The incidence of esophageal cancer increases, annually resulting in 572.000 new cases, and with 508.600 estimated esophageal cancer-related deaths worldwide.1 In the Netherlands, the standard curative treatment for patients with a diagnosis of esophageal carcinoma involves a combination of neoadjuvant chemoradiotherapy (or perioperative chemotherapy) followed by an esophagectomy with two-field lymphadenectomy. Both the neoadjuvant chemoradiotherapy and lymph node dissection (LND) during esophagectomy have the potential to eliminate possible metastatic lymph nodes.2

To date, no consensus exists regarding the optimal routine LND.3 The Dutch national guideline does not specify the number of lymph node stations (LNSs) to be dissected,4 and although the Dutch Upper GI Cancer Audit (DUCA) uses the quality indicator of at least 15 dissected lymph nodes in the resection specimen,5 the locations of the LNSs are not specified. Besides, different classification systems to address lymph node metastases of esophageal cancer are used worldwide.6

To overcome this problem, Schuring et al.6 proposed one uniform TIGER-study classification system with defined anatomic landmarks based on the 11th edition of the ‘Japanese Classification of Esophageal Cancer’ (JES) and the 8th edition of the ‘American Joint Committee on Cancer/Union for International Cancer Control’ (AJCC/UICC) classification.7 Even when consensus has been reached on which LNSs to dissect (the extent of LND), it is expected that discrepancies remain in the definition of anatomic boundaries of each LNS. This became apparent during the development of the procedure-specific competency assessment tool (MIE-CAT) by Ketel et al.8 to assess the surgical performance of minimally invasive esophagectomy (MIE) and to determine where most discrepancies were present in assessing the LND phases during MIE. Only with consensus on the anatomic boundaries can the resection of each LNS be adequately assessed.

To enhance comparability of the LND among surgeons and institutions worldwide for research purposes and to generate uniform treatment strategies, explicit anatomic boundaries of each LNS, including both superior/inferior, ventral/dorsal, and left/right borders, are needed.

This study aimed to investigate the perspectives of Dutch esophageal surgeons regarding the extent of their LND and their anatomic boundaries per LNS during LND in MIE. This insight will be the essential foundation toward a consensus on the definition of a complete LND during MIE for esophageal cancer.

Methods

Study Design

This cross-sectional survey study among Dutch MIE surgeons was the first step in a larger study project toward consensus on the optimal routine extent of LND (specific stations to dissect) and the definition of anatomic boundaries per LNS during LND in MIE for esophageal cancer.

Study Participants

A nationwide survey targeted esophageal surgeons in the Netherlands. From each of the 15 Dutch hospitals performing more than 20 (robot-assisted) MIEs per year according to the nationwide guideline,4 two surgeons were asked to participate. These surgeons were identified through previous esophageal surgery collaborations.9 In April 2023, a total of 30 Dutch esophageal cancer surgeons were invited to participate via extended e-mail invitations to participate in an English web-based survey. If the survey was not completed after 2 weeks, three subsequent reminders were sent (in April, May, and June) to encourage participation. Surgeons who did not respond or complete the survey after receiving these reminders (n = 6) were excluded.

Survey, Data Collection, and Outcome Measures

The survey was created using Castor Electronic Data Capture and contained five sections (Table S1). First, the participants provided information related to their experience, routine approach, and techniques of MIE. Second, the surgeon’s routine extent of the abdominal and thoracic LND for distal esophageal adenocarcinoma was collected, based on both the TIGER7 and the Japanese Classification of Esophageal Cancer (JES) classification. If at least 85% of the participating surgeons routinely dissected an LNS, it was considered that general agreement had been reached and therefore part of the consensus-based routine LND. Third, anatomic boundaries according to the TIGER definition7 of each indicated routinely dissected LNSs were displayed, and (dis)agreement with this definition was asked. Additionally, surgeons had the opportunity to provide in-practice superior, inferior, ventral, dorsal, left, and right anatomic boundaries per LNS. Fourth, the relevance of a complete LND and factors influencing the in-practice anatomic boundaries per LNS in LND were questioned. Fifth, the origin of each surgeon’s point of view regarding the anatomic boundaries of the LNSs he or she used were subtracted.

