Surgery is widely accepted as the first-line treatment for gastric cancer, and curative resection is the only therapy offering hope of a cure.

One of the most critical steps in surgery is the determination of the tumor margin and its distance from the resection line, which is usually established on the basis of the histotype and macroscopic type, being wider in non-differentiated and infiltrating carcinomas.

An infiltrated resection line (RL) in patients undergoing radical surgery indicates a significantly poorer outcome.1,2 However, in the event of delayed detection of margin infiltration, further surgery, even when possible, is not always advisable,1,3 and some authors report that retreatment may modify prognosis only in the absence of lymph node metastases.4

A singular setting is that of patients with early gastric cancer (EGC) and positive resection line is subsequently found. We describe our experience of 11 such patients who did not undergo surgical retreatment.

MATERIALS AND METHODS

From 1988 to 2001, 2740 patients underwent resection for gastric cancer in six Italian Surgical Units. One hundred and fifty-eight (5.7%) patients in whom a positive resection line was histologically confirmed were not surgically retreated for reasons of age, associated disease, or patient refusal. The present study evaluates 11 patients in this subgroup with mucosal or submucosal cancer and resection line involvement, representing 1.7 % of the 644 EGC patients operated on during this period (EGC constituted 23.5% of all the surgically treated gastric cancers).

Subtotal gastrectomy was performed for tumors located in the lower two thirds of the stomach, with removal of the greater and lesser omentum. This was associated with en-bloc resection of perigastric lymph nodes or dissection of first and second level lymph nodes, in accordance with Japanese Gastric Cancer Association (JGCA) recommendations.5 When possible, a macroscopic distance of more than 2 cm between the tumor and the upper resection margin was maintained. Hand-sewn anastomosis was generally performed, but staple guns were sometimes used for duodenal stump.

Histological type was based on Lauren’s classification.6 Death from postoperative complications was considered a hospital mortality. Patients were seen every 6 months for the first 5 years and once a year thereafter. Routine follow-up tests included abdominal scan, chest x-ray, and serum CEA/CA19-9 levels. Gastroscopy was performed 6 months after surgery and once a year thereafter. Survival was calculated from the date of surgery, and only cancer-related mortality was taken into account. Death from other causes was considered censored.

Survival was analyzed using the Kaplan-Meier product limit method.7

RESULTS

Eleven patients presented EGC and resection line involvement. Six patients were males and 5 were females, with a median age of 62.6 years (range: 49–75 years). All lesions were sited in the lower two thirds of the stomach and patients underwent subtotal gastrectomy with BII reconstruction (8 patients) or Roux-en-Y reconstruction (3 patients). The clinical pathological characteristics of these patients are summarized in Table 1.

Table 1 Patients with resection line (RL) involvement as a function of clinical pathological factors
Full size table

All patients were treated with curative resection intent, and microinvolvement of the resection line, not suspected at the time of surgical treatment, was subsequently histologically confirmed. Four patients presented a positive proximal resection line, and 7 had distal infiltration. Staple guns were used to resect the duodenum in 5 of the patients in the latter group. Specimens were always carefully opened by the surgeon, who removed the staple line to verify the macroscopic distance of the resection margin from the tumor.

Six patients operated on before the D2 procedure became standard practice in all our centers or elderly patients with associated diseases were subjected to D1 lymphadenectomy, whereas the remaining 5 patients underwent D2 lymphadenectomy, with a median number of 12.5 (range: 6–20) and 23 (range:15–52) lymph nodes dissected for D1 and D2 procedures, respectively. Three patients presented first level lymph node metastases according to JGCA criteria and were classified as N1 according to the new TNM classification. None of the patients had second level lymph node metastases.

The median lesion size was 25.1 mm (range: 10–65 mm). Seven patients presented mucosal EGC and 4 had submucosal invasion. With regard to macroscopic type, 2 patients had type I EGC, 4 had type II (1 type IIb and 3 type IIc), and 5 had type III.

On the basis of Lauren’s classification, 4 patients presented intestinal-type adenocarcinoma and 7 had diffuse-type carcinoma.

The gastric margin was involved in 4 patients and 7 had duodenal resection line involvement. No surgical complications or postoperative deaths were observed.

Surgical retreatment was not considered for these patients for reasons of age, associated diseases, patient refusal or, in the event of an infiltrated duodenal margin, because duodenocephalus pancreasectomy as an extension of the original resection was not considered achievable. In the absence of surgical alternatives, 3 of the 11 patients subsequently underwent complementary chemotherapy, decided upon in their own center and not based on standard protocols, because the impact of a positive resection margin on the prognosis of EGC patients was still not fully understood.

The median follow-up was 93.4 months (range: 24–151 months), and no patients were lost to follow-up. One of the 11 patients developed liver metastases and died 77 months after surgery. None of the remaining 10 patients presented relapse.

Five-year and 8-year survival was 100% and 86% (95% CI = 60–100), respectively (Table 2). The only patient who relapsed did not have lymph node metastases, whereas 3 patients with N1 disease showed an 8 year survival of 100%.

