Purpose

To assess the factors influencing multiparametric (MP) MR Imaging occuracy in estimating the volume of clinically significant prostate cancer (V_h) by using whole-mount step-section slides as standard of reference.

Methods

Thirty-six consecutive patients with biopsy-proven prostate cancer (PCa) were included. All patients received mpMRI of the prostate, including T2WI, DWI, ADC and DCE. Following radical prostatectomy, all specimens were processed as whole-mount step-section slides. MR images were interpreted according to PI-RADS v2 by an experienced radiologist who was blinded to clinical details and histopathology. Volumes of MR lesions were measures on T2WI (V_T2), on DWI (V_DWI), on ADC maps (V_ADC), and on DCE (V_DCE). The Vh and the MR-derived tumor volumes were compared by Mann-Whitney U test, and the correlations between these were computed by using Spearman coefficients. Generalized linear mixed models were used to test the influence on MP MR imaging accuracy of the tumor Gleason score, Vh, PI-RADS score.

Result

Thirty-six consecutive patients with forty-one clinically significant cancers (thirty-one peripheral zone (PZ) prostate cancers and ten transitional zone (TZ) prostate cancers) were included. The V_h had no statistically significant difference with the other mpMRI’s volumes, but not with the V_DCE (P = .012) compared with the V_h of PZ prostate cancers. The V_h of PZ prostate cancers showed strong correlation with V_DCE (R2 = 0.735, P < .0001). The V_h of TZ prostate cancers showed strong correlation with V_T2 (R2 = .733, P = .016). At generalized linear mixed analysis, V_h (P = .015) significantly influenced V_DCE of PZ prostate cancers accuracy. This accuracy was good in tumors with Vh less than 0.5cc. PI-RADS score (P < .0001) and V_h (P < .0001) significantly influenced V_T2 of TZ prostate cancers accuracy. This accuracy was good in tumors with PI-RADS score of 5 or V_h less than 0.5cc.

Discussion

This study divided the tumor into PZ prostate cancers and TZ prostate cancers based on the tumor’s locations, and the result emphasized the need to assess the factors influencing MP MR imaging occuracy before using any MR-derived volume as a surrogate for V_h. As a consequence, this study assessed a model taking into account not only V_DCE of PZ prostate cancers and V_T2 of TZ prostate cancers but also the PI-RADS score of the TZ tumors could improve Vh estimation.

Conclusion

Vh can be estimated by using V_DCE of PZ prostate cancers, and Vh or the tumors with PI-RADS score of 5 can be estimated by using V_T2 of TZ prostate cancers.

Acknowledgements

No acknowledgement found.

References

No reference found.