Complex segregation analysis of restless legs syndrome provides evidence for an autosomal dominant mode of inheritance in early age at onset families.код для вставкиСкачать
Complex Segregation Analysis of Restless Legs Syndrome Provides Evidence for an Autosomal Dominant Mode of Inheritance in Early Age at Onset Families Juliane Winkelmann, MD,1 Bertram Muller-Myhsok, MD,2 Hans-Ulrich Wittchen, PhD,3 Bettina Hock,1 Muriel Prager,1 Hildegard Pfister,3 Andreas Strohle, MD,4 Ilonka Eisensehr, MD,5 Martin Dichgans, MD,4 Thomas Gasser, MD,4 and Claudia Trenkwalder, MD1 A strong familial component of restless legs syndrome (RLS) is known. The objective of this study therefore was to investigate the likely mode of inheritance of RLS. RLS patients and their first-degree relatives were investigated and classified in RLS affected and RLS nonaffected subjects. Assessments were based on direct, personal standardized diagnostic interviews. Complex segregation analysis was performed with the families stratified according to the mean age at onset of the disease within the families. Two hundred thirty-eight RLS patients, 537 first-degree relatives, and 133 spouses were interviewed. Two groups of families were stratified: mean age at onset up to 30 years of age (Group A) and older than 30 years (Group B; p < 0.005). In Group A, segregation analysis strongly favored a single major gene acting autosomal dominant with a multifactorial component. Parameter estimates were 0.003 for the allele frequency, 1.0 for the penetrance, and 0.005 for the phenocopy rate. In Group B, no evidence for a major gene could be elucidated. The segregation pattern found in our families argues for an autosomal allele acting dominantly in RLS families with an early age at onset of symptoms and suggests that RLS is a causative heterogeneous disease. Ann Neurol 2002;52:297–302 The restless legs syndrome (RLS) is characterized by dysesthesias that usually are located in the lower limbs, mainly the calves, and are associated with an irresistible urge to move these limbs. Moving the affected leg or walking leads to a temporary relief of the symptoms. These occur predominantly at rest in the evening or at night, interfering with sleep onset and disrupting sleep.1–3 The diagnosis of the disease is based on the clinical description of these symptoms by the patient.1 A recent population-based study reported a high prevalence of 9.8% in individuals older than 65 years.4 A survey in the general population has confirmed this, pointing to marked age effects ranging from 3% in individuals aged younger than 30 years to approximately 20% in those aged 80 years and older.5 Clinical surveys of idiopathic RLS patients showed that up to 60%6 – 8 report a positive family history. Investigations of single families with RLS have suggested an autosomal domi- nant mode of inheritance9 –13 with variable expressivity.11–13 In addition, a younger age at onset was consistently found in familial compared with sporadic cases.6 – 8,14,15 In a single French-Canadian family, evidence for linkage of RLS to chromosome 12 recently has been found.16 Interestingly, this finding was obtained using an autosomal recessive model with very high allele frequency (q ⫽ 0.75), which is unlikely to be representative for RLS as a whole. The mode of inheritance in RLS and related genetic parameters such as penetrance and the rate of phenocopies are not yet known. An accurate assumption of these parameters, however, is an essential prerequisite for linkage studies. The objective of this study therefore was to investigate the likely mode of inheritance of RLS by examining a clinical population of RLS patients and their family members. From the 1Section of Neurology, Max Planck Institute of Psychiatry, Munich; 2Bernhard Nocht Institute for Tropical Medicine, Hamburg; Departments of 3Clinical Psychology and 4Psychiatry, Max Planck Institute of Psychiatry; 5Department of Neurology, Ludwig Maximilians University of Munich, Klinikum Grosshadern, Munich, Germany. Received Oct 22, 2001, and in revised form Mar 25, 2002. Accepted for publication Apr 20, 2002. Current address for Dr Wittchen: Institute of Clinical Psychology and Psychotherapy, Technical University of Dresden, Dresden, Germany. Published online Jun 7, 2002, in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ana.10282 Address correspondence to Dr Trenkwalder, Department of Clinical Neurophysiology, Georg-August-Universität Göttingen, Robert-KochStr. 