Application of proton density fat fraction imaging in risk stratification of prostate cancer.

Abstract

Background: Prostate cancer (PCa) as one of the most prevalent malignancies in men. We introduced a non-invasive quantitative measurement of intraprostatic fat content based on magnetic resonance proton density fat fraction (PDFF) imaging. The study aims to determine the fat fraction (FF) of PCa using proton density magnetic resonance imaging (MRI), gather clinical and routine MRI characteristics, and identify risk factors for high-risk PCa through multifactorial logistic regression.

Methods: Clinical and imaging data from 191 pathologically confirmed PCa patients were collected. Patients were stratified based on Gleason score (GS), with 63 in the intermediate- and low-risk group (GS =3+3, 3+4) and 128 in the high-risk group (GS ≥4+3). All patients underwent routine prostate MRI and FF imaging. Clinical and imaging data related to PCa were analyzed, including age, body mass index (BMI), prostate volume (PV) measured by MRI, smoking history, alcohol history, diabetes history, serum prostate-specific antigen (PSA) level, apparent diffusion coefficient (ADC) value, T2 signal intensity (T2SI), Prostate Imaging Reporting and Data System 2.1 (PI-RADS 2.1) score, GS, lesion FF, whole gland FF, periprostatic fat thickness (PPFT), and subcutaneous fat thickness (SFT). Independent risk factors for stratifying PCa risk were identified through multivariate logistic regression analysis, and a predictive model was established. Receiver operating characteristic (ROC) curve analysis was conducted for visual analysis.

Results: Significant differences were found in BMI, PV, PSA, tumor ADC value, standard T2SI, PI-RADS score, lesion FF, and PPFT between low- and medium-risk and high-risk groups (P<0.05). No significant differences were observed in age, smoking history, drinking history, diabetes history, and SFT between the two groups (P>0.05). GS correlated significantly with FF (ρ=0.6, P<0.001), PSA (ρ=0.432, P<0.001), ADC value (ρ=-0.379, P<0.001), and PI-RADS (ρ=0.366, P<0.001). Multiple logistic regression analysis revealed that an increase in FF, a PI-RADS score increase of 5 points, and a decrease in ADC value and PV were independent predictors of high-risk PCa (P<0.05). The ROC curve showed that the best cut-off value for the model was 0.67, with an area under the curve (AUC) of 0.907, sensitivity of 78.1%, and specificity of 88.9%.

Conclusions: The FF of PCa determined by proton density MRI is significantly associated with GS, serving as an independent predictor of high-risk PCa.