Shimoyama S, Mafune K, Kaminishi M. Indications for a Pylorus-Preserving Gastrectomy for Gastric Cancer With Proper Muscle Invasion. Arch Surg. 2003;138(11):1235-1239. doi:10.1001/archsurg.138.11.1235
Our previous demonstrations of the feasibility of a pylorus-preserving gastrectomy with a wider-scope lymphadenectomy for early gastric cancer (GC) have raised the possibility of applying pylorus-preserving gastrectomy to a broader segment of GC patients, such as those having GC with invasion extending into the proper muscle layer (GCpm).
Tertiary care center.
This study comprised 107 patients with solitary primary GCpm located in the middle or lower third of the stomach.
All patients underwent gastrectomy, using at least the former D2 dissection of the American Joint Committee on Cancer.
Main Outcome Measures
Patterns and sites of nodal involvement were analyzed in relationship to other clinicopathologic characteristics, including preoperative and intraoperative evaluations of tumor depth (cT), nodal involvement (cN), and serosal involvement. The conditions required were serosal negativity and measurements less than or equal to cT2 cN0 with histologic proof of node negativity, or if positive, the positive nodes had to be confined to the first or selective second tier. The selective second-tier nodes were defined as those along the left gastric, common hepatic, and celiac arteries.
The factors favoring node negativity were serosal negativity, cN0, or both. For tumors that were serosal-negative and less than or equal to cT2 cN0 to meet the above conditions, a tumor in the middle third of the stomach had to be less than or equal to 2 cm in the maximum diameter, and that in the lower third could be any size. No patients with serosal negativity and cT2 cN0 GCpm less than or equal to 2 cm in diameter died of cancer or had positive nodes beyond the selective second tier.
Considering the required distance between the pyloric ring and distal margin of the tumor to avoid a positive resection margin, pylorus-preserving gastrectomy with a selective second-tier node dissection is optimal for tumors that are serosal negative, less than or equal to cT2 cN0, and less than or equal to 2 cm in maximum diameter when located in the middle or lower third of the stomach.
CURRENT SURGICAL trends toward the treatment of gastric cancer (GC) recognize that the quality of postoperative life may be improved with less invasive surgery, including endoscopic mucosal resection, local resection with1 or without2- 4 adjacent lymphadenectomy, and gastrectomy with a modified D1 lymphadenectomy,5 especially for carefully selected patients with mucosal GC. In this context, we6 and other investigators7- 9 have introduced the pylorus-preserving gastrectomy (PPG), a less invasive surgical option, for the treatment of early GC. By preserving pyloric function, the PPG has been shown to have many advantages over distal gastrectomy with conventional Billroth I reconstruction in terms of the prevention of postgastrectomy disorders, such as dumping syndrome and alkaline regurgitation.6- 9 Most of the reported PPG procedures by other investigators, however, have not included suprapyloric lymph node dissection to preserve the pyloric branch of the vagal nerve and the right gastric artery; thus, indications for such a type of PPG are quite limited.7- 9 We have recently demonstrated that PPG with a wider-scope (perigastric and regional) lymphadenectomy, including the dissection of suprapyloric lymph nodes, is feasible, safe, and conducive to a better postoperative quality of life.6 We have rationally performed PPG on patients with submucosal GC, and it is possible to apply PPG for an even a broader segment of patients. Recent animal experiments have indicated no postoperative functional differences between PPG with and without dissection of the pyloric branch of the vagal nerve, supporting the idea that PPG with perigastric and regional lymphadenectomy does not compromise its postoperative advantages.10
For node-positive and/or GC with invasion extending into the proper muscle layer (GCpm) or farther, an extended lymphadenectomy has been the advocated treatment in Japan as well as in several Western institutes.11- 16 Since GCpm rarely exhibits peritoneal relapse,17,18 attempts to reduce the risks of regional node recurrence provide increased chances for successful cures in patients with GCpm. The ability of such risk reduction by an extended lymphadenectomy is supported by a recent randomized study in which a D2 dissection led to a reduction in local recurrence frequency compared with a D1 dissection.19 Therefore, we have raised the hypothesis that some segments of GCpm patients can also be treated by PPG when lymphadenectomy extends to the second tier. If this is true, these selected patients have a chance to avoid postgastrectomy disorders, a goal that can hardly be achieved by conventional Billroth I reconstruction. We demonstrate here that PPG with a priority second-tier node dissection (a modified D2 dissection) is sufficient for curative intent for serosal-negative, node-negative GCpm less than or equal to 2 cm in maximum diameter.