40, D-37075, Göttingen, Germany. E-mail: firstname.lastname@example.org © 2002 Wiley-Liss, Inc. 297 Patients and Methods Study Population RLS, and no positive RLS family history information was obtained for any relatives. All inpatients and outpatients of the Movement Disorder Unit at the Section of Neurology, Max Planck Institute of Psychiatry, between January 1996 and March 1999 who received a diagnosis of idiopathic RLS2 were included in the study as index patients. This sample was enriched by patients of the Department of Neurology, Klinikum Grosshadern, Ludwig Maximilian University, Munich who met the same study criteria. We excluded secondary cases of RLS, except for five uremic patients who had an onset of their RLS symptoms before the onset of their dialysis treatment. Study Design All index patients were contacted by letter describing the aim and design (see below) of the study. Patients were asked to contact as many first-degree relatives (parents, siblings, children) as possible to participate in the study and were requested to send back a prepaid envelope containing a list of all family members who had agreed to participate. We excluded family members younger than the age of 18 years. One subject aged 12 years was included as an exception because she presented as an index patient. Written consent was obtained from all subjects (Ludwig Maximilians University, ref. no. 97/99). Diagnostic Interview All subjects (index patients, spouses, and first-degree relatives) were contacted by telephone and assessed by means of a direct standardized computer-assisted personal interview using the Composite International Diagnostic Interview17 as a model. The subjects were randomly assigned to interviewers. In each family, the index patient was interviewed first, a pedigree then was constructed, and the participating relatives were contacted. The interview started with a section on demographic data, followed by a second section on the four diagnostic criteria.2 Questions pertaining to criteria 1 to 3 were coded with “yes” or “no.” The fourth criterion (circadian rhythm) had multiple response options: symptoms in the morning, afternoon, evening, night, or no specific time. Only the answers “afternoon,” “evening,” and “night” were counted as “positive” toward RLS diagnosis. Only if all questions referring to the minimal criteria were answered in the positive was the subject classified as RLS “affected.” The third section included data on clinical characteristics of the disease (eg, age at onset of the first symptom experienced). Frequency of Familial Restless Legs Syndrome The index patients were divided into the following subgroups. Patients with RLS were classified as “definite familial RLS” if at least one family member in addition to the index patient was classified as RLS affected by interview. Patients were classified as having “probable familial RLS” if family members were classified as nonaffected in the direct interview, but a positive RLS family history was ascertained (indirect assessment) in at least one additional first-degree family member. Patients were classified “nonfamilial RLS” if no directly interviewed family members met the criteria for 298 Annals of Neurology Vol 52 No 3 September 2002 Likely Mode of Inheritance Only probands in whom the phenotype information was assessed by direct interview were included in the following analysis. STRATIFICATION OF THE FAMILIES BY THE MEAN AGE AT ONSET OF THE DISEASE. A test for admixture (commin- gling analysis)18 was performed to assess whether the distribution of the mean ages at onset in the families could be better explained as a mixture of two (or more) distributions rather than a single distribution. In the case of significant admixture, the resulting groups were used for delineation of strata for complex segregation analysis. COMPLEX SEGREGATION ANALYSIS. In each of the resulting strata (see above), as well as for the whole sample, a complex segregation analysis was performed (program pedigree analysis package),19,20 and an ascertainment correction was applied.21 For each model, the likelihood and Akaike’s information criterion (AIC) were calculated.22 Significance testing was performed by the method of nested hypothesis testing. The specific principle of testing for a major gene consists of two steps. First, the most parsimonious major gene-containing model (MPMG) is tested against a multifactorial model (no MPMG). If this test shows that the MPMG is significantly better than the multifactorial model, a second step is performed. Here, the MPMG model is tested against the model of free transmission probabilities (⫽ general model). If the MPMG cannot be rejected