The Impact of D2 Versus D1 Lymphadenectomy in Siewert II Gastroesophageal Junction (GEJ) Cancer

The study identified 155 adult patients with a Siewert II diagnosis during the study period from 1 January 2012 through 30 June 2022. Of the 155 patients, 129 (83%) were male, and 26 (17%) were female. Ages at the date of diagnosis ranged from 28 to 84 years (mean, 64.2 ± 11.0 years). Of the 155 patients, 115 (74%) underwent a D1 LAD, and 40 (26%) underwent a D2 LAD. The baseline patient characteristics (age, race/ethnicity, sex, BMI at diagnosis, CCI score, smoking history, and history of alcohol abuse) did not differ significantly between the Siewert II patients who underwent a D1 LAD and those who underwent a D2 LAD (Table 1).

Both the patients undergoing a D1 LAD and those undergoing a D2 LAD most often had clinical stage III disease at the time of their diagnosis (Table 1) and pathologic stage III disease (Table 2) after surgical resection. The patients with a D2 LAD had more than 15 lymph nodes harvested more frequently than the patients with D1 LAD (83% vs 48%; p < 0.001), but the mean number of positive nodes did not differ significantly (2.8 ± 5.2 D2 vs 2.1 ± 4.2 D1; p = 0.355 (Table 2; Fig. 1).

In a multivariable logistic regression model, including patients treated with the transhiatal approach, no significant difference was demonstrated between D2 and D1 for obtaining at least one positive lymph node (adjusted odds ratio [aOR], 1.58; 95 % confidence interval [CI], 0.73–3.44; p = 0.246; Table 3). When the study excluded the nine patients who underwent the transhiatal approach, similar results were demonstrated, with no significant difference between D2 and D1 for obtaining at least one positive lymph node (aOR, 1.26; 95% CI, 0.55–2.93; p = 0.586; Table S1), The patients with a D2 LAD had a median operative time approximately 2 h longer than those who had a D1 LAD (362 vs 244 min; p < 0.001). No significant difference was observed between the D2 and D1 LAD patients in the proportion of patients upstaged (18% vs 10%, respectively; p = 0.26; Table 2).

Kaplan-Meier analysis did not show a significant difference in 5-year overall survival probabilities between the D2 and D1 LAD patients (0.60 D2 vs 0.44 D1; p = 0.11; Fig. 2). In the Kaplan-Meier analysis, 38 D1 patients (33%) versus 29 D2 patients (73%) were censored without 5 full years of follow-up evaluation, and 55 D1 patients (48%) versus 10 D2 patients (25%) experienced the event of interest (death) during the 5 years.

In a multivariable Cox regression model, with adjustment for clinical stage, anastomotic leak (yes/no), neoadjuvant therapy (chemotherapy and/or radiation [yes/no]), age, and CCI score, no significant difference in overall survival was observed for the patients undergoing a D2 versus D1 LAD (adjusted hazard ratio [aHR], 0.52; 95% CI, 0.25–1.00; p = 0.067). A statistically significant overall survival difference in clinical staging was observed (p = 0.016; Table 4). Regarding the impact of extensive lymphadenectomy on locoregional, D1 and D2 LAD did not differ significantly in terms of 3-year recurrence-free survival (p = 0.11; Fig. S1).

The proportional hazards assumption was not violated in this multivariable Cox regression model. A log-log plot graphically demonstrated no significant departures from the proportional hazards assumption for surgery type (D1 vs D2 LAD; Fig. 3). Additionally, plots of scaled Schoenfeld residuals and a global test of the Schoenfeld residuals for a slope of 0 (p = 0.28; chi-square, 9.78; df, 8) supported the proportional hazards assumption (Fig. S2).

Current literature regarding the extent of LAD in GEJ cancer remains controversial, with the majority of current data specific to gastric cancer or GEJ cancer with curative-intent gastrectomy.¹⁰ This is the first study to detail the outcomes of Siewert II GEJ tumors treated with D1 versus D2 LAD via a minimally invasive trans-thoracic, Ivor-Lewis esophagectomy approach.

