Subtyping Schizophrenia According to Outcome or Severity: A Search for Homogeneous Subgroups

Transcription

1 Subtyping Schizophrenia According to Outcome or Severity: A Search for Homogeneous Subgroups Abstract by MarC'Andre Roy, Chanted Merette, and Michel There is a growing consensus that current definitions of schizophrenia (SZ) include different disorders, or else different dimensions underlain by different pathophysiologies. This article reviews the evidence for the validity of three novel strategies to subtype SZ according to outcome or severity (deficit vs. nondeficit, Kraepelinian vs. non-kraepelinian, congenital vs. adult-onset). Medline and bibliographies were used to locate articles. The methodology of the studies was reviewed, and their results were grouped according to seven validating criteria. Several differences were found between subtypes, particularly for the deficit/nondeficit subtypes. However, for most of these differences, replications have yet to be undertaken. Important indicators of etiology from the environmental risk factors and genetic domains have received very little attention. These three subtyping strategies represent promising attempts to address the etiologic heterogeneity of SZ. However, one cannot conclude whether these strategies identify etiologically distinct SZ subgroups. We propose ten methodological and conceptual recommendations for future studies aimed at the identification of valid SZ subtypes according to outcome or severity. Keywords: Schizophrenia, genetics, etiology, subtypes, deficit syndrome, congenital schizophrenia, Kraepelinian, nosology. Schizophrenia Bulletin, 27(1): ,2001. SZ probably includes different diseases or reflects several dimensions of psychopathology, each underlain by its own pathophysiology (Tsuang et al. 1990; Andreasen and Carpenter 1993). This etiologic heterogeneity is one of the most likely explanations for the slow progress in the identification of susceptibility genes and consistent biological markers for SZ. In contrast, the identification of liability genes and pathophysiology has progressed more rapidly for other complex disorders such as breast cancer and Alzheimer dementias because epidemiological studies have identified valid subtypes. Therefore, etiologically homogeneous subtypes and/or valid dimensional phenotypes would facilitate the identification of the causes of SZ (Cloninger 1994; Tsuang 1994). Distinguishing SZ subjects according to outcome or severity may be one strategy for identifying such subtypes, as suggested by the following findings: (1) there is substantial variability in outcome and severity among SZ subjects (Carpenter and Kirkpatrick 1988), which is a prerequisite to using any variable to define subgroups; (2) many predictors of poor social functioning, such as poor premorbid adjustment and severe negative symptoms, correlate with each other (Roy and DeVriendt 1994), suggesting that these aspects may characterize a specific subgroup of very poor outcome (VPO) subjects; and (3) a few studies found that a family history positive for SZ predicted poor outcome and that a family history positive for affective disorders predicted good outcome in SZ probands (McGlashan 1986). This last finding suggests the implication of different familial/genetic factors in VPO vs. non-vpo SZ, although conflicting results and methodological limitations preclude considering these results definitive. In agreement with the above considerations, three recent strategies to subtype SZ according to outcome or severity (deficit, Kraepelinian, and congenital) have been proposed. First, deficit SZ (Carpenter et al. 1988; Kirkpatrick et al. 1989) is defined by the presence of at least two of six negative symptoms for at least 12 months. These symptoms should not be secondary to anxiety, depression, psychosis, drug side effects, mental retardation, or other causes (see Carpenter et al for a detailed discussion of the potential causes of secondary negative symptoms). SZ patients who do not meet these criteria are considered nondeficit. Send reprint requests to Dr. M.-A. Roy, Centre de Recherche, University Laval Robert-Giffard, 2601 de la Canardiere, Beauport, P. Quebec, Canada, G1J 2G3; marc-andre.roy@psa.ulaval.ca. 115

2 Schizophrenia Bulletin, Vol. 27, No. 1, 2001 M.-A. Roy et al. Second, Kraepelinian SZ (Keefe et al. 1987) is defined by (1) either continuous hospitalization, or if living outside the hospital, complete dependence on others for basic needs; (2) the lack of useful employment; and (3) the absence of a complete remission of symptoms. Patients who do not meet these criteria are considered non-kraepelinian. Third, congenital SZ (Murray et al. 1991) is defined by an insidious onset, occurring before age 25, preceded by poor premorbid adjustment. After onset, the course is chronic, with at least some symptoms persisting between episodes, and affective symptoms are typically absent. Optional features include a poor response to neuroleptics, the presence of cognitive impairment, and a history of obstetric complications. Conversely, adult-onset SZ is defined by an onset occurring abruptly, at any age throughout life, without significant premorbid impairment. The course is then characterized by acute psychotic episodes with very few symptoms, if any, between episodes. Optional features include a good response to neuroleptics, an absence of cognitive impairment, and no history of obstetric complications. Congenital SZ is posited to result from aberrant brain development during fetal and neonatal life, because of genetic factors or early environmental brain insults, and adult-onset SZ is posited to be genetically related to affective disorders. These three subtyping strategies share the assumption that subgroups of SZ differing on various aspects of outcome or severity may also differ on crucial etiologic aspects. However, these three strategies do not use the same aspects of outcome or severity to define subtypes. First, deficit SZ emphasizes diminished motivation and affectivity and is the only one of these three strategies that explicitly focuses on a single domain of psychopathology to define subgroups of patients. Second, Kraepelinian SZ emphasizes severe and persisting symptoms and self-care deficits. Third, congenital SZ is defined by its course and natural history (i.e., age and type of onset, premorbid adjustment, and course). Demonstrating the validity of any VPO SZ subtypes would dramatically change the design of genetic and biological studies of SZ, because such studies would then focus on more homogeneous subgroups within SZ. The goal of the present paper is to review the studies testing the validity of these three subtyping strategies. Methods Fifty-one articles comparing SZ subgroups defined according to any of these three subtyping strategies were identified through Medline searches, bibliographies of published articles, and contacts with some of the developers of these subtypes. These reports were grouped according to seven validators of psychiatric syndromes on which SZ subgroups can be compared (Robins and Guze 1970; see table 4). Within each of these seven validators, we assessed the consistency of findings across reports. While assessing the seven validators, the following methodological issues were taken into account, as they were in a previous paper on familial versus sporadic SZ (Roy and Crowe 1994): (1) sample size, which affects statistical power; (2) verification of the reliability of the variables used; (3) blindness of the assessment of the dependent variables to subtype assignment; (4) use of an appropriate control group, when needed to determine which subgroup differed from normality; (5) control for pertinent confounding factors; (6) sampling strategy (e.g., observation during acute relapse vs. stabilized stage); and (7) other methodological factors specific to each type of study (e.g., methods used to diagnose relatives in a family study). Results This section reviews the evidence for the validity of these three subtyping strategies according to our seven validators. Table 1 summarizes studies on deficit SZ; table 2, studies on Kraepelinian SZ; and table 3, studies on congenital SZ. Table 4 summarizes the main results for these three strategies. Reliability of the Subtyping Strategies. The Maryland Psychiatric Research Center (MPRC) group (Kirkpatrick et al. 1989) and other investigators (Harris et al. 1991; Fenton and McGlashan 1992; Ribeyre et al. 1994a; Fenton et al. 1997) achieved very good interrater reliability in distinguishing between deficit and nondeficit SZ with the Schedule for the Deficit Syndrome (SDS; Kirkpatrick et al. 1989) or based on a detailed review of extensive clinical information. In addition, the MPRC group has documented very good reliability with another research group (Amador et al. 1999). On the other hand, although the DSM-IV Field Trial did not assess the reliability of the deficit versus nondeficit subtypes, it has yielded poor reliability of a critical step in the identification of these subtypes: the distinction between primary and secondary negative symptoms (Flaum and Andreasen 1995). Three characteristics of the Field Trial may explain this poorer reliability (Flaum et al. 1997): (1) information from relatives, clinicians, and medical records is often critical for the primary/secondary distinction, but the collection of such additional information was not emphasized in the Field Trial; (2) the Field Trial raters did not receive detailed 116

3 I Table 1. Studies comparing deficit and nondeficit schizophrenia on 5' TO Statistically significant differences Information used to subtype Dependent variables Sample size Authors Population studied No difference in month of birth 53 deficit 148 nondeficit Environmental risk factors: season of birth DSM-IV patients SDS Dollfus et al. (1999)»' Higher cell density in deficit than in nondeficit and in normal controls 3 deficit 6 nondeficit 9 controls SDS Postmortem: Density of interstitial (information cells of the white matter in the inferior from relatives) parietal cortex DSM-III-R patients whose brains were in a brain collection Kirkpatrick et al. (1999) Better response of positive symptoms and better improvement in social functioning with clozapine vs. haloperidol in both deficit and nondeficit SZ; no differences between deficit and nondeficit SZ 21 deficit 53 nondeficit Response to treatment: changes in positive, negative, and extrapyramidal symptoms, and social functioning, during a randomized comparison of clozapine and haloperidol DSM-III-R SZ or SDS schizoaffective treatment-resistant MPRC outpatients Buchanan et al. (1998) Longer right parietal latency in deficit and nondeficit SZ vs. controls; longer left temporal and frontal latencies in nondeficit SZ 21 deficit 43 nondeficit 48 controls Auditory evoked potentials: P300 in frontal, parietal, and temporal areas DSM-III-R SZ inpatients Baseline, and outpatients followup research interviews Turestky et al. (1998) vs. controls; no comparison of deficit and nondeficit SZ Poorer performance in deficit vs. nondeficit SZ and controls on both tasks; poorer performance in nondeficit SZ vs. controls on the Span of Apprehension task Lesser education in deficit SZ 20 deficit 56 nondeficit 27 controls Neuropsychology: attention (continuous performance and span of apprehension tests) SDS DSM-///-RSZMPRC outpatients Buchanan et al. (1997) Clinical: thought disorder, positive symptoms Natural history: age of onset, education Neuropsychology: visual information processing Same as above SDS i 1 B' Bustillo et al. (1997) s»*^ 17 deficit 28 nondeficit 25 controls Clinical: positive and negative > Slower reaction time in deficit vs. nondeficit SZ and controls, and in nondeficit SZ vs. controls; abnormal asymmetry in nondeficit vs. deficit SZ and controls symptoms Natural history: age of onset Vol More severe negative, less severe positive symptoms in deficit SZ O More improvement in social skills and negative symptoms in nondeficit SZ 3 deficit 3 nondeficit Response to treatment: changes in social skills and negative symptoms after social skills training SDS to Kopelowicz et al. DSM-IV SZ outpatients (1997) 8

