Three-dimensional nuclear magnetic resonance spectroscopy: A complementary tool to multiparametric magnetic resonance imaging in the identification of aggressive prostate cancer at 3.0T

Abstract

Background: The limitations of the assessment of tumor aggressiveness by Prostate Imaging Reporting and Data System (PI-RADS) and biopsies suggest that the diagnostic algorithm could be improved by quantitative measurements in some chosen indications. We assessed the tumor high-risk predictive performance of 3.0 Tesla (3.0T) multiparametric magnetic resonance imaging (mp-MRI) combined with nuclear magnetic resonance spectroscopic sequences (NMR-S) in order to show that the metabolic analysis could bring out an evocative result for the aggressive form of prostate cancer. Methods: We conducted a retrospective study of 26 patients (mean age, 62.4 years) who had surgery for prostate cancer between 2009 and 2016 after pre-therapeutic assessment with 3.0T mp-MRI and NMR-S. Groups within the intermediate range of the D'Amico risk classification were divided into two categories, low risk (n=20) and high risk (n=6), according to the International Society of Urological Pathology (ISUP) 2-3 limit. Histoprognostic discordances within various risk groups were compared with the corresponding predictive MRI values. The performance of predictive models was assessed based on sensitivity, specificity, and the area under the curve (AUC) of receiver operating characteristic (ROC) curves. Results: After prostatectomy, histological analysis reclassified 18 patients as high-risk, including 16 who were T3 MRI grade, of whom 13 (81.3%) were found to be pT3. Among the patients who had cT1 or cT2 digital rectal examinations, the T3 MRI factor multiplied by 8.7 [odds ratio (OR), 8.7; 95% confidence interval (CI), 1.3-56.2; P=0.024] the relative risk of being pT3 and by 5.8 (OR, 5.8; 95% CI, 0.95-35.7; P=0.05) the relative risk of being pGleason (pGS) > GS-prostate biopsy. Spectroscopic data showed that the choline concentration was significantly higher (P=0.001) in aggressive disease. Conclusions: The predictive model of tumor aggressiveness combining mp-MRI plus NMR-S was better than the mp-MRI model alone (AUC, 0.95 vs. 0.86). Information obtained by mp-MRI coupled with spectroscopy may improve the detection of occult aggressive disease, helping in the discrimination of intermediate risks.