In vivo prostate cancer detection and grading using restriction spectrum imaging-MRI

K C McCammack, C J Kane, J K Parsons, N S White, N M Schenker-Ahmed, J M Kuperman, H Bartsch, R S Desikan, R A Rakow-Penner, D Adams, M A Liss, R F Mattrey, W G Bradley, D J A Margolis, S S Raman, A Shabaik, A M Dale, D S Karow, K C McCammack, C J Kane, J K Parsons, N S White, N M Schenker-Ahmed, J M Kuperman, H Bartsch, R S Desikan, R A Rakow-Penner, D Adams, M A Liss, R F Mattrey, W G Bradley, D J A Margolis, S S Raman, A Shabaik, A M Dale, D S Karow

Abstract

Background: Magnetic resonance imaging (MRI) is emerging as a robust, noninvasive method for detecting and characterizing prostate cancer (PCa), but limitations remain in its ability to distinguish cancerous from non-cancerous tissue. We evaluated the performance of a novel MRI technique, restriction spectrum imaging (RSI-MRI), to quantitatively detect and grade PCa compared with current standard-of-care MRI.

Methods: In a retrospective evaluation of 33 patients with biopsy-proven PCa who underwent RSI-MRI and standard MRI before radical prostatectomy, receiver-operating characteristic (ROC) curves were performed for RSI-MRI and each quantitative MRI term, with area under the ROC curve (AUC) used to compare each term's ability to differentiate between PCa and normal prostate. Spearman rank-order correlations were performed to assess each term's ability to predict PCa grade in the radical prostatectomy specimens.

Results: RSI-MRI demonstrated superior differentiation of PCa from normal tissue, with AUC of 0.94 and 0.85 for RSI-MRI and conventional diffusion MRI, respectively (P=0.04). RSI-MRI also demonstrated superior performance in predicting PCa aggressiveness, with Spearman rank-order correlation coefficients of 0.53 (P=0.002) and -0.42 (P=0.01) for RSI-MRI and conventional diffusion MRI, respectively, with tumor grade.

Conclusions: RSI-MRI significantly improves upon current noninvasive PCa imaging and may potentially enhance its diagnosis and characterization.

Conflict of interest statement

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Figures

Figure 1
Figure 1
Example tumor (red) and normal peripheral zone (green) region of interest (ROI) assignment to the patient’s T2 image (a), subsequently coregistered to the Ktrans map overlaid on T2 (b), conventional ADC map (c) and restriction spectrum imaging-magnetic resonance imaging (RSI-MRI) map overlaid on T2. The provided RSI-MRI color bar scale demonstrates corresponding z-score (d). ROI definitions were made on T2 images by an experienced radiologist based on the whole-mount hematoxylin and eosin section (e) tumor outlines provided by genitourinary pathology. Final pathology in this case demonstrated Gleason score 4+5 = 9.
Figure 2
Figure 2
Conventional multiparametric-magnetic resonance imaging components and restriction spectrum imaging-magnetic resonance imaging (RSI-MRI) as used for targeting biopsy. (a) 4+3 Gleason score (GS) prostate cancer (PCa) in the right anterior mid central gland, (b) 4+4 GS PCa in the right posterior base peripheral zone, (c) 3+4 GS PCa in the left posterior mid peripheral zone. In these cases, RSI-MRI more conspicuously demarcates PCa for targeted biopsy than T2, dynamic contrast enhancement or conventional diffusion-weighted imaging techniques as confirmed by post-prostatectomy whole-mount pathology.
Figure 3
Figure 3
Receiver-operating characteristic (ROC) curves demonstrating performance of restriction spectrum imaging (RSI)-magnetic resonance imaging z-score, conventional ADC values and Ktransfor the quantitative discrimination of prostate cancer from normal peripheral zone. Respective areas under the ROC curve are listed in the legend.

Source: PubMed

3
Subscribe