Further lack of association between the 5HT2A gene promoter polymorphism and susceptibility to eating disorders and a meta-analysis pertaining to anorexia nervosa

Please download to get full document.

View again

of 3
0 views
PDF
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Document Description
Further lack of association between the 5HT2A gene promoter polymorphism and susceptibility to eating disorders and a meta-analysis pertaining to anorexia nervosa
Document Share
Document Transcript
  Molecular Psychiatry (1999) 4,  410–417 ©  1999 Stockton Press All rights reserved 1359–4184/99 $15.00 SCIENTIFIC CORRESPONDENCE Further lack of associationbetween the 5-HT 2A  genepromoter polymorphism andsusceptibility to eatingdisorders and a meta-analysispertaining to anorexianervosa SIR – Collier and colleagues reported an allelic associ-ation between the  − 1438G/A promoter polymorphismof the 5-HT 2A  gene and anorexia nervosa (AN). 1 Whiletwo groups were unable to replicate the intial find-ing, 2,3 two other studies reported a positive result. 4,5 This paper presents data of a new association studybetween eating disorders and this polymorphism.Additionally, we performed a meta-analysis to investi-gate the inconsistent findings.We compared allele and genotype frequenciesbetween study groups of mostly adult patients with AN( n  =  78) or bulimia nervosa (BN;  n  =  99) and 170 normalcontrols. None of these individuals was included inour previous study. 2 AN and BN study groups werediagnosed according to DSM-IV and have beendescribed previously. 6 Seventy-seven per cent of thepatients with AN were of the restricting type. Writteninformed consent was given by all participants and, incase of minors, by their parents. The investigation wasapproved by the ethics committee of the University of Marburg. The  − 1438A/G polymorphism was analysedas previously described. 2 Genotype frequencies did notdeviate from Hardy–Weinberg equilibrium in either of the study groups or in controls (all  P -values   0.39).Eighteen (18.18%) and 55 (55.55%) of the 99 patientswith BN were homozygous for the  − 1438A/A and the − 1438G/G genotype, respectively. Data for patientswith AN and for controls are given in Table 1. Neitherallele nor genotype frequencies of patients with ANand BN were different from those of the controls (allnominal  P -values   0.75). Using the trend test, geno-typic odds ratios (OR) for patients with AN and BN  vs controls were 0.98 (95% CI: 0.66–1.45) and 1.05 (0.75–1.48), respectively. Our new data failed to replicate theinitial finding of Collier and colleagues. 1 However, oursample had power between 65% and 99% to detect anassociation between the  − 1438A/G polymorphism andanorexia nervosa using the continuity-corrected one-sided    2 -test at the 5% test level with the allele fre-quencies for the  − 1438A/G polymorphism as estimatedby Sorbi and colleagues, 4 Enoch and colleagues 5 andCollier. 7 Sorbi and colleagues 4 reported a strong positiveassociation ( P   0.0001) between the  − 1438A/G poly-morphism and AN of the restricting type. However, ourexploratory analyses for patients with AN of therestricting type  vs  controls revealed a nominal  P -value  0.45 (data not shown). Due to the inconsistent find-ings of the different studies, we performed a meta-analysis as recommended by Baron. 8 For this purpose,a logistic regression using the original data was con-ducted. The model included a random effect to adjustfor study of specific effects like ethnic diversity 9 because there was substantial deviation from studyhomogeneity ( P   0.0001). The srcinal table of Collierand colleagues contained an error. 10 Therefore, weused the corrected data of Collier. 7 There seems to be no methodological framework tocombine the results of family-based association studieswith those from case-control association studies. Thus,we are unable to include the findings of Hinney andcolleagues 2 who performed a family-based associationstudy and did not detect linkage and associationbetween the  − 1438A/G promoter polymorphism andAN.Table 1 shows that both genotype and allele fre-quencies vary substantially between the individualstudies resulting in dissimilar OR. For instance, the fre-quency for controls being homozygous for the A alleleranges from 8.70% 5 to 20.00%. 