Efficacy of prophylactic splenectomy for proximal advanced gastric cancer invading greater curvature

Initial cohort included 1309 patients who underwent total or subtotal gastrectomy for advanced proximal gastric carcinoma invading greater curvature at the Department of Gastroenterological Surgery, Toranomon Hospital, between January 1975 and December 2015. Among these, 108 patients were selected according to the following inclusion criteria: histopathologically proven adenocarcinoma, T2-4 in the upper 1/3 of the stomach invading the greater curvature, no gross nodal metastasis at the splenic hilum or along the splenic artery, no invasion to the spleen or the pancreas, and no evidence of distant metastases or peritoneal dissemination (both in macroscopic observation and peritoneal lavage). The definition of clinical lymph node-positive was enlarged lymph node (≧8 mm in the minor axis or ≧10 mm in the major axis) by contrast-enhanced abdominal computed tomography. The JCOG 0110 excluded Borrmann type 4 because of their aggressive malignant behavior. On the other hand, we included Borrmann type 4 because this type is thought to be associated with higher rate of peritoneal dissemination due to aggressive invasion of cancer cells to the gastric wall and it is a very important point that we should resect or preserve the spleen for these patients who had type 4 tumor. In the past report, a type 5 gastric cancer contained various types of tumors, from low-grade tumor to a high-grade tumor. However, the type 5 tumor generally has a large tumor diameter, high lymph node metastasis rate, and deeply invasive tumor. And also, these tumors has high rate of vascular invasion and high histological malignancy grade; therefore, Borrmann type 5 tumors were grouped with Borrmann types 3 and 4 [7, 8].

Then, the 108 patients were divided into two groups according to the concurrent splenectomy: the splenectomy group (n = 63) and spleen preservation group (n = 45), and clinical outcomes were compared.

D2 lymphadenectomy was routinely performed with total or subtotal gastrectomy for each patient. In the splenectomy group, lymphadenectomy along the splenic artery (No.11p and 11d lymph nodes) was completely performed with mobilizing the tail of the pancreas and only the spleen was excised while preserving the pancreatic parenchyma. In the spleen preservation group, No.11p and 11d lymph nodes were dissected without mobilizing the pancreatic tail and spleen. Lymph nodes at the splenic hilum (No.10 lymph nodes) were left untouched, but were dissected if judged easily removable in lean patients. In our hospital, treatment of advanced proximal gastric cancer was total gastrectomy with simultaneous splenectomy for adequate regional lymphadenectomy, especially splenic hilar lymph nodes. However, these indications were not established and so the surgeons finally determined whether it should be performed with or without splenectomy.

Clinical stage was determined according to the Union for International Cancer Control TNM Classification of Malignant Tumors, 7th edition on pathological examination [9]. A diagnosis of postoperative complications was made when we observed adverse events that corresponded to grade 2 or greater in the Clavien-Dindo classification [10].

Table 1 shows the clinicopathological characteristics of the patients. No significant intergroup difference was observed in clinical findings such as age, sex, tumor depth, and macroscopic or histological type. Operative duration and curability of surgery did not differ between the groups, whereas the amount of blood loss was significantly higher in the splenectomy group. As for the pathological findings, lymphatic and venous invasion, nodal metastasis, and pathological stage and postoperative chemotherapy did not differ significantly between the groups. In the spleen preservation group, No.10 dissection or sampling without splenectomy was performed in 11 patients (24%) and the median number of retrieved No.10 nodes in these patients was 3. There were no lymph node metastases. In the splenectomy group, four patients (6%) were histopathologically diagnosed with nodal involvement at the splenic hilum. Of these four patients, two patients developed recurrence in peritoneum and two patients have no recurrence. The mean survival time of these four patients was 62.0 months.

Table 2a compares the postoperative complication between the two groups. Splenectomy group showed higher rate of grade 2 or greater complications according to the Clavien-Dindo classification system (30.2 vs. 13.3%, p = 0.041). The rate of pancreas-related complications was particularly higher in the splenectomy group (17.4 vs. 0%, p = 0.003), while the rates of other types of complications were almost equivalent between the groups.

The median follow-up period was 135 months in the splenectomy group and 189 months in the spleen preservation group (p = 0.188) (Kaplan-Meier estimate). Table 2b shows recurrence rates and patterns, and RFS curves are shown in Fig. 1a. Five-year recurrence-free (RFS) rates did not differ significantly between the splenectomy and spleen preservation groups (60.2 vs. 67.3%, p = 0.609). As for the recurrence patterns, peritoneal dissemination occurred in 28.6% of patients in the splenectomy group and 22.2% in the spleen preservation group, with no significant intergroup difference. Isolated recurrence of nodal metastasis at the splenic hilum was observed in two (4.4%) patients in the spleen preservation group. Overall survival (OS) curves are shown in Fig. 1b. The 5-year OS was 73.6% in the spleen preservation group and 63.7% in the splenectomy group (p = 0.769). In a subset analysis, splenectomy was correlated with marginally better survival in patients with Borrmann type 1 or 2 gastric cancer (p = 0.072). No other subgroups showed potential advantage of splenectomy for OS (Fig. 2). In addition, of all 37 patients with recurrence, the rate of the peritoneal dissemination in patients with Borrmann type 1 or 2 (4 out of 8 patients; 50.0%) was lower than in patients with Borrmann type 3, 4, or 5 (24 out of 29 patients; 82.6%), but the difference between the two groups was not significant (p = 0.056).

For reference, 26 patients who were not included in this study due to therapeutic splenectomy for evident nodal involvement in No.10 or No.11p and 11d lymph nodes were additionally analyzed. Table 3 shows the clinicopathological characteristics of these patients stratified by tumor recurrence. Of the 26 patients, 17 (65%) patients developed recurrence in the peritoneum (n = 16) or para-aortic node (n = 1). The median survival time of these 17 patients was 22.9 months, while the remaining 9 patients were alive without recurrence with a median survival time of 140.9 months. No evidence difference was observed in clinicopathological findings between the patients who developed recurrence and those who did not.