Segregation analysis of schizophrenia and related disorders.

Segregation analysis was applied to 79 nuclear families ascertained through chronic schizophrenic probands. Analysis was performed on the diagnosis of schizophrenia alone and on schizophrenia and schizotypal personality disorder (milder phenotype) combined. The models used were the transmission probability model and the mixed model. Because the disease is associated with reduced fertility, all likelihoods were calculated conditional on parental phenotypes. However, compatibility of the mating-type distribution predicted by each model with the observed was also examined. In all analyses, results suggested consistency with genetic transmission. In the analysis of schizophrenia alone, discrimination among models was difficult. In the analysis including the milder phenotypes, all single-locus models without polygenic background were excluded, while pure polygenic inheritance could not be eliminated. The polygenic model also gave good agreement with supplementary observations (lifetime disease incidences, mating-type distribution, and monozygotic twin concordance). The estimated components of variance for the polygenic model were: polygenes (H) 81.9%; common sib environment (B) 6.9%; random environment (R) 11.2%. Although the polygenic model was parsimonious, segregation analysis and the supplementary observations were also consistent with a mixed model, with a single major locus making a large contribution to genetic liability. Such a locus is more likely to be recessive than dominant, with a high gene frequency and low penetrance. The most likely recessive mixed model gave the following partition of liability variance: major locus, 62.9%; polygenes, 19.5%; common sib environment, 6.6%; and random environment, 11.0%.