and if, in addition, the environmental transmission hypothesis is rejected against the general model, it is possible to conclude that a major gene exists. A test for heterogeneity between the results of the segregation analysis was performed.23,24 Gender differences between both groups were tested using a contingency table 2 test for heterogeneity of proportions. Results Response Rate and Study Population Recruitment letters were sent out to 397 RLS index patients. Eighty-six failed to respond: 9 patients had died, and 77 could not be contacted on at least 2 attempts. Three hundred eleven patients sent back the prepaid envelope: 239 agreed (1 patient died after his agreement; n ⫽ 238) and 72 patients refused to participate in the study. Of the 238 patients, 196 allowed us to contact their relatives (537 of 772 living firstdegree relatives and 133 spouses), and for 42 patients no family members were available for direct contact. Altogether, 908 interviews were conducted within 196 families and with an additional 42 index patients (Fig). Frequency of Familial Restless Legs Syndrome According to our classification, 89 of the 196 (45.4%) index patients showed “definite familial RLS.” Forty- Fig. Study design, response rate, participating subjects, and frequency of familial restless legs syndrome (RLS). Patients were classified as “definite familial RLS” if at least one family member in addition to the index patient was classified as having definite RLS by interview. They were only classified having “probable familial RLS” if family members interviewed were classified RLS nonaffected, but an additional first-degree family member was classified as having RLS by family history. They were classified “nonfamilial RLS” if no family members interviewed showed RLS and, additionally, no other family member was known to have symptoms of RLS by family history. three (21.9%) families showed “probable familial RLS” (indirect assessment) and 64 (32.7%) had a “nonfamilial RLS.” In the group in which only the index patients were interviewed, 16 (38.1%) index patients showed “probable familial RLS” (indirect assessment) and 26 (61.9%) had a “nonfamilial RLS” (see Fig). Demographic Data Of the 908 subjects interviewed, 513 were women and 395 were men. The mean age of respondents at the time of the evaluation was 54.7 ⫾ 16.8 years (mean ⫾ standard deviation [SD]; range, 12–93 years) in women and 53.6 ⫾ 16.5 years (mean ⫾ SD; range, 18 –91 years) in men. Of the 513 women, 240 were RLS positive (46.8%), and 273 (53.2%) were RLS negative. Of the 395 men, 153 were RLS positive (38.7%) and 242 (61.3%) were RLS negative. Commingling Analysis The distribution of the mean ages at onset in the families was better explained by a mixture of two normal distributions than a single normal one ( p ⬍ 0.005). The first distribution (mean age at onset in the families, 23.5 years; SD ⫽ 10.9 years) accounted for 38%; the second distribution (mean, 45.9 years; SD ⫽ 15 years) accounted for the remaining 62% of the cases. There was no significant evidence for additional admixture in the age at onset data. Using a Bayesian type of analysis, we found that a mean age at onset within the families of 30 years corresponds to a 50% chance of belonging to either the first or the second group. According to this analysis, two groups were defined: mean age at onset up to 30 years of age (Group A, n ⫽ 75 pedigrees) and older than 30 years (Group B, 163 pedigrees). There was no significant difference between Groups A and B for the gender of the interviewed persons (2 test for heterogeneity of proportions, 2 ⫽ 0.035; df ⫽ 1; p ⫽ 0.85). Complex Segregation Analysis of Groups A and B GROUP A. The MPMG is an autosomal dominant model (AIC, 230.34) with a significant multifactorial component (⫽ mixed model, dominant). This model is equivalent to the codominant mixed model that converged to identical parameter values (AIC, 232.34). Comparing the MPMG with a simple multifactorial model (AIC, 238.03), we found that the MPMG is significantly more likely (2 ⫽ 11.69; df ⫽ 2; p ⬍ 0.003). In contrast, the MPMG could not be rejected against the general model (2 ⫽ 2.90; df ⫽ 3; p ⬍ 0.4080). The hypothesis of environmental transmission is less likely than the general model (2 ⫽ 14.59; df ⫽ 2; p ⬍ 0.0007). The MPMG explains the segregation in Group A significantly better than a corresponding autosomal dominant model Winkelmann et al: Restless Legs Syndrome 299 without a multifactorial (h2 ⫽ 0) component (2 ⫽ 4.19; df ⫽ 1; p ⬍ 0.0407). The hypotheses of no major gene, a recessive gene, and