Quantitative trait loci analysis of high density schizophrenia families does not provide greater power than traditional linkage methods to detect susceptibility loci for schizophrenia

Background: Schizophrenia is multidimensional and heterogeneous. Its various clinical features maybe differentially influenced by separate genetic loci. Furthermore, symptom dimensions in affected probands may be etiologically continuous with personality features in their non-psychotic relatives. These aspects of the disorder may make quantitative trait loci (QTL) analysis a feasible method for detecting susceptibility loci, which may have greater statistical power than traditional linkage methods. Methods: The genotypic and phenotypic data in this study were collected as part of the Irish Study of High Density Schizophrenia Families (ISHDSF). Two hundred sixty-five pedigrees were ascertained on the basis of having two or more members with schizophrenia or poor outcome schizoaffective disorder. Diagnoses were made using a structured interview based on the SCID. All patients were rated on the Major Symptoms of Schizophrenia Scale (MSSS), integrating clinical and course features throughout the course of illness. Non-psychotic relatives were rated on seven dimensions of schizotypy based on the Structured Interview for Schizotypy (SIS). Unified scales of positive and negative symptoms were constructed using schizotypal symptoms at the lower ends and schizophrenic symptoms at the higher ends. All subjects were assigned scores on these two scales. Identity by descent (IBD) probabilities and Pi-hat were calculated for all sib pairs at a range of locations on chromosomes 6,8, and 10, in regions previously shown to be linked, using the program MAP-MAKER/SIBS. The effect of a QTL at a given location was assessed using the maximum likelihood approach of Fulker and Cherny. At each location, the log-likelihood of sibpairs at each location were computed, and these were summed to obtain the overall log-likelihood of the sample, LI. The log-likelihood of a model with zero QTL effect, L0, was also computed, and the likelihood ratio test was used to assess goodness of fit. This was implemented using the structural equation modeling package MX. Results: No QTL effect was detected at any location on any of the three chromosomes for either positive or negative symptoms. Conclusions: QTL methodology does not appear to have as much power to detect a susceptibility locus for schizophrenia as do traditional linkage methods in this sample of high density schizophrenia families from Ireland.