Original Article
Predicting the likelihood of early recurrence based on mRNA sequencing of pituitary adenomas
Abstract
Background: There is no comprehensive and objective method existing for predicting early recurrence of pituitary adenomas (PAs). The most advanced gene sequencing technology can be applied to build a prognostic model that can effectively predict early recurrence of PAs.
Methods: In this study, using mRNA-Seq data, the corresponding postoperative early recurrence status, and other clinical features of 107 PA samples were obtained and randomly divided into the training and validation groups. Cox regression and receiver operating characteristic (ROC) analysis accompanied by the risk score method was used to build a seven-gene prediction model.
Results: Area under curve values was 0.857 in the training group, 0.936 in the validation group, and 0.848 in all patients. Patients with low-risk scores had a significantly lower probability of early postoperative recurrence compared to those acquiring high-risk scores in the training group, validation group, and all patient (P<0.0001) groups. In addition, 6 out of these 7 significant genes were highly correlated to the early recurrence of PAs.
Conclusions: This prediction model derived from mRNA-Seq data may help in identifying the early recurrence of PAs, consequently aiding in the classification of patients with PAs and the administration of the appropriate therapeutic and follow-up strategy for these patients.
Methods: In this study, using mRNA-Seq data, the corresponding postoperative early recurrence status, and other clinical features of 107 PA samples were obtained and randomly divided into the training and validation groups. Cox regression and receiver operating characteristic (ROC) analysis accompanied by the risk score method was used to build a seven-gene prediction model.
Results: Area under curve values was 0.857 in the training group, 0.936 in the validation group, and 0.848 in all patients. Patients with low-risk scores had a significantly lower probability of early postoperative recurrence compared to those acquiring high-risk scores in the training group, validation group, and all patient (P<0.0001) groups. In addition, 6 out of these 7 significant genes were highly correlated to the early recurrence of PAs.
Conclusions: This prediction model derived from mRNA-Seq data may help in identifying the early recurrence of PAs, consequently aiding in the classification of patients with PAs and the administration of the appropriate therapeutic and follow-up strategy for these patients.