environmental transmission converge to an identical model. The simultaneous failure to reject mendelian transmission probabilities and the successful rejection of environmental transmission as well as the rejection of the nomajor-gene hypothesis fulfills the criteria required for the conclusion that a single major gene exists (Table). The genetic model found in Group A consists of an allele frequency of 0.003, complete penetrance in heterozygotes and homozygotes, and a phenocopy rate of 0.005. The estimation of the portion of intragenotype (at the major locus) variability (h2) attributable to multiple genes is maximized at 100% (h2 ⫽ 1). This implies that an additional genetic (multifactorial) influence besides the single major gene in determining disease status must be allowed for. GROUP B. Group B stands in contrast with Group A. The most parsimonious model is the general model (AIC, 354.46). All other models are rejected at a significance level of 0.01 or better. Therefore, there is no evidence for a major gene effect. Table. Genetic Models Tested in Group A, Group B, and in Whole Sample Model Group A Mean age at onset up to 30 yr of age No major gene Mixed model, dominant Mixed model, recessive Mixed model, codominant Dominant Recessive Codominant Environmental Environmental ⫹ multifactorial General (dominant) Group B Mean age at onset older than 30 yr No major gene Mixed model, dominant Mixed model, recessive Mixed model, codominant Dominant Recessive Codominant Environmental Environmental ⫹ multifactorial General (dominant) Whole sample No major gene Mixed model, dominant Mixed model, recessive Mixed model, codominant Dominant Recessive Codominant Environmental Environmental ⫹ multifactorial General (dominant) f2 AA AB ⫽f1 ⫽f2 ⫽f1 1d ⫽f2 ⫽f1 0.868 0.271 0.068 ⫽f2 ⫽f1 1d —e 1d 0.868 0.738 0.868 0.271 0.068 1d —c b b 1b b b b ⫽p ⫽p 1d —c [0.5] [0.5] [0.5] [0.5] [0.5] [0.5] ⫽p ⫽p 0.652 0.031 1E-04 0d 0.01 1E-04 0.008 1E-04 1d 0 0.016 ⫽f1 ⫽f2 ⫽f1 1d ⫽f2 ⫽f1 0.723 0.61 0d,e ⫽f2 ⫽f1 0.576 —e 1d 0.721 0.971 1d 0.136 §0.031d,e 1d —c b b b b b b ⫽p ⫽p 1d 0.042 0.007 0.042 0.007 5E-05 0d 7E-05 0.171 —e 0.023 ⫽f1 ⫽f2 ⫽f1 1d ⫽f2 ⫽f1 0.774 0d —e ⫽f2 ⫽f1 1d —e 1d 0.773 0.803 1d 0.877 0.042 1d —c b b b b b b ⫽p ⫽p 1d q f0 f1 b 0.003 0d 0.003 0.001 0.376 0.001 0.05 0.399 0.001 0.068 0.005 0.068 0.005 0.001 0d 0.001 0.271 0.068 0d b 2E-04 0d 0.001 1E-04 0.206 2E-04 0.935 0d 0.032 b 0.002 0d 0.002 2E-04 0.28 1E-04 0.439 1d 0.042 BB h2 Parameter ⫺2InL⫹C AIC⫹Ca —c       ⫽p ⫽p *0 1d 1d 1d 1d     1d 1d 2 4 4 5 3 3 4 4 5 7 234.03 222.34 234.03 222.34 226.53 244.67 226.53 262.96 234.03 219.44 238.03 230.34 242.03 232.34 232.53 250.67 234.53 270.96 244.03 233.44 —c [0.5] [0.5] [0.5] [0.5] [0.5] [0.5] ⫽p ⫽p 1d —c       ⫽p ⫽p 0d 1d 1d 1d 1d     1d 1d 2 4 4 5 3 3 4 4 5 7 362.75 371.82 362.75 358.28 380.39 377.39 380.34 440.71 362.75 340.46 366.75 379.82 370.75 368.28 386.39 383.39 388.34 448.71 372.75 354.46 —c [0.5] [0.5] [0.5] [0.5] [0.5] [0.5] ⫽p ⫽p 1d —c       ⫽p ⫽p 0.142 1d 1d 1d 1d     1d 1d 2 4 4 5 3 3 4 4 5 7 600.38 585.73 600.38 585.73 610.32 625.52 610.32 706.48 600.38 577.95 604.38 593.73 608.38 595.73 616.32 631.52 618.32 714.48 610.38 591.95 Column 1 shows all genetic models tested for each of the groups considered. The first is the no-major-gene hypothesis (equivalent to a multifactorial hypothesis), in which all the variation for the trait is concentrated on the intragenotype variation parameter h2. Next are mixed models in which in addition to a major gene, there is a multifactorial component (h2 is freely estimated within its boundaries, ranging from 0 to 1). These models are followed by major gene models without a multifactorial component (h2 ⫽ 0). These are followed by the models on environmental transmission (h2 ⫽ 0) and environmental ⫹ multifactorial component (h2 estimated freely). The final model is the general model, the most prominent feature of which is that the transmission probabilities are allowed to vary unrestrained. a AIC serves as a weighted measure of the fit of any given model. The lower the AIC, the more parsimonious is the hypothesis.22 Parameter values in brackets are values fixed because of the specification of the model. c Parameter value irrelevant for the model. d Parameter value estimated at a boundary of its possible range. e The particular value of a parameter due to the fixation of another parameter at a boundary is no longer needed for the calculation of the likelihood. In this case, the last value obtained before the fixation is given. b q ⫽ frequency of deleterious allele; f0, f1, f2⫽ the penetrances for carriers of 0, 1, and 2 copies of the deleterious allele, respectively; AA, AB, BB ⫽ transmission probabilities; h2 ⫽ proportion of intragenic variance attributable to polygenic effects. AIC ⫽ Akaike information criteria. 