ORIGINAL RESEARCH ARTICLE. H Ewald 1,2, T Flint 2, TA Kruse 3 and O Mors 1. Subjects and methods

Transcription

1 (2002) 7, Nature Publishing Group All rights reserved /02 $ ORIGINAL RESEARCH ARTICLE A genome-wide scan shows significant linkage between bipolar disorder and chromosome 12q24.3 and suggestive linkage to chromosomes 1p22 21, 4p16, 6q14 22, 10q26 and 16p13.3 H Ewald 1,2, T Flint 2, TA Kruse 3 and O Mors 1 1 Department of Psychiatric Demography and 2 Department of Biological Psychiatry, Institute for Basic Psychiatric Research, Psychiatric Hospital in Aarhus, Risskov, Denmark; 3 Department of Clinical Biochemistry and Genetics, Odense University Hospital, Denmark The present study reports a genomewide scan using linkage analysis for risk genes involved in bipolar disorder with 613 microsatellite markers including additional testing of promising regions. As previously published significant linkage was obtained at chromosome 12q24.3 with a two-point parametric lod score of 3.42 at D12S1639 including all members in both families (empirical P-value , genome-wide P-value ). The multipoint parametric lod score at D12S1639 was 3.63 (genome-wide P-value ). At chromosome 1p22 p21 a parametric, affecteds-only two-point lod score of 2.75 at marker D1S216 was found (empirical P- value , genome-wide P-value ). A three-point lod score of 2.98 (genome-wide P- value ) at D1S216, and a multipoint non-parametric analysis with a maximum NPL-all score of (P-value ) at D1S216 further supported this finding. A number of additional loci on chromosomes 4p16, 6q14 q22, 10q26 and 16p13.3 yielded parametric lod scores around or above 2. (2002) 7, doi: /sj.mp Keywords: manic depressive illness; bipolar disorder; chromosome 12; chromosome 1; linkage Introduction Bipolar disorder is a severe and common psychiatric disorder with a lifetime risk around %. The disease is characterized by affective episodes with manic, depressive and other psychiatric symptoms. Mania and melancholia have been known for more than 2300 years and are two of the earliest described human diseases. 1 The evidence for the involvement of genes in the etiology of bipolar disorder is based on family and twin studies of several thousand persons, and the genetic contribution seems stronger than for many other complex disorders. Since the first association and linkage studies of bipolar disorder were performed, around four and three decades ago respectively, 2 4 developments in diagnostic instruments and criteria, molecular genetics, computer programs and statistics have helped to identify more than 10 candidate chromosome regions potentially containing genes which increase susceptibility. The regions on chromosomes 4p16, q24, q32, 13,14 18p, q, q, and possibly Xq have been supported in a number of studies. Other positive findings have been reported on chromosome 4q35, 33,34 chromosome 10p 35 and chromosome Some regions have received further support from co-occurrence of bipolar disorder with monogenic disorders or cytogenetic abnormalities. 37 Linkage analyses of families with schizophrenia have also supported some of these regions. 38,39 The present study reports results from a genomewide scan and additional further testing of promising regions, including 613 unique DNA markers on Danish families with bipolar disorder. Markers from a number of chromosome regions, including chromosome 4p16, 12q24, 16p13.3 and 18, have been published previously. 8,10,40,41 Subjects and methods Correspondence: H Ewald, Consultant, Associate Professor, DrMedSc, Department of Psychiatric Demography, Institute for Basic Psychiatric Research, Psychiatric Hospital in Aarhus, Skovagervej 2, 8240 Risskov, Denmark. he psykiatri.aaa.dk Received 4 September 2001; revised 4 October 2001; accepted 21 November 2001 Family ascertainment and diagnosis Families were identified through bipolar probands in contact with the Psychiatric Hospital in Aarhus, Denmark. Two families with affected members in several generations were included in the study (Figure 1).

2 Linkage between bipolar disorder and chromosome 12q Figure 1 The haplotypes for markers D1S209, D1S3467, D1S2829, D1S216 and D1S1728 shown for relevant members of (a) family 4 and (b) family 2 using identical individual numbering as in previous studies. 8 The suggested disease haplotype is shown boxed and shaded. In family 4 a common haplotype including markers D1S3467 D1S2829 D1S216 of at least 7.3 cm is probably shared by all affected persons except person III 25 and the bilineal subbranch (V 2, V 2, V 3) and their father (IV 19). Recombinations between marker alleles are shown with an X.