One hundred fifty patients with primary solitary GCpm underwent curative resection (Union Internationale Contre le Cancer R0) between 1976 and 2000 in the department of gastrointestinal surgery, University of Tokyo (Tokyo, Japan). Exclusions were made for those who had GC located in the upper third or the whole stomach (28 patients) and for those undergoing a D1 lymphadenectomy (15 patients); thus, the remaining 107 patients exhibiting middle- or lower-third GCpm formed the study population. The clinicopathologic characteristics of these patients included age, sex, predominant tumor location, gross form, maximum tumor diameter, preoperative and intraoperative findings concerning tumor depth (T) and nodal involvement (N), serosal findings at laparotomy, lymphatic and vascular invasion, and Lauren histologic type. The maximum tumor diameter was evaluated histologically. Gross form and degree of N were expressed according to the Japanese Classification of Gastric Carcinoma (JRSGC).20 Gross form was divided into 3 types: localized (Borrmann types 1 and 2), invasive (Bormann types 3 and 4), and advanced. The advanced type was histologically proved to be an advanced GC but macroscopically resembled an early GC.21
The lymph node stations that correspond with specific lymph node tiers have recently been revised by the JRSGC.20 In addition to the second-tier nodes defined by the American Joint Committee on Cancer (AJCC) in 1987,22 the new JRSGC classification has recognized lymph nodes along the hepatoduodenal ligament (station 12a) and at the root of the superior mesenteric vein (station 14v) as the second-tier nodes. The grouping of nodal tiers can be modified slightly depending on the tumor location (Table 1). Since these 2 stations belonged to the third tier in the former AJCC classification (and also in the former JRSGC classification), they are considered more upstream nodes for both middle- and lower-third GC. In the new JRSGC classification, therefore, a "complete" D2 dissection has come to require clearance of these newly added stations. Under these circumstances, in the present study, the dissection of first-tier nodes as well as that of preferential lymph nodes along the left gastric (station 7), common hepatic (station 8), and celiac (station 9) arteries can be defined as a modified D2 dissection, and these 3 stations are defined as selective second-tier stations.
Following the new JRSGC definitions of lymph node stations and tiers, the extent of N was expressed as N0 if involvement was not observed, N1 if it was confined to the first-tier nodes, N2 if it extended to the second-tier nodes, and N3 if it extended to the third-tier nodes. Among the 107 patients, an AJCC-D222 was performed in 63 patients, a JRSGC-D220 in 35 patients, and a D3 in 9 patients. A distal6 and total gastrectomy was performed in 97 and 10 patients, respectively. Lymphadenectomy at the splenic hilum was achieved in 15 patients by splenectomy or by skeletonizing the short gastric artery and splenic hilum.23 The scope of lymphadenectomy and amount of resection depended on the surgeon's decision to achieve an R0 resection.
Clinical classification of T (cT) and N (cN) was determined by preoperative and intraoperative evaluations, including barium radiography, upper gastrointestinal tract endoscopy, abdominal ultrasonography, computed tomography when possible, and findings from surgical exploration. Endoscopic ultrasonography has been applied for the preoperative evaluation of cT since 1987. Evaluations of serosal involvement during surgery were also recorded and classified as serosal-negative, suspected involvement, and serosal-positive. The intraoperative cN classifications were evaluated by node palpation. Pathologic N classification (pN), determined postoperatively, was based on routine hematoxylin-eosin–based histologic examinations in the resected specimens.
The last date of follow-up was May 2002. Recurrence was confirmed by clinical and radiologic findings. The first site of detectable failure was recorded as the predominant recurrence patterns, which were classified into 4 groups: hematogenous metastasis, peritoneal dissemination, distant lymphatic spread, and regional lymph node recurrence.
An unpaired t test was used to test the equality between the 2 means of variables. The Fisher exact test or χ2 test was used to test the independence between the 2 groups. P = .05 was considered significant.
