Magnetic Resonance Imaging (MRI) Biomarkers for Therapeutic Response Prediction in Rectal Cancer

Abstract

Prediction of chemoradiotherapy (CRT) response in rectal cancer would enable stratification of management whereby responders could undergo ‘watch-and-wait’ to avoid surgical morbidity, and non-responders could have early treatment intensification to improve therapeutic outcomes. Functional MRI can assess tumour function and heterogeneity, and may improve therapeutic response prediction. The aims of this PhD were to (i) prospectively evaluate multi-parametric MRI at 3.0 tesla in vivo combining diffusion weighted imaging (DWI) and dynamic contrast enhanced (DCE) MRI for prediction of CRT response and 2 year disease-free survival (DFS), and (ii) examine diffusion tensor imaging (DTI) MRI biomarkers of rectal cancer extent and heterogeneity at ultra-high field 11.7 tesla ex vivo in order to establish a pipeline for MRI biomarker discovery from ultra-high field to clinical field.

Patients with locally advanced rectal cancer undergoing CRT followed by surgery underwent multi-parametric MRI before, during, and after CRT. A whole tumour voxelwise histogram analysis of apparent diffusion co-efficient (ADC) and Ktrans heterogeneity was performed and correlated with histopathology tumour regression grade. After CRT (before surgery) ADC 75th and 90th quantiles were significantly higher in responders than non-responders. Patients with higher Ktrans values after CRT or greater increase in Ktrans values from before to after CRT had a significantly higher risk of distant metastases, and lower 2 year DFS.

Biobank tissue from patients with rectal cancer were examined at 11.7 tesla and DTI-MRI results correlated with histopathology. This work established a discovery framework for screening Biobank cancer tissue for novel MRI biomarkers of tumour extent and heterogeneity, and resulted in good preservation of tissue integrity and MRI-histopathology alignment. DTI-MRI derived fractional anisotropy (FA) was able to differentiate between tumour and desmoplasia, fibrous tissue, and muscularis propria, allowing for more accurate delineation of rectal cancer tumour extent and stromal heterogeneity ex vivo.

In conclusion, DWI-MRI was predictive of CRT response, DCE-MRI was predictive of 2 year DFS, and DTI-MRI was able to more accurately define tumour extent and heterogeneity in rectal cancer. These findings could be useful for stratification of patients for individualised treatment based on accurate assessment of tumour extent and therapeutic response prediction.