Gastric cancer, one of the most common malignancies in the world, frequently reveals lymph node, peritoneum, and liver metastases. Most of gastric cancer patients present with lymph node metastasis when they were initially diagnosed or underwent surgical resection, which results in poor prognosis. Both the depth of tumor invasion and lymph node involvement are considered as the most important prognostic predictors of gastric cancer. Although extended lymphadenectomy was not considered a survival benefit procedure and was reported to be associated with high mortality and morbidity in two randomized controlled European trials, it showed significant superiority in terms of lower locoregional recurrence and disease related deaths compared to limited lymphadenectomy in a 15-year follow-up study. Almost all clinical investigators have reached a consensus that the predictive efficiency of the number of metastatic lymph nodes is far better than the extent of lymph node metastasis for the prognosis of gastric cancer worldwide, but other nodal metastatic classifications of gastric cancer have been proposed as alternatives to the number of metastatic lymph nodes for improving the predictive efficiency for patient prognosis. It is still controversial over whether the ratio between metastatic and examined lymph nodes is superior to the number of metastatic lymph nodes in prognostic evaluation of gastric cancer. Besides, the negative lymph node count has been increasingly recognized to be an important factor significantly associated with prognosis of gastric cancer.

More than half of gastric cancer patients have lymph node metastasis when they are initially diagnosed or underwent surgical resection, which results in poor prognosis[1-4]. Lymph node involvement is the most important indictor for overall survival (OS) of gastric cancer patients following curative resection (R0), and the survival rates markedly decrease with the increase in the number of metastatic lymph nodes[5-10]. Furthermore, many investigators demonstrated that lymph node metastasis was an independent risk factor for gastric cancer recurrence in patients following curative resection[11,12]. The overall survival of lymph node-negative gastric cancer patients was significantly longer than that of lymph node-positive patients, and the overall recurrence rate in lymph node-negative gastric cancer patients was significant lower than that in lymph node-positive patients[13-16].

It is certain that a meticulous and deliberate classification of lymph node metastasis can provide the guarantee of accurately evaluating the prognosis of gastric cancer patients. To evaluate lymph node metastasis exactly, two main classifications of metastatic lymph nodes have been recommended to use in clinic by many investigators. The first classification was originated from Japan in 1980s, which defined the status of lymph node metastasis by the location of metastatic lymph nodes relative to the primary tumor according to the Japanese Classification of Gastric Carcinoma (JCGC)[17,18]. In 1997, the International Union Against Cancer (UICC) proposed that lymph node involvement should be classified according to the number of metastatic lymph nodes[19-21]. Many researchers reported that the classification based on the number of metastatic lymph nodes was more sensitive in prognostic evaluation of gastric cancer than that based on the location of metastatic lymph nodes[22-25]. In addition, the ratio between metastatic and dissected lymph nodes was deemed as an important prognostic indictor of gastric cancer after surgery[26]. So far, there has not been a universally accepted classification of metastatic lymph nodes for evaluating the overall survival of gastric cancer worldwide.

In view of the fact that regional lymph node metastasis usually begins in early period of gastric cancer, regional lymph node dissection should be recommended as part of curative resection. However, it is still debatable over the optimal extent of lymph node dissection during the curative resection for gastric cancer. Limited lymphadenectomy (D1 according to JCGC[23]) is limited to removing the perigastric lymph nodes, which can not guarantee accurate staging of lymph node metastasis and avoid potentially rudimental metastatic lymph nodes[27]. When limited lymphadenectomy is performed, no information is obtained on lymph nodes other than the perigastric ones. Conversely, if only the number of metastatic lymph nodes is used as a criterion, extended lymphadenectomy (D2 or D3 according to JCGC) could result in stage migration[28,29]. Better survival rates in extended lymphadenectomy may result from greater staging accuracy and improvement of stage-specific survival rates owing to stage migration. Although two randomized controlled European trials that compared limited lymphadenectomy with extended lymphadenectomy failed to show a survival benefit from extended lymphadenectomy[30], the lack of experience with the surgical procedure and with postoperative care were most obvious reasons to account for the poor outcome of extended lymphadenectomy[31,32]. Recently, researchers demonstrated that extended lymphadenectomy showed significant superiority in terms of lower locoregional recurrence and disease related deaths to limited lymphadenectomy in a 15-year follow-up study[33]. However, the rationale for extended lymphadenectomy as the conventional procedure of curative gastrectomy has not been elucidated clearly.

