Is the Gly82Ser polymorphism in the RAGE gene relevant to schizophrenia and the personality trait psychoticism?

Personality trait study participants

The 2 population-based cohorts, one consisting of middle-aged women and the other of middle-aged men, investigated in this study were originally recruited from the National Population Register for studies on obesity, anthropometrics and cardiovascular risk factors. Clinical data and further details on these 2 cohorts have been reported previously.16,17 At the time of investigation, no exclusions owing to somatic or psychiatric disease were made in either cohort. We obtained blood samples from all women and men for genotyping. All volunteers gave their informed consent, and the study protocol was approved by the ethical committee at the University of Gothenburg.

Personality assessment

The participants in both cohorts were assessed using the Karolinska Scales of Personality (KSP).18 This inventory consists of a self-report questionnaire that has been widely used in studies involving biologic correlates of personality traits.14,18–20 Previous studies have investigated the validity and reliability of the KSP questionnaire and found that interindividual variations were both stable and partly heritable for several of the subscales.19,21 The KSP is based on 135 items forming 15 subscales that can further be classified into 4 factors covering different dimensions of temperament: extraversion (comprising the impulsiveness and monotony avoidance subscales), neuroticism (comprising the somatic anxiety, muscular tension, psychic anxiety, psychasthenia, inhibition of aggression, guilt and socialization subscales), psychoticism (comprising the detachment and suspicion subscales) and nonconformity (comprising the verbal aggression, indirect aggression, irritability and social desirability subscales).22 The KSP factors and subscales are standardized to adjust for age and sex using normative data to have a mean of 50 and standard deviation (SD) of 10 (i.e., T-scores).

Case–control study participants

We recruited all patients with schizophrenia in the northwestern part of Stockholm County, and they have been described in detail previously.23,24 All reviews of hospital case notes, interviews and diagnostic formulations were performed by psychiatrists trained in Sweden.

The case–control cohort could be divided into 2 groups. The first group consisted of patients recruited from psychiatric clinics in northwestern Stockholm County who were assessed for lifetime psychiatric diagnosis using the DSM-III-revised and for geographic origin using data from hospital case notes, clinical and/or structured interviews and parish register data. Further assessments were made with regards to clinical subtraits and subdiagnosis of schizophrenia based on the DSM-III-revised. The second group was recruited in a similar way, with the following differences: in addition to the Structured Clinical Interview for DSM-III-revised, patients were assessed using the psychosis module of the Schedules for Clinical Assessment in Neuropsychiatry. Evaluations of lifetime diagnosis were conducted by reviewing hospital records using DSM-IV, not DSM-III-revised, as the diagnostic tool. However, investigations of the comparability of the 2 diagnostic systems showed good agreement in the present sample.23 Identification of geographic origin in the second group was based on interview data only.

We selected participants based on diagnosis and race, and we excluded participants with an unknown subdiagnosis or one that was not included among the 5 subclassifications of schizophrenia (i.e., 295.1–3, 295.6 and 295.9). We defined age at onset of schizophrenia as the patient’s age at the time of the first appearance of psychotic symptoms; one of us (E.G.J.) assessed age at onset by reviewing the patients’ lifelong psychiatric medical records.

The controls were selected randomly from among the 1090 participants of the Kungsholmen project, which consisted of people aged 75 years and older living in Stockholm, Sweden.25 All participants in the case–control study provided informed consent, and the study was approved by the Ethics Committee of the Karolinska Hospital and the Karolinska Institutet.

Genotyping

Human genomic DNA was extracted from blood samples using the QIAamp DNA Blood Mini Kit (Qiagen). The studied polymorphism was analyzed using polymerase chain reaction (PCR) as the amplifying step, followed by genotype determination by pyrosequencing.

The Gly82Ser polymorphism is located in exon 3 of the gene coding for RAGE. The primers that were used were 5′-ATT TGG ATC CCC GTC ACT CT-3′ as the forward and 5′-biotin-GCC TGG CAC CGG AAA ATC-3′ as the reverse primer. The PCR step was performed using HotstarTaq polymerase (Qiagen) and GeneAmp PCR System 9700 (Applied Biosystems). We used a total volume of 20 μL containing 0.3 μM of primers, 1.5 mM of MgCl₂, about 50 ng of DNA and 200 μM of each deoxyribonucleotide triphosphate. An initial 15-minute denaturation step at 95°C was followed by 41 cycles of 15 seconds at 95°C, 30 seconds at 62°C and 15 seconds at 72°C. Once the cycles were completed, the reaction was incubated at 72°C for 7 minutes and then left at 4°C. The PCR product was genotyped using a Pyrosequencer PSQ 96 and the PSQ 96 SNP Reagent Kit (Qiagen). To identify the polymorphism, we used 15 pmol of the sequencing primers 5′-CGT GTC CTT CCC AAC-3′. We used a total of 20 μL of PCR product for pyrosequencing in accordance with the manufacturer’s instructions.

Statistical analysis

The association between KSP T-scores and genotype was assessed using an independent samples t test. The interaction between sex and genotype was assessed with personality traits as the dependent variable in a linear regression.

