Combined Morphologic and Molecular Classification for Predicting Lymph Node Metastasis in Early-stage Colorectal Adenocarcinoma

Background: Identifying reliable predictors of lymph node (LN) metastasis is clinically important, particularly for optimizing treatments for early colorectal cancer (ECC) patients. This study evaluated risk-predictive models of LN metastasis using several pathologic and molecular ECC parameters.

Methods: Tissue specimens from 179 patients with histologically confirmed ECC were enrolled. A total of 20 clinicopathological characteristics, including tumor budding, micropapillary structure, and mucinous differentiation, and 22 protein expressions related to cancer invasion in central and invasive front areas were evaluated for their predictive value for LN metastasis.

Results: Alongside conventional histopathological parameters, tumor budding and mucinous differentiation at the invasive front of ECCs and micropapillary structure were found to be independent predictive factors for LN metastasis. Immunohistochemical expressions of CXCL12 and p38-MAPK at the invasive front were also found to be associated with regional LN metastasis in ECC. Analytic logistic models, using combinations of statistically independent parameters, revealed their abilities to predict LN metastasis in ECC. Further, receiver operating characteristic analysis using combinations of 6 or 7 independent variables represented predictive performances (area under curve of 0.956 or 0.960, respectively) for LN metastasis in ECC.

Conclusions: The combined histomorphologic and molecular factors tested here might be able to predict for LN metastasis in ECC.