Predicting Schizophrenia: Findings from the Edinburgh High-Risk Study

BRITISH JOURNAL OF PSYCHIATRY (2005),(2005),186,18^25 186, 18^25

Predicting schizophrenia: findings schizophrenia. Two models of inheritance were considered, predicting that from 200 from the Edinburgh High-Risk Study individuals aged 16–24 years 19 and 29 persons, respectively, would develop schizophrenia within 10 years. The actual EVE C. JOHNSTONE, KLAUS P. EBMEIER, PATRICK MILLER, number would depend upon the ages of DAVIDG.C.OWENSandSTEPHENM.LAWRIEDAVID G. C. OWENS and STEPHEN M. LAWRIE the individuals in the sample and the relative frequencies of the stronger and weaker patterns of inheritance, but it appeared rea- sonable to predict that from 200 such high- risk individuals 20 would develop schizophrenia within 10 years. It is, of course, Background The hypothesis that Direct evidence for a neurodevelopmental the case that some individuals from the fa- schizophrenia is neurodevelopmental was origin of schizophrenia can only be acquired milies who have illnesses of very early onset by comparing individuals at risk with might be excluded and some might become investigatedininvestigated in a prospective study of normal controls before illness onset, and ill later, but the purpose of our study was young people with a postulated10^15% following both groups through the period not to acquire every case but to acquire suf- risk for the development of schizophrenia. of risk until the psychosis does or does ficient numbers for adequate comparisons. not develop. Several prospective studies There is little work sufficiently similar to Aims To determined e te r min e premorbidp r e m o r b i d have identified high-risk individuals as provide a basis for adequate power calcula- variables distinguishing high-risk people infants on the grounds that their mothers tions, but imaging was an important part of who will go on to develop schizophrenia had schizophrenia (for review, see Tarrant our considerations, and the study by Sud- from those who will not. & Jones,&Jones,1999). These studies encountered dathdath et aletal (1990) of monozygotic twins pro- difficultiesdifficultiesresulting from the 20-year inter- vided clear findings on 15 discordant pairs. MethodMethod Ahigh-risksampleof163A high-risk sample of163 val before the participants entered the We aimed, therefore, to acquire a high-risk period of maximum risk (Asarnow, 1988; sample ofsampleof200 persons (Johnstone et aletal,, young adults with two relatives with Cornblatt & Obuchowski, 1997). The 2000). Controlgroups comprised well schizophrenia was recruited.They and 36 Edinburgh High-Risk Study (EHRS) young people and individuals in the first controls were serially examined.Baseline (Lawrie(Lawrie et aletal, 1999; Johnstone et aletal, 2000),2000) episode of schizophrenia who did not have measures were compared betweenthose is a development of such work. It investi- a family risk of the disorder. who did develop schizophrenia, a well gates individuals at enhanced risk because The EHRS examined the pathogenesis they have two or more affected relatives. of schizophrenia by addressing the controlcontrolgroup, group, a wellwellhigh-risk high-risk group and These individuals were identified at age hypothesis that individuals from the high- high-risk participants with partial or 16–24 years as they entered the period of risk sample who eventually develop schizo- isolated psychotic symptoms. maximum risk andwere followed over 10 phrenia would, at ascertainment and long years, by whichtime most of those destined before the development of psychosis, differ ResultsResults Ofthose at high risk, 20 to develop schizophrenia would have done from high-risk individuals who do not 11 developed schizophrenia within 2//22 years. so.so. developdevelopschizophreniaschizophrenia and also from the More experienced isolated or partial well controlgroup, in terms of the clinical and neurobiological assessments used. We psychotic symptoms.Those who METHOD predicted that, although the high-risk developed schizophrenia differed from sample as a whole would differ from the those who did not on social anxiety, The purpose of the Edinburgh High-Risk control groups in terms of these indices, withdrawalandother schizotypalfeatures. Study is to determine the features that those who went on to develop schizo- The whole high-risk sample differed from distinguish high-risk individuals who go phrenia would show more marked differ- on to develop schizophrenia from those ences than those who did not. Previous the control group on developmental and who do not, and to compare relevant comparisons between this high-risk sample neuropsychological variables. variables in affected and unaffected and the two control groups have shown members of the high-risk sample with differences in clinical, psychopathological, Conclusions The genetic componentof matched controls. We sought to acquire a psychological, neurological, developmental schizophrenia affects manymore sample of young people aged 16–24 years and imaging variables (Hodges et aletal,, individualsthanwilldeveloptheillness,individuals than will develop the illness, and and considered to be well at ascertainment, 1999; Johnstone et aletal, 2000; Lawrie et aletal,, partial impairment can be found in them. who each had at least two first- or second- 20012001aa,,bb; Miller;Miller et aletal, 2002,2002aa; Byrne;Byrne et aletal,, degree relatives with schizophrenia. To 2003). One of the central comparisons to Highly significant predictors ofthe determine the number of high-risk indivi- be addressed in this data-set is the development of schizophrenia are duals that we would need to study in order comparison in terms of baseline data of detectable years before onset. to achieve a number who would become ill those who have and those who have adequate for relevant comparisons, we not gone on to develop schizophrenia. Declaration of interest None. considered data on age at onset from We are now in a position to examine this Funding detailed in Acknowledgements. 235 families multiply affected with issue.issue.

