASD and autistic disorder were estimated to be approximately 50%. These findings may inform the counseling of families with affected children.

A unified model for estimating and testing familial aggregation

Investigations of familial aggregation of disease can provide important clues for genetic mechanisms, and many such studies have been published in the epidemiological literature using various statistical methods. We developed a unified model for familial risk by extending a Cox regression model to enable estimation of the detailed effects of kinship. By appropriate parameterisation of the model, we show how the risks to all specific first-degree kinships can be estimated and formally compared using simple interaction terms and how the model can be extended to accommodate higher-degree relatives. The correlation due to observations from family members and from the potential for repeated observations is accommodated by a robust sandwich variance estimator or a bootstrap estimate. Hazard ratios for different kinships are formally compared using a robust Wald test. We illustrate the method with applications to studies of adult leukemia and non-Hodgkin's lymphoma in the Swedish population and display our results on a pedigree diagram. Our estimates are consistent with published work that used simpler stratified methods, and our model enabled the detection of a number of statistically significant effects of kinship. The recognition of such kindred-specific disease risk could be a first step in the design of more informative genetic biomarker studies.

Common variants of ZNF750, RPTOR and TRAF3IP2 genes and psoriasis risk

Psoriasis vulgaris is a genetically heterogenous disease with unclear molecular background. We assessed the association of psoriasis and its main clinical phenotypes with common variants of three potential psoriasis susceptibility genes: ZNF750, RPTOR and TRAF31P2. We genotyped 10 common variants in a cohort of 1,034 case–control individuals using Taqman genotyping assays and sequencing. Minor alleles of all four TRAF3IP2 variants were more frequent among cases. The strongest, significant association was observed for rs33980500 (OR = 2.5, p = 0.01790). Minor allele of this SNP was always present in two haplotypes found to be associated with increased psoriasis risk: rs13196377_G + rs13190932_G + rs33980500_T + rs13210247_A (OR = 2.7, p = 0.0054) and rs13196377_A + rs13190932_A + rs33980500_T + rs13210247_G (OR = 1.8, p = 0.0008). Analyses of clinically relevant phenotypes revealed association of rs33980500 with pustular psoriasis (OR = 1.2, p = 0.0109). We observed significant connection of severity of cutaneous disease with variation at rs13190932 and suggestive with three remaining TRAF3IP2 SNPs. Another positive associations were found between age of onset and familial aggregation of disease: smoking and younger age of onset, smoking and occurrence of pustular psoriasis, nail involvement and arthropatic psoriasis, nail involvement and more severe course of psoriasis. We found no statistically significant differences in the prevalence of the examined variants of RPTOR and ZNF750 genes among our cases and controls. We have replicated the association of TRAF3IP2-_rs33980500 variant with the susceptibility to psoriasis. We have found new associations with clinically relevant subphenotypes such as pustular psoriasis or moderate-to-severe cases. We ascertain no connection of RPTOR and ZNF750 variants with psoriasis or its subphenotypes.Electronic supplementary materialThe online version of this article (doi:10.1007/s00403-013-1407-9) contains supplementary material, which is available to authorized users.

The heritability of human disease: estimation, uses and abuses

Relatives provide the basic material for the study of inheritance of human disease. However, the methodologies for the estimation of heritability and the interpretation of the results have been controversial. The debate arises from the plethora of methods used, the validity of the methodological assumptions and the inconsistent and sometimes erroneous genetic interpretations made. We will discuss how to estimate disease heritability, how to interpret it, how biases in heritability estimates arise and how heritability relates to other measures of familial disease aggregation.

Expansion mutation in C9ORF72 does not influence plasma progranulin levels in frontotemporal dementia

A hexanucleotide repeat expansion in chromosome 9 open reading frame 72 (C9ORF72) gene has recently been described as a cause of familial and sporadic frontotemporal lobar degeneration. The aim of this study was to assess whether plasma progranulin (GRN) levels could be modulated by the presence of this repeat expansion. Sixty-five patients diagnosed with frontotemporal dementia and 10 family members with familial aggregation of disease were screened for the presence of the hexanucleotide repeat expansion in C9ORF72 gene, using a repeat-primed polymerase chain reaction method. Plasma GRN levels were measured in all subjects through enzyme-linked immunosorbent assay. Seven individuals with the repeat expansion were identified. No differences were found between C9ORF72 repeat expansion carriers and noncarriers (116.4 ± 21 ng/mL and 131.7 ± 36 ng/mL, respectively, p = 0.3). Analysis of family members did not disclose any difference in plasma GRN levels between carriers and noncarriers. In conclusion, plasma GRN levels are not influenced by the hexanucleotide repeat expansion in C9ORF72 gene, and therefore, cannot be used as a reliable biomarker to detect mutation carriers.

