Possible Evidence for a Common Risk Locus for Bipolar Affective Disorder

Molecular Psychiatry (2004) 9,93–98 & 2004 Nature Publishing Group All rights reserved 1359-4184/04 $25.00 www.nature.com/mp ORIGINAL RESEARCH ARTICLE Possible evidence for a common risk locus for bipolar affective disorder and schizophrenia on chromosome 4p16 in patients from the Faroe Islands TD Als1, HA Dahl2, TJ Flint3, AG Wang4,5, M Vang5, O Mors1, TA Kruse2 and H Ewald1,3 1Department of Psychiatric Demography, Institute for Basic Psychiatric Research, Psychiatric Hospital in Aarhus, Aarhus University Hospital, Risskov, Denmark; 2Department of Clinical Biochemistry and Genetics, Odense University Hospital, Odense Denmark; 3Department of Biological Psychiatry, Institute for Basic Psychiatric Research, Psychiatric Hospital in Aarhus, Aarhus University Hospital, Risskov, Denmark; 4Department of Psychiatry, Amager Hospital, Copenhagen University Hospital, Copenhagen, Denmark; 5Department of Psychiatry, Landssju´krahusid (National Hospital), Torshavn, Faroe Islands

Patients with schizophrenia (n = 11) and bipolar affective disorder (n = 17) from the relatively isolated population of the Faroe Islands were genotyped for 34 polymorphic markers on chromosome 4 in a search for allelic association and haplotype sharing among distantly related patients. When considering bipolar patients only, there was no clearcut support for any region on chromosome 4. The two-marker segment D4S394–D4S2983 at 4p16.1 was, however, supported by a P-value of 0.0162. For patients with schizophrenia, there was reasonable support for 4p16.1 as marker D4S2281 (P = 0.0019), a two-marker segment (D4S2281–D4S1605, P = 0.0009) and a three-marker segment (D4S2923–D4S2928–D4S1582, P = 0.0005) appeared to be associated with schizophrenia, with some alleles/haplotypes occurring with different frequencies in patients compared to controls. When combining both psychiatric disorders, chromosome 4p16.1 received further support from five partially overlapping two- and three- marker segments (D4S394–D4S2983, P = 0.0039; D4S2281–D4S1605, P = 0.0027 and D4S394– D4S2983–D4S2923, P = 0.006; D4S2923–D4S2928–D4S1582, P = 0.00007; D4S1582–D4S1599– D4S2281, P = 0.005). Increased haplotype sharing in patients with schizophrenia and in the combined data set was partly supported by Fisher’s exact test and tests based on the genealogy. Our study yields support for a common risk gene for schizophrenia and bipolar affective disorder on the short arm of chromosome 4, as suggested by previous findings in the neighbouring Scottish population. Molecular Psychiatry (2004) 9, 93–98. doi:10.1038/sj.mp.4001393 Keywords: genetic association; haplotype sharing; dopamine receptor; chromosome 4; susceptibility locus

