Editorial comment: Feasibility of central gastrectomy for gastric cancer☆☆☆

Patients and methods 

Included in the current study were 100 patients with gastric cancer who underwent CG at Shizuoka General Hospital from July 1992 to September 1996. Informed consent was obtained from all patients. For comparison, we also reviewed the hospital records of 100 patients who underwent conventional DG (including subtotal gastrectomy) from January 1986 to June 1992 and had indications for CG. This study was completed in September 1998; the CG group was followed-up for 24 to 76 months (median, 49 months) and the DG group was followed-up for 24 to 99 months (median, 55 months). At least 1 surgeon from a team of 6 members participated in all operations either as an operator or as the first assistant.

Table I shows indications and exclusion criteria for CG. The vertical location of the stomach was recorded as the upper, middle, or lower third of the stomach defined by the 2 lines dividing each curvature into 3 equal parts. When a tumor was spread over 2 portions, the main involved portion was described as the tumor location. When a tumor was located in the distal half of the upper third of the stomach, CG was offered to limited patients with a distance of more than 3 or 4 cm from the esophagogastric junction to the edge of the cancer. When an apparent nodal metastasis was found at the root of the gastric or gastroepiploic artery during operation, conventional DG was used to achieve more exhaustive lymphadenectomy.

Table I. Indications and exclusion criteria for central gastrectomy (CG)

	Indications
	1) Tumor invasion restricted to T1 or limited T2 (muscularis propria and macroscopically localized type)
	2) Tumor size ≤ 5 cm
	3) Tumor in the middle third, distal half of upper third, or proximal half and greater curvature side of lower third of the stomach
	Exclusion criteria
	4) Macroscopic tumor size < 2 cm and depth limited to mucosa → Endoscopic mucosal resection or wedge gastrectomy indicated
	5) Apparent node metastasis noted at the root of the gastric or gastroepiploic artery → Distal gastrectomy (DG) indicated

When a cancer was detected by routine endoscopic examination, longitudinal and horizontal step biopsies were done to map the precise extent of the cancer during a second endoscopic examination. Depth of tumor invasion was judged by either barium radiograph or endoscopic examination. Endoscopic ultrasonography was not used. When an accurate assessment of the site of the lesion could not be detected during operation, intraoperative endoscopic localization or direct visualization by gastrotomy was done early during the procedure (37 patients), and preoperative endoscopic clipping, 1 cm from the cancer edge, was usually done later (40 patients).

As shown in the Figure, A, the operative strategies for CG were: (1) excision more than 2 cm from the edge of T1 cancer and 3 cm from a T2 cancer; (2) preservation of hepatic and pyloric branches of the vagal nerve to maintain pyloric function; (3) the area around the right gastric artery was not disturbed to preserve the neurovascular supply to the distal antrum and pylorus, and lymph nodes at the proximal short part of the right gastroepiploic artery were dissected with preservation of the vessels to maintain the blood supply of the distal greater curvature of the stomach; (4) excluding the area around the right gastric artery, lymph nodes were principally dissected to near the same level as that in conventional DG; however, in lesser curvature mucosal cancers less than 3 cm in size, we occasionally omitted lymph node dissection along the proximal right gastroepiploic artery (Fig, B).

View Large Image Download to PowerPoint

Fig. Schematic illustration of CG. A, Basic procedure. B, Less invasive procedure, locating on the lesser curvature. C, Less invasive procedure, locating on the greater curvature. GA, Gastric artery; GEA, gastroepiploic artery; CHA, common hepatic artery; Ca, cancer.

Clinical, surgical, and pathologic findings were described on the basis of UICC TNM classification for gastric cancer.9 The pathologic N classification in the new UICC TNM staging system is defined according to number of metastases to the regional lymph nodes. The TNM classification recommends examination of 15 or more lymph nodes; in this study, we included patients in whom fewer than 15 nodes were excised. Three patients satisfying CG indications (T factor, tumor size, and location) were converted from CG to DG because of macroscopic findings with apparent lymph node metastasis at the root of the left gastric or the right gastroepiploic artery.