1818

Downloaded from https://www.cambridge.org/core. 27 Sep 2021 at 03:02:27, subject to the Cambridge Core terms of use. PREDICTING SCHIZOPHRENIA

Derivation of the sample Organization, 1992) and PSE (CATEGO some data and 156 provided complete data The study began in 1994. High-risk indivi- S+ or O+). Psychotic illness of a non-schizo- by closure of the recruiting period of the duals aged 16–25 years with no history of phrenic type is not covered by the scale, but programme in July 1999. There were 36 serious psychiatric problems were identified it did not occur. Points 2 and 3 are com- participants in the well control group. At throughout Scotland on the basis that they bined within this study and participantspantspartici ascertainment the mean age of the high-risk had at least two first- or second-degree rela- are referred to as having had psychotic oror group was 21.19 years (s.d.¼2.97) and it tives affected with schizophrenia (Hodges possibly psychotic symptoms. comprised 77 men and 79 women. The well et aletal, 1999). Participants for the well con- The neuropsychological test battery control group’s mean age was 21.17 years trol group were recruited from the social (Byrne(Byrne et aletal, 1999) consisted of tests of (s.d.(s.d.¼2.37) and there were 17 men and 19 network of the high-risk individuals them- general IQ, attention, motor speed, execu- women. On social class, however, the selves; they had no personal or family tive function, verbal learning and memory. samples did differ significantly, with 19 history of psychotic illness, but could have Brain structure was assessed (Whalley etet (53%) of the control group having fathers a family history of other psychiatric illness alal, 1999) by MRI scanning on a 1 T in non-manual occupations against only and otherwise were as similar to the high- Magnetom scanner (Siemens, Erlangen, 46 (29.5%) of the high-risk group 22 risk participants as possible (Hodges et aletal,, Germany). In addition to these measures, ((ww ¼6.9, Fisher’s exact test PP¼0.011).0.011). 1999). Participants for the first-episode we assessed the degree of genetic liability The updated results reported here case group were recruited from local hospi- of the high-risk participants by both catego- concern 173 participants (from both the tals and were balanced group-wise for age rical and continuous methods (Lawrie et aletal,, high-risk and the well control groups) with with the high-risk individuals. Both control 20012001aa). From 1999 to 2004 the assess- whom we remain in regular contact. Of groups were planned to consist of approxi- ments were continued every 18 months, these, 27 were members of the well control mately 35 persons each, the maximum with the addition of functional MRI. group, none of whom has developed schizo- number of the high-risk sample predicted The principal purpose of this study is phrenia. The high-risk group was divided to develop schizophrenia. twofold. First, we wished to determine into ‘high risk without psychotic or possibly variables that at baseline (i.e. at initial psychotic symptoms ever by July 2003’ Plan of the study and assessments ascertainment assessment) distinguish be- ((nn¼66), ‘high risk with psychotic or poss- usedused tween high-risk individuals who will fall ibly psychotic symptoms by July 2003’ The plan for the period 1994–1999 was to ill with schizophrenia, and those who will ((nn¼60) and ‘high risk ill by July 2003’ assess all participants at ascertainment in not do so but who will or will not show ((nn¼20). Occasionally the ‘high risk ill’ terms of clinical features, neuropsychology psychotic or possibly psychotic symptoms. participants were classified as ill at their and brain structure as determined by struc- To do this, we selected all the variables planned review, but – as might be ex- tural magnetic resonance imaging (MRI). from our previous studies (Lawrie et aletal,, pected – most developed schizophrenia People in the first-episode control group 20012001aa,,bb; Miller;Miller et aletal, 2001, 2002aa,,bb,,cc;; between assessments and were admitted to were assessed only on ascertainment; for ByrneByrne et aletal, 2003; Langsley et aletal, 2004) that a local service. Consultants in the areas clarity, findings in these individuals have at baseline either distinguished high-risk from which these patients came were colla- been omitted from this report, although individuals from well controls beyond the borators in the project. They and the high- their baseline data have been presented PP550.01 level of significance, or distin- risk participants themselves, and their elsewhere (Lawrie et aletal, 2001,2001aa,,bb; Byrne;Byrne etet guished high-risk individuals who experi- families, knew that we wished to be in- alal, 2003). In the high-risk and the well con- enced psychotic symptoms at an early formed of any deterioration. Through their trol groups psychopathological assessments stage from those who did not (PP550.01)0.01) cooperation we were able to obtain PSE were repeated every 18 months. The base- (see Table 1). We retested these variables ratings shortly after admission for line clinical measures included assessments to assess the usefulness of each one in mak- treatment of the first psychotic episode for of childhood behavioural traits (Miller etet ing the distinctions described. The sample 18 of 20 participants. All those rated alal, 2002,2002aa), schizotypal features (Miller etet retested consisted of all the participants fulfilled the PSE CATEGO criteria for alal, 2002,2002bb,,cc), and the neurodevelopmental with whom we were still in regular contact. schizoschizophreniaphrenia and paranoid psychoses variables of minor physical anomalies and We predicted that there would be a grada- and all 2020andall fulfilled ICD–10 criteria for neurological soft signs (Lawrie et aletal,, tion in the effects, from high-risk indivi- schizophrenia. 20012001bb), ocular hypertelorism (Boyes et aletal,, duals who fall ill followed by high-risk We were unable to obtain follow-up 2001), dermatoglyphics (Langsley et aletal,, individuals with psychotic symptoms only, data on 10 (6%) of the high-risk group 2004) and substance use (Miller et aletal,, high-risk individuals without psychotic and 9 controls (25%). 2001). Mental state was assessed at entry symptoms and well controls, in that order. Twenty-seven variables assessed at and at all the follow-up points by the Our second aim was to describe, for the baseline met our criteria for initial inclusion Present State Examination (PSE; Wing et aletal,, first time, some of the characteristics of set out above. Baseline scores for each of 1974), and from this we derived the follow- the high-risk participants who became ill these were subjected to one-way analysis ing five-point psychopathological scale with schizophrenia. of variance (ANOVA) (with log transfor- (Johnstone(Johnstone et aletal, 2000): 0, no psychotic or mations where warranted) within our neurotic symptoms; 1, neurotic symptoms RESULTSRESULTS sample of 173 participants divided as only; 2, partially held psychotic symptoms; above. These ANOVAs were followed up 3, definite but isolated and/or transient Predicting illness onset by three planned comparisons: psychotic symptoms; 4, schizophrenia A total of 229 high-risk participants were diagnosed by ICD–10 (World Health ascertained, of whom 163 had provided (a) ‘high risk ill’ vv. controls;.controls;

