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

We read with interest the recent study by Wojcieszynski and colleagues, who used the Surveillance, Epidemiology, and End Results (SEER) database to assess “whether the sequencing of chemoradiotherapy (CRT) has an effect on survival and cardiopulmonary mortality” for esophageal cancer.1 Although chemotherapy data are not available in SEER, the authors use the delivery of radiotherapy (RT) as a surrogate for CRT, assuming that RT is “almost always administered concurrently with chemotherapy.”1 The findings are striking: the use of preoperative RT was associated with a 10% improvement in 5-year survival compared with postoperative RT (P < .0001). The sequencing of CRT relative to surgery is an area of clinical uncertainty, with arguments supporting both approaches. One advantage of neoadjuvant CRT is that RT volumes are usually smaller; conversely, with the adjuvant approach, some patients may avoid CRT if their surgical staging indicates very early stage disease. The authors indicate that no randomized controlled trials (RCTs) are available to address this issue; but indeed there is at least 1, and the differences in results are quite informative. That trial randomized 238 patients who were undergoing surgery for locally advanced thoracic squamous cell esophageal carcinoma to preoperative CRT, postoperative CRT, or surgery alone. There were no differences in survival between the preoperative and postoperative arms, and both were superior to surgery.2 Although there are some differences between the baseline populations in the 2 studies, we are faced with a situation in which RCT data conflict with the results from a population-based study. Conceivably, the 2 studies may differ because of a lack of generalizability of the RCT, but it is more likely that the differences arise because randomization can control for both measured and unmeasured confounders, whereas an observational study cannot control for the latter. One unmeasured confounder that threatens the validity of the SEER study is comorbidity. In some situations, controlling for comorbidity can reverse the results of an observational study: for example, SEER data indicate that postoperative RT for N2 nonsmall cell lung cancer improves overall survival,3 whereas the more robust SEER-Medicare data (controlling for comorbidity) indicate a trend toward harm.4 Furthermore, radiation dose is not available in SEER, and some patients in the postoperative cohort in the study by Wojcieszynski et al actually may have received RT for palliation of early recurrence. The optimal timing CRT for esophageal cancer remains uncertain, and an ongoing RCT is comparing quality-of-life outcomes with the 2 approaches.5 We encourage additional RCTs to further examine survival outcomes between the 2 groups. David A. Palma, MD, MSc, PhD1George B. Rodrigues, MD, MSc1Richard Malthaner, MD, MSc2 1Department of Radiation Oncology, London Health Sciences Center, London, Ontario, Canada 2Department of Thoracic Surgery, London Health Sciences Center, London, Ontario, Canada

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.