There are several lines of evidence indicating that nia and bipolar affective disorder on chromosome 4, schizophrenia and bipolar affective disorder are in samples of distantly related patients from the unlikely to be caused by single major genes but rather genetically isolated population of the Faroe Islands. by multifactorial inheritance.1,2 The two disorders The present study identifies segments in chromosome may share genetic and nongenetic risk factors.3 regions 4p16 and 4p15 that may be involved in the Significant linkage of markers to 4p16 has been susceptibility to schizophrenia and bipolar affective found in several families segregating for bipolar disorder. In searching for a joint risk locus for the two disorder or schizophrenia,4–9 which makes this a disorders, a number of overlapping segments were candidate region for susceptibility gene(s). Perhaps supported by increased sharing of one or more the most interesting candidate gene in this region is haplotypes. the gene coding for the dopamine type 5 receptor Linkage-disequilibrium (LD)-based approaches for (DRD5),10 as malfunction of the dopamine system has mapping of complex neuropsychiatric disease genes been implicated in psychotic disorders.11 may be useful for locating susceptibility loci.12,13 The The present study focuses on obtaining further recent founding, isolation and relatively rapid expan- evidence to map susceptibility genes for schizophre- sion of the Faroese population14,15 is reflected in the population structure and the extensive background LD that has been observed in this population,16,17 Correspondence: Dr TD Als, Department of Psychiatric Demo- making this population suitable for LD-mapping of graphy, Institute for Basic Psychiatric Research, Psychiatric complex diseases. The affected individuals in this Hospital in Aarhus, Aarhus University Hospital, Skovagervej 2, study share common ancestry, and it is therefore DK-8240 Risskov, Denmark. E-mail: [email protected] Received 21 March 2003; revised 26 May 2003; accepted 02 June likely that they also share disease genes and alleles of 2003 surrounding markers. In LD-mapping, one searches A common risk locus on 4p16 TD Als et al 94 for markers that have different allele or haplotype be associated with schizophrenia. When combining frequencies when comparing cases and controls, both psychiatric disorders, chromosome 4p16.1 re- suggesting that markers where allele frequencies ceived further support from six partially overlapping differ are in LD and therefore presumably lie close two- and three-marker segments (D4S394–D4S2983, to a disease gene. D4S2923–D4S2928, D4S2281–D4S1605 and D4S394– Allele and haplotype frequencies of 34 microsatel- D4S2983–D4S2923; D4S2923–D4S2928–D4S1582, lite-markers were compared among cases and controls D4S1582–D4S1599–D4S2281). For most of these using Monte Carlo-based tests as implemented in the markers and segments, Fisher’s exact tests revealed computer program CLUMP.18 Association tests were P-values that were either nonsignificant, an order of performed for single markers, two- and three-marker magnitude higher than the one obtained by running segments (Table 1). Table 2 lists P-values for Fisher’s CLUMP or several alleles/haplotypes produced sig- exact tests performed for single alleles or haplotypes nificant P-values. of markers that revealed an interesting level of The DRD5 gene is located between markers association (Table 1). For a given marker or segment, D4S2923 and D4S2928 on 4p16.1. Fisher’s exact tests Fisher’s exact tests were only performed for specific for the combined data set yielded P-values below 0.08 alleles or haplotypes with large deviation in fre- for two haplotypes of this segment (Table 2), suggest- quency between cases and controls, in order to ing that the difference between cases and controls evaluate whether low P-values obtained by CLUMP (Table 1) might be explained by a combination of were preferentially caused by deviation in frequency these haplotypes. Only one of these (132–191) of a single allele or haplotype. Each separate Fisher’s showed increased frequency among cases. IBD prob- exact test, therefore, only evaluates differences in abilities suggest that IBD by chance is less likely frequencies of a specific allele or haplotype. If the considering the specified genealogy, and that this P-value obtained by running CLUMP is lower, this haplotype may show increased sharing among cases indicates that the overall difference between cases due to LD with a common risk allele. and controls is not caused by divergence in frequency For schizophrenia and for both disorders of a single allele or haplotype, but must be caused by combined, a single haplotype (132–191–124) of divergence in frequencies of more than one allele or D4S2923–D4S2928–D4S1582 turned out to have sig- haplotype. Affected individuals who are related may nificant differences in frequency between cases and share segments that are identical by decent (IBD) from controls (Table 1), although two other haplotypes also a common ancestor who does not necessarily contain contributed to the very low CLUMP P-value in the shared disease genes, and therefore represent false combined data set (Table 2). IBD calculations for the positives. The goal is to identify haplotypes shared combined data set suggest that haplotype sharing by more frequently by cases than would be expected by chance is less likely, and this segment could therefore chance given the genealogical relation of cases. The potentially contain a risk locus in association with the probability that affected individuals share specific two-marker haplotype. haplotypes that are IBD by chance from a known The two-marker segment D4S403–D4S3048 (4p15) ancestor is presented in Table 2. IBD probability appeared to be associated with schizophrenia and the calculations are performed for a single haplotype, and two disorders combined (Table 1). Fisher’s exact tests are therefore too conservative for evaluating situa- revealed differences in frequency of two haplotypes tions where low CLUMP P-values are caused by (Table 2). IBD probabilities calculated according to differences in frequencies of more than one haplo- Houwen et al22 indicate that each of these differences type. considered separately is less likely to be IBD by The gene for the DRD5 maps to chromosome chance. 4p16.110 and dysfunction of dopamine synthesis has The present study of patients from the Faroe Islands been implied in the pathophysiology of psychotic provides no support for the previously suggested disorders for decades.11 Linkage between D4S394 regions on chromosomees 4q31.21 and 4q35,5,8 (4p16.1) and bipolar affective disorder was initially neither for each of the disorders considered sepa- reported by Blackwood et al.4 Other surveys have rately nor for the combined data set (Table 1). provided additional evidence for linkage of bipolar Many of the markers and segments are in close affective disorder or schizophrenia to markers in this proximity and, based on a previous study of LD region and other regions on chromosome 4.5–9 There among markers on chromosome 22,17 presumably in are, however, also negative reports of linkage of extensive LD. The association tests performed in bipolar affective disorder or schizophrenia to chro- the present study cannot, therefore, be considered mosome 4.9,19–21 independent. There is no generally accepted method When considering bipolar patients, two segments to make the appropriate correction, and this issue (D4S394–D4S2983 and D4S2923–D4S2928–D4S1582) is further complicated by the fact that LD varies at 4p16.1 received some support (Table 1). For between populations and among different genomic patients with schizophrenia, there was reasonable regions.23–25 The P-values presented here are, there- support for 4p16.1, as marker D4S2281, a two-marker fore, not corrected for multiple tests. segment (D4S2281–D4S1605) and a three-marker The patients in this study are distantly related, and segment (D4S2923–D4S2928–D4S1582) appeared to can therefore, in principle, not be considered as

