Quantitative 3-T multi-parametric MRI and step-section pathology of recurrent prostate cancer patients after radiation therapy

Catarina Dinis Fernandes,Ivo Schoots,Floris J Pos,Milena Smolic,Petra J Van Houdt,Jeroen De Jong,Iris Walraven,Henk G Van Der Poel,Stijn W T P J Heijmink,Uulke A Van Der Heide,Ghazaleh Ghobadi

doi:10.1007/s00330-018-5819-y

Catarina Dinis Fernandes, Ivo Schoots + Show 9 more

Open Access

https://doi.org/10.1007/s00330-018-5819-y

Copy DOI

Abstract

ObjectivesDiagnosis of radio-recurrent prostate cancer using multi-parametric MRI (mp-MRI) can be challenging due to the presence of radiation effects. We aim to characterize imaging of prostate tissue after radiation therapy (RT), using histopathology as ground truth, and to investigate the visibility of tumor lesions on mp-MRI.MethodsTumor delineated histopathology slides from salvage radical prostatectomy patients, primarily treated with RT, were registered to MRI. Median T2-weighted, ADC, Ktrans, and kep values in tumor and other regions were calculated. Two radiologists independently performed mp-MRI-based tumor delineations which were compared with the true pathological extent. General linear mixed-effect modeling was used to establish the contribution of each imaging modality and combinations thereof in distinguishing tumor and benign voxels.ResultsNineteen of the 21 included patients had tumor in the available histopathology slides. Recurrence was predominantly multifocal with large tumor foci seen after external beam radiotherapy, whereas these were small and sparse after low-dose-rate brachytherapy. MRI-based delineations missed small foci and slightly underestimated tumor extent. The combination of T2-weighted, ADC, Ktrans, and kep had the best performance in distinguishing tumor and benign voxels.ConclusionsUsing high-resolution histopathology delineations, the real tumor extent and size were found to be underestimated on MRI. mp-MRI obtained the best performance in identifying tumor voxels. Appropriate margins around the visible tumor-suspected region should be included when designing focal salvage strategies. Recurrent tumor delineation guidelines are warranted.Key Points• Compared to the use of individual sequences, multi-parametric MRI obtained the best performance in distinguishing recurrent tumor from benign voxels.• Delineations based on mp-MRI miss smaller foci and slightly underestimate tumor volume of local recurrent prostate cancer.• Focal salvage strategies should include appropriate margins around the visible tumor.

Highlights

Post-radiation therapy (RT) benign confounders pose a challenge to Magnetic resonance imaging (MRI) interpretation
Compared to the use of individual sequences, multi-parametric MRI obtained the best performance in distinguishing recurrent tumor from benign voxels
Delineations based on multi-parametric MRI (mp-MRI) miss smaller foci and slightly underestimate tumor volume of local recurrent prostate cancer

Summary

Introduction

Prostate tissue shows diffuse signal intensity (SI) reduction on T2-weighted (T2w) MRI, complicating tumor detection [8]. The use of 1.5-T MR spectroscopy [9] as well as DWI- and DCE-MRI [10] has been found to surpass T2w in detecting local recurrence. Multi-parametric MRI (mpMRI) with diffusion-weighted imaging (DWI) and dynamic contrast-enhanced (DCE) MRI is often preferred. Zattoni et al [11] reported 3-T mp-MRI to have good accuracy in detecting recurrent disease extension. Donati et al [12] found no additional benefit of DCE when added to T2w and DWI. These studies used either biopsy or SRP samples for validation, but none attempted accurate registration of imaging and histopathology. No guidelines exist on how to score or delineate recurrent tumor-suspected regions

Objectives

Methods

Results

Conclusion