1919

Downloaded from https://www.cambridge.org/core. 27 Sep 2021 at 03:02:27, subject to the Cambridge Core terms of use. JOHNSTONE ET AL )) .. ContinuedContinued vv 0.024 0.023 (( NSNS 0.0010.001 NSNS NSNS NSNS NSNS NSNS NSNS NSNS NSNS NS NS NS elilelill fell illfell 55 high risk high risk High risk High risk with symptoms with symptoms 11 .. vv 0.015 NS 0.008 0.024 0.0010.001 0.044 0.0210.021 NS NS 0.001 0.023 NS NS NS NS NS NS NS elilelill fell illfell 55 high risk high risk High risk High risk no symptoms no symptoms .. vv Statistical comparisons Statistical comparisons 0.028 0.008 0.003 0.015 0.007 NS 0.022 NS 0.013 0.0350.035 NS 0.002 NS 0.26 NS 0.0010.001 0.002 NS 0.001 0.0010.001 NS 0.0010.001 NS 0.012 NS 0.016 NS 0.001 0.001 elilelill fell illfell 55 55 55 55 controls controls High risk High risk FF 0.027 0.271 0.027 0.26 0.034 0.022 0.003 0.007 0.003 0.0050.005 0.0090.009 NS 0.007 0.002 0.0010.001 0.006 0.013 0.0010.001 0.012 0.012 0.00 1 0.016 0.002 0.001 0.003 0.010 0.016 0.001 0.001 Overall Overall 55 55 0.01)0.01) 55 PP 20)20) 7.9 0.61 0.001 9.0 3.3 0.003 0.98 0.0010.001 0.028 6.9 0.006 6.9 5.6 0.003 6.0 0.271 3.5 0.012 ¼ 47.1 0.0410.041 0.0050.005 0.044 32.1 0.034 34.1 0.007 97.4 0.010 43.5 0.001 38.5 43.4 0.016 elilelill fell illfell nn (( High risk High risk 60)60) 7.7 6.9 0.03 0.98 0.02 0.61 5.0 9.0 3.7 6.9 3.2 5.6 2.6 3.3 8.7 7.9 6.5 6.0 2.3 3.5 ¼ with with 97.0 97.4 52.4 47.1 nn (( High risk High risk symptomssymptoms Mean values Mean values 67) 67) 0.16 0.16 0.03 1.8 2.6 2.1 3.2 0.22 9.6 8.7 3.0 2.3 4.2 5.0 2.7 3.7 0.22 0.02 8.4 7.7 6.4 6.5 nono ¼ 19.3 24.924.9 38.5 51.7 52.4 18.2 24.224.2 43.5 99.9 97.0 33.5 32.432.4 32.1 98.9 98.998.9 98.098.0 0.0680.068 nn 77 77 (( High risk High risk symptomssymptoms 7)7) 1.1 7.0 1.6 1.8 1.1 3.0 0.05 0.7 7.0 6.4 2.9 1.6 2.1 1.8 2.7 0.480.48 0.05 0.7 1.8 2.9 4.2 ¼ 11.6 37.9 11.6 9.6 10.9 37.9 35.235.2 35.635.6 34.1 10.9 18.2 10.7 18.7 48.1 54.4 36.0 10.7 8.4 18.7 19.3 36.0 33.5 48.1 45.745.7 45.845.8 43.4 54.4 51.7 nn (( 107.2 107.2 99.9 77 104.3 104.3 98.9 Controls Controls 0.010.01 55 PP No No No No No No No No No No No No No No No Yes Yes No Yes No Yes Yes Yes Yes No No No Yes Yes Yes Yes Yes Yes Yes Yes Yes .highrisk.highrisk vv beyond 19991999 beyond not psychotically symptomatic pre not psychotically symptomatic pre ypoai presymptomatic 1999 symptomatic pre 1999 High-risk psychotically 0.01 0.01 No No No No No No No No No No No No No No No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 55 PP differed from differed from controls beyond controls beyond All high-risk subjects All high-risk subjects High-risk psychotically , 2003),2003) et aletal )atage16years ) at age 16 years )) aa bb , 2002,2002 , 2002,2002 et aletal et aletal The study used variables and at their breakdown baseline to according current status (other variables assessed did not at achieve significance The study used variables and at their breakdown baseline to according current status (other variables assessed did not at achieve significance Delayed Immediate Delayed recall Immediate recall Total, trials 1^5 Trial 1 Delayed Immediate Delayed recall Immediate recall Total, trials 1^5 Trial 1 Total score Other problems Aggressive behaviour Total score Attention problems Anxiety^depression Odd behaviour factor (Miller(Miller (Miller(Miller Social withdrawal factor RBMT story Visual reproductions RAVLTRAVLT HSCT type B errors Total score NART pttewr test Spot-the-word Other problems WAIS^R Full-scale IQ Aggressive behaviour Total score Attention problems Anxiety^depression Odd behaviour factor Social withdrawal factor RBMT story Visual reproductions HSCT type B errors NART pttewr test Spot-the-word WAIS^R Full-scale IQ Childhood Behavior Checklist Structural Interview for Schizotypy Childhood Behavior Checklist Structural Interview for Schizotypy Neuropsychological tests (Byrne able 1a l e 1 Ta b l e 1Tab Baseline variable Neuropsychological tests (Byrne Baseline variable