The recent AJCC definition of Siewert II tumors deemed as esophageal tumors has important surgical and therapy implications that distinguish them from Siewert III or gastric tumors.⁷ This is an important distinction to make because the benefits of D2 LAD leading to decreased locoregional recurrence and disease-specific survival as seen in the largest prospective, randomized trial were limited to patients with GEJ tumors who underwent gastrectomy and not an esophagectomy with curative intent.⁹

In terms of obtaining an adequate LAD, Ivor-Lewis esophagectomy has been shown to adequately obtain thoracic and abdominal LN with feasibility to perform a D2 dissection at the discretion of the surgeon.^12,13 In addition to the differences in surgical management seen in our study compared with prior literature, major inconstancies exist in the optimal neoadjuvant and/or adjuvant therapies for Siewert II tumors. In our study, the Siewert II patients undergoing D1 LAD were more likely to have undergone neoadjuvant and concurrent chemoradiation (75 %) than those who underwent D2 LAD (33 %). Additionally, the patients undergoing D2 LAD were more likely to receive adjuvant chemotherapy (58 %) than those who underwent D1 LAD (30 %).

Further adding to the complexity of Siewert II tumor management, our patients were left to the discretion of the multidisciplinary tumor board review whether to use chemoradiotherapy for esophageal cancer as in the surgery study (CROSS) or to use perioperative chemotherapy with fluorouracil plus leucovorin, oxaliplatin, and docetaxel versus fluorouracil or capecitabine plus cisplatin and epirubicin for locally advanced, resectable gastric or gastro-oesophageal junction adenocarcinoma as in both the FLOT4 trial and the Medical Research Council Adjuvant Gastric Infusional Chemotherapy (MAGIC) trial.^14‐16 We found that the patients who underwent D1 dissection were more likely to receive neoadjuvant chemoradiation and thus follow the CROSS study regimen, whereas the patients who underwent D2 were more likely to receive treatment similar to gastric cancer management and to be treated with perioperative chemotherapy instead of chemoradiotherapy, in accordance with the FLOT4 trial, with the addition of staging diagnostic laparoscopy.¹⁶

Although it is difficult to say that the addition of neoadjuvant radiotherapy allows for less extensive LAD via D1 or that D2 may increase nodal yield to necessitate adjuvant chemotherapy, these differences in surgical management are important to note because current prospective randomized studies have not demonstrated a standardized approach, which may be ultimately influenced by differences in current oncologic therapy.

Although D2 LAD has the theoretical advantage of a more extensive LN harvest to influence pathologic staging, its clinical significance in terms of improving overall survival is controversial.^8,17 As expected, the Siewert II patients who underwent D1 LAD had a smaller nodal harvest than the patients who underwent D2 LAD, with 48 % of the D1 LAD procedures yielding more than 15 LNs versus 83 % of the D2 LAD procedures (p < 0.001). However, according to the AJCC, accurate staging of GEJ cancer is defined by the number of positive LNs in addition to R0 resection.^1,7 When LN positivity was evaluated between those receiving D1 LAD and those treated with D2 LAD, the two groups did not differ significantly (Table 2; Fig. 1). Additionally, D2 LAD had minimal to no impact on the Siewert II patients in terms of changing management because the patients who were clinically stage III or IVa remained in their respective stages. Stage migration occurred in 17.5% (n = 7) of the Siewert II patients undergoing D2, with six patients migrating from stage III to stage IV, and one patient migrating from stage II to stage IV.

The benefits seen with a larger nodal harvest for pathologic staging in prior studies has been largely limited to gastric cancer and has not been well-studied for GEJ cancer with an intent-to-cure esophagectomy. Although extended LAD with D2 may yield a larger LN harvest, its ability to influence stage migration and ultimately guide therapy is questionable because extended LAD did not lead to the capture of more malignant cells in Siewert II tumors.

