The K-W test was used to assess the presence o f any association between clinical characteristics measured for each patient in the two population samples and genotype. This test determines if an association exists but does not determine where the association lies. The results o f the K-W test are summarised in table 3.9. A significance level, a o f 0.05 was used. Neither the age-at-onset (AAO), negative symptom, movement disorders, other clinical ratings or measures o f overall severity o f disease showed a significant association with genotype. A slight trend was suggested between genotype and the PJM outcome ratings, P = 0.06. After applying the Bonferroni correction (see appendix B7) for the seven tests performed for the EA population sample, the trend between PJM scores and genotype no longer exists {P = 0.42 after correction).
The weak association between PJM outcome ratings and genotype suggested by the K- W test was further assessed using the Robust rank order test by performing two tailed pair-wise genotype comparisons. The pair-wise comparisons performed were: homozygosity for allele 1 versus homozygosity for allele 2, heterozygosity versus homozygosity for allele 1 and then allele 2 and finally homozygosity versus heterozygosity. The results are presented below table 3.9.
The D2-like dopamine receptors
Table 3.9: Results o f K-W test, testing the association between clinical characteristics and
M scl genotypes in patients. Degree of freedom = 2
Number of groups = 3 [1-1/ 1-2/ 2-2] Number of cases: EA = 66 and EY = 68
Characteristics He - corrected for ties Probability EA
AAO 1.45 0.50
Negative symptoms (HENs) 0.63 0.75
Parkinsonian rating 0.04 O j# Dyskinesia 3.79 0.15 Catatonia 3.05 0.25 PJM outcome ratings 5.26 0.06 Rogers score 3.95 0.15 EY AAO 1.07 0.58 HENs 2.21 0.33 Mini-Mental Score (MMS) 0.74 0.69 Global 1.51 0.47 1-1 v s . 2 -2 Ù P -1.489 =0. 13 P a ir w is e c o m p a r i s o n s o f g e n o t y p e 1-1 v s . 1 -2 1 -2 v s . 2 -2 Ù ^ Ù ^ -2.240 = 0.02 -0.365 >1 1 -1 /2 -2 v s . 1 -2 Ù P -1.943 = 0.06
The negative sign indicates the direction in which the association lies. The results o f the four comparisons above suggest that the association lies with the M scl 1-2 genotype. PJM outcome ratings o f schizophrenic patients who are heterozygous for the M scl polymorphism have a higher PJM outcome score than patients who are homozygous for allele 1 (ù = -2.240, P - 0.02). A similar trend was suggested comparing hom ozygosity versus heterozygosity (ù = -1.943, P - 0.06). These results suggest the association to be genotypic and not allelic. However, after correcting for the four tests, the association between genotype 1-2 and outcome scores lies at the border line o f significance {P - 0.05). PJM outcome ratings was measured on a scale o f 0 to 3. A score o f 0 meaning complete recovery, no active symptoms, no signs o f deterioration, no or minimal social impairment and are able to work at previous level, and a score o f 3 meaning moderate or severe ongoing active symptoms, signs o f deterioration, unable to work and patients might need long term hospitalisation. Therefore schizophrenic patients who are heterozygous are more severely affected than individuals who are homozygous for allele
1 or 2.
The D2-like dopamine receptors
3.2.4 DISCUSSION
The results o f this study did not provide evidence for an association between alleles and genotypes o f the M sc\ polymorphism o f the DRD3 and schizophrenia in the tw o population samples examined. The results o f this study does not support the original findings o fC ro cq e/ ah (1992, see section 1.12.2.2). The possibility that these negative results could be due to a type II error which could have been generated by a small sample size or population stratification must be considered. The power calculations by Nanko et al. (1993) suggested that their study consisting o f 91 patients and 91 controls had a power greater than 90% to detect an effect o f the magnitude o f that reported by Crocq et al. (1992). Therefore the EA sample size (83 patients and 77 controls), the EY sample size (72 cases and 85 controls) and the pooled sample size, according to the pow er calculations by Nanko et al. (1993), seems to have sufficient power to detect association. Population stratification is an important confounding factor in case/control association studies. However, it is unlikely that population stratification served as confounding factor in this study, since the H-W equilibrium test did not show any significant deviation. Therefore the patients and control samples represent a homogeneous population.
Unlike other studies that have reported weak associations between family history, response to clozapine and early AAO and homozygosity for the M scl polymorphism, such associations were not observed in this study. However, an association was observed between PJM outcome ratings and the genotype 1-2 before correcting for multiple testing {P = 0.02, Robust rank order test). Patients with the genotype 1-2 had higher PJM scores and are severely affected. Association with the 1-2 genotype has not been previously reported. This finding adds to the inconsistencies o f the results reported by the numerous studies o f the M scl polymorphism (see section 1.12.2.2, table 1.3).
The contradicting findings o f the large numbers of studies performed to date have led to the speculation that the associations reported may be chance findings. Conversely it has also been suggested that most studies involved multiple testing and as a result the significance levels would be unlikely to survive corrections in many cases (as was observed with the present study). This leads to type 11 errors. Furthermore, it has also been suggested that under-reporting o f negative results could also lead to an over representation o f positive associations (Ebstein et al., 1995). As a result it has been
The D2-like dopamine receptors proposed by several authors that these studies do support the possibility that the DRD3 or a locus (either downstream to exonl or upstream in the promoter region) in linkage disequilibrium with it may be associated with schizophrenia. It has been suggested that the effect caused is likely to be small and could be global, or as one which manifests in specific ethnic groups or is an aetiological factor in certain demographic or clinical subgroups (Ebstein et al., 1995; Mant et al., 1994; Owen and O'Donovan, 1997).
3.2.5 CONCLUSION
No association between the alleles or the genotypes o f the M sc\ polymorphism and schizophrenia was detected in the EA or the EY population samples.
The D2-like dopamine receptors
3.3 INVESTIGATING THE 5^-UPSTREAM REGION OF THE DRD3 IN