Table 2 presents the incidences of overall node-positive pN1, pN2, and pN3. The patients with pN2 or greater comprised 32.7% of the 52 node-positive patients. No significant difference was found in incidences of each pN between middle- and lower-third GCpm (P = .6). The lower-third GCpm exhibited more frequent incidences of intestinal-type histologic characteristics (P = .03) and vascular invasion (P = .04).
Patients with node-negative GCpm showed significantly higher incidences of cN0 (P = .006) and of the combined findings of serosal negativity and measurements less than or equal to cT2 cN0 (P = .01) than patients with node-positive GCpm (Table 3). Furthermore, the correlation between nodal status and serosal negativity was marginally significant (P = .06). On the other hand, neither the distributions of sex, predominant tumor location, gross form, histologic type, nor the mean values of age, maximum tumor diameter, and dissected nodes proved significant between node-positive and node-negative patients (Table 3). The distribution of pN was determined by preoperative and intraoperative findings and was dependent on tumor diameter. Patient distributions of each pN were analyzed according to the predominant tumor location, histologic type, maximum tumor diameter, serosal findings, and cT cN classifications (Table 4).
Among the middle-third GCpms, intestinal-type tumors that were serosal negative and less than or equal to cT2 cN0 revealed pN0 or, even if they were node-positive, the positive nodes were confined to the first or selective second tier when the maximum tumor diameter was less than or equal to 2 cm. Nodal involvement extended beyond the selective second tier if the maximum tumor diameter exceeded 2 cm. The same findings were obtained in diffuse-type tumors with serosal negativity and measurements less than or equal to cT2 cN0.
Among the lower-third GCpms, most of the tumors that were serosal-negative and less than or equal to cT2 cN0 were actually pN0 or pN1, irrespective of the histologic type and maximum tumor diameter. One exception was a patient with selective second-tier node-positive cancer with a histologically diffuse tumor greater than or equal to 5 cm in maximum diameter. Among the 62 patients with cN0, 23 patients (37%) were found to have node-positive cancer (≥pN1).
Overall, the mean and median follow-up periods were 100 and 108 months, respectively. Nine patients died of cancer, the documented failure patterns being hematogenous metastasis in 6 patients and a distant lymphatic spread in 3 patients. No patients exhibited peritoneal relapse or regional lymph node recurrence. In patients with serosal negativity and cT2 cN0 less than or equal to 2 cm in maximum diameter (13 patients), an AJCC-D2 dissection was performed in 8 patients and a JRSGC-D2 dissection in 5 patients. Five patients had positive nodes; however, they were confined to the first and selective second tier, and nodal involvement beyond that was not observed. No patients died of cancer during a mean follow-up of 121 months (range, 65-169 months).
Extensive archives of patients with GC in Japan demonstrate that the incidences of overall node-positive, pN1, and greater than or equal to pN2 in GCpm are 53%, 30%, and 23%, respectively.24 These incidences are consistent with our findings, suggesting that our sample of patients was not an eccentric population and had received a sufficient lymphadenectomy, and that a misinterpretation of the actual positive node location, ie, stage migration phenomenon, is unlikely. Our results also agree with the previous analyses of failure patterns, in which serosal-negative GC rarely developed peritoneal dissemination and hematogenous metastasis was the most frequent.17,18 These findings clearly suggest that the complete elimination of cancer-bearing lymph nodes is essential and provides a chance for successful cure in the GCpm patient. From the data demonstrated in the present study and others,24 there is no doubt that D1 dissection alone leaves positive nodes in the second tier, which account for approximately 23% to 33% of patients with node-positive GCpm. Therefore, the recommended scope of lymphadenectomy for GCpm, in Japan as well as in several Western institutes, has been at least a D2 dissection.11- 16
Pylorus-preserving gastrectomy has been reported to have many advantages over conventional Billroth I reconstruction. The PPG preserves pyloric function, which prevents dumping syndromes,6,7 preserves the reservoir function of the residual stomach,8 reduces alkaline fluid regurgitation,9 and improves postoperative nutritional status.6 The scope of lymphadenectomy in our PPG, in contrast with that of other investigators,7- 9 includes suprapyloric lymph nodes, making it possible to extend the scope to the second tier. Therefore, PPG can be theoretically applied to some segments of the GCpm patient population and, if so, PPG is preferable to conventional Billroth I reconstruction for such patients when a negative resection margin can be achieved. Among the second-tier nodes, the selective second-tier nodes should be targeted for preferential dissection (a modified D2 dissection) to increase the possibility of leaving an intact spleen and pancreas in accordance with the concept of less-invasive surgery. The above hypothesis has compelled us to search for a segment of GCpm patients who can be cured by PPG with a modified D2 dissection.