Using linkage analysis of large pedigrees to guide association analyses

To date, genome-wide association studies have yielded discoveries of common variants that partly explain familial aggregation of diseases and traits. Researchers are now turning their attention to less common variants because the price of sequencing has dropped drastically. However, because sequencing of the whole genome in large samples is costly, great care must be taken to prioritize which samples and which genomic regions are selected for sequencing. We are interested in identifying genomic regions for deep sequencing using large multiplex families collected as part of earlier linkage studies. We incorporate linkage analysis into our search for Q1-associated alleles. Overall, we found that power was low for both whole-exome and linkage-guided sequencing analysis. By restricting sequencing to regions with high LOD peaks, we found fewer associated single-nucleotide polymorphisms than by using whole-exome sequencing. However, incorporating linkage analysis enabled us to detect more than half of the associated susceptibility loci (52%) that would have been identified by whole-exome sequencing while examining only 2.5% of the exome. This result suggests that incorporating linkage results from large multiplex families might greatly increase the efficiency of sequencing to detect trait-associated alleles in complex disease.

Studies based on the Danish Adoption Register: Schizophrenia, BMI, smoking, and mortality in perspective

The genetic and familial environmental influences making up familial correlations in traits and familial aggregation of diseases may be disentangled by adoption studies. Thus, resemblance between the adoptee and the biological relatives indicates a contribution of the segregating gene they have in common, and resemblance between the adoptee and the rearing adoptive family members indicates influences of the shared family environment. The Danish Adoption Register includes information on all 14,425 non-familial adoptions of Danish children legally granted in Denmark from 1924 to 1947. To illustrate the adoption study approach in research, we present results achieved in schizophrenia, body fatness, smoking, and mortality. Significantly higher prevalence of schizophrenia and related conditions were found in biological relatives to adoptees with schizophrenia than among biological relatives to adoptees without schizophrenia. Clear trends in body mass index of biological relatives and the weight of adoptees were found in several settings. A genetic influence on smoking habits were found within a generation. A moderate genetic influence on the mortality rate has been found. The Danish Adoption Register has contributed to the knowledge about environmental and genetic influence in various fields. Genetic influence on the risk of schizophrenia, body fatness, smoking habits, and on the mortality rate has been found.

Quantification of the familial contribution to juvenile idiopathic arthritis

We previously demonstrated that there is familial aggregation of juvenile idiopathic arthritis (JIA). Using a large JIA cohort, we sought to identify additional clusters of JIA cases and to calculate robust estimates of the relative risk (RR) of JIA in the siblings and cousins of JIA probands. We also estimated the population attributable risk (PAR) of familial factors in JIA. A probabilistic record-linking analysis was performed by matching the records of 862 patients with JIA with the records of approximately 7 million individuals in the Utah Population Database (UPDB), a computerized genealogic database. For each patient, 5 control subjects matched for birth year and sex were selected from the UPDB. Specialized software was used to test for familial aggregation of disease, to estimate the magnitude of familial risks, and to identify families at high risk of disease. We identified 22 founders who had significantly more descendants with JIA than expected (5-13 descendants; P values ranged from <0.0001 to <0.008). The PAR of familial factors for JIA was approximately 13%. The RR of JIA in the siblings of patients was significantly increased (11.6, 95% confidence interval [95% CI] 4.9-27.5, P < 2.59 x 10(-8)). The RR of JIA in first cousins was also increased (5.82, 95% CI 2.5-13.8, P < 6.07 x 10(-5)). We have identified the largest sets of JIA pedigrees described to date. Approximately 13% of cases of JIA can be attributed to familial factors. Siblings and first cousins of probands with JIA have an increased risk of JIA. The observed decline in the magnitude of risk between siblings and cousins suggests that JIA is influenced by shared genetic factors.

