External validation of a clinical prediction model for mid-term mortality after video-assisted thoracoscopic surgery lobectomy for non-small cell lung cancer

Background: Few risk models designed to predict mid-term outcomes after thoracic surgery exist. Accurately predicting mid-term outcomes after lung cancer resection would be beneficial in clinical decision making. The objective of this study was to externally validate a previously developed clinical prediction model (Leeds model) for 2-year mortality after video-assisted thoracoscopic surgery (VATS) lobectomy. Methods: A multi-centre retrospective analysis of consecutive patients who underwent VATS lobectomy for primary lung cancer between 2012 and 2018 was performed. The primary outcome was 2-year mortality. Performance of the Leeds model was assessed using measures of discrimination and calibration. Cox proportional hazards regression analysis was used to identify factors independently associated with 2-year mortality in our cohort. Results: A total of 862 patients were included with a 2-year mortality rate of 12.9% (n=111). Patients were divided into three groups according to their class of risk as per the Leeds model. Log rank analysis demonstrated a significant difference in 2-year mortality between the three groups (P<0.001). After adjustment with Cox proportional hazards analysis, advanced age, lower percentage diffusion capacity of the lung for carbon monoxide (DLCO), higher systemic immune inflammation index (SII), higher tumour stage and the presence of nodal disease were all independently associated with 2-year mortality. Conclusions: The Leeds model demonstrated acceptable statistical performance. A number of additional pre-operative risk factors that are not included in the Leeds model were found to be independently associated with mid-term mortality after VATS lobectomy. Further work on predicting mid-term outcomes after VATS lobectomy for primary lung cancer is required.