Propensity score analysis of oesophageal cancer treatment with surgery or definitive chemoradiotherapy

Background The role of treatments involving surgery versus definitive chemoradiotherapy (dCRT) for oesophageal cancer remains controversial. Methods Consecutive patients with oesophageal cancer were studied. Those whose treatment involved surgery alone or who received neoadjuvant chemotherapy or chemoradiotherapy were compared with those receiving dCRT. Multiple regression models, including propensity scores, were developed to assess confounding factors associated with undergoing surgery or dCRT, and the risk-adjusted association between treatment and survival. Results From a total of 727 patients, regression adjustment to control for bias created a cohort of 521 patients available for comparison (277 in the surgery group and 244 in the dCRT group). Local and distant recurrence rates were 10·1 and 22·0 per cent respectively after surgery, compared with 26·2 and 11·9 per cent following dCRT (P < 0·001). Median survival, and 2- and 5-year survival rates after surgery were 27 months, 53·8 and 31·0 per cent respectively, compared with 28 months, 54·2 and 31·9 per cent after dCRT (P = 0·918). On multivariable analysis, disease-free survival was related to endosonographic tumour category (hazard ratio (HR) 0·76, 95 per cent confidence interval 0·10 to 6·04 for T1; HR 1·57, 0·21 to 11·58 for T2; HR 2·12, 0·29 to 15·49 for T3; HR 3·07, 0·41 to 23·16 for T4; P = 0·003, in relation to T0 as reference), lymph node metastasis count (HR 1·10, 1·04 to 1·15; P < 0·001) and total disease length (HR 0·96, 0·93 to 1·00; P = 0·041). Conclusion There was no difference in survival after oesophageal cancer treatment involving surgery or dCRT. A challenge to the role of surgery