Fully completed surveys were included for analyses, and the analyses were conducted anonymously. Data are presented as means with standard deviations (SDs) or as numbers with percentages (%). Statistical analysis was performed using IBM SPSS Statistics for Windows version 28.0 (IBM Corp., Orchard Road, Armonk, New York, USA).

Results

Surgeon Characteristics

In June 2023, 24 esophageal surgeons from 12 Dutch hospitals completed the survey (response rate, 80%). The characteristics of these surgeons are summarized in Table 1. Regarding surgical approach, the majority of the hospitals performed a transthoracic MIE (n = 23,95.8%), and an intrathoracic anastomosis was preferred in the majority of their cases (n = 20,83.3%). The average experience with MIE was 10 years, and the average date for starting MIE was 2013 (range, 2003–2019).

Table 1 Background questions regarding experience, routine approach, and techniques of MIE

Routine Extent of LND

The response regarding surgeons’ routine extent of abdominal and thoracic LND for distal esophageal adenocarcinoma is displayed in Fig. 1 and Table 2. All 24 surgeons (100%) agreed that the abdominal part of the LND should include at least the left and right paracardial, left gastric artery and vein, and celiac trunk stations. The majority also routinely dissected the proximal splenic artery (95.8%) and common hepatic artery lymph nodes (95.8%). All 24 surgeons (100%) agreed that LND in the thorax should include at least the subcarinal, middle, and lower mediastinal paraoesophageal LNSs, and most of the surgeons also routinely dissected the left and right pulmonary ligament (95.8%) as well as the upper mediastinal paraoesophageal (87.5%) lymph nodes. The LNSs not routinely dissected for a distal esophageal adenocarcinoma included the hepatoduodenal ligament (portal vein [66.7%], proper hepatic artery [56.5%]) and aortopulmonary window (37.5%), as well as the lower paratracheal (30.4%), distal splenic artery (20.8%), upper paratracheal (0%), and all cervical (0%) LNSs.

Fig. 1
figure 1

Adapted from the TIGER study protocol with permission of the authors.

Routine extent of the lymph node dissection (LND) for distal esophageal adenocarcinoma. The percentage indicates the number of Dutch surgeons dissecting the stations.

Table 2 Routine LND during MIE for distal adenocarcinoma

Interestingly, within the 12 respondents’ centers, differences between surgeons of the same center were found regarding their routine extent of LND. These differences between surgeons of the same center concerned abdominal LNSs of the hepatoduodenal ligament (the portal vein [n = 4, 33.3%] and proper hepatic artery [n = 5, 41.7%]), distal splenic artery (n = 3, 25.0%), proximal splenic artery (n = 1, 8.3%), and common hepatic artery (n = 1, 8.3%). In the thorax, this concerned LNSs of the lower paratracheal lymph nodes (n = 3, 25.0%), aortopulmonary window lymph nodes (n = 3, 25.0%), upper mediastinal paraoesophageal lymph nodes (n = 1, 8.3%), and pulmonary ligament lymph nodes (n = 1, 8.3%).

Influences on the Anatomic Boundaries and Relevance of the Boundaries

The surgeons indicated that the anatomic boundaries of each LNS (“100 percent complete dissection per LNS”; mean, 4.3 ± 0.6) were more relevant than the extent of the LND (“number of LNSs”; mean, 3.7 ± 0.9). Moreover, both the extent of LND and the anatomic boundaries per LNS were influenced mostly by patient factors (66.7% and 75%, respectively) and tumor location (79.2% and 62.5%, respectively), as shown in Table 3.

Table 3 Parameters influencing the extent of LND and anatomic boundaries

Anatomic Boundaries

Tables 4 and 5 display the six directions of the thoracic and abdominal anatomic boundaries provided by the surgeons, together with the TIGER definitions of these stations. Between these anatomic boundaries, both consensus and differences were found. For example, for the majority, the ventral anatomic boundary of the celiac trunk LNS was the left gastric artery (× 11), and for some, it was the stomach (× 1), the overlying peritoneum (× 1), or the specimen (× 1). However, more distinctive differences were found for the inferior anatomic boundary of the celiac trunk LNS. These included the left gastric splenic and hepatic artery (× 5), the pancreas (× 4), and the celiac trunk (× 3). In addition, a significant number of anatomic boundary answers included “arbitrary” or “not applicable or no clearly described boundary.”

