Pneumonectomy and broncho-pleural fistula: predicting factors and stratification of the risk.

The goal of the study is to evaluate the different risk factors and stratify the patients, before the surgery, into distinct risk classes. We retrospectively reviewed pre, peri, and postoperative outcomes of 366 consecutive patients who underwent pneumonectomy for lung cancer between the last 10 years (2009-2019). We classified the patients into four classes, depending on preoperative assessments. Differences between groups were assessed with the log-rank test. Multivariable Cox proportional hazards regression analysis was used to assess the independent prognostic significance of the variables associated with the development of BPF at univariate analysis. Finally, we performed non-linear [artificial neural network (ANN)] multiple regression analyses. All tests were two-sided, and p values < 0.05 were considered significant. Fifty-one patients (13.9%) out of 366 developed BPF. Male sex (p = 0.048), right side (p = 0.015), postoperative pulmonary complications (p = 0.0139) and adjuvant treatments (p = 0.0169) were the independent predicting factors of fistulas in multivariate analysis. The right side (p = 0.043) and adjuvant treatments (p = 0.032) were the independent predicting factors of BPF after the ANN analysis. Based on multivariate and artificial neural network analysis and our experience, we observed a trend of growing risk of BPF in the first 4 weeks (early fistula), considering the four classes. Preoperative differentiation of the patients into four risk classes could allow a correct stratification of the growing risk of developing early BPF. This information could be significant to share with patients and the other physicians during the decision-making process, to minimise the risk of BPF.