Molecular Psychiatry Table 1 Empirical P-values obtained from CLUMP18 for single markers, two- and three-loci haplotypes

Position Bipolar affective disorder (n¼17) Schizophrenia (n¼11) Both (n¼28)

Single marker Two-marker Three-marker Single marker Two-marker Three-marker Single marker Two-marker Three-marker

Marker cM kb T1 T4 T1 T4 T1 T4 T1 T4 T1 T4 T1 T4 T1 T4 T1 T4 T1 T4

D4S3038 1.48 1106 0.8290 0.9920 0.8180 0.7790 0.8770 0.9770 D4S2366 12.93 6402 0.9810 0.9820 0.9300 0.9660 0.1730 0.1360 0.1950 0.3350 0.6440 0.6340 0.8600 0.7850 D4S394 16.01 6865 0.3880 0.3100 0.7070 0.7990 0.8402 0.7592 0.2490 0.2080 0.4600 0.2550 0.0575 0.0788 0.1240 0.0720 0.6320 0.4150 0.5624 0.6104 D4S2983 17.49 7654 0.0590 0.1090 0.0162* 0.0240* — — 0.4640 0.3580 0.0320* 0.0130* 0.0124 0.0294 0.0760 0.0420 0.0039 0.0026 0.0277 0.0342 D4S2923 17.49 8176 0.0142 0.0074 0.0790 0.0420 — — 0.7090 0.7960 0.9000 0.9190 0.0327 0.0330 0.0860 0.1170 0.3910 0.2310 0.0058 0.0079 D4S2928 23.17 10287 0.0640 0.0580 0.1040 0.1440 — — 0.1530 0.2060 0.1470 0.1290 — — 0.0220 0.0360 0.0138* 0.0162* 0.1189 0.1019 D4S1582 23.76 10376 0.0790 0.0330 0.2880 0.3140 0.0110* 0.0010 0.2730 0.1400 0.0208 0.0392 0.0005 0.0011 0.0320 0.0140 0.0428 0.0380 6.9 Â 10À 5 6.6 Â 10À 5 D4S1599 23.17 10570 0.7320 0.7240 0.7050 0.6400 0.7762 0.6773 0.3080 0.3860 0.3700 0.3670 — — 0.1620 0.2500 0.3930 0.5300 0.6124 0.4915 D4S2281 23.76 10629 0.3620 0.3170 0.1550 0.1520 0.0190 0.0350 0.0019 0.0014 — — — — 0.0340 0.0270 0.0760 0.0470 0.0047 0.0141 D4S1605 23.17 10726 0.8260 0.8150 0.1550 0.1730 0.3307 0.5025 0.8750 0.8480 0.0009 0.0024 — — 0.8770 0.8280 0.0027 0.0059 0.1249 0.2078 D4S3009 23.17 10811 0.3020 0.2210 0.3690 0.2240 0.3736 0.3766 0.7420 0.4830 0.5090 0.4550 0.0451 0.0679 0.1420 0.0800 0.1770 0.0820 0.0639 0.0606 D4S3036 24.29 12061 0.6710 0.6030 0.9730 0.9900 0.9540 0.9880 0.3260 0.8010 0.9020 0.7170 0.8372 0.6334 0.4040 0.4280 0.9610 0.8910 0.7013 0.5564 D4S2311 27.52 13437 0.8190 0.9430 0.8120 0.8120 — — 0.7100 0.6010 0.7240 0.6740 0.9271 0.8681 0.9520 0.8750 0.6250 0.6490 0.8162 0.7642 D4S403 25.90 13613 0.9350 0.9360 0.9180 0.9510 0.7483 0.7183 0.1720 0.1670 0.2930 0.2430 0.5984 0.7632 0.3950 0.3220 0.6570 0.7380 0.4695 0.3666 D4S3048 29.14 15822 0.3680 0.4780 0.0500 0.0490 0.5385 0.5465 0.0760 0.0300 0.0101 0.0077 0.0685 0.0711 0.1100 0.2130 0.0018 0.0040 0.0986 0.1150 D4S2639 33.42 18660 0.7680 0.9000 0.1430 0.1460 0.1548 0.1598 0.0270 0.0580 0.0480 0.0700 0.5874 0.5295 0.4510 0.5230 0.1680 0.1060 0.4326 0.5075 D4S2397 42.74 27408 0.4680 0.5360 0.7220 0.7710 