4 Table 1. Studies comparing deficit and nondeficit schizophrenia Continued Authors Population studied Information used to subtype Dependent variables Sample size Statistically significant differences Spalleta et al. DSM-III-R SZ acutely SDS (1997) psychotic inpatients Clinical: positive, disorganized, negative, depressive symptoms Extrapyramidal symptoms Natural history: age of onset, education 18 deficit 40 nondeficit More severe negative symptoms in deficit SZ Lesser education in deficit SZ Waltrip et al. DSM-III-R SZ MPRC SDS (1997) outpatients Dollfus et al. DSM-III-R, Carpenter, SDS (1996) Schneider, or Langfeldt SZ Kirkpatrick et al. SZ subjects from DSM- DS/W-/V field (1996a, 1996ft, IVField Trial trial: PDS 1998) SZ MPRC outpatients MPRC: SDS Kirkpatrick et al. (1996c, 1998) Loas et al. (1996) Ross et al. (1996,997) DSM-III-R first episode SZ, schizoaffective, or schizophreniform cases DSM-III-R SZ inpatients and outpatients DSM-///-RSZMPRC outpatients PDS PDS SDS Environmental risk factors: presence of anti-borna disease antibodies Genetic: SZ or other disorders in relatives Clinical: positive and disorganized symptoms, suspiciousness, and depressive episodes Natural history: age of onset, premorbid and current functioning, drug and alcohol use Environmental risk factors: season of birth Natural history: social functioning at admission and followup, longitudinal stability of PDS Clinical: suspiciousness, positive and negative symptoms Environmental risk factors: season of birth Clinical: self-rated physical anhedonia and depressive symptoms 15 deficit 49 nondeficit 20 controls 22 deficit 49 nondeficit Field trial: 51 deficit, 101 nondeficit MPRC: 25 deficit, 66 nondeficit 32 deficit 56 nondeficit 29 deficit 121 nondeficit Eye tracking: position root mean 12 deficit square error during the pursuit of a 40 nondeficit sinusoidal target; latency to pursuit 25 controls onset, open-loop acceleration, velocity More frequent anti-borna disease antibodies in deficit and nondeficit SZ vs. controls, and in deficit vs. nondeficit SZ More frequent family history positive for any psychiatric disorder in nondeficit SZ More severe suspiciousness and trend for more frequent depressive episodes in nondeficit SZ More severe alcohol and drug use in nondeficit SZ Poorer current and premorbid adjustment in deficit SZ More frequent summer birth in deficit SZ Lesser severity of suspiciousness and greater severity of negative symptoms in deficit SZ Poorer social functioning in deficit SZ at admission and followup; increasing number of deficit SZ patients during followup More frequent summer birth in deficit SZ More severe anhedonia in deficit SZ; more severe depressive symptoms in nondeficit SZ More frequent abnormal tracking in deficit vs. nondeficit SZ, in deficit SZ vs. controls, and in nondeficit SZ vs. controls; lesser I I S' O I

5 after target onset Presence of tardive dyskinesia velocity in deficit SZ vs. nondeficit SZ and controls; lesser open-loop acceleration in deficit and nondeficit SZ vs. controls Bustillo (1995) Bustillo (1995) Nibuya (1995) etal. etal. etal. Buchanan et al. (1994) Fenton and al. DSM-III-R SZ outpatients DSM-III-R SZ outpatients DSM-III-R SZ inpatients MPRC MPRC long-term DS/W-///-flSZMPRC outpatients SZ subjects from McGlashan Chestnut Lodge (1994); Fenton et al. (1994, 1997) Gur et al. (1994); DSM-III-R SZ cases Mozley et al. from University of (1994);Turetsky Pennsylvania etal. (1995) SDS SDS SDS SDS Extrapyramidal symptoms Response to treatment: changes in negative symptoms during drug washout 20 deficit 38 nondeficit 7 deficit 17 nondeficit Clinical: symptom dimensions from the 20 deficit BPRS Natural history: age of onset; years in a job CT scan: ventricular brain-ratio Neuropsychological: IQ Biochemical: phva Neuropsychology: executive functioning; visuospatial, memory tasks; global IQ Natural history: education, age of onset Clinical: positive symptoms Medical records Natural history: premorbid, current and research functioning; age, type of onset; interview (mean response to life events followup = 19 Clinical: positive, disorganized, years) negative symptoms; suicidal behavior and ideation Spontaneous dyskinetic movements SZ in relatives, obstetric complications Baseline and MRI: brain, cranial, temporal, and follow-up frontal volumes; CSF volume: research ventricular, sulcal, temporal, and interview frontal areas 20 nondeficit 18 deficit 21 nondeficit 30 controls 46 deficit 141 nondeficit 23 deficit 57 nondeficit 81 controls No differences in severity of extrapyramidal symptoms No changes in severity of negative symptoms during drug washout in patients whose psychosis did not worsen More severe thinking disturbance, negative symptoms in deficit SZ Fewer years with a job in deficit SZ Larger ventricle-to-brain ratio in deficit SZ Lower IQ in deficit SZ Higher phva in deficit SZ Lower IQ, poorer performance on Stroop, Trail Making, and Mooney Faces Tests in deficit vs. nondeficit SZ; poorer performance in deficit and nondeficit SZ vs. controls on other tests More severe positive symptoms in nondeficit SZ Poorer premorbid and current functioning, more insidious onset and lesser response to life events in deficit SZ More severe negative and disorganized symptoms, less severe delusions, and less frequent suicide in deficit SZ Increased risk for spontaneous dyskinesia in deficit SZ Larger left temporal CSF volume in deficit vs. nondeficit SZ and controls More severe negative symptoms e S 5era on o f I" 5' Co S' o!

6 Table 1. Studies comparing deficit and nondeficit schizophrenia Continued Authors Population studied Information used to subtype Dependent variables Sample size Statistically significant differences Natural history: age of onset Clinical: positive, disorganized, negative symptoms Gender ratio in deficit SZ Greater proportion of males in deficit SZ Kirkpatrick et al. (1994) DSM-///-RSZMPRC outpatients SDS Clinical: clinician and self-rated depressive symptoms 24 deficit 76 nondeficit More severe clinician and selfrated depressive symptoms in nondeficit SZ Mayerhoff et al. First episode RDC SZ Research (1994) cases interview Ribeyre et al. DSM-III-R SZ inpatients SDS (1994b); Thibaut and outpatients etal. (1998) Natural history: premorbid, current 11 deficit functioning; age at onset 36 nondeficit Clinical: positive, negative, and disorganized symptoms Biochemical phva and MHPG 14 deficit Clinical: positive, negative symptoms 20 nondeficit Natural history: age of onset Earlier onset, poorer premorbid and current functioning in deficit SZ More severe formal thought disorder in deficit SZ Lower phva and higher MHPG levels in deficit SZ More severe negative symptoms in deficit SZ Buchanan et al. DSM-III-R SZ MPRC (1993) outpatients SDS Tamminga et al. DSM-III SZ MPRC drug- SDS (1992) free inpatients MRI: prefrontal, caudate, 17 deficit amygdala/hippocampus, and cranial 24 nondeficit volumes 30 controls Natural history: age of onset, education PET scan: glucose use in temporal, 4 deficit occipital, parietal, frontal lobes; 7 nondeficit cingulum; hippocampus; caudate; 11 controls pallidum; substantia nigra; thalamus Smaller prefrontal volume in nondeficit vs. deficit SZ and controls; larger left caudate and smaller amygdala-hippocampus in deficit and nondeficit SZ vs. controls Lesser education in deficit SZ Trends for lower thalamic, frontal, and parietal metabolism in deficit SZ vs. nondeficit SZ and normal controls Harris et al. DSM-III-R SZ stable SDS (1991) outpatients age a 46 Buchanan et al. DSM-III-R SZ MPRC SDS (1990) outpatients Clinical: positive, negative, depressive 17 deficit symptoms 20 nondeficit Neuropsychology: IQ; Halstead-Reitan global scores Natural history: age of onset, education Natural history: premorbid adjustment, 17 deficit age of onset 17 nondeficit More severe negative symptoms in deficit SZ More severe impairment on the Halstead Impairment Index and lower IQ in deficit SZ Poorer premorbid adjustment in deficit SZ I 3- I- fio O

7 Neurological soft signs: sequence, motor, and sensory signs and sensory integration More sensory signs, more impaired audiovisual integration, graphesthesia, face-hand test, and left-right confusion in deficit SZ Kirkpatrick and DSM-IIISZ MPRC Buchanan (1990) outpatients Thaker et al. Same as above (1989) Carpenter et al. Same as above (1988) Wagman et al. (1987) Same as above SDS Clinical: physical and social 16 deficit anhedonia, magical ideation, 42 nondeficit perceptual aberration, impulsive nonconformity Information Eye tracking: qualitative rating; 15 deficit from clinicians, volitional and reflex saccadic 27 nondeficit medical latencies; distractibility score; saccade 14 controls records, direct size interviews Tardive dyskinesia Same as above Natural history: stability of negative 15 deficit symptoms, age of onset, education, 64 nondeficit current functioning Clinical: positive, negative, depressive symptoms Same as above Neuropsychology: motor speed, visuo- 15 deficit motor, tactile-motor, constructional 15 nondeficit 15 controls skills; spatial memory; executive functioning; IQ Clinical: positive, negative symptoms Natural history: current functioning, age of onset, education More severe physical and social anhedonia in deficit SZ Longer latency for volitional saccade and larger size of the saccade in deficit vs. nondeficit SZ Greater stability of negative symptoms in deficit SZ More severe negative symptoms in deficit SZ Greater proportion of males in deficit SZ Poorer psychomotor performance in deficit and nondeficit SZ vs. controls; poorer general performance in deficit SZ vs. controls Poorer current functioning in deficit SZ More severe negative symptoms in deficit SZ Note. BPRS = Brief Psychiatric Rating Scale; CSF = cerebrospinal fluid; CT = computed tomography; MHPG = 3-methoxy-4-hydroxyphenytethyleneglycol; MPRC = Maryland Psychiatric Research Center; MRI = magnetic resonance imaging; PDS = Proxy Deficit Syndrome; PET = positron emission tomography; phva = plasma homovanillic acid; RDC = Research Diagnostic Criteria; SDS = Schedule for the Deficit Syndrome; SZ = schizophrenia. For studies published in more than one report, the order is determined by the year of the first report. 00 c o S I I 1

8 Schiz a- ~ Ci Co tin, to Table 2. Studies comparing Kraepeiinian and non-kraepelinian schizophrenia Authors Population studied Information used to subtype Dependent variables Sample size Statistically significant differences Davis et al. (1998) Keefe et al. (1996) Harvey et al. (1991) Keefe et al. (1987, 1991) Keefe et al. (1987, 1991) Male DSM-III-R SZ inpatients or outpatients Male DSM-III-R SZ or schizoaffective inpatients or outpatients RDC SZ or schizoaffective (mainly SZ subtype) or Feighner SZ inpatients Same as above Same as above Research interviews, medical records, and information from relatives CASH, medical records and information from relatives Research interviews, medical records, and information from relatives Same as above Same as above CT: changes in VBR over a 5- year period ;, Clinical: positive, negative symptoms; thought disorder; bizarre behavior; frequency of schizoaffective diagnosis Natural history: premorbid functioning, age of onset 22 Kraepeiinian 31 non-kraepelinian 13 controls 61 Kraepeiinian 80 non-kraepelinian Response to treatment: response! 8 Kraepeiinian to a 6-week discontinuation of 16 non-kraepelinian neuroleptic treatment, followed by a 4-week trial of flexible dose of haloperidol Response to neuroleptic treatment: 6-week flexible-dose trial of haloperidol < Clinical: positive, negative symptoms; thought disorder; presence of an affective syndrome Natural history: premorbid and current functioning, age of onset Genetic: frequency of SZ spectrum disorders in first and second degree relatives (FHRDC) 21 Kraepeiinian 54 acutely psychotic non-kraepelinian 26 Kraepeiinian 101 non-kraepelinian (15 remitted, 86 acutely psychotic) Greater increase in VBR in Kraepeiinian vs. non- Kraepelinian SZ and controls (no significant differences between groups at baseline) More severe negative symptoms and thought disorder in Kraepeiinian SZ; more frequent schizoaffective diagnosis in non- Kraepelinian SZ More severe increase in severity of symptoms during drug washout, and more important decrease after reintroduction of haloperidol in non- Kraepelinian SZ Better response to treatment in non- Kraepelinian SZ More severe thought disorder and negative symptoms, and less frequent affective syndrome in Kraepeiinian SZ Poorer current functioning in Kraepeiinian SZ Higher frequency of a positive family history for SZ spectrum disorder, and higher morbid risk for SZ in relatives of Kraepeiinian SZ. f 8 ^ p* *w *r A O

