Lymph node metastasis in early gastric cancer

Objective: to evaluate the incidence of lymph node metastasis in early gastric cancer, identifying risk factors for its development. Methods: we conducted a prospective study of patients with gastric cancer admitted to the Section of the Esophago-Gastric Surgery of the Surgery of Service HUCFF-UFRJ, from January 2006 to May 2012. Results: the rate of early gastric cancer was 16.3%. The incidence of nodal metastases was 30.8% and occurred more frequently in patients with tumors with involvement of the submucosa (42.9%), in those poorly differentiated (36.4%), in tumors larger than 2 cm (33.3%) and in type III ulcerated lesions (43.8%). Conclusion: the incidence of lymph node metastases in patients was very high and suggests that one should keep the radicality of resection in early gastric cancer, particularly in relation to D2 lymphadenectomy, recommended for advanced gastric cancer. Conservative resections, with lymphadenectomies smaller than D2, should be performed only in selected cases, well-studied as for the risk factors of lymph node metastasis. Despite the small number of cases did not permit to relate the rate of lymph node metastasis to the risk factors considered, we noted a strong tendency for the occurrence of these metastases in the poorly differentiated, type III, larger than 2 cm tumors, and in the Lauren diffuse types.


INTRODUCTION INTRODUCTION INTRODUCTION INTRODUCTION INTRODUCTION
L ymph node metastasis in gastric cancer is frequent and early, and depends on many variables, amongst which the depth of invasion of the gastric wall, the degree of tumor differentiation and tumor size [1][2][3][4] .Thus, early tumors may have incidence of lymph node metastases ranging from 0%, when restricted to the mucosa, well differentiated and less than 2cm in size, to more than 30%, when reaching the submucosa, being poorly differentiated and having more than 2cm diameter 5,6 .The knowledge of the parameters of risk for these metastases, as well as the use of diagnostic procedures able to identify them in the preoperative and intraoperative periods, made possible the realization of more conservative procedures, such as endoscopic resection and smaller gastrectomies with limited lymphadenectomy, in selected cases of early gastric cancer, thus individualizing surgery of gastric adenocarcinoma and reducing treatment complications, both early and late 7,8 .
The objective of this study was to evaluate the incidence of lymph node metastasis in early gastric cancer (EGC), identifying its risk factors for its development.

METHODS METHODS METHODS METHODS
We conducted a prospective study of patients with gastric cancer admitted to the Esophago-Gastric Section of the General Surgery Service of the Clementino Fraga Filho University Hospital, Federal University of Rio de Janeiro (HUCFF-UFRJ), from January 2006 to May 2012.All patients admitted with a diagnosis of gastric adenocarcinoma confirmed by histopathological examination of gastric tissue obtained by endoscopic biopsies fragment underwent clinical staging and imaging.Gastric cancer patients without criteria of inoperability or unresectability, confirmed by laparoscopy or laparotomy, were submitted to curative gastric resection with D2 lymphadenectomy.All patients were informed of the risks and benefits of the procedure and signed an informed consent form.
From 2008 on, with the introduction of endoscopic ultrasonography in HUCFF-UFRJ, patients with early gastric lesions suggestive of cancer at endoscopy underwent endoscopic ultrasound to confirm the degree of tumor penetration of the gastric wall and evaluation of suspect perigastric nodes.Those with echoendoscopic confirmation of early tumor without perigastric lymphadenopathy were operated and submitted to sentinel nodes research, which, when negative, provided the realization of smaller gastric resections with modified D1 lymphadenectomy, as recommended by the Japanese Society of Gastric Cancer 9,10 .Patients with endoscopic ultrasound doubtful as to the degree of penetration of the tumor and/or suspect perigastric lymph nodes were treated for advanced tumors.For all the patients operated the lymphadenectomy specimen was individualized as for the number of resected lymph nodes and the presence of nodal metastases.
The macroscopic appearance of the lesion was classified according to criteria of the Japanese Society of Gastric Cancer 11 in types I, IIa, IIb, IIc and III.Its dimensions and microscopic features, such as cellular differentiation and Lauren classification, were evaluated in surgical specimens.The parameters used for the risk of nodal metastases configuration were: tumor size, less than or equal to 2 cm versus larger than 2 cm; macroscopic classification: type I and II versu type III; degree of cell differentiation: well-or moderately differentiated versus poorly differentiated; Lauren classification: intestinal type versus diffuse type; degree of tumor penetration of the gastric wall: restricted to the mucosa (T1a) versus with involvement of the submucosa (T1b).
The staging of gastric cancer followed the criteria established by the American Joint Committee on Cancer (AJCC) in its seventh edition, 2010 12 .
The descriptive analysis of the observed data was presented in the form of tables, expressed by the frequency (n) and percentage (%) for categorical data and by median, minimum and maximum for numeric data.In order to check whether there was a significant association between the variables of the lesion with the presence of metastasis, we applied: the chi-square ( 2 ) or Fisher exact test for comparisons of categorical (qualitative) data and the Mann-Whitney test (nonparametric) to compare numerical data (lesion size in cm).
We applide the nonparametric method, because the size of the lesion did not show normal distribution (Gaussian distribution) due to the wide dispersion of the data and rejection of the hypothesis of normality according to the Kolmogorov-Smirnov test.The criterion for determining significance was set at 5%.