Downloaded from https://www.cambridge.org/core. 27 Sep 2021 at 03:02:27, subject to the Cambridge Core terms of use. PREDICTING SCHIZOPHRENIA .. vv 0.002 NS NS NS NS NS NS elilelill fell illfell fitisapplied. high risk fitisapplied. high risk High risk High risk with symptoms with symptoms 11 .. vv 0.001 NS NS 0.001 0.002 NS elilelill fell illfell 55 high risk high risk High risk High risk no symptoms no symptoms .. vv Statistical comparisons Statistical comparisons 0.003 NS 0.002 NS 0.0010.001 elilelill fell illfell 55 controls controls High risk High risk FF 0.151 0.151 0.141 0.212 0.141 0.212 0.026 0.0090.009 NS 0.562 0.013 0.003 0.843 0.0010.001 0.018 0.002 Overall Overall 55 20)20) 3.1 0.013 2.1 3.6 0.843 3.4 0.4 0.562 0.5 0.3 0.018 ¼ 39.4 elilelill fell illfell nn (( High risk High risk 60)60) 1.5 2.1 3.0 3.1 0.3 0.5 3.4 3.4 4.6 4.54.5 0.026 2.0 0.3 ¼ with with 30.7 39.4 nn (( High risk High risk symptomssymptoms Mean values Mean values 67)67) 0.1 0.20.2 0.4 1.4 1.5 1.9 2.0 2.9 3.0 4.7 4.6 3.5 3.63.6 3.6 nono ¼ 25.5 30.7 nn (( High risk High risk symptoms symptoms 7)7) 0.1 0.9 0.0 0.1 0.30.3 0.3 3.9 4.9 3.6 3.6 0.9 1.4 0.0 0.1 3.9 1.9 2.7 4.9 4.7 3.6 3.5 3.6 3.43.4 3.4 2.7 2.9 ¼ 25.8 25.8 25.5 nn (( Controls Controls 0.010.01 55 PP No No No No No No No No No No No No No No Yes Yes Yes Yes .highrisk .highrisk vv beyondbeyond 19991999 not psychotically not psychotically symptomatic pre symptomatic pre 0.0033.We do not regard this as necessary as we are testing specific hypotheses, restricted but it will be that evident we lose few results relatively i 0.0033. We do not regard this as necessary as we are testing specific restricted hypotheses, but it will be evident that we lose relatively few results i symptomatic pre 1999 ypoai presymptomatic 1999 High-risk psychotically 55 PP 0.01 0.01 No No No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 55 PP differed from differed from controls beyond controls beyond All high-risk subjects All high-risk subjects High-risk psychotically , 2004),2004) et aletal , 2001),2001) 1000 1000 66 et aletal )) )) )) bb aa cc )) 1000 1000 1000 1000 , 2001,2001 , 2001,2001 66 , 2002,2002 66 et aletal et aletal Continued Continued et aletal (( Total Other tests Sensory integration (Lawrie Binocular diameter (from SMRI), mm Left thalamic nucleus Right AHC etACetAHC Left AHCLeft (Lawrie Whorls, arches (Miller(Miller Full scale Total Other tests Sensory integration (Lawrie Binocular diameter (from SMRI), mm Yes Left thalamic nucleus Right AHC (Lawrie Whorls, arches Full scale Neurological Evaluation Scale yetlrs (Boyes Hypertelorism Brain volume proportions Dermatoglyphics (Langsley Rust Inventory of Schizoptypal Cognitions a ble 1a l e 1 Ta b l e 1Tab Baseline variable magnetic SMRI, structural MemoryTest; Behavioural RBMT,Rivermead LearningTest; Verbal Auditory RAVLT,Rey Adult ReadingTest; NART,National Sentence CompletionTest; complex; HSCT,Hayling AHC, amygdala^hippocampal resonance imaging; WAIS ^R, Wechsler1. Bonferroni Adult Intelligence correction Scale for multiple ^ testing requires Revised. a significance level of Neurological Evaluation Scale yetlrs (Boyes Hypertelorism Brain volume proportions Dermatoglyphics (Langsley Rust Inventory of Schizoptypal Cognitions H,aydl^ipcma ope;HC,HyigSnec opeinet ATNtoa dl ednTs;RVTRyAdtr eblLannTs;RM,iema eaiua eoyet MI tutrlmagnetic SMRI, structural MemoryTest; Behavioural RBMT,Rivermead LearningTest; Verbal Auditory RAVLT,Rey Adult ReadingTest; NART,National Sentence CompletionTest; complex; Hayling HSCT, AHC, amygdala^hippocampal resonance imaging; WAIS ^R, Wechsler1. Bonferroni Adult correction Intelligence Scale for multiple ^ testing requires Revised. a significance level of Baseline variable

Downloaded from https://www.cambridge.org/core. 27 Sep 2021 at 03:02:27, subject to the Cambridge Core terms of use. JOHNSTONE ET AL

