Development and validation of a nomogram prediction model for long-term efficacy of growth hormone-type pituitary adenoma patients after transsphenoidal endoscopic surgery

Objective To analyze the long-time prognostic factors of patients with growth hormone (GH)-type pituitary adenoma after transsphenoidal endoscopic surgery, and establish a nomogram prediction model for GH-type pituitary adenoma. Methods Patients with GH-type pituitary adenoma who were admitted to the Affiliated Hospital of Xuzhou Medical University from January 2014 to January 2020 were enrolled and their clinical data were analyzed. The optimal predictors were screened by LASSO regression analysis, and the risk predictors were further screened by multivariate Logistic regression analysis. A nomogram prediction model was established based on risk predictors. The discriminative performance of the model was evaluated by the area under the receiver operating curve (ROC), and the calibration curve of the model was tested using Hosmer-Lemeshow goodness of fit test. The calibration curve was evaluated by the discriminant consistency index (C-index) and internal validation of the bootstrap strategy. The decision curve analysis (DCA) was performed to assess the clinical validity of the model. Results A total of 96 patients with GH-type pituitary adenoma were consecutively included. At the final follow-up, 61 patients (63.5%) met the remission criteria. A prediction model for long-term non-remission after surgery was constructed based on four risk factors affecting postoperative remission (preoperative random GH, preoperative IGF-1, Knosp grade and MPAD). The area under the ROC curve of the model was 0.897 95%CI (0.828, 0.967), indicating that the model had good discriminative performance. The calibration curve of the model showed the similar trend with the actual prediction curve, after the Hosmer-Lemeshow goodness of fit test (P=0.615). The internal verification of the prediction model through the Bootstrap method showed that the one-time index (C-index) was 0.930, indicating that the model had a high degree of discrimination. Finally, a decision curve analysis (DCA) of the predictive model was plotted to determine the clinical validity of the model. Conclusions Preoperative random GH, preoperative IGF-1, Knosp grade, and MPAD are the independent predictors of long-term remission after transsphenoidal endscopic surgery in patients with GH-type pituitary adenomas. The constructed nomogram prediction model shows good discrimination performance and precision performance, which is helpful to predict postoperative remission and guide the development of individualized follow-up plans and treatment regimens.