Pancreas-Preserving Total Gastrectomy for Gastric Cancer

Background  Pancreas-preserving total gastrectomy for gastric cancer has been proposed to remove lymph nodes along the upper border of the pancreas without performing a distal pancreatic resection. However, the original technique includes the ligation of the splenic artery at its origin and thus carries the risk of pancreatic necrosis.

Hypothesis  A technique of pancreas-preserving total gastrectomy that includes ligation of the splenic artery approximately 5 cm distally from the root may reduce the risk of postoperative pancreatic necrosis.

Design  Case series.

Setting  Both primary and referral hospital care.

Patients  Hospital records of 228 consecutive patients who, according to a personal technique, underwent D3 pancreas-preserving total gastrectomy for gastric cancer from 1981 to 1997 were reviewed.

Main Outcome Measures  Surgical complications, postoperative deaths, and survival.

Results  Hospital morbidity and mortality were 33.3% and 3.9%, respectively. No patients experienced pancreatic necrosis. The 5-year survival rate after curative resection was 53.6%: 96.9% for stage IA, 76.3% for stage IB, 63.0% for stage II, 35.6% for stage IIIA, 27.0% for stage IIIB, and 20.3% for stage IV (N3-positive patients) disease.

Conclusion  Results of the present study show the efficacy of this method of radical resection for gastric cancer as demostrated by the low incidence of postoperative complications and high survival rates.

PANCREAS-PRESERVING total gastrectomy for gastric cancer has been proposed to remove lymph nodes along the upper border of the pancreas without performing a distal pancreatic resection.¹ The surgical technique includes the ligation of the splenic artery at its origin, which is effective in achieving a complete D2 nodal dissection and in reducing the incidence of complications related to pancreatic resection, but carries the risk of pancreatic necrosis.¹

We have developed a technique of pancreas-preserving total gastrectomy for gastric cancer² that reduces the risk of pancreatic necrosis.

The aim of the present study is to report early and long-term results of our technique of pancreas-preserving total gastrectomy for gastric cancer.

Between 1981 and 1997, 279 patients underwent total gastrectomy at our department for the primary surgical treatment of gastric adenocarcinoma.

This population accounts for 37.6% (279 of 742) of all patients with primary gastric cancer observed in our unit during the same period.

Patients undergoing pancreatic resection for macroscopic infiltration of the gland after both radical (n = 11) and palliative (n = 7) surgery and those undergoing simple D1 palliative gastrectomy (n = 33) were excluded from the analysis. The remaining 228 patients underwent D3 pancreas-preserving total gastrectomy and constitute the population of the present study.

The surgical procedure involves dissection of the entire greater omentum, the superior leaf of the mesocolon, and the serosa of the pancreatic surface; node dissection is then performed in the infraduodenal and supraduodenal areas, along the retropancreatic region (node region 13 according to the Japanese Research Society Committee on Histological Classification of Gastric Cancer [JRSGC]³), the hepatic pedicle (node region 12), the mesenteric root (node region 14), and along the common hepatic (node region 8), and celiac (node region 9) arteries. The left gastric artery (node region 7) is ligated at its origin, and node dissection (node region 11) is extended along the proximal third of the splenic artery, which is ligated distally approximately 5 cm from its origin. The spleen and the distal pancreas are then mobilized and retracted. The tail of the pancreas is carefully exposed, and the splenic vein and caudal pancreatic arteries are ligated and divided. Finally, the upper border of the pancreas is exposed, and the spleen (node region 10) and the middle and distal third of the splenic artery with the surrounding fatty connective tissue and nodes (node region 11) are removed en bloc with the stomach, gastric omentum, and perigastric nodes (node regions 1-6). The pancreatic parenchyma and splenic vein are preserved (Figure 1). The intestinal continuity was restored by a Roux-en-Y esophagojejunostomy in all cases.

Demographic data, tumor location and gross appearance, histological type according to the Lauren classification,⁴ regional extent of nodal involvement according to the JRSGC, and hospital morbidity and mortality were recorded. The staging of the disease was performed according to the TNM classification.

Comparison of the rate of occurrence of events was performed using the Fisher exact test. Survival was calculated on the basis of the actuarial life-table method.⁵

Patients' status was determined by follow-up examination or telephone contact. A total of 160 patients (70.1%) were followed up for at least 5 years after hospital discharge.

Patients' characteristics are shown in Table 1. Overall, the incidence of postoperative complications and postoperative mortality after pancreas-preserving total gastrectomy (Table 2) was lower than that observed among patients who underwent total gastrectomy associated with pancreatic resection: 76/228 (33.3%) vs 11/18 (61.1%) (P = .02) and 9/228 (3.9%) vs 4/18 (22.2%) (P = .01).

