F69. SYMPTOM PREDICTORS OF TREATMENT-RESISTANCE IN FIRST-EPISODE PATIENTS

Abstract Background Recent studies suggest that most individuals that will develop treatment-resistant schizophrenia (TRS) will not respond since the first episode of psychosis (FEP). These individuals may be part of a specific subtype of schizophrenia with distinct neurobiology and response to treatment. Once the improvement of prognosis of TRS patients is related to early clozapine initiation, it is relevant to identify clinical predictors that could forecast response failure to other antipsychotics. In a previous study, we found that a score comprising three PANSS (Positive and Negative Syndromes Scales) items can predict treatment resistance in a prospective inpatient cohort with chronic patients, result replicated in an independent outpatient sample. The purpose of this study is to identify whether this same score can predict early response failure in a cohort of antipsychotic naïve FEP patients. Methods Patients (n = 87) with FEP, without previous use of antipsychotics, were recruited at an emergency service in São Paulo, Brazil, between 2011 and 2017. Individuals were assessed at admission and after 10 weeks of follow-up. The diagnosis of all subjects was confirmed using the Structured Clinical Interview for DSM-IV Disorders (SCID-I). Patients were evaluated with the Positive and Negative Syndrome Scale (PANSS) at the baseline and after 10 weeks of treatment. A reduction of less than 30% in total baseline PANSS scores was considered response failure. To investigate the clinical predictors, the predictive capacity of the sum of three PANSS items was tested: P2 (conceptual disorganization), N5 (abstract thinking difficulty), and G9 (unusual mental thinking). The predictive power of these items was evaluated using a logistic regression model. The PANSS scores percentage reduction after treatment was correlated with these three PANSS items, controlled by sex, age, and duration of untreated psychosis (DUP). Results The mean age was 25.98 years (SD ± 7.2), the majority (58.62%) was male and the mean DUP was 212.40 days (SD ± 282.3). Higher scores on the P2 + N5 + G9 model predicted higher chance of response failure (β =0.2550, x2 = 6.3202, df = 1, p = 0.0120). In addition, a ROC curve was performed, resulting in an AUC of 0.675. Discussion We replicated previous findings of our group in chronic patients, showing that the sum score of P2, N5 and G9 can predict early treatment failure. In general, we believe that such items are associated to the disorganize