3 The meta-analysis combining the studies using thedata shown in Table 1 yields similar results for theallelic and the genotypic OR. Our meta-analysis did notdetect association between the  − 1438A/G promoterpolymorphism and AN (both nominal  P   0.11). Poss-ible explanations for the inconsistencies between studyresults are many-fold as discussed eg by Paterson. 11 These include genetic heterogeneity or ethnic admix-ture. 2 The latter can be circumvented by family-basedassociation studies. Acknowledgements We thank Mr Gutfleisch and Ms Jesse for their assist-ance. This study was supported by the Deutsche For-schungsgemeinschaft. A Ziegler 1 , J Hebebrand 2 , T Go ¨rg 1 , K Rosenkranz 2 ,MM Fichter 3 , B Herpertz-Dahlmann 4 , H Remschmidt 2 and A Hinney 2 1  Institute of Medical Biometry and Epidemiology,Philipps-University Marburg, Germany;  2  Department of Child and Adolescent Psychiatry, Philipps-University Marburg, Germany;  3 Klinik Roseneck, Hospital for  Behavioural Medicine, Prien, Germany;  4  Department of Child and Adolescent Psychiatry, University of  Aachen, Germany Correspondence should be addressed to Dr A Ziegler. E-mail: ziegler@mai-ler.uni-marburg.de   S   c i      e n t    i     fi   c  C  or  r   e  s   p on d   e n c  e  4   1   1    Table 1  Genotype frequencies, allele frequencies (percentages corresponding to reported frequencies in parenthesis) and odds ratio estimates (95% CI in parenthesis;trend test for genotypes) of the  − 1438A/G 5-HT 2A  promoter polymorphism  Reference Disease status Genotype-wise Allele-wise − 1438A/A  − 1438A/G  − 1438G/G Odds ratio  − 1438A  − 1438G Odds ratio Collier 7 Anorexia 25 (30.86) 33 (40.74) 23 (28.40) 1.52 (1.05–2.18) 83 (51.23) 79 (48.77) 1.52 (1.06–2.17)Control 34 (15.04) 117 (51.77) 75 (33.18) 185 (40.93) 267 (59.07)Hinney  et al 2,a Anorexia 20 (20.00) 39 (39.00) 41 (41.00) 0.89 (0.65–1.22) 79 (39.50) 121 (60.50) 0.89 (0.64–1.22)Control 62 (17.46) 177 (49.86) 116 (32.68) 301 (42.39) 409 (57.61)Campbell  et al 3 Anorexia 39 (25.66) 68 (44.74) 45 (29.61) 1.23 (0.91–1.68) 146 (48.03) 158 (51.97) 1.26 (0.91–1.74)Control 30 (20.00) 67 (44.67) 53 (35.33) 127 (42.33) 173 (57.67)Sorbi  et al 4 Anorexia 23 (29.87) 41 (53.24) 13 (16.88) 266 (1.64–4.31) 87 (56.49) 67 (43.51) 2.36 (1.54–3.61)Control 10 (9.35) 56 (52.34) 41 (38.32) 76 (35.51) 138 (64.49)Enoch  et al 5 Anorexia 17 (25.00) 35 (51.47) 16 (23.53 1.95 (1.15–3.32) 69 (50.74) 67 (49.26) 1.81 (1.11–2.95)Control 6 (8.70) 38 (55.07) 25 (36.23) 50 (36.23) 88 (63.77)New study Anorexia 7 (8.97) 32 (41.03) 39 (50.00) 0.98 (0.66–1.45) 46 (29.49) 110 (70.51) 0.98 (0.64–1.49)Control 21 (12.35) 60 (35.29) 89 (52.35) 102 (30.00) 238 (70.00)Meta-analysis 1.26 (1.00–1.90) 1.31 (0.99–1.73) a Data from obese and underweight individuals were pooled to serve as controls.  Scientific Correspondence 412 1 Collier DA  et al .  Lancet   1997;  350 : 412.2 Hinney A  et al .  Lancet   1997;  350 : 1324–1325.3 Campbell DA  et al .  Lancet   1998;  351 : 499.4 Sorbi S  et al .  Lancet   1998;  351 : 1785.5 Enoch MA  et al .  Lancet   1998;  351 : 1785–1786.6 Hinney A  et al .  Mol Psychiatry  1998;  3 : 539–543.7 Collier DA.  Lancet   1999;  13 : 929.8 Baron M.  Mol Psychiatry  1997;  2 : 278–281.9 Littell RC  et al .  SAS System for Mixed Models . SAS Institute: Cary,NC, 1996.10 Ziegler A, Go¨rg T.  Lancet   1999;  353 : 929.11 Paterson AD.  