Table 4 Anatomical boundaries from the thoracic LNS provided by surgeons.
Table 5 Anatomical boundaries from the abdominal LNS provided by surgeons.

To illustrate similarities and differences in the provided anatomic boundaries, Fig. 2a–f shows a graphic representation of the variations regarding the celiac trunk LNS. Every line represents the anatomic boundaries provided by one individual surgeon, with corresponding colors of the surgeons between the Figures. Lines directly next to each other demonstrate the same anatomic boundary provided by the surgeons.

Fig. 2
figure 2

The anatomic boundaries of the celiac trunk lymph node station (LNS). Each colored line represents the survey results of a particular surgeon. (A) Superiod, (B) inferior, (C) ventral, (D) dorsal, (E) left, and (F) right, after ligation (A, D, E, and F) and before ligation (B and C). LGA, left gastric artery; CHA, common hepatic artery; SA, splenic artery; CT, celiac trunk; PV, portal vein; AG, adrenal gland; VC, vena cava

Origin of Point of View

Overall, 20 surgeons (83.3%) indicated that they have changed their point of view regarding the extent (number of LNSs) and anatomic boundaries of their LND over time. Most (n = 11, 45.8%) found the extent more important over time, whereas others found the extent less important (n = 4, 16.7%). Moreover, other surgeons mentioned that the anatomic boundaries per LNS became more important (n = 4, 16.7%) or less important (n = 1, 4.2%). Furthermore, the surgeons indicated that both literature (n = 17, 70.8%) and sharing experiences with other surgeons/centers (n = 17, 70.8%) contributed most to their (changing) point of view regarding the anatomic boundaries of the LNSs, followed by training (n = 14, 58.3%), education (n = 13, 54.2%), own experience (n = 11, 45.8%), and congresses, symposia, and other meetings (n = 10, 41.7%).

Discussion

This web-based survey confirmed the expected variability among Dutch esophageal surgeons with respect to both the extent of LND and the applied anatomic boundaries per LNS during MIE. Surgeons commonly presume that they have the same understanding of the stations and perform similar dissections. However, this study demonstrated this is not necessarily true. The observed variation brings challenges in current discussions regarding the benefit or harm of extended LNDs, and underscores the necessity of establishing consensus regarding the optimal routine LND for enhanced comparability in surgical patient care.

The Dutch surgeons indicated that they dissect at least 15 lymph nodes within the DUCA,5 but the specific lymph node stations dissected routinely remained unclear. This study showed nationwide consensus (≥ 85% of the participating Dutch esophageal surgeons) on the routine extent for distal esophageal adenocarcinoma encompassing the left and right paracardial, left gastric artery, celiac trunk, proximal splenic artery, common hepatic artery, subcarinal, upper mediastinal paraesophageal, middle mediastinal paraoesophageal, lower mediastinal paraesophageal, and pulmonary ligament LNSs (TIGER LNS 14, 15, 16, 17; proximal 18, 9, 10, 11, 12 and 13). This largely corresponds to the two-field or abdominal and mediastinal LND3 (except for the higher thoracic LNSs 6, 7 and 8 and the abdominal LNSs 17, distal, and 19.7 Surprisingly, the upper mediastinal paraoesophageal LNS (no. 10) was indicated to be routinely dissected by almost all the surgeons, with most of the surgeons performing Ivor-Lewis esophagectomy. Interestingly, discrepancies in the routine LND extent among the surgeons in the same hospital were noted for five participating hospitals. Although this seems remarkable, it confirms our hypothesis previously based on conversations and meetings with esophageal surgeons. Variability within a surgical team probably arises from the complexity of LND in esophagectomy and the lack of guidelines clearly delineating this. The observed heterogeneity in LND among surgeons is consistent with earlier findings10 and emphasizes the need for additional research toward consensus on the definition of complete LND.

