The authors have declared that no competing interests exist.
Conceived and designed the experiments: SC BWZ YBC. Performed the experiments: SC BWZ YFL XYF XWS WL ZWZ YQZ YBC. Analyzed the data: SC BWZ YFL CNQ. Contributed reagents/materials/analysis tools: SC BWZ YBC. Wrote the paper: SC.
To investigate whether the recommendation to remove 15 lymph nodes that is used in the staging system is necessary to assess gastric cancer progression and to evaluate whether our metastatic lymph node ratio dividing method, adapted from the AJCC’s (American Joint Committee on Cancer) 7^{th} TNM staging system, is helpful for the patients with fewer than 15 harvested lymph nodes.
We performed a retrospective study of 1101 patients with histologically diagnosed gastric cancer who underwent a D2 gastrectomy at the Sun Yatsen University Cancer Center between January 2001 and December 2010. The Kappa and Chisquared tests were employed to compare the clinicopathological variables. The KaplanMeier method and Cox regression were employed for the univariate and multivariate survival analyses.
In the trial, 346, 601 and 154 patients had 0–14, 15–30 and more than 30 lymph nodes harvested, respectively. The median survival times of patients with different lymph nodes harvested in N0, N1, N2 and N3a groups were 45.43, 54.28 and 66.95 months (p = 0.068); 49.22, 44.25 and 56.72 months (p<0.001), 43.94, 47.97 and 35.19 months (p = 0.042); 32.88, 42.76 and 23.50 months (p = 0.016). Dividing the patients who had fewer than 15 lymph nodes harvested by the metastatic lymph node ratio at 0, 0.13 and 0.40, the median survival times of these 4 groups were 70.6, 50.5, 53.5 and 30.7 months (p<0.001). After recategorising these 4 groups into the N0, N1, N2, N3a groups, the histological grade, T staging, premier N staging, and restaged N staging were the independent prognostic factors.
Large numbers of lymph nodes harvested in radical gastrectomy do not cause stage migration. For those patients with a small number of harvested lymph nodes, their stage should be divided by the metastatic lymph node ratio, referred to in the TNM staging system, to assign them an accurate stage.
Approximately one million people are diagnosed with gastric cancer each year, making it the fourth most common cancer type and the second leading cause of cancerrelated death worldwide, with an estimated 800,000 deaths caused by the disease
In 2010, the AJCC’s (American Joint Committee on Cancer) 7th edition TNM classification of malignant tumours for gastric cancer was published
In our study, we investigated whether recommended 15 lymph nodes for use in the TMN staging system is sufficient for evaluating gastric cancer and whether 30 lymph nodes would be more accurate. Additionally, we assessed whether our metastatic lymph node ratio dividing method, adapted from the AJCC’s 7^{th} TNM staging system, is helpful for the patients with fewer than 15 harvested lymph nodes.
All patients were provided written informed consent for their information to be stored in the hospital database. Study approval was obtained from the independent ethics committees at Cancer Center of Sun YatSen University. The study was undertaken in accordance with the ethical standards of the World Medical Association Declaration of Helsinki.
The eligibility requirements included the following: (1) Patients had gastric carcinoma identified by histopathological examination, (2) underwent gastrectomy, (3) presented absence of identifiable distant metastasis, such as liver, lung and distal lymph nodes, (4) presented no history of another synchronous malignancy, (5) presented no recurrent gastric cancer or remnants of gastric cancer, (6) received no neoadjuvant therapy, (7) survived in the perioperative period and (7) had complete followup data collection. The procedures of tumour resection and the D2 lymphadenectomy performed by experienced surgeons were similar in all patients undergoing radical resection.
According to the guidelines of the Japanese Gastric Cancer Association (JGCA), the stomach was divided anatomically into upper, middle and lower portions. The three portions were defined by subdividing both the lesser and greater curvatures into three equal lengths
It is well known that in the AJCC’s TNM staging system, the cutoff values to divide the patients into different N groups are 0, 2 and 6 regional lymph nodes. Thus, we evaluated the N staging of patients by the metastatic lymph node ratio with 0, 0.13 (2/15), and 0.40 (6/15) as our cutoff values. According to the metastatic lymph node ratio, we separated the patients who had fewer than 15 lymph nodes harvested into 4 groups [0, 0–0.13 (2/15), 0.13–0.40 (6/15) and greater than 0.40] and recategorised these 4 groups into the N0, N1, N2, and N3a groups. These groups were then combined with other patients who had greater than 15 lymph nodes harvested to create a revised N staging.