3 Linkage between bipolar disorder and chromosome 12q IV). No one else in the blood line was diagnosed as having schizophrenia, schizotypal, delusional or schizoaffective disorders. In general, individuals who married into the families were excluded, along with their descendants if they themselves or their first degree relatives had bipolar disorder or any other major psychiatric disorder. An inbreeding loop was allowed in family 2, however. In family 4 the bilineal subbranch including person IV-19 and the three children V-2, V-3 and V-4 were not included in the statistical analyses except in additional analyses of seven markers as mentioned below. Persons with primary alcohol or substance abuse, and their descendants were also excluded from the study. Sex and birth order have been concealed to ensure confidentiality. Figure 1 Continued. Both families were of Danish Caucasian origin and from the same geographic region in Aarhus county. To obtain diagnoses, family members were interviewed by HE and OM using the semi-structured interview Present State Examination, 10th edition. 42 Further diagnostic information was obtained from relatives, hospital records, and general practitioners. A senior psychiatrist experienced in ICD-10 and DSM-IV, Dr A Bertelsen, who did not have access to the genealogical data or laboratory results, made the final best estimate diagnoses in accordance with ICD-10 Diagnostic Criteria for Research (DCR) 43 and DSM-IV. 44 Further details of the study concerning diagnostic procedure, attempts to minimise misclassification, interrater reliability, admissions and treatment have been reported earlier. 45 Diagnoses according to ICD-10 DCR can be seen in Figure 1. Clinical information was updated January 2001 and the following individuals changed diagnoses compared to the information given in Ewald et al 45 : V-5 and III- 25, family 4, from bipolar affective disorder with hypomania to bipolar affective disorder with mania (ICD- 10) and from bipolar II disorder to bipolar I disorder (DSM-IV). Individual IV-1, family 4, changed from bipolar affective disorder (ICD-10) to schizoaffective disorder, manic and depressed type (ICD-10 and DSM- Genotyping DNA was prepared from whole blood, using a standard triton lysis, nuclear lysis protocol with sodium chloride/isopropanol precipitation. A total of 666 DNA markers were genotyped, of which 613 were unique. With a total genome length around 3593 cm the average spacing between markers was 5.9 cm. The largest gap between any pair of markers was 12.9 cm. The position of and distances between markers are based on the Marshfield map ( org/genetics/). Additional markers were typed in areas with lod scores above 1 and in areas with a priori evidence for risk genes involved in bipolar affective disorder. Of the DNA markers 300 were genotyped in Aarhus, Denmark. For these markers DNA amplification was employed using a Perkin-Elmer thermocycler (Applied Biosystems, Foster City, CA, USA), with optimised reaction conditions for each primer set, and analyzed on the ABI Prism 310 Genetic Analyzer (Applied Biosystems). Three hundred and forty DNA markers were genotyped by Généthon, Paris, France 46 as part of the European Science Foundation (ESF) program On the Molecular Neurobiology of Mental Illness. Finally, 26 DNA markers were genotyped by other collaborators. When comparing the 1558 genotypes from 41 markers which were typed in both Aarhus and at Généthon, a total of 24 genotypings were discrepant corresponding to a 0.75% error rate. Seven errors were made in Aarhus, 13 were made by Généthon and four were of uncertain origin. The following markers D1S216, D4S394, D10S217, D12S1639 and D16S510 were genotyped twice in Aarhus, and no errors were found in the 270 genotypes which were typed twice except for marker D16S510. For marker D16S510 person III-25 in family 4, which was originally conservatively stated to be 1 1 as we were unable to unambiguously identify the second allele, 40 was found to be 1 5 by retyping this marker in both families. This corresponds to an error rate of 0.19%. Data analysis Two phenotypic models based on a nested diagnostic hierarchy were constructed. A narrow model included

4 as affected: individuals diagnosed as bipolar disorder, single episode mania or schizoaffective disorder, manic and depressed type. This phenotypic model was used in model 1 in the parametric linkage and the nonparametric linkage analyses mentioned below. A broader phenotype included, in addition to the individuals included by the narrow model, individuals with at least two depressive episodes of at least moderate severity. This phenotypic model was applied in the parametric linkage analyses model 2 and recessive model, and model 2 in the non-parametric linkage analyses. For all 613 markers, the following analyses were performed: two-point parametric linkage analyses using LINKAGE version and non-parametric multipoint analyses, NPL-all scores, using GENEHUNTER software package version A total of eight different analyses were performed on each of the 613 markers. Parametric two-point linkage analyses including all or only affected family members were calculated applying dominant model 1, 2 and the recessive model, and nonparametric multipoint NPL-all analyses were calculated for phenotypic models 1 and 2. As the phenotypic and dominant models overlap, the eight tests performed are not independent. Both families were primarily analysed together. (a) Lod score analyses Two-point lod scores were calculated using the computer program LINKAGE version 5.1. Corresponding empirical P-values were estimated using SIMULATE 49 and the genome-wide P-values were calculated according to Sham 50 as discussed below. The parametric lod score analyses were performed applying two dominant and one recessive model(s) with definitions of affected status based on a diagnostic hierarchy as mentioned above. We restricted the criteria for bipolar disorder by excluding individuals with only mild episodes (hypomania and mild depression). Conservative values of 0.60, 0.65 and 0.65 for maximum penetrance were chosen for dominant models 1 2 and the recessive model respectively. The penetrances for persons homozygous for the normal allele were set relatively high and age dependent, adding up to 0.005, 0.02 and 0.02 for dominant models 1 2 and the recessive model respectively. The exact age-dependent penetrances for the different models have been reported earlier. 