A nomogram for predicting tongue collapse in DISE based on UACT findings in OSA patients.

Objective: This study aimed to analyze the correlation between DISE and upper airway CT (UACT) in OSA patients and to establish and validate a predictive model for tongue base (TB) collapse in the upper airway of OSA patients.

Methods: From October 2014 to March 2022, a total of 117 OSA patients diagnosed by polysomnography were enrolled. Three-dimensional CT measurements of the upper airway were conducted, and preoperative DISE examinations were performed. Surgical plans were determined based on VOTE scoring. We used logistic regression analysis to identify key predictors of tongue base (TB) collapse in patients with obstructive sleep apnea (OSA). Subsequently, a predictive model was constructed and validated using the R language.

Results: Through univariate logistic regression analysis, distances between bilateral tonsils (BT), thickness of parapharyngeal space (PPS), thickness of lateral pharyngeal walls (LPW), distance from tongue to posterior pharyngeal wall (PAS), and transverse diameter of the epiglottis (TEP) (P ≤ 0.1) were included in the multivariate logistic regression analysis. Multivariate analysis revealed that the thickness of LPW (P < 0.05) and TEP (P < 0.001) were significant predictive factors for TB collapse in the predictive model. A nomogram model was created for predicting TB obstruction based on these factors, exhibiting an AUC of 0.729 (95% CI: 0.619 ~ 0.914). The internal validation of the model resulted in an AUC of 0.804 (95% CI: 0.688 ~ 0.878) within the validation population.

Conclusion: A significant correlation between TB obstruction observed under DISE and LPW (P < 0.05) and TEP as measured in UACT scans among OSA patients. Based on this, a predictive nomogram model for TB obstruction during DISE demonstrates good performance.