(b) ‘high risk ill’ vv. ‘high risk without generally greater than positive, being as DISCUSSION psychotic or possibly psychotic high as 97%, whereas the best positive symptoms’;symptoms’; predictive value is 50%. First, in this prospective study, the predicted number of participants becoming ill (c) ‘high risk ill’ vv. ‘high risk with psychotic with schizophrenia was arrived at within or possibly psychotic symptoms’. Characteristics of high-risk 8 years. We have been fortunate in succeed- Table 1 sets out these variables, indicates individuals who fell ill ing in retaining contact with the great the earlier results that justified their majority of the high-risk participants and Twelve men and eight women developed inclusion and gives the mean values and a lesser (although still high) proportion of schizophrenia. At ascertainment their the significance levels for the ANOVAs the control group over this time, and can mean age was 20.3 years (s.d.¼2.20) for2.20)for overall.overall. have confidence in our assessments of the men and 19.6 years (s.d.¼2.73) for women. On this basis the Structural Interview clinical status of the individuals concerned. The mean ages at which they became ill for Schizotypy (SIS; Miller et aletal, 2002,2002bb) so-)so- Second, we show that high-risk individuals were 22.8 years (s.d.¼2.5) for men and cial withdrawal score, the SIS total score who developed schizophrenia during the 22.8 years (s.d.¼2.50) (women) and the and the Rust Inventory of Schizotypal follow-up period clearly differed at ascer- mean length of time between ascertainment Cognitions (RISC; Miller et aletal, 2002,2002cc)) tainment (some years before the develop- and diagnosis of schizophrenia was 2.4 distinguish the high-risk group who fall ill ment of the psychosis) from the high-risk years (s.d.years(s.d.¼1.9) for men and 3.2 years from high-risk subjects who do and do participants who remained well and the (s.d.(s.d.¼0.9) for women. There was no not develop psychotic symptoms. Although normal control group on nine clinical significant gender difference on these results of the Rey Auditory Verbal Learning measures and to a lesser extent on neuro- variables and no significant difference on Test (RAVLT) total are just significant psychological assessments. On several other age at ascertainment between these (Table 1), it is the behavioural measures variables (see Table 1) trends are shown. individuals and the high-risk participants that clearly separate high-risk participants who did not become ill: mean age at who will become ill from the other two ascertainment 21.3 years (s.d.¼3.0) for3.0)for Issues of discrimination high-risk groups. Statistically significant men and 21.4 years (s.d.¼3.0) for3.0)for and numbers separation of those who will become ill women.women. from the controls is, however, additionally The idea that people at high risk who will According to our classification, at the achieved on a number of psychological become ill can be clearly distinguished from time of entry (i.e. at baseline), 11 people tests, on measures of childhood behaviour, those who do not become ill is an oversim- who fell ill described having or having on the developmental measures of ocular plification of this situation. Psychopatholo- had some psychotic or possibly psychotic hypertelorism and dermatoglyphic gical symptoms short of psychosis occurred symptoms, and 9 who fell ill did not. Table variables, and on left thalamic volume. in many more members of the high-risk 3 gives further details, dividing participants sample than were ever predicted to develop who subsequently fell ill according to their Clinical significance schizophrenia. It is extremely unlikely that symptom status on entry. A preponderance all the high-risk participants who have We then went on to consider the adequacy of men who fell ill showed psychotic or shown psychotic or possibly psychotic of the measures separating the high-risk possibly psychotic symptoms on entry, symptoms at some stage will actually devel- group who would become ill from the other whereas the group without such symptoms op schizophrenia. If this were to happen, two high-risk groups, for predicting schizo- contained mostly women. Those with such the frequency of the disorder would be phrenia within the high-risk sample as a symptoms were older on entry than those greatly in excess of what is usually re- whole (Table 2). Each scale is dichotomised without, and there was a slight tendency ported. Most of the men, at least, have with the cut-off points determined by for those without such symptoms to have now passed through the period of highest receiver operating characteristic analyses a higher proportion of ill parents or risk. Moreover, it has been repeatedly re- (Table 2). Negative predictive power is siblings.siblings. ported that the well-established gender differences in age at onset of schizophrenia are Ta b l e 2 Predictors of schizophrenia dividing thethe‘high ‘high risk ill’ group from other high-risk participants much less in familial cases (Hafner et aletal,, 1999). The Copenhagen High Risk Study Optimal Sensitivity Specificity Positive Negative (Parnas(Parnas et aletal, 1982; Cannon et aletal, 1994),,1994), which was similar in design to our study, cut-offcut-off (%) (%) predictive predictive power identified children of women with a psy- point power (%) (%) chotic disorder when the children were aged between 10–19 years in 1962, and RAVLT total trials 1^548.5 61.1 32.8 11.8 85.1 followed them up between 1972 and 1974 SIS at entry when they were about 25 years old (Parnas Social withdrawal factor1.10 44.4 90.2 40.0 91.7 et aletal, 1982) and again between 1986 and Oddness 0.8161.1 78.0 28.9 93.2 1989 when their mean age was 39 years Total scorescoreTotal 25.5 88.988.9 68.368.3 29.129.197.7 (Cannon(Cannon et aletal, 1994); the number of indi- RISC at entry 39.561.1 91.3 50.0 94.3 viduals with schizophrenia appeared to

RAVLT,ReyAuditoryVerbalLearningTest;RISC,RustInventoryofSchizotypalCognitions;SIS,StructuralInterviewfor increase by four (from 13 to 17) between Schizotypy. the two reports. On the basis of the

2222

Downloaded from https://www.cambridge.org/core. 27 Sep 2021 at 03:02:27, subject to the Cambridge Core terms of use. PREDICTING SCHIZOPHRENIA

