Differentiation of Prostate Cancer From Normal Prostate Tissue: Role of Hotspots in Pharmacokinetic MRI and Histologic Evaluation

OBJECTIVE. The purpose of this study was to investigate whether analysis of voxels with the highest perfusion and blood volume (hotspots) within a whole region affected by prostate cancer can improve differentiation of cancerous and normal prostate tissue at pharmacokinetic MRI. SUBJECTS AND METHODS. Fifty-three patients with biopsy-proven prostate cancer were examined with 1.5-T MRI performed with endorectal and body phased-array coils and a dynamic contrast-enhanced inversion-prepared dual-contrast gradient-echo sequence (temporal resolution, 1.65 seconds). Perfusion and blood volume maps were generated with a sequential three-compartment model. A total of 110 regions (62 prostate cancer, 48 normal prostate tissue) and their MRI perfusion and blood volume hotspots were analyzed and correlated with histologic mean vessel density and mean vessel area. RESULTS. In patients with prostate cancer, median perfusion in the entire region was 0.71 mL/cm(3) min(-1) (hotspot, 1.53 mL/cm(3) min(-1)) with a median blood volume of 1.06% (hotspot, 2.23%). In the corresponding histologic areas, median mean vessel density was 77 vessels/mm(2) (hotspot, 156 vessels/mm(2)) with a median mean vessel area of 1.61% (hotspot, 2.50%). In normal prostate tissue, median perfusion was 0.33 mL/cm(3) min(-1) (hotspot, 1.38 mL/cm(3) min(-1)) with a median blood volume of 0.62% (hotspot, 2.6%). In the corresponding histologic regions, median mean vessel density in the entire area was 57 vessels/mm(2) with a median mean vessel area of 1.21% (no hotspots). MRI perfusion in the entire region was the most suitable parameter for differentiating prostate cancer and facilitated correct classification in 61.8% of cases (blood volume hotspots, 58.2%; perfusion hotspots, 56.4%). CONCLUSION. In differentiation of cancerous and normal prostate tissue, use of perfusion in the entire region is superior to use of perfusion and blood volume in MRI hotspots.