Tsuang FY, Shih SR, Tseng HM, Wang HC. Perioperative growth hormone levels as an early predictor of new-onset secondary adrenal insufficiency following transsphenoidal pituitary tumor resection.
Asian J Surg 2024;
47:1746-1755. [PMID:
38148260 DOI:
10.1016/j.asjsur.2023.12.120]
[Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/11/2023] [Revised: 12/06/2023] [Accepted: 12/15/2023] [Indexed: 12/28/2023] Open
Abstract
OBJECTIVE
This study aims to predict new-onset secondary adrenal insufficiency (NOSAI) after transsphenoidal pituitary tumor resection surgery using perioperative growth hormone (GH) and prolactin (PRL) levels, among other factors.
METHODS
A cohort of 124 adult patients who underwent transsphenoidal resection for non-functioning pituitary adenoma, with routine perioperative glucocorticoid use, was used to develop the predictive regression model. An additional 46 patients served as the validation cohort. Generalized additive models were used to identify optimal cut-off points for the variables.
RESULTS
The GH level on postoperative day one (POD1) can be a simple predictor by implementing a cut-off point of 0.41 ng/ml. A value ≤ 0.41 ng/mL predicted NOSAI with 0.6316 sensitivity and 0.7810 specificity for the original cohort and 1.0000 sensitivity and 0.7143 specificity for the validation cohort. The multiple logistic regression model included perioperative PRL level difference, perioperative GH level difference, intraoperative cerebrospinal fluid (CSF) leakage, tumor size, and the combined effect of diabetes insipidus (DI) and relative perioperative GH level difference. The areas under the receiver operating characteristic curves were 0.9410 (original cohort) and 0.9494 (validation cohort) for the regression model.
CONCLUSION
Early morning GH level on POD1 can predict NOSAI with fair accuracy when perioperative stress dose glucocorticoid is administered. Prediction accuracy can be improved by considering CSF leakage, DI, and perioperative changes in GH and PRL in the final regression model.
Collapse