Ta b l e 33Tab Characteristics of individuals at high risk who fell ill, according to presence or absence of psychotic findings also have clear implications for symptoms on entry thethedesign of genetic studies, as it is apparent that people at high risk who develop schizophrenia closely resemble those who Without With psychotic OverallStatistic PP develop symptoms short of the diagnosis. psychotic symptoms What appears to be inherited is a state of symptoms on on entryonentry vulnerability rather than psychosis per seperse.. entry Other factors would seem to be involved in the development of florid schizophrenia. Gender, nn Clearly, such findings could be used to Male 212 10012 Fisher’s exact0.005 provide a guide to intervention strategies; FemaleFemale 717 188probability however, they raise important questions. Social class, nn Why do not all those with the vulnerability ManualManual 696 91515Fisher’s exact 0.6170.617 factors become ill? What can be done to try Non-manual 323 255probability to retain more individuals in a state in Illness present in which florid illness does not occur and Parent or sibling, nn 747 41111Fisher’s exact0.092 functioning remains good, even though Other relative, nn 272 799probability abnormalities can be demonstrated on Age on entry, years: mean18.38 21.0119.95 tt¼2.942.94 0.0090.009 detailed investigation? Details of the pro- gress of symptoms in our high-risk sample Time between entry and 2.87 2.212.21 2.582.58 tt¼0.867 0.3980.398 over the years are the subject of a separate illness onset, years: mean paper (Owens et aletal, 2005). These individuals volunteer no complaints, most of them are in work and, to the casual observer, they Copenhagen study we would not expect study and some of them progressed to ill- do not appear dysplastic or in any way more than a few more of our high-risk par- ness quite soon. Those who had psychotic impaired and are apparently normal young ticipants now to develop schizophrenia. In or possibly psychotic symptoms were older people who for everyday purposes function terms of simple behavioural measures from than those who did not. On the other hand, well. If they could be held at this stage, the SIS and RISC, the high-risk participants there was no difference in the average time their apparent inheritance of a state of who have become ill (Table 1) show ob- between recruitment and illness for those vulnerability to schizophrenia need be no vious and significant differences from those who did and did not have symptoms on real disadvantage to them. who have not become ill. However, it does entry according to our classification; this not seem likely that there is a clear separa- average time overall was 2½ years. Further- tion between these two groups in terms of more, although there are indeed highly Relationship to other research developmental measures, as there seems to significant differences on the RISC and the It is appropriate to consider these findings be a gradient of impairment. SIS between those who fell ill and the other in relation to the results of studies that The strongest discriminators identified high-risk participants, there is also con- define individuals as being of high risk of in our study between those who fell ill siderable overlap, i.e. many of the high-risk schizophrenia on the basis of symptomatic and the other high-risk participants are group who did not fall ill were just as criteria (e.g. Klosterkotter et aletal, 2001) oror,2001) found on the RISC and the SIS. The 26 symptomatic on entry as any of those who a combination of familial risk and sympto- items of the RISC are designed to measure did.did. matic criteria (Yung et aletal, 2003, 2004). schizotypal cognitions rather than overt Such individuals, in contrast to those in psychotic symptoms (examples are ‘I never our study, present seeking help for symp- use a lucky charm’ and ‘sometimes I get a Possible clinical applications toms. Much higher rates of transition to weird feeling that I am not really here’). The data in Table 2 indicate that in a psychosis (not necessarily schizophrenia) The SIS contains several scales, some of sample of high genetic liability, we could were found – 36% over 12 months in the which directly measure near-psychotic use some of the measures to successfully Australian study (Yung et aletal, 2004) and symptoms but most of which do not. The identify a group who are likely to remain 49.4% over 9.6 years in the Cologne study elements composing the social withdrawal well, and we could also identify a group (Klosterkotter et aletal, 2001) – and the positive factor, which gives the strongest result, in whom the chance of development of predictive value of certain variables was concern anxiety and introversion rather schizophrenia was 50%, rather than the ap- greater than we have found here. In the than anything of a psychotic nature (Miller proximately 10% risk conveyed by their help-seeking samples described by Yung etet et aletal, 2002,2002bb). However, the question is known familial high risk. Although replica- alal (2004) and Klosterkotter et aletal (2001) it(2001)it raised as to whether the individuals who la- tion would be important before this is is evident that the presence of sub-threshold ter suffered onset of schizophrenia were in applied in clinical practice, and the ethical psychotic symptoms was associated with the prodromal phase of the illness on re- issues would require careful consideration, the later development of psychotic illness, cruitment to the study. There is no easy an- the SIS and the RISC are simple measures and the suggestion that such symptoms swer. Just over half of those who fell ill that could be widely employed, and it is merit active treatment is reasonably made. (Table 3) were in the ‘psychotic or possibly possible that this could be helpful for In members of our study sample, who were psychotic symptoms’ group on entry to the clinicians, parents and individuals. The not seeking care, it is evident that transient

