NEW YORK – A combination of four readily available clinical and psychosocial tests can predict whether patients are at risk for converting to psychosis with fair to good positive predictive value, an investigator asserts.
Measures of thought content, suspiciousness, and subscales of the Brief Assessment of Cognition in Schizophrenia instrument, as well as total life events, can predict with fairly good sensitivity and specificity which patients will progress to psychosis, said Tyrone D. Cannon, Ph.D., professor of psychology and psychiatry at Yale University in New Haven, Conn.
"The foundation of any effective prevention program has to be begin with the goal of prediction; we have to find the people most at risk. And then another pillar of prevention is understanding something about the underlying mechanisms of illness," he said at the American Psychiatric Association annual meeting.
Neil Osterweil/Frontline Medical News
Tyrone D. Cannon, Ph.D.
The clinical high-risk paradigm (CHR) is based on the establishment of CHR criteria that can capture the population at risk of imminent onset of psychosis.
A recent systematic review risk of CHR research (JAMA Psychiatry 2013;70:107-20) found that among high-risk patients, rates of conversion to psychosis were about 15% by 1 year of follow-up and 30% by 2 years. Of those who convert to psychosis, about 80% develop a schizophrenia spectrum disorder, and about 20% develop mood or atypical psychoses.
A second study (Schizophr. Bull. 2012;38:1225-33) showed that of nonconverters, one-third recover symptomatically by 2 years, an additional third recover functionally, and another third recover both symptomatically and functionally.
Therefore, a good rule of thumb is that among CHR populations, about one-third will remit, one-third will convert to psychosis, and one-third will remain in a CHR state, Dr. Cannon said. Of the latter group, some might convert to schizophrenia or psychosis, and some might develop schizotypal disorders.
Currently available multivariate algorithms that involve specific combinations of symptoms and demographic factors generally have high positive predictive values and specificity but tend to have low sensitivity, and the accepted diagnostic profiles vary widely across different studies, he said.
Dr. Cannon and colleagues in the NAPLS (North American Prodrome Longitudinal Study) consortium have studied clinical predictors of psychosis in 360 of the 750 planned CHR patients recruited. Of this group, 60 converted to psychosis within 2 years, and the remaining nonconverters were followed for the same period.
The investigators examined the receiver operating characteristic (ROC) curves for predictive models for which there were 10 or more converters per predictor. Using this method, a value of 1 indicates a perfect test result, and a value of .5 area-under-the-curve (AUC) is a poor or failed result.
"There are a number of predictors that have emerged from the prior literature that in this new sample are indeed confirmed as very robustly predictive. The single best one is the rated function of what we call P1 and P2, which are the symptoms of unusual thought content and suspiciousness. Higher levels of those, still at a prepsychotic level, but at a higher, prodromal level of activity, predicts psychosis with pretty high accuracy on their own," Dr. Cannon said.
For P1 and P2, the AUC was 0.668 and was significantly predictive (P less than .0001). Other significant markers included the digital sequencing subscale of the Brief Assessment of Cognition in Schizophrenia (BACS) (AUC 0.614; P = .0018) and the Hopkins Verbal Learning Test (AUC 0.630; P = .0005). A less robust but still significant predictor was the total number of life events (AUC 0.579; P = .0495).
Combining the four markers yielded an AUC of 0.74, with a sensitivity of 70% and specificity of 73%. The positive predictive value was relatively low, at 46%, but the negative predictive value of the combination was fairly good, at 88%, Dr. Cannon pointed out.
The NAPLS investigators also are investigating whether specific biomarkers such as cortisol levels and change in prefrontal gray matter volume and event-related potential can identify patients who will convert to schizophrenia or other psychotic states.
The NAPLS consortium is supported by the National Institute of Mental Health. Dr. Cannon disclosed that he is a consultant to the Los Angeles County Department of Mental Health on early detection and prevention services.