9 Keefe et al. Same as above Same as above CT scan: VBR and asymmetry of 15 Kraepelinian Higher left-right ventricle j? (1987,1991) lateral ventricles 15 non-kraepelinian ratio in Kraepelinian SZ.f Note. CASH = Comprehensive Assessment of Symptoms and History; CT = computed tomography; FHRDC = Family History Research Diagnostic Criteria; RDC = Research 5' Diagnostic Criteria; SZ = schizophrenia; VBR = ventricle-to-brain ratio. *, f o Table 3. Studies comparing congenital and adult-onset schizophrenia Information used Statistically significant Authors Population studied to subtype Dependent variables Sample size differences Corrigall and Maudsley twin sample Case summaries Genetic: frequency of congenital 18 congenital psychotic Trends for more Murray (1994) and adult-onset psychoses and of 34 adult-onset psychotic frequent adult-onset nonpsychotic depression in co-twin 4 nonpsychotic depression psychosis and o, nonpsychotic depression jj; in co-twin of adult-onset > psychotic probands, and j " for more frequent 2. congenital psychoses in t» co-twin of congenital a: psychotic proband

10 Table 4. Summary of results Validator Deficit vs. nondeficit SZ Kraepelinian vs. non-kraepelinian SZ Good to very good (RF) Good to very good (RF) Reliability Clinical symptoms Hallucinations and delusions Disorganized symptoms Negative symptoms Depressive symptoms Natural history Premorbid adjustment Age of onset Type of onset Social functioning Temporal stability Gender ratio Response to treatment Social skills training Neuroleptics Biological markers Structural imaging Functional imaging Postmortem brain study Neuropsychology Eye tracking Auditory evoked potentials Biochemical markers Neurological soft signs Spontaneous dyskinesias Tardive dyskinesias Twin and family studies Environmental risk factors Equal or lesser severity in nondeficit (RF) More severe in deficit SZ (IF) More severe in deficit SZ (RF) More severe in nondeficit SZ (RF) Poorer in deficit SZ (RF) No differences (RF) More insidious in deficit SZ (UF) Poorer in deficit SZ (RF) High degree (RF) Possible association between male gender and deficit SZ (RF) Poorer response in deficit SZ (UF) Equal response (UF) Increased temporal spinal fluid volume in deficit SZ (UF) Greater degree of ventricular enlargement in deficit SZ (UF) Smaller brain frontal volume in nondeficit SZ (UF) Decreased thalamic, frontal, and parietal metabolism in deficit SZ (UF) Higher density of interstitial cells of the white matter of the inferior parietal lobe in deficit SZ (UF) Poorer performance in deficit SZ in several domains (UF) Abnormal asymmetry of visual processing in nondeficit (UF) Poorer tracking in deficit SZ (UF) No differences (UF) Higher or lower level of phva in deficit SZ (OF) Lower level of 3 MHPG in deficit SZ (UF) More frequent in deficit SZ (UF) More frequent in deficit SZ (UF) No differences in frequency (RF) More frequent Borna virus antibodies in deficit SZ (UF) More frequent summer birth in deficit SZ (RF) No differences during stabilized stage (RF) More severe in Kraepelinian (RF) More severe in Kraepelinian (RF) No differences (RF) No differences (RF) Poorer in Kraepelinian (RF) High degree (UF) Poorer response in Kraepelinian SZ (RF) Abnormal lateral ventricles asymmetry and higher degree of progressive ventricular enlargement in Kraepelinian SZ (UF) Higher risk for SZ in Kraepelinian Note. IF = inconsistent finding (a minority of studies reporting this finding, with a predominance of negative results); MHPG = 3-methoxy-4-hydroxyphenylethyleneglycol; OF = opposite finding (a single study with at least one other study reporting opposite results); phva = plasma homovanillic acid; RF = replicated finding (at least two studies that present concordant results; if there are some inconsistent results, they can be easily accounted for by methodological factors); SZ = schizophrenia; UF = unreplicated finding (a single study without any attempt at replicating results or with a single negative study). We did not include a separate column on congenital and adult-onset SZ because its validity was tested in a single study. This study revealed that co-twins of probands with congenital psychoses had a higher risk for congenital psychoses, and that co-twins of probands with adult-onset psychoses had a higher risk for adult-onset psychoses and nonpsychotic depression. In addition, that study revealed very good interrater reliability of these subtypes. These results have not been replicated. 3 S' Co c O 8

11 Subtyping Schizophrenia Schizophrenia Bulletin, Vol. 27, No. 1, 2001 training on the primary/secondary distinction of negative symptoms; and (3) 93.4 percent of the patients were undergoing a psychotic relapse at the time of the assessment, and the primary/secondary distinction is easier to perform during a stabilized stage between psychotic episodes (Spalletta et al. 1997). To overcome these difficulties, which are often encountered in large-scale studies, the Proxy Deficit Syndrome (PDS) was developed (Kirkpatrick et al. 1993). The PDS classifies SZ subjects into deficit or nondeficit subgroups based on a specific profile on the Brief Psychiatric Rating Scale (BPRS): that is, low scores on anxiety, guilt feelings, depressive mood, and hostility, and high scores on affective blunting, which were found to predict a deficit syndrome on the SDS. The developers of the Kraepelinian versus non- Kraepelinian dichotomy have documented very good interrater reliability for this distinction (kappa = 0.96) (Keefe et al. 1996), which was confirmed by the DSM-IV Field Trial (Amador, personal communication cited in Keefe et al. 1996). The developers of the congenital versus adult-onset SZ achieved very good interrater reliability (Corrigall and Murray 1994) (kappa = 0.80), but this reliability was not tested by other groups. Clinical Differences Outside the Subtype Definition. Finding that subtypes differ according to symptoms not included in their definition may add support to their validity. Congenital and adult-onset SZ could not be compared on such symptoms, because most of them were included in their definition. Hallucinations and delusions. Nine studies found no differences in the severity of hallucinations and delusions between deficit and nondeficit SZ (Wagman et al. 1987; Carpenter et al. 1988; Harris et al. 1991; Fenton and McGlashan 1994; Mayerhoff et al. 1994; Nibuya et al. 1995; Buchanan et al. 1997; Spalletta et al. 1997; Thibaut et al. 1998). Conversely, five studies found lesser severity of these symptoms in deficit SZ (Buchanan et al. 1994; Turetsky et al. 1995; Kirkpatrick et al. 1996ft, 1996c; Bustillo et al. 1997). Although we found no methodological features that could explain the discrepant conclusions of these two groups of studies, the studies unequivocally support the finding that people with deficit SZ do not have more severe psychotic symptoms than people with nondeficit SZ. Non-Kraepelinian SZ subjects, most of whom were undergoing an acute psychotic relapse, were found not to differ from Kraepelinian SZ subjects on the severity of psychotic symptoms in two studies (Keefe et al. 1987, 1991, 1996). However, the severity of psychotic symptoms has not been compared in these subgroups during the stabilized stage. Disorganized dimension. Three studies (Fenton and McGlashan 1994; Mayerhoff et al. 1994; Nibuya et al. 1995) found more severe disorganized symptoms (i.e., thought disorder and bizarre behavior) in deficit than in nondeficit SZ. However, these differences were not observed in four studies (Turetsky et al. 1995; Kirkpatrick et al. 1996/?; Buchanan et al. 1997; Spalletta et al. 1997). We found no methodological explanation for these discrepant results. Kraepelinian SZ subjects were found to have more severe formal thought disorder than non-kraepelinian SZ subjects in two independent samples (Keefe et al. 1987, 1991, 1996). Negative symptoms. Three noteworthy refinements can be found on the severity of negative symptoms in deficit versus nondeficit SZ beyond the obvious observation of more severe negative symptoms in deficit SZ: (1) one study (Spalletta et al. 1997) confirmed that the greater severity of negative symptoms in deficit SZ was also present during acute psychotic relapse; (2) two studies found that the greater severity of negative symptoms in deficit SZ observed by clinicians was concordant with patients' self-reports (Kirkpatrick and Buchanan 1990; Loas et al. 1996); and (3) one study found no change of negative symptoms in either deficit or nondeficit cases whose psychosis remained stable during a 2-week drug discontinuation (Bustillo et al. 1995). These findings suggest that the greater severity of negative symptoms in deficit SZ does not result from neuroleptic side effects, in agreement with two studies that did not find more severe extrapyramidal symptoms in deficit SZ (Bustillo et al. 1995; Spalletta et al. 1997). Finally, Kraepelinian SZ was found to display more severe negative symptoms than non-kraepelinian SZ in two independent studies (Keefe et al. 1987, 1996). Depressive symptoms. Three studies on stabilized outpatients found more severe depressive symptoms in nondeficit than in deficit SZ probands (Kirkpatrick et al. 1994, 1996ft; Loas et al. 1996), consistent with the observation of a higher suicide risk in nondeficit SZ probands (Fenton et al. 1997). On the other hand, three studies did not find such differences (Carpenter et al. 1988; Nibuya et al. 1995; Spalletta et al. 1997). Sampling differences may explain the negative results of one of the three latter studies (Spalletta et al. 1997) because it focused on acutely psychotic patients; depressive symptoms observed within such a context may have a different significance from those observed in stabilized outpatients. In addition, the studies finding differences between deficit and nondeficit SZ used more detailed assessments of depressive symptoms, which may have increased their sensitivity to differences between groups. Indeed, two of the studies finding differences used the Beck Depression Inventory (Kirkpatrick et al. 1994; Loas et al. 1996), and the third 125

