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 phenotype, implicitly assum- How heterogeneous is schizophrenia? ing a unitary view of the disorder. Phenotype Genetic heterogeneity and pleiotropy refinement through disaggregation into clinical sub- Genetic heterogeneity – ‘the existence of two or more types, or extension by covariate quantitative traits, genetically distinct entities with essentially one and has been a successful strategy in the genetic dissec- the same phenotype’24 – is a common attribute of tion of asthma,20 type I diabetes,19 or dementia.23 This human disease, characterizing not only the geneti- approach has had limited following in schizophrenia cally complex disorders, but inherently present in research. The failure to address the unresolved monogenic mendelian conditions as well. In the heterogeneity in schizophrenia continues to be a latter, locus heterogeneity is usually estimated as serious obstacle to the effective harnessing of novel

Molecular Psychiatry Subtyping schizophrenia A Jablensky 817 technologies, such as whole-genome association broad clinical definition of schizophrenia to carve out analysis, or joint association and expression studies, biologically homogeneous clinical populations for into an effort to ‘deconstruct’ schizophrenia. genetic analysis. Such concerns are not new – attempts to parse the Symptoms and course as criteria defining the complexity of schizophrenia into simpler component phenotype disorders or subtypes have been undertaken since the As its inception and to the present day, the clinical earliest formulation of the diagnostic concept, using entity of schizophrenia is diagnosed by analysis of the clinical or biological criteria, as well as a variety of subjective symptoms reported by patients, their statistical methods. In the overview, that follows, the history and course, observation of behaviour, affect terms ‘subtype’, ‘component disorder’ or ‘variant’ are and speech, and (to a lesser extent) by evaluation used near-synonymously to denote sets of phenotypic of premorbid development, personality traits, and attributes defining discrete subgroups of individuals family background. The diagnostic criteria of the likely to be internally more homogeneous for aetiolo- International Classification of Diseases ICD-102 and gically relevant genes than the whole of the clinical Diagnostic and Statistical Manual (DSM)-IV3 were population meeting broad diagnostic criteria for originally conceived with a view to achieving three schizophrenia. fundamentally different goals: (i) to identify groups of patients with broadly similar clinical presentation and prognosis; (ii) to facilitate early diagnosis and Putative schizophrenia subtypes based choice of treatment; and (iii) to define a homogeneous on clinical features heritable diagnostic category for genetic aetiological Kraepelin’s clinical forms of dementia praecox research.46 Whereas the first two goals have, by and Acknowledging the diversity of the clinical pictures large, been achieved as regards the clinical utility of subsumed under dementia praecox and the absence the criteria, attainment of the third goal remains of pathognomonic symptoms, Kraepelin1 articulated, remote. The estimated high heritability of schizo- ‘for the sake of a more lucid presentation’, nine phrenia (B80%, on the basis of twin studies using different ‘clinical forms’ (Table 1). However, he DSM-III criteria47) has not been matched by unambig- emphasized that ‘we everywhere meet the same uous genetic linkage findings in studies using the fundamental disorders in the different forms of diagnostic category as the phenotype. Two recent dementia praecox, in very varied conjunctions, even meta-analyses48,49 suggested greater consistency of though the clinical picture may appear at first sight linkage results when pooled samples from different ever so divergent’. The ‘fundamental disorders’, sources were examined, but the actual agreement holding together the disease entity, were cognitive between the two studies on specific linkage findings deficit (a ‘general decay of mental efficiency’) and was disappointingly low. The inconsistency of the executive dysfunction (‘loss of mastery over volitional results suggests extensive locus and allelic hetero- action’), most clearly manifested in the residual, geneity, as well as an admixture of phenotypically ‘terminal states’ of the illness. Kraepelin was reluc- varied clinical populations. tant to impute aetiological significance to the clinical Notwithstanding the reasonable level of inter-rater variants he described, and regarded the issue of a agreement that can be achieved on the broad unitary process versus multiple disease states within diagnosis, the symptoms of schizophrenia span a schizophrenia ‘an open question’. The renewal of wide range of psychopathology and display an interest in Kraepelin’s dementia praecox since the extraordinary amount of interindividual variability 1990’s has led researchers to attempt delineating a and temporal inconstancy, calling to mind Wittgen- ‘Kraepelinian’ subtype of schizophrenia, in terms of stein’s remark that classifying subjective psychologi- negative or disorganized symptoms, poor outcome, cal phenomena was akin to classifying clouds by their neuropsychological deficits, and risk factors.54–56 shape.50 As no symptom is pathognomonic or neces- However, as indicated in Table 1, Kraepelin’s original sary, but variable subsets of symptoms can be typology allowed for much greater heterogeneity in sufficient for the diagnosis, patients may be allocated the clinical manifestations of dementia praecox than to the diagnostic category of schizophrenia without it is currently assumed. having a single symptom in common. As a conse- quence, the phenomenological similarity of patients, Bleuler’s ‘group of schizophrenias’ selected for genetic and other biological research by Having coined the term ‘schizophrenia’ to replace the current criteria is modest at best, and disconcert- dementia praecox, Bleuler45 stated that schizophrenia ingly low at worst. This might be part of the reason for ‘is not a disease in the strict sense, but appears to be a the limited capacity of the diagnosis to predict group of diseases.Therefore, we should speak of accurately which biological or behavioural attributes schizophrenias in the plural’. He acknowledged that will be shared by the majority of individuals allocated the clinical subgroups of paranoid schizophrenia, to the diagnostic category, or to draw ‘zones of catatonia, hebephrenia and simple schizophrenia – rarity’51 that clearly demarcate schizophrenia from retained in the present DSM and ICD classifications – other disorders, such as bipolar affective disorder.52,53 were not ‘natural’ nosological entities. What were Such flaws raise doubts about the capacity of the then the multiple ‘schizophrenias’? Bleuler argued

Molecular Psychiatry Subtyping schizophrenia A Jablensky 818 Table 1 Emil Kraepelin’s ‘clinical forms’1

K Dementia praecox simplex (‘Impoverishment and devastation of the whole psychic life which is accomplished quite imperceptibly’) K Hebephrenia (Insidious change of personality with shallow capricious affect, senseless, and incoherent behaviour, poverty of thought, occasional hallucinations, and fragmentary delusions, progressing to profound dementia) K Depressive dementia praecox (simple and delusional form) (Initial state of depression followed by slowly progressive cognitive decline and avolition, with or without hypochondriacal or persecutory delusions) K Circular dementia praecox (Prodromal depression followed by gradual onset of auditory hallucinations, delusions, marked fluctuations of mood, and aimless impulsivity) K Agitated dementia praecox (Acute onset, perplexity, or exaltation, multimodal hallucinations, fantastic delusions) K Periodic dementia praecox (Recurrent acute, brief episodes of confused excitement with remissions) K Catatonia (‘Conjunction of peculiar excitement with catatonic stupor dominates the clinical picture’ in this form, but catatonic phenomena frequently occur in otherwise wholly different presentations of dementia praecox) K Paranoid dementia (mild and severe form) (The essential symptoms are delusions and hallucinations. The severe form results in a ‘peculiar disintegration of psychic life’, involving especially emotional and volitional disorders. The mild form is a very slowly evolving ‘paranoid or hallucinatory weak-mindedness’ which ‘makes it possible for the patient for a long time still to live as an apparently healthy individual’) K Schizophasia (confusional speech dementia praecox) (Cases meeting the general description of dementia praecox but resulting in an end state of ‘an unusually striking disorder of expression in speech, with relatively little impairment of the remaining psychic activities’)

that ‘the disease schizophrenia must be a much WHO cross-national studies, defined a ‘nuclear’ broader concept than the overt psychosis of the same schizophrenia (S þ ) characterized by presence of at name’. Along with the ‘latent’ schizophrenias, which least three out of six FRS. Familiality and modest to manifested mainly aberrant personality traits, he substantial heritability has been reported for the listed atypical depressive or manic states, Wernicke’s FRS,65,66 but a study comparing their occurrence in motility psychoses, reactive psychoses, and other affected subjects with and without linkage to chromo- nonorganic, nonaffective psychotic disorders as be- somes 5q, 6p, 8p, and 10p produced ambiguous longing to the broad group of schizophrenias, suggest- results.67 ing that ‘this is important for the studies of heredity’, thus foreshadowing the notion of schizophrenia Leonhard’s alternative classification of the spectrum disorders. ‘endogenous’ psychoses In a clinical tradition striving to group psychotic Post-Kraepelinian and post-Bleulerian subtypes and illnesses on the basis of presumed localized cerebral dichotomies dysfunction, Leonhard68 developed a classification of Further subnosological distinctions have been pro- the ‘endogenous’ psychoses which departed substan- posed, including schizoaffective disorder;57 schizo- tially from the Kraepelinian nosology. Leonhard phreniform psychoses;58 process – nonprocess;59 and defined sharply delineated disease entities, described paranoid – nonparanoid schizophrenia.60 Schneider61 by a detailed psychopathology emphasizing objective claimed that nine groups of psychotic symptoms, signs (e.g. psychomotor behaviour), course and out- designated as ‘first-rank symptoms’ (FRS), had a come, and family history. The nonaffective psychoses ‘decisive weight’ in the diagnosis of schizophrenia: were split into two groups of schizophrenias – audible thoughts; voices arguing about, or discussing ‘systematic’ and ‘unsystematic’ – and a third group the patient; voices commenting on the patient’s of ‘cycloid’ psychoses, each containing further sub- actions; experiences of influences on the body; types (Table 2), for which clinical homogeneity and thought withdrawal and other interference with distinct disease status were claimed. The resulting thought; thought broadcast (diffusion of thought); categorical taxonomy is the antithesis of the notion of delusional perception; and other experiences involv- a continuum. While the ‘unsystematic’ schizophre- ing ‘made’ impulses and feelings experienced as nias are considered to be primarily genetic, hereditary caused by an outside agency. Owing to the sharpness factors play a secondary role in the cycloid psychoses of their definition and the hope that they could be and in the ‘systematic’ schizophrenias, presumed to reliably ascertained, the FRS have been incorporated be exogenously determined by maternal obstetric in the Research Diagnostic Criteria, RDC;62 DSM III;63 complications or early failure of social learning. and ICD-10.2 The Catego algorithm,64 used in the Notably, Leonhard’s classification neither expands,

Molecular Psychiatry Subtyping schizophrenia A Jablensky 819 Table 2 Karl Leonhard’s classification of the nonaffective endogenous psychoses68

I. Group of systematic schizophrenias (Insidious onset, auditory and somatic hallucinations, delusions, early blunting of affect, continuous unremitting course, personality deterioration)

Paraphrenias (Auditory hallucinosis, audible thoughts, thought broadcast, passivity experiences, delusional misidentifications, falsifications of memory) Hebephrenias (Extreme autistic withdrawal, flat affect, impoverished or disorganized speech and behaviour) Catatonias (Excessive parakinesias, mannerisms, verbigeration, posturing, stereotypies, mutism, auditory hallucinations)

II. Group of unsystematic (atypical) schizophrenias (Rapid onset, relatively preserved affect, remitting course, mild personality deterioration)

Affect-laden paraphrenia (Paranoid delusions with affective loading) Cataphasia (schizophasia) (Incoherent, pressured speech but well-organized behaviour) Periodic catatonia (Episodic hyper- or hypokinesia, mixed excitatory and hallucinatory symptoms)

III. Group of cycloid psychoses (Sudden onset, pervasive delusional mood, multimodal hallucinations, labile affect, polarity of manifestations, typically complete recovery from episode)

Anxiety-happiness psychosis (Extreme shifts of affect, polarity e.g. intense fear – ecstatic elation) Motility psychosis (Impulsive hypermotility – psychomotor inhibition) Confusion psychosis (Incoherent pressure of speech – mutism)

nor constricts the outer boundaries of schizophrenia but not confirmed in an independent sample.77,78 as defined by ICD-10,2 and DSM-IIIR,69 but carves the Although the jury is still out, the apparent clinical schizophrenia spectrum in a different way. homogeneity of the syndrome of periodic catatonia Leonhard’s family studies, although richly descrip- may not be underpinned by genetic homogeneity. tive, may be methodologically outdated by present- day criteria. However, his conjectures have found The notion of a schizophrenia spectrum support in a twin study,70 in which the highest MZ/ The concept of a spectrum of schizophrenia-related DZ concordance rate differential was obtained with phenotypes originates in the observation that several ‘unsystematic’ schizophrenia (88.9% MZ, 25.0% DZ) ostensibly different disorders tend to cluster among and the lowest with cycloid psychoses (38.5% MZ, biological relatives of individuals with clinical schi- 36.4% DZ), consistent with the prediction about the zophrenia.45,79,80 Epidemiological and family study role of genetic factors in the aetiology of the different data suggest that the genetic liability to schizophrenia psychoses. A genome scan of 12 German multiplex is shared with liability to other related syndromes.81,82 pedigrees with periodic catatonia (one of the ‘un- The term ‘schizotypy’, introduced by Rado83 and systematic’ schizophrenias with a 26.9% recurrence Meehl,84 describes a personality characterized by risk71), revealed significant linkage on chromosome anhedonia, ambivalence, ‘interpersonal aversiveness’, 15q15, and suggestive evidence on 22q13.72,73 The body image distortion, ‘cognitive slippage’, and follow-up of the 15q finding has so far excluded the sensory, kinaesthetic or vestibular aberrations. nicotinic acetylcholine receptor alpha7 subunit gene, Chapman et al.85 designed scales to measure percep- the zinc transporter SLC30A4, and the NOTCH4 tual aberrations and ‘magical ideation’ as traits gene.74–76 The putative 22q13 locus contains the predicting ‘psychosis proneness’. These constructs MLC1 (WKL1) gene, coding for a cation channel were amalgamated with clinical descriptions from the expressed exclusively in the brain and causing, in its Danish-US adoptive study into the DSM-III diagnostic mutated form, a severe neurodegenerative disorder category of schizotypal personality disorder (SPD), (megaencephalic leukoencephalopathy). A rare mis- which is now central to the spectrum notion.8186 The sense mutation in this gene, cosegregating with frequent occurrence of SPD among first-degree rela- periodic catatonia in a single pedigree was detected, tives of probands with schizophrenia has been