Downloaded from https://www.cambridge.org/core. 27 Sep 2021 at 03:02:27, subject to the Cambridge Core terms of use. JOHNSTONE ET AL

and partial psychotic symptoms and psychotic-like experiences occur in many CLINICAL IMPLICATIONS more people than might be anticipated to develop schizophrenia. The filmed records && Among individuals at enhanced genetic risk of schizophrenia, a state of of the PSE interviews show that such symp- vulnerability, including transient and partial symptoms, will occur in many more toms often appear to have been associated individuals than will develop florid schizophrenia. with little distress or functional impairment, and we know that they may be && It is possible, using simple behavioural assessments of schizotypal and anxiety short-lived and followed by years in which cognitions, to predict with some accuracy those of a high-risk group who will (and they do not occur at all. with considerable accuracy those who will not) develop schizophrenia, some years before the development of the psychosis. What does this tell us about && schizophrenia? Neuropsychological and neurodevelopmental measures are more successful in distinguishing individuals at high risk from healthy controls than they are in The central finding of our study is that it is distinguishing high-risk individuals who will develop schizophrenia from those who the simple behavioural measures of the SIS and the RISC that provide the best measure will not. of distinguishing high-risk individuals who LIMITATIONS will develop schizophrenia from those who will not. None the less, there are a number && The findings of the study refer to a group of individuals with a substantial family of other distinguishing measures (particu- history of schizophrenia who have been willing to participate in repeated assessments larly in terms of neuropsychology) where over an 8-year period, and thus are not typical of the generality of individuals who highly significant results are obtained, may develop schizophrenia. especially on measures of episodic memory. Impairments in this task are suggestive of && Not all the participants have passed through the principal period of risk of temporal lobe dysfunction. We know from schizophrenia, and some who are currently well may yet develop the psychosis. the serial studies in the EHRS that both && memory function and temporal lobe The control group volunteers are partly selected by their willingness to continue volume, as demonstrated by structural with this ongoing study, despite having no personal interest in the issue; they are, MRI, deteriorate with the passage of time therefore, likely to be more socially responsible and persistent than average. (Cosway(Cosway et aletal, 2000; Lawrie et aletal, 2002;,2002; JobJob et aletal, 2003) in those with psychotic or possibly psychotic symptoms. We consider that the findings of the study as a whole are consistent with the view that schizo- EVE C. JOHNSTONE, FRCPsych, KLAUS P. EBMEIER, MD, PATRICK MILLER, PhD, DAVID G. C. OWENS, phrenia is primarily a disorder of temporal FRCPsych, STEPHEN M. LAWRIE, MRCPsych, Division of Psychiatry,University of Edinburgh, Edinburgh, UK lobe structure and function which develops slowly over several years. The exact nature CorrespCorrespondence:ondence: Professor Eve C.Johnstone,DivisionC. Johnstone, Division of PsychPsychiatry,Universityiatry,University of Edinburgh,Kennedy Tower,Royal Edinburgh Hospital, Morningside Park, Edinburgh EH10 5HF,UK. E-mail: of the change that pushes an individual into Norma.Brearley@@ed.ac.uked.ac.uk psychosis is not clear at this point, but our continuing studies, particularly of functional (First received 12 January 2004, final revision 11August 2004, accepted 26 August 2004) imaging, may reveal this.

Final comment Although imaging is integral to the high- from the Medical Research Council. The sustained controls: preliminary findings of the Edinburgh High Risk Study (EHRS). Psychological Medicine,, 29,1161^1173. risk study as a whole and is providing excit- cooperation of the families involved is acknowledged with gratitude, as are the consistenteffortsconsistent efforts of allstaff ing findings, the central features that are Byrne,Byrne,M.,Clafferty,B.A.,Cosway,R., M.,Clafferty, B. A.,Cosway, R., et aletal (2003) and collaborators. We also thank Norma Brearley Neuropsychology, genetic liability and psychotic presented here do not depend on advanced for the careful preparation of the manuscript. symptoms in those at high-risk of schizophrenia. JournalJournal technology. We suggest that this study of Abnormal Psychology,, 112,38^48. shows that with a clear hypothesis, a well- REFERENCES Cannon,T. D., Mednick, S. A., Parnas, J., et aletal (19 94) characterised sample and an appropriate Developmental brain abnormalities in the offspring of design, worthwhile new insights into a Asarnow, J. R. (1988) Children at risk for schizophrenic mothers. II. Structural brain common and crippling disorder can be schizophrenia: converging lines of evidence. characteristics of schizophrenia and schizotypal Schizophrenia Bulletin,, 1414,613^631., 613^631. personality disorder. Archives of General Psychiatry,, 5151,, obtained using simple clinical methods. 955^962.955^962. Boyes, J.,Whalley,H.J.,Whalley, H. C., Lawrie, S. M., et aletal (2001)(2001) AA MRI study of ocular hypertelorism in individuals at high Cornblatt, B & Obuchowski, M. (1997) Update of ACKNOWLEDGEMENTS risk of developing schizophrenia. Schizophrenia Research,, high-risk research: 1987^1997. International Review of 50,1^2. PsychiatryPsychiatry,, 99,,437^447. 437^447.

This study was carried out with the approval of the Byrne, M., Hodges, A., Grant, E., et aletal (1999)(19 9 9) Cosway,R.,Byrne,M.,Clafferty,R.,et aletal (2000) ethical committees of the relevant areas. It was Neuropsychological assessment of young people at high Neuropsychological change in young people at high risk supported financially by two Programme Grants genetic risk for developing schizophrenia compared with for schizophrenia: results from the first two