0.4555 0.6094 0.0920 0.1660 0.4320 0.2910 0.3337 0.3377 0.0920 0.1520 0.4620 0.3130 0.2438 0.2927 D4S2629 50.53 36522 0.9840 0.9750 0.9670 0.9340 — — 0.0660 0.1780 0.3670 0.1890 0.0300 0.1658 0.6350 0.7240 0.6130 0.3740 0.0629 0.0629 D4S1627 60.16 44253 0.9760 0.9810 0.4630 0.2820 — — 0.9390 0.9140 0.9010 0.8150 — — 1.0000 1.0000 0.4810 0.3340 — — omnrs ou n4p16 Als on TD locus risk common A D4S3255 61.42 52621 0.3830 0.1660 0.3010 0.3100 0.8142 0.8851 0.2860 0.1410 0.6680 0.7790 0.9301 0.9730 0.0580 0.0280 0.2060 0.1840 0.7752 0.7403 D4S3248 72.52 59759 0.2450 0.5290 0.3980 0.4440 0.4725 0.4545 0.4960 0.4090 0.8940 0.6780 0.7632 0.8651 0.5030 0.4860 0.5380 0.2860 0.5574 0.6533 D4S2367 78.43 68194 0.6860 0.6620 0.2210 0.3020 0.8661 0.8172 0.7170 0.8550 0.8830 0.7270 0.9650 0.9201 0.3710 0.3500 0.7610 0.8980 0.8462 0.6973 al et D4S2364 100.75 93675 1.0000 1.0000 0.4050 0.2440 0.2188 0.3107 0.7210 0.7210 0.7760 0.6890 0.5834 0.6713 0.7940 0.7940 0.2820 0.1790 0.3317 0.4555 D4S1647 104.94 99681 0.6100 0.5250 0.2120 0.1630 0.3087 0.3157 0.9410 0.8960 0.9660 0.9670 0.3686 0.4725 0.8390 0.8060 0.9360 0.9290 0.4206 0.5345 D4S2394 129.92 131237 0.0206 0.0240 0.0960 0.1380 0.2308 0.2408 0.3300 0.3040 0.6990 0.9110 0.9201 0.9161 0.0350 0.0630 0.2950 0.3790 0.6543 0.7792 D4S1520 140.64 141580 0.6360 0.6400 0.3060 0.2710 — — 0.8560 0.8680 0.1950 0.2450 — — 0.6260 0.6180 0.3860 0.4240 0.5495 0.7223 D4S1644 143.31 142615 0.7310 0.5560 0.6420 0.8770 0.1409 0.2098 0.6710 0.6170 0.5330 0.4680 0.4855 0.6404 0.7620 0.8750 0.7880 0.8540 0.3217 0.3846 D4S1625 145.98 144287 0.2470 0.0880 0.5340 0.4220 0.9830 0.9850 0.5180 0.5250 0.6890 0.6790 0.8821 0.4925 0.5650 0.5110 0.7190 0.6460 0.9870 0.9620 D4S1629 157.99 159216 0.3990 0.4050 0.3220 0.4090 0.6853 0.6723 0.9300 0.8660 0.7880 0.7750 0.8052 0.7602 0.5630 0.5140 0.4070 0.4520 0.7203 0.7483 D4S2431 176.19 175741 0.1300 0.1780 0.3650 0.7400 0.8312 0.8442 0.0600 0.0810 0.0410 0.0380 — — 0.4190 0.3570 0.3950 0.8020 0.6084 0.6494 D4S3028 177.33 177476 0.5820 0.6500 0.1810 0.2100 — — 0.7280 0.7380 0.1340 0.2260 0.4735 0.3566 0.7950 0.7320 0.1770 0.3700 0.2827 0.3407 D4S2417 181.93 181187 0.3350 0.6360 0.0860 0.0630 0.1409 0.1548 0.0110 0.0187 0.1340 0.1960 — — 0.1410 0.1050 0.0650 0.1130 0.1389 0.1798 D4S1530 185.28 182504 0.8840 0.8290 0.1370 0.3170 — — 0.6650 0.6290 0.2620 0.3400 0.1758 0.2288 0.8610 0.7200 0.3960 0.5140 0.8761 0.9231 D4S408 195.06 186179 0.7200 0.8810 0.5670 0.3330 0.2737 0.2707 0.0230 0.0090 0.1870 0.1670 0.3057 0.2887 0.1700 0.1220 0.4290 0.4160 0.3596 0.3786

