A risk model for predicting progression of pituitary tumors by blood and clinical factors

Abstract

Purpose To build a model that utilizes clinical and blood parameters to predict the recurrence or progression of pituitary tumors after surgery. Methods A training group (67,70%) and a validation group (29,30%) were formed from 96 individuals with recurrent pituitary tumors. The training group was screened for blood parameters and a blood-related risk score (BRS) was established. Clinical-related prognostic factors were also assessed through Cox regression analysis, which was used alongside the BRS to construct a clinical prognostic model. In the validation group to assess the stability and accuracy of the BRS and the clinical prognostic model. Additionally, a clinical and blood-based nomogram was developed. Result The preoperative blood parameters K+, cholinesterase (CHE), and 5-nucleotidase (NT5E) were found to be correlated with progression-free survival (PFS). The area under the curve (AUC) for the BRS was 0.788 (95% CI: 0.657-0.919) in the training group and 0.852 (95% CI: 0.706-0.997) in the validation group. For 1-, 3-, and 5-year intervals in the validation set, the clinical model's AUC was 0.718, 0.852, and 0.864, respectively. While in the test group, the AUCs were 0.600, 0.889, and 0.660. The nomogram had C indices of 0.732 and 0.704 in the training and validation groups, respectively, and its calibration curves and clinical decision curves (DCA) indicated good calibrations and clinical utility. ConclusionThis is a new non-invasive tool that provides a tool for predicting prognosis and risk stratification.