Optimal diagnostic method using multidetector-row computed tomography for predicting lymph node metastasis in colorectal cancer

Abstract Background Prediction of nodal involvement in colorectal cancer is an important aspect of preoperative workup to determine the necessity of preoperative treatment and the adequate extent of lymphadenectomy during surgery. This study aimed to investigate newer multidetector-row computed tomography (MDCT) findings for better predicting lymph node (LN) metastasis in colorectal cancer. Methods Seventy patients were enrolled in this study; all underwent MDCT prior to surgery and upfront curative resection for colorectal cancer. LNs with a short-axis diameter (SAD) ≥ 4 mm were identified on MDCT images, and the following measures were recorded by two radiologists independently: two-dimensional (2D) SAD, 2D long-axis diameter (LAD), 2D ratio of SAD to LAD, 2D CT attenuation value, three-dimensional (3D) SAD, 3D LAD, 3D SAD to LAD ratio, 3D CT attenuation value, LN volume, and presence of extranodal neoplastic spread (ENS), as defined by indistinct nodal margin, irregular capsular enhancement, or infiltration into adjacent structures. Results Forty-six patients presented 173 LNs with a SAD ≥ 4 mm, while 24 patients exhibited pathologically confirmed LN metastases. Receiver operating characteristic analysis revealed that 2D LAD was the most sensitive measure for LN metastases with an area under the curve of 0.752 (cut-off value, 7.05 mm). When combined with CT findings indicating ENS, 2D LAD (> or ≤ 7 mm) showed enhanced predictive power for LN metastases (area under the curve, 0.846; p < 0.001). Conclusions LAD in axial MDCT imaging is the most sensitive measure for predicting colorectal LN metastases, especially when MDCT findings of ENS are observed