Defining diagnostic criteria for ductal prostate cancer on prostatic MRI.

Abstract

e17532 Background: Ductal prostate adenocarcinoma (DAC) is an aggressive histologic variant of prostate cancer (PCa) which often be missed due to their low PSA secretion. Further, a large proportion of DACs have extra-prostatic extension and nodal disease at presentation warranting accurate diagnosis and treatment planning. However studies have yet to differentiate DACs from high grade acinar PCas (PAC) on MRI. Therefore we aimed to develop MRI criteria to identify DACs and assess its diagnostic accuracy. Methods: Patients with histologically proven DAC who had MRIs prior to RP were identified from January 2011 to November 2018. Histology-based MRI diagnostic criteria were developed using RP specimens from nine patients with a pure dominant DAC focus and corresponding MRIs. Sixty-eight DAC patients were compared to a matched cohort of 70 patients with Gleason Score 8 or 9 PAC using the pre-defined MRI criteria. Chi-Squared, T tests, Mann Whitney U tests and sensitivity analyses were performed. Results: The following features of DAC were defined on MRI after correlation with histology: 1) intermediate T2 signal 2)well-circumscribed 3) lobulated tumor and 4) a dark peripheral rim. Majority of DACs were lobulated (79.4% vs 5.7%), with a dark peripheral rim on T2 weighted imaging (55.9% vs 4.3%) and had ≥3 MRI features compared to PAC (73.6% vs 7.2%) (all p < 0.001). Moreover, a higher proportion of pure DACs were lobulated (100% vs 5.7%), had a dark peripheral rim (94.7% vs 4.3%) and ≥3 MRI features (100% vs 7.2%) compared to PAC (all P < 0.001). There were no differences in median T2 contrast enhancement, ADC values or ADC ratios between the groups. Using our criteria MRI demonstrated sensitivity of 73.5%, specificity of 92.9 %, PPV of 90.9%, and NPV of 78.3% in diagnosing DACs if ≥3 features were present. In the diagnosis of pure DACs, MRI demonstrated sensitivity of 100%, specificity of 92.9%, PPV of 95.2%, and NPV of 100%. The area under the curve (AUC) for the diagnosis of all DACs was 0.81 and 0.98 for pure DACs. Conclusions: The presence of ≥3 features (well-circumscribed, lobulations and a dark peripheral rim and intermediate signal on the T2 phase) on prostatic MRI can help differentiate DAC from PAC. While this is the largest cohort of DACs to be analyzed, further studies are needed to validate these findings.