Magnetization transfer MRI and T2-weighted MRI texture analysis allow non-invasive detection of intestinal fibrosis in a fibrosis-inducing chronic colitis mouse model

Introduction: Differentiation between inflammatory and fibrotic bowel strictures remains a holy grail in Crohn’s disease (CD) management, primarily because of the therapeutic implications. Stricture characterisation in CD patients is further hampered by the co-occurrence of inflammation and fibrosis in various degrees. Currently, there are no validated imaging tools to differentiate fibrotic from inflammatory or mixed strictures, but several new techniques, such as magnetization transfer magnetic resonance imaging (MT-MRI) and magnetic resonance texture analysis (MR-TA), seem promising. Aim: We investigated whether MT-MRI and MR-TA can detect and quantify intestinal fibrosis in an established mouse model of fibrosis-inducing chronic colitis. Methods: Chronic colitis was induced in C57BL/6 (n=16) mice by three consecutive cycles of administration of the colitis inducing agent dextran sodium sulphate (DSS) for 7 days, followed by a 14-day recovery period. Using a 7.0 Tesla scanner, MT and T2-weighted MR images were recorded for each mouse at baseline and weeks (wk) 1, 3, 4, 6, 7, and 9. Regions of interest were created over the bowel wall on both MT and T2-weighted images. The bowel wall to spine muscle normalized MT ratio was calculated using ImageJ. Textural features, including skewness, kurtosis and entropy, were extracted by a filtration histogram technique, enclosed in the TexRAD software. Masson’s trichrome stained colon sections were used as golden standard for the quantification of fibrosis. Multivariate mixed model analysis and receiver operating characteristic (ROC) curves were applied to statistically assess the MR parameters for evaluating intestinal fibrosis. Results: Conducting mixed model analysis, significant differences in MT-ratio between mixed strictures and pure fibrotic strictures were observed at wk4 vs. wk6 (p=0,012) and at wk7 vs. wk9 (p<0,0001), whereas the significance for TA-entropy was lower at these compared timepoints (p= 0,108 and p=0,012, respectively). Nonetheless, TA-entropy performed better when monitoring the increasing proportion of fibrotic tissue in mixed strictures between first and second (wk1 vs. wk4, p=0,001) and second and third DSS-cycle (wk4 vs. wk7, p=0,004), compared to the MT-ratio. ROC-curve analysis using histopathology as reference score, generated a sensitivity, specificity and area under the curve (AUC) for TA-entropy of 83%, 80% and 0,93 (p=0,018) respectively, while MT-ratio outperformed with a sensitivity and specificity of 100% and an AUC of 1 (p=0,006). Conclusions: To our knowledge, we have shown for the first time that MT-MRI and MR-TA can both accurately detect fibrosis in a fibrosis-induced chronic colitis mouse model. Where TA-entropy excels in detecting fibrosis build-up in an inflammatory environment, MT-ratio has higher sensitivity and specificity to detect fibrotic tissue as such. MT-ratio and TA-entropy should be seen as complementing techniques prompting comprehensive fibrosis detection.