12 Schizophrenia Bulletin, Vol. 27, No. 1, 2001 M.-A. Roy et al. sought diagnoses of previous depressive episodes (Kirkpatrick et al. 1996b) as opposed to the negative studies that relied on a cruder assessment of depressive symptoms (Carpenter et al. 1988; Nibuya et al. 1995; Spalletta et al. 1997), that is, the depressive symptoms assessed by the BPRS. Given these methodological issues, we conclude that there is strong evidence for more severe depressive symptoms in nondeficit SZ. Comparisons on depressive symptoms have not been reported for Kraepelinian versus non-kraepelinian SZ. However, a higher proportion of schizoaffective disorder was found in non-kraepelinian subjects (Keefe et al. 1987, 1996), which is in agreement with previous findings of schizoaffective disorder having a better outcome than SZ (Samson et al. 1988). Natural History. We examined whether the subtypes differed from each other on several natural history validators. Congenital and adult-onset SZ could not be compared on these aspects, as most of them were included in their definition. Premorbid adjustment. Poorer premorbid adjustment in deficit SZ can be considered well replicated because it was found in four studies (Buchanan et al. 1990; Fenton and McGlashan 1994; Mayerhoff et al. 1994; Kirkpatrick et al. 1996a) and was confirmed by the observation of a lesser level of education in deficit SZ (Buchanan et al. 1993, 1997; Spalletta et al. 1997), which is an indirect indicator of premorbid problems. Two studies found no differences in premorbid adjustment between Kraepelinian and non-kraepelinian SZ (Keefe et al. 1987, 1996). Age of onset. A single study (Mayerhoff et al. 1994) found that deficit SZ cases had an earlier onset than nondeficit SZ cases, but 14 studies found no differences (Wagman et al. 1987; Carpenter et al. 1988; Buchanan et al. 1990, 1993, 1994, 1997; Harris et al. 1991; Fenton and McGlashan 1994; Nibuya et al. 1995; Turetsky et al. 1995; Kirkpatrick et al. 1996b, 1996c; Bustillo et al. 1997; Spalletta et al. 1997). Therefore, a genuine difference in age of onset between deficit and nondeficit SZ appears unlikely. Two studies found no differences in age of onset between Kraepelinian and non-kraepelinian SZ (Keefe et al. 1987, 1996). Type of onset. The Chestnut Lodge Study yielded more insidious onset in deficit SZ (Fenton and McGlashan 1994). Social functioning. Poorer social functioning in deficit SZ was found in seven independent samples, using different sampling methods, different social functioning scales, and either concurrent or prospective assessment of social functioning (Wagman et al. 1987; Carpenter et al. 1988; Fenton and McGlashan 1994; Mayerhoff et al. 1994; Nibuya et al. 1995; Kirkpatrick et al. 1996a, 1996c). One study found less frequent substance abuse in deficit than in nondeficit SZ (Kirkpatrick et al. 1996a), and another study found that deficit SZ subjects were less reactive to difficult life events (Fenton and McGlashan 1994). Therefore, the association between deficit SZ and poorer functional outcome is well replicated, and this association seems to be an inherent consequence of the nature of the disorder rather than an artifact of a difficult living situation or substance abuse. Social functioning was found to be poorer in Kraepelinian SZ in two studies (Keefe et al. 1987, 1996), which is somewhat tautological because poor social functioning is the main criterion for Kraepelinian SZ. Temporal stability. One study found that the deficit syndrome became increasingly frequent during the first 5 years of illness and had a high degree of longitudinal stability once established (Fenton and McGlashan 1994). These results are consistent with those of two studies of first-episode SZ patients that found a low prevalence of the deficit syndrome in this population (Mayerhoff et al. 1994) and an increasing severity of the deficit syndrome over the first 2 years of illness (Kirkpatrick et al. 1996c). Moreover, the finding of a high degree of longitudinal stability has been replicated in a recent longitudinal prospective followup (mean duration = 3.8 years) study (Amador et al. 1999). A high degree of 1-year stability of the Kraepelinian/non-Kraepelinian dichotomy (intraclass correlation coefficient = 0.68) was observed in the DSM-IV Field Trial (Amador, personal communication cited in Keefe et al. 1996). Unfortunately, the methods and the results of this study have not been published in sufficient detail to assess the validity of these findings. Gender Ratio. The literature on gender differences in SZ suggests an association between male gender and VPO subtypes. Of 22 published papers examining gender ratio in deficit and nondeficit SZ (Carpenter et al. 1988; Buchanan et al. 1990, 1993, 1994, 1997; Kirkpatrick and Buchanan 1990; Harris et al. 1991; Tamminga et al. 1992; Kirkpatrick et al. 1993, 1994, 1996a, 1996b, 1996c; Fenton and McGlashan 1994; Turetsky et al. 1995, 1998; Dollfus et al. 1996; Ross et al. 1996; Bustillo et al. 1997; Spalletta et al. 1997; Waltrip et al. 1997; Thibaut et al. 1998), only 3 (Carpenter et al. 1988; Turetsky et al. 1995; Buchanan et al. 1997) found a statistically significant higher proportion of males among deficit SZ patients. On the other hand, all these studies yielded odds ratios for an association between male gender and deficit SZ that were equal to or greater than 1, suggesting a weak association between male gender and deficit SZ. To test this hypothesis, we have undertaken a meta-analysis (Roy et al., in press) that has confirmed a significant association 126

13 Subtyping Schizophrenia Schizophrenia Bulletin, Vol. 27, No. 1, 2001 between male gender and deficit SZ (Mantel-Haenszel pooled odds ratio = 1.75, p = ). Gender differences could not be examined in studies of Kraepelinian versus non-kraepelinian SZ because more than 90 percent of the subjects included were male. We did not find any comparison of gender ratio in congenital versus adult-onset SZ. Response to Treatment. Despite the modest specificity of psychiatric drugs, finding different response to treatment according to subtypes would add support to their validity. Congenital and adult-onset SZ could not be compared on this aspect because response to treatment is included in their definition. One study found no difference in response to either haloperidol or clozapine between deficit and nondeficit SZ (Buchanan et al. 1998). Another study found poorer response to social skills training in deficit versus nondeficit SZ (Kopelowicz et al. 1997), although the very small sample size (n = 6) made these results preliminary. Non-Kraepelinian SZ has yielded a more important increase in severity of symptoms during drug washout (Harvey et al. 1991) and a better response to haloperidol (Keefe et al. 1987, 1991; Harvey et al. 1991) than Kraepelinian SZ, suggesting that Kraepelinian SZ may be a form of treatment-resistant SZ. Biological Markers. We looked for differential patterns of association with biological markers in each subtype. We.found no comparison of congenital and adult-onset SZ for any of these aspects. Structural imaging. Three structural imaging studies compared deficit and nondeficit SZ. One study (Buchanan et al. 1993), using magnetic resonance imaging (MRI), yielded smaller prefrontal brain volumes in nondeficit SZ probands versus normal controls and deficit SZ probands, which was somewhat surprising and contradicted the a priori hypothesis that deficit SZ would be specifically characterized by changes in that area. In a second study using MRI, the only significant difference observed was an abnormal increase in temporal spinal fluid volume that was specific to the left side in deficit SZ patients (Gur et al. 1994; Mozley et al. 1994; Turetsky et al. 1995). A third study, using a computed tomography (CT) scan (Nibuya et al. 1995), found increased ventricular volume in deficit versus nondeficit SZ. Differences among these three studies in terms of imaging method (CT vs. MRI) and the structures measured prevented direct comparisons between them, and therefore, none of the findings can be considered replicated. Two CT scan studies have compared Kraepelinian and non-kraepelinian SZ. The first one yielded a greater degree of asymmetry of lateral ventricles (area of left ventricle divided by area of right ventricle) in Kraepelinian SZ (Keefe et al. 1987). However, these asymmetry findings are undermined by the use of area measurements that have been found sensitive to head positioning and to absolute volume of the ventricles (Zipursky et al. 1990). A second CT scan study (Davis et al. 1998) yielded a greater increase in ventricular volume in Kraepelinian SZ over a 5-year followup. However, in the latter study, the Kraepelinian SZ subjects were on average 5 years older than non-kraepelinian ones, which may partly account for the findings because the Kraepelinian subjects were entering the period that previous studies had found to be characterized by a greater increase in ventricular volume. Functional imaging. A positron emission tomography scan study (Tamminga et al. 1992) found lower thalamic, frontal, and parietal resting metabolism in deficit than in nondeficit SZ, but none of these differences achieved statistical significance, probably because of the small sample size, making these results preliminary. Another study (Gur et al. 1995) found no differences on resting brain metabolism between deficit and nondeficit SZ (total n - 42). The latter study was not included in table 1 because it did not focus on the comparisons between deficit and nondeficit SZ and because several important methodological aspects (e.g., the number of deficit versus nondeficit patients, the method used to assess the deficit syndrome) were not reported. Postmortem brain study. A single postmortem study (Kirkpatrick et al. 1999) has found a higher density of interstitial cells of the white matter of the inferior parietal lobe in deficit than in nondeficit SZ and normal controls. However, the very small sample size (9 people with SZ, 9 controls) warrants caution in the interpretation of these results. Neuropsychology. Compared to nondeficit SZ subjects and normal controls, deficit SZ subjects were found more impaired on (1) global intellectual functioning (Wagman et al. 1987; Harris et al. 1991; Buchanan et al. 1994; Nibuya et al. 1995); (2) executive functioning and visuospatial tasks (Buchanan et al. 1994); (3) attentional tasks (Buchanan et al. 1997); and (4) visuospatial information processing (Bustillo et al. 1997). Contrasting with the abundance of more abnormal findings in deficit SZ, the right-left asymmetry of visuospatial information processing (Bustillo et al. 1997) was the only task found more impaired in nondeficit SZ. Moreover, a detailed inspection of these studies revealed a systematic ordering of the performance: with very few exceptions, deficit SZ subjects had the poorest performance and normal controls the best, with nondeficit SZ subjects in between. Eye tracking. Two studies compared deficit and nondeficit SZ on eye tracking, but differences in measures used in these studies precluded comparing their results. First, during an antisaccade task (Thaker et al. 1989), 127

14 Schizophrenia Bulletin, Vol. 27, No. 1, 2001 M.-A. Roy et al. deficit SZ showed longer latency for volitional saccades and larger saccade size than nondeficit SZ. Second, regarding smooth pursuit movements, deficit SZ subjects showed lesser velocity (Ross et al. 1996) and qualitatively more impaired tracking (Ross et al. 1997) than nondeficit SZ subjects and controls. In the latter study, a systematic ordering of performance could be found, because on almost every measure, deficit SZ subjects had the poorest performance and normal controls the best, with nondeficit SZ subjects in between. Auditory evoked potentials. A single study compared auditory evoked potentials (P300) in deficit and nondeficit SZ subjects and controls (Turetsky et al. 1998). In that study, the electrode location (i.e., frontal, parietal, or temporal) that yielded statistically significant differences between SZ subgroups and controls differed, but no differences between deficit and nondeficit SZ were reported. Biochemical markers. One study found lower plasma homovanillic acid (phva, a metabolite of dopamine) level and higher 3-methoxy-4-hydroxyphenylethyleneglycol (MHPG, a metabolite of noradrenaline) levels (Ribeyre et al. 1994ft; Thibaut et al. 1998) in deficit versus nondeficit SZ. However, another study yielded opposite findings: that is, a higher level of phva in nondeficit SZ than in deficit SZ (Nibuya et al. 1995). Although it is difficult to reconcile such opposite findings, differences in sampling may partly explain these discrepancies, because the second study focused exclusively on long-term SZ patients, while the first study had a broader range of patients. Neurological soft signs. One study found that deficit SZ cases had more frequent abnormalities in sensory and audiovisual integration than nondeficit cases (Buchanan et al. 1990), suggesting more severe parietal dysfunction in deficit SZ. Spontaneous and tardive dyskinesias. One study revealed an increased risk for spontaneous dyskinesias in deficit versus nondeficit SZ (Fenton et al. 1994) among neuroleptic-naive SZ patients. While such results suggest that die deficit syndrome is a risk factor for tardive dyskinesia, two studies found no differences in risk for tardive dyskinesias in deficit versus nondeficit neuroleptic-treated SZ patients (Thaker et al. 1989; Ross et al. 1996). However, the modest sample size in the latter studies makes their negative results preliminary. Twin and Family Studies and Environmental Risk Factors. Given the role of genetic factors in the etiology of SZ, twin and family studies are essential to establishing validate subtypes. Therefore, we looked for evidence of different patterns of psychopathology in relatives according to probands' subtypes. We also looked for differential associations with environmental risk factors. Five published reports examined these aspects in deficit versus nondeficit SZ. A first study yielded no differences between deficit and nondeficit SZ on the frequency of obstetric complications and psychopatfiology in relatives (Fenton and McGlashan 1994). However, these aspects were rated from the records of the SZ probands, making these findings inconclusive. A second study (Dollfus et al. 1996) found that nondeficit SZ subjects were more likely to have a family history of any psychiatric disorder than deficit SZ subjects. Given the breadth of the disorders considered (see Roy and Crowe 1994 for a detailed discussion on this issue) and given that diagnoses were based exclusively on the family history (FH; i.e., information from relatives) method, these results should be considered preliminary. Indeed, FH has been shown to agree poorly with best estimate diagnoses (Roy et al. 1996) (i.e., diagnoses based on direct interview with the relatives themselves, FH, and medical records), which are considered more valid (Weissman et al. 1986; Maziade et al. 1992; Roy et al. 1997). A third study yielded an association between Borna virus antibodies and deficit SZ (Waltrip et al. 1997), which is particularly interesting, given the tropism of Borna viruses in rodents to brain structures likely to be involved in SZ (i.e., the hippocampus and dopaminergic tracts), altfiough the human pathogenic effects of this virus remain uncertain. A fourth report yielded an association between summer birth and deficit SZ in three independent samples (Kirkpatrick et al. 1998), although it was not replicated in a fourth sample from a different center (Dollfus et al. 1999). A qualitative comparison with months of birth in the general population in the first article suggested that the rate of summer birth in deficit SZ exceeded that of the general population, although there was no direct comparison of season of birth in SZ subgroups and population controls. Given the excess of summer birth in deficit SZ, it could be speculated that the excess of winter birth in SZ found in several studies may be specific to nondeficit SZ. Therefore, this association between summer birth and deficit SZ appears particularly promising and would be compatible, for example, with a pathogenic effect of the influenza virus in the deficit subgroup. First and second degree relatives of Kraepelinian and non-kraepelinian SZ cases have been compared on their risk for SZ, schizoaffective disorder, unspecified functional psychoses, and SZ spectrum personality disorders (affective disorders were not reported) (Keefe et al. 1987). Relatives of Kraepelinian SZ probands were found to have a higher morbid risk for SZ and a higher frequency for a positive family history of SZ spectrum disorders in any first or second degree relative. However, the diagnoses in relatives were based exclusively on the FH method, making these results preliminary, for the reasons discussed above. 128

