Schizophrenia Research 237 (2021) 31–39 Contents lists available at ScienceDirect Schizophrenia Research journal homepage: www.elsevier.com/locate/schres Cognitive clusters in first-episode psychosis Silvia Amoretti a, b, c, d, 1, Francisco Diego Rabelo-da-Ponte e, 1, Adriane Ribeiro Rosa e, f, Gisela Mezquida a, b, Ana M. Sánchez-Torres g, h, David Fraguas b, i, Bibiana Cabrera a, b, Antonio Lobo b, j, Ana González-Pinto b, k, Laura Pina-Camacho b, i, Iluminada Corripio b, l, Eduard Vieta b, c, Carla Torrent b, c, Elena de la Serna b, m, Daniel Bergé b, n, o, Miquel Bioque a, b, Marina Garriga b, c, Maria Serra c, Manuel J. Cuesta g, h, Miguel Bernardo a, b, *, PEPs Group a Barcelona Clinic Schizophrenia Unit, Neuroscience Institute, Hospital Clinic of Barcelona, Department of Medicine, University of Barcelona, Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Spain b Biomedical Research Networking Centre in Mental Health (CIBERSAM), Spain c Bipolar and Depressive Disorders Unit, Institute of Neurosciences, University of Barcelona, Barcelona, Catalonia, Spain d Department of Psychiatry, Hospital Universitari Vall d'Hebron, Group of Psychiatry, Mental Health and Addictions, Psychiatric Genetics Unit, Vall d'Hebron Research Institute (VHIR), Barcelona, Spain e Laboratory of Molecular Psychiatry, Hospital de Clínicas de Porto Alegre, Federal University of Rio Grande do Sul, Brazil f Department of Pharmacology, Postgraduate Program in Psychiatry and Behavioral Sciences, Universidade Federal do Rio Grande do Sul (UFRGS), Rua Ramiro Barcelos, Porto Alegre, RS, Brazil g Department of Psychiatry, Complejo Hospitalario de Navarra, Pamplona, Spain h IdiSNA, Navarra Institute for Health Research, Pamplona, Spain i Department of Child and Adolescent Psychiatry, Hospital General Universitario Gregorio Marañón, Instituto de Investigación Sanitaria Gregorio Marañón, School of Medicine, Universidad Complutense de Madrid, Spain j Department of Medicine and Psychiatry, Zaragoza University, Instituto de Investigación Sanitaria Aragón (IIS Aragón), Zaragoza, Spain k Department of Psychiatry, Araba University Hospital, Bioaraba Research Institute, Department of Neurociences, University of the Basque Country, Vitoria, Spain l Psychiatry Department, Institut d'Investigació Biomèdica-Sant Pau (IIB-SANT PAU), Hospital de la Santa Creu i Sant Pau, Universitat Autònoma de Barcelona (UAB), Barcelona, Spain m Department of Child and Adolescent Psychiatry and Psychology, 2017SGR881, Clínic Institute of Neurosciences, Hospital Clínic de Barcelona, IDIBAPS, Department of Medicine, University of Barcelona, Spain n Hospital del Mar Medical Research Institute (IMIM), Barcelona, Spain o Autonomous Universitiy of Barcelona (UAB), Barcelona, Spain A R T I C L E I N F O A B S T R A C T Keywords: Impairments in a broad range of cognitive domains have been consistently reported in some individuals with First episode first-episode psychosis (FEP). Cognitive deficits can be observed during the prodromal stage. However, the Cognitive reserve course of cognitive deficits is still unclear. The aim of this study was to identify cognitive subgroups over time Cognition and to compare their sociodemographic, clinical and functional profiles. A total of 114 patients with Schizo­ Neuropsychology Early intervention phrenia Spectrum Disorders were included in the present study. We assessed subjects through psychiatric scales and eight neuropsychological tests at baseline and at two-year follow-up visit. We performed the Partition Around Medoids algorithm with all cognitive variables. Furthermore, we performed a logistic regression to identify the predictors related to the different cognitive clusters at follow-up. Two distinct subgroups were found: the first cluster characterized by cognitive impairment and a second cluster had relatively intact cognition in comparison with norms. Up to 54.7% of patients with cognitive deficits at baseline tended to improve during the first two years of treatment. Patients with intact cognition at follow-up had a higher socioeconomic status, later age of onset, lower negative symptoms and a higher cognitive reserve (CR) at baseline. CR and age of onset were the baseline variables that predicted cognitive impairment. This research allows us to obtain a better under­ standing of the heterogeneous profile of psychotic disorders. Identifying the characteristics of patients who will * Corresponding author at: Department of Psychiatry and Psychology, Clinical Institute of Neuroscience, Hospital Clinic of Barcelona, Villarroel, 170, 08036 Barcelona, Spain. E-mail address: bernardo@clinic.cat (M. Bernardo). 1 SA and FDRP should be considered joint first authors. Both contributed equally to this work. https://doi.org/10.1016/j.schres.2021.08.021 Received 20 October 2020; Received in revised form 15 July 2021; Accepted 22 August 2021 Available online 1 September 2021 0920-9964/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). S. Amoretti et al. Schizophrenia Research 237 (2021) 31–39 present a cognitive impairment could improve early detection and intervention. These results suggest that enhancing CR could contribute to improving the course of the illness. 1. Introduction performance of mixed neurocognitive deficits (Heinrichs and Awad, 1993; Goldstein et al., 1998; Lewandowski et al., 2014; Burdick et al., Impairment in a broad range of cognitive domains has been consis­ 2014; Van Rheenen et al., 2017). Some differences have been showed in tently reported in individuals with first-episode psychosis (FEP), cognitive subtypes across diagnoses. For example Lewandowski et al. including attention, verbal memory, processing speed, working memory reported that subjects with bipolar disorder with psychosis were over- and executive functioning (Carrión et al., 2018). Cognitive deficits are represented in “neuropsychologically normal” cluster (63%) compared observed during the prodromal stage. The literature suggests that with subjects with schizoaffective disorder (26%) and schizophrenia cognition may be considered a predictor of patients' outcome (Green, (11%) (Lewandowski et al., 2014). Cognitive clusters are also associated 1996), and that cognitive dysfunction is associated with prominent with distinct clinical characteristics. The intact cognition group are functional impairment, which involves social, occupational and inde­ more likely to present predominantly positive symptoms, and better pendent living activities (van Winkel et al., 2007; Hall et al., 2019). premorbid and functional profiles, whereas cognitively impaired cluster Furthermore, a normal cognitive function is strongly associated with groups are more likely to present more prominent negative symptoms clinical improvement in FEP. A systematic review of associations be­ (Heinrichs and Awad, 1993; Goldstein et al., 1998; Lewandowski et al., tween psychotic psychopathology and measures of cognitive impair­ 2018; Sánchez-Torres et al., 2018). ment in subjects with a lifetime history of non-affective psychosis has Although there was evidence of distinct cognitive subgroups of pa­ reported that cognitive deficits are associated with negative and disor­ tients, the differences in clinical, functional and sociodemographic ganized dimensions rather than positive and depressive dimensions characteristics of each cognitive subgroup and their course in SSD sub­ (Dominguez et al., 2009). Other