As proposed by the Japanese Society of Gastroenterological Endoscopy in 1962, EGC is defined as adenocarcinoma that is limited to the gastric mucosa or submucosa regardless of the involvement of regional lymph nodes (T1) [13]. Many studies have clarified the status of LN metastasis in EGC. The overall incidence of LN metastases in T1 EGC is 10 to 20% [2,14-17]. The characteristics of the tumor such as the size, cancer depth, histologic type, and the presence of lymphovascular invasion are important determinants of the likelihood of spread [2,18-20]. For example, the Roviello et al. study evaluating 652 cases of resected EGC [2] showed that the incidence of LN metastasis to be 14.1% overall: 4.8% versus 23.6% for mucosal versus submucosal cancer. Smaller cancers were significantly less likely to be associated with positive nodes: 9% versus 20% and 30% for tumors <2 cm, 2 to 4 cm, and >4 cm in diameter, respectively. In the Sano et al. study, well-differentiated type I and IIa T1 tumors of less than 2 cm in diameter, and nonulcerative type IIc T1 tumors of less than 1 cm in diameter, were associated with a low risk of LN metastases (1.7%) [19]. Such a volume of sufficient data has contributed to the development of indications for endoscopic treatment.

The number of studies reviewing both the progression of LN metastasis from EGC to advanced gastric cancer (AGC) and the status of AGC is insufficient. A report from Japan suggested that >60% of untreated EGCs will progress to AGC within five years [21]. Nakajima et al. reported that the incidence of LN metastasis of gastric cancer with invasion to MP, SS, SE, SI were 52.2%, 66.9%, 74.4%, and 82.6%, respectively [22]. However, it is difficult to judge the presence and the extent of LN metastasis in AGC before operation. Two issues are related to evaluation of the incidence of LN metastasis in AGC: First, many factors, such as location, depth, size, macroscopic type, and histological type of the AGC, affect the incidence and distribution of LN metastasis. Second, the diagnosis of LN metastasis with resected specimens is affected by examination methods such as H and E staining, immunohistochemical staining, and reverse polymerase chain reaction. To predict the incidence and distribution of LN metastasis in detail before operation for AGC, a special modality such as the computer information system developed by Maruyama et al. is necessary [23].

There are few reports about the accuracy of preoperative LN status using abdominal ultrasonography. Isozaki et al. reported that the detection rate of LN metastasis by transabdominal US was 5% [24]. Due to problems with intraluminal gas, abdominal US of the gastrointestinal tract is not commonly used and has not been developed. Rather than abdominal US, a number of studies of the effectiveness of endoscopic ultrasonography (EUS) have been reported (described in section 2.3).

Dynamic CT scanning is usually performed early in the preoperative evaluation after a diagnosis of gastric cancer is made. CT is widely available and noninvasive. It is good for widely evaluating metastatic disease, especially hepatic metastases, ascites, and distant nodal spread. In 20 to 30% of patients with a negative CT, however, intraperitoneal disease will be found at either staging laparoscopy or at open exploration [25-27].

Another limitation of CT is its inability to accurately assess the depth of primary tumor invasion and the presence of LN involvement. CT accurately assesses the T stage of the primary tumor in only about 50 to 70% of cases [28-34]. The classification of nodal status is usually based on LN size, and sensitivity of CT for detecting regional nodal metastases is limited for involved nodes that are smaller than 0.8 cm [28,33]. Furthermore, false-positive findings may be attributed to inflammatory lymphadenopathy. In several series of patients undergoing staging CT for gastric cancer or gastric plus esophageal cancer, sensitivity and specificity rates for detection of regional nodal metastases ranged from 65 to 97% and 49 to 90%, respectively [35-39].

In comparative studies, EUS generally provides a more accurate prediction of T stage than does CT [40-42], although newer CT techniques (such as three-dimensional multidetector-row CT) and magnetic resonance imaging may achieve similar results in terms of diagnostic accuracy in T staging [39,43,44]. In contrast, accuracy for nodal staging (65 to 90%) is only slightly greater with EUS as compared to CT [40,45-50]. EUS-guided fine needle aspiration of suspicious nodes and regional areas adds to the accuracy of nodal staging [51].