15 Subtyping Schizophrenia Schizophrenia Bulletin, Vol. 27, No. 1, 2001 A reanalysis of the Maudsley twin sample (Corrigall and Murray 1994) found that co-twins of probands with congenital psychoses had a higher risk for congenital psychoses and that co-twins of probands with adult-onset psychoses had a higher risk for adult-onset psychoses and nonpsychotic depression. Although these results lent some support to a relationship between psychopathology in relatives and subtype of the probands, the evidence was weakened by the following limitations: (1) probably because of modest sample size, none of the comparisons reached statistical significance; (2) the study was not restricted to SZ probands, as some of the subjects suffered from psychotic affective or schizoaffective disorder (Farmer et al. 1987) and were probably classified in the adult-onset subgroup, which may have spuriously increased the evidence for a relationship between adultonset psychoses and affective disorders; and (3) the absence of a control group prevented the assessment of whether the risk for psychopathology in relatives of either subgroup of probands exceeded those of the general population. Discussion In this section, we summarize (table 4) and discuss the results for each subtyping strategy in the light of three current models of heterogeneity (McGuffin et al. 1987; Tsuang et al. 1990; Andreasen and Carpenter 1993; Buchanan and Carpenter 1997). Differences Between Deficit and Nondeficit SZ. The main findings were as follows. First, adequate reliability was achieved by several research groups, and one study found a high degree of concordance between the PDS and the SDS. Attempts at replicating the latter results are needed, although, as Kirkpatrick et al. (1998) argue, the similarities found between SDS- and PDS-defined deficit SZ suggest that these two instruments identify similar patients. Second, comparisons of the severity of the clinical symptoms outside the definition of the subtypes yielded equal or lesser severity of hallucinations and delusions in deficit than in nondeficit SZ and showed strong evidence for more severe depressive symptoms in nondeficit SZ. Third, natural history studies yielded poorer premorbid and current social functioning in deficit SZ, increasing frequency of the deficit syndrome during the first 5 years of illness, and stability of the deficit syndrome once established. Fourth, a meta-analysis yielded a statistically significant association between deficit SZ and male gender. Fifth, there was preliminary support for poorer response to social skills training in deficit SZ. Sixth, no replicated differences between deficit and nondeficit SZ were found on biological markers. However, deficit SZ showed more extreme abnormalities than nondeficit SZ on several biological markers in single, unreplicated studies. These aspects spanned structural and functional brain imaging, postmortem brain studies, neuropsychological and eye tracking performance, abnormalities in monoamine metabolites level (although some findings were conflicting), neurological soft signs, and spontaneous dyskinesias. Conversely, compared to deficit SZ, nondeficit SZ showed a greater decrease in prefrontal volume and abnormal lateral asymmetry of visual information processing. Seventh, an excess of summer birth in deficit SZ was found in three independent samples, and more frequent Borna virus antibodies in deficit SZ were found in a single study. However, there were no published family studies of these subtypes. (See addendum.) Heterogeneity Models and the Deficit Versus Nondeficit Subtypes. In this section, we describe three current models of heterogeneity (i.e., the distinct disorders, die continuum of severity, and the distinct domains models) and we examine the extent to which they were supported in published studies on deficit and nondeficit SZ. Distinguishing between these three models of heterogeneity is essential for interpreting the differences between SZ subtypes because these three models have very different implications for the design of studies on the pathophysiology of SZ. The pattern of differences between deficit and nondeficit subtypes may suggest that they represent distinct disorders (we termed this first pattern the "distinct disorders" model of heterogeneity). Indeed, deficit SZ appeared more severe than nondeficit SZ for several validators, and conversely, nondeficit SZ yielded more extreme findings for three validators (i.e., depressive symptoms, prefrontal atrophy, asymmetry of visual information processing). In addition, the association of a summer birth with the deficit syndrome would also fit with the distinct disorders model, as summer birth is not an indicator of severity. The distinct disorders model would suggest that deficit and nondeficit SZ represent mutually exclusive disorders with distinct pathophysiology and that investigations on the pathophysiology of SZ should study these subtypes independently. However, the three differences for which nondeficit SZ yielded more extreme findings should be considered as nondefinitive for three reasons. First, the greater severity of depressive symptoms in nondeficit SZ may be partially tautological. Indeed, in patients showing both depressive symptoms and negative symptoms, the latter are likely to be judged secondary to depressive symptoms. Consequently, these patients would be excluded from the deficit subgroup, which would 129

16 Schizophrenia Bulletin, Vol. 27, No. 1, 2001 M.-A. Roy et al. decrease the probability of patients with depressive symptoms being classified as deficit. Second, the greater degree of prefrontal atrophy in nondeficit SZ patients is difficult to reconcile with other lines of evidence suggesting a link between negative symptoms and frontal dysfunction. Therefore, the validity of this finding should be considered uncertain, although the relationship between prefrontal atrophy and nondeficit SZ may be valid and may illustrate that primary and secondary negative symptoms have different correlates. Replications of this finding are needed to determine the correct interpretation of the association between nondeficit SZ and prefrontal atrophy. Third, nondeficit SZ was found to be more severe than deficit SZ on only one neuropsychological variable (lateral asymmetry of visual processing), while deficit cases performed worse on several neuropsychological tasks. Therefore, this single finding of poorer neuropsychological performance in nondeficit SZ should be viewed cautiously until it is replicated, because a global view of neuropsychological differences between deficit and nondeficit SZ may suggest that this was a chance finding. If the three findings for which nondeficit SZ yielded more extreme abnormalities (i.e., depressive symptoms, prefrontal atrophy, asymmetry of visual information processing) were artifactual, the judgment on the validity of deficit and nondeficit SZ would change. Indeed, we would be left with a systematic pattern of more extreme abnormalities in deficit SZ. Such a pattern would rather suggest that deficit and nondeficit SZ represent variations on a continuum of severity of a single disorder (we termed this second pattern the "continuum of severity" model of heterogeneity [McGuffin et al. 1987]). In such a continuum of severity model, deficit and nondeficit SZ would have a similar etiology, the deficit subtype having a more severe loading for these etiologic factors. With this model, combining deficit and nondeficit patients in a case-control study would not diminish the power to identify the etiologic factors of SZ. In addition to the pattern of more extreme findings in deficit SZ, the systematic ordering of performance observed on neuropsychological and eye tracking tasks is also typical of a continuum of severity model. Indeed, with very few exceptions, deficit cases showed the worst performance, normal controls showed the best performance, and nondeficit cases displayed a performance between that of the other two groups. Finally, the deficit/nondeficit categories may identify a distinct psychopathological domain (we termed this third pattern the "distinct domains" model of heterogeneity) rather than distinct disorders. Other orthogonal psychopathological domains could be the positive and the disorganized ones, which can have their own pathophysiology, different from that of the deficit domain. As a consequence, deficit SZ would be expected to yield more extreme findings on the pathophysiology indicators of the deficit domain, but deficit and nondeficit SZ patients would not differ on the indicators of the other two domains. Taken in isolation, this pattern of findings is very similar to a continuum of severity model. Consequently, the distinction between the continuum of severity and the distinct domains models requires the simultaneous study of more than one putative domain. For example, if a neurobiological abnormality was found to be associated with the deficit syndrome and another neurobiological abnormality with persistent positive symptoms in the same cohort of patients, a strong point could be made for the distinct domains model. Unfortunately, studies on deficit versus nondeficit SZ have so far focused exclusively on the deficit syndrome, which does not allow a more definitive test of the distinct domains model for deficit SZ. (See addendum.) However, previous evidence (Andreasen et al. 1995) that positive, negative, and disorganized symptom dimensions have distinct correlates strongly suggests that the distinct domains model is a very plausible explanation for the differences found between deficit and nondeficit SZ. If the distinct domains model were right, the etiology of SZ would be easier to elucidate through the study of these distinct dimensions. However, contrary to the distinct disorders model, which posits distinct pathophysiologies in mutually exclusive subgroups of patients, the distinct domains model suggests that domains are not exclusive and can co-occur in a given patient. Therefore, a given group of patients could provide information on the pathophysiology of each of these domains. We are thus left with three plausible models of heterogeneity that could account for the differences observed between deficit and nondeficit SZ. The methodological recommendations made below may help to discriminate between these various models. Kraepelinian Versus Non-Kraepelinian SZ. The main findings were as follows. First, adequate reliability was achieved by the developers of this strategy and by one other group. Second, comparisons of the severity of clinical symptoms yielded equal severity of hallucinations and delusions in Kraepelinian versus non-kraepelinian SZ during psychotic relapse, and more severe negative symptoms and more severe thought disorder in Kraepelinian SZ. Third, there was limited evidence for differences between Kraepelinian and non-kraepelinian SZ on natural history validators. Fourth, we found no comparison on gender ratio. Fifth, there was replicated evidence for better response to neuroleptic treatment in non-kraepelinian SZ. Sixth, there was unreplicated preliminary evidence for abnormal asymmetry of lateral ventricles and for more 130