We included 1,101 patients who underwent gastrectomy at the Sun Yatsen University Cancer Center between January 2001 and December 2010. The postoperative pathological results include tumour size, histological type, margin, adjacent tissues and neighbouring organs, retrieved lymph nodes, metastatic lymph nodes, and pTNM staging. The eligibility criteria include histologically confirmed R0 resection, which was defined as no residual macroscopic or microscopic tumour. Patients with distant metastases or carcinoma of the gastric stump after a gastric resection for benign disease were excluded from the study.
After treatment, patients with advanced gastric cancer were monitored every 2–3 weeks for six months postoperatively and then every 3 months for the first 2 years. Patients with early stage gastric cancer were required to have a further consultation with the doctor every 3 months for the first 2 years. All of the patients were monitored every 6 months thereafter. Telephone calls and letters were used to assess patients who could not be physically present for followup. Complete data were collected from all 1101 patients from the time following treatment until July 2011. The followup period ranged from 6 to 120 months (median, 41 months).
The Kappa and Chisquared tests were used to compare the clinicopathological variables between the groups with different numbers of harvested lymph nodes. Univariate survival analysis was performed using the KaplanMeier method. Survival curves were compared with the logrank test. Multivariate statistical survival analysis was performed using the Cox regression. Analyses were performed with the SPSS software version 20.0 for Windows (SPSS, Inc., Chicago, IL). Statistical significance was defined as P<0.05.
In total, 2 patients died in the perioperative period, secondary to anastomotic leakage and renal failure. There were another 94 patients excluded from our study because of incomplete followup data. The total number of patients included in our study is 1101. The median age of them was 59 years old (range: 18–83). Of these patients, 752 were male and 349 were female. The 5year survival of the whole patient group was 41.0%, with a median survival of 61.2 months. The patient clinicopathological characteristics are presented in
Clinical pathological data  Fewer than 15 harvested lymph nodes (n = 346 cases)  Harvested lymph nodes (15–29) (n = 601 cases)  Greater than 30 harvested lymph nodes (n = 154 cases)  
Cases  %  Cases  %  Cases  %  P value  
Age (Years)  Median  61  57  58  0.316  
Range  28–83  18–82  22–76  
Sex  Male  245  70.8  401  66.7  106  68.8  
Female  101  29.2  200  33.3  48  31.2  0.424  
Tumour location  Gastric cardia  211  61.0  232  38.6  42  27.3  
Middle  29  8.4  115  19.1  31  20.1  
Antrum  84  24.3  234  38.9  75  48.7  
Total stomach  2  0.6  7  1.2  4  2.6  
Remnant stomach 
20  5.8  13  2.2  2  1.3  <0.001  
Surgery  Proximal gastrectomy  218  63.0  219  36.4  42  27.3  
Distal gastrectomy  93  26.9  293  48.8  85  55.2  
Total gastrectomy  35  10.1  89  14.8  27  17.5  <0.001  
Tumour size  <3 cm  52  15.0  101  16.8  14  9.1  
≥3 cm  294  85.0  500  83.2  140  90.9  0.059  
Borrmann type  I  12  3.8  18  3.4  2  1.3  
II  184  57.5  286  53.4  60  39.5  
III  107  33.4  210  39.2  77  50.7  
IV  17  5.3  22  4.1  13  8.6  0.002  
Histological grade  Highdifferentiation  4  1.2  6  1.0  0  0  
Mediandifferentiation  111  32.1  147  24.5  23  14.9  
Lowdifferentiation  164  47.4  330  54.9  102  66.2  
Poordifferentiation 
67  19.4  118  19.6  29  18.8  0.001  
T staging 
T1  26  7.5  65  10.8  2  1.3  
T2  16  4.6  45  7.5  4  2.6  
T3  9  2.6  32  5.3  0  0.0  
T4a  233  67.3  348  57.9  133  86.4  
T4b  62  17.9  111  18.5  15  9.7  <0.001  
N staging 
N0  133  38.4  185  30.8  33  21.4  
N1  79  22.8  119  19.8  21  13.6  
N2  94  27.2  87  14.5  43  27.9  
N3a  40  11.6  156  26.0  37  24.0  
N3b  0  0.0  54  9.0  20  13.0  <0.001  
Adjuvant Chemotherapy  Yes  254  73.4  442  73.5  121  78.6  
No  92  26.6  159  26.5  33  21.4  0.410 
These 35 remnant stomach patients all had a more than 5 years history of the gastrectomy because of the gastric ulcer.
Poorlydifferentiated cells: signet ring cell carcinoma, mucinous adenocarcinoma, undifferentiated carcinoma, etc.
The T and N stagings for this group of patients are defined according to the AJCC’s 7^{th} TNM staging system for gastric cancer.
There were 346, 601 and 154 patients who had 0–14, 15–30 and more than 30 lymph nodes harvested, respectively. Their median survival times were 44.94, 46.62 and 42.82 months, respectively (p = 0.003). The results are shown in
