Retrospective study using the propensity score to clarify the oncologic feasibility of thoracoscopic esophagectomy in patients with esophageal cancer.

Background: The present study aimed to clarify the long-term prognostic impact and oncologic feasibility of thoracoscopic esophagectomy (TSE) in patients with esophageal cancer in comparison with open thoracic esophagectomy (OTE).
Methods: Patients with esophageal cancer underwent surgically curative esophagectomy without neoadjuvant therapy from January 1991 to December 2008 and were analyzed retrospectively. Of 257 patients, 91 underwent TSE and 166 had OTE. Relations between the long-term prognosis after surgery, the surgical procedure, and clinicopathologic parameters were analyzed statistically. The propensity scores were calculated for all patients through a multiple logistic regression model that was optimized with Akaike's Information Criterion. Using Cox's proportional hazard model with prognostic variables and the propensity scores, we implemented a multivariate analysis for comparing the performance of two surgical methods.
Results: Patient characteristics and the incidence of perioperative morbidity or hospital death were similar for the TSE and OTE groups. Significantly more lymph nodes were dissected in the TSE group than in the OTE group (total p = 0.013; thoracic p = 0.0094; recurrent laryngeal p < 0.0001). The TSE group exhibited a more favorable prognosis after surgery than the OTE group in terms of overall survival (p = 0.011) and disease-specific survival (DSS) (p = 0.0040). Particularly in subgroup analysis of DSS, the TSE group had a favorable prognosis in upper thoracic esophageal cancer (p = 0.0053), invasive cancer (p = 0.046), node-positive cancer (p = 0.020), progressive cancer (p = 0.0052), cancer with lymphatic vessel invasion (p = 0.0019), and cancer without blood vessel invasion (p = 0.0081). In terms of DSS, the TSE group exhibited a more favorable prognosis than the OTE group regardless of the presence or absence of metastasis to lymph nodes around the thoracic (p < 0.0001) or recurrent laryngeal (p < 0.0001) nerves. TSE (p = 0.0430), lymph node metastasis (p = 0.0382), lymphatic invasion (p = 0.0418), and p stage (p = 0.0047) were independent prognostic parameters in the Cox's proportional hazard model with the propensity scores.
Conclusions: TSE can contribute to prolonged survival after surgery in patients with esophageal cancer by enabling precise thoracic lymph node dissection based on a magnified surgical field. TSE might have maximum oncologic benefit and minimum invasiveness for patients with esophageal cancer.