In this review, we will discuss the above-mentioned issues regarding lymph node metastasis in gastric cancer in detail.

In gastric cancer, the presence or absence of lymph node metastasis is one of the most important prognostic indicators in patients following curative resection. More than 50% of gastric cancer patients have lymph node metastases at diagnosis, which lead to a 5-year survival rate < 30%[4]. Although several investigators reported that the minority of lymph node-negative gastric cancer patients had recurrence and poor survival, most investigators demonstrated that the prognosis of lymph node-negative gastric cancer patients was significantly better than that of lymph node-positive patients[34-36]. Hocbwald et al[14] analyzed the data of actual five-year survivors of gastric cancer, and found that the nodal status was the most powerful prognostic factor of outcome within the first five years after curative surgery for gastric cancer. They also demonstrated that the number of positive lymph nodes was the most important determinant of survival (P < 0.001) by using multivariate analysis. Ichikura et al[37] demonstrated that the median postoperative survival of patients with < 4 metastatic lymph nodes was significantly longer than that of patients with 4 or more metastatic lymph nodes. Gunji et al[38] also found that early gastric cancer patients with 4 or more metastatic lymph nodes had a higher probability of recurrence and shorter short-term survival. In a previous study, we also demonstrated that gastric cancer patients with positive lymph nodes had much shorter median OS than those with negative lymph nodes[39].

Recurrence is the most common reason of cancer related death in gastric cancer patients. Owing to the fact that more than half of gastric cancers was advanced at the time of diagnosis, even when the curative resection (R0) was possible, the recurrence could occur in approximately 60% of patients[40]. The recurrence of surgically resectable gastric cancer is influenced significantly by the presence of lymph node metastasis. The more the number of metastatic lymph nodes, the worse the prognosis of gastric cancer patients[41]. Besides, we found that the number of metastatic lymph nodes was the most important indicator of recurrence after curative surgery[42]. In a further analysis, we demonstrated that the locoregional recurrence of gastric cancer was significantly associated with the number of metastatic lymph nodes (P = 0.025)[43]. Hepatic metastasis is the most frequent distant metastasis of gastric cancer following curative resection. The detail mechanism of hepatic metastasis from gastric cancer after curative resection is unclear. Although the hematogenous metastasis of gastric cancer was deemed as the most important route of hepatic metastasis following curative resection[44,45], some researchers demonstrated that lymphatic metastasis should be another reason of hepatic metastasis[46]. In our previous study, we demonstrated that lymph node metastasis was not only the most important factor of hepatic metastasis of gastric cancer following curative resection, but also the independent factor of the interval time between radical gastrectomy and hepatic metastasis[47]. Unlike in the West, peritoneal dissemination is the main pattern of recurrence after curative gastrectomy in the East[48]. We also demonstrated that peritoneal dissemination was the most usual recurrence type of gastric cancer after surgery[43]. Meanwhile, we found that the number of metastatic lymph nodes was significantly associated with the occurrence of peritoneal dissemination.

The first classification of metastatic lymph nodes was originated from Japan in 1980s, which defined the status of lymph node metastasis by the location of positive lymph nodes relative to the primary tumor according to the JCGC. Once extragastric lymph node metastasis is identified, the probability of systemic cancer dissemination significantly increases in theory. In our latest study, we demonstrated that extragastric lymph node was an important factor associated with the dismal prognosis of patients, even those undergoing extended lymphadenectomy[49]. However, many researchers reported that classification based on the number of metastatic lymph nodes was more sensitive in prognostic evaluation of gastric cancer than that based on the location of metastatic lymph nodes. Although the JCGC classification of lymph node metastasis can reflect the metastatic pathways of cancer cells from the primary tumor, the non-quantitative characteristics of this classification lead to inefficient evaluation of the prognosis of gastric cancer patients. As we know, comparatively accurate results of data statistics should be based on the continuous data or the elaborate classification of non-continuous data. Therefore, the UICC has still maintained the classification of lymph node metastasis based on the number of lymph node metastasis (the 5^th/6^th N stage) to provide the accurate prognostic evaluation for gastric cancer patients since 1997. However, it is still controversial over the best cutoffs of the number of metastatic lymph nodes for prognostic evaluation. So many researchers proposed that the N staging should be updated for improvement of its accuracy for prognostic evaluation in gastric cancer in clinical application[41,50]. Although the 5^th/6^th edition UICC TNM node staging system is deemed to have higher feasibility, objectivity, reproducibility, and increased strength of prognostic stratification than other edition node staging systems, it has resulted in inevitable controversies ultimately. There has not been a universally accepted node staging system for extremely precise evaluation of the relationship between the 5^th/6^th edition UICC N stage and prognosis of gastric cancer patients[50-52]. In 2010, we identified that the 7^th edition UICC N stage was superior to the 5^th/6^th UICC N stage and Japanese n stage for prognostic evaluation in gastric cancer patients[53]. Subsequently, many researchers reported that the 7^th edition UICC N stage could provide a more stratified survival difference in sub-staged gastric cancer than the 5^th/6^th edition UICC N stage, which should be considered to be much more reasonable compared with the previous edition N stage, especially between the N1- and N2-stage tumors[54-57]. On the other hand, a few authors did not recommend that the 7^th edition UICC N stage as the optimal classification of lymph node metastasis in gastric cancer[58,59].