300 Annals of Neurology Vol 52 No 3 September 2002 Whole Sample The most parsimonious model is a general model (AIC, 591.95). All other models are rejected at the 5% level, although note that the rejection of the dominant mixed model is borderline (2 ⫽ 7.78; df ⫽ 3; p ⫽ 0.0508). An overall test for heterogeneity between the results of the segregation analysis in the different age strata was highly significant (2 ⫽ 18.05; df ⫽ 7; p ⬍ 0.0118). Discussion This study provides strong evidence for a difference in the mode of inheritance of RLS between families with an early versus a late age at onset in a series of consecutive, unselected RLS index patients. Families with a mean age at onset of up to 30 years of age showed strong evidence for an autosomal dominant mode of inheritance transmitted by a single major gene acting in a highly penetrant fashion. Besides a single major gene in the early age at onset families, an additional multifactorial component is a part of this model. This implies that further genetic susceptibility factors determine whether nonallele carriers (phenocopies) also may become affected. The codominant model, in which heterozygote and homozygote effects can be distinguished, cannot be excluded but is less plausible. In contrast, in families with a mean age at onset above 30 years, the familial occurrence of RLS did not follow either a clear Mendelian or a multifactorial mode of inheritance. This suggests either an oligogenic mode of inheritance with an involvement of a few major genes or much stronger environmental influences. We excluded secondary RLS index patients, but because of the nature of the study we cannot exclude the occurrence of secondary RLS in family members. However, secondary RLS patients are more likely to have an older age at onset,8 and their families would be represented in Group B of our analysis. This further supports the exclusive existence of a single major gene transmitted by an autosomal dominant mode of inheritance in early age at onset families. The frequency of 45% definite and an additional 22% probable familial RLS cases is in line with previous studies displaying a positive family history in 43%6,7 to 64%.8 To our knowledge, this is the first study, however, that investigated all available firstdegree relatives by personal direct interview even if the index patients were primarily not aware of RLS symptoms in their relatives.6 – 8 We cannot exclude an ascertainment bias of the relatives depending on their degree of affection and their awareness of the disease. This also might explain the reluctance of family members to participate. Furthermore, none of the spouses received a diagnosis of RLS that would not have been expected in view of its prevalence in recent populationbased studies. We ourselves cannot explain this. Conclusions such as “assortative mating in RLS” are too speculative at the moment and need to be investigated in more detail. Besides the known variable expressivity of the symptoms within a single family,11–13 an earlier age at onset of the disease in the younger generation has been reported in three large RLS families, suggesting the possibility of anticipation.11,13 Comparing all these studies, one has to consider that recalling sensory symptoms retrospectively depends on subjective factors and may affect the ascertainment of the age at onset of the disease. For the penetrance of RLS, a twin study showed a high concordance rate of approximately 83%.25 Investigation of the parents suggested an autosomal dominant inheritance with a high penetrance.25 In our study, however, the clear difference in the mode of inheritance between early- and late-onset RLS families argues for causative heterogeneity between these two age groups. Similar patterns of inheritance have been found in other diseases with “complex” inheritance, for example, in breast cancer,26 Alzheimer’s disease,27 or Parkinson’s disease.28 An age at onset–related differentiation of probands provides a new approach for future linkage studies that may shed more light on the genetics and the pathophysiology of this disorder. This work was supported by a Habilitations Förderpreis grant from the Bavarian Ministry of Science, Culture and Art (J.W.). We are indebted to all patients and family members who participated in this study and to Dr Catherine Bonaiti-Pellie for her valuable comments on this manuscript. References 1. Trenkwalder C, Walters AS, Hening W. Periodic limb movements and restless legs syndrome. Neurol Clin 1996;14: 629 – 650. 