Molecular Psychiatry Subtyping schizophrenia A Jablensky 820 replicated in the Roscommon epidemiological basis.103 Its construct validity and neurobiological study,87 which added to the schizophrenia spectrum correlates remain ambiguous.104–106 As no theory- further disorders cosegregating within families. The based rule exists for classifying the symptoms of resulting ‘continuum of liability’ included: (i) ‘typi- schizophrenia as negative or positive, the distinction cal’ schizophrenia (ii) schizotypal and paranoid is supported by their differential loadings on separate personality disorders; (iii) schizoaffective disorder, factors. The argument is somewhat circular, since depressed type; (iv) other nonaffective psychotic cross-sectional symptomatology is typically assessed disorders (schizophreniform, atypical psychosis); by rating scales designed a priori to reflect such and (v) psychotic affective disorders. The correlation distinction.107 of liability to the five disorders between probands with schizophrenia and their first-degree relatives 87 Deficit–nondeficit schizophrenia was 0.36. Recent evidence from the Finnish family 108,109 adoptive study of schizophrenia suggests more Carpenter and co-workers proposed the delinea- restrictive genetic boundaries of the spectrum, by tion of a subtype of schizophrenia characterized by enduring ‘primary’ negative symptoms that could not excluding paranoid personality disorder and psycho- tic affective disorders.88 In all its variations, however, be construed as sequelae of other psychopathology the spectrum concept remains critically dependent on (Table 3). This clinical construct, evocative of the validity of the SPD concept. Accumulating Kraepelin’s dementia praecox, was termed ‘deficit evidence from family and twin data indicates that schizophrenia’ (DS) and hypothesized to be an aetiologically distinct ‘disease’ within the schizo- SPD is multidimensional and may be genetically 110 heterogeneous.89–92 Its manifestations fall into two phrenia spectrum. Studies comparing DS cases with ‘nondeficit’ (NDS) patients and controls, esti- genetically independent clusters: a ‘negative’ cluster (odd speech and behaviour, inappropriate affect and mate the prevalence of the DS subtype at 16.5% in social withdrawal), more common among relatives of unselected epidemiological samples of schizophrenia cases111 and 25–30% within samples of chronic schizophrenic probands, and a ‘positive’ cluster 110 (magical ideation, brief quasipsychotic episodes), schizophrenia. Compared to NDS, DS cases exhibit associated with increased incidence of affective less paranoid ideation and depression, less substance abuse, more prominent anhedonia, poor social func- disorders in relatives.93 ‘Negative’ schizotypy may tioning, treatment resistance and a higher schizo- indeed represent a personality-based counterpart of 111 schizophrenia,94 manifesting attenuated cognitive phrenia risk in relatives. DS and NDS do not differ deficits95–97 and brain structural abnormalities98 on age at onset and length of illness, which argues characteristic of schizophrenia. against a progression leading from NDS to DS. Supportive evidence for the DS construct has been Positive–negative schizophrenia (‘Type I’ and ‘Type II’) A general ‘weakening’ of mental processes resulting Table 3 Diagnostic criteria for the deficit syndrome of in a ‘defect’ was the cornerstone of Kraepelin’s 108,109 dementia praecox, who suggested that precursors of schizophrenia ‘defect’ could be detected early in the illness, 1. At least 2 of the following 6 negative symptoms must be coexisting with ‘productive’ or ‘florid’ symptoms. present: Since the 1970s, the terms ‘defect’ and ‘productive’ symptoms have been virtually replaced by ‘negative’ Restricted affect and ‘positive’ symptoms.99 Crow100 proposed a simple Diminished emotional range subclassification of schizophrenia, based on the Poverty of speech predominance of either positive or negative sympto- Curbing of interests matology. ‘Type I’ (positive) schizophrenia was Diminished sense of purpose characterized by hallucinations, delusions, and for- Diminished social drive mal thought disorder, with a presumed underlying 2. Some combination of two or more of the above negative dopaminergic dysfunction, while patients with ‘Type symptoms have been present for the preceding 12 months II’ (negative) schizophrenia displayed social with- and always present during periods of clinical stability. drawal, loss of volition, affective flattening, and poverty of speech, presumed to be associated with 3. The negative symptoms are primary, that is, not structural brain abnormalities. Criteria and rating secondary to factors other than the disease process, for scales for positive (SAPS) and negative (SANS) example schizophrenia were proposed by Andreasen and Olsen.101 The initial typology, implying pathogeneti- Anxiety cally discrete and mutually exclusive ‘types’, was Drug effects later replaced by a negative and a positive dimension, Suspiciousness or other psychotic symptoms Mental retardation allowing the two kinds of symptoms to co-occur in Depression the same individual and share a common aetiology.102 In essence, the positive–negative typology is 4. The patient meets DSM-III criteria for schizophrenia a descriptive device without a strong theoretical

Molecular Psychiatry Subtyping schizophrenia A Jablensky 821 provided by neuropsychological studies, which schizophrenia should be viewed with caution, con- found those patients to be particularly impaired on sidering the diversity of clinical populations and the the Stroop colour-word interference test,112 the limitations of the instruments used to generate the degraded stimulus version of the continuous per- input data. formance task,113 and tests of general ability.114 Whereas factor analysis groups variables, cluster Oculomotor control measures indicate deficient analysis groups individuals on the basis of maximum tracking and antisaccade performance115–117 and shared characteristics. Farmer et al.143 identified neurological examination suggests deficient sensory two clusters into which patients with schizophrenia integration.118 The overall pattern has been inter- could be fitted, based on their scores on a checklist preted as indicative of a fronto-temporo-parietal of 20 symptom and history items: one characterized dysfunction, against a background of a more global by good premorbid adjustment, later onset, and well impairment.119 Taxometric analysis of 238 schizo- organized delusions, and another including early phrenia patients120 suggests a discrete taxon status for onset, poor premorbid functioning, incoherent DS. The DS–NDS typology with its covariates is well speech, bizarre behaviour, and family history of replicated, yet rarely used in genetic research121,122 schizophrenia. However, using the Positive and despite its potential suitability as a phenotype. Negative Syndrome Scale,132 Dollfus et al.144 obtained four quite different distinct clusters, corresponding to Statistically derived symptom dimensions or clusters positive, negative, disorganized and mixed sympto- Factor analysis has been applied to psychiatric rating matology. Thus, cluster analysis is as dependent on scales since the 1960s.123 Essentially, factor analysis the selection of input variables as factor analysis. and related methods reduce the covariation of the Latent class analysis (LCA) assumes the existence primary data matrix to covariances of small numbers of a finite number of mutually exclusive and jointly of latent factors which account for the interrelation- exhaustive groups of individuals, within which ships among the primary variables and explain a person characteristics, for example, responses to proportion of their variance. Based on a relatively symptom items, are: (a) determined by class member- small number of input variables (SANS/SAPS scores), ship and (b) locally independent.145 A latent class a three-factor structure was proposed by Liddle124 and typology of schizophrenia, proposed by Sham replicated by other investigators.125–127 In this model, et al.,146 using data on 447 patients with nonaffective negative symptoms load on a single factor of ‘psy- psychoses, suggested the existence of three sub- chomotor poverty’, while positive symptoms split groups: a ‘neurodevelopmental’ subtype resembling into a delusions-and-hallucinations factor (‘reality the hebephrenic form of the disorder (poor premorbid distortion’) and a thought-and-speech disorder factor adjustment, early onset, prominent negative, and (‘disorganization’). The model has been shown to be disorganized features); a ‘paranoid’ subtype (less longitudinally stable128 and replicable in non-Eur- severe, better outcome); and a ‘schizoaffective’ sub- opean populations.129,130 It was incorporated in DSM- type (dysphoric symptoms). In an epidemiological IV3 and tested in field trials.131 sample of 343 probands with schizophrenia and Results of factor-analysis of symptomatology de- affective disorders, Kendler et al.147 found six latent pend strongly on the content of the clinical rating classes, broadly corresponding to the nosological scales used as input. Studies using the SANS and groups of ‘Kraepelinian’ schizophrenia; major depres- SAPS result in different solutions from those pro- sion, schizophreniform disorder; schizoaffective dis- duced by the PANSS,132 BPRS,133 or OPCRIT,134 order (manic), schizoaffective disorder (depressed), including a general neurotic syndrome factor;135 and hebephrenia. Increased risk for schizophrenia excitement and depression;136,137 paranoid, first-rank spectrum disorders was found among the relatives of delusions and first-rank hallucinations;138 premorbid subjects assigned to the schizophrenia and schizo- adjustment deficits factor;139 and autistic preoccupa- phreniform classes, while increased risk for affective tion factor.140 In a large sample of probands with disorders was only found in the relatives of patients schizophrenia, McGrath et al.141 identified five factors assigned to the major depression and schizoaffective (positive, negative, disorganized, affective, and early (depressed) classes. Similar results, using a combina- onset/developmental), which were associated with tion of principal component (factor) analysis and risks of psychoses and affective disorders in relatives. LCA in an epidemiologically ascertained sample of In a series of factor analyses based on an expanded 387 patients with psychoses, have been reported by list of 64 psychopathological symptoms, Cuesta and Murray et al.148 Peralta142 concluded that a hierarchical 10-dimen- In contrast to conventional LCA, a special form of sional model provided the best fit on statistical and latent structure analysis, the grade of membership clinical grounds. Factor solutions, therefore, are not (GoM) model, allows individuals to be members of unique and the question ‘how many factors parsimo- more than one class and represents the latent groups niously describe the symptomatology of schizophre- as ‘fuzzy sets’,149,150 where individuals can be mem- nia?’ can only be answered in the context of a bers of more than one set. The GoM model simulta- particular selection of symptoms and measure- neously extracts from the data matrix a number of ment methods. Therefore, factor-analytical studies latent ‘pure types’ and assigns each individual a set of suggesting ‘established’ dimensions or syndromes of numerical weights quantifying the degree to which

Molecular Psychiatry Subtyping schizophrenia A Jablensky 822 that individual resembles each one of the identified schizophrenia families.39 Moreover, simulation power pure types. When applied to the symptom profiles of analyses174 suggest that the F/S design can be useful 1065 cases in the WHO International Pilot Study of only in the context of very large samples of nuclear Schizophrenia,151 the method identified eight pure families. types of which five were related to schizophrenia, two to affective disorders and one to patients in remission, Subtyping based on genetic linkage or association data all showing significant associations with course and Several studies based on samples with well-charac- outcome variables used as external validators. terized clinical phenotypes have examined associa- tions between selected clinical features and previously established genetic linkage regions or Subtypes based on putative genetic indicators putative candidate genes. Various approaches have Familial–sporadic schizophrenia been employed to interrogate genome-wide linkage or Subtyping schizophrenia by the presence/absence of association data with a view to exploring pathways of a positive family history for schizophrenia spectrum disease expression. Thus, a potential 6p locus, disorders was proposed in the 1980s as a strategy associated with a quantitative trait assessing the expected to be more successful in resolving hetero- severity of psychotic symptoms, was reported by geneity than symptom-based typologies.152 Although Brzustowicz et al.175 Pulver et al.176 used diagnostic inclusion/exclusion criteria varied across studies, information to stratify 54 multiplex pedigrees by ‘familial’ (F) cases were typically defined as having diagnostic phenotypes cosegregating in nonschizo- X1 first-degree relative affected, while ‘sporadic’ (S) phrenic first-degree relatives of the probands and cases had none among either first- or second-degree reported genome-wide significant linkage to 8p21 and relatives.153 The F/S dichotomy rests on the assump- suggestive linkage to 1p21 for schizophrenia spec- tion that the familial aggregation of cases is primarily trum personality disorders. Subsequent analyses of of a genetic origin, while sporadic cases result from affected siblings from these families revealed that the environmental insults (e.g. maternal obstetric com- linkage evidence for 8p21 was mainly contributed by plications) or de novo somatic mutations. In the a subgroup of 30 affected siblings sharing two alleles majority of studies applying this classification, the identity by descent in the 8p21 region and one allele proportion of familial cases was in the range of at a locus on chromosome 14, suggesting an interac- 8–15%, that is, lower than the 19% prevalence tion effect. Phenotypically, this subgroup was char- estimated by Gottesman et al.154 from pooled Eur- acterized by a high prevalence of bizarre delusions, opean studies between 1920 and 1978. As the F/S affective symptoms early in the course of illness, subtypes were hypothesized to differ aetiologically, a history of seizures, and attendance of special number of studies, mostly of small to moderate school.177 sample size ( < 100), compared the phenotypic char- Using data from the Irish Study of High-Density acteristics of the two groups.155,156 No consistent and Schizophrenia Families, Kendler et al.67 reported significant differences have been found in age at high levels of positive thought disorder, affective onset, symptom patterns, severity, treatment response deterioration, and worse outcome in probands from and outcome,157–159 and the findings with regard to families with evidence of linkage to 8p22–21, suggest- obstetric complications are inconclusive.160–162 Spora- ing that a susceptibility gene in the region may be dic cases are more likely to be winter-born159,163,164 predisposing to a Kraepelinian, dementia praecox and have more electroencephalographic (EEG) type of illness. Within the same cohort of families, abnormalities153 and enlarged ventricles on CT scan family-based transmission disequilibrium tests pro- or MRI.152,165,166 Familial cases, on the other hand, duced suggestive evidence of association between have more neurological signs,164,167–169 poorer sus- affective symptoms and the His452Tyr polymorphism tained attention performance;170 cortical abnormal- in the serotonin 2A receptor; between negative ities on MRI171 and reduced temporoparietal resting symptoms and the brain-derived neurotrophic factor regional blood flow.172 (BDNF); and between negative symptoms and a high- By and large, the F/S classification has not been risk haplotype in the dystrobrevin-binding protein 1 successful in identifying homogeneous phenotype (dysbindin, DTNBP1) on 6p24–22.178,179 Recently, an groups for genetic research. The dichotomy, based on association between a six-locus haplotype in DTNBP1 recurrence of manifest schizophrenia among biologi- and psychometrically assessed generalized cognitive cal relatives, might easily result in a misclassification deficit in schizophrenia patients was reported by unless: (i) ascertainment of all family members is Burdick et al.180 complete; (ii) appropriate adjustments for family size, A special focus in the search for genetic subtypes of age, and lifetimes at risk are made; and (iii) the schizophrenia has been the B3 Mb region on chromo- spectrum of disorders counted as recurrence cases is some 22q11, which contains at least three genes stringently defined. However, even if such confound- implicated in schizophrenia (COMT, PRODH2, and ing factors are adequately controlled for, and ‘famili- ZDHHC8) and is hemizygously deleted in the velo- ality’ is represented as a continuous trait rather than a cardiofacial syndrome (VCFS, DiGeorge syndrome, or dichotomy,173 the method remains open to error due Shprintzen syndrome). The microdeletion has a to the likely presence of unexpressed genotypes in population frequency of B1 in 6000 births181 and is

