Clinical implications of lymphadenectomy for invasive ductal carcinoma of the body or tail of the pancreas

Abstract Aim The appropriate extent of lymphadenectomy for pancreatic cancer of the body/tail has not been standardized worldwide. The present study evaluated the optimal extent of harvesting lymph nodes. Methods Patients who underwent distal pancreatectomy for invasive ductal carcinoma of the pancreas between 2007 and 2018 were retrospectively reviewed. Patients were subclassified into three groups depending on the tumor location: pancreatic body (Pb), proximal pancreatic tail (Ptp), and distal pancreatic tail (Ptd). The pancreatic tail was further divided into even sections of Ptp and Ptd. Patterns of lymph node metastasis and the impact of lymph node metastasis on the prognosis were examined. Results A total of 120 patients were evaluated. Fifty‐eight patients had a tumor in the Pb, 38 in the Ptp, and 24 in the Ptd. No patients with a Ptd tumor had metastasis beyond the peripancreatic and splenic hilar lymph nodes (LN‐PSH). All patients with metastasis to the lymph nodes along the common hepatic artery (LN‐CHA) or along the left lateral superior mesenteric artery (LN‐SMA) also had metastasis to the LN‐PSH. Recurrence after surgery occurred significantly earlier in this population. In a multivariate analysis, metastasis to the LN‐CHA or LN‐SMA (hazard ratio [HR] 3.3; P = .04) was an independent risk factor for overall survival. Furthermore, high levels of preoperative serum CA19‐9 (HR 10.9; P = .013) were a predictive factor for metastasis to the LN‐CHA or LN‐SMA. Conclusions Metastasis to the LN‐CHA or LN‐SMA was rare but a significant prognostic factor in patients with pancreatic body/tail cancer.


| INTRODUC TI ON
Lymph node status is well known to be a significant prognostic marker in patients with pancreatic cancer. [1][2][3][4][5] Pancreatectomy with lymphadenectomy has been the standard procedure for treating pancreatic cancer. [6][7] However, the optimal extent of lymphadenectomy has been controversial. Based on previous randomized controlled trials, extended lymphadenectomy with pancreatoduodenectomy has not been recommended for pancreatic head cancer. [8][9][10][11][12][13][14] Particularly for patients with adenocarcinoma in the body or tail of the pancreas, few studies have focused on the influence of lymph node involvement on the prognosis.
The recommended extent of lymph node dissection during distal pancreatectomy (DP) for pancreatic cancer differs somewhat between the seventh edition of the rules of the Japan Pancreas Society (JPS) 15 and the consensus statement by the International Study Group on Pancreatic Surgery (ISGPS). 7 The JPS recommends harvesting lymph nodes along the common hepatic artery and the celiac axis for both pancreatic body and tail cancers. In contrast, the ISPGS recommends that the lymph nodes around the celiac axis be resected, particularly when the tumor is close to the celiac axis in the body of the pancreas, and the lymph nodes along the common hepatic artery not to be dissected for pancreatic body or tail cancers, as resection of these lymph nodes has been considered to constitute extended lymphadenectomy. 7 Clarifying the incidence of metastasis in a specific regional lymph node station and the impact of lymph node metastasis on the prognosis has proven useful for understanding the patterns of tumor spread and examining the extent of lymph node dissection. However, to our knowledge, few studies have investigated the rates of lymph node metastasis, especially for distal pancreatic cancer. [16][17] The present study evaluated the patterns of lymph node metastasis in patients with pancreatic cancer in the body or tail and proved the validity of the current extent of lymphadenectomy during DP.

| Patients
From January 2007 to December 2018, 305 consecutive patients underwent DP, including 17 who underwent DP with celiac axis resection (DP-CAR), in Shizuoka Cancer Center, Japan. Among them, 135 patients who were histologically proven to have invasive ductal carcinoma of the pancreatic body or tail were included in this study. Of these, patients who underwent R2 resection (n = 1), those who underwent DP as total remnant pancreatectomy (n = 9), and those with double cancers (n = 5) were excluded from this study.
Ultimately, 120 patients were included as subjects in this study. The clinical data of these patients were obtained from a prospectively collected database.
This study was approved by the Institutional Review Board of the Shizuoka Cancer Center (approval number: J2020-164-2020-1-3).

| Surgical procedures
All surgical procedures were performed with an open approach.
No laparoscopic surgery was conducted during the study period.
Peritoneal lavage cytology and sampling of the para-aortic lymph nodes were performed after laparotomy. If unresectable factors were found, the planned procedure was abandoned. The surgical procedures performed for DP and DP-CAR were described previously. 19 Indications for DP-CAR in our institution included (a) the celiac axis was involved, whereas the aorta, superior mesenteric artery, and gastroduodenal artery remained free from the tumor; or (b) preserving the splenic artery root was technically or oncologically difficult. 19 To achieve complete lymph node dissection around the splenic artery and the splenic hilum, the spleen was routinely resected in both procedures. The extent of lymph node dissection was either equal to or greater than that recommended by the ISGPS. 7 In detail, the lymph nodes along the common he-  Figure 1A. The intraoperative histological evaluation of the stump of the pancreas was always performed by pathologists to ensure that the surgical margin remained negative for cancer cells.