The data are expressed as mean ± SEM. Statistical analysis should be considered as descriptive because of the retrospective nature of our study and because the 2 groups of patients were from different time eras. We used a Statview software package (SAS Institute, Inc, Cary, NC). A P value of less than .05 was considered significant. Relative comparability of the CG and DG groups was analyzed using Student t test and chi-square statistics. Survival was estimated using the Kaplan-Meier method.

Results 

Pathologic characteristics 

Table II shows pathologic characteristics of patients who underwent CG.

Table II. Histopathologic characteristics of patients undergoing CG

	Tumor location		Node metastasis
	Upper third	15	pN0	86
	Middle third	67	pN1	14
	Lower third	18	Pathologic stage
	Tumor size		stage Ia	75
	2 cm	31	stage Ib	18
	≥2 cm, ≤5 cm	59	stage II	7
	>5 cm	10	Proximal margin
	Histologic typing		Mean (cm)	2.9 ± 0.2
	Well-differentiated	24	<1 cm	10
	Moderately differentiated	23	≥1 cm, <3 cm	47
	Poorly differentiated	28	≥3 cm	43
	Signet-ring cell	25	Distal margin
	Tumor invasion (depth)		Mean (cm)	2.6 ± 0.2
	pT1 (mucosal)	50	<1 cm	7
	pT1 (submucosal)	32	≥1 cm, <3 cm	62
	pT2	18	≥3 cm	31

Transverse tumor sizes greater than 5 cm were seen in 10 patients, whereas longitudinal tumor sizes greater than 5 cm occurred in only 2 patients. Pathologic depth of invasion of tumors less than 2 cm in size (n = 31) was confined to mucosal (n = 22), submucosal (n = 8), and muscularis propria (n = 1). Of 18 patients with pT2 tumors, 15 showed invasion to muscularis propria, whereas 3 showed invasion to the subserosal layer. Numbers of dissected lymph nodes along the left gastric artery, left gastroepiploic artery, and right gastroepiploic artery were 7.2 ± 0.5, 1.3 ± 0.2, and 7.8 ± 0.6, respectively. Number of dissected lymph nodes along the common hepatic artery (n = 69) was 1.2 ± 0.2. Total number of dissected nodes per patient was 17.3 ± 1.0. The 14 patients with lymph node metastasis included 11 with 1 positive node, 2 with 2 positive nodes, and 1 with 5 positive nodes. Metastatic nodal sites were along the lesser curvature (n = 4), the greater curvature (n = 9), and both areas (n = 1). The 14 patients who were lymph node positive showed histologic typing from well to moderately well-differentiated cancers (n = 6), and poorly differentiated or signet-ring cell cancers (n = 8); depth of tumor invasions were pT1 (mucosal, n = 1; submucosal, n = 6) and pT2 (n = 7). The size of mucosal cancer with lymph node metastasis was 3.0 × 2.0 cm. Pathologic features in relation to extent of lymph node dissections are shown in Table III.

Table III. Pathologic features in relation to extent of node dissection in patients undergoing CG

			Tumor size		pT			pN
		No. of patients	<3 cm	≥3 cm	pT1 (m)	pT1 (sm)	pT2	pN0	pN1
	(A) Basic LN dissection	59	28	31	24	20	15	47	12
			(2)	(10)	(1)	(5)	(6)
	(B) LN along the proximal right GEA; not dissected	19	16	3	10	7	2	18	1
				(1)		(1)
	(C) LN along the left GA trunk; not dissected	22	19	3	16	5	1	21	1
			(1)				(1)

Parentheses indicate number of patients with positive LN metastasis. LN, Lymph node; GA, gastric artery; GEA, gastroepiploic artery; m, mucosa; sm, submucosa.

Of 59 patients who underwent basic lymph node dissection (Fig, A), 17 patients received lymph node dissections along the trunk of the left gastric artery with excision of the descending branch of the artery.

