Abstract
BACKGROUND: Tumor budding is associated with lymph node (LN) metastasis in submucosal colorectal cancer (CRC). However, the rate of LN metastasis associated with the number of tumor buds is unknown. Here, we determined the optimal tumor budding cut-off number and developed a composite scoring system (CSS) for estimating LN metastasis of submucosal CRC. METHODS: In total, 395 patients with histologically confirmed T1N0-2M0 CRC were evaluated. The clinicopathological characteristics were subjected to univariate and multivariate analyses. The Akaike information criterion (AIC) values of the multivariate models were evaluated to identify the optimal cut-off number. A CSS for LN metastasis was developed using independent risk factors. RESULTS: The prevalence of LN metastasis was 13.2%. Histological differentiation, lymphatic or venous invasion, and tumor budding were associated with LN metastasis in univariate analyses. In multivariate models adjusted for histological differentiation and lymphatic or venous invasion, the AIC value was lowest for five tumor buds. Unfavorable differentiation (odds ratio [OR], 8.16; 95% confidence interval [CI], 1.80-36.89), lymphatic or venous invasion (OR, 5.91; 95% CI, 2.91-11.97), and five or more tumor buds (OR, 3.01; 95% CI, 1.21-7.69) were independent risk factors. In a CSS using these three risk factors, the rates of LN metastasis were 5.6%, 15.5%, 31.0%, and 52.4% for total composite scores of 0, 1, 2, and ≥ 3, respectively. CONCLUSIONS: For the estimation of LN metastasis in submucosal CRC, the optimal tumor budding cut-off number was five. Our CSS can be utilized to estimate LN metastasis.