Downloaded from https://www.cambridge.org/core. 27 Sep 2021 at 03:02:27, subject to the Cambridge Core terms of use. PREDICTING SCHIZOPHRENIA

neuropsychological assessments of the Edinburgh High Lawrie,Lawrie,S.M.,Byrne,M.,Miller,P., S. M., Byrne, M., Miller, P., et aletal (2001(2001bb)) spectrum. A prospective study. Archives of General Risk Study.Study.Risk Psychological Medicine,, 3030, 1111 ^ 1121.1121.,1111 Neurodevelopmental indices and the development of PsychiatryPsychiatry,, 39, 658^664.,658^664. psychotic symptoms in subjects at high risk of Hafner,H.,Hafner, H., Maurer, K., Loffler,W., et aletal (1999)(1999) Onset schizophrenia. British Journal of Psychiatry,, 178178, 524^530. and prodromal phase as determinants of the course. In: Suddath, R. L., Christison, G.W.,G.W.,Torrey, Torrey, E. F., et al Search for the Causes of Schizophrenia (eds W.F.Gattaz & Lawrie, S. M.,Whalley,M., Whalley, H. C., Abukmeil, S. S., et aletal (19 9 0) Anatomical abnormalities in the brains of H. Hafner), pp. 35^58. Berlin: Springer. (2002)(2002) Temporal lobe volume changes in people at high monozygotic twins discordant for schizophrenia. NewNew risk of schizophrenia with psychotic symptoms. BritishBritish England Journal of Medicine,, 322322,789^794., 789^794. Hodges, A., Byrne, M.,M.,Grant, Grant, E., et al (1999)(1999) PeoplePeople Journal of Psychiatry,, 181181,138^143. at risk of schizophrenia. Sample characteristics of the Tarrant,C.J.&Jones,P.B.(1999)Tarrant, C. J. & Jones, P. B. (1999) Precursors to first 100 cases in the Edinburgh high-risk study. BritishBritish Miller, P., Lawrie, S. M., Hodges, A., et aletal (2001)(2001) schizophrenia: do biological markers have specificity? Journal of Psychiatry,, 174174, 547^553.,547^553. Genetic liability, illicit drug use, life stress and psychotic Canadian Journal of Psychiatry,, 44, 335^349. symptoms: preliminary findings from the Edinburgh Job, D. E.,Whalley, H. C.,Yates, S. L., et aletal (2003) Voxel based morphometry of grey matter reductions study of people at high risk for schizophrenia. Social Psychiatry and Psychiatric Epidemiology,, 3636, 338^342.,338^342. Whalley, H. C., Kestelman, J. N., Rimmington, J. E., over time in subjects at high risk of schizophrenia. et aletal (1999)(1999) Methodological issues in volumetric Schizophrenia Research (suppl.), 60,198. Miller, P. M., Byrne, M., Hodges, A., et aletal (2002(2002aa)) magnetic resonance imaging of the brain in the Johnstone, E. C., Abukmeil, S. S., Byrne, M., et aletal Childhood behaviour, psychotic symptoms and psychosis EdinburghHighRiskProject.Psychiatry Research,, 9191,, (2000) Edinburgh high risk study ^ findings after four onset in young people at high risk of schizophrenia: early 31^431^44. 4. years. Demographic, attainment and psychopathological findings from the Edinburgh high risk study. Psychological MedicineMedicine,, 3232,,173^179. 173^179. issues. Schizophrenia Research,, 4646,1^15. Wing, J. K., Cooper, J. E. & Sartorius, N. (1974) TheThe Klosterkotter, J., Hellmich, M., Steinmeyer, E. M., etet Miller, P., Byrne, M., Hodges, A., et aletal (2002bb)) Description and Classification of Psychiatric Symptoms. An al (2001)(2001) Diagnosing schizophrenia in the initial Schizotypal components in people at high risk of Instruction Manual for the PSE and Catego Systems.. prodromal phase. Archives of General Psychiatry,, 58,, developing schizophrenia: early findings from the Cambridge: Cambridge University Press. 158158^164. ^16 4. Edinburgh High-Risk Study. British Journal of Psychiatry,, 180180,179^184.,179^184. WorldHealthOrganization(1992)The ICD ^10^10The Langsley, N., Miller, P., Byrne, M., et aletal (2004)(2004) Classification of Mental and Behavioural Disorders: Clinical Dermatoglyphics and schizophrenia: findings from the Miller,P.M.,Lawrie,S.M.,Byrne,M.,Miller, P.M., Lawrie, S. M., Byrne, M., et aletal (2002(2002cc)) Descriptions and Diagnostic Guidelines..Geneva:WHO. Geneva: WHO. EdinburghHighRiskStudy.Schizophrenia Research,in,in Self-rated schizotypal cognitions, psychotic symptoms press.press. and the onset of schizophrenia in young people at high risk of schizophrenia. Acta Psychiatrica Scandinavica,, 105105,, Yung, A. R., Phillips, L. J.,Yuen, H. P., et aletal (2003)(2003) Lawrie, S. M.,Whalley, H., Kestelman, J. N., et aletal 341^345.341^345. Psychosis prediction: 12-month follow up of a high-risk (19(1999) 9 9) Magnetic resonance imaging of brain in people at (prodromal) group. Schizophrenia Research,, 60,21^32.,21^32. high risk of developing schizophrenia. Lancet,, 353353,30^33.,30^33. Owens, D. G. C., Miller,Miller,P., P., Lawrie, S. M., et aletal (2005) Pathogenesis of schizophrenia ^ a psychopathological Lawrie, S. M.,M.,Whalley, Whalley, H. C., Abukmeil, S. S., et aletal Yung, A. R., Phillips, L. J.,Yuen, H. P., et aletal (2004)(2004) perspective. British Journal of Psychiatry,,inpress. in press. (2001(2001aa)) Brain structure, genetic liability and psychotic Risk factors for psychosis in an ultra high-risk group: symptoms in subjects at high risk of developing Parnas, J., Schulsinger, F., Schulsinger, H., et aletal psychopathology and clinical features. Schizophrenia schizophrenia. Biological Psychiatry,, 49,811^823., 811^823. (19 82) Behavioral precursors of schizophrenia Research,, 6767,131^142.

2525

Downloaded from https://www.cambridge.org/core. 27 Sep 2021 at 03:02:27, subject to the Cambridge Core terms of use.