45 Using the LINKAGE package we also performed an affecteds-only analysis for all models, with the same penetrance ratios for the affected individuals as in the models mentioned above. In these analyses all unaffected individuals are considered to have unknown phenotype, ie there is no possibility of including false negative cases. If the disease is caused by the joint effect of a number of genes, this also excludes unaffected individuals who may only carry the disease allele(s) at one of the loci. 51 The frequency of the disease allele was set to for model 1, to 0.01 for model 2 and to 0.1 in the recessive model. Individuals with affective disorders other Linkage between bipolar disorder and chromosome 12q24.3 than those included by the model in question were scored as phenotype unknown when analysing that model. The two-point analyses were performed with MLINK for all markers using identical phenotypic, genetic models and parameters as reported earlier. 45 Allele frequencies for the markers genotyped in Denmark were calculated from individuals who married into the families, while the marker allele frequencies from the ESF project mentioned above were estimated from all participating European families. Though some marker allele frequencies were determined from relatively few individuals, the power to detect true linkage should not be greatly affected. 52 In addition, for five markers yielding the highest lod scores allele frequencies were also determined from 114 Danish controls. (b) Non-parametric analyses Non-parametric multipoint linkage analysis were performed using the GENE- HUNTER software package version GENE- HUNTER can perform nonparametric multipoint analyses with a large number of markers but only analyse pedigrees of relatively limited size. It was possible to analyse family 2 as a three-generational family without the inbreeding loop. Family 4 was analysed including all affected individuals according to phenotypic model 1. When analysing family 4 with phenotypic model 2 it was necessary to exclude a few individuals. It was decided to base these exclusions on late age of onset for the patients with recurrent depression. The age of onset was 18, 35, 65 and 51 for persons IV-3, IV- 11, III-14 and III-22 in family 4 respectively, and 24 for person IV-1 in family 2, and therefore persons III-14 and III-22 were excluded from family 4 when analysing model 2. The calculations were performed on a computer from Silicon Graphics, Origin 2000, with 64 MIPS R12000 CPUs, 64 GB physical memory (RAM) and a 1000 GB disk. The NPL multipoint scores from GENEHUNTER were obtained using the all pairs function as this method considers all affected relatives simultaneously instead of just pairs. (c) Further analyses of markers with lod scores at or above 2 In order to investigate the robustness of the parametric lod score findings, a number of additional analyses were performed in six regions, 1p22 21, 4p16, 6q14 22, 10q26, 12q24.3 and 16p13.3, in which parametric two-point lod scores around or above 2 were found. In these chromosome regions three-point parametric lod scores were calculated using LINKMAP from the LINKAGE package. For the markers from each region which yielded the highest two-point parametric lod score; D1S216, D4S394, D6S1021, D6S1009, D10S217, D12S1639 and D16S510, a number of additional analyses were performed one at a time. In these analyses either: (a) the bilineal subbranch from family 4 was included; or (b) each affected person was coded phenotype unknown 737

5 738 Linkage between bipolar disorder and chromosome 12q24.3 one at a time. Furthermore for markers D1S216, D4S394, D10S217, D12S1639 and D16S510, we investigated if the lod scores were different if based on marker allele frequencies from the Danish population, determined from 114 psychiatrically unscreened Danish controls, rather than when calculated from persons who married into the Danish or European families. When the bilineal subbranch of family 4 was added, the three siblings with bipolar I disorder, V-2, V-3 and V-4, and both of their ill parents from family 4, were included in the phenotypic model. These patients were entered into the same liability class as the other affected persons, ie without making any special assumptions, when calculating the two-point parametric lod score analyses using LINKAGE. This approach is unbiased. It has been shown that including bilineal pedigrees, of similar size and structure to unilineal pedigrees, at most leads to some loss of power under homogeneity. 53 We did not include this bilineal subbranch when analysing all 613 DNA markers as this could falsely decrease the lod score if any of the three children wrongly were assumed to be recombinants. As suggested by Hodge and Greenberg 54 two-point lod score analyses were also performed after consecutively coding each affected person as phenotype unknown in order to identify patients who, if misclassified as false positives, had a major influence on the lod scores in either direction. In these analyses the same phenotypic and genetic models was used as when calculating the original lod scores. Power considerations Concerning the parametric lod scores it was possible to use SLINK 55,56 to investigate the power of the families, given the diagnostic models and genetic parameters chosen, and to use SIMULATE 49 to estimate the risk of obtaining positive lod scores with an unlinked marker. For both analyses replicates of the families were simulated with a marker assumed to have four equally frequent alleles as this reflects the informativity of microsatellites. The simulations using SLINK with a linked marker showed reasonable power. For the dominant model 1 and 2 for both families combined, 72% and 85% of lod scores were above 1. Similarly, for family 4, 56% and 71% of the lod scores were above 1. In the present study we decided to investigate additional markers in regions with lod scores above 1. The simulations with an unlinked marker using SIMULATE showed that for models 1, 2 or the recessive model, lod scores above 1 or 2 occurred with a frequency less than 0.02 and in the two families combined. Results All parametric two-point lod scores above 1.9 are shown in Table 1. All positive parametric two-point lod scores for models 1 and 2 are shown in Figure 2. In the two families combined the highest lod scores were 3.42 at D12S1639 and 2.75 at D1S216 (Table 1). In family 4 the highest lod scores were 2.53 at D11S935, 2.31 at D6S1021, 2.24 at D1S216 and 2.21 at D12S1639. In family 2 the highest parametric twopoint lod scores were 1.44 at D4S1551, 39 cm from the p-telomere, 1.37 at D8S136, 44 cm from the p-telomere, both including all and assuming a recessive mode of inheritance. In the affecteds-only analyses, only marker D21S1903 at 54 cm yielded a lod score above 1 in family , assuming a recessive mode of inheritance. NPL-all scores were calculated every two cm throughout the genome. Figure 2 shows minus the logarithm of the P-values corresponding to the NPL-all scores obtained from GENEHUNTER version 2.1. All markers from any specific chromosome could be analysed simultaneously, ie up