Introduction

Multiparametric MRI (mpMRI) have become the standard tool to diagnose prostate cancer (PCa). However, mpMRI suffers from high false-positive rate,1primarily due to overlapping apparent diffusion coefficient (ADC) of PCa, inflammation, and benign prostatic hyperplasia (BPH). We examined the heterogeneous pathologies within mpMRI-determined PCa region using a recently developed Diffusion MRI Histology (D-Histo) method, i.e.,a modification of diffusion basis spectrum imaging (DBSI).2-4 Specifically, D-Histo models diffusion-weighted MRI signals of prostate as an anisotropic diffusion tensor reflecting stromal tissue (Fig. 1F),and a spectrum of isotropic diffusion tensors of highly-restricted (modeling prostatitis, Fig. 1D), restricted (modeling PCa, Fig. 1E) and non-restricted (normal prostate, Fig. 1G) isotropic diffusion tensors, respectively.

Method and materials

Patients: One-hundred and seventy-eightpatients with clinical suspicion for prostate cancer (PSA> 4 ng/ml) were recruited for this study. The study was approved by the local Institutional Review Board. MRI: In vivo DWI data was obtained on a Siemens 3T Skyra scanner (Erlangen, Germany) with an 18-channel phased-array body receive coil in Changhai Hospital, Shanghai, China. The imaging parameters were: TR 5000 ms, TE 88 ms, 4 averages, FOV 112 x 140 mm2, in-plane resolution 2 x 2 mm2, 24 slices at 4-mm thick, 25-dir icosahedral diffusion encoding scheme with maximum b-value 1500 s/mm2. Standard mpMRI were obtained in all patients. Image analysis: DWI data were analyzed with D-Histo and conventional DTI analysis using an in-house Matlabsoftware. PCa foci, stromal BPH, epithelial BPH and benign peripheral zone tissue were identifiedby an experienced radiologist based on mpMRI and applied to DTI and D-Histo metrics maps.

Results and discussions

104 out of 178 patients with positive mpMRI finding or PSA > 10 ng/ml were scheduled for biopsy. 74 patients were considered as non-cancerous by negative mpMRI and PSA < 10 ng/ml. 56 of the 104 biopsy patients were PCa-positive and 48 were negative. Among the PCa negative patients, all were BPH-positive while 50% were both BPH and prostatitis positive. This suggests BPH and/or prostatitis could lead to false-positive PCa diagnosisby mpMRI. We identified 222 tumor foci from the 56 biopsy proven PCa patients at various locations within the prostate gland: 89 at peripheral zone (PCaPZ),66 at transition zone (PCaTZ) and 67 at both peripheral and transition zone (PCaPZ + TZ). We then identified39 stromal BPH regions, 34 epithelial BPH and 59 benign peripheral zone tissues from thosebiopsy-negative patients. The ADC values of PCa PZ (0.94± 0.23 mm2/ms), PCaTZ (0.83± 0.18mm2/ms) and PCaPZ+TZ (0.73 ± 0.14 mm2/ms) were significantly lower (p < 0.0001) than the ADC values of benign peripheral zone tissue (2.13 ± 0.20 mm2/ms) and epithelial BPH (1.65 ± 0.25 mm2/ms). The ADC value of stromal BPH (1.01 ± 0.14 mm2/ms) was not different from that of PCa PZ tissue (0.94± 0.23 mm2/ms). This could contribute to the false-positive PCa diagnosis by mpMRI, ~46% in the current study. Despite the high rate of false-positive PCa detection, mpMRI identified PCa regions should still contain high extent of tumor in biopsy-positive patients. Thus, we examined the putative D-Histo PCa marker, the restricted isotropic diffusion fraction,in PCa PZ (0.18 ± 0.10), PCa TZ (0.23 ± 0.09) and PCa PZ+TZ tissue (0.29 ± 0.09) comparing with thoseof PZ benign tissue (0.01 ± 0.02), epithelial BPH (0.04 ± 0.03) and stromal BPH (0.14 ± 0.07). Indeed, mpMRI identified PCa region exhibits higher restricted-isotropic-diffusion fraction in the biopsy-positive PCa patients. Stromal BPH is commonly seen in the vicinity of PCa. We also examined the extent of anisotropic diffusion fraction in PCa regions (PZ, 0.23 ± 0.14; TZ, 0.26 ± 0.10; PZ + TZ, 0.23 ± 0.05) and stromal BPH (0.30 ± 0.13) from non-PCa subjects. Results showed that anisotropic fraction in these tissues was higher than that of benign PZ tissue (0 ± 0.01). For highly-restricted isotropic diffusion fraction (putative inflammation marker), PCa tissues (PZ, 0.06 ± 0.02; TZ, 0.06 ± 0.03; PZ+TZ, 0.07 ± 0.03) and stromal BPH (0.04 ± 0.03) all exhibited higher content than benign PZ tissue (0.02 ± 0.02). The mpMRI-based PCa regions contain all three D-Histo classifications: PCa (restricted diffusion), inflammation (highly-restricted diffusion), and stroma (anisotropic diffusion). In contrast, mpMRI identified benign prostate was free from these complications. D-Histo results demonstrated the underpinning of false-positive PCa diagnosis using mpMRI.

Conclusion

Our results support that underlying prostate pathologies, including BPH and inflammation, contribute to mpMRI false-positive identification of PCa. If quantitatively validated, D-Histo-derived restricted fraction could distinguish PCa from benign tissue, epithelial and stromal BPH, and inflammation to improve PCa detection.

Acknowledgements

No acknowledgement found.