Learning curve for minimally invasive oesophagectomy of oesophageal cancer and survival analysis

The incidence of oesophageal carcinoma has increased significantly over the past 20 years, and it is currently the 6th leading cause of cancer death [1, 2]. Radical surgical resection remains the primary treatment for early and locally advanced lesions [3], but it has a relatively high rate of morbidity and mortality [4], which may be attributed to lymph node dissection [5]. Minimally invasive oesophagectomy (MIE) results in low morbidity and mortality rates with equal mid- and long-term oncological outcomes [6‐10].

The minimally invasive approach is performed via a distinct view of the surgical anatomy, and specialised surgical schemes and skills are required. Although learning curves and other parameters for lymph node dissection were reported over the past few years [11‐14], systematic analyses of these learning curves, the clinical indexes and survival outcomes using a large number of cases or long monitoring times remain limited. From July 2010 to August 2016, 214 patients underwent MIE for oesophageal cancer and were followed up. The clinical parameters were analysed to examine the learning curve and characteristics of MIE. A clinical comparative study of mature MIE versus open oesophagectomy (OE) for oesophageal carcinoma was performed, and a Cox proportional hazards regression analysis was used to determine the clinical risk factors for overall survival.

MIE reduces postoperative pain, drainage volume and inflammatory reactions and shortens the hospital stay. However, MIE requires surgeons to have a deeper understanding of endoscopic anatomy, surgical procedure, accidental bleeding, and lymph node dissection process, which may contribute to the quality of the surgery. MIE has been performed since July 2010 in our department. Our dataset showed that surgical process proficiency was achieved after at least 57 surgeries with adequate lymph node dissection and favourable safety outcomes.

Dhamija et al. [11] reported the learning curve for lymph node resection in MIE for oesophageal cancer and suggested that the ability to dissect lymph nodes completely during MIE was affected by the surgeons’ experience, which may accumulate over time. A significant increase in experience was achieved after the first 25 cases in their study. The number of lymph nodes was regarded as an important measurement of lymphadenectomy, and different lymphadenectomy thresholds were proposed based on various scholars’ findings. Dutkowski [16] found that the diagnostic sensitivity of lymph node metastasis increased with an increasing number of dissections. When the dissection number reached 12, the sensitivity reached at least 90%, and more lymph node dissections only increased the complication rate because the diagnostic sensitivity reached a plateau. Rize [17] believed that the ideal number of lymph node dissections should be no less than 18. There was no significant difference in the pathological characteristics between the 4 stages except for differentiation. The numbers of lymph nodes dissected in each stage were 12.65 ± 4.13, 15.91 ± 3.36, 20.16 ± 7.71, and 22.67 ± 7.39 with a significant difference between the stages. To intuitively observe the upward trend, scatter plots were drawn, and the cumulative sum was calculated. The inflection point was observed at the 57th case in the present study, which suggests that proficiency in lymphadenectomy was gained after 57 patients underwent the procedure. Oncological benefits were implied with this progress. These benefits were due to the actions implemented during the 2nd stage, including performing hybrid surgeries and attending academic conferences.

MIE is divided into the McKeown technique (cervical anastomosis) and Ivor-Lewis technique (chest anastomosis). Surgeons prefer the former technique because of the maturity of the surgical procedure [18], and the latter technique is not widely performed due to its difficulty and the risk of the occurrence of a thoracic fistula. The surgical position and anastomosis type were also different in each clinical unit [19]. However, the Ivor-Lewis method was suggested as less invasive, with a shorter operation time, lower anastomotic leakage rate and lower recurrent laryngeal nerve injury rate than the McKeown method [20]. An increase in the proportion of Ivor-Lewis procedures appeared during the 4th stage in our series, after which the lymphadenectomy proficiency reached a relatively high level.

The importance of the surgical procedure was for accurate staging and an improvement in survival. The differences in 5-year overall survival and disease-free survival between the 1st 3 stages were not significant (22.22%, 22.58%, 25.53%, P = 0.9359) (21.22%, 19.35%, 19.15%, P = 0.8361), which may be due to the standard surgery procedure and sufficient lymph node dissection, even when performed in the 1st stage (12.65 ± 4.13).

