Predicting the likelihood of postoperative seizure status based on mRNA sequencing in low-grade gliomas.

OBJECTIVE

No comprehensive and objective methods yet exist for predicting postoperative seizure.

METHODS

mRNA-seq data and corresponding postoperative seizure status of 109 low-grade glioma samples were obtained from Chinese Glioma Genome Atlas database and divided into two sets randomly. Logistic regression and receiver operating characteristic analysis with risk score method were used to develop a ten-gene prediction model.

RESULTS

Considering gene number and area under the curve of receiver operating characteristic, a ten-gene model was generated which showed an area under the curve of 0.9965 in training set. Patients with high-risk scores had higher probability of postoperative seizure compared with those with low-risk scores.

CONCLUSIONS

This is the first prediction model for postoperative seizures in gliomas, integrating multiple genes. Clinical application may help patients with postoperative seizure control.