A The median survival times of patients who had 0–14, 15–30 and more than 30 lymph nodes harvested in the study were 44.94, 46.62 and 42.82 months, respectively (p = 0.003). B The median survival times of patients who had 0–14, 15–30 and more than 30 lymph nodes harvested in the N0 group were 45.43, 54.28 and 66.95 months, respectively (p = 0.068). C The median survival times of patients who had 0–14, 15–30 and more than 30 lymph nodes harvested in the N1 group were 49.22, 44.25 and 56.72 months, respectively (p<0.001). D The median survival times of patients who had 0–14, 15–30 and more than 30 lymph nodes harvested in the N2 group were 43.94, 47.97 and 35.19 months, respectively (p = 0.042). E The median survival times of patients who had 0–14, 15–30 and more than 30 lymph nodes harvested in the N3a group were 32.88, 42.76 and 23.50 months, respectively (p = 0.016). F The median survival times of patients who had 15–30 and more than 30 lymph nodes harvested in the N3b group were 22.48 and 36.0 months, respectively (p = 0.199).
In the N0 group, there were 133, 186 and 32 patients with 0–14, 15–30 and more than 30 lymph nodes harvested, respectively. Their median survival times were 45.43, 54.28 and 66.95 months, respectively (p = 0.068). These results are shown in
In the N1 group, there were 79, 119 and 21 patients with 0–14, 15–30 and more than 30 lymph nodes harvested, respectively. Their median survival times were 49.22, 44.25 and 56.72 months, respectively (p<0.001). These results are shown in
In the N2 group, there were 94, 87 and 43 patients with 0–14, 15–30 and more than 30 lymph nodes harvested, respectively. Their median survival times were 43.94, 47.97 and 35.19 months, respectively (p = 0.042). These results are shown in
In the N3a group, there were 40, 156 and 37 patients with 0–14, 15–30 and more than 30 lymph nodes harvested, respectively. Their median survival times were 32.88, 42.76 and 23.50 months, respectively (p = 0.016). These results are shown in
In the N3b group, there were 53 and 21 patients with 15–30 and more than 30 lymph nodes harvested, respectively. Their median survival times were 22.48 and 36.0 months, respectively (p = 0.199). The results are shown in
We categorised the patients in this study by their metastatic lymph node ratios: 0, 0–0.13 (2/15), 0.13–0.40 (6/15) and greater than 0.40; there were 351, 195, 270 and 285 patients in these 4 groups, respectively. Their median survival times were 75.0, 62.5, 51.4 and 31.6 months, respectively (p<0.001). The survival curve is shown in
A The median survival times of patients who had metastatic lymph node ratios of 0, 0–0.13 (2/15), 0.13–0.40 (6/15) and more than 0.40 in the study were 75.0, 62.5, 51.4 and 31.6 months, respectively (p<0.001). B The median survival times of patients who had metastatic lymph node ratios of 0, 0–0.13 (2/15), 0.13–0.40 (6/15) and more than 0.40 in the group with fewer than 15 harvested lymph nodes group were 70.6, 50.5, 53.5 and 30.7 months, respectively (p<0.001). C The median survival times of patients in the N1, N2 and N3a groups, those patients who had at least one metastatic lymph node in the fewer than 15 nodes that were harvested, were 46.2, 41.8 and 30.9 months, respectively (p = 0.131). D The median survival times of patients whose metastatic lymph node ratios were 0, 0–0.13 (2/15), 0.13–0.40 (6/15) and more than 0.40 in the group who had at least one metastatic lymph node in the fewer than 15 nodes that were harvested were 53.5, 53.5 and 30.7 months, respectively (p<0.001).
We focused on the patients who had fewer than 15 lymph nodes harvested and were found to have at least one metastatic lymph node. There were 213 patients in this category. According to the AJCC’s TNM staging system, there were 79, 94 and 40 patients in N1, N2 and N3a groups, respectively; their median survival times were 46.2, 41.8 and 30.9 months, respectively (p = 0.131). A survival curve is shown in
According to the metastatic lymph node ratio, we separated the patients who had fewer than 15 lymph nodes harvested into 4 groups [0, 0–0.13 (2/15), 0.13–0.40 (6/15) and more than 0.40], and recategorised these patients into the N0, N1, N2 and N3a groups. There were 351, 161, 219, 296 and 74 patients in the N0, N1, N2, N3a and N3b groups, respectively. Their median survival times were 74.6, 64.9, 54.5 37.7 and 31.6 months, respectively (p<0.001). A survival curve is shown in
In the univariate analysis, the tumour size, tumour position, histological grade, Borrmann type, T staging, premier N staging, and restaged N staging significantly correlated with the overall survival (
Variables  n  5year survival rate %  Median survival (months)  P value 