to 44 marker loci as for chromosome 1. When including both families the highest NPL-all scores and related P-values were at 104 cm from the p-telomere on chromosome 1, 4.09 (P-value ) and (P-value ) for model 1 and 2 respectively. The second highest NPL-all scores combining both families were on chromosome 12q24.3 at 148 cm from the p-telomere, 2.29 (P-value ) and 2.05 (P-value ). Only two other regions produced NPL-all scores above 2 when combining both families; 2.14 (P-value ) at 206 cm on chromosome 3 and 2.10 (P-value ) at 58 cm on chromosome 9, both for model 1. The two highest NPL-all scores were 6.15 (P-value ) at 102 cm and (P-value ) at 104 cm from the p-telomere on chromosome 1 in family 4 for models 1 and 2. Only two other regions produced NPL-all scores above 2 in family 4; 3.40 (P-value ) at 206 cm from the p- telomere on chromosome 3 and 2.60 (P-value ) at 156 cm from the p-telomere on chromosome 10, both for model 1. In family 2 the highest NPL-all score was 2.72 at 148 cm from the p-telomere on chromosome 12q24.3 for model 2. This region only received some support from family 4 with maximal NPL-all scores of 0.71 (P-value ) and 0.18 (P-value ) for models 1 and 2 respectively also at 148 cm. Three-point analyses using LINKMAP including markers at or near the seven markers were performed applying the same phenotype and mode of inheritance that produced the highest two-point lod score. The highest lod score was 3.63 at D12S1639 for model 1, increasing to 4.19, when including the bilineal subbranch of family 4 in the three-point analyses. The second highest lod score was 2.98 at D1S216 for model 2 and affecteds-only analyses when including both families. For D12S1639 and D1S216 the three-point lod scores were higher than the two-point lod score and these loci were thus supported by the additional genetic information obtained from the multi-point analyses. A lod score of 2.52 at D6S1009 for model 1 and a lod score of 2.19 at D16S510 for the recessive model when including both families and all family members were found. Both of these lod scores were quite similar to the two-point lod scores at these markers. A lod score of 1.96 at 0.10 recombination fraction proximal to D10S217 was found for model 2 when including all affected family members in family 4. Finally, an affecteds-only lod score of 1.51 at D6S1021 was found for

6 Linkage between bipolar disorder and chromosome 12q24.3 Table 1 All parametric two-point lod scores above 1.9, when including all or when using affecteds-only analyses, in both families or in family 4 or family 2 are shown. 739 Marker cm Lod score Theta Model Families Including D1S both affecteds-only D1S family 4 affecteds-only D1S family 4 affecteds-only/all D3S family 4 affecteds-only D4S R both all D6S both affecteds-only D6S both all D6S family 4 affecteds-only D6S both affecteds-only D6S family 4 all D6S both all D10S family 4 all D11S family 4 affecteds-only/all D11S both affecteds-only D12S both all D12S family 4 all D12S both all D12S both all D16S R both all Model 1 includes as affected: individuals diagnosed as having bipolar disorder, single episode mania or schizoaffective disorder, manic and depressed type. Model 2 and the recessive model: as model 1 and including individuals with at least two depressive episodes of at least moderate severity. The markers position from the p-telomere is shown in cm. model 2 including both families and a lod score of 1.04 at 0.10 recombination fraction proximal to D4S394 was found, when including both families and all family members, and assuming a recessive mode of inheritance. Empirical P-values obtained from SIMULATE corresponding to the original parametric two-point lod scores for the seven markers are shown in Table 2, when using the marker allele frequencies estimated from persons marrying into the families. Caution is required when interpreting results from SIMULATE as the P-values obtained with this program will depend on the number of untyped individuals in the pedigree. As pointed out by Sham 50 low empirical P-values may be obtained with Monte Carlo methods when the observed data have much greater informativity than expected, given the pedigree structure and marker allele frequencies. Further additional parametric two-point lod scores for the seven markers are also shown in Table 2. No major change in lod scores was seen when using population-based marker allele frequencies. At D12S1639, adding the bilineal subbranch increased the two-point lod score to 3.89 and the three-point lod score to The lod scores at the other markers were either mostly unchanged or decreased (D1S216) when adding this subbranch. Concerning the influence on the lod score when affected individuals one at a time were coded phenotype unknown, it is seen that D1S216 was most robust and at most decreased 0.3. Furthermore the same individual, IV-1, led to an increase in the lod score at D4S394 and D12S1639 when coded phenotype unknown. The most likely haplotypes were constructed by hand using a minimum recombination strategy and by GENEHUNTER. Close markers were checked for double recombinants and for four markers, which did not yield interesting lod scores, a few individuals were coded as genotype unknown. The haplotypes for the markers on chromosome 1p22 p21 are shown in Figure 1. The parametric lod scores and haplotypes or marker alleles on chromosomes 12q24.3, 16p13.3, and 4p16 have been published previously. 8,10,41 Discussion We have chosen to apply parametric two-point linkage analyses as well as non-parametric multipoint analyses. We have investigated two families and thus have limited power to find weaker risk genes and to search for heterogeneity among families. Testing a large number of markers we have found a higher number of positive findings than expected. For families a lod score threshold of 3.3 has been suggested for significant linkage corresponding to an exact genome-wide significance threshold of P = Similarly, a lod score of around 1.86 would be expected to occur once by chance in a genome-wide scan. 58 These suggestions are based on a completely informative marker map. If a less dense marker map is used lower lod scores will correspond to the P-value thresholds suggested above. 59 At chromosome 12q24.3 a significant lod score by

7 Linkage between bipolar disorder and chromosome 12q Figure 2 Parametric two-point lod scores for both families combined plotted for each of the 613 DNA markers at zero recombination fraction: (a) model 1 including all individuals; (b) model 1 only including affected individuals; (c) model 2 including all individuals; (d) model 2 only including affected individuals. Furthermore minus log 10 of the P-values corresponding to the NPL-all scores using GENEHUNTER are also shown: (e) model 1; (f) model 2.