Many studies indicated that the ratio between metastatic and dissected lymph nodes, namely the number of metastatic lymph nodes to that of dissected lymph nodes, was an independent prognostic factor for gastric cancer and other malignant neoplasms, and was reported to be able to reduce the phenomenon of stage migration[60,61]. Nevertheless, the prognostic superiority of the ratio has been still controversial for many years[62]. In the previous study, we demonstrated that the ratio between metastatic and dissected lymph nodes was inferior to the N stage for predicting the OS of gastric cancer patients with 15 or more dissected lymph nodes after curative resection[50]. Kulig et al[63] demonstrated that the ratio between metastatic and dissected lymph nodes could not be regarded as a standard classification of lymph node metastasis alternative to other classifications after curative gastrectomy plus extended lymphadenectomy. Actually, the ratio between metastatic and dissected lymph nodes can not define the number of examined lymph nodes and formulate the extent of lymph node dissection, which is considered the potentially key factor interfering with the quality of curative surgery and the prognosis of patients. We also demonstrated that the ratio between metastatic and dissected lymph nodes could enhance the prognostic evaluation accuracy of the N stage, although it was identified to be inferior to the N stage for accurate evaluation of patient prognosis[42].

Lymph node metastasis can occur during the early stages of gastric cancer, and regional lymphadenectomy is recommended as part of radical gastrectomy. However, the extent of lymphadenectomy to achieve the optimal result is controversial, and there is no worldwide consensus. Once extragastric lymph node metastasis of gastric cancer is identified[49], the probability of systemic dissemination of tumor cells will significantly increase. The optimal extent of lymph node dissection is an unavoidable problem which can interfere with the accurate evaluation of the extent of lymph node metastasis. Primary tumors were conventionally resected en bloc with limited or extended lymphadenectomy (D1 or D2-3 according to the JCGC). The limited lymphadenectomy (D1) entails the removal of the perigastric nodes only, whereas extended lymphadenectomy (D2 or D3) involves the removal of both perigastric and extragastric nodes. Extended lymphadenectomy can afford more exhaustive information of nodal metastasis than limited lymphadenectomy theoretically, while it has not reached a consensus of the standard extent of lymph node dissection for gastric cancer. Early studies have reported that 30%-40% of patients with metastatic lymph nodes including the second tier lymph nodes have survived longer than 5 years after extended lymphadenectomy[64]. Although extended lymphadenectomy was not considered a survival benefit procedure as it was reported in two randomized controlled European trials[65,66], the lack of experience with surgical procedure and with postoperative care was thought to account for the poor outcome of patients who underwent extended lymphadenectomy[67-69]. Ultimately, extended lymphadenectomy showed significant superiority in terms of lower locoregional recurrence and disease related deaths to limited lymphadenectomy in the 15-year follow-up study[33]. Recently, many Western clinical researchers also demonstrated that extended lymphadenectomy could improve survival and decrease the perioperative morbidity of gastric cancer patients[70-72].