The patients undergoing DG (including subtotal gastrectomy) had tumor locations in the upper third (n = 12), middle third (n = 79), and lower third (n = 9); tumor sizes of < 2 cm (n = 19), ≥ 2cm-≤ 5 cm (n = 73), and > 5 cm (n = 8); histologic differentiated (n = 44) and undifferentiated lesions (n = 56); pT1 (n = 82) and pT2 (n = 18); lymph node metastases of pN0 (n = 93) and pN1 (n = 7); and final pathologic stage Ia (n = 80), stage Ib (n = 15) and stage II (n = 5). No important differences were observed between DG and CG groups. Patients undergoing CG compared with DG had lesser proximal margins (2.9 ± 0.2 vs 3.7 ± 0.2 cm) and distal margins (2.6 ± 0.2 vs 8.4 ± 0.2 cm). Number of dissected lymph nodes per patient undergoing DG was 21.6 ± 1.0.

A second lesion was found in pathologic specimens from 5 patients undergoing CG and 3 undergoing DG. In 2 of these 8 patients, the second lesion was detected preoperatively during the endoscopic examination.

Perioperative data and morbidity 

Operating time for CG was 207 ± 5 minutes, operative blood loss was 294 ± 21 mL, and postoperative hospital stay was 17.3 ± 0.7 days. There was no hospital mortality. Five patients had major postoperative complications (Table IV). Two patients of the 100 undergoing CG had a prolonged postoperative hospital stay (54 and 50 days) as a result of severe gastric stasis. As their food intake gradually increased, additional surgical or endoscopic treatment was not required.

Table IV. Morbidity and follow-up

		CG (n = 100)	DG (n = 100)
	Morbidity
	Anastomotic leakage	1	3
	Gastric stasis	2
	Anastomotic stenosis		2
	Ischemic gastric ulcer	1
	Intra-abdominal bleeding	1
	Late result
	Cancer recurrence		1
	Death	3	7
	Lost to follow-up		3
	Cause of death
	Cancer recurrence		1
	Other diseases	3	6
	Occurrence of new gastric cancer	3	1

Discussion 

Recently, pylorus-preserving gastrectomy has been used for small gastric cancer in Japan to preserve pyloric function. Although it can reportedly alleviate postgastrectomy symptoms such as dumping syndrome or duodenogastric reflux, the resected gastric volume is similar to that by conventional DG.5, 6 In pylorus-preserving gastrectomy, the suprapyloric node dissection is excluded because the right gastric artery must be preserved to supply blood to the prepyloric stomach and to maintain the pyloric vagal nerve branch.4, 5, 6 The suprapyloric node can be excluded because the metastasis to suprapyloric nodes is only rarely observed in early gastric cancer in the middle third of the stomach.4, 17, 18

On the basis of the finding that lymph nodes along the proximal right gastroepiploic artery can be sufficiently dissected without excising the artery, we suggest that the level of node dissection in CG is similar to that in conventional DG with the exception of suprapyloric node dissection. This reasoning led us to develop the CG procedure.12 Because CG appears to afford sufficient node dissection, we extended its indications to T1 and limited T2 cancers of 5 cm or less in size. This concept was on the basis of the fact that a gastric resection of maximum 9 to 11 cm (5-cm tumor size plus 2- to 3-cm oral and anal margins) maintains an adequate residual gastric volume. Ohwada et al10 proposed that CG is indicated for patients with any mucosal cancers and submucosal cancers less than 5 cm in size with differentiated types. Although, with cancers less than 3 cm and invasion confined to the mucosa, nodal metastasis is extremely rare.19 Thus, we devised less invasive options for mucosal cancers less than 3 cm in size, in which lymph node dissection along the trunk of the left gastric artery or the proximal right gastroepiploic artery is omitted (Fig, B and C). Furukawa et al14 recommended segmental gastrectomy with limited perigastric node dissection for mucosal cancer of any protruded type or depressed type less than 2 cm in the midstomach.

