Molecular Psychiatry (2006) 11, 815–836 & 2006 Nature Publishing Group All rights reserved 1359-4184/06 $30.00 www.nature.com/mp FEATURE REVIEW Subtyping schizophrenia: implications for genetic research A Jablensky Centre for Clinical Research in Neuropsychiatry, School of Psychiatry and Clinical Neurosciences, The University of Western Australia, Perth, WA, Australia Phenotypic variability and likely extensive genetic heterogeneity have been confounding the search for the causes of schizophrenia since the inception of the diagnostic category. The inconsistent results of genetic linkage and association studies using the diagnostic category as the sole schizophrenia phenotype suggest that the current broad concept of schizophrenia does not demarcate a homogeneous disease entity. Approaches involving subtyping and stratification by covariates to reduce heterogeneity have been successful in the genetic study of other complex disorders, but rarely applied in schizophrenia research. This article reviews past and present attempts at delineating schizophrenia subtypes based on clinical features, statistically derived measures, putative genetic indicators, and intermediate phenotypes, highlighting the potential utility of multidomain neurocognitive endophenotypes. Molecular Psychiatry (2006) 11, 815–836. doi:10.1038/sj.mp.4001857; published online 27 June 2006 Keywords: schizophrenia; heterogeneity; pleiotropy; subtypes; genetics; endophenotypes Introduction to replace it with a pre-Kraepelinian notion of a unitary ‘psychosis’.9 Given the protean nature of the More than a century since the delineation of dementia symptoms of schizophrenia and the poor coherence of praecox by Kraepelin,1 the aetiology, neuropathology, the clinical and biological findings, such doubts are and pathophysiology of schizophrenia remain elusive. not without reason. However, simply dismantling the Despite the availability of criteria2,3 allowing reliable concept is unlikely to beget an alternative model that diagnostic identification, schizophrenia essentially would account for the host of clinical phenomena and represents a broad clinical entity defined by subjective research data consistent with a disease hypothesis symptoms, behavioural signs, and patterns of course. of schizophrenia. A well replicated epidemiological Research has marked out numerous biological indica- finding is that, on average, about 1% of the popula- tors tentatively associated with the disorder, including tion develop schizophrenic disorders in their life- neurocognitive dysfunction, brain dysmorphology, time.10,11 Further evidence that schizophrenia is and neurochemical abnormalities. Yet none of these not an arbitrary construct comes from the relative variables has to date been definitively proven to invariance of its clinical presentation and incidence possess the sensitivity and specificity expected of across populations and over time, established by field a diagnostic test. Genetic linkage and association research conducted by the World Health Organization studies have targeted multiple candidate loci and (WHO) in over 20 diverse populations and cultures.12 genes, but failed to demonstrate that any specific gene In this article, we survey the evidence for phenotypic variant, or a combination of genes, is either necessary variation and likely aetiological heterogeneity as major or sufficient to cause schizophrenia.4–6 Thus, the sources of inconsistent findings and argue that, like existence of a specific brain disease underlying other complex disorders, schizophrenia is not a schizophrenia is still a hypothesis, for which no nosological monolith. We review past and present conclusive proof or refutation has yet been produced. attempts at delineating subtypes based on clinical The paradox of an ever increasing volume of features, statistically derived measures, putative genetic research data, and the apparent stalemate of the indicators, and endophenotypes. The end point is that search for causes of the disorder, has fuelled doubts conceptual and methodological constraints, well arti- about the validity of the schizophrenia construct, culated in the genetics of complex disorders, have not some leading to proposals to discard the category7,8 or been adequately considered in schizophrenia research. Correspondence: Dr A Jablensky, Medical Research Foundation Bldg, Level 3, The University of Western Australia, 50 Murray Arguments for and against heterogeneity in the Street, Perth, WA 6000, Australia. aetiology of schizophrenia E-mail: [email protected] Received 8 May 2006; accepted 23 May 2006; published online The debate on heterogeneity and the likely existence 27 June 2006 of aetiologically distinct subtypes has waxed and Subtyping schizophrenia A Jablensky 816 