studies showed that patients who jects were not clear. Given their significant impact on psychosocial experienced higher cognitive performance also had a greater reduction functioning and quality of life, identifying differences between people in the severity of negative symptoms (Allott et al., 2011; Rodríguez- with and without cognitive impairment can provide extremely useful Sánchez et al., 2013). Besides, a relationship between neurocognitive information for the definition of personalized interventions. The aims of performance and cognitive reserve (CR) was found (de la Serna et al., this study were 1) to identify cognitive profiles in SSD using Cluster 2013; Anaya et al., 2016; Amoretti et al., 2016, 2018). Analysis at baseline and follow-up; 2) to examine their stability or Several studies in patients with a FEP, schizophrenia or bipolar dis­ movement towards another group at follow-up; and 3) to identify the order suggest that higher CR is associated with a later onset of psychosis predictors related to cognitive clusters at follow-up. and better recovery, as well as it being considered a positive moderator of the impact of pathology on clinical course, functional outcome and 2. Material and methods cognitive performance (de la Serna et al., 2013; Anaya et al., 2016; Amoretti et al., 2016, 2018, 2019, 2020; Herrero et al., 2019). However, 2.1. Participants it has been shown that CR plays a differential role in the outcome of psychoses according to the diagnosis (Amoretti et al., 2018). In The sample of this study came from a multicenter, naturalistic and Schizophrenia Spectrum Disorders (SSD) patients, those with high CR longitudinal project called “Phenotype-genotype interaction: Applica­ were older, had higher socioeconomic status, shorter duration of un­ tion of a predictive model in first psychotic episodes” (PEPs Project) treated psychosis, and a later age of onset. They also showed greater (Bernardo et al., 2013, 2019). A total of 335 patients with a FEP and 253 performance in most cognitive domains. In affective patients, those with healthy controls (HC) were recruited from April 2009 to April 2011. To a greater CR showed a higher socioeconomic status, better functioning, ensure more homogeneous sample diagnoses, for the current study we and greater verbal memory performance (Amoretti et al., 2018). In fact, only included patients with SSD, as we considered affective first-episode the current data support the idea that affective and SSD patients show patients a subgroup displaying several specific characteristics in terms of some differences in clinical aspects, premorbid adjustment and cognitive clinical course, functional outcome and antipsychotic treatments. We function, particularly executive measures such as verbal fluency, which considered SSD diagnoses of schizophrenia, schizophreniform, schizo­ may influence global functioning at follow-up (Torrent et al., 2018). affective disorders and psychoses that are not otherwise specified ac­ Nonetheless, the course of cognitive deficits is still unclear. Different cording to DSM-IV-TR. We included also all those with less than 6% studies have shown that neuropsychological deficits after a FEP missing data in neuropsychological tests at baseline and follow-up for appeared to remain stable over time (Bozikas and Andreou, 2011; the cluster analyses, all the information needed to calculate CR (see Sánchez-Torres et al., 2018), whereas others have shown that cognitive Subsection 2.2. Assessments - Cognitive Reserve Assessment) and, dysfunction deteriorates further (Kurtz, 2005) or improves over time additionally, belonging to the SSD diagnostic category. The final sample (Jahshan et al., 2010). In any case, most of the affected functions seem to for this study consisted of 114 SSD patients and 128 HC. improve modestly after treatment (Hill et al., 2004). Cluster analysis For this study, the inclusion criteria for patients were: 1) between 18 provides an opportunity to group individuals using a data-driven and 35 years of age at the time of first evaluation; 2) presence of psy­ approach and permits individuals to be classified based on their neu­ chotic symptoms of less than twelve months' duration; 3) ability to speak rocognitive profiles. It may throw light on homogeneous phenotypic Spanish correctly; and 4) signed informed consent. Exclusion criteria targets to understand the subtle neuropathologic differences among were: 1) mental retardation according to DSM-IV-TR criteria; 2) history individuals with FEP, since individuals within the same cluster may of head trauma with loss of consciousness; and 3) organic disease with share cognitive, genetic, and neurophysiologic features. This approach mental repercussions. The patients matched with HC age (± 10%), has been widely used to demonstrate specific brain phenotypes, genetic gender and parental socioeconomic status (± 1 level). The exclusion alterations, and cognitive deficits among bipolar disorder and schizo­ criteria for controls were the same as for the patients, yet also included phrenia patients (Green et al., 2020). Different data collected through the presence of a current or past psychotic disorder or major depression cluster studies of patients with schizophrenia and bipolar disorder have and having a first degree relative with psychotic disorder history. shown that there are distinct cognitive subgroups of patients; one with This study was conducted in accordance with the ethical principles of intact cognition or “neuropsychologically normal”, another one with the Declaration of Helsinki and Good Clinical Practice and the Hospital severe and broad impairment and other groups with an intermediate Clinic of Barcelona Ethics and Research Board. All participants provided 32 S. Amoretti et al. Schizophrenia Research 237 (2021) 31–39 written informed consent prior to their inclusion in the study. center. A table of the ratings of WAIS Vocabulary subtest (0 to 2 for each item) was constructed and, for each pair, we consider agreement as a 2.2. Assessments dichotomous variable: “1” for agreement or “0” for disagreement. Those who failed the first evaluation, this mean that intraclass correlation 2.2.1. Clinical and sociodemographic assessment coefficient (ICC) was below 0.80, were reassessed (Bernardo et al., We gathered all the relevant clinical and sociodemographic data for 2013). A good to excellent inter-rater reliability among psychologists all participants. Parental socioeconomic status (SES) was determined was indicated by ICC > 0.80 in the WAIS Vocabulary subtest (Cuesta using Hollingshead's Two-Factor Index of Social Position (Hollingshead et al., 2015). Thus, the inter-judge reliability study guarantees that the and Redlich, 1958); pharmacological treatment was measured by neuropsychologists carried out a correct application and correction of chlorpromazine equivalents (CPZ) based on international consensus the test. (Gardner et al., 2010); and the Duration of Untreated Psychosis (DUP) was calculated as the number of days between the first manifestations of 2.2.4. Cognitive reserve assessment psychotic symptoms until the initiation of adequate treatment for psy­ We have used the three most commonly proposed proxy indicators of chosis. Drug misuse habits were also collected. In order to represent the CR to assess it (de la Serna et al., 2013; Anaya et al., 2016; Amoretti entire population of FEP, participants