T1 is calculated based on the raw 2 Â N table, while T4 is produced by clumping columns of the raw 2 Â N table together into a new 2 Â 2 table in such a way that the w- square value is maximised.18 Interesting markers and P-values below specific thresholds are in bold. Thresholds are as follows: Pr0.005 for single-marker and Pr0.01 for two-marker and three-marker segments, except low P-values marked with *, which are just above the threshold, representing markers that are located within a priori regions. Missing values (À) indicate that no haplotype was observed more than once among cases and that CLUMP p-values were therefore not calculated. The order and physical position of the 34 microsatellite markers was based on the ‘majority-rule’ consensus of four publicly available databases on the physical positions of oeua Psychiatry Molecular the markers in the sequence of the Human genome (http://www.ncbi.nlm.nih.gov/genome/seq; http://genome.cse.ucsc.edu; http://www.ensembl.org/Homo_sapiens; and the map published by DeCODE27 ). Genetic positions was based on Marshfield (http://research.marshfieldclinic.org/genetics/Map_Markers/). 95 A common risk locus on 4p16 TD Als et al 96 Table 2 Alleles and segments of most interest

Segment Most frequent Frequency Fisher’s Probability of shared haplotypes exact test to be IBD by chance

Haplotype Cases Controls (P) P0 P1 P2

Bipolar D4S394–D4S2983 201; 245 2/31 0/85 0.070 0.0805 1.8026 3.999 205; 246 0/34 9/80 0.056 — — — D4S2923–D4S1582 132; 191; 124 2/30 0/88 0.063 0.0391 1.1265 3.999 132; 187; 122 1/31 10/75 0.169 0.5720 15.0596 4 Schizophrenia D4S2281 396 4/16 2/74 0.008 — — — 408 2/16 0/74 0.030 — — — D4S394–D4S2983 207; 245 3/22 2/84 0.059 0.0147 0.3115 3.6097 D4S2281–D4S1605 408; 129 2/17 0/77 0.031 0.1065 2.2317 3.999 372; 129 2/17 0/76 0.032 0.1065 2.2317 3.999 376; 129 0/16 14/74 0.066 — — — D4S403–D4S3048 227; 247 4/22 1/85 0.006 6.58 Â 10À4 0.0199 0.8582 225; 239 4/22 0/80 0.002 6.58 Â 10À4 0.0199 0.8582 217; 241 0/22 8/80 0.196 — — — D4S2923–D4S1582 132; 191; 124 3/22 0/88 0.007 0.0063 0.2021 3.6097 132; 187; 114 0/14 11/71 0.198 — — —