8 Linkage between bipolar disorder and chromosome 12q Figure 2 Continued. Table 2 The original parametric two-point lod scores and corresponding empirical P-values obtained using at least replicates of both families with SIMULATE Marker Original lod Families and Empirical Control allele Plus four Phenotype Phenotype score model P-values frequencies bipolar patients unknown highest unknown lowest D1S , theta 0.00 afo, both, M , theta , theta , III , IV 2 D4S , theta 0.05 both, REC , theta , theta , IV , IV 18 D6S , theta 0.00 afo, both, M nt 2.41, theta , *IV 5, *IV , IV 1 D6S , theta 0.00 both, M nt 2.49, theta , III 25 D10S , theta 0.00 F4, M , theta , theta , IV 1 D12S , theta 0.01 both, M , theta , theta , IV , III 25 D16S , theta 0.00 both, REC , theta , theta , IV , IV 1 Additional lod score results from the markers were obtained using marker allele frequencies from population controls or including four additional patients with bipolar disorder. The highest and lowest lod score after consecutively coding each affected person as phenotype unknown is also shown. *is individuals in family 2. Afo is affecteds-only analyses, F4 is family 4, F2 is family 2, M1, M2 and REC are phenotypic models as in Table 1, nt is not tested. When including four bipolar patients from the bilineal subbranch of family 4, the same families and phenotypic models were used as when calculating the original lod score. The mod scores were calculated for the same families as the original lod score. Lander and Kruglyak s criteria for families was obtained. 57 This lod score was the highest in the two families combined when applying a narrow phenotype and including all family members and supported by other markers nearby (Table 1, Figure 2). 10 The lod score increased to 3.63 with three-point analyses and further to 4.19 when adding the bilineal subbranch. The genome-wide significance of lod scores of 3.42, 3.63 and 4.19 corresponds to P-values of , and according to formulas in Sham. 50 Nonparametric analyses with the NPL-all test also sup- ported this region, especially in family 2. True lod score peaks may be broader than false peaks. 60 In family 2 the NPL-all score was 2.53 for model 1 from cm and 2.72 for model 2 from cm from the p-telomere (Figure 2). This was the highest NPL-all score in family 2. In family 4 the NPL-all scores were lower with a maximal NPL-all score of 0.71 (P-value ) at 148 cm. Family 4 thus received less support than in the parametric lod score analyses. The lod score at D12S1639 including all family members in family 4 was 2.21 at theta 0.05, but only 0.83 at theta

9 742 Linkage between bipolar disorder and chromosome 12q in the affecteds-only analyses. The unaffected family members, who are not included in the NPL-all analyses, thus contribute substantially to the lod score and this may probably explain why the NPL-all scores are lower. We have found it reasonable to include unaffected persons in the parametric lod score analyses as unaffected individuals in family 4 are at least 45 years of age, and all except two are above 50 years, and have thus passed the 90th percentile for age of onset of bipolar disorder. Finally, this region has received support from previous studies of neighbouring populations from Great Britain, 61 the Faroe Islands 12 and from North American families. 11 In addition, a more proximal region including the distal part of chromosome 12q23 has been suggested. 9,62 The locus suggested on chromosome 1p22 p21 close to D1S216 at 104 cm from the p-telomere yielded the highest affecteds-only lod score of 2.75 (genome-wide P-value ). Two-point lod scores at other nearby markers supported the finding (Figure 2 and Table 1) and the lod score increased to 2.98 (genome-wide P- value ) when performing three-point analyses. Furthermore, this region was clearly the strongest supported region in the NPL-all analyses, also especially in family 4. A pointwise P-value around is expected to occur by chance slightly more than once every genome scan 57 and the findings on chromosome 1p22 p21 must be considered suggestive rather than significant. However, the significance of this finding is underestimated as the two persons with recurrent depression from family 4, whom it was necessary to exclude from the GENEHUNTER analyses when analysing model 2, probably carry the relevant haplotype (Figure 1). Finally, the P-values corresponding to the NPL-all scores obtained from the GENEHUNTER 2.1 program are conservative when, as in family 4, inheritance information is incomplete. When using population-based marker allele frequencies the lod score at D1S216 increased to A common haplotype was most likely present amongst the majority of cases in family 4 (Figure 1). The haplotype of interest was also present in a number of unaffected individuals possibly indicating that additional genetic or non-genetics risk factors were not present among these persons. Finally, this locus was rather robust to phenotypic misclassification. A single false positive case would not have any major effect as the lod score maximally dropped 0.3 when coding any affected individual as phenotype unknown. This region has been suggested in the NIMH Genetics Initiative Bipolar Pedigree study with lod scores around 2.5 peaking around 102 cm. 63 So, in contrast to the region on chromosome 12q24.3, this region has been supported by a genome-wide scan on a larger number of pedigrees. In addition, a lod score around 1.6 has been found at D1S224 in a genome-wide scan for risk genes for alcohol dependence. 