with current substance abuse/ et al., 2016, 2018, 2020), which include premorbid intellectual func­ dependence comorbid diagnosis were not excluded. tioning (IQ), education and lifetime participation in leisure, social and Diagnoses were determined with the Structured Clinical Interview physical activities. The premorbid IQ was calculated with the vocabu­ for DSM (SCID-I-II) (First et al., 1997a, 1997b) according to DSM-IV lary subtest of the WAIS-III as a measure reflecting premorbid crystal­ criteria. Taking into consideration potential changes across time and lized intelligence. ‘Education’, the second proxy, was assessed taking in order to ensure diagnostic stability, the diagnosis of the patients who into account the number of years of obligatory education that subjects completed the study was determined based on information gathered at completed as well as parents' educational level, and lifetime school 2-year follow-up visit. performance (assessed by PAS scale - scholastic performance). Finally, A psychopathological assessment was carried out with the Spanish lifetime participation in leisure, social and physical activities was validated versions of the following scales: Positive and Negative Syn­ assessed by FAST scale. Higher scores correspond to better performance. drome Scale (PANSS) (Peralta and Cuesta, 1994), the Young Mania Rating Scale (YMRS) (Colom et al., 2002) and the Montgomery-Asberg 2.3. Analysis Depression Rating Scale (MADRS) (Lobo et al., 2002). Higher scores indicate greater severity. 2.3.1. Pre-processing Traumatic events were assessed by the Traumatic Experiences in All cognitive variables were converted into T-score based on Psychiatric Questionnaire (Davidson and Smith, 1990), which was normative data of the general population, which comes from the coded as a dichotomous variable (yes/no). normative table of the validation studies of each cognitive test. We performed all the steps of this section in R (Version 3.5.3) and Rstudio 2.2.2. Functional assessment (Version 1.1.44) (https://www.Rproject.org/). We used Multivariate The overall functional outcome was assessed by means of the Func­ Imputation by Chained Equations algorithm (Van Buuren and tioning Assessment Short Test (FAST) (Rosa et al., 2007) and The Global Groothuis-Oudshoorn, 2011) to impute missing data, setting the number Assessment of Functioning (GAF) (Endicott et al., 1976). The FAST scale of iterations equal to 50 and used the Predictive Mean Matching comprises 24 items and higher scores indicate worse functioning. They Imputation (PMM) method for all cognitive variables. All cognitive are divided according to six specific areas of functioning: autonomy, variables contained less than 6% and 2% missing data at baseline and 2- occupational functioning, cognitive functioning, financial issues, inter­ years follow-up, respectively (Supplementary Fig. S1). We needed to personal relationships, and leisure time. The GAF is a scale designed to impute data because the clustering algorithm does not permit missing assess the severity of symptoms and the level of functioning. Higher values (Supplementary Fig. S2). scores correspond to better functioning. The Premorbid Adjustment Scale (PAS) (Cannon-Spoor et al., 1982) 2.3.2. Machine learning technique was applied retrospectively to assess premorbid adjustment. Only After, we performed the Partition Around Medoids (PAM) algorithm childhood and early adolescence life periods have been taken into ac­ (Schubert and Rousseeuw, 2019) to identify cognitive clusters of sub­ count since they are the two periods answered by all the participants. jects with FEP at baseline and 2-years follow-up using all neuropsy­ Higher scores on the test indicate worse premorbid adjustment. chological tests (TMT-A, TAVEC, LN, Stroop, FAS, and Animal), that is, we performed cluster analysis in only patient group. We decided to use 2.2.3. Neuropsychological assessment PAM algorithm rather than k-means, a classical clustering algorithm, The neuropsychological assessments were performed in the second because PAM is more robust to noise and outliers, minimizing the sum of month of evaluation in order to ensure the clinical stability of patients dissimilarities between data points in place of a sum of squared and were repeated in the two-year follow-up visit. The neuropsycho­ Euclidean distances. We used Gower's distance (Gower, 2012) to logical battery measured the following cognitive domains: 1) Processing calculate the dissimilarities between pairs of subjects. speed was tested with the Trail Making Test, form A (TMT-A) (Reitan The optimal number of clusters was determined by the average and Wolfson, 1993); 2) Verbal learning and memory, assessed with the silhouette width. The silhouette indicates how well objects are clustered, Verbal Learning Test Spain Complutense for adults (TAVEC) (Benedet, ranging from − 1 (poorly) to 1 (well clustered). Hence, the algorithm 1998); 3) Working memory was assessed with the Digit Span Subtest and suggests the number optimal of clusters based on the number of sub­ the Letter-Number (LN) Sequencing Subtest of the Wechsler Adult In­ groups with the highest value for the average silhouette width. Dunn telligence Scale (WAIS-III) (Wechsler, 1997); 4) The executive functions index means the ratio between compactness within cluster and separa­ were evaluated using the Stroop Test, word-color interference effect tion between clusters (Hassani and Seidl, 2017; Tomasini et al., 2017). (Golden and Freshwater, 1978); Controlled Oral Word Association Test We performed Dunn index to verify the internal quality of clustering at composed by phonemic verbal fluency: FAS, and semantic fluency: an­ baseline and follow-up. The packages used were “mice”, “cluster”, imal naming) (Peña-Casanova, 1990; Loonstra et al., 2001). Higher T- “factorextra”, “NbClust”, and “fpc”. scores correspond to better performance in all the cognitive domains. To evaluate the differences between raters, an interrater reliability 2.3.3. Statistical analysis study was also conducted among different neuropsychologists at each We used Statistical Package for the Social Sciences (SPSS) (Version 33 S. Amoretti et al. Schizophrenia Research 237 (2021) 31–39 19) to perform statistical analysis. Descriptive analyses were conducted average Silhouette width equals to 0.71, and Dunn index equal to 0.75 at using chi-square for categorical variables and Student's test for contin­ baseline. At 2-years follow-up, the measures were 0.77, and 0.80, uous variables. Demographic, clinical and neuropsychological differ­ respectively (Supplementary Fig. S3). The average Silhouette width is ences between the groups were examined and effect sizes were reported. used to determine the optimal number of clusters and the Dunn index is Cohen's d was used to indicate the standardized difference between two used for internal validation. means, Odds ratio (OR) was used for 2 × 2 contingency table analyses At baseline, two distinct clusters were found: the first cluster with and Cramer's V accounts for multi-categorical variables. A Principal mild to moderate cognitive impairments in processing speed, verbal Components Analysis (PCA) was performed to create a “Cognitive learning, working memory and verbal fluency (executive function task) reserve score” for each subject with the three main proxies. (n = 64, 56.1%) and the second one with relatively intact