Abstract
Structural abnormality is a well-recognized feature of malignancy. On the other hand, diffusion-weighted MRI (DWI) has been reported as a tool that can reflect tumor biology. The purpose of this study is to apply histogram analysis to DWI to quantify structural abnormality of colorectal cancer, and evaluate its biomarker value. This is a retrospective study of 80 (46 men and 34 women; median age: 68.0years) colorectal cancer patients who underwent DWI followed by curative surgery at the Chiba University Hospital between 2009 and 2011. Median follow-up time was 62.2months. Histogram parameters including signal intensity of kurtosis and skewness of the tumor were measured on DWI at b = 1000, and mean apparent diffusion coefficient value (ADC) of the tumor was also measured on ADC map generated by DWIs at b = 0 and 1000. Associations of tumor parameters (kurtosis, skewness, and ADC) with pathological features were analyzed, and these parameters were also compared with overall survival (OS) and relapse-free survival (RFS) using Cox regression and Kaplan-Meier analysis. ADC of the tumor did not have significant associations with any pathological factors, but kurtosis and skewness of signal intensity in the tumor was significantly different between tumors with distant metastases and those without (4.23 ± 1.31 vs. 3.24 ± 1.32, p = 0.04; 1.09 ± 0.39 vs. 0.57 ± 0.58, p = 0.03). Kurtosis of the tumor was significantly correlated with OS and RFS (p = 0.04, p = 0.03, respectively), and skewness was significantly correlated with OS (p = 0.03) in Cox regression analysis. Higher kurtosis or higher skewness of the tumor was associated with worse OS in Kaplan-Meier analysis (p = 0.01, p = 0.009, log-rank). In subset analysis, there were 50 patients (32 men and 18 women) of lymph node-negative colorectal cancers (≤ stage II); skewness of signal intensity in the tumor was associated with OS using univariate Cox regression analysis (p = 0.04). Histogram analysis of DWI can be a prognostic biomarker for colorectal cancer.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have