Tumor heterogeneity in anal cancer assessed with diffusion-weighted (DW) and T2-weighted MRI: Potential biomarker of treatment outcome.

Abstract

617 Background: Despite advances in the management of anal squamous cell carcinoma (ASCC), roughly 25% of patients undergoing chemoradiotherapy (CRT) will experience a disease recurrence. Currently, there is no validated way to predict treatment failure and guidance on the timing of follow-up scans differs. This study was carried out to assess the role of statistical and fractal heterogeneity parameters from pre-treatment and early post-CRT T2- and diffusion-weighted MR in predicting disease outcome. Methods: Baseline and follow-up MR scans from patients with histologically proven non-metastatic ASCC undergoing radical CRT were included in this study. T2-W and DWI MR images were analyzed by two readers, blinded to clinical data. Statistical and fractal heterogeneity parameters were derived using in-house software. Results: 40 patients were included in the final analysis. Follow-up MRI scan was performed at a median of 15 weeks from the start of radiotherapy. 31/40 patients achieved a radiological complete response (CR) and a further 7/40 partial response (PR). Median follow-up was 34.5 months. During the course of follow-up, 12 patients experienced disease recurrence (30%). Radiological CR was not significantly associated with disease outcome. First order statistical heterogeneity analysis demonstrated a significantly lower ROI Skewness (median -0.05±0.49 vs 0.45±1.04 p=0.01) on baseline T2-W imaging as well as lower ROI Entropy (3.95±0.46 vs 4.45±0.49, p=0.01) and higher ROI Energy (0.07±0.03 vs 0.03±0.03, p=0.01) on follow-up T2-W imaging in subjects who experienced disease recurrence. In these patients T2-W Homogeneity derived from Grey-Level Co-occurrence Matrices (GLCM) was lower at baseline (0.41±0.12 vs 0.58 ±0.15 p=0.02) and associated with a higher T2-W Sum Average (median 35±7 vs median 29±7 p=0.01), whereas T2-W Contrast and ADC Correlation derived from GLCM were lower on follow-up imaging (30±11 vs 38 ±35, p=0.01 and 0.77±0.12 vs 0.88±0.12 p=0.04, respectively). Conclusions: First order statistical and fractal heterogeneity analysis of baseline and early post CRT MR scans in patients with ASCC might have a role in predicting long-term disease outcome.