Impact of prior thoracoscopic experience on the learning curve of robotic McKeown esophagectomy: a multidimensional analysis

Purpose Left upper mediastinal lymph node dissection (UMLND)-a technically demanding step of McKeown esophagectomy-is frequently complicated by recurrent laryngeal nerve (RLN) palsy. Under the hypothesis that robotic esophagectomy (RE) could increase the safety and feasibility of UMLND, we retrospectively investigated the degree to which a pre-existing experience in video-assisted thoracoscopic esophagectomy (VATE) may affect the learning curves of this critical part of RE. Methods Surgeon A had previously performed > 150 VATE procedures before transitioning to RE. While surgeon B had previously assisted to 50 RE, his pre-existing VATE experience consisted of less than five procedures. A total of 103 and 76 McKeown RE procedures were performed by surgeons A and B, respectively. The learning curve of left UMLND for each surgeon was examined using the cumulative sum method. Results The inflection point of RLN palsy for surgeon A occurred at patient 31. While the nerve palsy rate decreased from 32.3 to 4.2% (p < 0.001), the number of nodes harvested during left UMLND did not appreciably change. Surgeon B showed a bimodal learning curve for RLN palsy with primary and secondary inflection points at patients 15 and 49, respectively. The RLN palsy rate initially decreased from 66.7% (patients 1-15) to 14.7% (patients 16-49), followed by an additional decline to 3.7% (patients 50-76). However, the number of nodes harvested during left UMLND showed a downtrend which was paralleled by decreasing rates of RLN palsy. These results indicate that surgeon B has not yet reached an ideal balance between an extensive UMLND and nerve protection. Conclusion The pre-existing VATE experience seems to affect the learning curves of left UMLND during RE.