Utility of diffusional kurtosis imaging as a marker of adverse pathologic outcomes among prostate cancer active surveillance candidates undergoing radical prostatectomy.

The purpose of this study was to compare findings at nongaussian diffusional kurtosis imaging and conventional diffusion-weighted MRI as markers of adverse pathologic outcomes among prostate cancer patients who are active surveillance candidates and choose to undergo prostatectomy. Fifty-eight active surveillance candidates (prostate-specific antigen concentration, < 10 ng/mL; clinical tumor category less than T2a; Gleason score, 3 + 3; ≤ 25% of biopsy cores positive for tumor; ≤ 50% tumor involvement of any individual core; ≤ 20% tumor involvement across all cores) who underwent prostatectomy and preoperative 3-T MRI including diffusional kurtosis imaging (b values, 0, 500, 1000, 1500, and 2000 s/mm(2)) were included. Adverse pathologic features at prostatectomy were defined using two schemes of varying stringency. One scheme (less stringent) was presence of a Gleason score greater than 6 or extracapsular extension (n = 19). The other scheme (more stringent) was presence of a Gleason score greater than 6, extracapsular extension, or an index tumor 10 mm or larger (n = 35). Parametric maps displaying standard apparent diffusion coefficient (ADC), kurtosis (K) representing nongaussian diffusion behavior, and diffusion (D) representing a diffusion coefficient adjusted for nongaussian (kurtosis) behavior were reviewed, and the most abnormal region was recorded for each metric. Associations between these metrics and the presence of adverse final pathologic findings were assessed with unpaired Student t tests and receiver operating characteristic analyses. For both schemes, only D was significantly lower in patients with adverse final pathologic findings (p = 0.006, p = 0.025). K tended to be greater in patients with adverse final pathologic findings for the more stringent scheme (p = 0.072). ADC was not significantly different in the presence of adverse final pathologic findings for either scheme (p = 0.357, p = 0.383). With either scheme, D had a larger area under the receiver operating characteristics curve (AUC) for predicting adverse final pathologic results (AUC, 0.691 and 0.743) than did ADC (AUC, 0.569 and 0.655) or K (AUC, 0.617 and 0.714), but the difference was not significant (p = 0.183, p = 0.734). Preliminary results suggest that diffusional kurtosis imaging findings may have more value than findings at conventional diffusion-weighted MRI as a marker of adverse final pathologic outcome among active surveillance candidates.