Table_2.DOCX (52.05 kB)

Table_2.DOCX

Download (52.05 kB)
dataset
posted on 06.03.2018, 04:16 by Jesus Sainz, Carlos Prieto, Fulgencio Ruso-Julve, Benedicto Crespo-Facorro

Antipsychotic drugs are one of the largest types of prescribed drugs and have large inter-individual differences in efficacy, but there is no methodology to predict their clinical effect. Here we show a four-gene blood expression profile to predict the response to antipsychotics in schizophrenia patients before treatment. We sequenced total mRNA from blood samples of antipsychotic naïve patients who, after 3 months of treatment, were in the top 40% with the best response (15 patients) and in the bottom 40% with the worst response (15 patients) according to the Brief Psychiatric Rating Scale (BPRS). We characterized the transcriptome before treatment of these 30 patients and found 130 genes with significant differential expression (Padj value < 0.01) associated with clinical response. Then, we used Random Forests, an ensemble learning method for classification and regression, to obtain a list of predictor genes. The expression of four genes can predict the response to antipsychotic medication with a cross-validation accuracy estimation of 0.83 and an area under the curve of 0.97 using a logistic regression. We anticipate that this approach is a gateway to select the specific antipsychotic that will produce the best response to treatment for each specific patient.

History

Licence

Exports