The progress of endoscopic diagnosis and techniques has resulted higher frequency of detection of early stage CRC. With the advances of endoscopic instruments and techniques, the number of endoscopic resections for early-stage CRC has increased providing better quality of life for patients. In the treatment of early CRC which tumor cell forming in mucosal layer, it is better to perform a local treatment without adjuvant therapy because such cancer rarely metastasizes to the lymph node or distant organs [
6]. With application of endoscopic treatments being widespread, an adverse outcome such as tumor recurrence was observed occasionally[
13,
14]. Therefore, this treatment is only adequate for early stage CRC without lymph node metastasis when the cancer cells have invaded submucosa [
5‐
7]. In previous studies reported, the depth of submucosal invasion (Sm3), poor histologic grade, lymphovascular invasion, and tumor cell dissociation are labeled to be the risk factors for lymph node metastasis in submucosal invasive CRC.
Among these factors, the depth of submucosal invasion is an important predictor of lymph node metastasis [
4‐
7,
10,
13‐
16]. There are two ways to measure the depth of submucosal invasion. First is to directly measure the vertical distance of submucosal invasion from muscularis mucosae. When the muscularis mucosae could not be identified due to carcinomatous invasion, the superficial aspect of the submucosal invasive carcinoma is used as a baseline. The other is by measuring a relative depth applying the Kudo classification [
4]. Kitajima et al. [
15] examined 865 submucosal invasive CRCs and analyzed the correlation between lymph node metastasis and depth of submucosal invasion. For pedunculated submucosal invasive CRC, the rate of lymph node metastasis was 0% in head invasion cases and stalk invasion cases with depth of invasion <3000 μm if lymphatic invasion was negative. For nonpedunculated submucosal invasive CRC, the rate of lymph node metastasis was also 0% if the depth of invasion was <1000 μm. Sung et al. [
6] demonstrated that the minimal extent of submucosal invasion in tumors with lymph node metastasis was 1840 μm. Based on this finding, they suggested that when submucosal invasion of an endoscopically resected tumor was <1000-1500 μm, a complete cure can be achieved by endoscopic resection alone. Kudo et al. [
4] proposed criteria for staging early CRCs in 1984 classifying degrees of submucosal invasion into three types based on the depth of invasion: sm1 (less than one-third of the submucosa is invaded), sm2 (intermediate), and sm3 (more than two-thirds of submucosa is invaded). For sm1, it is subclassified into sm1 into a, b, and c, according to the degree of horizontal extension of cancer in the submucosal layer. This staging of submucosal invasion is reflected to the prognosis of lesion. They reported that the lymph node metastasis was detected 3.7% in sm1c cancers, 10.9% in sm2 cancers, and 23.7% in sm3 cancers. In present study, eight patients had metastatic lymph node. However, data of malignant polyps in two patients who underwent colonoscopic polypectomy at a local clinic before radical surgery could not be found. There was no lymph node metastasis in submucosal invasive CRCs that the depth of invasion was <1000 μm. In terms of primary tumor with lymph node metastasis, four (67.7%) of six patients had a depth of invasion of ≥3000 μm while two patients had depth of 1000 μm and 1500 μm respectively. The lymph node metastasis was detected in 14.3% of sm1 cancers, 5.6% of sm2 cancers, and 23.1% of sm3 cancers. Comparing to other previous studies, the lymph node metastasis was more frequently detected in sm1 cancers. Based on this finding, although the depth of invasion is very important predictive factor for lymph node metastasis in submucosal invasive CRC, other factors such as microscopic grade, lymphovascular invasion, and tumor budding should be considered for complete cure by endoscopic resection alone. Ueno et al. [
16] demonstrated that to extend the criteria for curative endoscopic resection, the combination of a quantitative factor and qualitative factors is useful. The quantitative factor is the depth of submucoal invasion and the qualitative factors include tumor grade, lymphovascular invasion, tumor budding, etc. Several studies have reported that the lymphovascular invasion is one of the histological risk factors for lymph node metastasis of submucosal invasive CRC[
10,
15‐
