This is the first report of long-term survival of patients undergoing TME with NAC for locally advanced ESCC. Our study demonstrated that the combination of preoperative DCF therapy and TME for thoracic ESCC could be performed safely and might lead to better prognosis than the current standard treatments. In the JCOG9907 trial, which assessed NAC for ESCC, the 3-year OS was 62%, and the 5-year OS was 55%; these are the highest survival rates observed among similar randomized controlled trials (RCTs)2, 4, 22. The study by Natsugoe et al, which was an RCT assessing NACRT, had a 3-year OS of 57%; this is the highest survival among RCTs assessing NACRT2, 22, 23. Conversely, our study showed a 3-year OS of 78.8%. Although it is difficult to compare our results with those of RCTs because our study was a retrospective observational study, our results seemed to be as good or better than those of these RCTs. There are two possible reasons for this discrepancy: 1) the superiority of DCF therapy and 2) the curability and low invasiveness of the TME procedure.
Neoadjuvant DCF therapy in our study demonstrated relatively high toxicity; however, it had a remarkable tumour control effect, with 12.5% (2/16) of patients having complete response and 56.3% (9/16) having pathological response. In the JCOG9907 trial, the complete response rate of NAC (CF therapy) was 2.4%, and the clinical response rate was 38%4. Therefore, triplet DCF therapy seems to be more effective than doublet CF therapy. Hara et al reported a phase II study on preoperative DCF therapy and found good results, with 17.1% (7/41) achieving a Grade 3 response and 34.1% (14/41) achieving a Grade 2 response24. Therefore, they saw greater effectiveness of DCF than we did. However, in their study, DCF therapy consisted of docetaxel and cisplatin at 70 mg/m2 on day 1 and 5-FU at 750 mg/m2/day on days 1–5 repeated three times every 3 weeks, so their patients received a higher cumulative dose than those in our study24. In Natsugoe’s study, NACRT consisted of 40 Gy of radiation and CF therapy (cisplatin: 7 mg over 2 h and 5-FU: 350 mg over 24 h during radiotherapy), and in the NACRT arm, 13.6% (3/22) of patients had a Grade 3 response and 31.8% (7/22) had a Grade 2 response23. Therefore, we conclude that neoadjuvant triplet chemotherapy is superior to doublet chemotherapy and comparable to CRT in terms of patient response. In Japan, the JCOG1109 trial, which is a triple-arm RCT comparing CF versus DCF versus radiotherapy with CF as preoperative therapy, is now ongoing25.
In terms of AEs, DCF therapy showed a higher rate of Grade 3/4 toxicities than what has been previously observed with CF therapy or CRT, especially in hematological components (leucopoenia and neutropenia)4, 23, 24. However, leucopoenia and neutropenia were manageable, and neither our study nor Hara’s phase II study had any chemotherapy-related death24. Therefore, neoadjuvant triplet chemotherapy seems to be safe. Although the DCF therapy in Hara’s study had a higher dose than ours, there was no difference in the extent of AEs between these two studies23. Hence, we might have been able to introduce more powerful DCF therapy.
As we mentioned in the introduction, the standard treatment for locally advanced ESCC in Japan is esophagectomy with 3FD, and it is believed that 3FD requires TTE. Therefore, TTE was performed for esophagectomy in the JCOG9907 trial, and Hara’s study (Natsugoe’s study also suggested that TTE was performed; however, details were not provided)4, 23, 24. In contrast, we performed TME with 3FD in this study. TME does not require thoracotomy and therefore is associated with fewer respiratory complications than TTE11. Infectious complications, including respiratory complications, may be directly linked to prognosis after esophageal cancer surgery8, 9; this was shown in the exploratory analysis of JCOG990710. In our study, we observed only two cases of pneumonia as infectious complications, and this result might have led to a better prognosis.
Some argue that TME is less radical than TTE. Previously, it was difficult to dissect the mediastinal lymph nodes, especially the nodes around the left recurrent laryngeal nerve or carina, without a transthoracic approach. However, the development of videoscopy and surgical devices has enabled mediastinal dissection without a transthoracic approach15, 16. We performed upper to middle mediastinal dissection under single-port inflatable mediastinoscopy from the left neck. We dissected 57.5 lymph nodes per operation and could radically dissect lymph nodes around the left recurrent laryngeal nerve, carina, and bilateral main bronchi (Fig. 2). This less invasive procedure may be beneficial to patient prognosis. In Hara’s study, the 2-year progression-free survival was 74.5%, and the 2-year OS was 88.0%24; in contrast, in our study, the 2-year RFS was 73.3%, and the 2-year OS was 93.3%. Even though a higher dose of DCF therapy was administered in Hara’s study24, our study had comparable results. This might indicate the superiority and radicality of our TME procedure, despite its less invasive nature.
However, our procedure was not without complications. We had three cases of chylothorax that required re-operation. All three cases could be controlled by laparoscopic transhiatal thoracic duct ligation, and patients recovered quickly26. Our TME procedure usually preserved the thoracic duct; however, highly advanced tumours would have required ligation or excision of the thoracic duct. Another complication was anastomotic stricture. We did not observe anastomotic leakage during treatment; however, six cases of anastomotic stricture requiring balloon dilation were observed after discharge. We used a circular stapler for esophagogastric tube anastomosis; it might have been possible to reduce anastomotic stricture by using a linear stapler or hand-sewing.
Combined neoadjuvant DCF and radical esophagectomy under TME showed fairly good results; however, this was a single-center, retrospective study in a small number of patients. To truly demonstrate the usefulness of our treatment, it is necessary to perform RCTs to verify both the superiority of neoadjuvant DCF and the superiority of TME to TTE. Regarding neoadjuvant DCF, as we mentioned above, the JCOG1109 trial, which compares doublet chemotherapy, triplet chemotherapy, and chemoradiotherapy, is currently ongoing. Therefore, our future direction should be to promote an RCT comparing the superiority of TME to TTE for resectable locally advanced esophageal cancer.