Molecular Psychiatry Subtyping schizophrenia A Jablensky 823 associated with increased risk for several neuropsy- Table 4 ‘Candidate’ endophenotype markers in schizophre- chiatric syndromes, including schizophrenia,182 bipo- nia research lar disorder, learning disability, and ADHD.183 Neurophysiological markers and endophenotypes Approximately, 1% of adult schizophrenia patients 205 are carriers of the microdeletion184 but the frequency Electrodermal deviance Prepulse inhibition of the startle reflex206–208 may be as high as 5% among individuals with 209–211 185 Deficient gating of the auditory evoked response (P50) childhood onset of schizophrenia. No evidence P300 amplitude reduction and latency delay212 has been produced to date that such patients express N400 amplitude reduction (semantic context a schizophrenia phenotype that is clinically distin- underutilization)213 guishable from schizophrenia in the absence of Mismatch negativity (MMN)214–217 VCFS.186 However, schizophrenia patients with VCSF Smooth pursuit eye movement dysfunction (SPEM)218–221 have more severe cognitive deficits of spatial working Antisaccade error rate (AS)222–224 225 memory, visual recognition, and attention than VCFS Composite inhibitory phenotype (P50, AS, SPEM) 187 Multivariate electrophysiological endophenotype (MMN, individuals without schizophrenia. Compared to 226 matched normal controls, schizophrenia patients P50, P300, AS) with VCSF have smaller total grey matter volume, 188 Neuroimaging markers and endophenotypes larger lateral ventricles, and decreased gyrification Fronto-thalamic-cerebellar gray matter deficit227 189 in the frontal and parietal lobes. Considering the Fronto-striato-thalamic gray matter deficit228 potential to trace causal pathways linking specific MRI-derived three-factor phenotype229 gene effects with intermediate phenotypes and clin- MRI whole-brain nonlinear pattern classification230 ical disease expression, studies of rare genetic defects Frontal hypoactivation in response to cognitive tasks have a special place in schizophrenia research. The (hypofrontality)231 discovery of a translocation break point cosegregating Atrophic and static (neurodevelopmental) schizophrenia 232 with schizophrenia in a Scottish pedigree190 has endophenotypes facilitated the recent identification of Disrupted- Cognitive markers and endophenotypes In-Schizophrenia 1 (DISC1) as a positional candidate Continuous performance tests (CPT, signal/noise ratio)233–235 gene with likely effects on brain development and Attention and vigilance-based subtype18 191,192 neurocognition in schizophrenia. Verbal dysmnesic subtype18 Verbal memory deficit, cortical or subcortical type236,237 18 Endophenotypes: signposts to a biological Dysexecutive subtype Prefrontal executive/working memory phenotype238 subclassification of schizophrenia? Frontal/abstraction deficit profile239 Spatial working memory240 Amidst growing doubts in the capacity of the broad 239,241–243 diagnostic category to serve as a reliable phenotype Generalized (diffuse, pervasive) cognitive deficit, CD for gene discovery,193–196 the concept of endopheno- Other markers and endophenotypes type (intermediate, elementary, alternative, or corre- Neurological soft signs244–247 lated phenotype) offered a novel perspective on Composite laterality phenotype248 subtyping schizophrenia that could be either an Nailfold plexus visibility249 alternative or a complement to symptom-based Minor physical anomalies250,251 phenotypes. The term, originating in early 20th century plant and insect genetics, was introduced into schizophrenia genetics by Gottesman and haemochromatosis,201 and juvenile myoclonic Shields.197 As ‘measurable components unseen by epilepsy.202 In schizophrenia research, an endophe- the unaided eye along the pathway between disease notype approach, based on smooth pursuit eye move- and distal genotype’,198 endophenotypes must meet ments (SPEM) was first explored by Holzman203,204 criteria of being: (i) associated with the clinical and followed by a growing number of studies utilizing disorder but not necessarily part of its diagnosis; (ii) psychophysiological, brain imaging, and cognitive heritable; (iii) state-independent (i.e. present before measures (Table 4). the onset of active illness or during remissions); (iv) cosegregating with illness in families; and (v) found Cognitive dysfunction as an endophenotype in unaffected family members at a higher rate than in Cognitive deficits are widely regarded as a core the general population.193,198 Earlier desiderata, for feature of schizophrenia and not an epiphenomenon example that endophenotypes have a mendelian of the illness.251-253 There is remarkable agreement in genetic architecture, may be unrealistic. An important the literature that deficits in multiple cognitive requirement, however, is that an endophenotype domains predate the onset of clinical symptoms;254–258 should be a quantitatively measurable trait (on a rank are not attributable to antipsychotic medications;259 scale as a minimum, but preferably on an interval persist over the course of the illness; are unrelated scale). Examples of endophenotypes meeting the to its duration;260–262 and behave like a stable above criteria and successfully used in gene identi- trait.233,263,264 Pervasive cognitive dysfunction has fication include the long QT syndrome,199 familial been reported in > 50% of schizophrenia patients in adenomatous intestinal polyposis,200 idiopathic a community-based survey in Scotland,265 and there

Molecular Psychiatry Subtyping schizophrenia A Jablensky 824 is compelling evidence that cognitive deficits are distribute patients into groups of severely compro- significantly correlated with impairments in activi- mised, intermediate and mildly affected perfor- ties of daily living (ADL) in patients with schizo- mance,291,292 ‘classical’ fine-grain neuropsychological phrenia,266–270 but only weakly associated with analyses (case studies of individual profiles rather psychotic symptoms.271 Patients with paranoid than group means; delineation of generalized/differ- schizophrenia and pronounced positive symptoms ential deficits; search for ‘double dissociations’) have tend to show better cognitive functioning compared identified patterns of dysfunction that parallel the to patients with undifferentiated or disorganized amnestic syndromes in coarse brain disease, such as schizophrenia.272–274 Huntington’s (HD), PD, or AD. Population-based, longitudinal cohort stu- In a series of 175 schizophrenia patients, compared dies254,275,276 have found that compromised general with 229 normal controls on the performance of the cognitive ability in late adolescence is a strong California Verbal Learning Test (CVLT), Paulsen predictor of subsequent schizophrenia risk. Family et al.236 elicited from 50% of the patients a subcortical studies indicate that a proportion of the unaffected (striatal, HD/PD-type) memory profile combining first-degree relatives of index cases of schizophrenia prominent retrieval deficit (poor free-format recall of display similar patterns of deficit in an attenuated word lists, improving substantially on presentation of form,234,235,277–279 and an epidemiologically-based cues) with absence of storage deficits (lack of rapid study of 111 DZ and MZ twin pairs discordant for forgetting). Another 15% had a cortical (hippocam- schizophrenia280 found that deficits in working pal-thalamic, AD-type) profile (primary encoding and memory, attention, reaction time and word recall storage impairment, with an excess of irrelevant word intrusions were highly heritable. Thus, the balance of intrusions on free and cued recall), while the profiles the evidence suggests that cognitive dysfunction of the remaining 35% did not deviate significantly meets most of the criteria198 of an endophenotype in from those of the controls. These findings were schizophrenia. This conclusion is underscored by the replicated by Turetsky et al.237 and supported by meta-analysis by Heinrichs and Zakzanis281 of 204 neuroimaging data suggesting ventricular enlarge- studies published between 1980 and 1994 (a total of ment with preserved temporal lobe grey matter and 7420 schizophrenia patients and 5865 controls), in no significant metabolic abnormalities in the sub- which effect sizes (Cohen’s d) and the U statistic cortical group. In contrast, the cortical group was (degree of nonoverlap) were calculated for 22 neuro- characterized by a left-hemisphere temporal and cognitive test variables ranging from IQ, verbal frontal volume reduction, and metabolic abnormal- memory, and attention to executive function, and ities in the superior temporal gyrus, hippocampus, language. Neurocognitive deficit was found to be a and thalamus. Using a different statistical approach, reliable and well replicated finding in schizophrenia, Dickinson et al.241 estimated that over 30% of the although no single test or cognitive construct was variance in cognitive test performance by schizophre- capable of separating perfectly schizophrenia patients nia patients could be explained by a large-effect ‘g’ from normal controls. Seven widely used measures factor, affecting fundamental processes that integrate achieved effect sizes greater than 1.0 (60–70% non- multiple intermodal brain functions into ‘core’ cog- overlap between the cases and controls): global verbal nitive operations such as concept formation and memory (1.41), bilateral motor skills (1.30), perfor- reasoning skills. Further variance, however, can be mance IQ (1.26), the continuous performance task explained by a number of independent, small-effect (1.16), word fluency (1.15), the Stroop task (1.11), and variables selectively affecting specific functions, such WAIS-R IQ (1.10). Although a subset of B50% of as processing speed and visual memory. Table 4 patients had nearly normal performance, significant provides an overview of proposed cognitive subtypes cognitive impairment was common in schizophrenia associated with schizophrenia. and exceeded the deficits found in some neurological Promising as they are, these approaches to ‘splitting disorders, justifying the view that ‘schizophrenia is a schizophrenia’293 are limited by sample size, as well neurological disorder that manifests itself in beha- as by insufficient efforts to integrate multidomain viour’.281 data (e.g. neuroimaging and neurophysiological mea- surements) that might increase their capacity to parse Subtypes of cognitive dysfunction the deficits characterizing schizophrenia. Cognitive deficits in schizophrenia are heteroge- Cognitive phenotypes have rarely been tested as neous, ranging from pervasive generalized dysfunc- phenotypes in molecular genetic studies. In one such tion through patchy focal disorders to mild focal study, Egan et al.294 used working memory/executive deficits or nearly normal performance.282–288 Yet function (assessed by the Wisconsin Card Sorting test, amidst seemingly extensive heterogeneity, converging WCST) as the phenotype in a functional investigation evidence points to specific deficits in verbal declara- of the val158/108met polymorphism in the catechol- tive memory and working memory (mainly in the o-methyltransferase (COMT) gene. Allele dosage early encoding stage) as major sources of var- effect of worsening WCST performance and reduced iance.289,290 This observation has prompted attempts fMRI activation response in the prefrontal cortex at delineating particular profiles or subtypes. was found for the val/met and val/val genotypes, While conventional cluster analyses tend to simply likely due to a more rapid inactivation of synaptic

Molecular Psychiatry Subtyping schizophrenia A Jablensky 825 dopamine by the COMT-val variant. No effect on first-degree relatives, and controls. The test results sustained attention was detected.238 In another study, were analysed for complex patterns of dysfunction using a more inclusive neurocognitive battery, Bilder using a GoM model, a form of latent structure analysis et al.295 found greater impact of the val158/108met which defines a parsimonious number of latent genotype on processing speed and attention than on groups or patterns of responses (‘pure types’), allow- executive function, suggesting that the effect of the ing individuals to resemble each group to varying COMT polymorphism on cognition may not be degrees, rather than allocating them to mutually exclusively mediated by prefrontal dopamine. Con- exclusive clusters as performed in standard sidering that the COMT-val allele has shown an asso- LCA.149,150 GoM resolves sample heterogeneity by ciation with schizophrenia in some reports,296–298 assigning to each individual GoM scores of affinity although not in others,299–301 the possibility that to each one of several pure types. The resulting this polymorphism might contribute to the risk of classification identified two distinct pure types of schizophrenia requires further study. multivariate neurocognitive profiles (Figure 1), which In a study of 168 Finnish families with schizo- comprised over 90% of the schizophrenia patients in phrenia, Paunio et al.302 applied a variance compo- the sample, as well as 23% of their clinically nent analysis (SOLAR) to genome scan data, using 11 unaffected first-degree relatives: a cognitive deficit neuropsychological test battery scores obtained from (CD) subtype and a cognitively spared (CS) subtype. probands and relatives as quantitative trait pheno- In the CD subtype verbal memory impairment was the types before linkage analysis. Compared with diag- most consistently observed cognitive deficit. Multiple nosis only as the phenotype, use of quantitative traits (3 þ ) cognitive deficits were exhibited by all CD resulted in a stronger signal and evidence of linkage cases, and estimated current IQ was low in the for verbal learning and memory over a 30 cM region majority of these patients. Consequently, generalized on 4q13–25 (Zmp = 3.84), as well as suggestive cognitive deficiency was the most salient character- evidence for visual working memory on 2q36 istic of this subtype, in contrast to mild or patchy (Zmp = 2.08), visual attention on 15q22 and executive deficits in the CS subtype. function on 9p22. Although the advantages of using To test the hypothesis that the two subtypes are quantitative traits are demonstrated by this study, the genetically distinct, a 10 cM whole genome scan was interpretation of specific linkage findings for sub- performed using 380 microsatellite markers in 93 component cognitive processes will be tenuous until schizophrenia families (34 of which had been as- replication or convergent evidence from endopheno- signed the CD subtype). Linkage analysis revealed typing studies become available. evidence for linkage on several chromosomes, with the most significant finding at 6p25–24.243 The data The Western Australian family study of schizophrenia were then assessed by ordered subsets analyses The Western Australian family study of schizophre- (OSA), where the families were rank-ordered by their nia (WAFSS) was designed as a testbed for exploring CD and CS quantitative trait scores (highest to lowest) heterogeneity in schizophrenia, delineating genetic and compared for changes in the logarithm of odds variants, or subtypes, and putting them to the test of (LOD) scores between the subset and the overall genetic linkage analysis. Our group adopted from the sample. The greatest increase in the LOD score, at outset the conjecture that: (a) the broad syndrome of 6p24, was accounted for by the families ranked from schizophrenia is a conflation of several underlying 1st to 47th by the proband’s CD score. Fifteen disorders that may be aetiologically distinct; and (b) additional microsatellite markers genotyped in this endophenotypes anchored in objective measures of region (6p25–22) increased the maximal LOD score in brain dysfunction might separate out such disorders the CD families to 3.32. Linkage was excluded in the more clearly than clinical symptoms alone. The non-CD families for the entire chromosome 6. The available evidence was pointing to cognitive and 6p25–24 locus coincides almost exactly with the neurobehavioural measures as being particularly linkage findings in 270 Irish schizophrenia families, sensitive to dysfunction associated with schizo- previously reported by Straub et al.303 The coincident phrenia, as well as being longitudinally stable, state- linkage region and the common ancestry (Anglo-Irish) independent, and heritable. As most neurocognitive of our families lead us to believe that they share the tasks typically engage several component processes, same susceptibility allele(s). Important independent we reasoned that a linear composite of such variables support for these findings was recently provided by a would be an appropriate endophenotype for genetic study of general cognitive abilities in 634 healthy sib studies and developed a design allowing simulta- pairs,304 which reported quantitative trait loci evi- neous analysis of performance in the various cogni- dence of linkage of full-scale IQ and verbal IQ to the tive domains for shared patterns of dysfunction, same 6p region. A notable aspect of the linkage results rather than for isolated deficits. is that they were predicted with considerable accu- In this study, involving 112 families (388 members, racy by the composite endophenotypes defined prior of which 138 affected with schizophrenia or schizo- to genetic analysis (Figure 2), and that the distribu- phrenia spectrum disorders) and 143 population tions of several of the cognitive measures differentiat- controls, we employed a comprehensive clinical and ing the two subtypes suggest bimodality, supporting cognitive assessment protocol243 to evaluate patients, their relative independence.

Molecular Psychiatry Subtyping schizophrenia A Jablensky 826 2.0 cognitive deficit (CD)

1.5 cognitively spared (CS)

1.0

0.5

0.0 z-scores

-0.5

-1.0

Baseline: standardised performance of controls -1.5

-2.0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

neurocognitive measures neurobehavioural measures 1 - Premorbid IQ 9 - Schizotypal symptoms 2 - Current IQ 10 - Neurological soft signs 3 - Verbal memory (immediate recall) 11 - Handedness lateralisation quotient 4 - Verbal memory (delayed recall) 5 - Verbal fluency personality factors (TCI) 6 - Sustained attention CPT-IP 12 - Harm-avoidance 7 - Sustained attention CPT-DS 13 - Self-transcendence 8 - Speed of information processing (inspection time) 14 - Self-directedness Figure 1 Profiles of the two cognitive subtypes (CD and CS), identified in the Western Australian Family Study of Schizophrenia.243

Clinically, the CD subtype exhibits some affinity relationship between the symptom profiles and a with the deficit syndrome described by Carpenter validating variable, such as outcome or heritability.51 et al.108 and Kirkpatrick et al.109 with predominance of The inconsistent and poorly replicated results of negative symptoms, relative paucity of complex genetic linkage and association studies using the delusions, and poor social functioning. Subsequent diagnostic category as the sole schizophrenia pheno- statistical analyses by our group indicate that a fair type are kindling discontent with the current noso- approximation to the CD/CS typology can be achieved logy of schizophrenia, based on the recognition that with an abridged, cost-efficient neurocognitive bat- ‘current nosology, now embedded in DSM-IV, tery, which should facilitate the replication of this although useful for other purposes, does not define approach to subtyping schizophrenia and its genetic phenotypes for genetic study’.194 It is now almost underpinnings. certain that the current broad diagnostic concept of schizophrenia does not demarcate a specific genetic entity. Deconstructing a complex disease? Schizophrenia geneticists are facing a particularly The overview of the evidence suggests that pheno- difficult situation, seeking to discover specific genes typic variability has been confounding the search for contributing to an overinclusive diagnostic category the causes of schizophrenia since the inception of the for which no specific biological substrate has yet been diagnostic category. Attempts at redefining its bound- identified – most likely due to extensive genetic aries by either ‘lumping’ or ‘splitting’ strategies24 have heterogeneity and an admixture of different under- been undertaken over decades, with limited success. lying disease subtypes. Many ‘top-down’ attempts Most such attempts, based on various rearrangements have been made to define an overarching disturbance of clinical symptoms and syndromes have ended in a in schizophrenia, sought in ‘a weakening of the failure to find natural boundaries between proposed mainsprings of volition’ and ‘loss of inner unity of clinical subtypes, either by locating a ‘zone of rarity’ mental activities’;1 ‘structural loosening of associa- between them, or by demonstrating a nonlinear tions’;45,305 ‘intrapsychic ataxia’;306 ‘neurointegrative

