We congratulate Dr Yang and colleagues for an important analysis investigating the utility of adding mediastinal lymph node dissection (MLND) to pulmonary metastasectomy for colorectal cancer (CRC). Using propensity score matching they convincingly show that while well-recognized factors — preoperative carcinoembryonic antigen (CEA) and higher metastatic number — have a significant adverse effect on survival, resection of mediastinal lymph nodes does not.1

This is important because the practice of MLND is being increadingly used during lung metastasectomy. Yang et al. comment that in lung cancer “the therapeutic effect remains unclear”.1 This is a correct reading of the evidence. There have been five randomized controlled trials (RCTs) of MLND in lung cancer surgery. There were clinically important “surgically-related morbidities” and the “high risk of bias in these trials makes the overall conclusion insecure”.2 This new evidence from a well conducted study by Yang et al. should be heeded.

However, we disagree with their statement about the PulMiCC trial: “The results of that study indicated that patients who had undergone surgical intervention had better survival outcomes than those who had been actively monitored”.1 PulMiCC has now been published in full.3, 4 There was good recruitment of 512 patients. After early exclusion of 28 patients whose nodule were found not to be CRC there was a cohort of 484 patients with data collected prospectively to trial standards. Randomization was difficult because of strongly held views based on uncontrolled clinical reports. As a result, 391 patients were managed by clinical decision.4 Their survival is shown in the upper panel of Figure 1. Survival of patients selected for metastasectomy was ~60%, similar to the best of reported series and better than those selected to not have an operation. There were clear differences in prognostic factors between the groups, all of which favored survival of those having metastasectomy. More had a solitary metastasis 69 versus 35 Fewer metastases overall (median) 1 versus 2 Lower median CEA 2 versus 3 ng/l Fewer with elevated CEA 12% versus 20% Fewer had liver metastases 28% versus 36% Better lung function (% predicted FEV1) 96% versus 87% More with zero ECOG impairment 68% versus 36% They were younger 67 versus 72 years

Survival of 391 patients with colorectal cancer-related lung metastases treated electively and 93 whose treatments was randomly assigned

All of these factors were balanced in the two arms of the PulMiCC RCT.3 The survival curves overlap and although the median survival was longer in the control arm at 45.6 versus 42.0 months there was no significant difference at any time point.3

In their insightful discussion, Yang et al. suggest that propensity score matching might be a way around the obstacles to randomization encountered in PulMiCC. But these were due to the reluctance of the clinicians (not the patients) to randomize because of their strong prior beliefs. A better understanding of the whole PulMiCC study shows that the belief in a survival benefit from metastasectomy is insecure and should make thoughtful clinicians reconsider belief in its effectiveness.

The authors have no conflicts of interest to declare.