Risk-score Model to Predict Prognosis of Malignant Airway Obstruction after Interventional Bronchoscopy.
Minlin Jiang,Hao Xu,Dongmei Yu,Li Yang,Wenhui Wu,Hao Wang,Hui Sun,Jun Zhu,Wencheng Zhao,Qiyu Fang,Jia Yu,Peixin Chen,Shengyu Wu,Zixuan Zheng,Liping Zhang,Likun Hou,Huixian Zhang,Ye Gu,Yayi He
DOI: https://doi.org/10.21037/tlcr-21-301
2021-01-01
Translational Lung Cancer Research
Abstract:Background Interventional bronchoscopy exhibits substantial effects for patients with malignant airway obstruction (MAO), while little information is available regarding the potential prognostic factors for these patients. Methods Between October 31, 2016, and July 31, 2019, a total of 150 patients undergoing interventional bronchoscopy and histologically-confirmed MAO were collected, in which 112 eligible participants formed the cohort for survival study. External validation cohort from another independent institution comprised 33 MAO patients with therapeutic bronchoscopy. The least absolute shrinkage and selection operator regression (LASSO) was applied to the model development dataset for selecting features correlated with MAO survival for inclusion in the Cox regression from which we elaborated the risk score system. A nomogram algorithm was also utilized. Results In our study, we observed a significant decline of stenosis rate after interventional bronchoscopy from 71.7%±2.1% to 36.6%±2.7% (P<0.001) and interventional bronchoscopy dilated airway effectively. Patients in our study undergoing interventional bronchoscopy had a median survival time of 614.000 days (95% CI: 269.876–958.124). Patients receiving distinct therapeutic methods of interventional bronchoscopy had different prognosis (P=0.022), and patients receiving treatment of electrocoagulation in combination with stenting and electrosurgical snare had worse survival than those receiving other options. Multivariate Cox analysis revealed that nonsmoking status, adenoid cystic carcinoma, and low preoperative stenosis length, as independent predictive factors for better overall survival (OS) of MAO patients. Then, the nomogram based on Cox regression and risk score system based on results from LASSO regression were elaborated respectively. Importantly, this risk score system was proved to have better performance than the nomogram and other single biomarkers such as traditional staging system (area under the curve 0.855 vs. 0.392–0.739). Survival curves showed that patients with the higher risk-score had poorer prognosis than those with lower risk-score (third quantile of OS: 126.000 days, 95% CI: 73.588–178.412 vs. 532.000 days, 95% CI: 0.000–1,110.372; P<0.001). Conclusions Nonsmoking status, adenoid cystic carcinoma, and low preoperative stenosis length, were independent predictive factors for better OS of MAO patients. We proposed a nomogram and risk score system for survival prediction of MAO patients undergoing interventional bronchoscopy with good performance.