cognition Backward stepwise logistic regression analyses were performed in (results within the typical limits in all scores on standardized tests) (n = this study to identify the factors related to cognitive clusters at follow- 50, 43.9%). Table 1 contains a summary of the baseline characteristics up. The threshold for statistical significance was p < 0.05. of patients and the differences between clusters. The relatively cogni­ tively intact group performed worse than HC on LN (p = 0.032), verbal 3. Results memory (p < 0.001) and verbal fluency (Animals, p = 0.001 and FAS, p < 0.001). No significant differences were found on Digits (p = 0.087), 3.1. Sociodemographic characteristics of the sample TMT-A (p = 0.566) and Stroop (p = 0.120). At 2-year follow-up, the same two clusters were maintained: one A total of 114 patients with SSD and 128 HC were included in the with cognitive impairment (n = 36, 31.6%) and another one with rela­ present study and were re-evaluated at 2-year follow-up. Sixty-seven tively intact cognitive function (n = 78, 68.4%). As shown in Fig. 1, 63% percent of patients were male with a mean age of 25.22 years (age at of patients (n = 72) were unchanged, showing a certain degree of onset of psychosis was 25 years). The mean dose of antipsychotic cognitive stability over time. Thirty-five patients (30.7% of the whole medication was equivalent to 596.40 ± 408.96 mg/day of CPZ and the sample and 54.7% of those with cognitive impairment at baseline) mean of DUP was determined as 109.40 ± 126.19 days. See Supple­ improved their cognitive performance, crossing from cognitive impair­ mentary Table 1 for details of cognitive characteristics of patients and ment to the relatively cognitively intact group after two years, while HC, and the differences between groups. seven patients (6.1% of the whole sample and 14% of those with intact cognition) crossed from the relatively cognitively intact group to the 3.2. Cognitive clusters in SSD patients and their stability cognitive impairment cluster at follow-up. See Supplementary Table 2 for details of sociodemographic, clinical, functional and cognitive In relation to clustering performance, PAM algorithm achieved an characteristics of patients and the differences between groups. The Table 1 Baseline sociodemographic, clinical, functional and cognitive reserve for non-affective patients and among clusters. Cognitive impairment (n = 64) Relatively cognitively intact (n = 50) P Cohen's d/Odds Ratio/Cramer's V Sociodemographic variables Gender: Male N (%) 40 (63) 36 (72) 0.286 OR = 1.54 Age (x±SD) 24.44 ± 5.16 26.22 ± 5.06 0.068 0.35 SES (%) 0.002 Cramer's V = 0.39 High 12 (19) 14 (28) Medium-High 1 (2) 11 (22) Medium 15 (23) 8 (16) Medium-Low 26 (41) 15 (30) Low 10 (16) 2 (4) DUP 128.41 ± 144.72 86.42 ± 95.85 0.088 0.34 Age of onset 24.11 ± 5.26 26.11 ± 5.67 0.135 0.37 CPZ 641.98 ± 449.45 531.75 ± 338.06 0.177 0.28 Tobacco: Yes N (%) 44 (68) 31 (62) 0.289 OR = 0.74 Can-bis: Yes N (%) 32 (50) 18 (36) 0.096 OR = 0.56 Trauma: Yes N (%) 37 (58) 29(58) 0.961 OR = 1.02 Clinical variables PANSS positive 19.13 ± 8.58 16.64 ± 7.05 0.100 0.32 PANSS negative 20.98 ± 7.65 17.04 ± 6.55 0.004 0.55 PANSS general 38.34 ± 12.40 36.28 ± 12.28 0.378 0.17 PANSS total 78.45 ± 25.32 69.96 ± 22.73 0.066 0.35 YMRS 8.02 ± 9.88 5.90 ± 8.11 0.223 0.23 MADRS 11.33 ± 10.10 11.78 ± 8.46 0.800 0.05 FAST 31.17 ± 16.96 24.62 ± 14.61 0.032 0.41 GAF 52.84 ± 19.12 56.68 ± 17.50 0.272 0.21 Neuropsychological performance Cognitive reserve 70.55 ± 10.12 81.42 ± 9.85 <0.001 1.01 LN 39.17 ± 9.04 48.94 ± 10.32 <0.001 1.01 Digits 41.03 ± 7.24 48.92 ± 8.51 <0.001 1.00 TAVEC 30.78 ± 11.45 46.40 ± 9.64 <0.001 1.48 TMT-A 33.94 ± 12.92 47.74 ± 11.33 <0.001 1.18 Stroop interference 50.11 ± 6.64 52.98 ± 10.90 0.105 0.32 FAS 34.27 ± 5.91 44.64 ± 7.82 <0.001 1.50 Animal 37.83 ± 9.12 48.56 ± 9.58 <0.001 1.15 Abbreviations: SES=Socioeconomic status, DUP = Duration of Untreated Psychosis, CPZ = Chlorpromazine equivalents, PANSS = Positive and Negative Symptom Scale, YMRS = Young Mania Rating Scale, MADRS = Montgomery-Asberg Depression Rating Scale, FAST = Functioning Assessment Short Test, GAF = Global Assessment of Functioning, LN = Letter-Number, TAVEC = Verbal Learning Test Spain Complutense for adults, TMT-A = Trail Making Test, form A, FAS = Phonemic Verbal Fluency. Significant differences (p < 0.05) marked in bold. 34 S. Amoretti et al. Schizophrenia Research 237 (2021) 31–39 Fig. 1. Longitudinal change pattern of clusters. relatively cognitively intact group performed worse than HC on all evaluated, except for Interference Stroop (executive task), demon­ cognitive domains evaluated. strating a good split among groups (see Fig. 2). There were no differ­ ences between clusters at baseline in terms of age, gender, DUP, age of 3.3. Cluster characteristics and follow-up predictors onset, CPZ, tobacco or cannabis use and history of traumatic events/ experiences (see Table 1). For clusters classified after two years of When comparing the clinical and sociodemographic profiles we follow-up, subjects with relatively intact cognition had higher SES, later found significant differences in SES, baseline negative symptoms, age of onset, lower baseline negative symptoms, lower scores of PANSS functional outcomes (autonomy of FAST scale, p = 0.020, and total FAST follow-up in all subscales and a higher CR at baseline (see Table 2). They scale, p = 0.032) and CR between both clusters classified at baseline. showed differences in all areas of functioning (autonomy (p < 0.001), They also showed significant differences in the entire cognitive subtest cognitive functioning (p = 0.001), financial issues (p = 0.022), Fig. 2. Cluster distribution at baseline and 2-year follow-up visit. a) Cognitive performance among cognitive clusters during baseline and follow-up. Each line represents the mean of cluster. b) Cluster distribution at baseline. c) Cluster distribution at 2-year follow-up. The graph “b” and “c” are 2D visualizations of the unsupervised reduced dimension representation through t-Distributed Stochastic Neighbor Embedding (t-SNE) (van der Maaten and Hinton, 2008). Each axis means a non-linear reduction of cognitive tests and each point is a subject. 35 S. Amoretti et al. Schizophrenia Research 237 (2021) 31–39 Table 2 Sociodemographic, clinical, functional and cognitive performance among clusters at 2-year follow-up. Cognitive impairment (n = Relatively cognitively intact (n = p Cohen's d/Odds Ratio/Cramer's Healthy p 36) 78) V controls Sociodemographic variables Gender: Male N (%) 25 (69) 51 (63) 0.419 OR = 1.71 84 (66) 0.448 Age (x±SD) 24.78 ± 4.99 25.42 ± 5.27 0.531 0.14 25.77 ± 5.83 0.244 SES (%) 0.002 Cramer's V = 0.381 0.119 High 5 (8) 21 (42) 33 (26) – Medium-High 0 (0) 12 (24) 29 (23) – Medium 5 (8) 18 (36) 29 (23) – Medium-Low 19 (30) 22 (44) 30 (23) – Low 7 (11) 5 (10) 7 (5) – DUP 133.11 ± 118.96 97.70 ± 128.80 0.175 0.29 – – Age of onset 22.77 ± 5.15 26.02 ± 6.20 0.022 0.57 – – CPZ 677.91 ± 425.63 561.67 ± 399.10 0.216 0.28 – – CPZ follow-up 494.48 ± 276.76 391.31 ± 220.40 0.152 0.41 – – Tobacco: Yes N (%) 24 (67) 51 (65) 0.534 OR = 0.94 52 (41) <0.001 Tobacco follow-up 23 (64) 47 (60) 0.149 OR = 0.82 44 (34) 0.001 Can-bis: Yes N (%) 17 (47) 33 (42) 0.386 OR = 1.29 25 (20) <0.001 Can-bis follow-up 6 (17) 9 (12) 0.093 OR = 0.65 25 (20) 0.141 Trauma: Yes N (%) 18 (50) 48 (63) 0.132 OR = 1.71 68 (53) 0.268 Clinical variables