Both D4S394–D4S2983 207; 245 5/52 2/84 0.106 1.36 Â 10À4 0.0026 0.218 207; 254 3/53 1/85 0.158 0.0251 0.5189 3.9574 201; 245 2/52 0/85 0.142 0.2084 4.7059 4.0000 D4S2923–D4S2928 132; 187 14/49 33/72 0.061 2.31 Â 10À20 4.36 Â 10À14 2.04 Â 10À14 132; 191 6/51 3/88 0.074 1.15 Â 10À6 1.25 Â 10À4 0.0150 D4S2281–D4S1605 380; 129 5/47 5/71 0.517 1.12 Â 10À4 0.0017 0.1362 396; 129 4/45 1/76 0.063 0.0015 0.0227 1.1581 372; 129 4/47 0/76 0.020 0.0017 0.0269 1.3217 D4S403–D4S3048 227; 247 6/55 1/85 0.015 5.09 Â 10À6 1.96 Â 10À4 0.0232 225; 239 5/55 0/82 0.009 9.88 Â 10À5 0.0034 0.2808 D4S394–D4S2923 207; 254; 132 2/53 0/85 0.146 0.2156 3.2511 4.0000 207; 245; 128 2/51 0/86 0.137 0.2013 3.0323 4.0000 205; 246; 132 0/53 6/80 0.081 — — — D4S2923–D4S1582 132; 191; 124 5/52 0/88 0.006 2.58 Â 10À5 0.0017 0.2177 132; 187; 122 1/49 10/75 0.049 — — — 132; 187; 114 1/42 11/71 0.030 — — — 134; 187; 114 0/51 5/74 0.079 — — — D4S1582–D4S2281 124; 148; 396 3/53 0/86 0.053 0.0334 0.3332 3.9575 124; 142; 392 3/52 0/87 0.050 0.0317 0.3159 3.9475 114; 154; 376 0/55 6/86 0.081 — — —

Fisher’s exact test identifies differences in allele or haplotype frequencies between cases and controls for specific alleles or

haplotypes. P0 is the probability of all cases having the most frequent haplotype IBD by chance from a known ancestor (i.e., 22 false-positive probability) according to Houwen et al. P1 is the same probability summed over all segments typed on 28 chromosome 4. P2 is the genome-wide probability of IBD according to the method by Durham and Feingold. The presence or absence of the most frequent haplotype could not always be unambiguously determined due to phase uncertainties and the

total number of chromosomes typed was therefore lowered accordingly. Due to the conservative way P1 and P2 are calculated,

they may exceed a value of 1 when the observed value of the haplotype of interest are small. P2 is considered to be the most conservative.

independent samples. All affected individuals are, samples of the same panmictic population and share however, related more than three generations ago and the same genetic history, seems valid for our sam- the bias introduced is considered to be relatively ple.16,17 small. As we have detailed knowledge of the In conclusion, our study supports chromosome genealogy of the patients, this allows us to calculate region 4p16.1 for both bipolar affective disorder and the likelihood of IBD by chance and estimate the schizophrenia. A single marker and two- and false-positive probability for specific haplotypes. three-marker segments, with varying degrees of We have previously shown that the essential overlap within the 7.75 cM region between markers assumption, that cases and controls represent sub- D4S394 and D4S1605, provide evidence for