Molecular Psychiatry Subtyping schizophrenia A Jablensky 827 a 0.9 shared with predisposition to other disorders, although they primarily express a ‘common final 0.8 pathway’ within the schizophrenia spectrum. Such polymorphisms and deficits need not be intrinsically 0.7 pathological and may represent extreme variants of normal structure and function. Above a certain 0.6 -1 SD density threshold, their additive or nonlinear inter- MEAN action could give rise to the diagnostic symptoms in 0.5 + 1SD probands, but subclinical manifestations as endophe- notype traits will be detectable in otherwise healthy people, with a higher relative risk in biological

CD giks 0.4 relatives of probands. 0.3 While reasoning along such lines is increasingly common among researchers, the approaches proposed 0.2 to deal with the phenotype bottleneck in schizophre- -1SD nia research differ substantially. On one hand, there 0.1 MEAN are proposals to abandon the ‘Kraepelinian dichot- + 1SD omy’ of schizophrenic and affective disorders in 9,311 0.0 favour of a ‘psychosis-spectrum illness’ or a ‘shift from narrow phenotypes to broad endophenotypes’, CD CS associated with an even broader spectrum of abnor- 312 b mal behaviours and emotions. On the other hand, 1.0 there is an emerging ‘splitting’ agenda seeking and testing narrowly constrained phenotypes that may tag distinct variants or subtypes of schizophrenia,313 0.8 -1SD resolving at least part of its aetiological heterogeneity. MEAN ‘Candidate’ endophenotypes or markers of pathoge- netic processes affecting cognition, brain morphology + 1SD and neurophysiology constitute the mainstay of this 0.6 approach. Several genetic linkage and association studies employing such endophenotypes have pro- duced promising results175,176,179,180,191,192,243,302,314 that CS giks 0.4 set a high priority for replication. -1SD In the absence of direct evidence that schizophrenia MEAN is either a homogeneous multifactorial disease or an + 1SD amalgamation of aetiologically distinct component 0.2 disorders, both ‘lumping’ and ‘splitting’ strategies are legitimate and should be put to the test. The question is, which approach holds at present greater promise 0.0 for advancing schizophrenia genetics? Two arguments CD CS reinforce doubts that greater power for genetic studies would be achieved by redefining the clinical bound- Figure 2 Genetic findings (linkage to chromosome 6p25– aries of the phenotype. First, lumping different 22) in the Western Australian Family Study of Schizo- 243 disorders into an expanding phenotype of ‘psychosis’ phrenia for 93 fully characterized families phenotypically runs against the grain of a large body of clinical classified into a CD and a CS subtype, based on the research indicating that psychotic symptoms in the probands’ GoM (gik) scores on composite quantitative traits CD and CS. ’ = linked family; J = nonlinked family. (a) context of schizophrenia, other nonaffective psycho- Distribution of the families on trait CD. (b) Distribution of tic illnesses, and affective disorders are phenomen- the families on trait CS. ologically different315 and may be influenced by different genetic mechanisms, notwithstanding par- tial overlap in their effects. This would increase, defect’;307 ‘cognitive dysmetria’;308 and ‘dysconnec- rather than decrease, heterogeneity. Secondly, despite tion disorder’.309,310 Although intuitively appealing, the availability of diagnostic criteria for research and such formulations achieve little more than high- structured diagnostic instruments, misclassification lighting one or another of the many facets of a error in the fine-grain assessment of symptoms is complex syndrome. It is doubtful that a specific likely to remain a factor compounding further the genetic basis for a causa prima explaining the heterogeneity of family or case–control samples phenomenology of schizophrenia will ever be found. collected at different sites and at different times. In contrast, it appears almost certain that the genetic Such heterogeneity is likely to be a serious problem in polymorphisms and neurobiological deficits under- whole-genome association studies, which require lying schizophrenia are multiple, varied, and partly very large case-control samples, feasible only by

Molecular Psychiatry Subtyping schizophrenia A Jablensky 828 pooling data collections from multiple sites. In cohesion and a characteristic evolution over time. contrast, subtyping strategies are supported by The dissection of the syndrome into modular en- mounting evidence that sample stratification, parti- dophenotypes with specific neurocognitive or neuro- cularly using quantitative traits as covariates, can physiological underpinnings is beginning to be reduce heterogeneity and substantially increase perceived as a promising approach in schizophrenia power.316–321 This approach has scored successes in genetics. The current evidence is neither final nor the genetics of other complex diseases and its static, and needs to be re-examined as new concepts application to schizophrenia genetics will bring the and technologies coming from molecular genetics, disorder into the mainstream of current research into cognitive science, or brain imaging bring forth new the common genetic diseases. perspectives on disease causation and brain function. What kind of data would constitute supportive This must be complemented by a refined, reliable, evidence for distinct component disorders or sub- and valid phenotyping not only at the level of types within schizophrenia? Converging evidence symptoms, but involving correlated neurobiological from endophenotype-based studies suggests that features. The study of endophenotypes cutting across measures of neurocognitive dysfunction arguably the conventional diagnostic boundaries may reveal provide the largest effect sizes and increases in unexpected patterns of associations with symptoms, relative risk to relatives among a host of ‘candidate’ personality traits, or behaviour. The mapping of endophenotypes,236,237,241,281,294,295 being also cost effi- clinical phenomenology on specific brain dysfunction cient for phenotyping large samples. In particular, (and vice versa) is becoming feasible and the resulting several characteristic patterns of short-term and functional psychopathology322,326 may in the future working memory impairment against a background substantially recast the present nosology. of generalized cognitive deficit have been replicated across studies and are present in a substantial proportion (B50%) of schizophrenia patients. As Acknowledgments many of neurocognitive tests tap into several compo- Work on the Western Australian Family Study of nent processes, composite endophenotypes, integrat- Schizophrenia has been supported by grants (to AJ) ing multiple neurocognitive measures, are more likely from the National Health and Medical Research to capture variation that is genetically influenced Council, Australia, and the North Metropolitan than single-feature endophenotypes. The subtypes Health Services, Perth, Western Australia. Invaluable generated by such approaches should be capable of assistance with the bibliography and illustrations for classifying individuals, rather than variables, and the this review was provided by Lyn Kløve, Milan resulting classification is likely to be polythetic (based Dragovic´ and Vera Morgan. on subsets of correlated features, rather than on the presence of all defining attributes). Whether subtypes are discrete taxa, that is, identifiable by marked areas References of discontinuity with other subtypes; dimensional, 1 Kraepelin E. Psychiatrie. 8 Auflage. Barth: Leipzig, 1909. representing continua with fuzzy boundaries; or [Reprinted English translation: Dementia Praecox and Paraphre- hybrid (class-quantitative, with dimensions super- nia. Krieger Publishing: Huntington, New York, 1971]. imposed on discrete categories), is testable with 2 World Health Organization. The ICD-10 Classification of Mental taxometric methods common in biological classifica- and Behavioural Disorders. Diagnostic Criteria for Research. 323,324 WHO: Geneva, 1993. tions. In the context of genetic research, the most 3 American Psychiatric Association. Diagnostic and Statistical significant criterion of their validity will be the gain Manual of Mental Disorders, 4th edn. American Psychiatric in predictive power and ‘process understanding’325 in Association: Washington, DC, 1994. the sense of mechanistic explanation of disease 4 Harrison PJ, Owen MJ. Genes for schizophrenia? Recent findings phenomena. and their pathophysiological implications. Lancet 2003; 361: 417–419. To sum up, we do not know whether schizophrenia 5 Harrison PJ, Weinberger DR. Schizophrenia genes, gene expres- is a single process with pleiotropic manifestations at sion, and neuropathology: on the matter of their convergence. Mol the level of cerebral organization, or a collection of Psychiatry 2004; 10: 40–68. aetiologically unrelated but dynamically interacting 6 Norton N, Williams HJ, Owen MJ. An update on the genetics of schizophrenia. Curr Opin Psychiatry 2006; 19: 158–164. processes. Although there are good grounds for the 7 Bentall RP, Jackson HF, Pilgrim D. Abandoning the concept of suspicion that schizophrenia is not a homogeneous ‘schizophrenia’: some implications of validity arguments for entity, this has never been directly demonstrated, psychological research into psychotic phenomena. Br J Clin mainly because few studies of the appropriate kind Psychol 1988; 27: 303–324. have ever been undertaken. Its manifestations in toto 8 Boyle M. Schizophrenia. A Scientific Delusion?. Routledge: London and New York, 1990. do not fit neatly into the proposed disease models, 9 Craddock N, Owen MJ. The beginning of the end for the although reasoning by analogy suggests an affinity to Kraepelinian dichotomy. Br J Psychiatry 2005; 186: 364–366. other complex multifactorial diseases, such as cancer, 10 Jablensky A. Epidemiology of schizophrenia: the global burden of ischemic heart disease, or diabetes. For the time disease and disability. Eur Arch Psychiatry Clin Neurosci 2000; 250: 274–285. being, the clinical concept of schizophrenia is 11 Jablensky A. The epidemiological horizon. In: Hirsch SR, supported by empirical evidence that its multiple Weinberger DR (eds). Schizophrenia, 2nd edn. Blackwell facets form a broad syndrome with some internal Publishing: Massachusetts, USA, 2003, pp. 203–231.

Molecular Psychiatry Subtyping schizophrenia A Jablensky 829 12 Jablensky A, Sartorius N, Ernberg G, Anker M, Korten A, Cooper 36 Chesler EJ, Lu L, Shou S, Qu Y, Gu J, Wang J et al. Complex trait JE et al. Schizophrenia: manifestations, incidence and course in analysis of gene expression uncovers polygenic and pleiotropic different cultures. A World Health Organization ten-country networks that modulate nervous system function. Nat Genet study. Psychol Med Nonogr Suppl 1992; 20: 1–97. 2005; 37: 233–242. 13 DeLisi LE, Nasrallah HA. Current controversies in schizophrenia 37 Zubenko GS. Do susceptibility loci contribute to the expression research. I. Is schizophrenia a heterogeneous disorder? Schizophr of more than one mental disorder? A view from the genetics of Res 1995; 17: 133. Alzheimer’s disease. Mol Psychiatry 2000; 5: 131–136. 14 Crow TJ. A continuum of psychosis, one human gene, and not 38 Holliday E, Mowry B, Chant D, Nyholt D. The importance of much else – the case for homogeneity. Schizophr Res 1995; 17: modelling heterogeneity in complex disease: application to 135–145. NIMH schizophrenia genetics initiative data. Hum Genet 2005; 15 Goldberg TE, Weinberger DR. A case against subtyping in 117: 160–167. schizophrenia. Schizophr Res 1995; 17: 147–152. 39 Gottesman II, Bertelsen A. Confirming unexpressed genotypes for 16 Cardno AG, Farmer AE. The case for or against heterogeneity in schizophrenia. Risks in the offspring of Fischer’s Danish the etiology of schizophrenia. The genetic evidence. Schizophr identical and fraternal discordant twins. Arch Gen Psychiatry Res 1995; 17: 153–159. 1989; 46: 867–872. 17 Tsuang MT, Faraone SV. The case for heterogeneity in the etiology 40 Moffitt TE, Caspi A, Rutter M. Strategy for investigating of schizophrenia. Schizophr Res 1995; 17: 161–175. interactions between measured genes and measured environ- 18 Heinrichs RW. Meta-analysis and the science of schizophrenia: ments. Arch Gen Psychiatry 2005; 62: 473–481. variant evidence or evidence of variants? Neurosci Biobehav Rev 41 Bearden CE, Reus VI, Freimer NB. Why genetic investigation of 2004; 28: 379–394. psychiatric disorders is so difficult. Curr Opin Genet Dev 2004; 19 Ghosh S, Watanabe RM, Valle TT, Hauser ER, Magnuson VL, 14: 280–286. Langefeld CD et al. The Finland-United States Investigation of 42 Tsuang MT, Lyons MJ, Faraone SV. Heterogeneity of schizo- Non-Insulin-Dependent Diabetes Mellitus Genetics (FUSION) phrenia. Conceptual models and analytic strategies. Br J Study. I. An autosomal genome scan for genes that predispose Psychiatry 1990; 156: 17–26. to Type 2 diabetes. Am J Hum Genet 2000; 67: 1174–1185. 43 Gibbons RD, Dorus E, Ostrow DG, Pandey GN, Davis JM, Levy DL. 20 Laitinen T, Daly MJ, Rioux JD, Kauppi P, Laprise C, Peta¨ys T et al. Mixture distributions in psychiatric research. Biol Psychiatry A susceptibility locus for asthma-related traits on chromosome 7 1984; 19: 935–961. revealed by genome-wide scan in a founder population. Nat 44 Moldin SO, Rice JP, Gottesman II, Erlenmeyer-Kimling L. Genet 2001; 28: 87–91 (letter). Psychometric deviance in offspring at risk for schizophrenia: II. 21 Dekker MCJ, Bonifati V, van Duijn CM. Parkinson’s disease: Resolving heterogeneity through admixture analysis. Psychiatry piecing together a genetic jigsaw. Brain 2003; 126: 1722–1733. Res 1990; 32: 311–322. 22 Shao Y, Cuccaro ML, Hauser ER, Raiford KL, Menold MM, 45 Bleuler E. Lehrbuch der Psychiatrie. Springer Verlag: Berlin, 1920 Wolpert CM et al. Fine mapping of autistic disorder to [Reprinted English translation: Textbook of Psychiatry. Arno chromosome 15q11–q13 by use of phenotypic subtypes. Am J Press: New York, 1976.]. Hum Genet 2003; 72: 539–548. 46 Feighner JP, Robins E, Guze SB, Woodruff RA, Winokur G, Munoz 23 Scott WK, Hauser ER, Schmechel DE, Welsh-Bohmer KA, Small R. Diagnostic criteria for use in psychiatric research. Arch Gen GW, Roses AD et al. Ordered-subsets linkage analysis detects Psychiatry 1972; 26: 57–63. novel Alzheimer disease loci on chromosomes 2q34 and 15q22. 47 Sullivan PF, Kendler KS, Neale MC. Schizophrenia as a complex Am J Hum Genet 2003; 73: 1041–1051. trait. Evidence from a meta-analysis of twin studies. Arch Gen 24 McKusick VA. On lumpers and splitters, or the nosology of Psychiatry 2003; 60: 1187–1192. genetic disease. Perspect Biol Med 1969; Winter: 298–312. 48 Badner JA, Gershon ES. Meta-analysis of whole-genome linkage 25 Smith CAB. Testing for heterogeneity of recombinant values in scans of bipolar disorder and schizophrenia. Mol Psychiatry human genetics. Ann Hum Genet 1963; 27: 175–182. 2002; 7: 405–411. 26 Risch N. A new statistical test for linkage heterogeneity. Am J 49 Lewis CM, Levinson DF, Wise LH, DeLisi LE, Straub RE, Hovatta I Hum Genet 1988; 42: 353–364. et al. Genome scan meta-analysis of schizophrenia and 27 Rivolta C, Sharon D, De Angelis MM, Dryja TP. Retinitis bipolar disorder, part II: Schizophrenia. Am J Hum Genet 2003; pigmentosa and allied diseases: numerous diseases, genes, and 73: 34–48. inheritance patterns. Hum Mol Genet 2002; 11: 1219–1227. 50 Wittgenstein L. Philosophische Bemerkunen II Band. Springer 28 Duerr RH. The genetics of inflammatory bowel disease. Verlag: Vienna, 1964. Gastroenterol Clin North Am 2002; 31: 63–76. 51 Kendell R, Jablensky A. Distinguishing between the validity 29 Miterski B, Drynda S, Boschow G, Klein W, Oppermann J, Kekow and utility of psychiatric diagnoses. Am J Psychiatry 2003; 160: J et al. Complex genetic predisposition in adult and juvenile 4–12. rheumatoid arthritis. BMC Genet 2004; 5:2. 52 Stassen HH, Scharfetter C, Winokur G, Angst J. Familial 30 Shen H, Liu Y, Liu P, Recker RR, Deng HW. Nonreplication in syndrome patterns in schizophrenia, schizoaffective disorder, genetic studies of complex diseases – lessons learned from mania, and depression. Eur Arch Psychiatry Clin Neurosci 1988; studies of osteoporosis and tentative remedies. J Bone Miner Res 237: 115–123. 2005; 20: 365–376. 53 Bebbington P. Recent findings in bipolar affective disorder. 31 Ammar N, Nelis E, Merlini L, Barisic N, Amouri R, Ceuterick C Psychol Med 2004; 34: 767–776. et al. Identification of novel GDAP1 mutations causing autosomal 54 Keefe RSE, Frescka E, Apter SH, Davidson M, Macaluso JM, recessive Charcot-Marie-Tooth disease. Neuromuscul Disord Hirschowitz J et al. Clinical characteristics of Kraepelinian 2003; 13: 720–728. schizophrenia: replication and extension of previous findings. 32 Stojkovic T, Latour P, Viet G, de Seze J, Hurtevent JF, Am J Psychiatry 1996; 153: 806–811. Vandenberghe A et al. Vocal cord and diaphragm paralysis, as 55 Bralet M-C, Loas G, Yon V, Mare´chal V. Clinical characteristics clinical features of a French family with autosomal recessive and risk factors for Kraepelinian subtype of schizophrenia: Charcot-Marie-Tooth disease, associated with a new mutation in replication of previous findings and relation to summer birth. the GDAP1 gene. Neuromuscul Disord 2004; 14: 261–264. Psychiatry Res 2002; 111: 147–154. 33 Cutting GR. Modifier genetics: cystic fibrosis. Annu Rev 56 Roy M-A, Lehoux C, E´ mond C, Laplante L, Bouchard R-H, Everett Genomics Hum Genet 2005; 6: 237–260. J et al. A pilot neuropsychological study of Kraepelinian and non- 34 Cohn JA, Neoptolemos JP, Feng J, Yan J, Jiang Z, Greenhalf W et Kraepelinian schizophrenia. Schizophr Res 2003; 62: 155–163. al. Increased risk of idiopathic chronic pancreatitis in cystic 57 Kasanin J. The acute schizoaffective psychosis. Am J Psychiatry fibrosis carriers. Hum Mutat 2005; 26: 303–307. 1933; 90: 97–126. 35 Larriba S, Bonache S, Sarquella J, Ramos MD, Gimenez J, Bassas L 58 Langfeld G. The prognosis of schizophrenia. Acta Psychiatria et al. Molecular evaluation of CFTR sequence variants in male Neurologica Scandinavica (Suppl 110). Munksgaard: Copenha- infertility of testicular origin. Int J Androl 2005; 28: 284–290. gen, 1956.

