Quantitative Pharmacokinetic Analysis of Dynamic Contrast-Enhanced Perfusion in Multiparametric Prostate MRI: An...

Background/Objectives: To quantitatively evaluate pharmacokinetic parameters derived from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in differentiating prostate malignant lesions from normal appearing prostatic tissue, and to analyze their correlation with Gleason scores (GS). Methods: A retrospective analysis was performed on 51 patients with initial biopsy-confirmed prostate cancer who underwent multiparametric MRI (mpMRI) between May 2021 and April 2024 for elevated serum prostate-specific antigen (PSA) levels and/or suspicious digital rectal examination (DRE) findings. Forty-three eligible patients with histopathologically confirmed prostate adenocarcinoma (GS ≥ 6) who met all strict imaging criteria were included. Demographic parameters, serum PSA, prostate volume, PSA density (manual: PSA-D_m, workstation: PSA-D_w), and lesion dimensions were recorded. Free-hand 2D regions of interest (ROIs) were placed on index lesions and normal peripheral zone tissue on a dedicated workstation (1.5-T MRI, syngo.via Tofts model) to compute quantitative and semi-quantitative pharmacokinetic parameters (K^trans, K_ep, iAUC, V_e, χ²). Statistical evaluations utilized non-parametric tests, Spearman's rank correlation, and clustered ROC analysis with patient-level bootstrapping. Results: Mean patient age was 64.21 ± 8.82 years. Significant differences were identified across GS groups regarding serum PSA, PSA-D_m, and PSA-D_w (p < 0.05). Group 3 (GS ≥ 8) demonstrated significantly higher median PSA (12.99 ng/mL) compared to Groups 1 and 2 (p = 0.004). For predicting high-risk disease (GS ≥ 8), ROC analysis for PSA (cut-off 7.25 ng/mL) yielded an area under the curve (AUC) of 0.798, whereas K^trans performed poorly (AUC = 0.560). Conversely, for intra-patient paired comparisons distinguishing malignant lesions from normal background parenchyma, all quantitative DCE-MRI parameters (K^trans, K_ep, iAUC, V_e, χ²) were significantly higher in malignant lesions (all p < 0.001, surviving stringent Bonferroni correction). Among pharmacokinetic parameters, K^trans (cut-off: 0.105, AUC = 0.853, sensitivity: 81.4%, specificity: 76.7%) and χ² (cut-off: 0.049, AUC = 0.844, sensitivity: 79.1%, specificity: 76.7%) exhibited the highest discriminative performance for identifying malignancy. Crucially, however, DCE-MRI parameters demonstrated no statistically significant correlation with Gleason scores (all p > 0.05) and failed to independently predict high-grade disease. Conclusions: Quantitative DCE-MRI pharmacokinetic parameters effectively differentiate malignant prostate lesions from normal parenchyma within this specific cohort, but they do not correlate with histological grade. Furthermore, the derived diagnostic cut-offs are highly protocol-dependent, reflecting 1.5-T hardware and 2D ROI sampling. Robust cross-platform standardization and prospective 3D whole-gland validations are strictly required before these functional metrics can be reliably incorporated into clinical workflows.