64 On proximal chromosome 6q14 q22 the positive lod scores at D6S1021 and D6S1009 did not receive much support from the NPL-all analyses. The NIMH Genetics Initiative Bipolar Pedigree study found lod scores above 2 with maximum at D6S Studies of families with schizophrenia have supported this region On chromosome 16p13.3 the lod score at D16S510 was slightly lower than previously reported 45 as person II-25 was regenotyped as 1 5 instead of 1 1. The empirical P-value was quite low (Table 2). The region did not receive support from the NPL-all test. One reason for this may be that NPL-all test performs less well than the NPL-pair test for common recessive alleles as suggested by Haines. 68 This region has received limited support by a few studies as mentioned in Detera- Wadleigh. 61 On chromosome 10q26 a positive lod score was found at D10S217 in family 4 and the third highest NPL-all score in family 4 anywhere in the genome was 2.60 (P-value ), 2 cm proximal to D10S217. This region has received some support from other studies of bipolar disorder and schizophrenia. 72 On chromosome 4p16 we have previously reported a lod score of 2.00 at a theta of 0.03 from D4S394. This region were not supported by the non-parametric analyses. However, as mentioned above, NPL-all analyses may be less able to detect common recessive alleles. Several regions on chromosome 18 have been implied and reviewed by Potash and Depaulo Jr. 73 We have previously reported a lod score of 1.83 at D18S67 in family We did not find much support for this region with non-parametric analyses. A number of other regions received some support mostly from individual analyses. For model 1 in family 4, marker D3S1265 at 229 cm yielded an affecteds-only lod score of 1.90, and a NPL-all score of 3.40 (P-value ) was found at 206 cm on chromosome 3. On chromosome 11p12, marker D11S935 yielded the highest parametric two-point lod score in family 4. Both inspection of haplotypes and three-point analyses suggested that this was most likely a false positive finding as probable recombinants were present. Finally, the highest affecteds-only lod score in family 2 was 1.04 at D21S1903 at 53.6 cm, around 2 cm distal to PFKL where significant linkage has been reported and replicated. 25,28 This area did not receive any support from the non-parametric analyses. We believe that the above mentioned regions on chromosomes 1 and 12 are the most interesting found in our study as they are supported by parametric as well as non-parametric analyses. Concerning other less significant findings from our study the a priori evidence is by far strongest on chromosome 4p16. In family 4 the only affected individual diagnosed as schizoaffective disorder, manic and depressed type, person IV 1, reduces the lod score for D12S1639 and D4S394 (Table 2). If this person is a false positive case then a locus for bipolar disorder on chromosome 12q24.3 and a locus on 4p16 for bipolar disorder and recurrent depression could be implicated. This is, however, speculative. Acknowledgements The authors wish to thank director and senior psychiatrist Aksel Bertelsen, WHO Reference and Training

10 Centre for ICD-10 Classification of Mental and Behavioural Disorders, WHO Reference and Training Centre for Schedules for Clinical Assessment in Neuropsychiatry (SCAN), Department of Psychiatric Demography, Psychiatric Hospital in Aarhus, for thorough reading of all reports and medical records, as well as performing the diagnostic evaluations; and Agata El Daoud, Department of Biological Psychiatry, Psychiatric Hospital in Aarhus, for assisting with the genotypings. This work has been supported by the Danish Medical Council grant No and grant No , the Lundbeck Foundation, grant No. 21/91, grants from Psykiatrisk Forskningsfond, Ejner Geert-Jørgensens foundation, a grant from NovoCare (Novo Nordisk Farmaka Denmark) and Ingeborg and Leo Dannins foundation. We thank the organising committees of the European Science Foundation concerted action, The Molecular Neurobiology of Mental Illness and the scientific and technical staff of Généthon, Paris, France. References 1 Marneros AAJ. Bipolar disorders: roots and evolution. In: Marneros AAJ (ed). Bipolar Disorders 1000 Years After Manic-depressive Insanity. Kluwer Academic Publishers: Dordrecht, 2000, pp Parker JB, Theilie A, Spielberger CD. Frequency of blood types in a homogenous group of manic-depressive patients. J Mental Sci 1961; 107: Reich T, Clayton PJ, Winokur G. Family history studies: V. The genetics of mania. Am J Psychiatry 1969; 125: Winokur G, Tanna VL. Possible role of X-linked dominant factor in manic depressive disease. Dis Nerv Syst 1969; 30: Blackwood DH, He L, Morris SW, McLean A, Whitton C, Thomson M et al. A locus for bipolar affective disorder on chromosome 4p. Nat Genet 1996; 12: Ginns EI, St Jean P, Philibert RA, Galdzicka M, Damschroder- Williams P, Thiel B et al. A genome-wide search for chromosomal loci linked to mental health wellness in relatives at high risk for bipolar affective disorder among the Old Order Amish. Proc Natl Acad Sci USA 1998; 95: Asherson P, Mant R, Williams N, Cardno A, Jones L, Murphy K et al. A study of chromosome 4p markers and dopamine D5 receptor gene in schizophrenia and bipolar disorder. Mol Psychiatry 1998; 3: Ewald H, Degn B, Mors O, Kruse TA. Support for the possible locus on chromosome 4p 16 for bipolar affective disorder. Mol Psychiatry 1998; 3: Craddock N, Owen