We compared genotype frequencies between patients with schizophrenia and controls using the Fisher exact test (1-tailed), and we tested homogeneity between odds ratios (ORs) using the Breslow–Day statistic. The possible influence of the Gly82Ser genotype on age at onset of schizophrenia was investigated using an independent samples t test (1-tailed).

Deviation from Hardy–Weinberg equilibrium for the Gly82Ser allele was assessed using Haploview (version 4.2). Power analyses in terms of effect-size calculations were assessed in all cohorts using the G*Power software. We performed the statistical analyses using SPSS Statistics for Mac, version 18.0.0. We considered results to be significant at p < 0.05, and p values are nominal except when explicitly stated as corrected.

Personality trait study participants

The 2 population-based cohorts comprised 270 women aged 42 years and 247 men aged 51 years. The KSP questionnaires were returned by 204 women and 153 men; however, 26 women and 20 men left some questions blank, which resulted in missing values in their personality profiles.

Case–control study participants

We recruited 173 patients with schizophrenia to participate in the study. Of these, 137 patients were assessed using the DSM-III-revised, reviews of hospital case notes, clinical and/or structured interviews and parish register data. Further assessments were made with regards to clinical subtraits and subdiagnosis of schizophrenia based on DSM-III-revised. The remaining 36 patients were assessed using the Structured Clinical Interview for DSM-III-revised and the Schedules for Clinical Assessment in Neuropsychiatry. In addition, evaluations of lifetime diagnosis were conducted using the DSM-IV, rather than the DSM-III-revised, as a diagnostic tool.

The mean age of all 173 patients was 43.7 (SD 16.5) years. We excluded 32 patients with an unknown subdiagnosis or one that was not included among the 5 subclassifications of schizophrenia. One patient with ancestors from Sudan was also excluded, resulting in a sample of 140 white European patients (53 women and 87 men) with a mean age of 44.9 (SD 16.7) years. The age at onset of schizophrenia was available for 136 patients: mean 24.2 (SD 7.6) years.

We recruited 258 controls (109 women and 149 men) for participation in the study. Their mean age was 80.5 (SD 4.6) years and they were all white Europeans.

Genetic characteristics

Of the 270 participants in the female personality trait cohort, 250 were homozygous and 20 were heterozygous for the 82Gly allele. None was found to be homozygous for the 82Ser variant. In the male personality trait cohort, 220 participants were homozygous and 22 were heterozygous for the 82Gly allele, and 1 was homozygous for the 82Ser variant. In the statistical analysis, this latter participant was grouped with those who were heterozygous for the 82Gly allele. Genotyping of 4 samples in the male cohort failed owing to lack of DNA. Power calculations revealed that with a 5% significance level and 80% power, we would be able to detect an overall mean difference between the genotype groups as low as Cohen’s d = 0.54.

Compared with carriers of the 82Gly/Gly genotype, the presence of the 82Ser variant was found to be associated with significantly higher scores in the psychoticism factor of the KSP (Table 1 and Fig. 1) in the combined population. Analysis of the subscales comprising this factor revealed significantly higher scores in detachment and suspicion for 82Ser carriers. When stratifying for sex, significant outcomes were seen only among women. However, we did not find any significant effect of interactions between sex and genotype on personality traits when analyzing the combined group.

Genotype frequencies from the case–control study are shown in Table 2. We were unable to genotype 2 samples in the patient group. Power calculations revealed that with a 5% significance level and 80% power we would be able to detect an OR of 2.4 and an earlier age of onset (Cohen’s d = 0.63) among 82Ser carriers compared with participants homozygous for 82Gly.

Compared with Gly homozygotes the 82Gly/Ser genotype was associated with an increased risk for schizophrenia (OR 2.4, 95% confidence interval 1.2–4.9; Table 2). When stratifying for sex, the increased risk was significant only in men, but a test of homogeneity of the OR in men (3.7) and women (1.4) did not reveal any significant differences (p = 0.24). Part of the difference could be explained by the discrepancies in genotype frequencies between the sexes in the control group. Using the entire control group, we found ORs of 2.7 and 1.9 for men and women, respectively. The Gly82Ser polymorphism was not found to significantly affect the age of onset in the patient group, but when men and women were analyzed separately, male carriers of the 82Ser variant showed a borderline significance of earlier age at onset of disease compared with the 82Gly/Gly group (Table 3). There was no significant effect of the interaction between sex and genotype on age at onset when analyzing the combined group (p = 0.39).

The distribution of the genotypes did not differ significantly from the Hardy–Weinberg equilibrium in either of the cohorts (all p > 0.05).

Limitations

The current study had some limitations. In the personality trait study, limitations included the selection process, the limited number of assessed participants and the use of a self-report measure. The limitations in the case–control study included the modest cohort size and population stratifications; however, to minimize the possibility of the latter, we ensured that both patients and controls were of the same race (i.e., white European), were unrelated and were recruited from the same area of Sweden. An additional limitation was the difference in age between patients and controls. Although unlikely, we cannot presently exclude the possibility that the 82Ser allele is associated with lower life expectancy, and therefore is less common in the older control group, rather than in patients with schizophrenia specifically.