Heliyon (Apr 2024)

Gender potentially affects early postoperative hyponatremia in pituitary adenoma: XGBoost-based predictive modeling

  • Zheng Peng,
  • Xiao-Jian Li,
  • Yun-feng Wang,
  • Zhuo-Yuan Li,
  • Jie Wang,
  • Chun-Lei Chen,
  • Hui-Ying Yan,
  • Wei Jin,
  • Yue Lu,
  • Zong Zhuang,
  • Chun-Hua Hang,
  • Wei Li

Journal volume & issue
Vol. 10, no. 7
p. e28958

Abstract

Read online

Purpose: The occurrence of hyponatremia is a prevalent complication following transnasal transsphenoidal surgery for pituitary adenoma surgery, which adversely affects patient prognosis, hospitalization duration, and rehospitalization risk. The primary objective of this study is to strengthen the correlation between clinical factors associated with pituitary adenoma and postoperative hyponatremia. Additionally, the study aims to develop a predictive model for postoperative hyponatremia in patients with pituitary adenoma, with the ultimate goal of establishing a basis for reducing the occurrence of postoperative hyponatremia following surgical interventions. Methods: The chi-square test or Fisher test was employed for nominal data, while the t-test or Mann-Whitney test was utilized for continuous data analysis. In cases where the data exhibited statistical differences, binary logistic analysis was conducted to examine the risk and protective factors associated with postoperative hyponatremia. XGBoost was employed to construct predictive models for hyponatremia in this study. The patients were partitioned into training and test sets, and the most suitable parameters were determined through five-fold cross-validation and subsequently utilized for training on the training set. The discriminatory capability was assessed on the internal validation set. Results and conclusions: Out of the total 280 patients included in this investigation, 82 patients experienced early postoperative hyponatremia. Among these individuals, male gender (P = 0.02, odds ratio = 1.98) was identified as a risk factor for early postoperative hyponatremia, while preoperative chloride levels (P = 0.021, odds ratio = 0.866) and surgery time (P = 0.039, odds ratio = 0.990) were identified as protective factors against postoperative hyponatremia. The XGBoost model exhibited a sensitivity of 94.2%, a specificity of 61.5%, a positive predictive value of 51.6%, a negative predictive value of 96%, and identified male gender, preoperative sodium, and preoperative cortisol as the most significant predictors. Our findings indicate that gender may have influence in the development of early postoperative hyponatremia in patients with pituitary adenomas.

Keywords