Although multiple studies exist regarding the increasing Western popularity of D2 LAD in the management of GEJ cancer, the benefits have been limited to patients undergoing intent-to cure gastrectomies.^8,9,18 Additionally, in multiple prospective, randomized trials, patients undergoing D2 LAD had greater perioperative morbidity and mortality with no change in overall survival for patients with GEJ cancer.^9,18

Although D2 LAD has gained more interest recently, it remains an additional technically demanding step, ultimately increasing the patient’s time under anesthesia.⁸ Evaluation of operative time showed that Siewert patients with a D2 LAD had a median operative time approximately 2 h longer than patients with D1 LAD. Longer operative time and time under anesthesia for esophagectomies have been known to lead to postoperative complications such has increased incidences of pneumonia, length of stay, unplanned re-intubation, and mortality.¹⁹ These increased short-term complications with D2 LAD may be outweighed by the improved long-term benefits of decreased locoregional recurrence and gastric cancer-related deaths, although this remains controversial.⁹ Several studies have demonstrated long-term benefit from D2 LAD performed for gastric cancer, but no study has demonstrated improved overall survival.^8,9,18

In our study, after adjustment for patient and clinical characteristics, overall survival did not differ significantly between the Siewert II patients undergoing D1 LAD and those treated with D2 LAD. Similar to other studies for gastric cancer, D2 LAD did not confer a significant difference in survival while adding additional 2 h of operating time when performed as an intent-to-cure esophagectomy. As expected, increasing clinical stage at diagnosis rather than performance of a D1 or D2 LAD demonstrated an overall survival difference that was statistically significant in Cox multivariable regression. Although improvements in surgical technique and medical therapy have made D2 LAD more feasible for the treatment of gastric cancer, its routine use in Siewert II esophagectomies remains controversial, particularly while similar concerns of lack of overall survival benefit still exist.

This study had several limitations. The study was retrospective in nature and limited to a single tertiary center. However, this is the largest of all the regionalized referral centers for esophagectomy in the entire region, routinely performing more than 40 minimally invasive esophagectomies (MIE) per year. Specific chemoradiation regimens were difficult to characterize accurately as radiotherapy, which is part of our center’s treatment algorithm. The difficulty was because outsourcing of radiation oncology referrals was dependent on the geographic region in which the patient resided.

Additionally, granular data regarding dosing, timing, and number of chemotherapy cycles were difficult to characterize, which may have led to inaccurate characterization of adjuvant therapy. In the modeling for overall survival, the diagnosis in the D2 LAD group occurred more often between 2018 and 2022 and therefore were more often censored in the model without a full 5 years of follow-up time available. Higher levels of censorship may have led to wider confidence intervals, limiting our findings.

The convergence of the log-log plot (Fig. 3) demonstrates that the relative hazards of the D1 and D2 LADs approached each other as time progressed. Similarly, individual scaled Schoenfeld residual tests (Fig. S2) indicate potential time-varying effects for D1 versus D2 (p = 0.07) and clinical stage (p = 0.06). Further exploration is suggested to examine whether this is due to a decreasing effect over time or to unbalanced censorship.

It also is important to note the operative time discrepancy between the D1 and D2 groups. TheD2 LAD requires more meticulous dissection, and our study demonstrated an operative time difference of approximately 2 h. This likely was due to the addition of a separate surgical team to perform the D2 dissection according to the protocol of our institution, in which different surgical trays and operative room turnover may account for this additional time difference, perhaps skewing the results regarding the operative disadvantage of performing a D2 versus D1 dissection. To further mitigate this discrepancy, we plan to convert all laparoscopic trays to one centralized universal tray set with the amount of the abdominal portion of the operation by different surgical teams consolidated from start to finish, with hopes of decreasing operative times and improving efficiency while maintaining similar outcomes.

In conclusion, despite the associated high mortality associated with GEJ cancer, few consensus guidelines exist regarding the optimal preoperative and operative management. Although D2 LAD may theoretically lead a larger lymph node harvest and may be warranted in select cases, its routine use for Siewert II tumors may not necessarily be mandatory because it did not increase the number of positive nodes and had no significant impact on survival.