Estimating Disease Prevalence Using Relatives of Case and Control Probands

We introduce a method of estimating disease prevalence from case-control family study data. Case-control family studies are performed to investigate the familial aggregation of disease; families are sampled via either a case or a control proband, and the resulting data contain information on disease status and covariates for the probands and their relatives. Here, we introduce estimators for overall prevalence and for covariate-stratum-specific (e.g., sex-specific) prevalence. These estimators combine the proportion of affected relatives of control probands with the proportion of affected relatives of case probands and are designed to yield approximately unbiased estimates of their population counterparts under certain commonly made assumptions. We also introduce corresponding confidence intervals designed to have good coverage properties even for small prevalences. Next, we describe simulation experiments where our estimators and intervals were applied to case-control family data sampled from fictional populations with various levels of familial aggregation. At all aggregation levels, the resulting estimates varied closely and symmetrically around their population counterparts, and the resulting intervals had good coverage properties, even for small sample sizes. Finally, we discuss the assumptions required for our estimators to be approximately unbiased, highlighting situations where an alternative estimator based only on relatives of control probands may perform better.

Methods for Assessing Familial Aggregation: Family History Measures and Confounding in the Standard Cohort, Reconstructed Cohort and Case-Control Designs

Objectives: To test whether case-control-based familial aggregation studies produce estimates of risk to relatives that are inherently biased or confounded by age and family size, and to compare case-control-derived estimates with those from the reconstructed cohort method. In addition, we test if the definition of family history affects the accuracy of results obtained from either design. We use simulated data, which allows us to know the true data origin. Methods: We simulated populations of three generation families. Both a dominant genetic disease and a non-genetic disease were present in the population. We compared the effect estimates from different measures of family history with those derived from the actual genetic cause of disease. Results: Effect estimates from family history measures that used multiple family members were more accurate than those derived from measures based on a single relative. Neither family size nor age of family members defining family history were confounders in the case-control design. Conclusion: The case-control and reconstructed cohort designs are equally valid in assessing familial aggregation of disease.

Familial aggregation of age-related macular degeneration in the Utah population

We examined familial aggregation and risk of age-related macular degeneration in the Utah population using a population-based case-control study. Over one million unique patient records were searched within the University of Utah Health Sciences Center and the Utah Population Database (UPDB), identifying 4764 patients with AMD. Specialized kinship analysis software was used to test for familial aggregation of disease, estimate the magnitude of familial risks, and identify families at high risk for disease. The population-attributable risk (PAR) for AMD was calculated to be 0.34. Recurrence risks in relatives indicate increased relative risks in siblings (2.95), first cousins (1.29), second cousins (1.13), and parents (5.66) of affected cases. There were 16 extended large families with AMD identified for potential use in genetic studies. Each family had five or more living affected members. The familial aggregation of AMD shown in this study exemplifies the merit of the UPDB and supports recent research demonstrating significant genetic contribution to disease development and progression.

“Population Lab”: The Creation of Virtual Populations for Genetic Epidemiology Research

Studies of familial aggregation of disease routinely use linked population registers to construct retrospective cohorts. Although such resources have provided numerous estimates of familial risk, little is known regarding the sensitivity of the estimates to assumed disease models, changing demographics and incidence, and incompleteness of the data. Furthermore, there are no standard tools for testing the validity of estimates from standard epidemiologic designs and from new analytic strategies using register data. We present a method and a software package for simulating realistic populations of related individuals, using easily available vital statistics (population counts and fertility and mortality rates). The virtual population is stored in a pedigree file, allowing for easy retrieval of relatives and family structures. We simulate breast cancer in our population using age-specific incidence rates. The Swedish population is simulated as dynamically evolving over the calendar period 1955-2002. The simulated and real population agree well on important features such as age profile, sibship size distribution, and average age at first birth. Using breast cancer as an example, we present several models of familial disease aggregation and show that the parameters used in the simulations are faithfully estimated. In addition, we illustrate how our simulated population provides insight into how incomplete family history in real register data can affect estimates of familial risk. This simulation method can be used to investigate various underlying models of disease aggregation in families and enhance the development of optimal approaches for family studies. The software package, Population Lab, is available for free download (http://www.meb.ki.se/ approximately marrei/software/poplab/ and http://cran.at.r-project.org/).