<0.001  
Gastric cardia  485  40  60.1  
Middle  175  27  48.6  
Antrum  393  51  60.3  
Total stomach  13  32  41.6  
Remnant stomach  35  0  37.4  

<0.001  
<3 cm  167  64  67.4  
≥3 cm  934  37  58.9  

<0.001  
I  32  44  53.4  
II  530  38  63.6  
III  394  36  49.8  
IV  52  18  32.4  
0.007  
Highdifferentiation  10  90  76.4  
Mediandifferentiation  281  50  58.4  
Lowdifferentiation  596  36  60.9  
Poordifferentiation  214  33  52.4  

<0.001  
T1  93  93  80.1  
T2  65  77  65.9  
T3  41  77  70.3  
T4a  714  35  58.9  
T4b  188  21  39.6  

<0.001  
N0  351  49  64.5  
N1  219  53  75.7  
N2  224  47  48.5  
N3a  233  29  39.9  
N3b  74  19  31.5  

0.003  
0–14  346  33  44.1  
15–29  601  38  64.0  
30–  154  38  50.6  

<0.001  
N0  351  49  74.6  
N1  161  57  64.9  
N2  219  53  54.5  
N3a  296  27  37.7  
N3b  74  19  31.6 
Lymph node metastasis is one of the most important prognostic factors in gastric cancer. The AJCC’s 7^{th} TNM staging system for gastric cancer is commonly used in the clinic. However, this system recommends the dissection of more than 15 lymph nodes for N staging, except for the N0 patients. It is unclear whether the 15 nodes can be a standard for the D2 radical gastrectomy, or whether 30 would be better. Another argument is the loose definition of the D2 dissection
Variable  HR  95% CI  P value 
OS in gastric cancer patients  
Tumour location  0.976  0.889–1.071  0.608 
Tumour size  1.226  0.863–1.742  0.256 
Borrmann type  1.103  0.943–1.289  0.220 
Histological grade  1.148  0.999–1.320  0.051 
T staging  1.501  1.285–1.753  <0.001 
N staging  1.363  1.257–1.479  <0.001 
Lymph nodes harvested  0.724  0.622–0.843  <0.001 
In our study, there were 346 patients with fewer than 15 lymph nodes harvested and 154 patients with more than 30 lymph nodes harvested. Some investigators have found that if the number of harvested lymph nodes is smaller, downmigration of the N stage may occur and conversely if the number is larger, upmigration of the N stage may occur
Variable  HR  95% CI  P value 


Tumour location  0.992  0.904–1.089  0.868 
Tumour size  1.218  0.858–1.730  0.270 
Borrmann type  1.115  0.954–1.303  0.171 
Histological grade  1.153  1.004–1.324  0.044 
T staging  1.491  1.276–1.741  <0.001 
N staging  0.696  0.525–0.924  0.012 
Lymph nodes harvested  0.853  0.721–1.008  0.063 

1.960  1.495–2.571  <0.001 
The other issue is how to stage gastric cancer patients with fewer than 15 lymph nodes harvested, especially patients with lymph node metastases. Empirical treatments from different oncologists make a substantial difference on the therapy and prognosis of those patients. The lymph node metastasis ratio is widely reported as an independent prognostic factor in gastric cancer
We found that the number of lymph nodes harvested was not an independent predictor. The reason may be that the predictive function of the lymph nodes harvested is covered by the recategorised N staging in the multivariate analysis. If the recategorised N staging is removed from the Cox’s regression analysis, the number of lymph nodes harvested is an independent prognostic factor. This also demonstrates the advantage of the recategorised N staging system compared to the original N staging system. The function of the original N staging is preserved in the recategorised N staging.
As a retrospective study, there were confounding factors that influenced the statistical analyses and conclusions. Lymph node metastasis is one of the most important prognostic factors for patients with gastric cancer. The staging of the disease may be improved by identifying micrometastases in the lymph nodes and identifying extranodal metastasis of the disease. Although our study showed that the recategorised N staging system is more accurate than the traditional N staging system, further prospective studies would provide additional evidence supporting the use of our recategorised N staging system and metastatic lymph node ratio as a standard for the N staging of gastric cancer.
Harvesting a large number of lymph nodes (more than 30) in radical gastrectomy would not cause Nstage migration. Additionally, for those patients who had fewer than 15 lymph nodes harvested, accurate staging would best be accomplished by dividing by the metastatic lymph node ratio that is mentioned in the TNM staging system.
We thank Dr. JiaFeng Fang of the Third Affiliated Hospital, Sun Yatsen University and Chen Chen from the National University of Singapore for their critical reading.