2. Walters AS. The International Restless Legs Syndrome Study Group. Toward a better definition of the restless legs syndrome. Mov Disord 1995;10:634 – 642. 3. Chokroverty S. Jankovic J. Restless legs syndrome: a disease in search of identity. Neurology 1999;52:907–910. 4. Rothdach AJ, Trenkwalder C, Haberstock J, et al. Prevalence and risk factors of RLS in an elderly population: the MEMO study. Neurology 2000;54:1064 –1068. 5. Phillips B, Young T, Finn L, et al. Epidemiology of restless legs symptoms in adults. Arch Int Med 2000;160:2137–2141. 6. Montplaisir J, Boucher S, Poirier G, et al. Clinical, polysomnographic, and genetic characteristics of restless legs syndrome: a study of 133 patients diagnosed with new standard criteria. Mov Disord 1997;12:61– 65. 7. Winkelmann J, Wetter TC, Collado-Seidel V, et al. Clinical characteristics and frequency of the hereditary restless legs syndrome in a population of 300 patients. Sleep 2000;23: 597– 602. Winkelmann et al: Restless Legs Syndrome 301 8. Ondo W, Jankovic J. Restless legs syndrome: clinicoetiologic correlates. Neurology 1996;47:1435–1441. 9. Montagna P, Coccagna G, Cirignotta F, Lugaresi E. Familial restless legs syndrome: long-term evolution. New York: Raven Press, 1983:231–235. 10. Montplaisir J, Godbout R, Boghen D, et al. Familial restless legs with periodic movements in sleep: electrophysiologic, biochemical, and pharmacologic study. Neurology 1985;35:130 –134. 11. Trenkwalder C, Collado-Seidel V, Gasser T, Oertel WH. Clinical symptoms and possible anticipation in a large kindred of familial restless legs syndrome. Mov Disord 1996;11:389 –394. 12. Walters AS, Picchietti D, Hening W, Lazzarini A. Variable expressivity in familial restless legs syndrome. Arch Neurol 1990; 47:1219 –1220. 13. Lazzarini A, Walters AS, Hickey K, et al. Studies of penetrance and anticipation in five autosomal-dominant restless legs syndrome pedigrees. Mov Disord 1999;14:111–116. 14. Walters AS, Hickey K, Maltzman J, et al. A questionnaire study of 138 patients with restless legs syndrome: the “NightWalkers” survey. Neurology 1996;46:92–95. 15. Allen RP, Earley CJ. Defining the phenotype of the restless legs syndrome (RLS) using age-of-symptom-onset. Sleep Med 2000; 1:11–19. 16. Desautels A, Turecki G, Montplaisir J, et al. Identification of a major susceptibility locus for restless legs syndrome on chromosome 12 q. Am J Hum Genet 2001;69:1266 –1270. 17. Wittchen H-U, Pfister H. Dia-X-Interviews: Manual für Screening-Verfahren und Interview; Interviewheft Längsschnittuntersuchung (DIA-X-Lifetime); Ergänzungsheft (Dia-xLifetime); Interviewheft Querschnittuntersuchung (Dia-X-12 Monate); Ergänzungsheft (Dia-X- 12 Monate); PC-Programm zur Durchführung des Interviews (Längs-und Querschnittsuntersuchung); Auswerteprogramm. Frankfurt: Swets & Zeitlinger, 1997. 302 Annals of Neurology Vol 52 No 3 September 2002 18. Hasselblad V. Estimation of parameters for a mixture of normal distributions. Technometrics 1966;8:431– 444. 19. “pedigree analysis package” PAP, version 4, 1994. ftp://ftp.genetics.utah.edu/pub/software/pub 20. Hassstedt SJ. A variance components/major locus likelihood approximation for quantitative, polychotomos, and multivariate data. Genet Epidemiol 1993;10:145–158. 21. Thompson EA, Cannings C. Sampling schemes and ascertainment. Prog Clin Biol Res 1979;32:363–382. 22. Akaike H. A new look at the statistical model identification. IEEE Trans Automatic Control AC 1974;19:716 – 723. 23. Williams WR, Anderson DE. Genetic epidemiology of breast cancer: segregation analysis of 200 Danish pedigrees. Genet Epidemiol 1984;1:7–20. 24. Khoury MJ, James LM. Population and familial relative risks of disease associated with environmental factors in the presence of gene-environment interaction. Am J Epidemiol 1993;137: 1241–1250. 25. Ondo WG, Vuong KV, Wang Q. Restless legs syndrome in monocygotic twins: clinical correlates. Neurology 2000;55: 1404 –1406. 26. Newman B, Austin MA, Lee M, King MC. Inheritance of human breast cancer: evidence for autosomal dominant transmission in high-risk families. Proc Natl Acad Sci USA 1988;85: 3044 –3048. 27. Rao VS, van Duijn CM, Connor-Lacke L, et al. Multiple etiologies for Alzheimer disease are revealed by segregation analysis. Am J Hum Genet 1994;55:991–1000. 28. Gasser T, Muller-Myhsok B, Wszolek ZK, et al. Genetic complexity and Parkinson’s disease. Science 1997;277:388 – 389.