Background: Several lines of evidence suggest that the clinical heterogeneity of schizophrenia is due to genetic heterogeneity. Genetic heterogeneity may decrease the signal-to-noise ratio in linkage and association studies. Therefore, linkage studies of clinically homogeneous classes of psychotic illness may result in greater power to detect at least some loci.
Methods: Latent class analysis was used to divide psychotic subjects from 270 Irish high-density families (N = 755) into six classes based on the Operational Criteria Checklist for Psychotic Illness. We heuristically call them Bipolar, Schizoaffective, Mania, Schizomania, Deficit Syndrome, and Core Schizophrenia. The latter four had prevalences of greater than .08 and were individually tested for linkage in a 10-cM nonparametric autosomal genomewide scan. Empirical significance was determined using 200 simulated genome scans.
Results: Seven regions achieved empirical criteria for suggestive significance for at least one latent class: 5q23.2-q35.3, 8q13.1-q23.1, 10q23.33-q26.3, 12q21.2-q24.32, 19q13.32-q13.43, 20p13-q22.3, and 21q11.2-q22.3. Five of 200 simulated scans resulted in seven suggestively significant loci (experiment-wide p = .03). Furthermore, at 20p13-p12.2, the Mania and Schizomania classes individually achieved criteria, whereas Deficit Syndrome had a suggestive logarithm of the odds peak 28 cM centromeric to this locus.
Conclusions: Using empirically derived, clinically homogeneous phenotypes, four chromosomal regions were suggestively linked but provided little evidence of linkage using traditional operationalized criteria. This approach was particularly fruitful on chromosome 20, which had previously yielded little evidence of linkage. Future studies of psychiatric illness may increase their ability to detect linkage or association by using clinically homogeneous phenotypes.