PANSS positive 18.94 ± 7.81 17.62 ± 8.11 0.412 0.17 – – PANSS positive follow-up 11.90 ± 5.34 9.74 ± 3.54 0.016 0.48 – – PANSS negative 21.53 ± 7.91 18.21 ± 6.99 0.026 0.45 – – PANSS negative follow-up 17.23 ± 6.26 12.93 ± 5.70 0.001 0.82 – – PANSS general 39.28 ± 11.41 36.59 ± 12.72 0.281 0.23 – – PANSS general follow-up 27.61 ± 9.99 24.41 ± 7.73 0.077 0.36 – – PANSS total 79.75 ± 24.05 72.41 ± 24.49 0.137 0.30 – – PANSS total follow-up 56.74 ± 19.50 47.08 ± 15.38 0.008 0.55 – – YMRS 6.72 ± 8.32 7.26 ± 9.59 0.774 0.06 – – YMRS follow-up 2.39 ± 4.23 1.26 ± 2.56 0.095 0.32 – – MADRS 11.19 ± 9.08 11.68 ± 9.57 0.799 0.05 – – MADRS follow-up 7.61 ± 7.37 5.03 ± 5.98 0.062 0.38 – – FAST 30.61 ± 16.27 27.23 ± 16.21 0.303 0.21 2.19 ± 5.16 <0.001 FAST follow-up 25.39 ± 15.09 15.64 ± 13.71 0.002 0.67 2.29 ± 6.63 <0.001 GAF 55.72 ± 16.69 53.97 ± 19.27 0.640 0.10 93.48 ± 5.00 <0.001 GAF follow-up 68.52 ± 11.61 75.14 ± 12.76 0.017 0.54 92.57 ± 4.14 <0.001 Neuropsychological performance Cognitive Reserve 69.36 ± 10.73 78.08 ± 10.47 <0.001 0.82 89.99 ± 10.01 <0.001 LN 37.39 ± 9.27 46.25 ± 10.25 <0.001 0.91 52.53 ± 9.49 <0.001 LN follow-up 40.28 ± 8.89 48.53 ± 10.14 <0.001 0.87 52.77 ± 10.76 <0.001 Digits 41.22 ± 7.89 46 ± 8.72 0.006 0.57 53.63 ± 9.44 <0.001 Digits follow-up 42.14 ± 8.49 47.40 ± 8.49 0.003 0.61 52.62 ± 9.18 <0.001 TAVEC 30.56 ± 13.72 40.90 ± 11.64 <0.001 0.81 53.31 ± 9.35 <0.001 TAVEC follow-up 26.39 ± 6.39 48.97 ± 8.31 <0.001 3.05 56.59 ± 9.31 <0.001 TMT-A 34.83 ± 13.24 42.37 ± 13.78 0.007 0.56 49.00 ± 13.88 <0.001 TMT-A follow-up 37.83 ± 12.44 47.52 ± 10.81 <0.001 0.83 54.19 ± 8.65 <0.001 Stroop Interference 48.06 ± 8.04 52.90 ± 8.82 0.006 0.57 55.51 ± 9.50 0.004 Stroop Interference follow- 50.25 ± 9.14 54.45 ± 7.56 0.011 0.50 57.11 ± 8.00 0.001 up FAS 35.50 ± 7.38 40.34 ± 8.62 0.004 0.60 49.13 ± 8.97 <0.001 FAS follow-up 34.64 ± 7.46 45.40 ± 9.44 <0.001 1.27 51.47 ± 9.49 <0.001 Animal 37.53 ± 10.60 44.85 ± 10.01 0.001 0.71 57.81 ± 12.06 <0.001 Animal follow-up 36.97 ± 8.38 49.10 ± 10.35 <0.001 1.29 59.41 ± 12.07 <0.001 Abbreviations: SES=Socioeconomic status, DUP = Duration of Untreated Psychosis, CPZ = Chlorpromazine equivalents, PANSS = Positive and Negative Symptom Scale, YMRS = Young Mania Rating Scale, MADRS = Montgomery-Asberg Depression Rating Scale, FAST = Functioning Assessment Short Test, GAF = Global Assessment of Functioning, LN = Letter-Number, TAVEC = Verbal Learning Test Spain Complutense for adults, TMT-A = Trail Making Test, form A, FAS = Phonemic Verbal Fluency. Significant differences (p < 0.05) marked in bold. interpersonal relationships (p = 0.008), and leisure time (p = 0.004)) membership at follow-up. The results showed that the adjusted logistic except occupational functioning (p = 0.381). They also differed in terms regression had a good model fit (x2 = 14.192, p < 0.001, Nagelkerke R2 of neurocognitive performance, showing a lower cognitive functioning = 0.221) and allowed for a correct classification of 75.9% of the cases. in all subtests evaluated at baseline and follow-up in patients with However, it is possible that changes in cognitive performance may be cognitive impairment than those with intact cognitive functioning. related to changes in the severity of symptoms or functionality. For that To identify the predictors related to cognitive clusters at follow-up a reason, we included in the subsequent model these clinical and func­ backward stepwise logistic regression analysis was performed. The tional changes as a predictors. CR (OR = 1.071, 95%CI [1.005–1.141], p cluster group was the outcome variable and the predictors included in = 0.035), FAST at baseline (OR = 0.948, 95%CI [0.900–0.999], p = the model were those that showed statistically significant differences 0.046), age of onset (OR = 1.133, 95%CI [1.012–1.268], p = 0.031), between the groups at baseline in the univariate analysis: SES, age of changes in FAST (OR = 0.931, 95%CI [0.883–0.983], p = 0.009) and onset, negative symptoms and CR. Only CR (OR = 1.073, 95%CI changes in negative symptoms (OR = 1.161, 95%CI [1.035–1.303], p = [1.020–1.130], p = 0.007) was significantly associated with cluster 0.011) were significantly associated with cluster cognition at follow-up. 36 S. Amoretti et al. Schizophrenia Research 237 (2021) 31–39 The results showed that the adjusted logistic regression had also a good in which the CR and cognitive performance has been analyzed in a FEP model fit (x2 = 26.369, p < 0.001, Nagelkerke R2 = 0.446) and allowed sample (de la Serna et al., 2013; Amoretti et al., 2016, 2018, 2020). for a correct classification of 83.6% of the cases. Together with previous studies, our results lead us to consider that it can be very helpful to evaluate CR as it may aid in the stratification of pa­ 4. Discussion tients with FEP. Based on these results, if conducted in the early stages of the illness or even on people with a high risk of suffering psychosis, the Four main findings emerged from the present study: 1) There are two implementation of early interventions centered on CR stimulation and well defined cognitive clusters at baseline in patients with SSD (cogni­ engaging lifestyle could be beneficial in preventing or reducing the tive impairment and relatively intact cognition); 2) At follow-up, the impact of illness. However, future and longer follow-up studies are same clusters were found showing a certain degree of cognitive stability needed to further explore the implementation of these interventions. among FEP. However, 35/64 (54.7%) of them improved their perfor­ Also, as suggested in earlier papers (Amoretti et al., 2018, 2019), pa­ mance (changed from the cognitively impaired cluster to the cognitively tients with a FEP and low CR could benefit from a cognitive rehabili­ intact group) and only 7/50 (14%) patients crossed from the intact tation program. Since cognitive deficits predicted long-term group to that with cognitive impairment; 3) The cognitively impaired functioning, they serve as natural targets. Nevertheless, this study em­ group at two-year follow-up was more likely to present lower SES, early phasizes those with low CR who will present a cognitive impairment. age of onset, more prominent negative symptoms (baseline), greater Because previous studies have shown that patients who demonstrate total PANSS score (follow-up) and lower CR; 4) Finally, CR was the only a better cognitive performance also display a greater reduction in the variable that predicted cognitive impairment after adjusted analysis. severity of negative symptoms (Milev et al., 2005; Rodríguez-Sánchez However, when clinical and functional changes were included in the et al., 2013) and improvements in cognitive domains seem to be model, CR, FAST at baseline, age of onset, changes in FAST and changes significantly correlated with better functioning in SSD patients (Rodrí­ in negative symptoms were significantly associated with cluster cogni­ guez-Sánchez et al., 2013), we introduced changes in the severity of tion at follow-up. symptoms and functionality to the predictive model. The results showed Previous publications have reported evidence of a cognitively intact that the cognitive functioning of our patients at follow-up could be cluster in patients with psychosis that did not differ from normative attributed to CR, age of onset and improvement in the severity of means (Uren et al., 2017; Lewandowski et al., 2018). In a FEP sample, negative symptoms and functioning. These results suggest two impor­ Uren et al. (2017) described three clusters (widespread cognitive im­ tant findings that should be further explored: 1) The concept of CR has pairments, moderately impaired cognitive functioning, and pattern of been defined as the ability of a brain to cope with brain pathology in cognitively intact performance across all domains). Furthermore, Lew­ order to minimize symptoms (Stern, 2002). Thus, probably patients with andowski et al. (2018) have recently demonstrated 4 cognitive clusters relatively intact cognition may also have experienced neuroprogressive in patients with psychosis (a neuropsychologically normal cluster, a effects, but starting from a higher baseline due to CR (Lewandowski globally impaired cluster, and two clusters of mixed profiles). The et al., 2018). Besides, CR, cognition and functionality are associated strengths of our study were that the FEP sample was very well charac­ concepts and the results obtained confirm this association; and 2) terized and the period of follow-up was longer than the previously Improvement in the severity of negative symptoms and psychosocial mentioned studies (24 months vs. 6 months or without follow-up). Our functioning in early stages could predict better cognitive performance. results showed that relatively cognitively intact group showed a worse Thus, we consider that early implementation of cognitive-behavioral cognitive profile than HC, thus, it cannot be considered as an intact therapy interventions could be beneficial to reducing the impact of cluster as that group performs in the normal range but there is no illness (Granholm et al., 2018). confirmation of them not having declined from premorbid levels. At 2- Some limitations of our work should be taken into consideration year follow-up, the same two groups were maintained. Findings in our before translating these findings into clinical practice. Firstly, a limita­ sample showed that up to 54.7% of patients with cognitive deficits at tion present in all CR studies undertaken on a psychiatric population is baseline tended to improve during the first two years of treatment and at that at this time there was no validated instrument to measure CR so follow-up just 31.6% of patients presented cognitive impairment. Thus, criteria established and replicated in previous studies were followed. In these results suggest that cognitive impairment is relatively stable after a 2019, the Cognitive Reserve Assessment Scale in Health (CRASH) FEP, but when patients do show changes they are more likely to improve (Amoretti et al., 2019) has been validated. It is the first measure than to decline. In fact, longitudinal studies in first episode patients have developed specifically for patients with severe mental illness with shown that most of the functions affected seem to improve modestly optimal psychometric properties, facilitating reliable and valid mea­ after the start of treatment (Hill et al., 2004; Haring et al., 2017). Our surement of CR. A second limitation is the diagnostic time instability of study suggests that in SSD patients, those subjects with a lower SES, the FEP. However, the evidence suggests a high prospective consistency early age of onset, higher negative symptoms, lower CR and worse for schizophrenia and bipolar disorder (Fusar-Poli et al., 2016). The cognitive performance have an increased risk of cognitive impairment at diagnosis was established based on data collected at a two-year follow- follow-up, which is in accordance with previous publications in which a up visit. In spite of its limitations, the study shows innovative and sig­ large correlation between CR, SES, negative symptoms and cognition nificant results that can be implemented in daily clinical practice. have also been reported (Puig et al., 2017; Amoretti et al., 2018). In summary, our results showed that CR and age of onset were Therefore, identifying differences between people with and without identified as the only baseline variables that predicted cognitive profile. cognitive impairment can provide extremely useful information in order This research allowed us to obtain a better understanding of the het­ to define personalized interventions in FEP (Sánchez-Torres et al., erogeneous profile of FEP and also suggested that CR measure should be 2018). Thus, the improvement could be attributed to protective factors considered as a tool that is supplemental to the comprehensive assess­ (CR and age of onset) and the effects of treatment (greater reduction in ment of this group of patients and may be useful for prognosis and the severity of negative symptoms and better functioning), which are in treatment. Future research should be conducted to explore cognitive line with the findings of longitudinal studies performed in adults with profiles using Cluster Analysis at baseline and follow-up according to FEP (Davidson et al., 2009). diagnosis (SSD vs. affective FEP). Despite the fact that several variables have been associated with cognition, CR was the only variable at baseline that predicted cognitive Funding impairment at follow-up and allowed for a correct classification of 75.9% of the cases. These patients with relatively intact cognition pro­ This research received no external funding. files exhibited greater CR, which is in accordance with previous studies 37 S. Amoretti et al. Schizophrenia Research 237 (2021) 31–39 CRediT authorship contribution statement Anmella, G., Rosa, A.R., Contreras, F., Safont, G., Vieta, E., Bernardo, M., 2019. Cognitive reserve assessment scale in health (CRASH): its validity and reliability. J. Clin. Med. 28, 8. SA and FDR designed the study, managed the literature searches and Amoretti, S., Rosa, A.R., Mezquida, G., Cabrera, B., Ribeiro, M., Molina, M., Bioque, M., analyses, undertook the statistical analysis, and wrote the first draft of Lobo, A., González-Pinto, A., Fraguas, D., Corripio, I., Vieta, E., de la Serna, E., the manuscript. Morro, L., Garriga, M., Torrent, C., Cuesta, M.J., Bernardo, M., PEPs Group, 2020. The impact of cognitive reserve, cognition and clinical symptoms on psychosocial ARR, EV and MBe revised the first draft and added critical comments functioning in first-episode psychoses. Psychol. Med. 1–12. to guide the redaction of the final manuscript. Anaya, C., Torrent, C., Caballero, F.F., Vieta, E., Bonnín, C.M., Ayuso-Mateos, J.L., GM, AMS, DF, BC, AL, AG, LP, IC, CT, ES, DB, MBi, MG, MS and MJC CIBERSAM Functional Remediation Group, 2016. Cognitive reserve in bipolar disorder: relation to cognition, psychosocial functioning and quality of life. Acta revise the second draft of the article and provided critical comments to Psychiatr. Scand. 133 (5), 386–398. guide the redaction of the final manuscript. Benedet, M.J., 1998. Test de Aprendizaje Verbal España-Complutense (TAVEC). TEA All the authors within the PEPs Group recluted patients and healthy Ediciones, Madrid. Bernardo, M., Bioque, M., Parellada, M., Saiz Ruiz, J., Cuesta, M.J., Llerena, A., controls at their centers, provided the anonymous data and revise the Sanjuán, J., Castro-Fornieles, J., Arango, C., Cabrera, B., 2013. Assessing clinical and final manuscript. functional outcomes in a gene-environment interaction study in first episode of All authors approved the final manuscript. psychosis (PEPs). Rev. Psiquiatr. Salud Ment. 6 (1), 4–16. Bernardo, M., Cabrera, B., Arango, C., Bioque, M., Castro-Fornieles, J., Cuesta, M.J., Lafuente, A., Parellada, M., Saiz-Ruiz, J., Vieta, E., 2019. One decade of the first Declaration of competing interest episodes project (PEPs): advancing towards a precision psychiatry. Rev. Psiquiatr. Salud Ment. 12 (3), 135–140. Bozikas, V.P., Andreou, C., 2011. Longitudinal studies of cognition in first episode MBi has received honoraria from talks and consultancy of Adamed, psychosis: a systematic review of the literature. Aust. N. Z. J. Psychiatry 45 (2), has received honoraria from consultancy of Ferrer, has received research 93–108. Burdick, K.E., Russo, M., Frangou, S., Mahon, K., Braga, R.J., Shanahan, M., Malhotra, A. support and honoraria from talks and consultancy of Janssen-Cilag, has K., 2014. Empirical evidence for discrete neurocognitive subgroups in bipolar received honoraria from talks and consultancy of Lundbeck, has disorder: clinical implications. Psychol. Med. 44 (14), 3083–3096. received honoraria from talks and consultancy of Otsuka, and a research Cannon-Spoor, H., Potkin, S., Wyatt, R., 1982. Measurement of premorbid adjustment in chronic schizophrenia. Schizophr. Bull. 8 (3), 470–484. prize from Pfizer. Carrión, R.E., Walder, D.J., Auther, A.M., McLaughlin, D., Zyla, H.O., Adelsheim, S., MBe has been a consultant for, received grant/research support and Calkins, R., Carter, C.S., McFarland, B., Melton, R., Niendam, T., Ragland, J.D., honoraria from, and been on the speakers/advisory board of ABBiotics, Sale, T.G., Taylor, S.F., McFarlane, W.R., Cornblatt, B.A., 2018. From the psychosis prodrome to the first-episode of psychosis: no evidence of a cognitive decline. Adamed, Angelini, Casen Recordati, Janssen-Cilag, Lundbeck, Otsuka, J. Psychiatr. Res. 96, 231–238. Menarini and Takeda. Colom, F., Vieta, E., Martinez-Aran, A., Garcia-Garcia, M., Reinares, M., Torrent, C., RR has been a consultant for, spoken in activities of, or received Goikolea, J.M., Banús, S., Salamero, M., 2002. Spanish version of a scale for the assessment of mania: validity and reliability of the young mania rating scale. Med. grants from: Instituto de Salud Carlos III, Fondo de Investigación Sani­ Clin. 119, 366–371. taria (FIS), Centro de Investigación Biomédica en Red de Salud Mental Cuesta, M.J., Sánchez-Torres, A.M., Cabrera, B., Bioque, M., Merchán-Naranjo, J., (CIBERSAM), Madrid Regional Government (S2010/BMD-2422 AGES; Corripio, I., González-Pinto, A., Lobo, A., Bombín, I., de la Serna, E., Sanjuan, J., S2017/BMD-3740), JanssenCilag, Lundbeck, Otsuka, Pfizer, Ferrer, Parellada, M., Saiz-Ruiz, J., Bernardo, M., PEPs Group, 2015. Premorbid adjustment and clinical correlates of cognitive impairment in first-episode psychosis. The Juste, Takeda, Exeltis, Angelini. PEPsCog Study. Schizophr. Res. 164 (1-3), 65–73. MGu has been on the speakers/advisory board of Janssen-Cilag. Davidson, J., Smith, R., 1990. Traumatic experiences in psychiatric outpatients. EV has received grants and served as consultant, advisor or CME J. Trauma. Stress. 3, 459–475. Davidson, M., Galderisi, S., Weiser, M., Werbeloff, N., Fleischhacker, W.W., Keefe, R.S., speaker for the following entities: AB-Biotics, Abbott, Allergan, Angelini, Boter, H., Keet, I.P., Prelipceanu, D., Rybakowski, J.K., Libiger, J., Hummer, M., AstraZeneca, Bristol-Myers Squibb, Dainippon Sumitomo Pharma, Dollfus, S., Hummer, M., Dollfus, S., López-Ibor, J.J., Hranov, L.G., Gaebel, W., Farmindustria, Ferrer, Forest Research Institute, Gedeon Richter, Glaxo- Peuskens, J., Lindefors, N., Riecher-Rössler, A., Kahn, R.S., 2009. Cognitive effects of antipsychotic drugs in first-episode schizophrenia and schizophreniform disorder: a Smith-Kline, Janssen, Lundbeck, Otsuka, Pfizer, Roche, SAGE, Sanofi- randomized, open-label clinical trial (EUFEST). Am. J. Psychiatry 166 (6), 675–682. Aventis, Servier, Shire, Sunovion, Takeda, the Brain and Behaviour De la Serna, E., Andrés-Perpiñá, S., Puig, O., Baeza, I., Bombin, I., Bartrés-Faz, D., Foundation, the Generalitat de Catalunya (PERIS), the Spanish Ministry Arango, C., Gonzalez-Pinto, A., Parellada, M., Mayoral, M., Graell, M., Otero, S., Guardia, J., Castro-Fornieles, J., 2013. Cognitive reserve as a predictor of two year of Science and Innovation (CIBERSAM), EU Horizon 2020, and the neuropsychological performance in early onset first-episode schizophrenia. Stanley Medical Research Institute. BS declares no conflict of interest Schizophr. Res. 143 (1), 125–131. related to this manuscript. Dominguez, Mde.G., Viechtbauer, W., Simons, C.J., van Os, J., Krabbendam, L., 2009. Are psychotic psychopathology and neurocognition orthogonal? A systematic review The rest of authors report no biomedical financial interests or po­ of their associations. Psychol. Bull. 135 (1), 157–171. tential conflicts of interest. Endicott, J., Spitzer, R.L., Fleiss, J.L., Cohen, J., 1976. The global assessment scale. a procedure for measuring overall severity of psychiatric disturbance. Arch. Gen. Psychiatry 33 (6), 766–771. Appendix A. Supplementary data First, M., Gibbon, M., Spitzer, R., Williams, J., Benjamin, L., 1997a. Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II). American Psychiatric Supplementary data to this article can be found online at https://doi. Press, Washington DC. First, M., Spitzer, R., Gibbon, M., Williams, J., 1997b. Structured Clinical Interview for org/10.1016/j.schres.2021.08.021. DSM-IV Axis I Disorders-Clinician (SCID-I). American Psychiatric Press, Washington DC. References Fusar-Poli, P., Cappucciati, M., Rutigliano, G., Heslin, M., Stahl, D., Brittenden, Z., Caverzasi, E., McGuire, P., Carpenter, W.T., 2016. Diagnostic stability of ICD/DSM first episode psychosis diagnoses: meta-analysis. Schizophr. Bull. 42 (6), 1395–1406. Allott, K., Liu, P., Proffitt, T.M., Killackey, E., 2011. Cognition at illness onset as a Gardner, D., Murphy, A., O'Donnell, H., Centorrino, F., Baldessarini, R.J., 2010. predictor of later functional outcome in early psychosis: systematic review and International consensus study of antipsychotic dosing. Am. J. Psychiatry 167 (6), methodological critique. Schizophr. Res. 125 (2–3), 221–235. 686–693. Amoretti, S., Bernardo, M., Bonnin, C.M., Bioque, M., Cabrera, B., Mezquida, G., Solé, B., Golden, C.J., Freshwater, S.M., 1978. The Stroop Color and Word Test: A Manual for Vieta, E., Torrent, C., 2016. The impact of cognitive reserve in the outcome of first- Clinical and Experimental Uses. Stoelting Co, Chicago. episode psychoses: 2-year follow-up study. Eur. Neuropsychopharmacol. 26 (10), Goldstein, G., Allen, D.N., Seaton, B.E., 1998. A comparison of clustering solutions for 1638–1648. cognitive heterogeneity in schizophrenia. J. Int. Neuropsychol. Soc. 4 (4), 353–362. Amoretti, S., Cabrera, B., Torrent, C., Mezquida, G., Lobo, A., González-Pinto, A., Gower, J.C., 2012. A general coefficient of similarity and some of its properties. Parellada, M., Corripio, I., Vieta, E., de la Serna, E., Butjosa, A., Contreras, F., Biometrics 27, 857–871. Sarró, S., Penadés, R., Sánchez-Torres, A.M., Cuesta, M., Bernardo, M., PEPs Group, Granholm, E., Holden, J., Worley, M., 2018. Improvement in negative symptoms and 2018. Cognitive reserve as an outcome predictor: first-episode affective versus non- functioning in cognitive-behavioral social skills training for schizophrenia: affective psychosis. Acta Psychiatr. Scand. 