The complexity of LND also appeared in the given answers when the surgeons were asked to provide anatomic boundaries for each LNS in six anatomic orientations. Even when indicating agreement with the TIGER definition, they provided anatomical boundaries for some stations that slightly deviated from the definition. Although consistency among surgeons was observed for certain stations in the description of anatomic boundaries per LNS (e.g., for 11 of 12 surgeons, the right border of the common hepatic artery LNS included the proper or common hepatic artery), the majority of LNSs showed (slight) variations (e.g., the inferior border of the common hepatic artery LNS included the pancreas [× 5], the common hepatic artery [× 2], the gastroduodenal artery [× 1], and the crus and vena cava [× 1]). The surgeons indicated that the anatomic boundaries they used were based primarily on literature and shared experiences among other surgeons in the field.11

The observed variations in the extent of the LND and the anatomic boundaries of each LNS hold significant clinical implications. These variations may explain the different outcomes in the existing literature. An extensive LND is currently advocated for enhancing tumor staging and potential long-term survival benefits,2,12,13 whereas there are concerns regarding the possible negative impact on morbidity and compromised quality of life due to excessive lymph node removal.14

Differences in anatomic boundaries hinder a comprehensive understanding of the precise components removed during the procedure. During dissection of the same LNSs, variations in boundaries may lead to diverse resected specimens and potentially result in distinct clinical outcomes for patients. It is anticipated that a clear understanding of anatomic boundaries will improve the comparability of LND among surgeons for research purposes. This includes comparison of performance assessments and their reliability among different reviewers8 as well as investigation of complications and survival benefits associated with LND, with the aim to generate uniform treatment strategies. Nevertheless, the observed variations in extent of LND and anatomic boundaries per LNS could serve as a foundation to establish consensus.

Our next step involves an international explorative study to investigate in-practice anatomic boundaries per LNS, incorporating assessments of short video clips of the LND with both the competency assessment tool8 and the surgical quality assessment tool.15 Research on this matter is indispensable because variation hinders comparing outcomes, and thereby valid assessments. Subsequently, we plan a Delphi study to achieve consensus on the LND extent and LNS anatomic boundaries.

This study was the first to investigate and describe the anatomic boundaries for the LND during MIE used by Dutch esophageal surgeons. However, it is important to acknowledge some limitations. First, describing the three-dimensional (3D) anatomic structure of LNSs in two-dimensional (2D) planes poses challenges. Although we selected this approach to provide a detailed and comprehensive overview, it may be worth exploring other visually representative methods that closely mirror clinical practice (e.g., drawing lines in a 3D anatomic model).

Second, described anatomic boundaries might differ from those of clinical practice because the dissection is based mostly on individual patient and tumor characteristics. For instance, the surgeon’s approach might be more radical, leading to removal of additional surrounding tissue beyond the currently defined boundaries. However, the surgeons were instructed to outline their anatomic boundaries for a “routine” LND as reference for the LND for distal adenocarcinoma, thereby eliminating the variation that tumor location could have in answers to the questions. Also, the vast majority of patients who undergo surgery for distal esophageal adenocarcinoma in the Netherlands present with stage cT3 and cN+ disease. Given the intermediate accuracy of current clinical tumor staging and the regular discordance between clinical and pathologic nodal staging, it may not be reliable enough to direct toward a certain extent of LND.16

Third, a complete overview of the current practices in the Netherlands was not established because three Dutch hospitals did not participate. Nevertheless, the insight of the 12 participating hospitals provided useful and valuable variations in LND extent and anatomic boundaries of LNSs.

Fourth, components of the web-based survey could have been described more clearly. Dividing the answer “patient and tumor factors” for the section discussing influences on the extent of the LND into “patient factors” and “tumor factors” might have yielded more comprehensive information. In addition, three LNSs (hepatoduodenal ligament, gastric artery, and splenic artery LNSs) received a more detailed description in the survey than in the TIGER definition. Because this could have caused some ambiguity in the description of the anatomic boundaries, these results were combined.

Conclusion

This study demonstrated the routine extent and the exploration of anatomic boundaries for the LND during MIE, as performed by esophageal surgeons in the Netherlands. The lack of consensus on both the extent of the LND and the anatomic boundaries presents a challenge in gaining a comprehensive understanding of the precise removed components during the LND of MIE and complicates investigation of the survival benefits and risks of morbidity associated with LND. Further investigation is necessary to establish consensus regarding the anatomic boundaries for each LNS, and thereby the definition of a “complete” LND. These efforts will enhance comparability of LND practices among surgeons and institutions worldwide, facilitating the implementation of standardized treatment strategies and aiding research.