17]. In light of the clinical and biological importance of this feature in T1 tumors, American Joint Committee on Cancer recommended that the following modification of the T1 category (tumor invades the submucosa) to be used: T1a (no evidence of lymphatic or venous invasion) or T1b (lymphatic or venous invasion is present). However, it will remain problematic as long as diagnosis continues to differ occasionally among pathologists. Controversies over the detection of lymphovascular invasion arise mainly from the difficulty in visualizing the lymphatic vessel wall. It is difficult to detect lymphovascular invasion with conventional hematoxylin and eosin(HE) staining [
18]. To overcome this limitation, Ishii et al.[
19] recommended the use of immunostaining with some materials such as monoclonal antibody D2-40 demonstrating that such method would provide important information about the risk of lymph node metastasis and may prove useful in evaluating the necessity of an additional resection after local excision in T1 CRCs. In the present study, lymphovacular invasion was diagnosed in 7 (14.6%) of 48 cases that were examined only with HE staining. Lymph node metastasis could be found in 2 (28.6%) of these 7 cases. From our results, we conclude that a close evaluation for lymphovascular invasion with immunostaining may be helpful for identifying the risk factor for lymph node metastasis in submucosal invasive CRC. Tumor budding is also seen to be another important parameter in the evaluation of submucosal invasive CRC. This evaluation for tumor budding has merits in that evaluation is possible by ordinary HE staining [
20]. Also, tumor budding may be a factor related to lymphovascular invasion. Morodomi et al. [
21] have suggested that by making serial sections of specimen, budding is associated with lymphovascular invasion. Moreover, some reports suggest that the most important factor regarding the presence of budding is lymphovascular invasion.[
22,
23] High-grade tumor budding has been consistently linked to lymph node metastasis, distant metastasis, local recurrence and correlates with the distance of tumor invasion beyond the outer border of the muscularis propria. Tumor budding is proposed as a useful indicator for isolated tumor cells in lymph nodes in patients with early CRCs and could indicate curative surgery in patients with locally excised T1 tumors.[
24] Also, microacinar structures (buds of tumor emanating from larger glands) and undifferentiated cells in stroma may be the antecedent stages of lymphatic invasion.[
21] Goldstein et al.[
12] suggested the association of extensive budding, microacinar structure and undifferentiated cells along the advancing edge with lymph node metastases. To sum up the findings, it is suggested that one of qualitative factors may not be detected independently, but any of those may be detected at progression of tumor. In other words, the detection of any qualitative factor is important in predicting lymph node metastasis. In this study, tumor budding (P = 0.042, Hazard ratio 13.285, Confidence interval 1.094 – 161.297) was the only independent factor for predicting lymph node metastasis. Also, although there was no lymph node metastasis in 15 patients with tumor budding (positive predictive value was 0.25) and not appropriated due to small sample size of this study, tumor budding seems to have a high sensitivity (83.3%), acceptable specificity (60.5%), and a high negative predictive value (0.958). We think that a high negative predictive value may be more important than a positive predictive value for clinician in order to decide whether an additional radical resection after local excision or colonocopic excision for submucosal invasive T1 colorectal cancer would be required or not. This finding may suggested that a close examination of histopathologic finding would provide important information about the risk of lymph node metastasis and may prove useful in evaluating the necessity of an additional resection after local excision in T1 CRC cases.
Our study does have some limitations. It was a retrospective study using the medical records of small number of patients, and the pathologic review of primary tumor could not be performed in all patients. Hence, our results for depth of invasion and lymphovascular invasion as a risk factor for lymph node metastasis were somewhat different compared with previous reported studies. Therefore, a large scaled prospective trial is necessary to verify the management strategy for submucosal invasive CRC.