We retrospectively analysed 327 patients who underwent oesophagectomy for oesophageal cancer at our institution. A total of 157 patients underwent MIE (from the 3rd stage), and 170 patients underwent open oesophagectomy (OE). The MIE group had a smaller bleeding volume (P = 0.0000) and shorter operation time (P = 0.0132) than the OE group. There was no significant difference in the complication rates between the two groups. However, the MIE group had a lower incidence of postoperative pulmonary inflammation (12.1% vs. 29.4%, P = 0.0000), which is very similar to the results of the randomised controlled trial of Biere et al. [21] (12% vs. 34%, P = 0.005). According to Parameswaran et al. [22] and Verhage et al. [23], the postoperative respiratory complication rate of OE was higher than MIE. This result may be due to the large incision, rib fractures, respiratory muscle detachment, injury and repair of the diaphragm, and retention of sputum. Therefore, MIE was superior to OE for protecting respiratory function.

The open group had more lymph nodes dissected than the MIE group (23.99 ± 10.15 vs. 21.26 ± 7.72, P = 0.0069), but the number of lymph nodes dissected in the MIE group was similar to a previous study [24]. Lymph nodes were dissected from more sites in the MIE group than the OE group (4.87 ± 1.43 vs. 4.16 ± 1.26, P = 0.0000), which may be due to the comprehensive view available with the thoracoscope. Subgroup analysis showed that the dissection rates of the No. 106rec (right or left) lymph nodes reached 78.4% in the MIE group and 23.5% in the open group (P = 0.0000). The No. 106rec (both right and left) lymph node dissection rate in the MIE group was 64.7%, which was significantly different from the 7.1% in the open group (P = 0.0000). No.106tbL lymph node dissection followed a similar trend (31.8% vs. 16.5%, P = 0.001). However, the open group had a statistical advantage (47.1% vs. 12.1%, P = 0.0000) in left gastric artery lymph node dissection, which reflects the capability limitations of the thoracic surgeon during laparoscopic abdominal lymphadenectomy.

Survival outcomes are crucial for the treatment of carcinoma. The 1-year survival rates between MIE and OE were not significantly different [25]. Li et al. [26] reported that the 1-, 2-, and 3-year survival rates of patients who underwent MIE were not significantly different from patients who underwent the open procedure, but the proportion of early-stage disease in the MIE group was higher than open procedure group. A systematic review [27] also provided information that the proportion of early-stage disease in the MIE group exceeded the open procedure group, which may be inevitable while the surgeons are still learning. Mitzman et al. [28] performed a propensity analysis using data from the National Cancer Database of the United States, and equivalent oncological outcomes and survival outcomes were found for the MIE and open procedure groups. No significant differences in long-term survival were found in patients who underwent robotic-assisted minimally invasive oesophagectomy (RAMIE), MIE or OE. Therefore, surgeon expertise and experience may be the most important aspect [29]. The overall 3-year survival rate did not differ significantly between the groups in our series (MIE group 46.00% vs. open group 37.14%, P = 0.2035). This study did not provide evidence of improvements in the survival outcomes of patients who underwent MIE. Therefore, a Cox proportional hazards regression was used to evaluate the effects of clinical factors on survival for patients who underwent radical surgery with curative intent. The clinical parameters included age, surgical procedure (MIE or open), differentiation degree (poor or middle-high), tumour excision area (≥ 5 cm or < 5 cm), infiltration depth (T1, T2, T3), and the number of positive lymph nodes. The maximum diameter of the tumour excision area and the number of positive lymph nodes significantly influenced postoperative survival. There was no evidence that the surgical procedure (MIE or open) exerted significant influence on the prognosis of patients with oesophageal cancer.

There are some limitations to this study. The management of chest drainage and nasogastric drainage and inflammatory factors were not considered, but these factors are important indicators for presenting a learning curve. The number of lymph nodes was used as a reflection of lymphadenectomy completeness, and this number may be influenced by a fracturing of the nodes or pathological diligence. Our study was a retrospective analysis, and challenging cases were avoided in the early stage of the learning process. Therefore, selection bias was inevitable. Only a few patients received neoadjuvant therapy, which was recommended by the medical guidelines.

Our findings provide evidence of the learning curve for MIE and its short-term benefits. However, the results provide no evidence that MIE and OE contribute differently to postoperative survival outcomes.