17 Subtyping Schizophrenia Schizophrenia Bulletin, Vol. 27, No. 1, 2001 severe progressive ventricular enlargement in Kraepelinian SZ. Seventh, there was preliminary evidence for a higher risk for SZ in relatives of Kraepelinian versus non-kraepelinian SZ. This pattern of differences between subtypes supports a continuum of severity model, because Kraepelinian subjects yielded more extreme findings across several validators, while non-kraepelinian SZ subjects were not found more severe on any of the validators. However, because few of these findings were replicated, and because there are several other validators on which Kraepelinian SZ and non-kraepelinian SZ have yet to be compared, this conclusion cannot be definitive. It should be stressed that if a continuum of severity model were appropriate to account for the differences between Kraepelinian and non-kraepelinian SZ, these subtypes could still have important implications for studies on the etiology and pathophysiology of SZ. Indeed, because Kraepelinian cases would have a more important loading on etiologic factors (Greenberg 1992), comparing Kraepelinian cases to controls could increase statistical power. In addition, the continuum of severity model would suggest that comparing Kraepelinian and non- Kraepelinian subtypes could yield important information on the determinants of outcome. Congenital Versus Adult-Onset SZ. The main findings were as follows. First, adequate reliability was achieved by the developers of this strategy but was not tested by other groups. Second, there were no comparisons on the severity of the clinical symptoms outside the definition of the subtypes, natural history validators, gender ratio, treatment response, and biological markers. Third, there was preliminary evidence for differential patterns of familial psychopathology in congenital versus adult-onset SZ. These differences between these subtypes fit a distinct disorders model because different types of psychopathology are found in relatives according to proband subtype. However, given the methodological limitations discussed above, these results are preliminary and warrant further studies of these subtypes. Overall Conclusions on the Three Subtyping Strategies None of the three subtyping strategies yielded replicated differences between subtypes across all validators. This shortage of replicated differences may be due only to the novelty of these subtypes; for most validators, available information stems from single, unreplicated studies. The amount of available information appears more or less sufficient depending on the perspective from which the subtypes are. judged. For example, the clinical relevance of the deficit and nondeficit subtypes has been unequivocally established by strong evidence for differences in outcome between these subgroups, even if the information is not sufficient to definitively establish the validity of these subtypes from an etiologic perspective. At this stage, these subtyping strategies can be regarded as promising strategies to identify homogeneous SZ subtypes, although additional studies are needed to further test their validity. The accumulated evidence for the validity of these subtyping strategies, particularly the deficit/nondeficit one, is certainly sufficient to warrant their use in future case-control or genetic studies of SZ, because the subtyping of patients is relatively inexpensive when compared to some of the high-tech procedures (e.g., functional brain imaging). The use of such subtypes in molecular genetic studies of SZ is particularly appealing, because such studies are likely to provide the most definitive evidence on the validity of any subtyping strategy. Priorities for Future Studies The following recommendations can be made for future studies. First, family studies of these subtypes are direly needed. Not a single published family study of these subtypes used blind best estimate diagnoses in relatives (see addendum), although there was preliminary evidence that non-vpo (i.e., adult-onset) SZ may represent a severe expression of genetic liability to affective disorders. Such familial co-aggregation of affective disorders and non- VPO SZ could reflect genetic anticipation (Meiette et al., 2000). However, an association between outcome and the type of psychopathology in relatives may reflect diagnostic misclassifications. Indeed, some non-vpo SZ cases may represent severe affective disorders misdiagnosed as SZ because such cases are often more difficult to diagnose than prototypical VPO SZ and nonpsychotic affective disorders (Roy et al. 1997). Hence, an increased risk for affective disorders in relatives of non-vpo SZ probands could be an artifact of the misclassification of psychotic affective disorders within that subgroup. Future studies should make every possible effort to exclude such psychotic affective disorder patients from the proband groups. Second, because the three subtyping strategies reviewed denned subtypes a priori, it is not known which of the criteria used in these various definitions are the most crucial. Instead, it may be advantageous to empirically derive definitions of VPO and non-vpo SZ, based on criteria that are recognized as essential for the validity of psychiatric syndromes (Robins and Guze 1970). For example, variables predicting VPO in probands and those predicting SZ in relatives could be used to define a VPO subtype, while variables predicting better outcome in probands and those predicting affective disorders in relatives could be used to define a non-vpo subtype. In addi- 131

18 Schizophrenia Bulletin, Vol. 27, No. 1, 2001 M.-A. Roy et al. tion, multivariate statistical classification techniques (i.e., latent class and cluster analyses) can be used to define subtypes based on patterns of similarities and differences (Goldstein et al. 1990; Castle et al. 1994; Sham et al. 1996). However, several complex issues are raised by the use of these techniques (Everitt and Merette 1990): (1) the definition of the classes or clusters may depend on the statistical technique used; (2) the definition of subgroups may depend on the choice of variables; and (3) it may be difficult to decide on the number of classes or clusters to retain, because several statistical criteria exist to guide this decision and these various criteria may lead to different conclusions. Third, future studies have to take into account dimensional models of heterogeneity by studying simultaneously the determinants of several dimensions of psychopathology or physiopathology (Andreasen et al. 1995). Indeed, as was discussed above, such models are an essential alternative to categorical models of heterogeneity (Andreasen and Carpenter 1993). In addition, recent studies (Van Os et al. 1996, 1999) have suggested that psychopathological dimensions yield better predictive power than diagnoses for several aspects of outcome in psychotic patients. Such dimensional models of heterogeneity can be used to test for possible continuities across current diagnostic categories (e.g., bipolar disorder and SZ) (Maziade et al. 1995; Toomey et al. 1997). Fourth, it is of interest to study the degree of overlap between the various strategies to subtype SZ according to outcome, in order to determine the extent to which results obtained with one approach to subtyping generalize to other approaches. Some degree of overlap may be suggested by the finding that poor outcome and greater severity of negative symptoms are common to Kraepelinian and deficit SZ. In addition, the overlap between VPO and treatment-resistant SZ needs to be studied because treatment response may also be pertinent to identify etiologically homogeneous subtypes (Brown and Herz 1989). Fifth, attention must be paid to sample size and power, because studies of heterogeneity are expected to require large samples to achieve satisfactory power. Some sampling procedure could be used to increase power. For example, it may be advantageous to sample SZ with outcome toward the ends of the severity spectrum. Such a sampling procedure would increase the proportion of subjects with extreme values on variables such as premorbid adjustment or response to treatment, thereby increasing the power to identify meaningful relationships. In addition, while age of onset did not appear to differentiate any of these subtypes, most studies of these subtypes probably did not include a sizable proportion of cases with very early onset (VEO; i.e., before 15 years of age) SZ, as is usual in samples of adult SZ cases. Available evidence that VEO SZ yields a particularly poor outcome (Maziade et al. 1996a, 1996b) would suggest that such cases are more likely to be from the deficit, the Kraepelinian, or the congenital subtype, providing an impetus for studying these subtypes in samples of VEO SZ. Another precaution that could increase the power to find differences between subtypes would be to exclude patients with an ambiguous classification from either subtype. This misclassification problem, which affects other subtyping strategies as well (Roy and Crowe 1994; Roy et al. 1995), could explain the systematic ordering of neuropsychological performance found in deficit SZ subjects, nondeficit SZ subjects, and normal controls. An example of an ambiguous classification that may reduce the homogeneity of the nondeficit subgroup is the difficulty in distinguishing between primary and secondary negative symptoms in the presence of severe and persistent disorganization symptoms. In case of doubt, the SDS recommends considering the negative symptoms as being secondary. As a consequence, some of these cases classified as nondeficit may in fact suffer from deficit SZ. Sixth, longitudinal studies suggest that functional outcome and deficit symptoms stabilize after the first 5 years of illness (Westermeyer and Harrow 1988; Fenton and McGlashan 1994; Kirkpatrick et al. 1994). Consequently, to ensure that poor outcome is not transient, to allow for symptoms such as the deficit ones to emerge, and because the risk for diagnostic unreliability decreases with longer followup (Roy et al. 1997), it appears preferable to use a sample of cases who are at least 5 years past onset. Alternatively, prospective followup studies of incident cases for at least 5 years are needed, because the prospective followup studies could allow more accurate measurements of several clinical variables. Seventh, attention should be paid to the effect of neuroleptic treatment on course of illness because insufficient treatment is clearly a factor that could result in a poor outcome. Therefore, it could be useful to require, for example, that people with VPO SZ have had at least two adequate trials with neuroleptics, to make sure they are not simply insufficiently treated patients. Moreover, to assess the deficit syndrome, investigators would ideally control the treatment, which could make easier the distinction between primary and secondary negative symptoms. Controlling the treatment could allow the use of the so-called atypical neuroleptics, which recent studies have shown to be superior to classical ones in terms of decreasing the severity of negative symptoms and in being less likely to cause parkinsonian syndromes (Marder et al. 1997; Tollefson and Sanger 1997). Eighth, sampling strategies with a well-defined sampling frame, such as the Suffolk study (Kirkpatrick et al. 1996), would help to assess the generalizability of findings and 132

19 Subtyping Schizophrenia Schizophrenia Bulletin, Vol. 27, No. 1, 2001 might prevent sampling biases that may occur in convenience samples such as those used in most of the studies reviewed. Ninth, the developers of these three subtyping strategies and other researchers (Brockington et al. 1992; Maziade et al. 1995) agree on the need to use longitudinal data to derive meaningful subtypes. Because the goal of these subtyping strategies is to find subtypes differing on crucial etiologic aspects, it is important to emphasize lifetime measurements that are less likely than cross-sectional measurements to be overly influenced by short-term fluctuations. Therefore, this emphasis on longitudinally based clinical assessment is warranted. Tenth, replications of previous findings should obviously be sought to establish the validity of those findings. Providing detailed information on methodological aspects and on the clinical characteristics of the subjects under investigation is of utmost importance to assess the potential sources of discrepancies among studies. The replication of the studies that found nondeficit SZ more severe on some validators would be particularly important, because such replications would significantly strengthen the case for deficit and nondeficit SZ being distinct disorders. References Amador, X.F.; Kirkpatrick, B.; Buchanan, R.W.; Carpenter, W.T.; Marcinko, L.; and Yale, S.A. Stability of the diagnosis of deficit syndrome in schizophrenia. American Journal of Psychiatry, 156: , Andreasen, N.C., and Carpenter, W.T., Jr. Diagnosis and classification of schizophrenia. Schizophrenia Bulletin, 19(2): , Andreasen, N.C.; Roy, M.-A.; and Flaum, M.A. Positive and negative symptoms of schizophrenia. In: Hirsch, S., and Weinberger, D. Handbook of Schizophrenia. Cambridge, MA: Basel/Blackwell Scientific Publications, pp. 28^5. Brockington, I.; Roper, A.; Edmunds, E.; Kaufman, C; and Meltzer, H.Y. A longitudinal psychopathological schedule. Psychological Medicine, 22: , Brown, W.A., and Herz, L.R. Response to neuroleptic drugs as a device for classifying schizophrenia. Schizophrenia Bulletin, 15(1): , Buchanan, R.W.; Breier, A.; Kirkpatrick, B.; Ball, P.; and Carpenter, W.T. Positive and negative symptom response to clozapine in schizophrenic patients with and without the deficit syndrome. American Journal of Psychiatry, 155: , Buchanan, R.W.; Breier, A.; Kirkpatrick, B.; Elkashef, A.; Munson, R.C.; Gellad, F.; and Carpenter, W.T., Jr. Structural abnormalities in deficit and nondeficit schizophrenia. American Journal of Psychiatry, 150:59-65, Buchanan, R.W., and Carpenter, W.T., Jr. The neuroanatomies of schizophrenia. Schizophrenia Bulletin, 23(3): , Buchanan, R.W.; Kirkpatrick, B.; Heinrichs, D.W.; and Carpenter, W.T, Jr. Clinical correlates of the deficit syndrome of schizophrenia. American Journal of Psychiatry, 147: , Buchanan, R.W.; Strauss, M.E.; Breier, A.; Kirkpatrick, B.; and Carpenter, W.T. Attentional impairments in deficit and nondeficit forms of schizophrenia. American Journal of Psychiatry, 15: ,1997. Buchanan, R.W.; Strauss, M.E.; Kirkpatrick, B.; Holsein, C; Breier, A.; and Carpenter, W.T, Jr. Neuropsychological impairments in deficit vs. nondeficit forms of schizophrenia. Archives of General Psychiatry, 51: ,1994. Bustillo, J.R.; Kirkpatrick, B.; and Buchanan, R.W. Neuroleptic treatment and negative symptoms in deficit and nondeficit schizophrenia. Biological Psychiatry, 38:64-67, Bustillo, J.R.; Thaker, G.; Buchanan, R.W.; Moran, M.; Kirkpatrick, B.; and Carpenter, W.T. Visual information processing impairments in deficit and nondeficit schizophrenia. American Journal of Psychiatry, 154: , Carpenter, W.T, Jr.; Heinrichs, D.W.; and Alphs, L.D. Treatment of negative symptoms. Schizophrenia Bulletin, H(3): , Carpenter, W.T, Jr.; Heinrichs, D.W.; and Wagman, A.M. Deficit and nondeficit forms of schizophrenia: The concept. American Journal of Psychiatry, 145: , Carpenter, W.T, Jr., and Kirkpatrick, B. The heterogeneity of the long-term course of schizophrenia. Schizophrenia Bulletin, 14(4): , Castle, D.J.; Sham, PC; Wessely, S.; and Murray, R.M. The subtyping of schizophrenia in men and women: A latent class analysis. Psychological Medicine, 24:41-51, Cloninger, C.R. Cloninger turning point in the design of linkage studies of schizophrenia. American Journal of Human Genetics (Neuropsychiatric Genetics), 54:83-92,1994. Corrigall, R.J., and Murray, R.M. Twin concordance for congenital and adult-onset psychosis: A preliminary study of the validity of a novel classification of schizophrenia. Ada Psychiatrica Scandinavica, 89: , Davis, K.L.; Buchsbaum, M.S.; Shihabuddin, L.; Spiegel-Cohen, J.; Metzger, M.; Frecska, E.; Keefe, R.S.; and Powchik, P. Ventricular enlargement in pooroutcome schizophrenia. Biological Psychiatry, 43: ,