Mol Psychiatry  1997;  2 : 277–278. Sampling and analyticconsiderations in moleculargenetics studies of ADHD. Theadvantages of predictingphenotype from genotype inADHD research SIR – The articles on ADHD and the DRD4 in the Sep-tember 1998 issue of   Molecular Psychiatry  were veryinformative. What impressed me most was the expec-tation that the authors of these articles use as a separategroup, 1 or explain why they did not use as a separategroup, 2 potential control group recruits which metsymptom requirements for ADHD. It has been myexperience as a reader of ADHD research that ADHDand control participants tend to be recruited differently(clinical  vs  non-clinical populations) and that controlstend to be excluded for exhibiting comorbid conditions(ie, conduct disorder) commonly diagnosed in theADHD participants.Recruiting the experimental and control groups fromdifferent populations may reveal group differences inrepeat frequencies unrelated to participant statuswhich may result in both false-positive and false nega-tive results. 3 Secondly, though Castellanos  et al ’s 1 report of a 71% comorbidity rate is quite high, theredoes tend to be a higher rate of comorbidity in clinic-referred than in non-referred children diagnosed withADHD. 4 Since it has been put forward that comorbiditymay moderate the phenotypic expression of ADHD, 5,6 it may be necessary to account for the effects of comor-bidity on the expression of ADHD when doing gen-etic research.I postulate that the best way to avoid the statisticalerrors associated with the sampling biases listed aboveis, instead of recruiting on the basis of phenotype andtesting for genotype, we should recruit on the basis of genotype and test for phenotype. Considering thepresent interest in the relationship between the DRD4and ADHD 7 and the suggestion that the DRD2 is amajor moderator of ADHD, 4 I propose that we employa 5  ×  3 factorial design using only those individualswith the 2,2, 2,4, 4,4, 4,7 and 7,7 repeats of the DRD4gene and the A1,A1, A1,A2 and A2,A2 alleles of theDRD2 gene. The proposed design will not only showthe independent effect of each gene on the expressionof ADHD and various comorbid conditions (as well aspossible interactions), but will also provide a largegroup of (DRD4) 4,4 repeat, (DRD2) A2,A2 alleles parti-cipants that may be used to test the influence of otherproposed ADHD genes, such as Cook   et al ’s 3 DAT1. V Lavallee 26 Larchwood PlaceWinnipeg, ManitobaCanada R2H 1M1 E-mail: vaudree@hotmail.com 1 Castellanos F  et al .  Mol Psychiatry  1998;  3 : 431–434.2 Rowe D  et al .  Mol Psychiatry  1998;  3 : 419–426.3 Cook E  et al .  Am J Am Genet   1995;  56 : 993–998.4 Weiss G, Hechtman L.  Hyperactive Children Grown Up , 2nd edn.The Guilford Press, London, 1993, pp 335, 346.5 Pisecco S  et al .  J Am Acad Child Adolesc Psychiatry  1996;  35 :1477–1484.6 Gjone H  et al .  J Am Acad Child Adolesc Psychiatry  1996;  35 :588–596.7 Thaper A.  Mol Psychiatry  1998;  3 : 370–372. Waldman and Rowe reply SIR – In her recent letter to  Molecular Psychiatry , MsLavallee raises a number of interesting issues in geneticresearch on ADHD. In this comment, we focus on threeof these issues regarding the sampling of participantsin molecular genetic studies of ADHD. Specifically, weaddress the issues of choice of controls, the impact of overlapping diagnostic conditions, and the viability of recruiting participants based on genotype  vs  pheno-type.Ms Lavallee raises the appropriate concern thatADHD and comparison groups recruited from differentpopulations might differ in allele frequencies forreasons other than linkage between the disorder andthe genetic marker. Although careful matching of casesand controls may overcome this problem to a degree,many potential sources of population stratificationexist, some heretofore unknown, that could lead toassociations that are not due to the casual effects of thegene. In order to counteract this problem, a number of recent molecular genetic studies of ADHD 1–5 have useda within-family statistical procedure, the TransmissionDisequilibrium Test (TDT), 6 that protects against thebiasing effects of population stratification. The TDTcontrasts the transmission of high-risk   vs  low-risk alleles from heterozygous parents to their affected chil-dren. Use of the TDT should eliminate any populationstratification biases incurred in the comparison of ADHD cases with controls, which may be drawn frompopulations with characteristics that differ inimportant ways.
Similar documents
View more...
Search Related
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks