Automated quantification of interstitial lung abnormalities and emphysema on computed tomography: a predictive marker for postoperative pulmonary complications after esophagectomy

Background: Postoperative pulmonary complications (PPCs), including pneumonia, acute lung injury, and acute respiratory distress syndrome, are common morbidities associated with mortality following esophagectomy. This study aimed to assess the association of chest computed tomography (CT) texture features with PPCs following esophagectomy. Methods: Between 2016 and 2022, data from 765 patients who underwent upfront esophagectomy were analyzed. Deep learning–based automatic quantifi cation was used to identify interstitial lung abnormalities (ILAs) and emphysema on the preoperative chest CT. Logistic regression analyses were performed to identify risk factors for PPC. Results: The mean age of the patients was 64.72 ± 8.27 years, and 698 (91.2%) patients were male. PPCs developed in 129 (16.2%) patients. Patients with PPCs were more likely to have current smoking status, lower lung function, and open esophagectomies than patients without PPCs. The PPC group also exhibited more emphysema (0.236% vs. 0.123%, p= 0.005) and ILAs (0.342% vs. 0.149%, p 0.001) on chest CT scans compared with patients without PPCs. Multivariable logistic analysis demonstrated that emphysema (odds ratio [OR] 1.158, p = 0.004) and ILA (OR 1.364, p 0.001) were risk factors for PPC after adjusting for other confounding factors. Conclusions: The extent of emphysema and ILA, quantifi ed by automated software, was signifi cantly associated with PPC following esophagectomy. Future research should focus on perioperative management strategies for patients with emphysema or ILA and esophageal cancer.