Nodal metastatic stage redefinition and local relapse decrease were deemed as the most important reasons for improvement in the prognosis of gastric cancer patients who underwent extended lymphadenectomy, even in patients with node-negative gastric cancer[73,74]. Theoretically, extended lymphadenectomy allows to harvest more lymph nodes and eliminate more broadly perigastric lymphatic tissues which were potentially invaded by tumor cells, compared with limited lymphadenectomy[75]. The extent of surgery will especially influence locoregional control. In a Dutch trial, locoregional recurrence was registered in 58% of cases in the limited lymphadenectomy group and in 45% of cases in the extended lymphadenectomy group. However, the Japan clinical oncology group 9206-1 trial demonstrated that local relapse rate was only < 1% following curative gastrectomy with extended lymphadenectomy[76]. Yoshikawa et al[77] analyzed 1041 early gastric cancer patients who underwent extended lymphadenectomy with curative intent, and found that 15 cases had died of recurrence (1.44%) at the last follow-up.

Although high morbidity and mortality rates have been reported in patients requiring extensive gastric resection[78], extended lymphadenectomy has a much lower morbidity and mortality in Japan than in the West, with a mortality rate being less than 3%[79]. Biffi et al[80] analyzed 250 gastric cancer patients in a single center and demonstrated that extended gastrectomy with spleen and pancreas routine preservation can be considered a safe treatment for gastric cancer in Western patients, at least in experienced centers. In the Japan clinical oncology group study 9501, the reported hospital mortality was only 0.80%, which is significantly lower compared with Western reports[81]. Recently, an Italy randomized clinical trial demonstrated that extended lymphadenectomy could be considered a safe option, with a mortality of only 3.0% in Western gastric cancer patients[82]. Wu et al[67] reported that extended lymphadenectomy was a sufficiently safe procedure for gastric cancer patients, even without hospital morbidity.

In 1997 the UICC/American Joint Committee on Cancer (AJCC) recommended, in the 5^th edition of the staging manual, that a minimum of 15 lymph nodes need to be assessed per patient[83]. However, the 7^th edition TNM classification of N stages of gastric cancer states that ‘‘histological examination of a regional lymphadenectomy specimen will ordinarily include 16 or more lymph nodes’’. In this classification, it is only a recommendation and no longer a prerequisite that there is to be no less than 16 dissected lymph nodes for adequate N stage classification. Of note, the 7^th edition N staging does specify that at least 16 metastatic lymph nodes are required to assign N3b stage to gastric cancer patients[84]. Studies have shown that stage migration occurs in patients with a lower number of lymph nodes examined, creating inaccuracies in survival prediction[85-87]. Smith et al[31] proposed that the total lymph node number analyzed is an important and powerful qualifier of staging information and survival evaluation for gastric cancer, regardless of the underlying mechanisms that influence this survival impact of lymph node counts. They demonstrated that the trend toward superior survival outcome could be followed to lymph node counts greater than 40, although there was no unambiguous cutoff point for numerical lymph node analysis after curative gastrectomy.

Theoretically, a consensus should be reached that inadequate lymph node assessment directly affects patient survival[88-90]. Stage migration was occasionally identified to arise in patients with even great number of dissected lymph nodes, although many researchers demonstrated that the effect of stage migration is most usually shown in patients with a small number of lymph nodes examined[31,91]. It is certain that lymph node counts reflect not only the actual number of lymph nodes removed intraoperatively but also (or especially) the number of lymph nodes identified and properly examined during macroscopic and microscopic pathologic analysis[92]. Several investigators reported that the more the number of lymph node dissection, the longer the disease-free survival (DFS) of gastric cancer patients[89,90]. Besides, there was a significantly negative correlation between the number of dissected lymph nodes and the local recurrence rate in patients after curative gastrectomy in above-mentioned studies. In the previous study, we have demonstrated that patients presented with no less than 15 dissected lymph nodes had comparatively longer median OS after recurrence, compared to those with less than 15 dissected lymph nodes[90]. Scartozzi et al[91] reported a remarkable decrease in local recurrence rate, from 23% in patients with less than 25 lymph nodes assessed to 4.7% in patients with no less than 25 lymph nodes assessed. They considered that abundant lymph node dissection should be associated with good locoregional control. Another example, in the 15-year analysis of the Dutch D1/D2 randomized controlled trial, local recurrence was 22% in the D1 group, compared with 12% in the D2 group; while regional recurrence was 19% in the D1 group, and 13% in the D2 group (P = 0.015)[33]. In 2012, Xu et al[93] demonstrated that it is necessary to examine at least 16 lymph nodes for accurate pathological examination of gastric cancer, even in node-negative gastric cancer patients.

