Prediction of extraprostatic extension on multi-parametric magnetic resonance imaging in patients with anterior prostate cancer

초록

Objectives To validate how established markers of extraprostatic extension (EPE) are applied to anterior prostate cancers (APCs), and to investigate other novel markers if available. Methods Among 614 histopathologically confirmed APCs from 2011 to 2016, 221 lesions with PiRADS (verion 2) scores >= 4 on 3-T multi-parametric MRI were analyzed retrospectively. Two radiologists independently assessed capsular morphology qualitatively with 5-point scale (normal, thinning, bulging, loss, extracapsular disease), and capsule contact length (arc), tumor dimension, and their ratio (arc-dimension ratio) quantitatively. Reproducibility in measurement was assessed with kappa and intra-class correlation coefficients (ICCs). Logistic regression analysis was done to find meaningful indicators of EPE. Diagnostic performance of markers was compared to one another with generalized linear model and multi-reader multi-case ROC analysis. Results Reproducibility was moderate to substantial (kappa 0.45-0.73) for qualitative, and moderate to almost perfect (ICC 0.50-0.87) for quantitative features of EPE. Capsular morphology (odds ratio [OR] 1.818), capsule contact length (OR 1.115), tumor dimension (OR 1.035), and arc-dimension ratio (OR 1.846) were independently associated with EPE (p <= 0.019). Capsular bulging and capsule contact length of 10 mm as thresholds of EPE demonstrated sensitivity/specificity of 0.58/0.85 and 0.77/0.68, respectively. Capsule contact length yielded greatest AUC (0.784), followed by capsular morphology (0.778), arc-dimension ratio (0.749), and tumor dimension (0.741). Diagnostic performance of capsular morphology, capsule contact length, and arc-dimension ratio was comparable in predicting EPE. Conclusions Existing markers of EPE applicable regardless of locations of tumors apply similarly to APCs. Arc-dimension ratio may be a novel marker of EPE of APCs.

키워드