M, Burge S, Kurian B, Thomas P, McGuffin P. Familial cosegregation of major affective disorder and Darier s disease (keratosis follicularis) Br J Psychiatry 1994; 164: Ewald H, Degn B, Mors O, Kruse TA. Significant linkage between bipolar affective disorder and chromosome 12q24. Psychiatr Genet 1998; 8: Morissette J, Villeneuve A, Bordeleau L, Rochette D, Laberge C, Gagne B et al. Genome-wide search for linkage of bipolar affective disorders in a very large pedigree derived from a homogeneous population in Quebec points to a locus of major effect on chromosome 12q23 q24. Am J Med Genet 1999; 88: Degn B, Lundorf MD, Wang A, Vang M, Mors O, Kruse TA et al. Further evidence for a bipolar risk gene on chromosome 12q24 suggested by investigation of haplotype sharing and allelic association in patients from the Faroe Islands. Mol Psychiatry 2001; 6: Detera-Wadleigh SD, Badner JA, Berrettini WH, Yoshikawa T, Goldin LR, Turner G et al. A high-density genome scan detects evidence for a bipolar-disorder susceptibility locus on 13q32 and other potential loci on 1q32 and 18p11.2. Proc Natl Acad Sci USA 1999; 96: Badenhop RF, Moses MJ, Scimone A, Mitchell PB, Ewen KR, Rosso Linkage between bipolar disorder and chromosome 12q24.3 A et al. A genome screen of a large bipolar affective disorder pedigree supports evidence for a susceptibility locus on chromosome 13q. Mol Psychiatry 2001; 6: Berrettini WH, Ferraro TN, Goldin LR, Weeks DE, Detera-Wadleigh S, Nurnberger JI Jr et al. Chromosome 18 DNA markers and manicdepressive illness: evidence for a susceptibility gene. Proc Natl Acad Sci USA 1994; 91: Berrettini WH, Ferraro TN, Goldin LR, Detera-Wadleigh SD, Choi H, Muniec D et al. A linkage study of bipolar illness. Arch Gen Psychiatry 1997; 54: Nothen MM, Cichon S, Rohleder H, Hemmer S, Franzek E, Fritze J et al. Evaluation of linkage of bipolar affective disorder to chromosome 18 in a sample of 57 German families. Mol Psychiatry 1999; 4: Stine OC, Xu J, Koskela R, McMahon FJ, Gschwend M, Friddle C et al. Evidence for linkage of bipolar disorder to chromosome 18 with a parent-of-origin effect. Am J Hum Genet 1995; 57: Coon H, Hoff M, Holik J, Hadley D, Fang N, Reimherr F et al. Analysis of chromosome 18 DNA markers in multiplex pedigrees with manic depression. Biol Psychiatry 1996; 39: De Bruyn A, Souery D, Mendelbaum K, Mendlewicz J, Van Broeckhoven C. Linkage analysis of families with bipolar illness and chromosome 18 markers. Biol Psychiatry 1996; 39: Ewald H, Wang AG, Vang M, Mors O, Nyegaard M, Kruse TA. A haplotype-based study of lithium responding patients with bipolar affective disorder on the Faroe Islands. Psychiatr Genet 1999; 9: Freimer NB, Reus VI, Escamilla MA, McInnes LA, Spesny M, Leon P et al. Genetic mapping using haplotype, association and linkage methods suggests a locus for severe bipolar disorder (BPI) at 18q22 q23. Nat Genet 1996; 12: McMahon FJ, Hopkins PJ, Xu J, McInnis MG, Shaw S, Cardon L et al. Linkage of bipolar affective disorder to chromosome 18 markers in a new pedigree series. Am J Hum Genet 1997; 61: Detera-Wadleigh SD, Badner JA, Goldin LR, Berrettini WH, Sanders AR, Rollins DY et al. Affected-sib-pair analyses reveal support of prior evidence for a susceptibility locus for bipolar disorder, on 21q. Am J Hum Genet 1996; 58: Straub RE, Lehner T, Luo Y, Loth JE, Shao W, Sharpe L et al. A possible vulnerability locus for bipolar affective disorder on chromosome 21q22.3. Nat Genet 1994; 8: Byerley W, Holik J, Hoff M, Coon H. Search for a gene predisposing to manic-depression on chromosome 21. Am J Med Genet 1995; 60: Gurling H, Smyth C, Kalsi G, Moloney E, Rifkin L, O Neill J et al. Linkage findings in bipolar disorder. Nat Genet 1995; 10: Aita VM, Liu J, Knowles JA, Terwilliger JD, Baltazar R, Grunn A et al. A comprehensive linkage analysis of chromosome 21q22 supports prior evidence for a putative bipolar affective disorder locus. Am J Hum Genet 1999; 64: Mendlewicz J, Fleiss JL. Linkage studies with X-chromosome markers in bipolar (manic-depressive) and unipolar (depressive) illnesses. Biol Psychiatry 1974; 9: Lucotte G, Landoulsi A, Berriche S, David F, Babron MC. Manic depressive illness is linked to factor IX in a French pedigree. Ann Genet 1992; 35: Baron M, Freimer NF, Risch N, Lerer B, Alexander JR, Straub RE et al. Diminished support for linkage between manic depressive illness and X-chromosome markers in three Israeli pedigrees. Nat Genet 1993; 3: Pekkarinen P, Terwilliger J, Bredbacka PE, Lonnqvist J, Peltonen L. Evidence of a predisposing locus to bipolar disorder on Xq24 q27.1 in an extended Finnish pedigree. Genome Res 1995; 5: Adams LJ, Mitchell PB, Fielder SL, Rosso A, Donald JA, Schofield PR. A susceptibility locus for bipolar affective disorder on chromosome 4q35. Am J Hum Genet 1998; 62: Friddle C, Koskela R, Ranade K, Hebert J, Cargill M, Clark CD et al. Full-genome scan for linkage in 50 families segregating the bipolar affective disease phenotype. Am J Hum Genet 2000; 66: Foroud T, Castelluccio PF, Koller DL, Edenberg HJ, Miller M, Bowman E et al. Suggestive evidence of a locus on chromosome 10p 743