waned in the schizophrenia literature without ever the proportion of families linked to a given locus reaching a closure. Over 10 years ago, the editors of using a Bayesian model25 or likelihood-ratio tests.26 Schizophrenia Research invited researchers to state The approaches to resolving such heterogeneity reasons for and against schizophrenia being a homo- include segregation analysis and search for subclini- geneous single disease.13 In summary, the arguments cal markers that may index aetiological subtypes. were as follows. Retinitis pigmentosa is an apt example, where the identification of mutations in over 30 genes has A. The case for homogeneity resulted in a genetic classification of the retinal degeneration syndromes.27 Extensive heterogeneity There are no disease entities in psychiatry, only 14 is the rule in many common complex disorders, continua of variation. Even when the aetiology of including inflammatory bowel disease,28 rheumatoid a disorder is known and is unitary, the presentation arthritis,29 and osteoporosis.30 The obverse phenom- and outcome may be remarkably varied. enon of pleiotropy, that is, multiple phenotypes Phenotypic variation in schizophrenia reflects arising from one genetic factor, is equally common, a continuum of severity in which patients vary 15 for example, in demyelinating peripheral neuropathy along cognitive and neurobiological dimensions. or axonal neuropathy with vocal cord paresis caused Phenotypic variation is compatible with aetiologi- by mutations in the same GDAP1 gene.31,32 A range of cal homogeneity resulting from a liability-threshold 16 dissimilar syndromes may be varying expressions of a process. single genetic defect, such as severe neonatal intest- B. The case for heterogeneity inal obstruction, bronchiectatic lung disease, idio- The candidate gene findings reject the parsimo- pathic pancreatitis, and male infertility, caused by 33–35 nious hypothesis that all schizophrenia is caused mutations in the CFTR gene. Both heterogeneity by the same pattern of genetic mutations, birth and pleiotropy are implicated in the phenotypic 36 complications and viral infections. variation of common brain disorders, where their There is already evidence for several subtypes of effects may be aggravated by unknown or poorly schizophrenia associated with specific chromo- understood environmental contributions to the 37 some abnormalities. It may be worthwhile to phenotype. subdivide schizophrenic patients on the presence Schizophrenia cannot be an exception from these or absence of a putative aetiological factor, as in laws. Although heterogeneity is generally acknow- other complex diseases, such as diabetes. ledged and genetic linkage analysis often performed The question ‘heterogeneity: yes or no?’ should be under the assumption of heterogeneity, this is usually reworded to ‘heterogeneity: how much?’17 performed post hoc, that is, after the data have been collected, or by default, when difficult to interpret In a review of the meta-analytical evidence, results have been obtained.38 Apart from locus and Heinrichs18 reformulated the ‘continuum of severity’ allelic heterogeneity, commonly suspected sources of argument in terms of a ‘pathological shift’, which ‘nuisance’ variance in schizophrenia include a poorly holds that a single aetiological factor may, in each understood, poly- or oligogenic transmission; incom- individual case, affect differentially various dimen- plete penetrance; variable phenotype expression; sions of psychopathology and biology, with a result- unknown environmental contribution; phenocopies; ing net effect mimicking heterogeneity. He argued that misspecification of the genetic model; and measure- such a model would be difficult to refute by direct ment or classification error.39,40 Less often acknow- evidence. The contrasting hypothesis of biological ledged, but potentially critical sources are a fallible subtypes offers an interpretation of the evidence that phenotype;41 existence of latent disease subtypes is refutable. The demonstrated success in resolving that may be aetiologically different;42 and popula- heterogeneity by phenotype subtyping in a number of tion admixture (of subtypes and ethnic variation) complex disorders, such as noninsulin-dependent that could seriously compromise the power of the diabetes;19 asthma;20 Parkinson’s disease (PD);21 available analytic methods.43,44 autism;22 and familial Alzheimer disease (AD)23 The likely existence of aetiologically different provides prima facie support to a ‘splitting’ approach subtypes of the disorder (Bleuler’s notion of a ‘group in schizophrenia genetics. of schizophrenias’45), is rarely considered in genetic studies, which tend to be predicated on the broad clinical diagnosis as the