A Preoperative Nomogram for Prediction of Postoperative Hypocortisolism in Patients with Pituitary Adenomas: A...

Patients with pituitary adenomas (PAs) are at an increased risk preoperatively and postoperatively for hypopituitarism. Postoperative hypocortisolism is associated with increased mortality and morbidity as well as poor quality of life. However, research about the risk factors for postoperative hypocortisolism is limited, and a predictive nomogram for postoperative hypocortisolism has not yet been developed. We aimed to investigate the predictive factors for postoperative hypocortisolism and construct a dynamic online nomogram.

Our database included 438 consecutive PA patients who were hospitalized and treated with transsphenoidal surgery by experienced neurosurgeons from the different medical teams in the Neurosurgery Department, Jinling Hospital, between January 2018 and October 2020. The final study group included 238 eligible patients. Data on possible predictors, including age, sex, treatment history of PAs, preoperative signs and symptoms, primary recurrence subtype, and clinical subtypes, were collected. Univariable and multivariable logistic regression analyses were applied to identify independent predictors, which were included in constructing the nomogram model. The calibration curve and receiver operating characteristic curve were computed to evaluate the predictive performance of the nomogram model.

The incidence of postoperative hypocortisolism was 12.08%. Three preoperative predictors were identified to construct the nomogram: surgical type (microscopic or endoscopic, with endoscopic surgery proven to be the protective factor) (odds ratio, 0.24; 95% confidence interval [CI], 0.093-0.610; P = 0.003), prothrombin time (odds ratio, 2.40; 95% CI, 1.332-4.326; P = 0.004), and basophil cell count (odds ratio, 5.25; 95% CI, 1.270-21.816; P = 0.022,). The area under the curve of receiver operating characteristic curve for the constructed nomogram was 0.749 (95% CI, 0.640-0.763); a well-fixed calibration curve was generated for the nomogram model. An interactive web-based dynamic nomogram application was also constructed.

In this study, surgical type, prothrombin time, and basophil cell count were the most relevant predictive factors for postoperative hypocortisolism. A predictive nomogram that can preoperatively assess the risk of hypocortisolism after surgical treatment of PAs was developed. This nomogram could be helpful in identifying high-risk patients who require close monitoring of serum cortisol levels and initiating clinical procedures for patients requiring cortisol administration therapy as a lifesaving strategy.