Development And Validation Of Prognostic Prediction Model To Estimate The Long-term Postoperative Outcome Of Pituitary GH Adenoma Patients

Background:According to the clinical guidelines,transsphenoidal surgery(TSS)is the first choice for the treatment of pituitary growth hormone adenomas(acromegaly).It is of great clinical significance to predict the long-term prognosis of patients during the perioperative period.At present,there are few studies on the prognostic model of long-term postoperative outcome of pituitary growth hormone adenomas.Objective:To develop and validate the prediction model of long-term outcome after transsphenoidal surgery in patients with pituitary growth hormone adenomas,Design:This study is a single center retrospective cohort study.Objects:According to the inclusion and exclusion criteria,this study included 328 newly diagnosed adult patients with complete follow-up data who met the diagnosis of acromegaly.All the data came from the medical records of the inpatients in the Department of Neurosurgery of Peking Union Medical College Hospital.Methods:In this study,stepwise variable selection,LASSO,SVM machine learning algorithm and optimal subset variable selection methods were used for multivariable screening combined with clinical application significance.Logistic regression equation was used to build models,and random split validation,K-fold cross-validation and Bootstrap resampling validation were used for internal validation.C statistics are used to evaluate the discrimination of the model,and Brier scores and calibration curves are used to evaluate the calibration of the model.Finally,the decision curve analysis is used to evaluate the best prediction range of the model.Results:In this study,two predictive models were developed and validated to predict the probability of long-term outcome after transsphenoidal surgery in patients with pituitary growth hormone adenomas.Model A includes 6 preoperative predictive factors,such as preoperative GH nadir,maximum tumor diameter,Knosp grade,empty sella syndrome,BMI and acromegaly appearance,which can be used to predict the probability of remission after surgery in patients with acromegaly.Model B contains 4 predictors,including immediate postoperative GH levels,BMI,p53 and Knosp grade,which can be used to predict the probability of long-term remission in the early postoperative stage(within 1 week after surgery,usually at the time of discharge).Validated by three internal validation methods,that is random split verification,K-fold cross-validation and Bootstrap resampling validation,the two models have good discrimination and calibration,and model B performs better.Conclusion:The prediction model of long-term prognosis of pituitary growth hormone adenomas after transsphenoidal surgery is helpful for clinicians and patients to make follow-up schedules and personalized treatment plans.