Molecular Psychiatry Subtyping schizophrenia A Jablensky 830 59 Stephens JH, Astrup C. Prognosis in ‘process’ and ‘non-process’ 80 Kallman FJ. The Genetics of Schizophrenia. Augustin: New York, schizophrenia. Am J Psychiatry 1963; 119: 945–953. 1938. 60 Tsuang MT, Winokur G. Criteria for subtyping schizophrenia. 81 Kendler KS, Gruenberg AM. An independent analysis of the Arch Gen Psychiatry 1974; 31: 43–47. Danish Adoption Study of Schizophrenia. VI. The relationship 61 Schneider K. Klinische Psychopathologie, 8th edn. Thieme: between psychiatric disorders as defined by DSM-III in the Stuttgart, 1950 [English translation Hamilton MW, Anderson relatives and adoptees. Arch Gen Psychiatry 1984; 41: 555–564. EW: Clinical Psychopathology. Grune and Stratton: New York, 82 Baron M, Risch N. The spectrum concept of schizophrenia: 1959.]. evidence for a genetic-environmental continuum. J Psychiatry 62 Spitzer RL, Endicott J, Robins E. Research diagnostic criteria. Res 1987; 21: 257–267. Rationale and reliability. Arch Gen Psychiatry 1978; 35: 773–782. 83 Rado S. Theory and therapy: the theory of schizotypal organiza- 63 American Psychiatric Association. Diagnostic and Statistical tion and its application to the treatment of decompensated Manual of Mental Disorders, 3rd edn. American Psychiatric schizotypal behaviour. In: Scher SC, Davis HR (eds) The Association: Washington, DC, 1980. Outpatient Treatment of Schizophrenia. Grune and Stratton: 64 Wing JK, Cooper JE, Sartorius N. Measurement and Classification New York, 1960. of Psychiatric Symptoms. An Instruction Manual for the PSE and 84 Meehl PE. Schizotaxia, schizotypy, schizophrenia. American CATEGO Program. Cambridge University Press: London, Psychologist 1962; 17: 827–838 [Reprinted in Meehl PE. Psycho- England, 1974. diagnosis: selected papers. University of Minnesota Press: 65 Loftus J, DeLisi LE, Crow TJ. Factor structure and familiality of Minneapolis, 1973, pp 135–155.]. first-rank symptoms in sibling pairs with schizophrenia and 85 Chapman LJ, Chapman JP. Scales for rating psychotic and schizoaffective disorder. Br J Psychiatry 2000; 177: 15–19. psychotic-like experiences as continua. Schizophr Bull 1980; 6: 66 Cardno AG, Sham PC, Farmer AE, Murray RM, McGuffin P. 476–489. Heritability of Schneider’s first-rank symptoms. Br J Psychiatry 86 Kety SS, Wender PH, Jacobsen B, Ingraham LJ, Jansson L, Faber B. 2002; 180: 35–38. Mental illness in the biological and adoptive relatives of 67 Kendler KS, Myers JM, O’Neill FA, Martin R, Murphy B, schizophrenic adoptees. Arch Gen Psychiatry 1994; 51: 442–455. MacLean CJ, et al. Clinical features of schizophrenia and linkage 87 Kendler KS, Neale MC, Walsh D. Evaluating the spectrum to chromosomes 5q, 6p, 8p, and 10p in the Irish Study of concept of schizophrenia in the Roscommon Family Study. High-Density Schizophrenia Families. Am J Psychiatry 2000; 157: Am J Psychiatry 1995; 152: 749–754. 402–408. 88 Tienari P, Wynne LC, La¨ksy K, Moring J, Nieminen P, Sorri A et al. 68 Leonhard K. Classification of Endogenous Psychoses and their Genetic boundaries of the schizophrenia spectrum: evidence Differentiated Etiology, 2nd edn. Springer Verlag: Vienna, New from the Finnish Adoptive Family Study of Schizophrenia. Am J York, 1999. Psychiatry 2003; 160: 1587–1594. 69 American Psychiatric Association. Diagnostic and Statistical 89 Fogelson DL, Nuechterlein KH, Asarnow RF, Payne DL, Subotnik Manual of Mental Disorders, 3rd edn, Revised. American KL, Giannini CA. The factor structure of schizophrenia spectrum Psychiatric Association: Washington, DC, 1987. personality disorders: signs and symptoms in relatives of 70 Franzek E, Beckmann H. Different genetic background of psychotic patients from the UCLA family members study. schizophrenia spectrum psychoses: a twin study. Am J Psychiatry Psychiatry Res 1999; 87: 137–146. 1998; 155: 76–83. 90 Fosatti A, Maffei C, Battaglia M, Bagnato M, Donati D, Donini M 71 Beckmann H, Franzek E, Sto¨ber G. Genetic heterogeneity in et al. Latent class analysis of DSM-IV schizotypal personality catatonic schizophrenia: a family study. Am J Med Genet disorder criteria in psychiatric patients. Schizophr Bull 2001; 27: Neuropsychiatry Genet 1996; 67: 289–300. 59–71. 72 Sto¨ber G, Saar K, Ru¨ schendorf F, Meyer J, Nu¨ rnberg G, Jatzke S et 91 Suhr JA, Spitznagel MB. Factor versus cluster models of al. Splitting schizophrenia: periodic catatonia-susceptibility schizotypal traits. I. A comparison of unselected and highly locus on chromosome 15q15. Am J Hum Genet 2000; 67: schizotypal samples. Schizophr Res 2001; 52: 231–239. 1201–1207. 92 Kendler KS, McGuire M, Gruenberg AM, Walsh D. Schizotypal 73 Sto¨ber G, Seelow D, Ru¨ schendorf F, Ekici A, Beckmann H, Reis A. symptoms and signs in the Roscommon Family Study. Arch Gen Periodic catatonia: confirmation of linkage to chromosome 15 Psychiatry 1995; 52: 296–303. and further evidence for genetic heterogeneity. Hum Genet 2002; 93 Torgersen S, Onstad S, Skre I, Edvardsen J, Kringlen E. ‘True’ 111: 323–330. schizotypal personality disorder: a study of co-twins and 74 Meyer J, Ortega G, Schraut K, Nu¨ rnberg G, Ru¨ schendorf F, Saar K relatives of schizophrenic probands. Am J Psychiatry 1993; 150: et al. Exclusion of the neuronal nicotinic acetylcholine receptor 1661–1667. alpha7 subunit gene as a candidate for catatonic schizophrenia in 94 Fanous A, Gardner C, Walsh D, Kendler KS. Relationship a large family supporting the chromosome 15q13–22 locus. between positive and negative symptoms of schizophrenia and Mol Psychiatry 2002; 7: 220–223. schizotypal symptoms in nonpsychotic relatives. Arch Gen 75 Kury S, Rubie C, Moisan JP, Sto¨ber G. Mutation analysis of the Psychiatry 2001; 58: 669–673. zinc transporter gene SLC30A4 reveals no association with 95 Cadenhead KS, Perry W, Shafer K, Braff DL. Cognitive functions periodic catatonia on chromosome 15q15. J Neural Transm in schizotypal personality disorder. Schizophr Res 1999; 37: 2003; 110: 1329–1332. 123–132. 76 McKeane DP, Meyer J, Dobrin SE, Melmed KM, Ekawardhani S, 96 Nuechterlein KH, Asarnow RF, Subotnik KL, Fogelson DL, Payne Tracy NA et al. No causative DLL4 mutations in periodic DL, Kendler KS et al. The structure of schizotypy: relationships catatonia patients from 15q15 linked families. Schizophr Res between neurocognitive and personality disorder features in 2005; 75: 1–3. relatives of schizophrenic patients in the UCLA Family Study. 77 Meyer J, Huberth A, Ortega G, Syagailo YV, Jatzke S, Mo¨ssner R Schizophr Res 2002; 54: 121–130. et al. A missense mutation in a novel gene encoding a putative 97 Asarnow RF, Nuechterlein KH, Asamen J, Fogelson D, Subotnik cation channel is associated with catatonic schizophrenia in a KL, Zaucha K et al. Neurocognitive functioning and schizo- large pedigree. Mol Psychiatry 2001; 6: 302–306. phrenia spectrum disorders can be independent expressions of 78 Devaney JM, Donarum EA, Brown KM, Meyer J, Sto¨ber G, Lesch familial liability for schizophrenia in community control chil- KP et al. No missense mutation of WKL1 in a subgroup of dren: the UCLA Family Study. Schizophr Res 2002; 54: 111–120. probands with schizophrenia. Mol Psychiatry 2002; 7: 419–423. 98 Dickey CD, McCarley RW, Shenton ME. The brain in schizotypal 79 Ru¨ din E. Studien u¨ ber Vererbung und Enstehung geistiger personality disorder: a review of structural MRI and CT findings. Sto¨rungen, I: Zur Vererbung und Neuenstehung der Dementia Harv Rev Psychiatry 2002; 10: 1–15. praecox (Studies on the inheritance and origin of mental Illness, 99 Strauss JS, Carpenter Jr WT, Bartko JJ. The diagnosis and I: The problem of the inheritance and primary origin of Dementia understanding of schizophrenia. Part III. Speculations on the praecox). Monographien aus dem Gesamtgebeit der Neurologie processes that underlie schizophrenic symptoms and signs. und Psychiatrie, No. 12. Springer Verlag: Berlin, 1916. Schizophr Bull 1974; Winter: 61–69.