Recent immunohistochemical analyses have revealed micrometastases of GC in the dissected nodes, which cannot be detected by routine hematoxylin-eosin–based testing. The reported incidences of micrometastases of 12% to 30% in mucosal GC,25- 27 15% to 25% in submucosal GC,25- 29 and 17% to 24% in all early GC25,26,30 suggest a potential risk of nodal microinvolvement even for early GC. Although the adverse survival effect on micrometastases remains controversial,26- 30 the existence of micrometastases provides, at the moment, a theoretical basis for proposing that the scope of lymphadenectomy should be larger than the scope of nodal involvement (D>cN).27 Therefore, PPG with a modified D2 dissection for GCpm is justified for tumors with serosal negativity, measurements less than or equal to cT2 cN0 (also proved to be pN0) or, if node-positive, the positive stations should be within the first or selective second tier. Such combined analyses of preoperative, intraoperative, and postoperative findings are indispensable to establish indications for less invasive surgery, since surgical procedures should be decided preoperatively and intraoperatively. Under these conditions, tumors with suspicious serosal positivity and/or cT3 and/or measurements greater than or equal to cN1 are not candidates for PPG, even if they are proved to be GCpm and pN0∼pN1.
Our combined analyses reveal that, for middle-third GCpm, tumors that are serosal negative, less than or equal to cT2, and cN0 less than or equal to 2 cm in maximum diameter exhibited less than or equal to pN1 or restricted involvement in the selective second-tier nodes regardless of the histologic type. Conversely, nodal involvement beyond the selective second-tier stations was present in GCpm tumors that exceeded the above cutoff diameter. For lower-third GCpm, all tumors that were serosal negative, less than or equal to cT2, and cN0 showed less than or equal to pN1 or restricted involvement in the selective second-tier nodes. Since the 2.5-cm length of the pyloric cuff has been demonstrated to be the most desirable length in a recent comparative study,31 the distance between the pyloric ring and the distal margin of the tumor requires greater than or equal to 4.5 cm to nullify the risks of a positive resection margin. Therefore, for lower-third GCpm, larger tumors should not be treated by PPG to avoid a positive resection margin; thus, a cutoff diameter of 2 cm is realistic. Taking all of these factors together, we find that among tumors of middle- or lower-third GCpm that are serosal negative, less than or equal to cT2 cN0, and less than or equal to 2 cm in maximum diameter with the distal margin greater than or equal to 4.5 cm from the pyloric ring are rationally treated with curative intent by PPG with a modified D2 dissection. These criteria are practical since those who could enjoy the benefits of PPG (13 patients) comprised one quarter of the patients with tumors that were serosal negative and less than or equal to cT2 cN0 (53 patients), and 12% of the 107 patients with GCpm.
Based on the concept that complete clearance of lymph nodes that have a risk of metastasizing is necessary in a curative approach, one may doubt that a modified D2 dissection is really sufficient for patients with pN1 or pN2 even if they fulfill the previously mentioned cutoff criteria. In our series of patients, the possibility of residual diseases following a modified D2 dissection is unlikely, as those who fulfilled the criteria did not die of cancer, survived more than 5 years, and did not have nodal involvement in the stations beyond the selective second tier. As has been demonstrated by this study and others,15 the diagnostic accuracy cN is unreliable; thus, careful evaluation of serosal involvement, cT, and cN should be a prerequisite, and frozen-section examinations in suspicious involved nodes or sentinel nodes may sometimes be necessary. Under these strict evaluations, carefully selected GCpm patients can enjoy the benefits of PPG.
Corresponding author and reprints: Shouji Shimoyama, MD, Department of Gastrointestinal Surgery, University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo 113-8655, Japan (e-mail: firstname.lastname@example.org).
Accepted for publication March 23, 2003.