Some Molecular and Clinical Aspects of Genetic Predisposition to Malignant Melanoma and Tumours of Various Site of Origin

Based on epidemiological data we can assume that at least some malignant melanoma (MM) and breast cancer cases can be caused by the same genetic factors. CDKN2A, which encodes the p16 protein, a cyclin-dependent kinase inhibitor suppressing cell proliferation, is regarded as a major melanoma susceptibility gene and the literature has also implicated this gene in predisposition to breast cancer. Genes also known to predispose to MM include XPD and MC1R. We studied CDKN2A/ARF, XPD and MC1R for their associations with melanoma and breast cancer risk in Polish patients and controls. We found that CDKN2A and ARF do not contribute significantly to either familial melanoma or malignant melanoma within the context of a cancer familial aggregation of disease with breast cancer. However, the common variant of the CDKN2A gene A148T, previously regarded as non-pathogenic, may predispose to malignant melanoma, early-onset breast cancer and lung cancer. Compound carriers of common XPD variants may be at slightly increased risk of breast cancer or late–onset malignant melanoma. Common recurrent variants of the MC1R gene (V60L, R151C, R163Q and R160W) may predispose to malignant melanoma. In general, the establishment of surveillance protocols proposed as an option for carriers of common alterations in CDKN2A, XPD or MC1R variants requires additional studies. It is possible that missense variants of genes for which truncating mutations are clearly pathogenic may also be deleterious, but with reduced penetrance. This may be overlooked unless large numbers of patients and controls are studied. A registry that includes 2000 consecutive breast cancer cases, 3500 early onset breast cancer patients, 500 unselected malignant melanoma and over 700 colorectal cancer patients has been established in the International Hereditary Cancer Centre and can contribute to these types of large association studies.

Role of Genetic Analyses in Cardiology

“When a variation is of the slightest use to a being, we cannot tell how much of it to attribute to the accumulative action of natural selection, and how much to the conditions of life.” — Charles Darwin, The Origin of Species, Chapter 5: Laws of Variation, 1859 Genetic testing is a growing discipline in clinical practice, and it is being used to make treatment decisions and to counsel patients. Before genetic testing can be done, reliable information needs to be available on which genes are involved in a disease. Research into genetic determinants of disease often is based on articles reporting the magnitude of the heritability of disease. Here, we address limitations in the use of heritability as a point of departure, such as a basis for power calculations, for genetic studies. Familial aggregation of diseases generally is taken as evidence for a role of genetic factors in the origin of a disorder or trait. A commonly used measure to quantify the extent to which this familial aggregation of diseases is due to genetic factors is the heritability. Heritability estimates were first used by plant and animal breeders to predict the effectiveness of their selection of a desired phenotype such as an increase in milk production or resistance to yeast infections. In human studies, heritability has the intuitive appeal of quantifying genetic effects without necessarily knowing the mode of inheritance of a trait. Heritability is frequently estimated for complex diseases or traits such as diabetes mellitus, coronary artery disease, and serum lipid levels, which are phenotypes reflecting the interplay of genetic and environmental factors. Continuous traits or phenotypes (height, body mass index, serum lipid levels, etc) in a population usually have an approximately normal distribution characterized by a mean and a variance. The total variance can be partitioned mathematically …

Analysis of Familial Aggregation in the Presence of Varying Family Sizes

Summary Family studies are frequently undertaken as the first step in the search for genetic and/or environmental determinants of disease. Significant familial aggregation of disease is suggestive of a genetic aetiology for the disease and may lead to more focused genetic analysis. Of course, it may also be due to shared environmental factors. Many methods have been proposed in the literature for the analysis of family studies. One model that is appealing for the simplicity of its computation and the conditional interpretation of its parameters is the quadratic exponential model. However, a limiting factor in its application is that it is not reproducible, meaning that all families must be of the same size. To increase the applicability of this model, we propose a hybrid approach in which analysis is based on the assumption of the quadratic exponential model for a selected family size and combines a missing data approach for smaller families with a marginalization approach for larger families. We apply our approach to a family study of colorectal cancer that was sponsored by the Cancer Genetics Network of the National Institutes of Health. We investigate the properties of our approach in simulation studies. Our approach applies more generally to clustered binary data.

On the Multivariate Probit Model for Exchangeable Binary Data with Covariates

This paper considers the use of a multivariate binomial probit model for the analysis of correlated exchangeable binary data. The model can naturally accommodate both cluster and individual level covariates, while keeping a fairly flexible intracluster association structure. We discuss Bayesian estimation when a sample of independent clusters of varying sizes are available, and show how Gibbs sampling may be used to derive the posterior densities of parameters. The methodology is illustrated with two examples: the first involves epidemiological data from a study of familial disease aggregation; the second uses teratological data from a developmental toxicity application.