Improved individualized prediction of schizophrenia in subjects at familial high risk, based on neuroanatomical data, schizotypal and neurocognitive features

Abstract

To date, there are no reliable markers for predicting onset of schizophrenia in individuals at high risk (HR). Substantial promise is, however, shown by a variety of pattern classification approaches to neuroimaging data. Here, we examined the predictive accuracy of support vector machine (SVM) in later diagnosing schizophrenia, at a single-subject level, using a cohort of HR individuals drawn from multiply affected families and a combination of neuroanatomical, schizotypal and neurocognitive variables. Baseline structural magnetic resonance imaging (MRI), schizotypal and neurocognitive data from 17 HR subjects, who subsequently developed schizophrenia and a matched group of 17 HR subjects who did not make the transition, yet had psychotic symptoms, were included in the analysis. We employed recursive feature elimination (RFE), in a nested cross-validation scheme to identify the most significant predictors of disease transition and enhance diagnostic performance. Classification accuracy was 94% when a self-completed measure of schizotypy, a declarative memory test and structural MRI data were combined into a single learning algorithm; higher than when either quantitative measure was used alone. The discriminative neuroanatomical pattern involved gray matter volume differences in frontal, orbito-frontal and occipital lobe regions bilaterally as well as parts of the superior, medial temporal lobe and cerebellar regions. Our findings suggest that an early SVM-based prediction of schizophrenia is possible and can be improved by combining schizotypal and neurocognitive features with neuroanatomical variables. However, our predictive model needs to be tested by classifying a new, independent HR cohort in order to estimate its validity.

Individualized Diagnostic and Prognostic Models for Patients With Psychosis Risk Syndromes: A Meta-analytic View on the State of the Art

“…A total of 19 ML studies (73%) employed a support vector machine algorithm (10,30,(37)(38)(39)(40)(41)(42)(43)(44)(45)(46)(47)(48)(49)(50)(51)(52)(53), while the rest used Gaussian process (11) or convex hull classification (54), randomized trees (55), greedy algorithm (20), random forest (5), or LASSO regression (56,57). All ML models were computed with CV, whereas studies using Cox regression applied bootstrapping (28,(58)(59)(60)(61)(62), reported apparent results (i.e., the model is tested in the same sample from which it was derived) (63-68), or lacked a validation procedure.…”

“…Clinical models were trained on prodromal positive and negative symptoms, functioning, and family risk associated with functional decline; the neurocognitive modality was based on executive functions and verbal IQ (41) or speech features (54,73). Multimodal approaches included different combinations of clinical, neuropsychological, and demographic variables as well as genetic risk (28,51,52,54,74). One model was built on P300 amplitude from event-related potentials and sociopersonal adjustment (62).…”

“….10) (Table S2). CHR in 86% of the studies fulfilled the UHR criteria (75), while 6 models were based on the genetic risk syndromes 22q11.2DS (38,48,49) or FR (39,52,53). Because of this imbalance, we could not statistically test the effects of this variable, yet results did not change when excluding patients with 22q11.2DS and FR (Supplement).…”

“…This result may mirror a real lack of significant differences in biomarker type when distinguishing the CHR state from the norm or predicted outcome. However, because only 4 prognostic studies tested the relative and combined predictive ability of different data modalities on the same individuals (10,30,52,56), and because data modalities are overall underor overrepresented, the currently available studies do not allow this conclusion to be drawn. Further research directly comparing performance across data modalities, followed by meta-analytic evaluation, is warranted.…”

“…Most of the at-risk individuals in our sample fulfilled the UHR criteria, while a minority (5.7%) had an FR or a 22q11.2DS diagnosis, which prevented us from quantitatively estimating the effects of risk group designation. However, it is noteworthy that two of the instruments operationalizing UHR criteria (i.e., SIPS Meta-analysis of Psychosis Risk Syndrome Prediction Models include a genetic risk group (i.e., the genetic risk deterioration syndrome) and that two studies in our sample included FRs and deletion syndrome patients with subthreshold psychotic symptoms (49,52). This diagnostic overlap might create, on the one hand, a further source of variability and, on the other, a tangible bridge to the well-known heterogeneity among CHR individuals.…”