Biological phenotypes and genetic research on schizophrenia.

Archival family, twin and adoption studies have demonstrated that the lion's share of susceptibility to manifest schizophrenia is determined by genetic factors. These same archival studies have also showed that the inheritance of risk for schizophrenia does not follow simple Mendelian patterns and, like other common medical conditions such as heart disease and adult onset diabetes, it is likely caused by multiple genes and environmental factors. Schizophrenia, and probably all psychiatric illnesses, are polygenic disorders, which have a complex genetic architecture involving locus and allelic heterogeneity (i.e. multiple genes and more than one variation in the DNA sequence within a given gene), epistasis (i.e. nonadditive interactions between genes), pleiotropy (protean phenotypic manifestations of the same allele), incomplete penetrance, and environmental modification. This complexity means that there is a weak predictive relationship between the clinical diagnosis and an underlying genotype. Stated another way, the effect size of a specific allele or genotype on prediction of clinical diagnosis is likely to be small. These predictions have been confirmed by family based linkage studies of schizophrenia. The so-called genome scan linkage approach, which has been effective in mapping genetic loci related to rare inherited disorders caused by major effect genes, has been much less powerful in mapping common polygenes of minor pathogenic effect involved in complex genetic disorders. In the case of schizophrenia, family studies using non-coding DNA markers spanning the genome (i.e. the 'genome scan') have identified a few significant chromosomal susceptibility loci, that have been difficult to replicate across families even within the positive linkage studies and that do not segregate consistently with illness even within families that show linkage to one of these markers (1,2). Pedigree based linkage studies are exquisitely model dependent and, as the mode of inheritance of schizophrenia is unknown, the validity of findings from such studies is controversial. The use of nonparametric approaches, such as affected sib pair methods, while an important advance, has unfortunately not solved the problems of genetic complexity and even thousands of affected sib pairs may be underpowered to find susceptibility loci (3). The polygenic model of schizophrenia articulated by Gottesman and Shields (4) hypothesizes that qualitatively different genetic factors underlie different clinical subtypes or dimensions of schizophrenia, and that the genetics of subtypes may be simpler than the genetics of the complex phenotype. It has been argued that evidence for familial heterogeneity in the schizophrenic syndrome is of potential significance in the search for susceptibility loci, as it may allow a division of the sample before linkage analysis into etiologically distinct subgroups and, thus, increase power (5). These traditional arguments were based on clinical symptoms, not on underlying biology, and the clinical symptoms may not be sufficiently distinct to predict genotype. Genes do not encode hallucinations, delusions, or thought disorganization per se. Genes determine the structure of simple molecules in cells, usually proteins, and these proteins affect how cells process and respond to stimuli. A variation in the sequence of a gene that affects either the expression or activity of a protein might be expected to slightly vary the functional characteristics of a cell, which could lead to changes in the interactions that the cell has with other cells, in the connections and cell assemblies that develop, and in how such assemblies and networks operate as functional systems. Such biological effects at the level of cell and neural system function are far removed in biological space and time from the clinical psychopathology of schizophrenia and from phenotype at the level of clinical diagnosis. The diagnostic symptoms of schizophrenia are likely emergent phenomena related to underlying abnormalities in brain information processing, and the biological effects of genes are likely to be more predictable in terms of the underlying abnormalities in brain function. Because susceptibility genes bias towards the expression of a susceptible biology, biological abnormalities related to genetic risk for illness may represent aspects of biological susceptibility and, therefore, more direct gene effects. Indeed, the parsing of a complex phenotype into component biological traits is an increasingly recognized approach to complex genetic medical disorders (e.g. heart disease, obesity).