Locally advanced rectal cancer: post-chemoradiotherapy ADC histogram analysis for predicting a complete response

• Published in: Acta radiologica (1987) Vol. 56; no. 9; p. 1042
• Main Authors: Cho, Seung Hyun, Kim, Gab Chul, Jang, Yun-Jin, Ryeom, Hunkyu, Kim, Hye Jung, Shin, Kyung-Min, Park, Jun Seok, Choi, Gyu-Seog, Kim, See Hyung
• Format: Journal Article
• Language: English
• Published: England 01.09.2015
• Subjects: Adult; Aged; Aged, 80 and over; Chemoradiotherapy; Combined Modality Therapy; Diffusion Magnetic Resonance Imaging - methods; Digestive System Surgical Procedures; Female; Humans; Image Interpretation, Computer-Assisted - methods; Male; Middle Aged; Neoplasm Staging; Predictive Value of Tests; Rectal Neoplasms - pathology; Rectal Neoplasms - therapy; Retrospective Studies; Treatment Outcome; histogram; magnetic resonance imaging (MRI); apparent diffusion coefficient (ADC); Rectal cancer; chemoradiation
• ISSN: 1600-0455
• DOI: 10.1177/0284185114550193

The value of diffusion-weighted imaging (DWI) for reliable differentiation between pathologic complete response (pCR) and residual tumor is still unclear. Recently, a few studies reported that histogram analysis can be helpful to monitor the therapeutic response in various cancer research. To investigate whether post-chemoradiotherapy (CRT) apparent diffusion coefficient (ADC) histogram analysis can be helpful to predict a pCR in locally advanced rectal cancer (LARC). Fifty patients who underwent preoperative CRT followed by surgery were enrolled in this retrospective study, non-pCR (n = 41) and pCR (n = 9), respectively. ADC histogram analysis encompassing the whole tumor was performed on two post-CRT ADC600 and ADC1000 (b factors 0, 600 vs. 0, 1000 s/mm(2)) maps. Mean, minimum, maximum, SD, mode, 10th, 25th, 50th, 75th, 90th percentile ADCs, skewness, and kurtosis were derived. Diagnostic performance for predicting pCR was evaluated and compared. On both maps, 10th and 25th ADCs showed better diagnostic performance than that using mean ADC. Tenth percentile ADCs revealed the best diagnostic performance on both ADC600 (AZ 0.841, sensitivity 100%, specificity 70.7%) and ADC1000 (AZ 0.821, sensitivity 77.8%, specificity 87.8%) maps. In comparison between 10th percentile and mean ADC, the specificity was significantly improved on both ADC600 (70.7% vs. 53.7%; P = 0.031) and ADC1000 (87.8% vs. 73.2%; P = 0.039) maps. Post-CRT ADC histogram analysis is helpful for predicting pCR in LARC, especially, in improving the specificity, compared with mean ADC.