Negative lymph node count has been fully given importance for its significant association with the prognosis of patients with a malignant disease[94-96]. It is undoubted that patients with negative node metastasis present with a higher 5-year survival rate (5-YSR) and lower recurrence rate than those with positive node metastasis after curative resection for gastric cancer[97]. Owing to the comparatively better bio-behaviors of tumor, gastric cancer patients who present with no nodal involvement can have a higher 5-YSR regardless of lymphadenectomy[98]. Besides, the negative lymph node count is significantly associated with the prognosis of patients. Theoretically, the negative lymph node count should be deemed as the important guarantee for real R0 resection of gastric cancer and significantly associated with the OS of gastric cancer patients owing to micro-metastasis in negative nodes which could not be identified by conventional pathological examination[99,100]. It is so crucial that the total harvested lymph nodes must comprise negative lymph nodes, which is the basic guarantee for really radical resection of gastric cancer without leaving over any potential metastasis theoretically. By increasing the number of negative lymph nodes evaluated, the chance of leaving the micro-metastasis within negative lymph nodes decreases.

Recent data demonstrate that the immunohistochemical evaluation of histologically negative lymph nodes detects micro-metastatic disease with varying frequencies even in early gastric cancer. Kim et al[99] demonstrated that the incidence of lymph node micro-metastasis in early gastric cancer patients with negative lymph nodes was less than 10%. Saito et al[100] demonstrated that the micrometastases within lymph nodes which were determined by using the immunohistochemical anti-cytokeratin antibody in node-negative early gastric cancer patients were closely associated with recurrence following curative gastrectomy. Endo et al[101] used anti-cytokeratin antibody to immunohistochemically detect nodal micrometastasis that was not identified by routine pathological examination in 162 patients with apparent node-negative submucosal gastric cancer. The micrometastasis was detected in 45 of 2048 nodes (2.2%), representing 31 of 162 patients (19%). Although the micro-metastasis and isolated tumor cells in negative lymph nodes are considered as the key factors that could lead to an adverse effect on the OS of patients, patients with negative lymph node metastasis who were identified to have isolated tumor cells or micro-metastasis did not demonstrate a significantly worse prognosis than those who did not have isolated tumor cells after curative gastrectomy with extended lymphadenectomy[102]. Harrison et al[103] demonstrated that T3N0M0 gastric cancer patients who underwent extended lymphadenectomy had significantly more negative lymph nodes than those who underwent limited lymphadenectomy, indicating that extended lymphadenectomy can improve the OS of T3N0M0 patients. This result is potentially associated with the elimination of micro-metastasis in negative lymph nodes.

In a previous study, we also found that the negative lymph node count was an independent predictor of OS in gastric cancer patients with perigastric node metastasis, as was the positive lymph node count[104]. Further analysis demonstrated that the negative node count was the most intensive predictor of OS in patients with perigastric lymph node involvement after extended lymphadenectomy. Ultimately, we deduced that extended lymphadenectomy improved the prognosis of patients with perigastric lymph node involvement possibly by dissecting more negative nodes, compared to the limited lymphadenectomy. Recently, the sentinel node navigation was demonstrated to be an important factor influencing the diagnosis and treatment of lymph node metastasis from gastric cancer, especially in early stage patients[105]. Clinical application of the sentinel node navigation in gastric cancer is still restricted to early gastric cancer patients, owing to the complexity of the lymphatic drainage pathways of the stomach. The results of two noted random control trails (JCOG0302 and SNNS) of the sentinel node navigation in Japan failed to demonstrate that the detection of sentinel lymph nodes could exactly predict the status of lymph node metastasis intraoperatively, which implied that lymphadenectomy should be irreplaceable for the elimination of nodal metastasis[106].

Lymph node metastasis, as one of the most intensively prognostic indictor of gastric cancer, needs to be further investigated to elucidate the optimal significance for improving the prognostic evaluation and clinical treatment. Although the number of metastatic lymph nodes is considered the most appropriate classification of nodal metastasis for evaluation of OS, the negative lymph node count is an important variable that can provide important information contributing to prognostic evaluation and clinical treatment of gastric cancer. In addition, the extent of lymphadenectomy should be deeply investigated by studying the basic metastatic mechanism of cancer cells.