Multiparametric magnetic resonance imaging for the detection and localization of prostate cancer: combination of T2-weighted, dynamic contrast-enhanced and diffusion-weighted imaging

OBJECTIVE To evaluate the combination of multiple magnetic resonance imaging (MRI) techniques, including T2-weighted imaging (T2W), dynamic contrast-enhanced imaging (DCE) and diffusion-weighted imaging (DWI), for the detection and localization of prostate cancer. PATIENTS AND METHODS In all, 57 patients underwent endorectal MRI at 1.5 T before radical prostatectomy (RP) for localized prostate cancer. On T2W images and histological whole-mount analysis, the peripheral zone (PZ) and transition zone (TZ) were divided into upper and lower glands, as well as left and right halves, thus yielding four quadrants for each zone. On histological analysis, the total number of tumour foci, their location and larger diameter were recorded. T2W alone, T2W + DWI, T2W + DCE and all three techniques combined were scored for the likelihood of tumour in each area and results were compared with whole-mount analysis. The area under the receiver operating characteristic curve (A(z)) was used to evaluate accuracy for tumour detection. The association between MR accuracy and Gleason score was statistically assessed. RESULTS Of the 456 prostate octants analysed, 145 showed cancer on whole-mount analysis, 120 (83%) of them with a diameter assumed to correspond to a volume >0.2 cm(3). Gleason score was >= 7 in 68 (47%) tumours. In the PZ, the A(z) value was significantly higher for T2W + DWI, T2W + DCE and all three techniques combined than for T2W alone (P < 0.05). In the TZ, the A(z) value was higher for T2W + DWI than for T2W alone, but the difference was not significant. The A(z) value for T2W + DWI was significantly higher than that for T2W + DCE or for the three sequences combined. Gleason score was significantly associated with cancer detection in the PZ. CONCLUSIONS Adding DWI and DCE to T2W imaging increased MRI performance in cancer detection in the PZ significantly. However, this multiparametric model failed to improve performance in the TZ. Gleason score significantly influenced cancer detection in the PZ but not in the TZ.