| Histological evaluation and numbering of lymph nodes
A histological assessment was carried out by at least two special-

| Subclassification of the tumor location
A schematic illustration of the subclassification of the tumor location is also described in Figure 1B. Tumors located at the tail of the pancreas were classified into two groups: proximal pancreatic tail (Ptp) and distal pancreatic tail (Ptd). The boundary between Ptp and Ptd was defined as the line that equally divided the left border of the abdominal aorta and the end of the pancreatic tail. If the tumor was located in more than two areas, classification was performed according to the location of the center of the tumor.
Preoperative computed tomography (CT) images were used for this analysis.

| Statistical analyses
Categorical variables were compared using the chi-square test or Fisher's exact test, as appropriate. Continuous variables were compared using the Mann-Whitney U-test. The survival was analyzed using Kaplan-Meier curves and the log-rank test. The optimum cutoff values of each continuous parameter for the overall survival (OS) and predicting metastasis to the LN-CHA or LN-SMA were determined using the minimum P values calculated using the logrank test. Especially, as to tumor marker, cutoff values were shown to be 15.0 ng/mL for CEA (P = .0029) and 400 U/mL for CA19-9 (P = .00047) ( Figure S1A, B). Hazard ratios were estimated by univariate and multivariate survival analyses using the Cox regression model. Variables with P < .05 using the univariate log-rank test were further explored in the multivariate setting. Differences were considered statistically significant at P < .05. All analyses were performed using the SPSS software program, v. 25.0 (IBM, Armonk, NY, USA).

| RE SULTS
Patients' demographics and operative characteristics are summarized in Table 1. Fifty-eight patients had tumors in the Pb, 38 in the Ptp, and 24 in the Ptd. Patients with tumors in the Ptd were younger than those with tumors in the Pb (P < .05). All patients with tumors in the Ptd had resectable lesions and underwent DP. DP-CAR was performed in 17 patients with tumors in the Pb or Ptp. There were no other significant differences among these three groups.
No significant difference was shown in the OS and the diseasefree survival (DFS) for patients in the Pb, Ptp, and Ptd groups ( Figure   S2). Pathologic characteristics are also shown in Table 1. Nodal involvement was observed in 64 (53%) patients. The median number of examined regional lymph nodes was 16. R1 resection was

| Prognostic factors for OS and DFS
Multivariate analyses revealed that lymph node metastasis to the LN-CHA or LN-SMA, serosal invasion, portal venous system invasion, and a lack of adjuvant chemotherapy were risk factors for OS (Table 3).
Similarly, a high level of serum CA19-9, large tumor, lymph node metastasis, portal venous system invasion, and no adjuvant chemotherapy were shown to be risk factors for DFS by multivariate analyses (Table   S1).

| Predictive factors for metastasis to the LN-CHA or LN-SMA
Univariate analysis showed that high levels of preoperative serum CA19-9 were a predictive factor for lymph node metastasis to the LN-CHA or LN-SMA ( Table 4).

| DISCUSS ION
LN-CHA and LN-SMA are considered appropriate for dissection, regardless of tumor location, according to the classification of pancreatic carcinoma in Japan. 15 However, few studies have described the metastasis rate of those stations and the effect of dissection of those lymph nodes, especially for pancreatic tail cancer. [16][17] This study describes the patterns of lymph node metastasis for patients with pancreatic body/tail cancer who underwent DP.
Specifically, it revealed that LN-CHA and LN-SMA metastasis was rare but still a significant prognostic factor in patients with pancreatic body/tail cancer. According to the mapping of the meta- cancer without lymph node metastasis as a preoperative diagnosis, a low extent of lymphadenectomy has been recommended. 23 For breast cancer without clinically lymph node metastasis, as confirmed by a sentinel node biopsy, axillary lymph node dissection has been omitted. 24 These treatments have been supported by an accurate diagnosis for tumor staging. Regarding pancreatic cancer, in general, the concept of the sentinel lymph node hypothesis has not been adopted, and a preoperative diagnosis for staging is sometimes difficult to make, compared to cases of stomach or breast cancer. Further advances in imaging studies along with the accumulation of evidence will help resolve this issue.
The pancreatic resection line during DP is determined by con-   surgery. This might also be associated with our institutional policy, where the LN-SMA is usually dissected only in cases with Pb tumors. Thus, given these potential biases, we recognize that we cannot draw any absolute conclusions from these data. To confirm the current results, a further multicenter study including data from high-volume centers should be conducted. Nevertheless, we believe that the results of the study will help refine classical procedures.
In conclusion, metastasis to the LN-CHA or LN-SMA was rare but still a significant prognostic factor in patients with pancreatic body/tail cancer.