Molecular Psychiatry Subtyping schizophrenia A Jablensky 831 100 Crow TJ. Molecular pathology of schizophrenia: more than one 125 Johnstone EC, Frith CD. Validation of three dimensions of disease process? BMJ 1980; 280: 66–68. schizophrenic symptoms in a large unselected sample of 101 Andreasen NC, Olsen S. Negative v positive schizophrenia. patients. Psychol Med 1996; 26: 669–679. Definition and validation. Arch Gen Psychiatry 1982; 39: 126 Grube BS, Bilder RM, Goldman RS. Meta-analysis of symptom 789–794. factors in schizophrenia. Schizophr Res 1998; 31: 113–120. 102 Crow TJ. The two-syndrome concept: origins and current status. 127 Smith DA, Mar CM, Turoff BK. The structure of schizophrenic Schizophr Bull 1985; 11: 471–486. symptoms: a meta-analytic confirmatory factor analysis. Schi- 103 Berrios GE. Positive and negative symptoms and Jackson. A zophr Res 1998; 31: 57–70. conceptual history. Arch Gen Psychiatry 1985; 42: 95–97. 128 Arndt S, Andreasen NC, Flaum M, Miller D, Nopoulos P. A 104 Sommers AA. ‘Negative symptoms’: conceptual and methodolo- longitudinal study of symptom dimensions in schizophrenia. gical problems. Schizophr Bull 1985; 11: 364–379. Prediction and patterns of change. Arch Gen Psychiatry 1995; 52: 105 Tandon R, Greden JF. Negative symptoms of schizophrenia: the 352–360. need for conceptual clarity. Biol Psychiatry 1991; 30: 321–325. 129 Gureje O, Aderibigbe YA, Obikoya O. Three syndromes in 106 Peralta V, Cuesta MJ, De Leon J. Positive and negative symptoms/ schizophrenia: validity in young patients with recent onset of syndromes in schizophrenia: reliability and validity of different illness. Psychol Med 1995; 25: 715–725. diagnostic systems. Psychol Med 1995; 25: 43–50. 130 Emsley RA, Niehaus DJH, Mbanga NI, Oosthuizen PP, Stein DJ, 107 Jablensky A. Symptoms of schizophrenia. In: Henn F, Sartorius Maritz JS et al. The factor structure for positive and negative N, Helmchen H, Lauter H (eds). Contemporary Psychiatry. Vol 3. symptoms in South African Xhosa patients with schizophrenia. Specific Psychiatric Disorders. Springer Verlag: Berlin, 2001, Schizophr Res 2001; 47: 149–157. pp. 3–36. 131 Ratakonda S, Gorman JM, Yale SA, Amador XF. Characterization 108 Carpenter Jr WT, Heinrichs DW, Wagman AMI. Deficit and non- of psychotic conditions. Use of the domains of psychopathology deficit forms of schizophrenia: the concept. Am J Psychiatry model. Arch Gen Psychiatry 1998; 55: 75–81. 1988; 145: 578–583. 132 Kay SR, Fizbein A, Opler LA. The Positive and Negative 109 Kirkpatrick B, Buchanan RW, Breier A, Carpenter Jr WT. Case Syndrome Scale (PANSS) for schizophrenia. Schizophr Bull identification and stability of the deficit syndrome of schizo- 1987; 13: 261–276. phrenia. Psychiatry Res 1993; 47: 47–56. 133 Overall JE, Gorham DR. The Brief Psychiatric Rating Scale 110 Kirkpatrick B, Buchanan RW, Ross DE, Carpenter Jr WT. A (BPRS): recent developments in ascertainment and scaling. separate disease within the syndrome of schizophrenia. Psychopharmacol Bull 1988; 24: 97–99. Arch Gen Psychiatry 2001; 58: 165–171. 134 McGuffin P, Farmer AE, Harvey IH. A polydiagnostic application 111 Kirkpatrick B, Ross DE, Walsh D, Karkowski L, Kendler KS. of operational criteria in studies of psychotic illness: develop- Family characteristics of deficit and nondeficit schizophrenia in ment and reliability of the OPCRIT system. Arch Gen Psychiatry the Roscommon family study. Schizophr Res 2000; 45: 57–64. 1991a; 48: 764–770. 112 Buchanan RW, Strauss ME, Kirkpatrick B, Holstein C, Breier A, 135 Rey E-R, Bailer J, Bra¨uer W, Ha¨ndel M, Laubenstein D, Stein A. Carpenter Jr WT. Neuropsychological impairments in deficit vs Stability trends and longitudinal correlations of negative and nondeficit forms of schizophrenia. Arch Gen Psychiatry 1994; 51: positive syndromes within a three-year follow-up of initially 804–811. hospitalized schizophrenics. Acta Psychiatry Scand 1994; 90: 113 Buchanan RW, Strauss ME, Breier A, Kirkpatrick B, Carpenter Jr 405–412. WT. Attentional impairments in deficit and nondeficit forms of 136 Lindenmayer J-P, Grochowski S, Hyman RB. Five factor model of schizophrenia. Am J Psychiatry 1997; 154: 363–370. schizophrenia: replication across samples. Schizophr Res 1995; 114 Addington J, Addington D, Maticka-Tyndale E. Cognitive 14: 229–234. functioning and positive and negative symptoms in schizophre- 137 Serretti A, Rietschel M, Lattauda E, Krauss H, Schulze TG, Mu¨ ller nia. Schizophr Res 1991; 5: 123–134. DJ et al. Major psychoses symptomatology: factor analysis of 2241 115 Ross DE. The deficit syndrome and eye tracking disorder may psychotic subjects. Eur Arch Psychiatry Clin Neurosci 2001; 251: reflect a distinct subtype within the syndrome of schizophrenia. 193–198. Schizophr Bull 2000; 26: 855–866. 138 Cardno AG, Jones LA, Murphy KC, Asherson P, Scott LC, 116 Nkam I, Thibaut F, Denise P, Van Der Elst A, Se´gard L, Brazo P Williams J et al. Factor analysis of schizophrenic symptoms et al. Saccadic and smooth-pursuit eye movements in deficit and using the OPCRIT checklist. Schizophr Res 1996; 22: 233–239. non-deficit schizophrenia. Schizophr Res 2001; 48: 145–153. 139 Lenzenweger MF, Dworkin RH. The dimensions of schizophrenia 117 Hong LE, Avila MT, Adami H, Elliot A, Thaker GK. Components phenomenology. Not one or two, at least three, perhaps four. Br J of the smooth pursuit function in deficit and nondeficit Psychiatry 1996; 168: 432–440. schizophrenia. Schizophr Res 2003; 63: 39–48. 140 White L, Harvey PD, Opler L, Lindenmayer JP, The PANSS Study 118 Arango C, Kirkpatrick B, Buchanan RW. Neurological signs and Group. Empirical assessment of the factorial structure of clinical the heterogeneity of schizophrenia. Am J Psychiatry 2000; 157: symptoms in schizophrenia. Psychopathology 1997; 30: 263–274. 560–565. 141 McGrath JA, Nestadt G, Liang KY, Lasseter VK, Wolyniec PS, 119 Galderisi S, Maj M, Mucci A, Cassano GB, Invernizzi G, Rossi A Fallin MD et al. Five latent factors underlying schizophrenia: et al. Historical, psychopathological, neurological, and neuro- analysis and relationship to illnesses in relatives. Schizophr Bull psychological aspects of deficit schizophrenia: a multicenter 2004; 30: 855–873. study. Am J Psychiatry 2002; 159: 983–990. 142 Cuesta MJ, Peralta V. Integrating psychopathological dimensions 120 Blanchard JJ, Horan WP, Collins LM. Examining the latent in functional psychoses: a hierarchical approach. Schizophr Res structure of negative symptoms: is there a distinct subtype 2001; 52: 215–229. of negative symptom schizophrenia? Schizophr Res 2005; 77: 143 Farmer AE, McGuffin P, Spitznagel EL. Heterogeneity in 151–165. schizophrenia: a cluster-analytic approach. Psychiatry Res 121 Bakker SC, Hoogendoorn MLC, Selten J-P, Verduijn W, Pearson 1983; 8: 1–12. PL, Sinke RJ et al. Neuregulin 1: genetic support for schizo- 144 Dollfus S, Everitt B, Ribeyre JM, Assouly-Besse F, Sharp C, Petit phrenia subtypes. Mol Psychiatry 2004; 9: 1061–1063 (letter). M. Identifying subtypes of schizophrenia by cluster analyses. 122 Hong LE, Wonodi I, Avila MT, Buchanan RW, McMahon RP, Schizophr Bull 1996; 22: 545–555. Mitchell BD et al. Dihydropyrimidinase-related protein 2 (DRP-2) 145 McCutcheon AL. Latent Class Analysis. Sage Publications: gene and association to deficit and nondeficit schizophrenia. Am Beverly Hills, California, 1987. J Med Genet B Neuropsychiatry Genet 2005; 136B: 8–11. 146 Sham PC, Castle DJ, Wessely S, Farmer AE, Murray RM. Further 123 Lorr M, Klett CJ, McNair DM. Syndromes of Psychosis. Pergamon: exploration of a latent class typology of schizophrenia. Schizophr New York, 1963. Res 1996; 20: 105–115. 124 Liddle PF. The symptoms of chronic schizophrenia. A 147 Kendler KS, Karkowski LM, Walsh D. The structure of psychosis. re-examination of the positive-negative dichotomy. Br J Latent class analysis of probands from the Roscommon Family Psychiatry 1987; 151: 145–151. Study. Arch Gen Psychiatry 1998; 55: 492–499.

Molecular Psychiatry Subtyping schizophrenia A Jablensky 832 148 Murray V, McKee I, Miller PM, Young D, Muir WJ, Pelosi AJ et al. 171 Jones PB, Owen MJ, Lewis SW, Murray RM. A case-control study Dimensions and classes of psychosis in a population cohort: a of family history and cerebral cortical abnormalities in four-class, four-dimension model of schizophrenia and affective schizophrenia. Acta Psychiatry Scand 1993; 87: 6–12. psychoses. Psychol Med 2005; 35: 499–510. 172 Malaspina D, Harkavy-Friedman J, Corcoran C, Mujica-Parodi L, 149 Woodbury MA, Clive J, Garson A. Mathematical typology: a grade Printz D, Gorman JM et al. Resting neural activity distinguishes of membership technique for obtaining disease definition. subgroups of schizophrenic patients. Biol Psychiatry 2004; 56: Comput Biomed Res 1978; 11: 277–298. 931–937. 150 Manton KG, Woodbury MA, Tolley DH. Statistical Applications 173 Weiss KM, Chakraborty R, Majumder PP. Problems in the Using Fuzzy Sets. John Wiley: New York, 1994. assessment of relative risk of chronic disease among biological 151 Manton KG, Korten A, Woodbury MA, Anker M, Jablensky A. relatives of affected individuals. J Chron Dis 1982; 35: 539–551. Symptom profiles of psychiatric disorders based on graded 174 Eaves LJ, Kendler KS, Schulz SC. The familial sporadic disease classes: an illustration using data from the WHO classification: its power for the resolution of genetic International Pilot Study of Schizophrenia. Psychol Med 1994; and environmental etiologic factors. J Psychiatry Res 1986; 20: 24: 133–144. 115–130. 152 Lewis SW, Reveley AM, Reveley MA, Chitkara B, Murray RM. 175 Brzustowicz LM, Honer WG, Chow EWC, Hogan J, Hodgkinson K, The familial/sporadic distinction as a strategy in schizophrenia Bassett AS. Use of a quantitative trait to map a locus associated research. Br J Psychiatry 1987; 151: 306–313. with severity of positive symptoms in familial schizophrenia to 153 Kendler KS, Hayes P. Familial and sporadic schizophrenia: a chromosome 6p. Am J Hum Genet 1997; 61: 1388–1396. symptomatic, prognostic, and EEG comparison. Am J Psychiatry 176 Pulver AE, Mulle J, Nestadt G, Swartz KL, Blouin J-L, Dombroski 1982; 139: 1557–1562. B et al. Genetic heterogeneity in schizophrenia: stratification of 154 Gottesman II, McGuffin P, Farmer AE. Clinical genetics as clues genome scan data using co-segregating related phenotypes. Mol to the ‘real’ genetics of schizophrenia (A decade of modest gains Psychiatry 2000; 5: 650–653. while playing for time). Schizophr Bull 1987; 13: 23–47. 177 Chiu YF, McGrath JA, Thornquist MH, Wolyniec PS, Nestadt G, 155 Lyons MJ, Kremen WS, Tsuang MT, Faraone SV. Investigating Swartz KL et al. Genetic heterogeneity in schizophrenia II: putative genetic and environmental forms of schizophrenia: conditional analyses of affected schizophrenia sibling pairs methods and findings. Int Rev Psychiatr 1989; 1: 259–276. provide evidence for an interaction between markers on 156 Roy M-A, Crowe RR. Validity of the familial and sporadic chromosome 8p and 14q. Mol Psychiatry 2002; 7: 658–664. subtypes of schizophrenia. Am J Psychiatry 1994; 151: 805–814. 178 Fanous AH, Neale MC, Straub RE, Webb BT, O’Neill AF, Walsh D 157 Gorwood P, Leboyer M, Jay M, Payan C, Feingold J. Gender and et al. Clinical features of psychotic disorders and polymorphisms age at onset in schizophrenia: impact of family history. Am J in HT2A, DRD2, DRD4, SLC6A3 (DAT1), and BDNF: a family Psychiatry 1995; 152: 208–212. based association study. Am J Med Genet B Neuropsychiatry 158 Alda M, Ahrens B, Lit W, Dvorakova M, Labelle A, Zvolsky P Genet 2004; 125B: 69–78. et al. Age at onset in familial and sporadic schizophrenia. Acta 179 Fanous A, van den Oord E, Riley B, Aggen SH, Neale MC, O’Neill Psychiatry Scand 1996; 93: 447–450. FA et al. Relationship between a high-risk haplotype in the 159 Roy M-A, Flaum MA, Gupta S, Jaramillo L, Andreasen NC. dystrobrevin-binding protein 1 (DTNBP1) gene and clinical Epidemiological and clinical correlates of familial and sporadic features of schizophrenia. Am J Psychiatry 2005; 162: 1–9. schizophrenia. Acta Psychiatry Scand 1994; 89: 324–328. 180 Burdick KE, Lencz T, Funke B, Finn CT, Szeszko PR, Kane JM 160 Lewis SW, Murray RM. Obstetric complications, neurodevelop- et al. Genetic variation in DTNBP1 influences general cognitive mental deviance, and risk of schizophrenia. J Psychiatry Res ability. Hum Mol Genet 2006; 15: 1563–1568. 1987; 21: 413–421. 181 Murphy KC. The behavioural phenotype in velo-cardio-facial 161 Schwarzkopf SB, Nasrallah HA, Olson SC. Perinatal complica- syndrome. J Intellect Disabil Res 2004; 48: 524–530. tions and genetic loading in schizophrenia: preliminary findings. 182 Liu H, Abecasis GR, Heath SC, Knowles A, Demars S, Chen YJ Psychiatry Res 1989; 27: 233–239. et al. Genetic variation in the 22q11 locus and susceptibility to 162 O’Callaghan E, Larkin C, Kinsella A, Waddington JL. Obstetric schizophrenia. Proc Natl Acad Sci USA 2002; 99: 16859–16864. complications, the putative familial-sporadic distinction, and 183 Swillen A, Vogels A, Devriendt K, Fryns JP. Chromosome 22q11 tardive dyskinesia in schizophrenia. Br J Psychiatry 1990; 157: deletion syndrome: update and review of the clinical features, 578–584. cognitive-behavioral spectrum, and psychiatric complications. 163 Shur E. Season of birth in high and low genetic risk schizo- Am J Med Genet C Semin Med Genet 2000; 97: 128–135. phrenics. Br J Psychiatry 1982; 140: 410–415. 184 Horowitz A, Shifman S, Rivlin N, Pisante´ A, Darvasi A. A survey 164 O’Callaghan E, Gibson T, Colohan HA, Walshe D, Buckley P, of the 22q11 microdeletion in a large cohort of schizophrenia Larkin C et al. Season of birth in schizophrenia: evidence for patients. Schizophr Res 2005; 73: 263–267. confinement of an excess of winter births to patients without 185 Sporn A, Addington A, Reiss AL, Dean M, Gogtay N, Potocnik U a family history of mental disorder. Br J Psychiatry 1991; 158: et al. 22q11 deletion syndrome in childhood onset schizophre- 764–769. nia: an update. Mol Psychiatry 2004; 9: 225–226 (letter). 165 Reveley AM, Reveley MA, Murray RM. Cerebral ventricular 186 Bassett AS, Chow EWC, Abdel Malik P, Gheorghiu M, Husted J, enlargement in non-genetic schizophrenia: a controlled twin Weksberg R. The schizophrenia phenotype in 22q11 deletion study. Br J Psychiatry 1984; 144: 89–93. syndrome. Am J Psychiatry 2003; 160: 1580–1586. 166 Schwarzkopf SB, Nasrallah HA, Olson SC, Bogerts B, McLaugh- 187 van Amelsvoort T, Henry J, Morris R, Owen M, Linszen D, lin JA, Mitra T. Family history and brain morphology in Murphy K et al. Cognitive deficits associated with schizophrenia schizophrenia: an MRI study. Psychiatry Res 1991; 40: 49–60. in velo-cardio-facial syndrome. Schizophr Res 2004; 70: 223–232. 167 Bartels M, Mann K, Friedrich W. Tardive dyskinesia: marked 188 Chow EWC, Zipursky RB, Mikulis DJ, Bassett AS. Structural predominance of nongenetic schizophrenia. Biol Psychiatry brain abnormalities in patients with schizophrenia and 22q11 1985; 20: 94–119. deletion syndrome. Biol Psychiatry 2002; 51: 208–215. 168 McCreadie RG, Hall DJ, Berry IJ, Robertson LJ, Ewing JI, Geals 189 Schaer M, Schmitt JE, Glaser B, Lazeyras F, Delavelle J, Eliez S. MF. The Nithdale schizophrenia surveys: X. Obstetrical compli- Abnormal patterns of cortical gyrification in velo-cardio-facial cations, family history and abnormal movements. Br J Psychiatry syndrome (deletion 22q11.2): an MRI study. Psychiatry Res 1992; 161: 799–805. Neuroim 2006; 146: 1–11. 169 Griffiths TD, Sigmundsson T, Takei N, Frangou S, Birkett PB, 190 Millar JK, Wilson-Annan JC, Anderson S, Christie S, Taylor MS, Sharma T et al. Minor physical anomalies in familial and Semple CAM et al. Disruption of two novel genes by a sporadic schizophrenia: the Maudsley family study. J Neurol translocation co-segregating with schizophrenia. Hum Mol Genet Neurosurg Psychiatry 1998; 64: 56–60. 2000; 9: 1415–1423. 170 Walker E, Shaye J. Familial schizophrenia. A predictor of 191 Cannon TD, Hennah W, van Erp TGM, Thompson PM, Lo¨nnqvist neuromotor and attentional abnormalities in schizophrenia. Arch J, Huttunen M et al. Association of DISC1/TRAX haplotypes with Gen Psychiatry 1982; 39: 1153–1156. schizophrenia, reduced prefrontal gray matter, and impaired