No patients developed postoperative pancreatic necrosis after pancreas-preserving total gastrectomy, and only 2 patients (0.9%) experienced pancreatic fistula. By contrast, the incidence of pancreatic fistula was higher after total gastrectomy associated with pancreatic resection: 2/18 (11.1%; P = .02), as was the incidence of esophagojejunostomy leak 5/18 (27.8%) vs 22/228 (9.6%) (P = .03).

The mean ± SD number of dissected nodes per operative specimen was 43.4 ± 22.3; the incidence of nodal metastasis was 60.9% (139/228): 37.7% (86/228) for N1, 16.2% (38/228) for N2, and 6.6% (15/228) for N3 involvement. Table 3 shows the mean number of dissected nodes and the percentage of metastatic involvement to individual node stations.

Five-year survival rate after curative resection (228 patients) was 53.6%: 96.9% for stage IA, 76.3% for stage IB, 63.0% for stage II, 35.6% for stage IIIA, 27.0% for stage IIIB, and 20.3% for stage IV (N3-positive patients) disease (Figure 2).

Despite the unfavorable results of recent studies,^7,8 the extended D2 lymphadenectomy is commonly accepted as the treatment of choice of curable gastric cancer not only by Japanese authors, but also by the majority of Western authors.^9- 13

According to Japanese nomenclature, a complete D2 dissection for tumor located at the upper and middle third of the stomach necessarily requires the dissection of nodes along the splenic artery (JRSGC node region 11) and at the splenic hilum (JRSGC node region 10). For that reason, many authors have routinely associated the resection of the distal portion of the pancreas with total gastrectomy for the radical treatment of upper and middle curable gastric tumors.^14,15 However, complications related to the pancreatic resection were frequently experienced, such as pancreas juice leakage, acute pancreatitis, left subphrenic abscess, and postoperative diabetes.¹⁵ Even in recent prospective studies on lymphadenectomy for gastric cancer, pancreatic resection had a significant adverse effect on both morbidity and mortality. The Medical Research Council (MRC) trial⁸ reported significantly higher postoperative morbidity and mortality rates after pancreas removal: 56% vs 28% (P<.001) and 16% vs 7% (P = .01), respectively. The Dutch Gastric Cancer Trial⁷ found by univariate and multivariate analyses of postoperative risk factors¹⁶ that distal pancreatectomy had the largest relative risk for postoperative complications both in the univariate (5.04) and multivariate (3.34) model. As shown by the present study, the distal pancreatectomy, performed "de necessitate" only in patients with direct infiltration of the gland, also had negative effects on postoperative outcome. Moreover, even in the absence of overt pancreatic fistula, the subclinical leakage of pancreatic juice after pancreatic resection would accumulate in close proximity to the proximal reconstructive anastomosis, thus affecting the healing process. In this context, Japanese authors have lately stressed that extended lymphadenectomy for gastric cancer should, whenever possible, be undertaken without distal pancreatectomy—the pancreas-preserving technique, with ligation of the splenic artery at its origin, has been proposed.¹ The original technique, however, carries the risk of pancreatic necrosis; in fact, the dorsal pancreatic artery usually arises from the proximal third of the splenic artery^17,18 and joins the posterosuperior pancreaticoduodenal artery (so-called Kirk arcade) after the emergence of the transverse pancreatic artery. In the absence of the Kirk arcade (about 40% of cases), the dorsal pancreatic artery is the sole blood supply to the left pancreas (Figure 3). Thus, ligation of the origin of the splenic artery exposes 2 of 10 patients to the risk of pancreatic necrosis.

For that reason, we preserve the proximal third of the splenic artery in all cases of pancreas-preserving total gastrectomy by ligating the splenic artery approximately 5 cm distally from the root to preserve the blood supply to the left pancreas through the dorsal pancreatic artery. Proximal splenic nodes are removed en bloc with those of the left gastric artery, and distal splenic nodes are removed together with the splenic artery, the surrounding fatty tissue, and the spleen. The pancreatic parenchyma and the splenic vein are preserved.

Results of the present study show the efficacy of this method of radical resection for curable gastric cancer as demonstrated by the low incidence of postoperative complications and high survival rates.

Reprints: Fabio Pacelli, MD, Istituto di Clinica Chirurgica, Universitá Cattolica del Sacro Cuore, Largo A. Gemelli n. 8, 00168 Roma, Italy (e-mail: fpacelli@hotmail.com).