Reply to “Neoadjuvant Therapy and Lymphadenectomy in Esophageal Cancer: Both Are Essential to Maximize Survival Benefit”
We thank Dr Pandey and colleagues for their insightful comments and the opportunity to clarify a number of points from our article.1 Their main point of criticism is that the number of removed lymph nodes is only a surrogate marker for systematic lymphadenectomy. Indeed, we have addressed this issue in the discussion. An adequate lymphadenectomy is also determined by the relevance of the removed lymph node stations and hence the surgical approach. One has to remove the relevant lymph nodes to have an impact on survival.
Our study has a strong design because it is based on the CROSS trial.2 In the CROSS trial, patients were randomized between surgery alone and neoadjuvant chemoradiotherapy (nCRT) plus surgery. Therefore, we can conclude that after surgery alone a higher number of resected nodes were associated with survival whereas this association was lost in a comparable group of patients who underwent surgery after nCRT. The randomization was stratified for treatment center, clinical N-stage and histology, which makes asymmetry between both treatment arms in, for example, disease extension at baseline, surgical technique or lymph node examination by the pathology department unlikely. Moreover, statistical adjustments excluded these characteristics as an explanation of the observed associations.
Pandey and colleagues refer to studies with weaker designs. In the study by Solomon et al,3 the effect of systematic lymphadenectomy was stratified for pathological N-stage (pN0 vs pN+), which carries the potential bias of stage migration. In the studies by Altorki et al4 and Portale et al,5 only a small minority underwent neoadjuvant therapy, which is exactly the matter that is at stake. The advantage of the study by Portale et al is that it compared surgical approaches (en bloc vs transhiatal), but the design of the trial was not randomized.
We agree that secondary analyses within a randomized trial can merely generate new hypotheses. We need confirmation from observational studies, and ideally randomized controlled trials, to exclude potential biases such as stage migration. Retrospective analyses on the extent of lymphadenectomy from observational data may have only limited value because various selection biases may play up. Groups probably differ for many more variables (tumor type, diagnostic workup, inclusion and exclusion criteria, etc), besides the extent of lymphadenectomy. A secondary analysis of a randomized controlled trial supplies a higher level of evidence than a retrospective comparison.
In the surgery-alone era, we have previously performed a randomized controlled trial comparing transhiatal resection with limited lymphadenectomy versus transthoracic resection with extended lymphadenectomy.6 The mean number of removed lymph nodes doubled after extended resection (from 16 ± 9 to 31 ± 14 nodes), and there was a nonsignificant trend toward improved long-term survival. In line with this, the present study supports a potential therapeutic benefit of extended lymphadenectomy in the surgery-alone group.
To quantify the therapeutic impact of extended lymphadenectomy for patients after neoadjuvant treatment, we need a randomized controlled trial in which patients receive nCRT, followed by a limited transhiatal or extended transthoracic resection. But the current evidence from our study suggests that this impact is less important than that from the surgery-alone era.