Reply to When randomized trials and observational data disagree: The case of preoperative versus postoperative chemoradiotherapy for esophageal cancer


We thank Palma et al for bringing to our attention the 3-arm randomized controlled trial (RCT) from China by Lv et al.[1] We agree with many of the limitations of any Surveillance, Epidemiology, and End Results (SEER) study that they point out. Specifically, confounding because of unmeasured covariates is always a potential source of error in observational studies. However, in our study, we conducted a thorough sensitivity analysis to assess the effects of unmeasured variables, such as clinical tumor staging.[2] A similar sensitivity analysis demonstrated that, even if the preoperative patients were half as likely to have comorbid disease as postoperative patients, our statistically significant results would hold.

Palma et al suggest that some of the postoperative patients may have received radiotherapy for palliation of early recurrence after surgery, confounding the results. This is unlikely, because SEER data only capture first course of treatment information within 4 months of the cancer diagnosis, and the likelihood of patients developing recurrence within the first few months after surgery is less than 5% according to recent randomized data.[3]

More important, the RCT by Lv et al warrants deeper exploration of the results beyond reading the results and conclusions of the abstract. The lack of difference in outcomes between preoperative and postoperative patients was noted for the entire cohort of patients, which also included patients receiving a “palliative resection or esophageal bypass.” When looking at the overall survival data pertaining only to patients who underwent curative, radical resection (see Lv et al, Fig. 1D), it is clear that there is a trend toward improved overall survival with preoperative treatment. The authors also acknowledge these findings in their discussion section. Therefore, we believe that the data as reported in this RCT, if anything, may lean in favor of preoperative therapy, particularly when considering the lower local recurrence rate with preoperative treatment.

Given the limitations of this RCT and our own observational study, we acknowledge that, without a well designed and sufficiently large RCT comparing preoperative versus postoperative treatment, the question of the optimal sequencing of perioperative treatment cannot be definitively answered. We applaud the investigators who are actively studying this topic in the ongoing Quality of Life in Neoadjuvant Versus Adjuvant Therapy of Esophageal Cancer Treatment Trial (QUINTETT) RCT and eagerly await their results.