Abstract
BackgroundSeparation of individuals into schizophrenia and bipolar diagnoses has long been questioned, with some suggesting that the classification impairs the understanding of etiology, the accuracy of prognoses, and treatment selection. In this study, we employed unbiased statistical techniques to identify subgroups of individuals with chronic illness using a large array of variables commonly evaluated at the bedside. We then validated the resulting groups by investigating age of onset, schizophrenia polygenic risk scores (PRS), and functional outcomes at a 1-year follow-up period. Our hypothesis was that transdiagnostic subgroups would be stratified based on illness onset whereby individuals with earlier onset would have higher genetic risk loading and poorer functional outcomes.MethodsParticipants were selected from a longitudinal, naturalistic, multi-site project (PsyCourse) designed to investigate psychiatric illness course and outcomes. A total of 329 participants (age(SD)=45.7(12.6); 54% female; years of illness duration(SD) = 13.7(10.3)) with a DSM-IV diagnosis of schizophrenia, schizoaffective disorder, or bipolar disorder were assessed from 17 centers at baseline and 1-year follow-up periods. A clinical battery measuring sociodemographic, illness history, symptoms, cognition, and personality questionnaires (199 variables) was used to subgroup individuals. A non-negative factor analytic consensus clustering MATLAB toolbox was created based on previous methodological work in oncology. PRS were generated using widely used strategies, and differences between resulting subgroups were investigated with MANCOVA controlling for ancestry effects. Differences in functional outcomes were investigated with repeated measures ANOVA.ResultsA 4-subgroup solution was robustly defined as the optimal solution using resampling techniques and cluster validity indices. Diagnoses were mixed in two subgroups, but predominantly bipolar or schizophrenia in the other two. All subgroups had equal illness durations (p>0.05), but the age of onset showed a decreasing trend with the earliest age being linked to two subgroups: a mixed bipolar-schizophrenia group with intermediate levels of general functioning and in a schizophrenia group with low levels of functioning (p<0.001). PRS scores were significantly increased in the early-onset, mixed bipolar-schizophrenia subgroup (p=0.007, uncorrected) and in the schizophrenia group (p=0.025, uncorrected). Prognoses differed between the four groups (p=0.003), with the greatest increases in functional outcomes in a late-onset mixed diagnostic subgroup (p=0.006) and in the schizophrenia group (p=0.002).DiscussionFour subgroups were detected and our hypothesis was supported by a relationship between earlier illness onset and higher schizophrenia genetic risk loading. While one of the subgroups with an earlier onset mostly consisted of individuals with schizophrenia, the other subgroup was diagnostically mixed. Our results tentatively suggest that transdiagnostic clustering may identify subgroups that could be effectively used to understand etiology and prognoses. Future research will investigate the possibility of differential treatment effects in these subgroups.
Highlights
Separation of individuals into schizophrenia and bipolar diagnoses has long been questioned, with some suggesting that the classification impairs the understanding of etiology, the accuracy of prognoses, and treatment selection
Psychotic symptoms were not limited to patients with an ICD-10 diagnosis of a psychotic disorder and were present in the control group
We found that psychotic symptoms were not limited to patients with a specific ICD-10 diagnosis and were present in a wide range of ICD10 disorders
Summary
The clinical classification of psychotic disorders has remained largely unchanged and is based on criterion-based diagnostic systems (such as ICD-10 and DSM-5) which do not necessarily reflect their underlying aetiology and pathophysiology. Automated information extraction methods such as natural language processing (NLP) offer the opportunity to quickly extract and analyse large volumes of clinical data from EHRs. We sought to characterise the range of presenting symptoms in a large sample of patients with psychotic disorders using NLP. Psychotic symptoms were not limited to patients with an ICD-10 diagnosis of a psychotic disorder and were present in the control group. Discussion: We found that psychotic symptoms were not limited to patients with a specific ICD-10 diagnosis and were present in a wide range of ICD10 disorders. These findings highlight the utility of detailed NLP-derived symptom data to better characterise psychotic disorders. Nikolaos Koutsouleris1 1Ludwig Maximilian University; 2University Medical Centre Göttingen; 3Institute of Computational Biology
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
