Prostate cancer remains the most common cancer and the second cause of cancer-related death among men in the US [1]. Multiparametric magnetic resonance imaging (mpMRI), which includes T2-weighted (T2W), diffusion-weighted (DWI), and dynamic contrast-enhanced (DCE) imaging, is now widely adopted in clinical practice for the detection of prostate cancer [2]. Identification of an index lesion (i.e. dominant nodule) on mpMRI is of most importance as it usually drives the decision to pursue a targeted biopsy [3]. Although interpretation is based almost exclusively on subjective assessment of imaging findings, quantitative analysis of mpMRI data can potentially provide objective assessment of tumor characteristics such as tumor vascularity and cellularity that can add to both the detection and characterization of the index lesion. However, integration of these quantitative tools in the clinical assessment of patients with known or suspected prostate cancer are lacking. The goal of this study was to investigate the use of quantitative DWI and DCE measurements in MRI-visible index lesions as a surrogate for aggressiveness in prostate cancer patients.

In this institutional review board-approved retrospective study, consecutive prostate mpMRI exams performed on a 3T MRI scanner (Achieva, Philips Medical Systems) with a 6-channel cardiac coil (Philips Medical Systems) and an endorectal coil (eCoil, Bayer Healthcare) at our institution between February and October 2014 were reviewed. A total of 43 patients who had prostate acinar adenocarcinoma and radical prostatectomy were included in this study. Patients were stratified into low risk (Gleason score 6, or 3+4 with cancer in <20% of the prostate; n=13), intermediate risk (Gleason score 3+4 with cancer in ≥20% of the prostate; n=6), and high risk (Gleason score ≥4+3; n=14) [4,5]. Our clinical protocol included: (1) T2W fast spin-echo anatomical imaging; (2) Single-shot spin-echo echo-planar DWI with b-values of 0, 10, 25, 50, 100, 250, 450, 1000, 1500, and 2000 s/mm²; (3) DCE using a 3D spoiled gradient-echo sequence before, during, and after a bolus injection of 0.1 mmol/kg body weight of gadobutrol (Gadavist; Bayer Healthcare Pharmaceuticals) using a power injector at a rate of 3 mL/sec followed by a 20 mL saline flush at the same rate.

All patient studies were reviewed on VersaVue Enterprise (iCAD Inc.). With the knowledge of patient’s surgical pathology, a region of interest (ROI) of the dominant tumor was manually drawn on ADC map generated from the scanner. Circular ROIs representing normal central gland and normal peripheral zone were also defined. These ROIs were copied to DCE parametric maps calculated from the Tofts model to obtain K^trans (transfer constant), k_ep (rate constant), v_e (extravascular extracellular volume fraction), and iAUC (initial area under the curve). Quantitative diffusion parameters were computed with different diffusion models: (1) monoexponential (mono): $$S=S_{0}\cdot{e^{-b\cdot{ADC}}}$$ (2) biexponential intravolxel incoherent motion (IVIM) (biexp): $$S=S_{0}\cdot({(1-f)\cdot{e^{-b\cdot{D_{t}}}}}+f\cdot{e^{-b\cdot{D_{p}}}})$$ and (3) simplified IVIM (sIVIM): $$S=S_{0}\cdot({(1-f)\cdot{e^{-b\cdot{D_{t}}}}}+f\cdot{\delta_{0}{(b)}})$$ in which the perfusion effect was modeled by a Delta function for b = 0 s/mm² [6]. ADC is the apparent diffusion coefficient, D_t represents the pure tissue diffusion coefficient, D_p is the pseudo-diffusion coefficient, and f is the perfusion fraction. All b-values were used for mono and biexp models, whereas only b-values of 0, 250, 450, 1000, 1500, and 2000 s/mm² were used for the sIVIM model.

Analysis of variance (SAS 9.3) was used to test the difference in mpMRI parameters between tumor and normal tissues, and between low and intermediate-to-high risk tumors. Logistic regression using the stepwise backward elimination [7] was performed to evaluate the association of mpMRI parameters with tumor risk. The area under the receiver operating characteristic (ROC) curve (AUC) was calculated for the comparison of the effect of mpMRI measures on risk category.

Figure 1 shows representative mpMRI images of tumors with low, intermediate, and high risk. Tumors with intermediate-to-high risk showed more restricted diffusion compared to low risk tumors (Table 1). Higher DCE measures were observed in tumors with intermediate-to-high risk, but did not reach statistical significance. Logistic regression with backward selection suggested on average, a model of iAUC and D_t from sIVIM had best performance in discriminating tumor risk categories. ROC analysis showed higher AUC value when these two measures were combined (Figure 2).Our results showed combination of quantitative DWI and DCE analysis improved the characterization of prostate cancer compared to either technique alone for the distinction of low and intermediate-to-high risk cancer. Anatomic and functional mpMRI study may provide a more reliable assessment of the aggressiveness of prostate cancer in patients.