Genome Scans For Genetic Predisposition To Alcoholism By Use Of Transmission Disequilibrium Test Analyses

Transcription

1 Genetic Epidemiology 17(Suppl 1): S277-S281 (1999) Genome Scans For Genetic Predisposition To Alcoholism By Use Of Transmission Disequilibrium Test Analyses Grier P. Page, Terri M. King, Jill S. Barnholtz, Mariza de Andrade, Leif E. Peterson, and Christopher I. Amos Department of Epidemiology (G.P.P., T.M.K., J.S.B., M.d.A., C.I.A), Division of Cancer Prevention, MD Anderson Cancer Center, Houston, Texas; Department of Biometry (J.S.B.), University of Texas School of Public Health, and Department of Medicine (L.E.P.), Baylor College of Medicine, Houston, Texas We report the results of the analysis of three measures of alcoholism and six associated symptoms using transmission disequilibrium (TDT) analysis on data from the Collaborative Study on the Genetics of Alcoholism data set. Implementation of identity-by-state (IBS) routines for error checking revealed 10 reported full siblings that were rejected as a full sibling to all of their purported full siblings with p < TDT analysis revealed two loci with significant transmission disequilibrium (p < 0.001) on chromosomes 1 and 7. Analysis by parental origin found alleles at three loci displaying significant disequilibrium in the transmission of the paternal alleles for at least three of the nine tested traits. These loci are on chromosomes 6,9, and 13. Analyses of Caucasian families alone and the use of a single affected individual from each family also yielded significant results for the loci on chromosomes 6, 9, and Wiley-Liss, Inc. Key words: HTR-2A, identity-by-state, linkage, linkage disequilibrium INTRODUCTION Alcoholism is a common complex disease and identifying causative genetic factors will require a variety of statistical methods. Here we chose to apply the transmission disequilibrium test (TDT). In the presence of disequilibrium the TDT method has been shown to have much more power than usual linkage methods that assume linkage Address reprint requests to Dr. G.P. Page, Department of Biometry and Epidemiology, Medical University of S. Carolina, 135 Rutledge Avenue, Suite 1148, P.O. Box , Charleston, SC Wiley-Liss, Inc.

2 S278 Page et al. equilibrium [Risch and Merikangas, 1996]. The TDT method was designed to be a test of association and linkage that is independent of population substructure [Spielman et al., 1993]. TDT analyses check for alleles that are transmitted more often than not from parents to affected children. TDT analyses have been used to find genes influencing other complex diseases such as type I diabetes [Spielman et al., 1993]. METHODS A complete description of the sample ascertainment and pedigree checking has been described previously [Reich et al., 1998]. Because several sibships did not have any parental information, the identity-by-state (IBS)-based test SibError [Ehm and Wagner, 1998] was used to check the reported familial relationships. The families were divided into nuclear families and analyzed using 285 autosomal markers. All known half siblings were reclassified as full siblings to establish an estimate of the false negative rate. To conclude that a purported full sibling was inconsistent with the pedigree structure we required that he/she be rejected in comparison with all other full siblings with p < The disease classifications used were based upon DSM-III-R and Feighner (COGA) and ICD-10. According to these classifications, only affected individuals were used for TDT analysis. Pure unaffected and unaffected with symptoms were considered to be unaffected. Unknown individuals and those who have never drunk alcohol were considered to be missing. The final disease classification grouped individuals who were affected and unaffected with symptoms according to DSM-III-R and Feighner as symptomatic. The six symptom variables that were analyzed included: persistent desire to stop drinking, morning drinking, alcohol craving, giving up activities to drink, physical health problems, and emotional/psychological problems resulting from drinking. Among the 11 variables distributed, multivariate logistic regression identified these six as most predictive of alcoholism (data not shown). All variables but persistent desire to quit drinking are binary discrete variables. For persistent desire to quit drinking, only individuals who had the desire to quit drinking more than three times were considered affected. The SAGE 3.1 [S.A.G.E., 1997] routine TDTEX was used to carry out the TDT analysis for the three diagnoses and six symptom variables. The transmitted alleles were analyzed as individual alleles rather than by the offspring's genotype. The p-values reported are Markov chain Monte Carlo estimates of the significance. The first analysis included all affected individuals for whom both parents had been genotyped. Analysis was also performed on the basis of the parental origin of transmitted alleles. To verify that significant findings were not the result of including multiple affected individuals within the same sibship, the three loci with the most significant results,,, and, were reanalyzed using the oldest affected offspring in each family. This reanalysis was made on the entire data and for a subset of self-reported Caucasian families only. The other ethnicities had sample sizes too small to allow meaningful conclusions. RESULTS AND DISCUSSION All known half siblings were correctly identified as not being full siblings (data not shown) with at least a 5% level of significance. Table I shows all the individuals that are rejected when compared with all other purported siblings at p < Interestingly, in two sibships, two individuals were rejected. In family 42, the sibship consists of individuals