20 Schizophrenia Bulletin, Vol. 27, No. 1, 2001 M.-A. Roy et al. Dollfus, S.; Brazo, P.; Langlois, S.; Gourevitch, R.; Dassa, D.; Besse, F; Vand Der Elst, R; Delamillieure, P.; Chabot, B.; Gelfi, J.D.; and Petit, M. Month of birth in deficit and non-deficit schizophrenic patients. European Psychiatry, (France) 14: , Dollfus, S.; Ribeyre, J.M.; and Petit, M. Family history and deficit form in schizophrenia. European Psychiatry, 16: , Everitt, B.S., and Merette, C. The clusteing of mixed model data: A comparison of possible approaches. Journal of Applied Statistics, 17: , Farmer, A.E.; McGuffin, P.; and Gottesman, I.I. Twin concordance for DSM-IH schizophrenia. Archives of General Psychiatry, 44: , Fenton, W.S., and McGlashan, T.H. Testing systems for assessment of negative symptoms in schizophrenia. Archives of General Psychiatry, 49: ,1992. Fenton, W.S., and McGlashan, T.H. Antecedents, symptom progression, and long-term outcome of deficit syndrome in schizophrenia. American Journal of Psychiatry, 151: , Fenton, W.S.; McGlashan, T.H.; Victor, B.J.; and Blyler, C.R. Symptoms, subtype, and suicidality in patients with schizophrenia spectrum disorders. American Journal of Psychiatry, 154: , Fenton, W.S.; Wyatt, R.J.; and McGlashan, T.H. Risk factors for spontaneous dyskinesia in schizophrenia. Archives of General Psychiatry, 51: , Flaum, M.; Amador, X.; Gorman, J.; Bracha, H.S.; Edell, W.; McGlashan, T.; Pandurangi, A.; Kendler, K.S.; Robinson, D.; Lieberman, J.; Ontiveros, A.; Tohen, M.; McGorry, P.; Tyrrell, G.; Arndt, S.; and Andreasen, N.C. The DSM-IVfield trial for schizophrenia. In: Widiger, T.A.; Frances, A.J.; Pincus, H.A.; Ross, R.; First, M.B.; and Wakefield-Davis, W., eds. DSM-IV Sourcebook: Volume IV. Washington, DC: American Psychiatric Association, pp Flaum, M., and Andreasen, N. The reliability of distinguishing primary versus secondary negative symptoms. Comprehensive Psychiatry, 36:42I^t27, Goldstein, J.M.; Santangelo, S.L.; Simpson, J.C.; and Tsuang, M.T. The role of gender in identifying subtypes in schizophrenia: A latent class analytic approach. Schizophrenia Bulletin, 16(2): , Greenberg, D.A. There is more than one way to collect data for linkage analysis: What a study of epilepsy can tell us about linkage strategy for psychiatric disease. Archives of General Psychiatry, 49: , Gur, R.E.; Mozley, P.D.; Resnick, S.M.; Mozley, L.H.; Shtasel, D.L.; Gallacher, F; Arnold, S.E.; Karp, J.S.; Alavi, A.; Reivich, M.; and Gur, R.C. Resting cerebral glucose metabolism in first-episode and previously treated patients with schizophrenia relates to clinical features. Archives of General Psychiatry, 52: , Gur, R.E.; Mozley, P.D.; Shtasel, D.L.; Cannon, T.D.; Gallacher, F; Turetsky, B.; Grossman, R.; and Gur, R.C. Clinical subtypes of schizophrenia: Differences in brain and CSF volume. American Journal of Psychiatry, 151: , Harris, M.J.; Jeste, D.V.; Krull, A.; Montague, J.; and Heaton, R.K. Deficit syndrome in older schizophrenic patients. Psychiatry Research, 39: , Harvey, P.D.; Putnam, K.M.; Davidson, M.; Kahn, R.S.; Powchik, P.; McQueeney, R.; Keefe, S.E.; and Davis, K.L. Brief neuroleptic discontinuation and clinical symptoms in Kraepelinian and non-kraepelinian chronic schizophrenic patients. Psychiatry Research, 38: , Keefe, R.S.E.; Frescka, E.; Apter, S.H.; Davidson, M.; Macaluso, J.M.; Hirschowitz, J.; and Davis, K.L. Clinical characteristics of Kraepelinian schizophrenia: Replication and extension of previous findings. American Journal of Psychiatry, 153: , Keefe, R.S.E.; Lobel, D.S.; Mohs, R.C; Silverman, J.M.; Harvey, P.D.; Davidson, M.; Losonczy, M.F; and Davis, K.L. Diagnosis issue in chronic schizophrenia: Kraepelinian schizophrenia, undifferentiated schizophrenia, and state-independent negative symptoms. Schizophrenia Research, 4:71-79, Keefe, R.S.E.; Mohs, R.C; Losonczy, M.F.; Davidson, M.; Silverman, J.M.; Kendler, K.S.; Horvath, T.B.; Nora, R.; and Davis, K.L. Characteristics of very poor outcome schizophrenia. American Journal of Psychiatry, 144: , Kirkpatrick, B.; Amador, X.; Flaum, M.; Yale, S.A.; Gorman, J.M.; Carpenter, W.T.; Tohen, M.; and McGlashan, T. The deficit syndrome in the DSM-IV Field Trial: I. Alcohol and other drug abuse. Schizophrenia Research, 20:69-77, 1996a. Kirkpatrick, B.; Amador, X.; Yale, S.A.; Bustillo, J.R.; Buchanan, R.W.; and Tohen, M. The deficit syndrome in the DSM-IV Field Trial: II. Depressive episodes and persecutory beliefs. Schizophrenia Research, 20:79-90, 1996ft. Kirkpatrick, B., and Buchanan, R.W. Anhedonia and the deficit syndrome of schizophrenia. Psychiatry Research, 31:25-30, Kirkpatrick, B.; Buchanan, R.W.; Breier, A.; and Carpenter, W.T. Case identification and stability of the deficit syndrome of schizophrenia. Psychiatry Research, 47:47-56,

21 Subtyping Schizophrenia Schizophrenia Bulletin, Vol. 27, No. 1, 2001 Kirkpatrick, B.; Buchanan, R.W.; Breier, A.; and Carpenter, W.T. Depressive symptoms and the deficit syndrome of schizophrenia. Journal of Nervous and Mental Disease, 182:452^55, Kirkpatrick, B.; Buchanan, R.W.; McKenney, P.D.; Alphs, L.D.; and Carpenter, W.T. The schedule for the deficit syndrome: An instrument for research in schizophrenia. Psychiatry Research, 30: , Kirkpatrick, B.; Conley, R.C.; Kakoyannis, A.; Reep, R.L.; and Roberts, R.C. Interstitial cells of the white matter in the inferior parietal cortex in schizophrenia: An unbiased cell-counting study. Synapse, 34:95-102, Kirkpatrick, B.; Ram, R.; Amador, X.F.; Buchanan, R.W.; McGlashan, T.; Tohen, M.; and Bromet, E. Summer birth and the deficit syndrome of schizophrenia. American Journal of Psychiatry, 155: , Kirkpatrick, B.; Ram, R.; and Bromet, E. The deficit syndrome in the Suffolk County Mental Health Project. Schizophrenia Research, 22: , 1996c. Kopelowicz, A.; Liberman, R.P.; Mintz, J.; and Zarate, R. Comparison of efficacy of social skills training for deficit and nondeficit negative symptoms in schizophrenia. American Journal of Psychiatry, 154: , Loas, G.; Noisette, C; Legrand, A.; and Boyer, P. Anhedonia, depression, and the deficit syndrome of schizophrenia. Acta Psychiatrica Scandinavica, 94:477^79,1996. Marder, S.R.; Davis, J.M.; and Chouinard, G. The effects of risperidone on the five dimensions of schizophrenia derived by factor analysis: Combined results of the North American Trials. Journal of Clinical Psychiatry, 58(12): , Mayerhoff, D.I.; Loebel, A.D.; Alvir, J.M.J.; Szymanski, S.R.; Geisler, S.H.; Borenstein, M.; and Lieberman, J.A. The deficit state in first-episode schizophrenia. American Journal of Psychiatry, 151: , Maziade, M.; Bouchard, S.; Gingras, N.; Roy, M.-A.; Gauthier, B.; Tremblay, G.; C6te, S.; Fournier, C; Boutin, P.; Hamel, M.; MeYette, C; and Martinez, M. Long-term stability of diagnosis and symptom dimensions in a systematic sample of childhood and early-adolescence onset schizophrenia: n. Positive/negative distinction and childhood predictors of adult outcome. British Journal of Psychiatry, 169: , 1996a. Maziade, M.; Gingras, N.; Bouchard, S.; Gauthier, B.; Tremblay, G.; Cote", S.; Fournier, C; Cardinal, A.; Boutin, P.; Charron, L.; Hamel, M.; Roy, M.-A.; Martinez, M.; and Merette, C. Long-term stability of diagnosis and symptom dimensions in a systematic sample of childhood and early-adolescence onset schizophrenia: I. Nosology, sex and age of onset patterns. British Journal of Psychiatry, 169: , Maziade, M.; Roy, M.A.; Fournier, J.P.; Cliche, D.; Merette, C; Caron, C; Garneau, Y.; Montgrain, N.; Shriqui, C; Dion, C; Nicole, L.; Potvin, A.; Lavallee, J.C.; Pires, A.; and Raymond, V. Best estimate diagnosis in genetic studies of psychotic disorders. American Journal of Psychiatry, 149: , Maziade, M.; Roy, M.-A.; Martinez, M.; Cliche, D.; Fournier, J.-P.; Garneau, Y.; Nicole, L.; Montgrain, N.; Dion, C; Ponton, A.-M.; Potvin, A.; Lavall6e, J.-C; Bouchard, S.; Boutin, P.; Brisebois, F.; and Merette, C. The negative, the psychoticism and the disorganized dimensions in a familial sample of schizophrenia and bipolar disorder: Continuity and discontinuity between the two major psychoses. American Journal of Psychiatry, 150: ,1995. McGlashan, T.H. Predictors of shorter-, medium-, and longer-term outcome in schizophrenia. American Journal of Psychiatry, 143:50-55, McGuffin, P.; Farmer, A.; and Gottesman, I.I. Is there really a split in schizophrenia? The genetic evidence. British Journal of Psychiatry, 150: , Merette, C; Ghazzali, N.; Savard, F.; Boutin, P.; Roy, M.- A.; and Maziade, M. An investigation of anticipation in schizophrenia and bipolar disorder using various severity indexes and controlling for statistical artifacts. American Journal of Medical Genetics (Neuropsychiatric Genetics), 96:61-68, Mozley, P.D.; Gur, R.E.; Resnick, S.M.; Shtasel, D.L.; Richards, J.; Kohn, M.; Grossman, R.; Herman, G.; and Gur, R.C. Magnetic resonance imaging in schizophrenia: Relationship with clinical measures. Schizophrenia Research, 12: , Murray, R.M.; Castle, D.; O'Callaghan, E.; Lewis, S.W.; Sham, P.; and Wilkins, S. The neurodevelopmental and adult schizophrenias: Kraepelin lost and Kraepelin found. Schizophrenia Research, 4: , Nibuya, M.; Kanba, S.; Sekiya, U.; Suzuki, E.; Matsuo, Y; Kinoshita, N.; Shintani, F.; Yagi, G.; and Asai, M. Schizophrenic patients with deficit syndrome have higher plasma homovanillic acid concentrations and ventricular enlargement. Biological Psychiatry, 38:50-56, Ribeyre, J.M.; Dollfus, S.; Lesieur, P.; Menard, J.F.; and Petit, M. Instrument devaluation du syndrome deficitaire de la schizophrenic: Schedule for the deficit syndrome (SDS): Interet et pertinence de la SDS. Presentation de le version Francaise. L'Encephale, 20:413^19, 1994a. Ribeyre, J.-M.; Lesieur, P.; Varoquaux, O.; Dollfus, S.; Pays, M.; and Petit, M. A comparison of plasma homovanillic acid in the deficit and nondeficit subtypes of schizophrenia. Biological Psychiatry, 36: , 135

