D2 lymphadenectomy without routine splenectomy and pancreatic tail resection is now becoming the standard operation method for advanced resectable gastric cancer. The Dutch trial did not support the benefit of D2 lymphadenectomy as the complication rate and the postoperative death was significantly high in D2 lymphadenectomy group compared to D1 lymphadenectomy group [1] . However, as the long term survival analysis of this trial has proven that the locoregional recurrence and gastric cancer related death rate were lower in D2 lymphadenectomy group, the author recommended D2 lymphadenectomy as a standard procedure at a high volume center for advanced resectable gastric cancer [2] . The result of JCOG9501 has demonstrated that extended D3 lymphadenectomy did not improve survival compared to D2 lymphadenectomy [3] . Additionally, as the removal of No.10 and 11 lymph nodes by splenectomy showed no survival benefit, D2 lymphadenectomy without routine splenectomy and pancreatic tail resection in experienced hands is considered to be the standard for advanced resectable gastric cancer, both in Asian and in Western patients [4] [5] .

On the other hand, limited surgery and less invasive surgery has become a standard procedure for early gastric cancer (EGC) recently. In the treatment guidelines for gastric cancer in Japan (3rd edition, 2010) [6] , the indication of lymphadenectomy for EGC was defined as in Table 1. According to this guideline, D2 lymphadenectomy is required when lymph node metastasis was suspected by preoperative examination. However, Yoshikawa T et al. claims that D2 lymphadenectomy should be limited for EGC as there was little survival benefit [7] . The problems to solve are as follows. First, as the target is “early”, the curability should not be ruined. Second, which type of tumor can be the indication of limited surgery? Third, which type of tumor requires D2 lymphadenectomy? Forth, which type of tumor cannot be cured by surgery alone?

We, herein, report an analysis of 1141 EGC cases and reached to one direction that may help to solve these problems.

Surgically resected consecutive cases from January 1991 through December 2013 in our institution were analyzed. The size, macroscopic type, histological type, presence of ulcer, depth of wall invasion, lymphovascular invasion, operation method, metastasis of resected lymph node and their number are described according to the Japanese classification of gastric carcinoma. The lymph node number according to the Japanese classification of gastric carcinoma and the definition of lymphadenectomy according to the treatment guidelines for gastric cancer in Japan (3rd edition), namely D1, D1+ and D2 for each operative method are shown in Figure 1. Relation between the characteristics of the tumor and the cancer positive lymph node number was evaluated.

The dilemma in treating EGC is how to take balance between curability and invasiveness of the treatment strategy. There is no question that endoscopic mucosal resection (EMR) and endoscopic submucosal dissection (ESD) are less invasive therapy. However, the indication is basically limited to T1a cancers. The standard procedure for EGC that is out of indication of EMR/ESD will be surgical resection with lymphadenectomy as the possibility of lymph node metastasis rises. Nevertheless, as the frequency of lymph node metastasis of EGC is low, it has to say that unnecessary lymphadenectomy was performed in many cases. Sentinel node navigation surgery has been expected as an effective therapy to reduce such unnecessary lymphadenectomy. However, this method still requires further research and has difficulty in avoiding false negative judgement. It needs more time to become a standard procedure [8] [9] . Thus, at present, we have to determine the range of lymphadenectomy by the information obtained from the preoperative examination.

In this study, the cancer positive lymph nodes of T1a were all within D1 range. This means, D1 lymphadenectomy is enough for T1a cancer that is out of indication of EMR/ESD. This meets the criterion of the treatment guidelines for gastric cancer in Japan (3rd edition). However, most T1a cancer were negative for lymph node metastasis (667 out of 678: 98.4% in this study). Further studies are necessary to determine the condition for N0. For 82 cases of lymph node metastasis positive T1b cases, 77 cases (93.9%) were within D1+ range. One case did not meet the criterion of the guideline. This case was differentiated type, which size was smaller than 1.5 cm, but had cancer positive lymph node at No.8a which is out of D1 range (within D1+ range). Can we ignore this case? The operation of this case was distal gastrectomy and the difference between D1 and D1+ is the resection of No.8a and 9. Concerning surgical technique, as the difference between D1 and D1+ is not significant, we conclude that D1+ is reasonable for T1b cancer.

There were 5 cases (1.1%) which had cancer positive lymph node further than D1+. Two cases had No.12a metastasis. These cases also had multiple cancer positive lymph nodes at No.3. Thus, if the cancer is adjacent to the lesser curvature, undifferentiated type, and is larger than 4 cm in size, surgeons should examine the No.3 lymph nodes by pathological diagnosis during surgery. If No.3 lymph node metastasis is positive, No.12a lymph node should be resected to achieve D2. These cases require D2 lymphadenectomy. However, though one of these cases earned long term recurrence free survival (19 years after surgery), in another case life threatening lymph node recurrence occurred after one year. The question, whether D2 lymphadenectomy is necessary or not, still remains. Three cases had cancer positive lymph node further than D2. One case had No.16 lymph node metastasis. This case was suspected for No.16 lymph node metastasis by preoperative examination and was confirmed by resection of this lymph node. Multiple bone metastasis appeared, and this patient died after 4 years. Similar to the StageIV case in this study, there are some cases that cannot be cured by surgery alone among T1b cases. Two cases had lymph node metastasis at No.14v. In both of these cases, the tumor was at the L area and adjacent to the greater curvature and they had multiple cancer positive No.6 lymph nodes. Thus, these cases require No.6 lymph node by pathological diagnosis during surgery. When No.6 lymph node is cancer positive, No.14v resection should be performed. However, multiple bone metastasis appeared 5 years after operation in one case, and another case died of chronic heart failure and hemiplegia after cerebral infarction. The effect of lymphadenectomy further than D2 is not clear. However, as nearly 5 year survival was obtained in one case, there is a possibility that No.14v resection might be effective.

Considering the results of this study, we concluded as follow. D1 is enough for T1a cancer which is out of indication of EMR/ESD. D1+ is reasonable for T1b cancer. These are good indication for laparoscopic gastrectomy. D2 is required when the tumor exists adjacent to the lesser curvature, is undifferentiated type, is larger than 4 cm in size. Pathological diagnosis of No.3 lymph node should be performed during surgery in this case. No.14v resection might be effective in case, when the tumor is adjacent to the greater curvature of L area and also when No.6 lymph node metastasis is confirmed during surgery.