Molecular Psychiatry A common risk locus on 4p16 TD Als et al 97 association with schizophrenia and/or the two disorders combined. In some of the segments, specific haplotypes occur with increased frequency in both disorders. This supports the possibility that the locus on 4p16 may predispose to schizophrenia as well as bipolar affective disorder. Based on historical records and geography, the Faroese population is likely to be genetically related to parts of the Scottish and Irish populations, and it is thus possible that the same risk loci, and possibly even haplotypes, may be shared for complex disorders. It is therefore not surprising that chromosome 4p16 has been implied Figure 1 Genealogy of cases: one of several ways to in studies of patients collected from the Scottish connect 17 individuals with bipolar affective disorder (a) population4,9 and other neighbouring populations.5–7 and 11 individuals with schizophrenia (b). The sex of This study is the first to support these findings, individuals in the pedigree is not indicated. providing evidence for a possible shared risk locus on 4p16 for the two disorders on the basis of two- and Acknowledgements three-marker haplotypes. The present study will also make the comparison with specific haplotypes in We thank senior psychiatrist Aksel Bertelsen, Psy- affected individuals from Scotland and Ireland chiatric Hospital in Aarhus, for thorough reading of possible, as standardised allele sizes have been used. medical records and reports for diagnostic evaluation. This could potentially assist in further delineation of This work was supported by The Danish Medical a minimum critical region that harbours the risk Research Council (Grant Number 9601887); The locus. Lundbeck Foundation (Grants Numbers 39/98 and 59/99); Puljen til styrkelse af Psykiatrisk forskning i Aarhus Amt. Psykiatrisk forsknings fond; Ministry of Methods Education, Culture and Research, The Faroese Gov- Well-documented cases of severe schizophrenia or ernment; Research Fund of the Faroese Savings Bank, bipolar affective disorder were searched for among Torshavn, Faroe Islands; Danish Hospital Foundation patients treated at the Department of Psychiatry, for Medical Research, Region of Copenhagen, the National Hospital in Torshavn, Faroe Islands.15,16 Faroe Islands and Greenland. The genealogies of cases were deduced from informa- tion in church and civic records of the Faroe Islands. References The average number of generations relating two patients in the genealogically shortest possible way 1 Craddock N, Jones I. Genetics of bipolar disorder. J Med Genet through one of their parents was six for 11 patients 1999; 36: 585–594. 2 Pulver AE. Search for schizophrenia susceptibility genes. Biol with schizophrenia. 17 individuals with bipolar Psychiatry 2000; 47: 221–230. affective disorder were, on average, related seven 3 Berrettini WH. Are schizophrenic and bipolar disorders related? A generations ago. The average number of generations review of family and molecular studies. Biol Psychiatry 2000; 48: relating two patients with either bipolar affective 531–538. 4 Blackwood DH, He L, Morris SW, McLean A, Whitton C, Thomson disorder or schizophrenia was seven for 28 patients M. A locus for bipolar affective disorder on chromosome 4p. Nat (Figure 1). A total of 44 unrelated individuals without Genet 1996; 12: 427–430. a history of psychiatric disease were chosen as the 5No¨then MM, Cichon S, Franzek E, Albus M, Borrmann M, control population. Rietschel M. Systematic search for susceptibility genes in bipolar Haplotypes for chromosomal segments consisting affective disorder evidence for disease loci at 18p and 4p. JAm Hum Genet 1997; 61: 1679. of two or three neighbouring markers were deter- 6 Ewald H, Degn B, Mors O, Kruse TA. Support for the possible mined for cases on the basis of either available locus on chromosome 4p16 for bipolar affective disorder. Mol parental genotypes or genotypes of spouse and a Psychiatry 1998; 3: 442–448. child when available.15,16 All controls had their 7 Asherson P, Mant R, Williams N, Cardno A, Jones L, Murphy K. A study of chromosome 4p markers and dopamine D5 receptor gene haplotype reconstructed from the genotypes of their in schizophrenia and bipolar disorder. Mol Psychiatry 1998; 3: offspring. This method will reconstruct the majority 310–320. of relatively short haplotypes correctly.14,26 8 Adams LJ, Mitchell PB, Fielder SL, Rosso A, Donald JA, Schofield CLUMP18 produces four statistical measures that PR. A susceptibility locus for bipolar affective disorder on are modifications of a w2 test where the distribution of chromosome 4q35. Am. J Hum Genet 1998; 62: 1084–1091. 9 Muir WJ, Thomsom ML, McKeon P, Mynett-Johnson L, Whitton C, the test statistic is simulated using a Monte Carlo Evans KL. Markers close to the dopamine D5 receptor gene (DRD5) approach. This approach evaluates all alleles/haplo- show significant association with schizophrenia but not bipolar types in the same test and therefore also takes into disorder. Am J Med Genet 2001; 105: 152–158. account the situations where more than one allele/ 10 Sherrington R, Mankoo B, Attwood J, Kalsi G, Curtis D, Buetow K. Cloning of the human dopamine D5 receptor gene and identifica- haplotype differs in frequency between cases and tion of a highly polymorphic microsatellite for the DRD5 locus that controls. Two statistical measures are reported in this shows tight linkage to the chromosome 4p reference marker study (Table 1). RAF1P1. Genomics 1993; 18: 423–425.

Molecular Psychiatry A common risk locus on 4p16 TD Als et al 98 11 Harrison PJ. Dopamine and schizophrenia—proof at last? Lancet 20 Rice JP, Goate A, Nurnberger JI, DePaulo JR, Gershon ES, Blehar 2000; 356: 958–959. MC. Hot spots from a genome screen for bipolar disorder. Am J 12 Risch NJ. Searching for genetic determinants in the new Med Genet 1997; 74: 586–587. millennium. Nature 2000; 405: 847–856. 21 Nancarrow DJ, Levinson DF, Taylor JM, Hayward NK, Walters MK, 13 Cardon LR, Bell JI. Association study designs for complex Lennon DP. No support for linkage to the bipolar regions on diseases. Nat Rev Genet 2001; 2: 91–99. chromosomes 4p, 18p, or 18q in 43 schizophrenia pedigrees. Am J 14 Ewald H, Wang AG, Vang M, Mors O, Nyegaard M, Kruse T. A Med Genet 2000; 96: 224–227. haplotype-based study of lithium responding patients with bipolar 22 Houwen RH, Baharloo S, Blankenship K, Raeymaekers P, Juyn J, affective disorder on the Faroe Islands. Psychiatr Genet 1999; 9: Sandkuijl LA. Genome screening by searching for shared seg- 208. ments: mapping a gene for benign recurrent intrahepatic choles- 15 Ewald H, Flint TJ, Jørgensen TH, Wang AG, Jensen P, Vang M. tasis. Nat Genet 1994; 8: 380–386. Search for a shared segment on chromosome 10q26 in patients 23 Weiss KM, Clark AG. Linkage disequilibrium and the mapping of with bipolar affective disorder or schizophrenia from the Faroe complex human traits. Trends Genet 2002; 18: 19–24. Islands. Am J Med Genet 2002; 114: 196–204. 24 Ardlie KG, Kruglyak L, Seielstad M. Patterns of linkage dis- 16 Jørgensen TH, Børglum A, Mors O, Wang AG, Pinaud M, Flint TJ. equilibrium in the human genome. Nat Rev Genet 2002; 3: search for common haplotypes on 22q in patients with schizo- 299–309. phrenia or bipolar disorder from the Faroe Islands. Am J Med 25 Petes TD. Meiotic recombination hot spots and cold spots. Nat Rev Genet 2002; 114: 245–252. Genet 2001; 2: 360–369. 17 Jørgensen TH, Degn B, Wang AG, Vang M, Gurling H, Kalsi G. 26 Escamilla MA, Mclnnes LA, Service SK, Spesny M, Reus VI, Linkage disequilibrium and demographic history of the isolated Molina J. Genome screening for linkage disequilibrium in a Costa population of the Faroe Islands. Eur. J Hum Genet 2002; 10: Rican sample of patients with bipolar-I disorder: a follow-up study 381–387. on chromosome 18. Am J Med Genet 2001; 105: 207–213. 18 Sham PC, Curtis D. Monte-Carlo tests for associations between 27 Kong A, Gudbjartsson DF, Sainz J, Jonsdottir GM, Gudjonsson SA, disease and alleles at highly polymorphic loci. Ann Hum Genet Richardsson B. A high-resolution recombination map of the 1995; 59: 97–105. human genome. Nat Genet 2002; 31: 241–247. 19 Raeymaekers P, Verheyen GR, VanZand K, Souery D, Claes S, 28 Durham LK, Feingold E. Genome scanning for segments shared Godderis J. Linkage studies with chromosomal candidate regions identical by descent among distant relatives in isolated popula- in bipolar families. Am J Med Genet 1997; 74: 587. tions. Am J Hum Genet 1997; 61: 830–842.

Molecular Psychiatry