22 Schizophrenia Bulletin, Vol. 27, No. 1, 2001 M.-A. Roy et al. Robins, E., and Guze, S.B. Establishment of diagnostic validity in psychiatric illness: Its application to schizophrenia. American Journal of Psychiatry, 126: ,1970. Ross, D.E.; Thaker, G.K.; Buchanan, R.W.; Kirkpatrick, B.; Lahti, A.C.; Medoff, D.; Bartko, J.J.; Goodman, J.; and Tien, A. Eye tracking disorder in schizophrenia is characterised by specific ocular motor defects and is associated with the deficit syndrome. Biological Psychiatry, 42: , Ross, D.E.; Thaker, G.K.; Buchanan, R.W.; Lahti, A.C.; Medoff, D.; Bartko, J.J.; Moran, M.; and Hartley, J. Association of abnormal smooth pursuit eye movements with the deficit syndrome in schizophrenic patients. American Journal of Psychiatry, 153: , Roy, M.-A., and Crowe, R.R. Validity of familial and sporadic subtypes of schizophrenia. American Journal of Psychiatry, 151: , Roy, M.-A., and DeVriendt, X. Symptomes positifs et negatifs de la schizophrenic: Une mise a jour. Canadian Journal of Psychiatry, 40:407^14, Roy, M.-A.; Flaum, M.; and Andreasen, N.C. No difference found between winter-born and non-winter-born schizophrenic cases. Schizophrenia Research, 17: , Roy, M.-A.; Labbe", A.; Maziade, M.; and Merette, C. Deficit schizophrenia is associated with male gender: A meta-analysis. Schizophrenia Research, in press. Roy, M.-A.; Lanctot, G.; Martinez, M.; Cliche, D.; Fournier, J.-R; Boutin, P.; Rodrigue, C; Charron, L.; Gamache, L.; Rouillard, E.; Montgrain, N.; Pires, A.; Wallot, H.; Ponton, A.M.; Garneau, Y.; Dion, C; Nicole, L.; Raymond, V; Cardinal, A.; MeYette, C; and Maziade, M. Factors affecting reliability of best estimate diagnosis of major psychoses in pedigree studies. American Journal of Psychiatry, 154: , Roy, M.-A.; Walsh, D.; and Kendler, K.S. Accuracies and inaccuracies of the family history method: A multivariate investigation. Acta Psychiatrica Scandinavica, 93: , Samson, J.A.; Simpson, J.C.; and Tsuang, M.T. Outcome studies of schizoaffective disorders. Schizophrenia Bulletin, 14(4): , Sham, PC; Castle, D.J.; Wessely, S.; Farmer, A.E.; and Murray, R.M. Further exploration of a latent class typology of schizophrenia. Schizophrenia Research, 20: , Spalletta, G.; Pasini, A.; De Angelis, F; and Troisi, A. Patients with deficit, nondeficit, and negative symptom schizophrenia: Do they differ during episodes of acute psychotic decompensation? Schizophrenia Research, 24: , Tamminga, C; Thaker, G.; Buchanan, R.W.; Kirkpatrick, B.; Breier, A.; Alphs, L.; Chase, T.; and Carpenter, W.T., Jr. Limbic system abnormalities identified in schizophrenia using positron emission tomography with fluorodeoxyglucose and neocortical alterations with deficit syndrome. Archives of General Psychiatry, 49: , Thaker, G.; Kirkpatrick, B.; Buchanan, R.W.; Ellsberry, R.; Lahti, A.; and Tamminga, C. Oculomotor abnormalities and their clinical correlates in schizophrenia. Psychopharmacology Bulletin, 25: , Thibaut, F.; Ribeyre, J.-M.; Dourmap, N.; M6nard, J.-F; Dollfus, S.; and Petit, M. Plasma 3-methoxy-4-hydroxyphenylglycol and homovanillic acid measurements in deficit and nondeficit forms of schizophrenia. Biological Psychiatry, 43:24-30, Tollefson, G.D., and Sanger, T.M. Negative symptoms: A path analytic approach to a double-blind, placeboand haloperidol-controlled clinical trial with olanzapine. American Journal of Psychiatry, 154: , Toomey, R.; Kremen, W.S.; Simpson, J.C.; Samson, J.A.; Seidman, L.J.; Lyons, M.J.; Faraone, S.V.; and Tsuang, M.T. Revisiting the factor structure for positive and negative symptoms: evidence from a large heterogeneous group of psychiatric patients. American Journal of Psychiatry, 154: , Tsuang, M.T. Genetics, epidemiology, and the search for causes of schizophrenia. American Journal of Psychiatry, 151:3-6, Tsuang, M.T.; Lyons, M.L.; and Faraone, S.V. Heterogeneity of schizophrenia: Conceptual models and analytic strategies. British Journal of Psychiatry, 156:17-26,1990. Turetsky, B.; Cowell, P.E.; Gur, R.E.; Grossman, R.I.; Shtasel, D.L.; and Gur, R.E. Frontal and temporal lobe brain volumes in schizophrenia: Relationship to symptoms and clinical subtype. Archives of General Psychiatry, 52: , Turetsky, B.I.; Colbath, E.A.; and Gur, R.E. P300 subcomponent abnormalities in schizophrenia: I. Physiological evidence for gender and subtype specific differences in regional pathology. Biological Psychiatry, 43:84-96, Van Os, J.; Fahy, T.A.; Jones, P.; Harvey, I.; Sham, P.; Lewis, S.; Bebbington, P.; Toone, B.; Williams, M.; and Murray, R. Psychopathological syndromes in the functional psychoses: Associations with course and outcome. Psychological Medicine, 26: ,

23 Subtyping Schizophrenia Schizophrenia Bulletin, Vol. 27, No. 1, 2001 Van Os, J.; Gilvarry, C; Bale, R.; Van Horn, E.; Tattan, T; White, I.; and Murray, R.M. A comparison of the utility of dimensional and categorical representations of psychosis. Psychological Medicine, 29: , Wagman, A.M.I.; Heinrichs, D.W.; and Carpenter, W.T. Deficit and non-deficit forms of schizophrenia: Neuropsychological correlates. Psychiatric Research, 22: , Waltrip II, R.W.; Buchanan, R.W.; Carpenter, W.T.; Kirkpatrick, B.; Summerfelt, A.; Breier, A.; Rubin, S.A.; and Carbone, K.M. Borna disease antibodies and the deficit syndrome of schizophrenia. Schizophrenia Research, 23: , Weissman, M.M.; Merikangas, K.; John, K.; Wickramaratne, P.; Prusoff, B.A.; and Kidd, K.K. Familygenetic studies of psychiatric disorders. Archives of General Psychiatry, 43: , Westermeyer, J.F., and Harrow, M. Course and outcome in schizophrenia. In: Tsuang, M.T., and Simpson, J.C., eds. Handbook of Schizophrenia. New York, NY: Elsevier Science, pp Zipursky, R.B.; Lim, K.O.; and Pfefferbaum, A. Volumetric assessment of cerebral asymmetry from CT scans. Psychiatry Research, 35:71-89, Acknowledgments Drs. Roy and Merette are supported by scientist awards from the Fonds de la recherche en sante du Quebec. This work is supported by grants from the Fonds de la recherche en Sante du Quebec ( ; principal investigator: Dr. Roy), the EJLB Foundation (Dr. Merette), and the Medical Research Council of Canada (MT-14187; principal investigator: Dr. Roy); and by a group grant from the Medical Research Council of Canada (GR14501; principal investigators: Dr. Maziade, Dr. M6rette, Roberta Palmour, Dr. Roy, Peter Szatmari). Addendum Since the completion of this manuscript, additional articles comparing deficit and nondeficit SZ have been published. Their most important among findings include the following. First, data from the Camberwell Register Psychosis Series (Kirkpatrick et al. 2000a), using the PDS, found (1) an association between summer birth deficit SZ, further replicating the association already observed in three previous data sets; and (2) an association between deficit SZ and the presence of SZ in relatives, and between nondeficit SZ and the presence of other psychiatric disorders in relatives. However, the reliance on information from probands' charts to diagnose relatives precludes considering these findings definitive. Second, data from the Irish High-Density Study of Schizophrenia (Ross et al. 2000) found evidence that deficit vs. nondeficit subtypes "breed true" by observing a significant rate of concordance for deficit vs. nondeficit subtypes among sibling pairs affected by SZ. Third, data from the Roscommon Family Study of Schizophrenia (Kirkpatrick et al ) found a trend for an association between deficit subtype of SZ probands and SZ in relatives, and a significant association between deficit subtype of the proband and "deficit-like" symptoms in unaffected relatives. Fourth, we found a first paper simultaneously seeking the correlates (e.g., various types of neurological soft signs) of psychotic and disorganized as well as deficit symptoms (Arango et al. 2000), as proposed in the present review. This paper reported distinct patterns of correlations for each of these three symptom dimensions, thereby lending support to the distinct dimension models. Fifth, two functional imaging studies have recently reported (Heckers et al. 1999; Delamillieure et al. 2000) an association between hypofrontality and deficit SZ. Altogether, these studies significantly add to the evidence for the validity of the distinction between deficit vs. nondeficit subtypes and clearly warrant a more widespread use of these subtypes. Bibliography Arango, C; Kirkpartick, B.; and Buchanan, R.W. Neurological signs and the heterogeneity of schizophrenia. American Journal of Psychiatry, 157: , Delamillieure, P.; Fernandez, J.; Constans, J.-M.; Brazo, P.; Benali, K.; Abadie, P.; Vasse, T.; Thibaut, F.; Courtheoux, P.; Petit, M.; and Dollfus, S. Proton magnetic resonance spectroscopy of the medial prefrontal cortex in patients with deficit schizophrenia: preliminary report. American Journal of Psychiatry, 157: , Heckers, S.; Goff, d.; Schacter, D.L.; Savage, C.R.; Fischman, A.J.; Alpert, N.M.; and Rauch, S.L. Functional imaging of memory retrieval in deficit vs nondeficit schizophrenia. Archives of General Psychiatry, 56: , Kirkpatrick, B.; Castle, D.; Murray, R.M.; and Carpenter, W.T. Risk factors for the deficit syndrome of schizophrenia. Schizophrenia Bulletin, 26: , 2000a. Kirkpatrick, B.; Ross, D.E.; Walsh, D.; Karkowski, L.; and Kendler, K.S. Family characteristics of deficit and nondeficit schizophrenia in the Roscommon Family Study. Schizophrenia_Research, 45:57-64,