P107 Predictors of mortality in patients undergoing lung cancer surgery. (15th November 2016)

• Record Type: Journal Article
• Authors: Law, H; Foden, P; Evison, M; Wallace, F; Ashworth, A; Shah, R; Crosbie, P; Booton, R
• Is Part Of: Thorax. Volume 71(2016)Supplement 3
• Page Start: A141
• Page End: A141
• Publication Date: 2016-11-15
• Subjects: Chest -- Diseases -- Periodicals; Thorax; Chest -- Diseases; Periodicals; Periodicals; 617.54
• DOI: 10.1136/thoraxjnl-2016-209333.250
• Languages: English
• ISSNs: 0040-6376

Abstract

Abstract : Introduction: Surgical resection is a treatment of choice for patients with early stage lung cancer and physiological measurements are routinely used to help predict post-operative risk, particularly in the 'high-risk' patient group. At present, there is a lack of concordance between current available guidelines incorporating the use of such parameters aimed to guide decisions on surgery for high-risk patients with lung cancer. As a result, the decision to operate will differ for a particular patient depending on which guidelines are consulted. We aim to identify which parameters best predicts post-operative mortality and whether this information can be used to construct a more encompassing pre-operative risk prediction model to help guide these difficult decision processes. Methods: Retrospective analysis of all patients undergoing CPET (cardio-pulmonary exercise testing) prior to lung cancer surgery between 01/01/2012 and 31/12/2015 was carried out. Age, BMI along with pre-operative and post-operative predicted physiological parameters were reviewed and statistical analysis performed. We also looked at survival based on type of surgery (sub-lobar, lobar, pneumonectomy), histology and cancer staging. Results: Single variable analysis of the 178 patients identified that low BMI (p = 0.005) and PPO DLCO% (p = 0.004) were associated with greater post-operative mortality risk. There was a statistically significant difference between different cancer stage and type of surgery as expected. Using the probabilities from the logistical regression model to predict one-year mortality gives an AUC of 0.764. A probability cut-off of 0.167 used to predict whether a patient will die within one year of surgery provides a sensitivity of 76.5%, specificity 66.4%, PPV 35.1% and NPV 92.2%. Conclusions: Contrary to current guidelines, CPET data did not seem to carry statistically significant weighting in determining post-operative mortality outcomes in our patient group with BMI and PPO DLCO% showing a stronger, statistically significant association. Absolute% change between pre and PPO FEV1 values appears to be a good predictor of one-year mortality following surgery. Further work is required but early analysis suggested that parameters such as BMI, PPO DLCO% and absolute post-operative change in FEV1% can be used to construct a pre-surgical prediction model for 'high-risk' patients undergoing surgery for lung cancer.