Table 2 Patient survival
Full size table

DISCUSSION

A positive resection line has been reported in varying percentages (0.8%–10.2%) of surgically treated gastric cancer patients.1,8,9 The impossibility of manual palpation of the lesion in EGC renders the determination of resection line more difficult. In recent years, chromoendoscopy and magnifying endoscopy have led to a more accurate localization of the lesion; when this is near the surgical resection line, the use of an extemporary histological examination is always advisable.

Several studies have evaluated the extension of gastric cancer invasion when the esophageal or duodenal margin is involved to guarantee optimal surgical management. A 2 cm esophageal resection from the gastric cancer margin is generally considered safe when a histologically well-defined lesion is resected, but when undifferentiated tumor invasion is present, 4–7 cm of the esophagus may have to be resected10 because manual palpation or frozen sections may not suffice to detect the presence of skip submucosal foci.11

Duodenal invasion, when present, generally extends about 2 cm beyond the pyloric ring, but a 4 cm invasion through submucosal or subserosal layers or lymphatics has also been described.11–14

Clinical pathological factors that usually indicate a risk for positive resection line are macroscopic Borrmann types III and IV,9 T and N stage, location, and histologically nondifferentiated tumors.1,14

With regard to EGC macroscopic types, all patients with a positive resection line described by Fujmoto had excavated or superficial-type lesions.9 Huguier et al., described 4 EGC patients with macroscopic flat-type tumors and RL involvement.15 Yamaguchi, commenting on the study of Hugier et al., suggested that the use of routine frozen sections might be indicated only in flat-type EGC tumors.15 Conversely, we observed all macroscopic types except IIa, with a high prevalence of excavated lesions.

Resection line involvement is generally reported to cause high surgical morbility, and Keighley et al., in their 1981 paper, reported 4 anastomotic leakages in 4 total gastrectomies.16 Recently, Chan et al., reviewing 137 total or proximal gastrectomies, reported an 18.2% incidence of positive esophageal margins but did not observe an increased morbility or anastomotic leak incidence compared to patients with negative margins.17 Our experience confirms these data, and no surgical complications were observed in any of the 11 EGC patients with RL involvement.

Microscopic residual tumor, which is generally described as a significant prognostic factor,1,2,18 indicates poor survival, and several authors have proposed surgical retreatment when achieving a tumor-free resection line seems realistic.1,4 Other authors such as Papachristou et al., after reporting unsatisfactory results from additional treatment, suggested monitoring these patients in order to treat only anastomotic recurrence.3 Cascinu et al., comparing resection line-positive and negative patients treated with D1 lymphadenectomy, observed lower survival only in N0 patients with RL involvement and concluded that surgical retreatment may be indicated only in this subset of patients.4 Kim et al., in a study of 47 resected patients with positive margins, maintained that this negative prognostic factor lost its significance in those who underwent D2 or D3 lymphadenectomy. Examining this subset, they found that the survival of patients who had ≤5 positive lymph nodes was significantly worsened by a microscopically involved margin and concluded that outcome is determined by the presence of >5 lymph nodes rather than by macroscopic residual cancer at the margin.19 Although D2 dissection is the standard surgical treatment for EGC in all our centers, in this retrospective study, 6 D1 and 5 D2 lymphadenectomies were performed for reasons explained previously.

With regard to EGC, Nakamura et al., observed a good survival rate in 29 EGC patients and concluded that positive margins in EGC were of insignificant prognostic value.18 Our results confirm these findings and, despite the very modest survival rates in the literature for resected advanced gastric cancer patients with RL involvement, which our experience shows to be 9% at 5 years (95% CI 4–14) and 7% at 8 years (95% CI 2–11), we observed a 5 and 8 year survival of 100% and 86%, respectively, for the EGC patient subset in the present study. These rates are similar to those reported in a previous multicenter study of 584 patients in whom 5 year survival was 95% for N0 patients and 77% for N1 patients.20 Furthermore, in 3 of our patients evaluated as N1 on the basis of both JGCA criteria and the 1997 TNM classification, prognosis was not influenced by the presence of positive lymph nodes and they were alive and disease-free at a median follow-up of 69 months.

Only 1 of the 11 EGC patients presented distant disease recurrence and subsequently died during the follow-up period, whereas the remaining 10 did not show any signs of relapse. It is somewhat difficult to evaluate these data, especially with regard to the increased risk of relapse observed in patients undergoing endoscopic resections, when complete or en-bloc resection is not guaranteed. It is not known whether the use of staple guns or a more extended lymphadenectomy or deeper surgical margins than those used for endoscopic resection could lead to better results in surgically treated ECG patients, and further studies investigating this issue are warranted.

In conclusion, resection line involvement after gastrectomy generally indicates a very poor prognosis and extemporary histological examination of the margin is always advisable for suspicious lesions. A new margin resection is always advisable, when possible, to ensure optimal results. Notwithstanding, the 11 EGC patients considered in the present study presented good survival at 5 and 8 years. Furthermore, no postoperative complications or mortality were observed. The decision to submit this category of patients to surgical retreatment, should, however, take into account the above-mentioned observations.