INTRODUCTION AND OBJECTIVE:
The accuracy of multi-parametric MRI (mpMRI) in pre-operative staging and recommending nerve-sparing in radical prostatectomy (RP) remains controversial. There is considerable heterogeneity in the performance and the reliability of mpMRI in predicting extra-prostatic extension (EPE), with reduced sensitivity identified as a key limitation in previous systematic reviews. The objective of this study was to evaluate the ability of expert genitourinary radiologists at a high-volume RP centre to predict prostate cancer EPE on a side-specific basis using a Likert scale.
We included 124 men who underwent robot-assisted RP (RARP) as part of the NeuroSAFE PROOF study at our centre (NCT03317990). Three genitourinary radiologists retrospectively reviewed mpMRI blinded to prostatectomy pathology results, assigned a Likert score of suspicion for EPE (1-5) for each side of the prostate on the basis of sequential MRI information and then after clinical information was revealed. We present the detection rates for EPE at each level of the Likert scale, test performance when Likert score is dichotomised, and the value of additional imaging and clinical information in EPE prediction.
A total of 231 lobes from 121 men (mean age 56.9 years±7) were evaluated. 39 men (32.2%), or 43 lobes (18.6%) had EPE. 47.7% of lobes were assigned a Likert score 1-2 (i.e. had a negative diagnostic test). Detection rates of EPE increased with each increment in the Likert score assigned (Table 1, p<0.001). For detection of EPE, Likert score ≥3 had sensitivity, specificity, NPV, PPV of 90.2%, 52%, 96%, 29.1%, respectively. Kappa coefficient for agreement between radiologists on the final Likert was 0.62, 0.64, and 064 between readers 1&2, 1&3, and 2&3, respectively, indicating substantial agreement. Diagnostic accuracy improved as MRI sequences and clinical information were revealed. The AUCs for T2WI alone, +DWI, +DCE, +PSA, and +biopsy information, were 0.612 (CI; 0.55-0.674), 0.683 (CI: 0.624-0.741), 0.757 (CI: 0.704-0.810), 0.784 (CI: 0.734-0.833) and 0.813 (CI: 0.77-0.86), respectively.
Likert score is a practical and simple system that can be used effectively by experienced readers to rule out EPE. We report high sensitivity, which is important as the clinical imperative to not miss EPE in order to provide best chance cure with RP. Diagnostic accuracy of EPE prediction using mpMRI improves with additional MRI sequences and with sight of clinical parameters, which emphasises the importance of biopsy information and the multidisciplinary team when using mpMRI for prostate cancer staging.
Source of Funding:
NeuroSAFE PROOF trial funded by a charitable grant from JP Moulton Foundation