11 744 Linkage between bipolar disorder and chromosome 12q24.3 using the NIMH genetics initiative bipolar affective disorder pedigrees. Am J Med Genet 2000; 96: Kelsoe JR, Spence MA, Loetscher E, Foguet M, Sadovnick AD, Remick RA et al. A genome survey indicates a possible susceptibility locus for bipolar disorder on chromosome 22. Proc Natl Acad Sci USA 2001; 98: Ewald H. Genetics of bipolar affective disorders. In: Marneros AAJ (ed). Bipolar Disorders 100 Years After Manic-Depressive Insanity. Kluwer Academic Publishers: Dordrecht, 2000, pp Baron M. Genetics of schizophrenia and the new millennium: progress and pitfalls. Am J Hum Genet 2001; 68: Berrettini WH. Are schizophrenic and bipolar disorders related? A review of family and molecular studies. Biol Psychiatry 2000; 48: Ewald H, Mors O, Flint T, Koed K, Eiberg H, Kruse TA. A possible locus for manic depressive illness on chromosome 16p13. Psychiatr Genet 1995; 5: Ewald H, Mors O, Koed K, Eiberg H, Kruse TA. Susceptibility loci for bipolar affective disorder on chromosome 18? A review and a study of Danish families. Psychiatr Genet 1997; 7: Wing JK, Babor T, Brugha T, Burke J, Cooper JE, Giel R et al. SCAN. Schedules for Clinical Assessment in Neuropsychiatry. Arch Gen Psychiatry 1990; 47: World Health Organization. The ICD-10 classification of mental and behavioural disorders. Diagnostic criteria for research. World Health Organization: Geneva American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th edn. American Psychiatric Association: Washington, Ewald H, Mors O, Friedrich U, Flint T, Kruse T. Exclusion of linkage between manic depressive illness and tyrosine hydroxylase and dopamine D2 receptor genes. Psychiatr Genet 1994; 4: Gyapay G, Morissette J, Vignal A, Dib C, Fizames C, Millasseau P et al. The Genethon human genetic linkage map. Nat Genet 1994; 7: Lathrop GM, Lalouel JM, Julier C, Ott J. Strategies for multilocus linkage analysis in humans. Proc Natl Acad Sci USA 1984; 81: Markianos K, Daly MJ, Kruglyak L. Efficient multipoint linkage analysis through reduction of inheritance space. Am J Hum Genet 2001; 68: Ott J, Terwilliger JD. Assessing the evidence for linkage in psychiatric genetics. In: Mendlewicz J, Hippius H (eds). Genetic Research in Psychiatry. Springer-Verlag: Berlin, 1992, pp Sham PC. Statistics in Human Genetics. John Wiley & Sons: New York, Terwilliger JD, Ott J. Handbook of Human Genetic Linkage. The Johns Hopkins University Press: Bathmore, London, Freimer NB, Sandkuijl LA, Blower SM. Incorrect specification of marker allele frequencies: effects on linkage analysis. Am J Hum Genet 1993; 52: Hodge SE. Do bilineal pedigrees represent a problem for linkage analysis? Basic principles and simultation results for single-gene diseases with no heterogeneity. Genet Epidemiol 1992; 9: Hodge SE, Greenberg DA. Sensitivity of lod scores to changes in diagnostic status. Am J Hum Genet 1992; 50: Ott J. Computer-simulation methods in human linkage analysis. Proc Natl Acad Sci USA 1989; 86: Weeks DE, Ott J, Lathrop GM. SLINK: a general simulation program for linkage analysis. Am J Hum Genet 1990; 47 Suppl: A Lander ES, Kruglyak L. Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet 1995; 11: Nyholt DR. All LODs are not created equal. Am J Hum Genet 2000; 67: Rao DC, Gu C. False positives and false negatives in genome scans. Adv Genet 2001; 42: Terwilliger JD, Shannon WD, Lathrop GM, Nolan JP, Goldin LR, Chase GA et al. True and false positive peaks in genomewide scans: applications of length-biased sampling to linkage mapping. Am J Hum Genet 1997; 61: Detera-Wadleigh SD. Chromosomes 12 and 16 workshop. Am J Med Genet 1999; 88: Dawson E, Parfitt E, Roberts Q, Daniels J, Lim L, Sham P et al. Linkage studies of bipolar disorder in the region of the Darier s disease gene on chromosome 12q Am J Med Genet 1995; 60: Rice JP, Goate A, Williams JT, Bierut L, Dorr D, Wu W et al. Initial genome scan of the NIMH genetics initiative bipolar pedigrees: chromosomes 1, 6, 8, 10, and 12. Am J Med Genet 1997; 74: Reich T, Edenberg HJ, Goate A, Williams JT, Rice JP, Van Eerdewegh P et al. Genome-wide search for genes affecting the risk for alcohol dependence. Am J Med Genet 1998; 81: Cao Q, Martinez M, Zhang J, Sanders AR, Badner JA, Cravchik A et al. Suggestive evidence for a schizophrenia susceptibility locus on chromosome 6q and a confirmation in an independent series of pedigrees. Genomics 1997; 43: Martinez M, Goldin LR, Cao Q, Zhang J, Sanders AR, Nancarrow DJ et al. Follow-up study on a susceptibility locus for schizophrenia on chromosome 6q. Am J Med Genet 1999; 88: Levinson DF, Holmans P, Straub RE, Owen MJ, Wildenauer DB, Gejman PV et al. Multicenter linkage study of schizophrenia candidate regions on chromosomes 5q, 6q, 10p, and 13q: schizophrenia linkage collaborative group III. Am J Hum Genet 2000; 67: Haines JL. Affected relative pair analysis. In: Haines JL, Pericak- Vance MA (eds). Approaches to Gene Mapping in Complex Human Diseases. Wiley-Liss: New York, 1998, pp Cichon S, Schmidt-Wolf G, Schumacher J, Muller DJ, Hurter M, Schulze TG et al. A possible susceptibility locus for bipolar affective disorder in chromosomal region 10q25 q26. Mol Psychiatry 2001; 6: Kelsoe JR, Loetscher E, Spence MA, Fouget M, Sadovnick AD, Remick RA et al. A genome survey of bipolar disorder indicates a susceptibility locus on chromosome 22. Am J Med Genet 1998; 81: Ewald H, Flint T, Wang AG, Mors O, Vang M, Jensen P, Kruse TA. Evidence for a shared segment on chromosome 10q26 in patients with bipolar affective disorder from the Faroe Islands. Am J Med Genet (Neuropsychiatr Genet) 2002; 114: Levinson DF, Mahtani MM, Nancarrow DJ, Brown DM, Kruglyak L, Kirby A et al. Genome scan of schizophrenia. Am J Psychiatry 1998; 155: Potash JB, DePaulo JR Jr. Searching high and low: a review of the genetics of bipolar disorder. Bipolar Disord 2000; 2: 8 26.