We performed CG on the basis of our criteria (Table I), but postoperative histologic findings of tumor size and depth often differed from those diagnosed preoperatively. CG was carried out in 22 patients with mucosal cancers of less than 2 cm in size. Specifically, wedge gastrectomy or endoscopic mucosal resection was undertaken for patients whose histologic diagnosis was expected to coincide with the preoperative diagnosis. Recently, tumor depth was assessed using preoperative endoscopic ultrasonography. However, as its diagnostic accuracy for early gastric cancer is relatively low (44%20 or 55%21), we have not yet adopted this method to aid in preoperative staging.

In 41 patients, we did not use a lymph node dissection along the trunk of the left gastric artery or the proximal right gastroepiploic artery. We believe it is possible to conduct a focused node dissection for mucosal cancers less than 3 cm in size. However, we would emphasize that the basic node dissection shown in Fig, A, should usually be done in CG because precise preoperative diagnosis about the depth of tumor invasion remains difficult and remote lymph node metastasis may occur even in submucosal cancer.22

Two of our patients had severe gastric stasis develop postoperatively, probably as a result of injury to the hepatic or pyloric branch of the vagal nerve during operation. No cancer recurrence was experienced after CG. In 5 patients undergoing CG and 3 undergoing DG in our series, a second, small cancer lesion was detected by preoperative endoscopic examination or by postoperative pathologic examination. During follow-up after CG, a new cancer was detected in the residual stomach of 3 patients (distal stomach, 2; proximal stomach, 1). As with all localized resections for gastric cancer, this is a serious problem with CG. To prevent overlooking new lesions, postoperative endoscopy should be done once or twice a year. Takeda et al23 recommend periodic endoscopy to detect early gastric cancer arising from the remnant stomach. Another important concern is the direction of lymph drainage after CG. Strategies for the range of lymph node dissection for patients with a new gastric cancer during the follow-up of CG must be proposed hereafter.

Some studies15, 16 report abnormalities in postoperative gastric emptying and symptomatology after segmental gastrectomy in patients with benign gastric ulcer, though only a few deal with postoperative gastric cancer.11, 12 It is important to evaluate postoperative symptoms and physiologic parameters such as weight change, food intake, and gastric emptying with patients considering CG. Koufuji et al11 found that early satiety is high (44%) even 6 months after segmental gastrectomy. Ohwada et al24 reported that emptying of a liquid meal returns to a level similar to healthy individuals but emptying of a solid meal remains prolonged 1 year after segmental gastrectomy. In contrast, the 1-year postoperative/preoperative body weight after pylorus-preserving gastrectomy varies from 91% to 99%.5, 6 Tomita et al25 found that all patients after pylorus-preserving gastrectomy had early satiety and that gastric emptying of a semisolid diet was slower than in healthy control participants. Nakane et al26 reported that the early satiety after pylorus-preserving gastrectomy decreased from 35% (at 1 year after operation) to 16% (2 years after operation).

In 87% of our patients after CG, food intake 1 year postoperatively was 80% or more of the preoperative food intake. However, 63% of our patients had early satiety at 1 year after operation. The percentage of postoperative to preoperative body weight was 95% in CG group, and 91% in DG group. There are several potential technical variations of CG that would preserve either the perigastric artery or gastric branches of the vagal nerve. A much more limited procedure would be expected to result in a better gastric function postoperatively. One would assume that by using the basic operative procedure shown in Fig, A, a larger volume of the remaining proximal or distal stomach would improve postoperative early satiety and decreased food intake quantity. Although postoperative changes in body weight of our patients undergoing CG were minimal, study of the volume of retained stomach to postoperative gastric symptoms would be very interesting to see if this assumption was true.

In summary, CG with a distal margin of 2 to 3 cm from the edge of the cancer makes it possible to dissect almost the same extent of lymph nodes as that by conventional DG. Patients undergoing CG showed excellent postoperative survival and minimal weight loss. We conclude that CG is a useful operative procedure for selected patients with gastric cancer. However, it will be necessary to investigate postoperative gastric emptying and the incidence of new cancer developing in the residual stomach.