138 (5), 411–455. mediation by defeatist performance attitudes and asocial beliefs. Schizophr. Bull. 44 Amoretti, S., Cabrera, B., Torrent, C., Bonnín, C.M., Mezquida, G., Garriga, M., (3), 653–661. Jiménez, E., Martínez-Arán, A., Solé, B., Reinares, M., Varo, C., Penadés, R., Grande, I., Salagre, E., Parellada, E., Bioque, M., Garcia-Rizo, C., Meseguer, A., 38 S. Amoretti et al. Schizophrenia Research 237 (2021) 31–39 Green, M.F., 1996. What are the functional consequences of neurocognitive deficits in Persistent negative symptoms in first-episode psychosis: early cognitive and social schizophrenia? Am. J. Psychiatr. 153 (3), 321–330. functioning correlates and differences between early and adult onset. J. Clin. Green, M.J., Girshkin, L., Kremerskothen, K., Watkeys, O., Quidé, Y., 2020. A systematic Psychiatry 78 (9), 1414–1422. review of studies reporting data-driven cognitive subtypes across the psychosis Reitan, R.M., Wolfson, D., 1993. The Halstead-Reitan Neuropsychological Test Battery: spectrum. Neuropsychol. Rev. 30 (4), 446–460. Theory and Clinical Interpretation. Neuropsychology Press, Tucson, AZ. Hall, M.H., Holton, K.M., Öngür, D., Montrose, D., Keshavan, M.S., 2019. Longitudinal Rodríguez-Sánchez, J.M., Ayesa-Arriola, R., Pérez-Iglesias, R., Periañez, J.A., Martinez- trajectory of early functional recovery in patients with first episode psychosis. Garcia, O., Gomez-Ruiz, E., Tabares-Seisdedos, R., Crespo-Facorro, B., 2013. Course Schizophr. Res. 209, 234–244. of cognitive deficits in first episode of non-affective psychosis: a 3-year follow-up Haring, L., Mõttus, R., Kajalaid, K., Koch, K., Uppin, K., Maron, E., Vasar, E., 2017. The study. Schizophr. Res. 150 (1), 121–128. course of cognitive functioning after first-episode of psychosis: a six month follow-up Rosa, A., Sánchez-Moreno, J., Martínez-Aran, A., Salamero, M., Torrent, C., Reinares, M., study. Schizophr. Res. 182, 31–41. Comes, M., Colom, F., Van Riel, W., Ayuso-Mateos, J., Kapczinski, F., Vieta, E., 2007. Hassani, M., Seidl, T., 2017. Using internal evaluation measures to validate the quality of Validity and reliability of the functioning assessment short test (FAST) in bipolar diverse stream clustering algorithms. Vietnam J. Comput. Sci. 4, 171–183. disorder. Clin. Pract. Epidemiol. Ment. Health 7, 5. Heinrichs, R.W., Awad, A.G., 1993. Neurocognitive subtypes of chronic schizophrenia. Sánchez-Torres, A.M., Moreno-Izco, L., Lorente-Omeñaca, R., Cabrera, B., Lobo, A., Schizophr. Res. 9 (1), 49–58. González-Pinto, A., Merchán-Naranjo, J., Corripio, I., Vieta, E., de la Serna, E., Herrero, P., Contador, I., Stern, Y., Fernández-Calvo, B., Sánchez, A., Ramos, F., 2019. Butjosa, A., Contreras, F., Sarró, S., Mezquida, G., Ribeiro, M., Bernardo, M., Influence of cognitive reserve in schizophrenia: a systematic review. Neurosci. Cuesta, M.J., PEPs group, 2018. Individual trajectories of cognitive performance in Biobehav. Rev. 2, 149–159. first episode psychosis: a 2-year follow-up study. Eur. Arch. Psychiatry Clin. Hill, S.K., Schuepbach, D., Herbener, E.S., Keshavan, M.S., Sweeney, J.A., 2004. Neurosci. 268 (7), 699–711. Pretreatment and longitudinal studies of neuropsychological deficits in Schubert, E., Rousseeuw, P.J., 2019. Faster k-medoids clustering: improving the pam, antipsychotic-naive patients with schizophrenia. Schizophr. Res. 68 (1), 49–63. clara, and clarans algorithms. In: International Conference on Similarity Search and Hollingshead, A., Redlich, F., 1958. Social class and mental illness: a community study. Applications, pp. 171–187. 1958. Am. J. Public Health 97 (10), 1756–1757. Stern, Y., 2002. What is cognitive reserve? Theory and research application of the reserve Jahshan, C., Heaton, R.K., Golshan, S., Cadenhead, K.S., 2010. Course of neurocognitive concept. J. Int. Neuropsychol. Soc. 8 (3), 448–460. deficits in the prodrome and first episode of schizophrenia. Neuropsychology 24 (1), Tomasini, C., Borges, E.N., Machado, K., Emmendorfer, L., 2017. In: A Study on the 109–120. Relationship between Internal and External Validity Indices Applied to Partitioning Kurtz, M.M., 2005. Neurocognitive impairment across the lifespan in schizophrenia: an and Density-based Clustering Algorithms. Proceedings of the 19th International update. Schizophr. Res. 74 (1), 15–26. Conference on Enterprise Information Systems (ICEIS 2017), 1, pp. 89–98. Lewandowski, K.E., Sperry, S.H., Cohen, B.M., Ongür, D., 2014. Cognitive variability in Torrent, C., Reinares, M., Martinez-Arán, A., Cabrera, B., Amoretti, S., Corripio, I., psychotic disorders: a cross-diagnostic cluster analysis. Psychol. Med. 44 (15), Contreras, F., Sarró, S., González-Pinto, A., Lobo, A., Cuesta, M.J., Sánchez- 3239–3248. Torres, A., Bergé, D., Castro-Fornieles, J., Moreno, C., Bernardo, M., Vieta, E., PEPS Lewandowski, K.E., Baker, J.T., McCarthy, J.M., Norris, L.A., Öngür, D., 2018. group, 2018. Affective versus non-affective first episode psychoses: a longitudinal Reproducibility of cognitive profiles in psychosis using cluster analysis. J. Int. study. J. Affect. Disord. 238, 297–304. Neuropsychol. Soc. 24 (4), 382–390. Uren, J., Cotton, S.M., Killackey, E., Saling, M.M., Allott, K., 2017. Cognitive clusters in Lobo, A., Chamorro, L., Luque, A., Dal-Ré, R., Badia, X., Baró, E., Grupo de Validación en first-episode psychosis: overlap with healthy controls and relationship to concurrent Español de Escalas Psicométricas (GVEEP), 2002. Validation of the Spanish versions and prospective symptoms and functioning. Neuropsychology 31 (7), 787–797. of the Montgomery-Asberg depression and Hamilton anxiety rating scales. Med. Clin. van Buuren, S., Groothuis-Oudshoorn, K., 2011. Mice: Multivariate Imputation by 118, 493–499. Chained Equations in R. J. Stat. Softw. 45, 1–67. Loonstra, A.S., Tarlow, A.R., Sellers, A.H., 2001. COWAT metanorms across age, van der Maaten, L., Hinton, G., 2008. Visualizing High-Dimensional Data using t-SNE. education, and gender. Appl. Neuropsychol. 8 (3), 161–166. J. Mach. Learn. Res. 9, 2579–2605. Milev, P., Ho, B.C., Arndt, S., Andreasen, N.C., 2005. Predictive values of neurocognition Van Rheenen, T.E., Lewandowski, K.E., Tan, E.J., Ospina, L.H., Ongur, D., Neill, E., and negative symptoms on functional outcome in schizophrenia: a longitudinal first- Gurvich, C., Pantelis, C., Malhotra, A.K., Rossell, S.L., Burdick, K.E., 2017. episode study with 7-year follow-up. Am. J. Psychiatry 162 (3), 495–506. Characterizing cognitive heterogeneity on the schizophrenia-bipolar disorder Peña-Casanova, J., 1990. Programa Integrado de Exploración Neuropsicológica - Test spectrum. Psychol. Med. 47 (10), 1848–1864. Barcelona. Manual. Masson. Barcelona. van Winkel, R., Myin-Germeys, I., De Hert, M., Delespaul, P., Peuskens, J., van Os, J., Peralta, V., Cuesta, M.J., 1994. Psychometric properties of the positive and negative 2007. The association between cognition and functional outcome in first-episode syndrome scale (PANSS) in schizophrenia. Psychiatry Res. 53 (1), 31–40. patients with schizophrenia: mystery resolved? Acta Psychiatr. Scand. 116 (2), Puig, O., Baeza, I., de la Serna, E., Cabrera, B., Mezquida, G., Bioque, M., Lobo, A., 119–124. González-Pinto, A., Parellada, M., Corripio, I., Vieta, E., Bobes, J., Usall, J., Wechsler, D., 1997. Weschsler Adult Intelligence Scale, 3r Edn. The Psychological Contreras, F., Cuesta, M.J., Bernardo, M., Castro-Fornieles, J., PEPs Group, 2017. Corporation, San Antonio, TX, USA. 39