RESULTS RESULTS
During the study period, 160 patients with gastric cancer were admitted to staging.Of these, nine (5.6%) were considered inoperable, 12 (7.5%)unresectable at laparotomy, and 14 (8.8%) were submitted to palliative gastrojejunostomy due to the presence of pyloric obstruction.The remaining 125 (78.1%) underwent gastric resections, of which 83 (66.4%) with curative intent, with D2 lymphadenectomy, 30 (24%) with palliative resections for locally advanced disease without possibility of R0 resection, and 12 (9.6%)with atypical resections and gastric D1 modified lymphadenectomy (perigastric nodes plus 7 and/or 8 and/or 9 chains) (Table 1).Among the latter, 11 were patients with early tumors without evidence of lymph node metastasis at sentinel nodes research and one had tumor in situ.
The mean number of resected lymph nodes in D2 and modified D1 lymphadenectomies was 25.Of the patients who underwent gastrectomy with D2 lymphadenectomy, D1 (palliative resections) and modified D1, 116 could be properly staged.The incidence of lymph node metastases (N) and staging based on the degree of penetration of the tumor in the gastric wall (T) can be seen in Table 2.There was a significant association (p<0.0001) between the degree of penetration of the tumor in the gastric wall (T) and lymph node metastasis (N +).The list of patients with EGC and the tumors features, such as the degree of penetration of the gastric wall (T1a -restricted to the mucosa -or T1b -with submucosal involvement), the size of the lesion, macroscopic classification and Lauren, the degree of tumor differentiation and the presence of lymph node metastases are shown in Table 3.
The frequency of EGC was 16.3% (26 patients), 7.5% restricted to the mucosa (12 patients) and 8.8% with involvement of the submucosa (14 patients).The incidence of nodal metastases in patients with EGC was 30.8% and occurred more frequently in patients with tumors with involvement of the submucosa (42.9%), those moderately/poorly differentiated (38.9%), greater than 2cm (33.3%) and ulcerated type III (43.8%) (Tables 4 and 5).The relationship between the rates of lymph node metastasis and the variables of  the lesion, such as the degree of penetration of the gastric wall (T), lesion size, macroscopic classification, tumor differentiation grade and histological type of Lauren are shown in Table 6.There was no significant association between the variables of the lesion with the presence of metastases due to the small number of cases studied.There was, however, a trend toward an increased risk of metastasis in tumors larger than 2cm, poorly differentiated, diffuse type of Lauren, and particularly of the subgroup macroscopic classification of Type III.
In early tumors, ie, those in which there is invasion only of the mucousa or at maximum of the submucosa, nodal metastases rates are lower, ranging from 0 to 20.3% (mean 3.2%) when the tumor is confined to the mucosa and from 10.2 to 33.3% (mean 19.2%) 2 when reaching the submucosa.These variations are determined by the degree of depth of invasion in each of these layers, with the lowest rates of metastases in tumors with involvement of only the top 1/3 of the mucosa (m 1 ) and highest in those with invasion of the lower 1/3 of the submucosa (sm 3 ), where the lymphatic network is richer 15 .Moreover, most of these lymph node metastases in EGC reach the level 1, perigastric lymph nodes, with an incidence of implants in level 2 chains ranging from 10% to 32% 16,17 .These data suggest that extensive gastric resection with broad D2 lymphadenectomy may be unnecessary for many patients with EGC 16 .Smaller gastrectomies, with modified D1 lymphadenectomy, have lower postoperative morbidity and mortality and provide better quality of life for this group of patients, without compromising the oncological radicality necessary for the treatment of gastric cancer 18,19 .The research of sentinel lymph node in EGC can help determine these nodal metastases [20][21][22][23] .Abe et al. proposed the combination of EGC endoscopic resection with laparoscopic lymphadenectomy in patients with EGC with risk criteria for lymph node metastasis 24 .
In our study, the number of patients diagnosed with EGC, 26 (16.3%), was relatively high by Western standards, where this ratio is around 12-15%, but low compared to Eastern data 2,25 .The overall incidence of lymph node metastasis, 30.8% in our patients, was also very high when compared with both Eastern and Western patient populations.A German study with 126 patients with EGC showed rates of lymph node metastases around 18%, ranging from 10.6% in restricted mucosal lesions to 25.3% in those with submucosal invasion 15 .Our rates of metastasis of tumors confined to the mucosa and with the invasion of the submucosa were 16.7% and 42.9%, respectively, also high, and statistically significant difference between them.However, the degree of tumor penetration of the gastric wall is just one of the factors determining the risk of secondary lymph node implants.Other important factors are the macroscopic appearance of the lesion, tumor size, presence of lymphatic or vascular invasion and tumor differentiation grade 1,2,3,[5][6][7]13,15 . Earlytype III tumors, or ulcerated, or mixed and partially ulcerated, display higher rates of metastasis than other types of EGC, as attested in our sample.A study of 684 patients with EGC showed a lymph node metastasis rate of 3.4% in tumors confined to the mucosa and 19% in those with invasion of the submucosa: in all patients with lymph node metastases the tumor had areas of ulceration 16 .
Tumor size is another risk factor for the development of lymph node metastases.The greater the injury, the greater the risk of metastases [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16] .In our series, the incidence of metastasis in tumors larger than 2cm was greater than in tumors smaller than 2cm.This parameter, although significant in univariate analyzes, is best evaluated when considered in conjunction with other risk factors for metastasis.Hölscher et al., for example, demonstrated that the probability of lymph node metastases is very low in T1a tumors smaller than 2cm and in T1b lesions smaller than 1cm 15 .Hirasawa et al.studied over 3800 patients and did not observe lymph node metastasis in intramucosal, well differentiated, tumors of less than 3cm, without angiolymphatic invasion, whether ulcerated or not 6 .Regarding the degree of cell differentiation, welldifferentiated tumors have a lower rate of metastasis than their moderately differentiated and poorly differentiated counterparts.Accordingly, as for Lauren's classification, the intestinal type tends to have a lower rate of lymph node metastasis than the diffuse one 26 .These data are confirmed in our work and can be seen in Table 5.Still from the histological point of view, tubular and papillary adenocarcinomas have better prognosis than carcinomas with signet ring cells 13,15 .Thus, it can be stated that the incidence of lymph node metastasis is null in tumors smaller than 2cm, type I (elevated), well-differentiated, intestinal type of Lauren and restricted to the mucosa, allowing the safe conduct of endoscopic resection.On the other hand, tumors poorly differentiated or undifferentiated, diffuse type of Lauren, larger than 3cm, even restricted to the mucosa, have rates of lymph node metastasis higher than 8%, reaching more than 20% when there is involvement of the submucosa 6.In our series, the incidence of nodal metastases in EGC in these three conditions was 38.9%, 41.7% and 33.3%, respectively.
Gotoda 8 analyzed a large database of over 5000 patients who underwent gastrectomy with meticulous D2 lymphadenectomy, confidently establishing the risk of lymph node metastases in EGC and listing four groups of patients node metastasis in early gastric cancer with a metastases rate of 0%: 1-tumor restricted to mucosa, well or moderately differentiated, without lymphatic vascular invasion, with or without ulceration, less than 3cm in size; 2-tumor restricted to the mucosa, well or moderately differentiated, without lymphatic vascular invasion, without ulceration, of any size; 3-tumor limited to the mucosa, undifferentiated or poorly differentiated, without lymphatic vascular invasion, without ulceration, less than 2cm in size; and 4-tumor with superficial invasion of the submucosa (sm 1 ) well or moderately differentiated, without lymphatic vascular invasion, less than 3cm in size.
Kwee et al., in a meta-analysis, sought to identify predictive factors of nodal metastases in EGC 2 .The main variables significantly associated with metastases in tumors restricted to the mucosa were: age less than 57 years, tumor located in the middle 1/3 of the stomach, large lesions, depressed and ulcerated, undifferentiated, tumors, diffuse Lauren type, and presence of lymphatic invasion.The main variables significantly associated with metastasis in tumors with invasion of the submucosa were, on their turn, were: female gender, tumor located in the lower 1/3 of the stomach, large lesions, tumors with deeper submucosal invasion, undifferentiated tumors, and lymphatic or vascular invasion.
Other aspects related to lymph node metastasis still not well understood and that may have negative impact on the survival of patients suffering from EGC undergoing resections and more conservative lymphadenectomy are the micrometastases and the skip metastases.The former are not detected in conventional histopathology, but only with immunohistochemical tests, and may therefore go unnoticed on frozen section exams, not being resected in D2 lymphadenectomy.
Kim et al. studied 90 patients staged as T1N0 and found micrometastases in 10% of them 27 .None of them presented with tumor recurrence in more than five years of follow-up.The main independent risk factors associated with these micrometastases were the presence of lymphatic invasion and tumor size.The skip metastases, those that occur on more distant lymph nodes chains without involvement of lymph nodes closest to the tumor, bring the risk of false negativity for metastases upon research for sentinel lymph nodes.Kitagawa 29 .Most of these skip metastases concentrated on lymph node chains 7 and 8, which should therefore be included in the lymphadenectomy of patients with negative sentinel nodes undergoing minor gastric resection with modified D1 lymphadenectomy 10 .
Our rate of 16.3% of EGC amongst the 160 patients studied was relatively high for Western standards, but still very low when compared with Eastern standards.The incidence of lymph node metastasis in these patients was very high and suggests that one should keep the radicality of surgery for EGC, particularly in relation to D2 lymphadenectomy, recommended for advanced gastric cancer.Conservative resections with lymphadenectomies lower than D2 should be performed only in selected cases, well studied for the risk factors of lymph node metastasis.Despite the small number of cases did not allow to establish relation between the rate of lymph node metastasis and the risk factors studied, there was a strong tendency for the occurrence of these metastases in tumors with invasion of the submucosa, larger than 2cm, type III, poorly differentiated and diffuse type of Lauren, as already demonstrated in other works.
size in cm was expressed as median (minimum -maximum) and compared by Mann-Whitney test.

Table 2 -Table 2 -Table 2 -Table 2 -Table 2 -
Incidence of lymph node metastasis and N staging according to the degree of tumor penetration of the gastric wall (t).
p< 0.0001 (significant Association between the degree of tumor penetration of the gastric wall -T -and the presence of lymph node metastasis -N). *

Table 3 -
Tumor Characteristics and N staging of patients with EGC.

Table 4 -Table 4 -Table 4 -Table 4 -Table 4 -
Lymph node metastasis and N staging according to T, size of the lesion, macrocospic classification, degree of tumoral differentiation and histological type of Lauren.

Table 5 -Table 5 -Table 5 -Table 5 -Table 5 -
Relationship between rates of lymph node metastasis according to T, size of the lesion, macroscopic classification, degree of tumor differentiation and histological type of Lauren.

Table 6 -Table 6 -Table 6 -Table 6 -Table 6 -
Relationship between rates of lymph node metastasis and injury variables: T, lesion size, macroscopic classification, tumor differentiation grade and histological type of Lauren.
et al. performed a study on sentinel lymph nodes and drew attention to a 5.1% rate of skip metastasis 28 .Saito et al. analyzed 313 patients with lymph node metastasis in N2 chains and found 21 (6.7%) without metastases in level 1 lymph nodes