Using proteomics to predict clinical high-risk patient outcomes
Proteomics can predict the development of psychotic disorders for patients in the clinical high-risk state.
The clinical high-risk (CHR) state is characterised by the presence of early or prodromal signs and symptoms of psychosis. Identification of CHR patients is vital as between 16–35% of these individuals progress to develop their first episode of psychosis within three years.
Early recognition of this state may allow clinicians to identify vulnerable individuals who may also present with depression, anxiety and functional impairment. Identifying these individuals also offers the opportunity for early clinical intervention.
A second group at risk for the future development of psychosis are those people who have psychotic experiences (PEs), many of whom do not seek help. PEs are known to be associated with psychotic and non-psychotic disorders, suicidal behaviour and reduced functioning.
In a new study available here, the team of researchers led by Professor David Cotter at RCSI University of Medicine and Health Sciences, alongside a large group of international collaborators from 12 countries, have now identified a panel of proteomic biomarkers that can predict, with a high degree of accuracy, which patients with the CHR state will transition to psychosis.
Discovery of biomarkers that may aid in the prediction of outcomes for these two at-risk groups would be pivotal in improving patient care. There is already evidence that supports an association between dysregulation of inflammatory and immune processes and the development of psychosis.
The study sought to use proteomic methods to compare protein expression in CHR individuals who did and did not go on to develop psychosis. Similar methods were used in individuals who did and did not go on to develop PEs.
The CHR cohort consisted of 49 patients who transitioned to psychosis and 84 patients who did not transition. The group was derived from a pool of 344 patients that were prospectively followed for up to 6 years across 11 sites in Europe, Australia and Brazil.
Prof. Cotter and his team took plasma samples at baseline and compared samples between those who transitioned to psychosis on follow-up and those who did not.
35 proteins were found to have significantly different levels of expression between the two cohorts. The most implicated pathways were the complement and coagulation cascades. The accuracy of these proteins was tested on 30 patients who transitioned and 50 who did not transition. The AUC was found to be 0.97 (p<0.004) with a sensitivity of 86.7% and a specificity of 94.0%.
The PE cohort consisted of 55 cases who reported PEs between age 12 and 18. The cases were matched with 66 controls who did not report PEs between age 12 and 18. Age 12 plasma samples were compared between the cohorts using mass spectrometry. Five proteins were expressed differentially between the groups. The SVM model based on proteomic features from age 12 plasma samples was able to predict PE status at age 18 with an AUC of 0.76 (p<0.004).
In the study, the team demonstrated that proteomic features had greater predictive value than clinical features when predicting the risk of transition from the CHR state to psychosis. The researchers were also able to predict a 2-year functional outcome with an AUC of 0.72 (p=0.008).
Using these ten highly predictive proteins, a clinician may accurately predict which patients have an increased risk of transition to psychosis and target these patients with preventative measures.
Of particular interest was the strong implication of the complement and coagulation cascades as this gives some insight into the pathogenesis of psychosis. The strongest predictor of transition was reduced expression of A2M, a protease inhibitor with diverse functions including inhibition of pro-inflammatory cytokines such as IL-Iß; it also functions as a critical coagulation inhibitor. There appears to be a pro-coagulant phenotype in those who transition from CHR to psychosis.
Dr David Mongan, RCSI PhD student and Irish Clinical Academic Training (ICAT) Fellow, analysed the data with supervision by Professor David Cotter and Professor Mary Cannon in RCSI Department of Psychiatry. The proteomic analyses were performed by Dr Melanie Focking, RCSI Department of Psychiatry and Dr Gerard Cagney, UCD Conway Institute.
Following on from this study, Professor Cotter and his team have been awarded a prestigious Wellcome Trust Flagship Grant of over €1.3 million to further investigate markers derived from proteomics to help predict the transition from the ultra-high risk state to psychotic disorder.
Journal Article Information:
Development of proteomic prediction models for outcomes in the clinical high risk state and psychotic experiences in adolescence: machine learning analyses in two nested case-control studies
JAMA Psychiatry. 2020 Aug 26.
https://doi.org/10.1001/jamapsychiatry.2020.2459