3 TDT Analyses of Alcoholism Traits S279 TABLE I. Analysis of Individuals with Doubtful Relationships' Family ID Siblings < z Scores ,-9.2, -9.2, -4.46,-3.15,-3.4, , , , -2.74, , -6.48, , -2.92, , , -6.84, , -2.67, ,-9.55 a The columns are: the family number, individual ID number, total number of putative full siblings, number of putative siblings that rejected the individual in question with p-value < 0.01 (2 sided z-test), the number with p-values between 0.01 and 0.05, and the z-values for rejection. 435 to 438. Individuals 437 and 438 are full siblings, and 435 and 436 appear to be half siblings of each other and of all others in the sibship. In the sibship consisting of individuals 725 to 728, 726 and 728 are full siblings while 725 and 727 are half siblings of each other and of all others in the sibship. SibError found several possible problems in reported pedigree structures that had been missed by previous analyses; but, the test only rejects individuals as full siblings without affirming that they are instead half siblings. While we are confident that several individuals are actually half siblings, e.g., 822, others are more problematic. While they did meet our criterion for exclusion as full siblings, we would feel more sure of the relationships if we knew that the rejected sibling actually fit the expected IBS distribution for a half sibling. For the genome scan using all possible family members, significant results were found for D1S1602 (COGA: p-value = 0.001, symptomatic: p-value = 7) and for (emotional/psychological problems: p-value = ). The results from the TDT analyses with respect to parental origin of alleles are given in Table II. A total of 16 loci were found to be significant with p < under at least one of the tested models. The significances of the results for all the tested models are reported for these 16 loci. We performed a total of 7,731 tests. Assuming each test was independent, we expected a total of seven or eight tests with p < 0.001, but 42 were found. Three loci (and their most significant alleles), (181), (106), and (196), were significant for transmission disequilibrium of a paternal allele for three or more of the tested models. When using all affected individuals in a family the TDT method is only a valid test of linkage, but not of disequilibrium.,, and were re-analyzed by using the oldest affected offspring from each family for the three disease classifications. This group was analyzed as a whole and using only Caucasians. For all loci except, in Caucasians using the Symptomatic classification (p-value = 0.087) there was significant evidence (p < 0.05) of disequilibrium in the transmission of the paternal allele. We found evidence of a candidate gene near. is located at -48 cm from pter while the human serotonin receptor type 2A [HTR2A] is located between 40 and 47 cm from pter, according to the Whitehead Institute for Biomedicai Research [www-genome.wi.mit.edu] chromosome 13 STS map. Work in rats has indicated that HTR2A antagonists such as amperiozide inhibit alcohol consumption [Overstreet et al., 1997]. There is also evidence of paternal imprinting in the 5-HT2A region [Kato et al., 1996].

4 S280 Page et al. TABLE II. TDT Analysis on the Basis of Parental Origin of Transmitted Alleles Using Three Alcoholism Classification and Six Alcoholism-Associated Symptoms Using All Affected Offspring with the Reported p-values Unadjusted for Multiple Comparisons COGA ICD-10 Svmpti jmatic Marker Marker Marker Persistent desire to quit drinking N/D N/D Gave up activities to drink Morning drinking Physical health problems Alcohol craving Emotional/psych, problems

5 TDT Analyses of Alcoholism Traits S281 A genome scan using TDT analysis with and without parental origin suggests that there may be several loci involved in alcohol dependence. The regions near these loci would merit genotyping of additional closely linked markers or replication in another sample. The primary limitation of these TDT analyses is that the density of the markers is insufficient to provide coverage of the entire genome with any appreciable power. ACKNOWLEDGMENTS The work was funded by grants GM52607 and MH Portions of the results from this study were obtained by using the program package S.A.G.E., which is supported by U.S. Public Health Service Resource grant RR03655 from the National Center for Research Resources. We thank two anonymous reviewers from their helpful comments. REFERENCES Ehm MG, Wagner M (1998): A test statistic to detect errors in sib-pair relationships. Am J Hum Genet 62: Kato MV, Shimizu T, Nagayoshi M, Kaneko A, Saskai MS, Ikawa Y (1996): Genomie imprinting of the human serotonin-receptor [HTR2] gene involved in development of retinoblastoma. Am J Hum Genet 59: Overstreet DH, McArthur RA, Rezvani AH, Post C (1997): Selective inhibition of alcohol intake in diverse alcohol-preferring rat strains by the 5-HTR2A antagonist amperozide and FG5974. Alcohol Clin Exp Res 21: Reich T, Edenberg HJ, Goate A, Williamson JT, Rice JP, Van Eerdewegh P, Foroud T, Hesselbrock V, Schuckit MA, Bucholz K, Porjesz B, Li TK, Conneally M, Nürnberger JI Jr, Tischfield JA, Crowe RR, Cloninger CR, Wu W, Shears S, Carr K, Crose C, Willig C, Begleiter.H (1998): Genome-wide Search for genes affecting the risk for alcohol dependence. Am J Med Gen 81: Risch N, Merikangas K (1996): The future of genetic studies of complex human diseases. Science 283: S.A.G.E. (1997): Statistical analysis for Genetic Epidemiology, Release 3.1. Computer package available fro the Department of Epidemiology and Biostatistics, MetroHealth Campus, Case Western Reserve University, Cleveland, Ohio. Spielman R, McGinnis RE, Ewens WJ (1993): Transmission test for linkage disequilibrium: the insulin region and insulin-dependent diabetes mellitus [IDDM]. Am J Hum Genet 52: