Cannabinoid receptor 1 gene (CNR1) and susceptibility to a quantitative phenotype for hebephrenic schizophrenia.код для вставкиСкачать
American Journal of Medical Genetics Part B (Neuropsychiatric Genetics) 147B:279 –284 (2008) Cannabinoid Receptor 1 Gene (CNR1) and Susceptibility to a Quantitative Phenotype for Hebephrenic Schizophrenia Iván Chavarrı́a-Siles,1,2 Javier Contreras-Rojas,1,2 Elizabeth Hare,1 Consuelo Walss-Bass,1 Paulina Quezada,1 Albana Dassori,1 Salvador Contreras,1 Rolando Medina,1 Mercedes Ramı́rez,1 Rodolfo Salazar,2 Henriette Raventos,2 and Michael A. Escamilla1* 1 Department of Psychiatry, Psychiatric Genetics Research Center, University of Texas Health Science Center at San Antonio, San Antonio, Texas 2 Centro de Investigación en Biologı´a Celular y Molecular, Universidad de Costa Rica, San Jose´, Costa Rica Functional alterations of components of the endogenous cannabinoid system, in particular of the cannabinoid receptor 1 protein (CB1), are hypothetical contributors to many of the symptoms seen in schizophrenia. Variants within the cannabinoid receptor 1 gene (CNR1) have been shown to be directly associated with the hebephrenic form of schizophrenia in a Japanese population. This finding, however, has yet to be replicated. In the present study we sought to study the same (AAT)n-repeat microsatellite of the CNR1 gene which showed association to hebephrenic schizophrenia in Japan, and to investigate whether this microsatellite showed association to a hebephrenic type of schizophrenia in a family-based association study in a population of the Central Valley of Costa Rica. The Lifetime Dimensions of Psychosis Scale and a best estimate consensus process were utilized to identify subjects with schizophrenia who had an elevated lifetime dimensional score for negative and disorganized symptoms, which we used as a proxy for ‘‘hebephrenia.’’ Using the Family Based Association Test we found association of these hebephrenic subjects and the (AAT)n-repeat marker of the CNR1 (multi-allelic P ¼ 0.0368). Our hypothesis that an association with the (AAT)n-repeat marker of CNR1 would not be found with the more general type of schizophrenia was also confirmed. Schizophrenic subjects with prominent lifetime scores for disorganization and negative symptoms (dimension for hebephrenia) are associated with the CNR1 gene and present a type of symptomatology that resembles chronic cannabinoid-induced psychosis. The current finding points to the possibility of different genetic and pathophysiologic mechanisms underlying different types of schizophrenia. ß 2008 Wiley-Liss, Inc. Grant sponsor: NIH Research; Grant number: D43 TW06152. *Correspondence to: Michael A. Escamilla, Department of Psychiatry, University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive, San Antonio, TX 78229. E-mail: email@example.com Received 15 March 2007; Accepted 20 June 2007 DOI 10.1002/ajmg.b.30592 ß 2008 Wiley-Liss, Inc. KEY WORDS: CNR1; hebephrenic schizophrenia; central Valley of Costa Rica; disorganization; negative symptoms Please cite this article as follows: Chavarrı́a-Siles I, Contreras-Rojas J, Hare E, Walss-Bass C, Quezada P, Dassori A, Contreras S, Medina R, Ramı́rez M, Salazar R, Raventos H, Escamilla MA. 2008. Cannabinoid Receptor 1 Gene (CNR1) and Susceptibility to a Quantitative Phenotype for Hebephrenic Schizophrenia. Am J Med Genet Part B 147B:279–284. INTRODUCTION Schizophrenia (SC) is a debilitating psychiatric disease which affects 1% of the population worldwide [Kessler et al., 1994], and is thought to be caused in part by genetic factors [Gottesman, 1994]. The cannabinoid receptor 1 gene (CNR1), which encodes the cannabinoid receptor 1 protein (CB1), a G-coupled protein receptor that controls neuronal activity, is an attractive candidate gene for SC, based on genetic mapping studies as well as several experimental and clinical findings. The CNR1 gene is located on chromosome 6q14–15, a region which has been designated as the Schizophrenia 5 locus (OMIM603175). CB1 is one of the most abundant receptors in the human brain, distributed mainly in the cortex, hippocampus, basal ganglia, and cerebellum. In all these structures, CB1 has a presynaptic localization which could be related to its role in synaptic neurotransmission [Pazos et al., 2005]. Due to the distribution of the CB1 in GABAergic and glutametergic synapses, its activation produces a local net effect of hyperpolarization of the presynaptic neuron that leads to a general inhibitory effect [Howlett et al., 2004; Rodriguez de et al., 2005]. CB1 receptors also co-localize with dopamine D1/D2 receptors in the brain; this co-localization could influence the activity of the dopamine signaling in psychotic related disorders [Rodriguez De et al., 2001]. CNR1 knockout mice have been proposed as animal models for SC because they exhibit D2 dopamine receptor hyperactivation and their behavioral alterations mimic symptoms in SC [Fritzsche, 2001]. Several clinical studies indicate that cannabis abuse may have psychomimetic effects in previously non-psychotic subjects and, in schizophrenic patients, cannabis abuse can worsen symptoms and result in liability to relapse [Ujike and Morita, 2004]. Most recently, Solowij and Michie  have summarized the similarities and common neurophysiology underlying the cognitive dysfunction associated with long time cannabis use and the cognitive dysfunction seen in schizophrenia. 280 Chavarrı́a-Siles et al. Despite several lines of genetic and biological evidence that support the hypothesis that alterations of the CB1 may contribute to SC there is conflicting evidence regarding whether polymorphisms in or near the CNR1 gene are associated with SC. All previous studies have utilized case/ control designs, which are vulnerable to false positive or negative results due to population stratification. For the phenotype of SC, two previous studies; one in the Chinese population [Tsai et al., 2000] and the other in the French Caucasian population [Leroy et al., 2001], showed no association between polymorphisms in CNR1 and the phenotype of SC; while two studies, one study in the Spanish population [Martinez-Gras et al., 2006] and one in the Japanese population [Ujike et al., 2002] showed association of the (AAT)n CNR1 microsatellite with SC. In the Japanese study, this microsatellite was strongly associated with the hebephrenic type of SC. In the present study, we utilized family-based association analyses to investigate whether a hebephrenic type of schizophrenia would be associated with the CNR1 gene in the Central Valley of Costa Rican population (CVCR). Hebephrenic schizophrenia derives from a description of one subtype of what Eugen Bleuler called ‘‘the schizophrenias.’’ We hypothesized that a hebephrenic type of schizophrenia most closely approximated the type of cognitive dysfunction associated with the use of cannabinoids [Solowij and Michie, 2007], and that this type of schizophrenia would thus be most likely to show association to CNR1. We also hypothesized that CNR1 gene would not be associated to the broader phenotype of SC in our sample, as the association would be diluted when using the more diverse set of phenotypes classified under the broad DSM-IV classification of Schizophrenia [APA, 1994]. METHODS Sample Population All participants were recruited in accordance with the principles of the Declaration of Helsinki and with approval from the Institutional Review Boards of the University of Costa Rica and the University of Texas Health Science Center at San Antonio. Families were recruited through assessment of a proband, who was recruited from psychiatric hospitals and clinics in the Central Valley of Costa Rica; all probands had a history of at least one hospitalization with discharge diagnosis of SC according to the ICD 9 or ICD 10 system [Montero et al., 2002]. Diagnostic Procedures The original sample consisted of 490 subjects with diagnosis of SC (476 families, 1,246 subjects) All probands and any other family member with a history of psychosis picked up by the Family Interview for Genetic Studies (FIGS) were interviewed by a local psychiatrist using the Spanish version of the Diagnostic Interview for Genetics Studies (DIGS) [Nurnberger et al., 1994]. In addition, at least one healthy relative was interviewed using the Spanish version of the FIGS to gather additional information about each affected subject. All available medical records were also gathered for each affected subject (here ‘‘affected’’ refers to a subject with a history of psychosis). For each affected subject, all three sources of information (DIGS, FIGS, and medical records) were analyzed by two bilingual psychiatrists (best estimate raters) blind to the previous history of the subject and to the family relationships between subjects. The best estimating team arrived at a lifetime consensus diagnosis using DSM-IV criteria, as described in previous studies of the CVCR [Escamilla et al., 1996, 1999]. After this process, only 244 affected subjects had consensus DSM-IV diagnoses of schizophrenia (Table I). We also were able to obtain a diagnosis of abuse or dependence of different substances in these 244 subjects using the DSM-IV criteria (Table II). Previous history of substance abuse or dependence was not an exclusion criteria for this study. Each affected subject was also scored by each best estimate rater for lifetime dimensions of psychosis (LDPS), using the scale developed by Levinson et al. . The LDPS creates a profile of the lifetime characteristics of each case based on retrospective ratings, encompassing dimensions of positive symptoms, Shneiderian symptoms, depressive, manic, negative, and disorganized symptoms. A consensus score for each dimension of psychosis was arrived at through averaging of the two individual raters’ scores. Since the hebephrenic subtype of SC reported on by Ujike et al.  utilized the ICD-10 classification, and a hebephrenic subtype is not defined per se by the DSM-IV, we utilized the LDPS to define which subjects with a DSM-IV consensus diagnosis of SC scored highly on disorganized and negative symptoms over their lifetime. The ICD-10, used by Ujike et al. in their study, characterizes the hebephrenic type of schizophrenia largely based on predominance of disorganized, and negative symptoms [World Health Organization, 2003]. As noted in Ujike et al. , ‘‘hebephrenic type schizophrenia is characterized by predominant negative symptoms such as blunt affect, disorganized thought, and deterioration of personality.’’ ‘‘Deterioration of personality’’ is not specifically mentioned in the ICD-10 classification of hebephrenic schizophrenia, but can perhaps be implied in the context of a person with prominent negative and disorganized behaviors and communication skills. Defining Schizophrenic Subjects With Hebephrenia As the DSM-IV, which we used to make categorical diagnoses in our study, does not have a category for ‘‘hebephrenia,’’ we utilized the LDPS scale to create a quantitative ‘‘hebephrenia’’ score for each affected subject, with the intention to analyze persons who scored high on this trait separately from the overall sample. To obtain a quantitative trait score for ‘‘hebephrenia,’’ we used all of the items for rating negative, and disorganized symptoms in the LDPS; we multiplied severity, and duration score for the items N-1 (Blunted, restricted affect), N-2 (Poverty of speech), D-1 (Formal thought disorder) and D-2 (Bizarre behavior), and then summed these four products. The overall score for the ‘‘LDPS-hebephrenia’’ dimension could range from 0 to 64 points. For the 244 schizophrenic subjects these scores were normally distributed (Shapiro–Wilk Statistic ¼ 0.99) with a mean of 29.1 and standard deviation of 13.9. For our study we arbitrarily chose subjects who scored 30 or higher on this ‘‘hebephrenic’’ dimension, as those who had hebephrenia as a prominent component of their illness. As duration of the symptoms is an important factor in the ‘‘hebephrenia’’ scores (each dimension is scored by multiplying severity of the symptom with duration of the symptom), we also TABLE I. Demographics of the Sample Sample (diagnostic method) Schizophrenia—hospital discharge diagnosis (ICD10) SC-best-estimation (DSM-IV) Hebephrenic-SC (quantitative trait) Families (N) Subjects (N) Affected (N) Age (mean) Sex (M/F, %) 476 237 66 1246 724 231 490 244 66 44.83 37.12 36.54 58/42 70/30 79/21 CNR1 and Susceptibility to Hebephrenic Schizophrenia TABLE II. Sustance Abuse and Dependence Among Subjects With DSM-IV Diagnosis of Schizophrenia (N ¼ 244) Abuse Drug Alcohol Cannabis Cocaine Heroine LSD Opiates Inhalants N 15 20 8 0 1 0 6 Dependence % 6 8 3 0 0.4 0 2 N % 45 14 8 0 2 0 3 18 5 3 0 0.8 0 1 limited our association study of the ‘‘hebephrenic’’ type to those between age 25 and 45. Subjects who are much younger or older than this range (by virtue of their age) would be biased to score lower (for young persons) or higher (for older persons) on the hebephrenia score, compared to those in the range of 25– 45 years of age. As a secondary analysis, we used a quantitative analysis test to investigate whether the quantitative trait of ‘‘hebephrenia,’’ as defined by us using the LDPS scale, showed association to the CNR1 gene. Only the 148 SC subjects who were between age 25 and 45 were included in this quantitative association analysis. Genotyping Genomic DNA was extracted from blood samples using a Puregene DNA purification kit (Gentra, Minneapolis, MN). DNA from 725 subjects (244 subjects with SC and their relatives) was genotyped for the (AAT)n repeat polymorphism using the fluorescently labeled designed primers 50 GAAAGCTGCAAGAGCCC30 and 50 TTTTCCTGTGCTGCCAGGG30 (Applied Biosystems, Foster City, CA). Standard PCR was performed using GeneAmp PCR system 9700 (Applied Biosystems). Amplified fragments were analyzed on the ABI 3100 Genetic Analyzer (Applied Biosystems) and genotypes were assigned using GeneMapper v3.5 (Applied Biosystems). Two individuals blind to the diagnosis scored each genotype separately. Discrepancies were discussed with review of the peaks of the original run to obtain a final genotype. All genotypes were checked for mendelian errors using the program INFER in PEDSYS (http://www.sfbr.org/software/ pedsys/pedsys.html). If mendelian errors were found, genotype diagrams for these families were reviewed and, if necessary, re-genotyped. Any families which still showed non-mendelian inheritance for the microsatellite were not included in further analyses. Statistical Analysis The genotypes were analyzed for Hardy–Weinberg disequilibrium using the PEDSTATS program among unrelated individuals (Wigginton and Abecasis, 2005). All association analyses were performed using the Family Based Association Test (FBAT) (http://www.biostat.harvard.edu/fbat/ default.html), with the following settings: model additive, test bi-allelic (provides asymptotic P values of the Z score function, which looks at the transmitted alleles to affected offspring) and multi-allelic, and minimum size 10 (only alleles that were present in at least 10 informative families were tested for association). This program provides empirical P values for association studies and allows for multiple affected subjects and any available relatives to be included in the analyses. Affected subjects were defined as being between the ages of 281 25 and 45, and having a consensus DSM-IV diagnosis of Schizophrenia and a LDPS-Hebephrenia score of 30 or higher. Transmitted alleles (those going from parents to the affected subjects) were compared to non-transmitted alleles (those not transmitted from parents to affected subjects). When both parents were not available, additional siblings of the affected subjects were genotyped to permit inference of the parental alleles using the program INFER in PEDSYS. For the statistical analysis of whether a ‘‘hebephrenic’’ type of schizophrenia was associated with CNR1, we only analyzed affected subjects with ages ranging from 25 to 45 years old at the time of the interview, to obtain a more homogeneous group and avoid biases for the duration of the symptoms (as mentioned before the LDPS incorporates both severity and duration in determining the score of a lifetime dimension). The LDPS-hebephrenia scores were used as a quantitative trait to test for association with the CNR1 gene using FBAT. The program computes both bi-allelic tests and multi-allelic tests of association for microsatellite markers with quantitative traits, using the following commands in FBAT: trait LDPS; offset 0.000; model additive; test bi-allelic, and multiallelic; minsize 10; min_freq 0.000; P 1.000. RESULTS Genotyping of the AAT-repeat in our sample revealed nine different alleles, with product sizes ranging from 219 to 243 bp at 3 bp intervals. This is the same number of alleles found by Ujike et al.  in the Japanese population. The family structure of our sample consisted of 126 complete trios, 118 families with only one parent (for these families additional siblings were used to reconstruct the genotypes of the missing parents); 661 subjects were successfully genotyped with a genotyping completion rate of 0.91 (230 affected subjects, 431 relatives); in addition 29 genotypes were inferred for missing parents using siblings genotypes. The genotypes distribution were within the expected values of Hardy–Weinberg equilibrium (chi-squared ¼ 12.6078; P value ¼ 0.2464). Out of the 244 subjects with consensus diagnosis of schizophrenia only 148 were in the age range between 25 and 45 years old and 66 subjects satisfied the criteria for ‘‘hebephrenia’’: DSM-IV consensus diagnosis of schizophrenia with ‘‘LDPS -hebephrenia’’ scores of 30 or higher (Mean: 40.3, SD: 8.45). The cutoff point of 30 points is the percentile 58 of the normal distribution of the scores. This cutoff point was arbitrarily chosen and was the only one used in the present analyses. The vast majority of these 66 subjects were classified as undifferentiated or disorganized subtype at the time of the consensus diagnosis using the DSM-IV (Table III). In order to test if this score can be used as a quantitative trait for association analysis, we analyzed the normal distribution of the scores in the subjects with age range between 25 and 45 (N ¼ 148); the scores were normally distributed (Shapiro–Wilk Statistic ¼ 0.99) with a mean of 28.6 and a standard deviation of 13.4. Only the 148 subjects between the ages of 25 and 45 (plus their relatives) underwent quantitative trait analyses for the hebephrenia trait. TABLE III. DSM-IV Diagnoses for the Hebephrenic Subjects According to the Lifetime Dimensions of Psychosis Scale (LDPS) Score DSM-IV diagnoses Schizophrenia undifferentiated Schizophrenia disorganized Schizophrenia paranoid Schizophrenia residual Schizophrenia NOS N ¼ 66 % 37 13 7 2 7 56 20 10 4 10 282 Chavarrı́a-Siles et al. Using FBAT we found association of the ‘‘hebephrenic’’ type of SC and the (AAT)n-repeat marker of the CNR1 (multiallelic P value of 0.0368) (Table IV). None of the individual alleles showed positive evidence of association with hebephrenic schizophrenia, although allele 8 showed a trend towards positive association (bi-allelic P value ¼ 0.06333, Z value ¼ 1.805). Allele 4 showed a negative association to the hebephrenic subjects (bi-allelic P value ¼ 0.0243, Z value ¼ 2.251) (Table I). Using FBAT to test for association in all 244 subjects with consensus DSM-IV diagnosis of SC (broad phenotype), we found no association with the (AAT)n-repeat polymorphism (multi-allelic P value ¼ 0.238), although allele 8 showed a trend towards positive association with schizophrenia (bi-allelic P value ¼ 0.0710, Z value ¼ 1.805) and allele 4 showed a negative trend towards association with schizophrenia (bi-allelic P value ¼ 0.069, Z value ¼ 18.18). After confirming that the trait we used had a normal distribution in the age group 25–45, we used quantitative FBAT analysis to test if this trait was associated to the CNR1 (AAT)n polymorphism in this group (N ¼ 148) (Table V). The LDPS-hebephrenia trait was significantly associated with the (AAT)n-repeat marker of the CNR1 gene (multi-allelic P value ¼ 0.0282). Independently only allele 8 showed positive evidence of association with the trait (bi-allelic P value ¼ 0.0100, Z ¼ 2.573) and allele 4 showed a trend to negative association with the trait (bi-allelic P value ¼ 0.0606, Z ¼ 1.876) DISCUSSION Available biological and clinical evidence supports the hypothesis that CNR1 may be one of the genes which contributes to the pathophysiology of SC. The findings of Ujike et al.  that genetic variation near the CNR1 gene might be associated with hebephrenic schizophrenia, whose predominant symptoms in the tenth revision of the International Classification of Diseases (ICD-10) include negative and disorganized symptoms [World Health Organization, 2003] support this hypothesis. The hebephrenic subtype of SC is a distinct and more quantifiable phenotype than the broader category of SC described in the DSM-IV; it is characterized by predominant negative symptoms and disorganization. As early as Bleuler  in an attempt to classify ‘‘the schizophrenias’’ into more homogeneous groups, which he believed would one day be subdivided into their ‘‘natural subdivisions,’’ followed the work of Kraepelin and others and defined this type as having, among other qualities, ‘‘dull’’ emotions, ‘‘blunted’’ affect, ‘‘slackening’’ of attention, and disorganized emotions, and behaviors. The hebephrenic type of SC is a distinct subtype of schizophrenia [Jabs et al., 2002; World Health Organization, 2003] which is no longer in the DSM-IV, but which is still included in the ICD10. The patients in our sample were initially diagnosed using the DSM-IV criteria, making direct comparison with studies that use ICD-10 classification impossible. Nevertheless, in the present study, the Lifetime Dimensions of Psychosis Scale (LDPS) [Levinson et al., 2002] was scored for each subject; with this scale different symptoms can be scored to obtain quantitative phenotypes for genetic studies, and, as in the case of the present study, to describe different ‘‘subtypes’’ of schizophrenia, which can be analyzed for association to specific genes. Our study indirectly confirms (we used an alternative method to diagnose hebephrenia) the previous report of Ujike et al. that the (AAT)n microsatellite, located in the 30 untranslated region of the CNR1, is associated with hebephrenic schizophrenia. In both of these studies, the (AAT)n microsatellite was significantly associated with this particular phenotype. As classically defined [Bleuler, 1954] this type of schizophrenia is characterized by predominant negative symptoms and disorganization, which resembles what is seen in animal models of cannabinoid exposure [Viveros et al., 2005] and in chronic cannabinoid-induced psychosis [Halikas et al., 1971]. The current study is also the first study of the CNR1 gene and schizophrenia to control for potential stratification by utilizing family based association analyses. In sum, the current study provides clear evidence that supports the finding of Ujike et al.’s  that variation in the CNR1 gene confers risk for a hebephrenic type of schizophrenia. Our secondary hypothesis, that the same polymorphic marker from the CNR1 gene would not show association to the more general phenotype of SC, was also supported, as the global test of association was not significant, despite the fact that the affected sample of SC subjects (N ¼ 244) was almost four times as large as the sample of SC subjects who scored high on a lifetime dimension of hebephrenia (N ¼ 66). Our results when testing the SC phenotype, suggest that inclusion of other types of SC (other than the hebephrenic type) may dilute the power to find association of SC with the CNR1 gene. Out of four previous studies testing for association between SC as a general phenotype and the CNR1 gene, only one [Ujike et al., 2002] found significant association for the (AAT)n microsatellite (after correcting for multiple testing), and in this latter study, the association was driven primarily by the hebephrenic subtype of SC. The specific alleles associated with hebephrenia in our sample were not the same as in the Japanese study of Ujike et al., although the number of alleles for the AAT repeat were the same in both populations. The fact that different alleles are associated in these two populations might indicate that the specific allele is not the causal polymorphism for the trait, but that rather it is in linkage disequilibrium with a disease causing variant. If this were the case, this may reflect either different variants in the gene associated TABLE IV. Association Analysis of CNR1 (AAT)n-Repeat Marker With Hebephrenic Schizophrenia (HSC) and Schizophrenia (SC) in the Central Valley of Costa Rica population HSC (N ¼ 66) (AAT)-repeat a Frequency 4 5 6 7 8 Multiallelic P: 0.233 0.053 0.179 0.204 0.255 Alleles P value <0.05 are in bold. a Only informative alleles are shown. Z value P value 2.251 0.0243 1.732 0.0832 1.460 0.1444 0.115 0.9080 1.857 0.0633 P ¼ 0.0368 SC (N ¼ 244) Z value P value 1.818 0.0690 0.573 0.5668 0.115 0.9086 0.164 0.8697 1.805 0.0710 P ¼ 0.2382 CNR1 and Susceptibility to Hebephrenic Schizophrenia TABLE V. Association Analysis of CNR1 (AAT)n-Repeat Marker With the LDPS-Hebephrenia Quantitative Trait in the Central Valley of Costa Rica population LDPS-Hebephrenia trait (N ¼ 148b) (AAT)-repeat (allelesa) Z value P Value 4 5 6 7 8 Multi-allelic P: 1.876 0.952 1.905 0.358 2.573 0.060666 0.340885 0.056825 0.720442 0.010084 0.028279 P value <0.05 are in bold. a Only informative alleles are shown. b Only subjects in age range 25–45 were used for this analysis. with the disease (independent variants which arose by mutation in the two different populations) or the effect of recombination between this marker and a shared disease causing variant, which occurred at some point during the historical divergence of the two populations. The study by Martinez-Gras et al.  did not present corrected (for multiple testing) association results for the (AAT)n microsatellite analyses, although they did show a strong negative association of one particular allele (allele 4) with the phenotype of Schizophrenia. In the current study, we show a trend towards significance for that particular allele being negatively associated with SC (P ¼ .069, Table I). Interestingly, in our sample, allele 4 is also negatively associated with the hebephrenic type of SC (P ¼ .024, Table I). Since the CVCR population shares common ancestry with the Spanish population [Escamilla et al., 1996], it is possible that both populations share particular genetic variants in the CNR1 gene which contribute to schizophrenia or hebephrenic schizophrenia in particular, which are in tight linkage disequilibrium with the (AAT)n microsatellite. In such a situation, the two populations might share common alleles in this microsatellite which predict either increased or decreased risk for schizophrenic illnesses. As a secondary analysis we tested the trait ‘‘LDPShebephrenia’’ for those subjects in the age range 25–45 for association with the CNR1 in order to confirm if this trait can be used as a quantitative trait for association studies. The results confirm the association of the trait with the (AAT)n repeat marker (multi-allelic P value ¼ 0.028); allele 8 which previously showed a positive trend towards association with SC and the hebephrenic type of SC, is positively associated with the trait (bi-allelic P value ¼ 0.010); and allele 4 showed a trend toward negative association with the trait (bi-allelic P value ¼ 0.060), which is in relation to what we found for the phenotype hebephrenia in the CVCR population. Our clinical results also illuminate the complexity of the schizophrenic phenotype, as seen in clinical practice. Out of over 400 subjects being treated for Schizophrenia in the public health system of Costa Rica (by ICD-10 criteria, which includes a schizoaffective type of schizophrenia), only approximately half of these subjects met DSM-IV criteria for SC after a best estimate consensus process. We have also shown that subjects with a major lifetime component of what in the past would have been considered hebephrenia are now classified with a variety of diagnostic types under the DSM-IV, predominantly undifferentiated and disorganized (Table II). Since subtypes of schizophrenia may change over the course of a subject’s life [Kendler et al., 1985], lifetime dimensional ratings, such as those obtained by the LDPS, may play a special role in better characterizing biological types of schizophrenia. Furthermore, 283 our data suggest that dimensions, rather than categories of schizophrenic subtypes, might prove especially useful in teasing out which genes underlie what Eugen Bleuler termed ‘‘the Schizophrenias.’’ As pointed out previously by Martinez-Gras et al.; although the (AAT)n-repeat CNR1 polymorphism is associated with (in our study) hebephrenic schizophrenia, this polymorphism may not be the functional polymorphism directly responsible for this association. It is more likely that this polymorphism is in linkage disequilibrium with other polymorphisms in the CNR1 gene, which may directly cause psychopathologic changes contributing to schizophrenia [Martinez-Gras et al., 2006]. Targeted sequencing and association analyses are a logical next step to identify direct causal variants of the CNR1 gene. The present study has several limitations. First, we used an alternative approach to define which subjects had a phenotype of ‘‘hebephrenic’’ schizophrenia, which differs from the diagnostic procedure used in the original study by Ujike et al.  (ICD-10 classification of hebephrenic schizophrenia). Since the ICD-10 classification is based on clinical observation, we were able to use the LDPS retrospectively to obtain similar clinical information. Second, as this was the first time the LDPS has been used to try to define a ‘‘hebephrenic’’ type of schizophrenia, we used our best judgement to define procedures and cutoff points for defining this phenotype. Clearly other cutoff points could be selected, and ‘‘best fit’’ post hoc analyses may eventually be useful to find the best fit for defining a particular phenotype associated to the CNR1 gene. Such analyses, however, are beyond the scope of the present study, which was focused primarily on testing whether a ‘‘hebephrenic’’ type of schizophrenia was associated with the CNR1 gene. All in all, our current findings suggest that different genetic and pathophysiologic mechanisms may underlie different forms of SC, and that these different forms may be quantifiable using dimensional rating systems, in addition to classical categorical systems. ACKNOWLEDGMENTS This project was supported by NIH Research Grant # D43 TW06152 funded by the Fogarty International Center, The National Institute on Drug Abuse and the National Institute of Mental Health. We are indebted to the patients and family member who participated in this study. We also want to thank the psychiatry departments of hospitals and clinics in Costa Rica that collaborated in this project (in particular Hospital Nacional Psiquiatrico in Pavas). We thank the personnel of the CIBCM at the University of Costa Rica for their assistance collecting the sample. REFERENCES American Psychiatric Association. 1994. Diagnostic and statistical manual of mental disorders, 4th ed. Washington, DC: American Psychiatric Press. Bleuler E. 1911. Originally published in German as Dementia Praecox oder die Gruppe der Schizophrenien. English version. 1950.Dementia Praecox or the Group of Schizophrenias. New York: International Universities Press. Bleuler M. 1954. The concept of schizophrenia. Am J Psychiatry 111:382– 383. Escamilla MA, Spesny M, Reus VI, Gallegos A, Meza L, Molina J, Sandkuijl LA, Fournier E, Leon PE, Smith LB, Freimer NB. 1996. Use of linkage disequilibrium approaches to map genes for bipolar disorder in the Costa Rican population. Am J Med Genet 67:244–253. Escamilla MA, McInnes LA, Spesny M, Reus VI, Service SK, Shimayoshi N, Tyler DJ, Silva S, Molina J, Gallegos A, Meza L, Cruz ML, Batki S, Vinogradov S, Neylan T, Nguyen JB, Fournier E, Araya C, Barondes SH, Leon P, Sandkuijl LA, Freimer NB. 1999. Assessing the feasibility of 284 Chavarrı́a-Siles et al. linkage disequilibrium methods for mapping complex traits: An initial screen for bipolar disorder loci on chromosome 18. Am J Hum Genet 64: 1670–1678. Fritzsche M. 2001. Are cannabinoid receptor knockout mice animal models for schizophrenia? Med Hypotheses 56:638–643. Gottesman II. 1994. Schizophrenia epigenesis: Past, present, and future. Acta Psychiatr Scand Suppl 384:26–33. Halikas JA, Goodwin DW, Guze SB. 1971. Marihuana effects. A survey of regular users. JAMA 217:692–694. Howlett AC, Breivogel CS, Childers SR, Deadwyler SA, Hampson RE, Porrino LJ. 2004. Cannabinoid physiology and pharmacology: 30 years of progress. Neuropharmacology 47:345–358. Jabs BE, Verdaguer MF, Pfuhlmann B, Bartsch AJ, Beckmann H. 2002. The concept of hebephrenia over the course of time with particular reference to the Wernicke-Kleist-Leonhard School. World J Biol Psychiatry 3:200– 206. Kendler KS, Gruenberg AM, Tsuang MT. 1985. Subtype stability in schizophrenia. Am.J.Psychiatry 142:827–832. Kessler RC, McGonagle KA, Zhao S, Nelson CB, Hughes M, Eshleman S, Wittchen HU, Kendler KS. 1994. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States. Results from the National Comorbidity Survey. Arch Gen Psychiatry 51:8–19. Leroy S, Griffon N, Bourdel MC, Olie JP, Poirier MF, Krebs MO. 2001. Schizophrenia and the cannabinoid receptor type 1 (CB1): Association study using a single-base polymorphism in coding exon 1. Am J Med Genet 105:749–752. Levinson DF, Mowry BJ, Escamilla MA, Faraone SV. 2002. The Lifetime Dimensions of Psychosis Scale (LDPS): Description and interrater reliability. Schizophr Bull 28:683–695. Martinez-Gras I, Hoenicka J, Ponce G, Rodriguez-Jimenez R, JimenezArriero MA, Perez-Hernandez E, Ampuero I, Ramos-Atance JA, Palomo T, Rubio G. 2006. (AAT)n repeat in the cannabinoid receptor gene, CN R1:Association with schizophrenia in a Spanish population. Eur Arch Psychiatry Clin Neurosci 256:437–441. Montero AP, Balderas T, Pereira M, Lizano A, Medina R, Contreras S, Dassori A, Katz I, Levinson D, Garbarz J, Escamilla MA, Raventos H. 2002. Resultados de la categorización de una muestra de pacientes con diagnostico de egreso de esquizofrenia. Acta Medica Costarricense 44: 74–78. Nurnberger JI Jr, Blehar MC, Kaufmann CA, York-Cooler C, Simpson SG, Harkavy-Friedman J, Severe JB, Malaspina D, Reich T. 1994. Diagnostic interview for genetic studies. Rationale, unique features, and training. NIMH Genetics Initiative. Arch Gen Psychiatry 51:849– 859. Pazos MR, Nunez E, Benito C, Tolon RM, Romero J. 2005. Functional neuroanatomy of the endocannabinoid system. Pharmacol Biochem Behav 81:239–247. Rodriguez De FF, Gorriti MA, Bilbao A, Escuredo L, Garcia-Segura LM, Piomelli D, Navarro M. 2001. Role of the endogenous cannabinoid system as a modulator of dopamine transmission: Implications for Parkinson’s disease and schizophrenia. Neurotox Res 3:23–35. Rodriguez de FF, Del A I, Bermudez-Silva FJ, Bilbao A, Cippitelli A, Navarro M. 2005. The endocannabinoid system: Physiology and pharmacology. Alcohol Alcohol 40:2–14. Solowij N, Michie P. 2007. Cannabis and cognitive dysfunction: Parallels with endophenotypes of schizophrenia? J Psychiatr Neurosci 32:30–52. Tsai SJ, Wang YC, Hong CJ. 2000. Association study of a cannabinoid receptor gene (CNR1) polymorphism and schizophrenia. Psychiatr Genet 10:149–151. Ujike H, Morita Y. 2004. New perspectives in the studies on endocannabinoid and cannabis: Cannabinoid receptors and schizophrenia. J Pharmacol Sci 4:376–381. Ujike H, Takaki M, Nakata K, Tanaka Y, Takeda T, Kodama M, Fujiwara Y, Sakai A, Kuroda S. 2002. CNR1, central cannabinoid receptor gene, associated with susceptibility to hebephrenic schizophrenia. Mol Psychiatry 7:515–518. Viveros MP, Llorente R, Moreno E, Marco EM. 2005. Behavioural and neuroendocrine effects of cannabinoids in critical developmental periods. Behav Pharmacol 16:353–362. Wigginton JE, Abecasis GR. 2005. PEDSTATS: Descriptive statistics, graphics and quality assessment for gene mapping data. Bioinformatics 21:3445–3447. World Health Organization. 2003. International Statistical Classification of Diseases Then Revision 2nd ed. Geneva: World Health Organization.