Most errors in staging with EUS are due to understaging of nodal involvement and the depth of primary tumor invasion; however, overstaging can also occur that is attributed to inflammation around the tumor or within the LNs [50]. EUS is not recommended for pretreatment evaluation of gastric cancer in the guidelines from the National Comprehensive Cancer Network (NCCN) [52].

The role of PET using 18-fluorodeoxyglucose (FDG) in the preoperative staging of gastric adenocarcinoma is evolving. From the standpoint of locoregional staging, integrated PET/CT imaging can be useful to confirm malignant involvement of CT-detected lymphadenopathy [53]. However, this usually does not impact the decision to proceed to surgery. Furthermore, a negative PET scan is not helpful because even large tumors with a diameter of several centimeters can be falsely negative if the tumor cells have fairly low metabolic activity. Furthermore, most diffuse-type gastric cancers (signet ring carcinomas) are not FDG avid [54-58]. The main benefit of PET is that it is more sensitive than CT for the detection of distant metastases [42,58-60]. An important caveat is that the sensitivity of PET scanning for peritoneal carcinomatosis is only approximately 50% [61]. Thus, PET is not an adequate replacement for staging laparoscopy. NCCN guidelines for preoperative evaluation of gastric cancer suggest integrated PET/CT [52].

The application of the SLN technique in gastric cancer began in the late 1990s. Intraoperative subserosal or preoperative endoscopic submucosal injections can be used for the administration of blue dye or radioactive tracer. Identification of the SLN by means of a radiolabeled colloid and perioperative detection with a gamma probe has the disadvantage of radioactive tracing not only from LNs but also from the adjacent injection site. Most experience has therefore been gained with blue dye, but blue dye flows through and travels to the next LNs in line. The results reported in the literature on SLN biopsy in gastric cancer are widely divergent. Many authors from Asia reported an accuracy of more than 98% [62-64], in particular in early stages (T1-T2) [65], whereas other series from Western countries, the accuracy was about 80% [66-68], with the false negative SLN rate ranging from 15% to 20% [66-68]. The main reason for the poor accuracy could be the variability of the lymphatic routes in the gastric region, resulting in a high rate of skip metastases. Regarding the utility of SLN navigation in an attempt to detect the nodal basin, many issues are still to be resolved and further studies are recommended before this method can be introduced into daily practice.

Endoscopic resection is currently the standard treatment for EGC without the possibility of LN metastasis in Japan [69], as in the other countries, and is increasingly gaining acceptance as a therapy for EGC [70,71]. On the other hand, gastrectomy with LN dissection is required in cases of possible node metastasis [72] because the presence of LN metastasis has a strong adverse influence on patient prognosis [73,74]. In Japan, endoscopic submucosal dissection (ESD) is indicated for a differentiated mucosal cancer smaller than 2 cm in diameter [75] because risk for LN metastasis is negligible [17]. Recently, by using a large database involving more than 5000 patients who underwent gastrectomy with meticulous D2 level LN dissection, Gotoda and colleagues [17] were able to define the risk of LN metastasis. They revealed that submucosally invasive gastric cancer (similar to mucosal cancers) and tumor size larger than 3 cm with lymphatic or vessel involvement are significantly correlated with an increased risk of LN metastasis, and cancers penetrating deeply into the submucosal layer are most likely to be associated with LN metastasis. The extended indication including: (i) differentiated-type mucosal cancers without ulcerative findings, regardless of tumour size; (ii) differentiated-type mucosal cancers with ulceration findings, <30 mm; (iii) undifferentiated-type mucosal cancers without ulceration findings, <20 mm; and (iv) differentiated-type minute submucosal cancers (SM1) without ulceration findings, <30 mm was proposed by several reports [76,77].

From analysis of data from 118 patients with submucosal invasion, Yasuda et al. suggested that optimal LN dissection levels are as follows: (1) local resection (D0) for lesions of <1 cm; (2) limited LN dissection (D1) for 1- to 4-cm lesions, and (3) extended LN dissection (D2) for lesions >4 cm in diameter. When submucosal invasion of a tumor resected locally by ESD extends more than 300 mm, additional gastrectomy and LN dissection are necessary [78]. However, patients with submucosal invasion are not necessary to undergo D2 lymph node dissection in the Japanese guideline. Further study of the optimal extent of LN dissection for early gastric cancer is expected.

