The reliability of the computed tomography perfusion parameters as a predictor of colorectal cancer grade: influence of tumor region of interest position

• Main Authors: Lamiaa M. R. Khalaf, Peter E. Megal, Mostafa El-Sharkawy, Afaf Abd Elkader, Mona M. Sayed, Marwa T. Hussien, Hanan A. Eltyb
• Format: Article
• Language: English
• Published: SpringerOpen 2020-03-01
• Series: The Egyptian Journal of Radiology and Nuclear Medicine
• Subjects: Colorectal cancer grades; Multidetector computed tomography; Perfusion
• Online Access: http://link.springer.com/article/10.1186/s43055-020-00166-6

Abstract Background The aim of this study is to prospectively determine whether the computed tomography perfusion (CTP) can predict the tumor grade in colorectal cancer (CRC), and investigation of the best position of the tumor region of interest (ROI) could improve the tumor grade differentiation. Results The mean blood flow (mBF) at the tumor edge was significantly different among different tumor grades with the highest value noted in moderately differentiated and the lowest value in poorly differentiated CRC (111.67 ± 38.91 vs. 52.25 ± 21.97 ml/min/100 g, respectively, P = 0.03). At the tumor center, the BF and blood volume (BV) showed highest value in moderately differentiated and the lowest value in poorly differentiated CRC (P < 0.001, 0.02 respectively). The ROC curve revealed that the BF at the tumor edge showed its best results as regard its sensitivity 73.9% and specificity 82.1% in the characterization of moderately differentiated CRC from the other subtypes at > 92.2 ml/100 g/min cut off point and their AUC and P value were (0.82 and < 0.001 respectively), while the AUC of the BF at the tumor center was 0.81 and its P value was 0.03. In the characterization of poorly differentiated CRC from the other subtypes, BF cut off value at < 67.32 ml/100 g/min at the tumor edge showed the best performance of its sensitivity 88% and specificity 88% with 0.92 AUC and P value was < 0.001, while the AUC of the BF at the tumor center was 0.89. BF was higher at the edge than at the center of the tumor (88.53 ± 22.45 vs. 78.79 ± 28.98 ml/min/100 g; P < 0.001 respectively). Conclusions PCT could help in prediction of the poorly and the moderately differentiated CRC and the mBF was the most reliable parameter in prediction of the grade. The variation in ROI position has no substantial difference in the prediction of the tumor grade; however, this variation has a significant impact on the measurements of the PCT parameters especially the BF.