Differentiation of central gland prostate cancer from benign prostatic hyperplasia using monoexponential and...

To investigate biexponential apparent diffusion parameters of prostate central gland (CG) cancer, stromal hyperplasia (SH), and glandular hyperplasia (GH) and compare with monoexponential apparent diffusion coefficient (ADC) value for discriminating prostate cancer from benign hyperplasia.

Twenty-one CG cancer foci, 23 SH and 26 GH nodules in the CG were analyzed in 39 patients (19 with CG cancer, 20 with peripheral zone cancer but no CG cancer) who underwent preoperative conventional DWI (b-value 0, 1000s/mm(2)) and a 10 b-value (range 0 to 3000s/mm(2)) DWI. All of the cancer and hyperplastic foci on MR images were localized on the basis of histopathologic correlation. The ADC value of the monoexponential DWI, and the fast apparent diffusion coefficient (ADCf), slow apparent diffusion coefficient (ADCs) value and the fraction of ADCf (f) of the biexponential DWI were calculated for all of the lesions. Receiver operating characteristic (ROC) analysis was performed for the differentiation of CG cancer from SH and GH.

The ADC values (×10(-3)mm(2)/s) were 0.87±0.11, 1.06±0.15, and 1.61±0.27 in CG cancer, SH and GH foci, respectively, and differed significantly, yielding areas under the ROC curve (AUCs) of 1.00 and 0.80 for the differentiation of carcinoma from GH and SH, respectively. The ADCf (×10(-3)mm(2)/s), ADCs (×10(-3)mm(2)/s) and f for cancer were 1.92±0.38, 0.53±0.17, and 47.7±6.1%, respectively, which were lower than the same values for GH (3.43 ±0.65, 1.12±0.21, 61.1±8.7%) (all p<0.01). The ADCf and ADCs for cancer were also lower than those for SH (3.11±0.30, 0.79±0.21) (all p<0.01). The ADCf yielded AUCs (1.00, p>0.01) that were comparable to those from ADC for the differentiation of cancer from GH, while ADCf yielded higher AUCs (0.92) compared with ADC (p<0.01) for the differentiation of cancer from SH. ADCs and f revealed AUCs of 0.97 and 0.90, respectively, for the differentiation of cancer from GH, and the ADCs offered relatively lower AUCs (0.68) for differentiating cancer from SH.

Biexponential DWI could potentially improve the differentiation of prostate cancer in CG, and the ADCf of the biexponential model offers better accuracy than ADC.