Although many retrospective studies and only one randomized controlled trial (RCT) by a single institution in Taiwan suggest that extended LN dissection improves survival [87-89], multiple prospective randomized trials both in Asian and Western populations have failed to show a survival benefit with D2 versus D1 lymphadenectomy [90-92] or with D3 compared to D2 LN dissection [3,93-95]. The findings of the three largest trials are as follows.

The presence of LN metastasis (pN) is one of the most significant prognostic factors in patients with gastric cancer. However the classification of LN metastasis in patients with gastric carcinoma is controversial. In 1981, the Japanese Research Society for Gastric Carcinoma first proposed a classification based on the anatomical location of positive nodes, which was reviewed by the Japanese Gastric Cancer Association in 1998. In 1997 and 2002, the UICC and the AJCC proposed a new classification for N categories that was based on the number of metastatic LNs (N stage) [107,108]. Now, the UICC/AJCC classification is used most widely for the staging of gastric cancer [107-109]. It can provide a more accurate estimation of prognosis than the classification based on anatomical lymphatic spread. Some authors have pointed out the superiority of UICC/AJCC classification on the grounds of simplicity, reliability, and stratification; they have also mentioned some of the problems associated with it, such as stage migration [110-114]. In 2010, the Japanese guideline has adopted the N categories based on numbers [80].

A new prognostic tool, for the ratio between metastatic LNs and the total number of LNs examined (N ratio), was proposed. This new classification reflects the degree of LN metastasis and reduces stage migration [84,85,115,116]. However, the significance of the N ratio has not been evaluated in patients with <15 examined LNs. Xu et al. evaluated the prognostic value of the N ratio staging system compared with the N stage classification when <15 LNs were examined in gastric cancer patients. N ratio categories were identified as follows: N ration 0, 0%; N ratio 1, 1% to 9%; N ration 2, 10% to 25%; N ration 3, >25%. They concluded that the positive N ratio is an independent prognostic factor, regardless of the number of LNs examined [117].

Micrometastasis was defined as the presence of tumor cells—single or in small clusters—detected only by cytokeratin specific immunostaining that could not be detected by ordinary H and E staining. There are specificities of several different antibodies, such AE1/AE3 (Boehringer Mannheim, Indianapolis, IN, USA), KL-1 (Immunotech, Marseilles, France), and CAM5.2 (Becton Dickinson, San Jose, CA, USA). Yasuda et al. demonstrated that LN micrometastasis is an independent prognostic indicator for patients with histologically node-negative gastric cancer invading the muscularis propria or deeper (T2 or T3) [4]. In addition to the presence of LN micrometastasis, the number and level of micrometastases in the LNs were strongly associated with the survival time of patients. It is controversial whether LN micrometastasis detected by immunohistochemistry predicts the clinical outcome of patients with histologically node-negative gastric cancer [5-9]. Nakajo et al. reported that LN micrometastasis correlated with a significantly worse survival rate in patients with T1 or T2 tumors [8]. Cai et al. also found a significant relation between LN micrometastasis and poor prognosis in patients with T3 gastric cancer [9]. However, Fukagawa et al. showed that the presence of LN micrometastasis did not affect survival in a large number of patients with T2 gastric cancer [6]. Nevertheless, although immunohistochemical detection of micrometastasis has not spread worldwide because of the complexity of the immunohistochemical technique used in Japan, this parameter may be helpful for deciding treatment strategies for adjuvant chemotherapy.

Extranodal metastasis, comprising cancer cells in soft tissue discontinuous with the primary lesion, is found during routine examination of about 10–28% of resected gastric carcinoma specimens [118]. According to the UICC, this type of tumor spread should be regarded as LN metastasis if the nodule has the form and smooth contour of a LN, but should otherwise be regarded as part of the primary tumor [107]. Some studies have, however, suggested that such tumor extension represents peritoneal seeding from either the primary tumor or metastatic LNs. Etoh et al. reported that extranodal metastasis was closely related to a poor prognosis [119]. Moreover, Nakamura et al. described that classification of patients into a capsule rupture group or no capsule rupture group, on the basis of the status of extranodal spread, was important [120]. These reports support the notion that extranodal metastasis should be included in the clinical classification of gastric cancer.