An overview of the results is given in t1. Two allelic subgroups within each diagnostic group were formed by placing subjects homozygous for allele 1 into one group and collapsing allele 2 carriers (1/2 and 2/2) into the other. These subgroups (1/1 versus 1/2 and 2/2) did not differ significantly with respect to age, height, weight, education, verbal IQ, nicotine or alcohol consumption, age at onset of illness, or duration of illness (each t<1.05, df=90, p>0.30). Interleukin-1β genotype was not associated with diagnosis. The sample size allowed for sufficient power (0.80) to correctly accept the null hypothesis if the effect size was 0.60 (Cohen’s d) or larger.
Diagnosis and genotype had no main effects on either gray or white matter across the whole brain (t2). However, for gray matter volumes, the three-way interaction of diagnosis, genotype, and region indicated that there could be a region-specific effect. Separate three-factorial ANOVAs for each region revealed interaction effects of diagnosis and genotype in the frontal lobe (F=5.31, df=1, 88, p<0.03) and temporal lobe (F=4.26, df=1, 88, p=0.04). Further F tests showed that there was no genotype effect in the comparison subjects for frontal gray matter (F=0.97, df=1, 46, p=0.33) or temporal gray matter (F=0.07, df=1, 46, p=0.80), but among the schizophrenic patients allele 2 carriers had significantly smaller gray matter volumes in these areas than did 1/1 homozygotes (frontal: F=6.06, df=1, 42, p<0.02; temporal: F=11.62, df=1, 42, p=0.001). The ANOVA for white matter volumes showed a diagnosis-by-genotype interaction that was independent of brain region. Again, white matter volumes in comparison subjects were not affected by interleukin-1β genotype (F=1.36, df=1, 46, p=0.25), whereas schizophrenic patients had overall white matter deficits if they were allele 2 carriers (F=5.54, df=1, 42, p<0.03).