Quantitative evaluation of the relative apparent diffusion coefficient values on multiparametric magnetic resonance imaging to predict higher Gleason score prostate cancer

Abstract

Introduction: Apparent diffusion coefficient (ADC) values on multiparametric magnetic resonance imaging (mpMRI) have been reported to correlate with high-Gleason score (GS) prostate cancer. However, the relative ADC values between tumor lesions and normal tissue have been suggested as more suitable than the absolute ADC values for evaluation of diffusion abnormalities, because absolute ADC values are susceptible to differences in scanners or scanner settings. The present study evaluated the usefulness of the relative assessment of ADC values between tumor lesions and normal tissue on preoperative mpMRI for the prediction of high-risk prostate cancer on radical prostatectomy specimens.

Materials and Methods: A retrospective analysis of 48 men who underwent radical prostatectomy between January 2013 and December 2014 was conducted. MpMRI was performed with a 3.0-T scanner using b-values of 0 and 1500 s/mm². ADC values of the tumor (ADC_TUMOR) and normal prostate and the relative ADC tumor/normal ratio (ADC_TNR) were evaluated by two radiologists.

Results: The inter-rater reliability between two radiologists for ADC_TUMOR measurement was high, with Pearson’s r = 0.982. There was no difference in ADC_TUMOR between GS ≤7 and GS ≥8. In contrast, ADC_TNR was significantly lower in GS ≥8 than in GS ≤7. ROC curves of ADC_TNR to predict higher GS (≥8) showed better classification performance (AUC = 0.8243, p = .0012 by radiologist A and AUC = 0.7961, p = .0031 by radiologist B) than of ADC_TUMOR.

Conclusions: The relative assessment of ADC values between tumor lesions and normal tissue could improve the detection rate of high-risk prostate cancers.

Keywords:

• Prostate cancer
• multiparametric MRI
• apparent diffusion coefficient

Acknowledgements

The authors would like to thank their colleague Hiroyuki Nakazawa, who provided expertise that greatly assisted the research.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by JSPS KAKENHI [Grant Number 15K06882] and [Grant Number 16K11030]. Hiroyoshi Suzuki received research grants from Astellas Pharma Inc., Takeda Pharmaceutical Co., Ltd., Novartis, Pfizer Inc., Daiichi Sankyo Co., Ltd., Taiho Pharmaceutical Co., Ltd., Kissei Pharmaceutical Co., Ltd., and Nippon Kayaku Co., Ltd. Hiroyoshi Suzuki received lecture fees from AstraZeneca, Astellas Pharma Inc., Takeda Pharmaceutical Co., Ltd., Sanofi S.A., Daiichi Sankyo Co., Ltd., Bayer AG, and Janssen Pharmaceuticals, Inc.