Molecular Psychiatry Subtyping schizophrenia A Jablensky 833 short-and long-term memory. Arch Gen Psychiatry 2005; 62: schizophrenia: unifying basic research and clinical aspects. Eur 1205–1213. Arch Psychiatry Clin Neurosci 1999; 249(Suppl 4): IV/69–IV/82. 192 Burdick KE, Hodgkinson CA, Szeszko PR, Lencz T, Ekholm JM, 214 Michie PT, Innes-Brown H, Todd J, Jablensky AV. Duration Kane JM et al. DISC1 and neurocognitive function in schizo- mismatch negativity in biological relatives of patients with phrenia. Neuroreport 2005; 16: 1399–1402. schizophrenia spectrum disorders. Biol Psychiatry 2002; 52: 193 Cromwell RL, Elkins IJ, McCarthy ME, O’Neil TS. Searching for 749–758. the phenotypes of schizophrenia. Acta Psychiatry Scand 1994; 215 Baldeweg T, Klugman A, Gruzelier J, Hirsch SR. Mismatch 90 (Suppl 384): 34–39. negativity potentials and cognitive impairment in schizophrenia. 194 Ginsburg BE, Werick TM, Escobar JI, Kugelmass S, Treanor JJ, Schizophr Res 2004; 69: 203–217. Wendtland L. Molecular genetics of psychopathologies: a search 216 Butler PD, Zemon V, Schechter I, Saperstein AM, Hoptman MJ, for simple answers to complex problems. Behav Genet 1996; 26: Lim KO et al. Early-stage visual processing and cortical 325–333. amplification deficits in schizophrenia. Arch Gen Psychiatry 195 Leboyer M, Bellivier F, Nosten-Bertrand M, Jouvent R, Pauls D, 2005; 62: 495–504. Mallet J. Psychiatric genetics: search for phenotypes. Trends 217 Umbricht D, Krljes S. Mismatch negativity in schizophrenia: a Neurosci 1998; 21: 102–105. meta-analysis. Schizophr Res 2005; 76: 1–23. 196 Almasy L, Blangero J. Endophenotypes as quantitative risk 218 Holzman PS, Kringlen E, Matthysse S, Flanagan SD, Lipton RB, factors for psychiatric disease: rationale and study design. Am J Cramer G et al. A single dominant gene can account for eye Med Genet Neuropsychiatry Genet 2001; 105: 42–44. tracking dysfunctions and schizophrenia in offspring of discor- 197 Gottesman II, Shields J. Genetic theorizing and schizophrenia. Br dant twins. Arch Gen Psychiatry 1988; 45: 641–647. J Psychiatry 1973; 122: 15–30. 219 Levy DL, Lajonchere CM, Dorogusker B, Min DK, Lee S, 198 Gottesman II, Gould TD. The endophenotype concept in Tartaglini A et al. Quantitative characterization of eye tracking psychiatry: etymology and strategic intentions. Am J Psychiatry dysfunction in schizophrenia. Schizophr Res 2000; 42: 171–185. 2003; 160: 636–645. 220 Tregellas JR, Tanabe JI, Miller DE, Ross RG, Olincy A, Freedman 199 Keating M, Dunn C, Atkinson D, Timothy K, Vincent GM, Leppert R. Neurobiology of smooth pursuit eye movement deficits M. Consistent linkage of the long-QT syndrome to the Harvey in schizophrenia: an fMRI study. Am J Psychiatry 2004; 161: ras-1 locus on chromosome 11. Am J Hum Genet 1991; 49: 315–321. 1335–1339. 221 Matthysse S, Holzman PS, Gusella JF, Levy DL, Harte CB, 200 Leppert M, Burt R, Hughes JP, Samowitz W, Nakamura Y, Jørgensen A˚ et al. Linkage of eye movement dysfunction to Woodward S et al. Genetic analysis of an inherited predisposi- chromosome 6p in schizophrenia: Additional evidence. Am J tion to colon cancer in a family with a variable number of Med Genet B Psychiatry Genet 2004; 128B: 30–36. adenomatous polyps. New Engl J Med 1990; 322: 904–908. 222 Clementz BA, McDowell JE, Zisook S. Saccadic system function- 201 Lalouel JM, Le Mignon L, Simon M, Fauchet R, Bourel M, Rao DC ing among schizophrenia patients and their first-degree relatives. et al. Genetic analysis of idiopathic hemochromatosis using J Abnorm Psychol 1994; 103: 277–287. both qualitative (disease status) and quantitative (serum iron) 223 Malone SM, Iacono WG. Error rate on the antisaccade task: information. Am J Hum Genet 1985; 37: 700–718. heritability and developmental change in performance among 202 Greenberg DA, Delgado-Escueta AV, Widelitz H, Sparkes RS, preadolescent and late-adolescent female twin youth. Psycho- Treiman L, Maldonado HM et al. Juvenile myoclonic epilepsy physiology 2002; 39: 664–673. (JME) may be linked to the BF and HLA loci on human 224 Levy DL, O’Driscoll G, Matthysse S, Cook SR, Holzman PS, chromosome 6. Am J Med Genet 1988; 31: 185–192. Mendell NR. Antisaccade performance in biological relatives of 203 Holzman PS, Solomon CM, Levin S, Waternaux CS. Pursuit eye schizophrenia patients: a meta-analysis. Schizophr Res 2004; 71: movement dysfunctions in schizophrenia. Family evidence for 113–125. specificity. Arch Gen Psychiatry 1984; 41: 136–139. 225 Louchart-de la Chapelle S, Nkam I, Houy E, Belmont A, Me´nard 204 Holzman PS. Behavioral markers of schizophrenia useful for J-F, Roussignol A-C et al. A concordance study of three genetic studies. J Psychiatry Res 1992; 26: 427–445. electrophysiological measures in schizophrenia. Am J Psychiatry 205 Sponheim SR, Iacono WG, Thuras PD, Beiser M. Using biological 2005; 162: 466–474. indices to classify schizophrenia and other psychotic patients. 226 Price GW, Michie PT, Johnston J, Innes-Brown H, Kent A, Clissa P Schizophr Res 2001; 50: 139–150. et al. A multivariate electrophysiological endophenotype, from a 206 Braff DL, Geyer MA, Light GA, Sprock J, Perry W, Cadenhead KS unitary cohort, shows greater research utility than any single et al. Impact of prepulse characteristics on the detection of feature in the Western Australian Family Study of Schizophrenia. sensorimotor gating deficits in schizophrenia. Schizophr Res Biol Psychiatry, Pre-publication version available online 20 2001; 49: 171–178. December 2005, http:///www.sciencedirect.com/science/article/ 207 Joober R, Zarate J-M, Rouleau G-A, Skamene E, Boksa P. B6T4S-4HVDJK7-3/2/da823e5eacd44ad891541fe0c0c3acfe. Provisional mapping of quantitative trait loci modulating the 227 Marcelis M, Suckling J, Woodruff P, Hofman P, Bullmore E, van acoustic startle response and prepulse inhibition of acoustic Os J. Searching for a structural endophenotype in psychosis startle. Neuropsychopharmacology 2002; 27: 765–781. using computational morphometry. Psychiatry Res Neuroim 208 Kumari V, Das M, Zachariah E, Ettinger U, Sharma T. Reduced 2003; 122: 153–167. prepulse inhibition in unaffected siblings of schizophrenia 228 McDonald C, Bullmore ET, Sham PC, Chitnis X, Wickham H, patients. Psychophysiology 2005; 42: 588–594. Bramon E et al. Association of genetic risks for schizophrenia and 209 Adler LE, Freedman R, Ross RG, Olincy A, Waldo MC. bipolar disorder with specific and generic brain structural Elementary phenotypes in the neurobiological and genetic study endophenotypes. Arch Gen Psychiatry 2004; 61: 974–984. of schizophrenia. Biol Psychiatry 1999; 46: 8–18. 229 Faraone SV, Seidman LJ, Kremen WS, Kennedy D, Makris N, 210 Myles-Worsley M, Coon H, McDowell J, Brenner C, Hoff M, Lind Caviness VS et al. Structural brain abnormalities among relatives B et al. Linkage of a composite inhibitory phenotype to a of patients with schizophrenia: implications for linkage studies. chromosome 22q locus in eight Utah families. Am J Med Genet Schizophr Res 2003; 60: 125–140. Neuropsychiatry Genet 1999; 88: 544–550. 230 Davatzikos C, Shen D, Gur RC, Wu X, Liu D, Fan Y et al. Whole- 211 Freedman R, Adams CE, Adler LE, Bickford PC, Gault J, Harris JG brain morphometric study of schizophrenia revealing a spatially et al. Inhibitory neurophysiological deficit as a phenotype for complex set of focal abnormalities. Arch Gen Psychiatry 2005; 62: genetic investigation of schizophrenia. Am J Med Genet C Semin 1218–1227. Med Genet 2000; 97: 58–64. 231 Davidson LL, Heinrichs RW. Quantification of frontal and 212 Bramon E, McDonald C, Croft RJ, Landau S, Filbey F, Gruzelier JH temporal lobe brain-imaging findings in schizophrenia: a meta- et al. Is the P300 wave an endophenotype for schizophrenia? A analysis. Psychiatry Res Neuroim 2003; 122: 69–87. meta-analysis and a family study. NeuroImage 2005; 27: 960–968. 232 Garver DL, Nair TR, Christensen JD, Holcomb J, Ramberg J, 213 McCarley RW, Niznikiewicz MA, Salisbury DF, Nestor PG, Kingsbury S. Atrophic and static (neurodevelopmental) O’Donnell BF, Hirayasu Y et al. Cognitive dysfunction in schizophrenic psychoses: premorbid functioning, symptoms

Molecular Psychiatry Subtyping schizophrenia A Jablensky 834 and neuroleptic response. Neuropsychopharmacology 1999; 21: 252 Phillips WA, Silverstein SM. Convergence of biological and 82–92. psychological perspectives on cognitive coordination in schizo- 233 Nuechterlein KH, Asarnow RF, Subotnik KL, Fogelson DL, phrenia. Behav Brain Sci 2003; 26: 65–138. Ventura J, Torquato RD et al. Neurocognitive vulnerability factors 253 Heinrichs RW. The primacy of cognition in schizophrenia. Am for schizophrenia: convergence across genetic risk studies and Psychol 2005; 60: 229–242. longitudinal trait-state studies. In: Lenzenweger MF and Dworkin 254 Kremen WS, Buka SL, Seidman LJ, Goldstein JM, Koren D, RH (eds). Origins and Development of Schizophrenia: Advances Tsuang MT. IQ decline during childhood and adult psychotic in Experimental Psychopathology. American Psychological symptoms in a community sample: a 19-year longitudinal study. Association: Washington, DC, 1998, pp 299–327. Am J Psychiatry 1998; 155: 672–677. 234 Egan MF, Goldberg TE, Gscheidle T, Weirich M, Bigelow LB, 255 Mohamed S, Paulsen JS, O’Leary D, Arndt S, Andreasen N. Weinberger DR. Relative risk of attention deficits in siblings Generalized cognitive deficits in schizophrenia. A study of first- of patients with schizophrenia. Am J Psychiatry 2000; 157: episode patients. Arch Gen Psychiatry 1999; 56: 749–754. 1309–1316. 256 Bilder RM, Goldman RS, Robinson D, Reiter G, Bell L, Bates JA 235 Chen WJ, Chang C-H, Liu SK, Hwang TJ, Hwu H-G. Sustained et al. Neuropsychology of first-episode schizophrenia: initial attention deficits in nonpsychotic relatives of schizophrenic characterization and clinical correlates. Am J Psychiatry 2000; patients: a recurrence risk ratio analysis. Biol Psychiatry 2004; 55: 157: 549–559. 995–1000. 257 Johnstone EC, Ebmeier KP, Miller P, Owens DGC, Lawrie SM. 236 Paulsen JS, Heaton RK, Sadek JR, Perry W, Delis DC, Braff D Predicting schizophrenia: findings from the Edinburgh High-Risk et al. The nature of learning and memory impairments in Study. Br J Psychiatry 2005; 186: 18–25. schizophrenia. J Int Neuropsychol Soc 1995; 1: 88–99. 258 Brewer WJ, Francey SM, Wood SJ, Jackson HJ, Pantelis C, Phillips 237 Turetsky BI, Moberg PJ, Mozley LH, Moelter ST, Agrin RN, Gur LJ et al. Memory impairments identified in people at ultra-high RC et al. Memory-delineated subtypes of schizophrenia: relation- risk for psychosis who later develop first-episode psychosis. Am ship to clinical, neuroanatonical, and neurophysiological J Psychiatry 2005; 162: 71–78. measures. Neuropsychology 2002; 16: 481–490. 259 Torrey EF. Studies of individuals never treated with antipsycho- 238 Goldberg TE, Egan MF, Gscheidle T, Coppola R, Weickert T, tic medications: a review. Schizophr Res 2002; 58: 101–115. Kolachana BS et al. Executive subprocesses in working memory. 260 Hoff AL, Sakuma M, Wieneke M, Horon R, Kushner M, DeLisi LE. Relationship to catechol-O-methyltransferase Val158Met geno- Longitudinal neuropsychological follow-up study of patients type and schizophrenia. Arch Gen Psychiatry 2003; 60: 889–896. with first-episode schizophrenia. Am J Psychiatry 1999; 156: 239 Kremen WS, Seidman LJ, Faraone SV, Toomey R, Tsuang MT. 1336–1341. Heterogeneity of schizophrenia: a study of individual neuropsy- 261 Heaton RK, Gladsjo JA, Palmer BW, Kuck J, Marcotte TD, Jeste chological profiles. Schizophr Res 2004; 71: 307–321. DV. Stability and course of neuropsychological deficits in 240 Glahn DC, Therman S, Manninen M, Huttunen M, Kaprio J, schizophrenia. Arch Gen Psychiatry 2001; 58: 24–32. Lo¨nnqvist J et al. Spatial working memory as an endophenotype 262 Kurtz MM. Neurocognitive impairment across the lifespan in for schizophrenia. Biol Psychiatry 2003; 53: 624–626. schizophrenia: an update. Schizophr Res 2005; 74: 15–26. 241 Dickinson D, Iannone VN, Wilk CM, Gold JM. General and 263 Heinrichs RW, Ruttan L, Zakzanis KK, Case D. Parsing schizo- specific cognitive deficits in schizophrenia. Biol Psychiatry 2004; phrenia with neurocognitive tests: evidence of stability and 55: 826–833. validity. Brain Cogn 1997; 35: 207–224. 242 Bruder GE, Wexler BE, Sage MM, Gil RB, Gorman JM. Verbal 264 Caspi A, Reichenberg A, Weiser M, Rabinowitz J, Kaplan Z, memory in schizophrenia: additional evidence of subtypes Knobler H et al. Cognitive performance in schizophrenia patients having different cognitive deficits. Schizophr Res 2004; 68: assessed before and following the first psychotic episode. 137–147. Schizophr Res 2003; 65: 87–94. 243 Hallmayer JF, Kalaydjieva L, Badcock J, Dragovic M, Howell S, 265 Kelly C, Sharkey V, Morrison G, Allardyce J, McCreadie RG. Michie PT et al. Genetic evidence for a distinct subtype of Nithsdale Schizophrenia Surveys 20. Cognitive function in a schizophrenia characterized by pervasive cognitive deficit. Am J catchment-area-based population of patients with schizophrenia. Hum Genet 2005; 77: 468–476. Br J Psychiatry 2000; 177: 348–353. 244 Griffiths TD, Sigmundsson T, Takei N, Rowe D, Murray RM. 266 Green MF. What are the functional consequences of neurocog- Neurological abnormalities in familial and sporadic schizophre- nitive deficits in schizophrenia? Am J Psychiatry 1996; 153: nia. Brain 1998; 121: 191–203. 321–330. 245 Kinney DK, Yurgelun-Todd DA, Woods BT. Neurologic signs of 267 Gold JM, Goldberg RW, McNary SW, Dixon LB, Lehman AF. cerebellar and cortical sensory dysfunction in schizophrenics Cognitive correlates of job tenure among patients with severe and their relatives. Schizophr Res 1999; 35: 99–104. mental illness. Am J Psychiatry 2002; 159: 1395–1402. 246 Gourion D, Goldberger C, Bourdel M-C, Bayle FJ, Millet B, 268 Racenstein JM, Harrow M, Reed R, Martin E, Herbener E, Penn Olie J-P et al. Neurological soft-signs and minor physical DL. The relationship between positive symptoms and instru- anomalies in schizophrenia: differential transmission within mental work functioning in schizophrenia: a 10 year follow-up families. Schizophr Res 2003; 63: 181–187. study. Schizophr Res 2002; 56: 95–103. 247 Keshavan MS, Sanders RD, Sweeney JA, Diwadkar VA, Goldstein 269 Rempfer MV, Hamera EK, Brown CE, Cromwell RL. The relations G, Pettegrew JW et al. Diagnostic specificity and neuroanatomical between cognition and the independent living skill of shopping validity of neurological abnormalities in first-episode psychoses. in people with schizophrenia. Psychiatry Res 2003; 117: 103–112. Am J Psychiatry 2003; 160: 1298–1304. 270 Revheim N, Schecter I, Kim D, Silipo G, Allingham B, Butler P 248 Dragovic M, Hammond G, Badcock JC, Jablensky A. Laterality et al. Neurocognitive and symptom correlates of daily problem- phenotypes in patients with schizophrenia, their siblings and solving skills in schizophrenia. Schizophr Res 2006; 83: 237–245. controls: associations with clinical and cognitive variables. 271 O’Leary DS, Flaum M, Kesler ML, Flashman LA, Arndt S, Br J Psychiatry 2005; 187: 221–228. Andreasen NC. Cognitive correlates of the negative, disorganized, 249 Clementz BA, Iacono WG, Ficken J, Beiser M. A family study of and psychotic symptom dimensions of schizophrenia. J Neurop- nailfold plexus visibility in psychotic disorders. Biol Psychiatry sychiatry Clin Neurosci 2000; 12: 4–15. 1992b; 31: 378–390. 272 Hill SK, Ragland JD, Gur RC, Gur RE. Neuropsychological 250 McGrath J, El-Saadi O, Grim V, Cardy S, Chapple B, Chant D et al. differences among empirically derived clinical subtypes of Minor physical anomalies and quantitative measures of the head schizophrenia. Neuropsychology 2001; 15: 492–501. and face in patients with psychosis. Arch Gen Psychiatry 2002; 273 Brazo P, Marie´ RM, Halbecq I, Benali K, Segard L, Delamillieure P 59: 458–464. et al. Cognitive patterns in subtypes of schizophrenia. 251 Goldberg TE, David A, Gold JM. Neurocognitive deficits in Eur Psychiatry 2002; 17: 155–162. schizophrenia. In: Hirsch SR and Weinberger DR (eds) 274 Goldstein G, Shemansky WJ, Allen DN. Cognitive function in Schizophrenia, 2nd edn. Blackwell Publishing: Cambridge, schizoaffective disorder and clinical subtypes of schizophrenia. Massachusetts, 2003, pp. 168–184. Arch Clin Neuropsychol 2005; 20: 153–159.

