Minimally invasive versus open oesophagectomy for cancer: propensity score analysis improves results' reliability but matching criteria need to be selected accurately.

Minimally invasive oesophagectomy has been increasingly implemented for surgical treatment of resectable oesophageal cancer, becoming the most widely used technique [1]. However, because of the discordant results obtained in major studies [2–5], specifically with regard to pulmonary complications and anastomotic leak rates, there is still no agreement in the scientific community whether fully minimally invasive or hybrid procedures should be considered as the gold standard at present. Hence, in the last years, increasing attention has been also focusing on long-term results [6]. In this study, Chen et al. [7] have used data from the Taiwanese Cancer Registry comparing overall survival of patients undergoing minimally invasive or open oesophagectomy for squamous cell carcinoma after neoadjuvant chemoradiotherapy. In total, 2250 patients met inclusion criteria with 487 subjects being treated with open and 1763 with thoracoscopic oesophagectomy. Overall survival analysis was conducted both before and after propensity score matching. Notably, after matching no difference between procedures was observed with regard to survival. Nevertheless, subgroup analysis showed thoracoscopic oesophagectomy as superior in the y-pathologic complete response stage, y-pathologic III stage and y-pathologic T0N+ stage. This is a very important and well-designed study that highlights the importance of both adequate patients’ selection and bias reduction. As already mentioned by the authors in their discussion, previous randomized controlled trials and meta-analyses have shown contradictory results, probably because including patients with different histologic entities, with or without neoadjuvant treatment and if neoadjuvant treated, without differentiation between the protocols [2, 8, 9]. Contrarily, Chen et al. [7] decided to focus their investigation only on patients with squamous cell carcinoma undergoing resection after chemoradiotherapy. Furthermore, the authors applied a propensity score matching, a statistical method that has been increasingly used in observational, retrospective studies to reduce the confounders between investigated groups, hence allowing a meaningful comparison between them [10]. However, matching criteria have to be selected consciously and appropriately. In the study by Chen et al., age, gender, Charlson score, tumour location, tumour length, y-pathologic T stage, y-pathologic N stage, y-pathologic TNM-stage, resection margin status, neoadjuvant radiotherapy dose and even chemotherapy regimen were selected as matching criteria [7].