UC San Diego

Abstract Background: The vast majority of studies of neuropsychological (NP) functioning in Clinical High Risk (CHR) cohorts have examined group averages, possibly concealing a range of subgroups ranging from very impaired to high functioning. Our objective was to assess NP profiles and to explore associations with conversion to psychosis, functional and diagnostic outcome. Methods: Data were acquired from individuals (mean age 18.4, SD = 4.6) participating in the longitudinal North American Prodrome Longitudinal Study-I (NAPLS-I), a multi-site consortium following individuals at CHR for developing psychosis for up to 2½ years. By applying the Hierarchical Clustering Ward’s method including 8 different neuropsychological tests, we clustered data of 166 CHR individuals, 49 persons with a family history of psychosis without prodromal symptoms, and 109 healthy controls. We then tested whether cluster profiles with more severe NP impairments were associated with higher conversion rates, lower social and role functioning scores, and/or more chronic diagnostic outcomes compared to the lesser-impaired profiles. To examine clinical utility, analyses were repeated after data were clustered based on clinical decision rules that were established by clinical experts in the field. Results: Four distinctive profile clusters best described the level of NP performance in our CHR cohort: Severely Impaired (n = 33); Clearly Abnormal (n = 82); Borderline (n = 145) and Normal (n = 64). The Severely Impaired cluster largely distinguished itself from the rest of the clusters by larger deviations on processing speed and memory tasks. We found compelling differences in outcome between cluster profiles. Importantly, those assigned to the most impaired profile had a conversion rate of 42.4%, had a 40% chance of developing a diagnosis in the schizophrenia spectrum (as compared to 24.4% in the Clearly impaired, 7.4 % in the Borderline impaired and 2.9% in the Normal functioning group), and had significantly worse social (P < .001) and role (P < .001) functioning scores at baseline and 12-month follow-up. Similar results were obtained when data were clustered following clinical decision rules. Conclusion: Despite extensive neuropsychological investigations within CHR cohorts, this is one of the first studies to investigate NP clustering profiles as a contributor to heterogeneity in outcome. Our results indicate that the four NP profiles vary substantially in their outcome, underscoring the relevance of cognitive functioning in the prediction of illness progression. Our findings may tentatively suggest that individualized cognitive profiling should be explored in clinical settings, and my point to important directions for personalized treatment.