Molecular Psychiatry Subtyping schizophrenia A Jablensky 835 275 David AS, Malmberg A, Brandt L, Allebeck P, Lewis G. IQ and and schizophrenia: meta-analysis of case-control and family- risk for schizophrenia: a population-based cohort study. Psychol based studies. Am J Psychiatry 2003; 160: 469–476. Med 1997; 27: 1311–1323. 298 Bray NJ, Buckland PR, Williams NM, Williams HJ, Norton N, 276 Reichenberg A, Weiser M, Rapp MA, Rabinowitz J, Caspi A, Owen MJ et al. A haplotype implicated in schizophrenia Schmeidler J et al. Elaboration on premorbid intellectual susceptibility is associated with reduced COMT expression in performance in schizophrenia. Arch Gen Psychiatry 2005; 62: human brain. Am J Hum Genet 2003; 73: 152–161. 1297–1304. 299 Galderisi S, Maj M, Kirkpatrick B, Piccardi P, Mucci A, Invernizzi 277 Sitskoorn MM, Aleman A, Ebisch SJH, Appels MCM, Kahn RS. G et al. COMT Val158 Met and BDNF C270T polymorphisms Cognitive deficits in relatives of patients with schizophrenia: a in schizophrenia: a case-control study. Schizophr Res 2005; 73: meta-analysis. Schizophr Res 2004; 71: 285–295. 27–30. 278 Szo¨ke A, Schu¨ rhoff F, Mathieu F, Meary A, Ionescu S, Leboyer M. 300 Munafo MR, Bowes L, Clark TG, Flint J. Lack of association of the Tests of executive functions in first-degree relatives of COMT (Val 158/108Met) gene and schizophrenia: a meta-analysis of schizophrenic patients: a meta-analysis. Psychol Med 2005; 35: case-control studies. Mol Psychiatry 2005; 10: 765–770. 771–782. 301 Fan J-B, Zhang C-S, Gu N-F, Li X-W, Sun W-W, Wang H-Y et al. 279 Thompson JL, Watson JR, Steinhauer SR, Goldstein G, Pogue- Catechol-O-methyltransferase gene Val/Met functional poly- Geile MF. Indicators of genetic liability to schizophrenia: a morphism and risk of schizophrenia: a large-scale association sibling study of neuropsychological performance. Schizophr Bull study plus meta-analysis. Biol Psychiatry 2005; 57: 139–144. 2005; 31: 85–96. 302 Paunio T, Tuulio-Henriksson A, Hiekkalinna T, Perola M, Varilo 280 Cannon TD, Huttunen MO, Lo¨nnqvist J, Tuullo-Henriksson A, T, Partonen T et al. Search for cognitive trait components of Pirkola T, Glahn D et al. The inheritance of neuropsychological schizophrenia reveals a locus for verbal learning and memory on dysfunction in twins discordant for schizophrenia. Am J Hum 4q and for visual working memory on 2q. Hum Mol Genet 2004; Genet 2000; 67: 369–382. 13: 1693–1702. 281 Heinrichs RW, Zakzanis KK. Neurocognitive deficit in schizo- 303 Straub RE, MacLean CJ, O’Neill FA, Burke J, Murphy B, Duke F phrenia: a quantitative review of the evidence. Neuropsychology et al. A potential vulnerability locus for schizophrenia on 1998; 12: 426–445. chromosome 6p24–22: evidence for genetic heterogeneity. 282 Chapman LJ, Chapman JP. Strategies for resolving the here- Nat Genet 1995; 11: 287–293. togeneity of schizophrenics and their relatives using cognitive 304 Posthuma D, Luciano M, de Geus EJC, Wright MJ, Slagboom PE, measures. J Abnorm Psychol 1989; 98: 357–366. Montgomery GW et al. A genomewide scan for intelligence 283 Goldstein G. Neuropsychological heterogeneity in schizophrenia: identifies quantitative trait loci on 2q and 6p. Am J Hum Genet a consideration of abstraction and problem-solving abilities. Arch 2005; 77: 318–326. Clin Neuropsychol 1990; 5: 251–264. 305 Wernicke C. Grundriss der Psychiatrie in klinischen Vorlesungen. 284 Elliott R, Sahakian BJ. The neuropsychology of schizophrenia: Thieme: Leipzig, 1900. relations with clinical and neurobiological dimensions. Psychol 306 Stransky E. Auffassung gewisser Symptome der Dementia Med 1995; 25: 581–594. praecox. Neurol Centralblatt 1904; 23: 1137–1143 [English 285 Jæger J, Czobor P, Berns SM. Basic neuropsychological translation In: Cutting J, Shepherd M (eds). The Clinical Roots dimensions in schizophrenia. Schizophr Res 2003; 65: 105–116. of the Schizophrenia Concept. Cambridge University Press: 286 Badcock JC, Dragovic M, Waters FA, Jablensky A. Dimensions of Cambridge, 1985, pp 37–41.]. intelligence in schizophrenia: evidence from patients with 307 Meehl PE. Toward an integrated theory of schizotaxia, schizoty- preserved, deteriorated and compromised intellect. J Psychiatry py, and schizophrenia. J Personal Disord 1990; 4: 1–99. Res 2005; 39: 11–19. 308 Andreasen NC, Paradiso S, O’Leary DS. ‘Cognitive dysmetria’ as 287 Nuechterlein KH, Barch DM, Gold JM, Goldberg TE, Green MF, an integrated theory of schizophrenia: a dysfunction in cortical- Heaton RK. Identification of separable cognitive factors in subcortical-cerebellar circuitry? Schizophr Bull 1988; 24: 203–218. schizophrenia. Schizophr Res 2004; 72: 29–39. 309 Catani M, ffytche DH. The rises and falls of disconnection 288 Wilk CM, Gold JM, McMahon RP, Humber K, Iannone VN, syndromes. Brain 2005; 128: 2224–2239. Buchanan RW. No, it is not possible to be schizophrenic 310 Stephan KE, Baldeweg T, Friston KJ. Synaptic plasticity yet neuropsychologically normal. Neuropsychology 2005; 19: and dysconnection in schizophrenia. Biol Psychiatry 2006; 59: 778–786. 929–939. 289 Cirillo MA, Seidman LJ. Verbal declarative memory dysfunction 311 Craddock N, O’Donovan MC, Owen MJ. Genes for schizophrenia in schizophrenia: from clinical assessment to genetics and brain and bipolar disorder? Implications for psychiatric nosology. mechanisms. Neuropsychol Rev 2003; 13: 43–77. Schizophr Bull 2006; 32: 9–16. 290 Lee J, Park S. Working memory impairments in schizophrenia: a 312 Weiser M, van Os J, Davidson M. Time for a shift in focus in meta-analysis. J Abnorm Psychol 2005; 114: 599–611. schizophrenia: from narrow phenotypes to broad endopheno- 291 Seaton BE, Goldstein G, Allen DN. Sources of heterogeneity in types. Br J Psychiatry 2005; 187: 203–205. schizophrenia: the role of neuropsychological functioning. 313 Heinrichs RW. In Search of Madness. Schizophrenia and Neuropsychol Rev 2001; 11: 45–67. Neuroscience. Oxford University Press: Oxford, England, 2001. 292 Hoti F, Tuulio-Henriksson A, Haukka J, Partonen T, Holmstro¨mL, 314 Wilcox MA, Faraone SV, Su J, Van Eedewegh P, Tsuang MT. Lo¨nnqvist J. Family-based clusters of cognitive test performance Genome scan of three quantitative traits in schizophrenia in familial schizophrenia. BMC Psychiatry 2004; 4: 20 (doi: pedigrees. Biol Psychiatry 2002; 52: 847–854. 10.1186/1471-244X-4-20). 315 Jaspers K. General Psychopathology. Manchester University 293 Lander ES. Splitting schizophrenia. Nature 1988; 336: 105–106. Press: Manchester, 1963. 294 Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, 316 Moldin SO. Indicators of liability to schizophrenia: perspectives Straub RE et al. Effect of COMT Val108/158 Met genotype on frontal from genetic epidemiology. Schizophr Bull 1994; 20: 169–184. lobe function and risk for schizophrenia. Proc Natl Acad Sci USA 317 Allison DB, Thiel B, St. Jean P, Elston RC, Infante MC, Schork NJ. 2001; 98: 6917–6922. Multiple phenotype modeling in gene-mapping studies of 295 Bilder RM, Volavka J, Czobor P, Malhotra AK, Kennedy JL, Ni X quantitative traits: power advantages. Am J Hum Genet 1998; et al. Neurocognitive correlates of the COMT Val158 Met 63: 1190–1201. polymorphism in chronic schizophrenia. Biol Psychiatry 2002; 318 Boomsma DI, Dolan CV. A comparison of power to detect a QTL 52: 701–707. in sib-pair data using multivariate phenotypes, mean 296 Shifman S, Bronstein M, Sternfeld M, Pisante´-Shalom A, phenotypes, and factor scores. Behav Genet 1998; 28: 329–340. Lev-Lehman E, Weizman A et al. A highly significant association 319 Kovac I, Rouillard E, Me´rette C, Palmour R. Exploring the impact between a COMT haplotype and schizophrenia. Am J Hum Genet of extended phenotype in stratified samples. Genet Epidemiol 2002; 71: 1296–1302. 1999; 17(Suppl 1): S211–S216. 297 Glatt SJ, Faraone SV, Tsuang MT. Association between a 320 Marlow AJ, Fisher SE, Francks C, MacPhie IL, Cherny SS, functional catechol O-methyltransferase gene polymorphism Richardson AJ et al. Use of multivariate linkage analysis for

Molecular Psychiatry Subtyping schizophrenia A Jablensky 836 dissection of a complex cognitive trait. Am J Hum Genet 2003; 72: Toward the DSM-IV. The Guilford Press: New York, London, 561–570. 1986, pp 363–400. 321 Thornton-Wells TA, Moore JH, Haines JL. Genetics, statistics and 324 Skinner HA. Construct validation approach to psychiatric human disease: analytical retooling for complexity. Trends Genet classification. In: Millon T, Klerman GL (eds). Contemporary 2004; 20: 640–647. Directions in Psychopathology. Toward the DSM-IV. The Guilford 322 Zandi PP, DePaulo JR. Resolving the heterogeneity of psychiatric Press: New York, London, 1986, pp 307–330. disorders: clinical and statistical approaches. Clin Neurosci Res 325 Corning WC, Steffy RA. Taximetric strategies applied to psychia- 2005; 5: 45–51. tric classification. Schizophr Bull 1979; 5: 294–305. 323 Blashfield RK. Structural approaches to classification. In: Millon 326 Tamminga CA, Holcomb HH. Phenotype of schizophrenia: a T, Klerman GL (eds). Contemporary Directions in Psychopathology. review and formulation. Mol Psychiatry 2004; 10: 27–39.

Molecular Psychiatry