Multiparametric Magnetic Resonance Imaging

A Delphi consensus process was undertaken to formulate recommendations regarding three different areas in the diagnostic MRI pathway of PCa: (1) image quality assessment of mpMRI; (2) interpretation and reporting of mpMRI; and (3) reader experience and training requirements. The Delphi method is a technique of structured and systematic information gathering from experts on a specific topic using a series of questionnaires [13 (link)]. In this study, the diagnostic role of mpMRI in biopsy-naïve men with a suspicion of PCa was considered.
The Delphi process was carried out in four phases (Fig. 1). (1) Panellists from the European Society of Urogenital Radiology (ESUR) and EAU Section of Urologic Imaging (ESUI) were selected based on expertise and publication record in the PCa diagnosis, and on their involvement in guideline development. (2) A questionnaire was created with items that were identified by a subcommittee of the ESUR, based on the statements from a recent UK consensus paper on implementation of mpMRI for PCa detection [14 (link)]. (3) Panel-based consensus findings were determined using an online Delphi process. For this purpose, an internet survey was generated and sent by email to the members of the group (created in Google Forms). In the second round, a reminder to complete the questionnaires was sent by email. The panellists anonymously completed two rounds of a questionnaire consisting of 39 items (including 55 subquestions). Based on the knowledge of the entire group’s responses in the first round, second round voting was performed. Outcomes of the multiple-choice and open questions were graphically displayed, so the results could be reflected before selecting a response in the second round. For inclusion in the final recommendations, each survey item required to have reached group consensus by the end of the two survey rounds. (4) The items of the questionnaires were analysed, and consensus statements were formulated based on the outcomes. In total, 31 of 55 items were rated for agreement on a 9-point Likert scale.Fig. 1

Details of the stages of the Delphi process (flow chart)

An item scored as ‘agree’ (score 7–9) by ≥ 70% of participants and disagree (score 1–3) by ≤ 15% constituted ‘consensus agreement’ for an item. An item scored as ‘disagree’ (score 1–3) by ≥ 70% of participants and agree (score 7–9) by ≤ 15% was considered as ‘consensus disagreement’. The other items (24 of 55) were multiple-choice or open questions and were presented graphically. For the multiple-choice or open questions to reach consensus, a panel majority scoring of ≥ 50% was required.

Data were reported as medians, interquartile ranges (IQRs) and frequencies. A correlation between PI-RADS 3 lesions on mpMRI and fusion-TB results as in a per-lesion analysis was made. The diagnostic performance was evaluated using sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). The Mann–Whitney U test was used. The AUC was computed together with the 95% confidence interval (95% CI) and the asymptotic test for null hypothesis: true area=0.5. The best cut-off was calculated using maximisation of the Youden Index. Univariate and multivariate analyses were carried out using logistic regression. Odd ratios (ORs) were calculated together with their 95% CIs. The forward stepwise likelihood ratio method was used for variable selection in the multivariate analysis. The AUC was computed to evaluate the predictive accuracy of the multivariate model. All the tests were 2-tailed, and a p value of <0.05 was considered statistically significant. All the statistical analyses were carried out using IBM SPSS 25.0 (SPSS Inc., Armonk, NY, USA).

All the PI-RADS 3 lesions were biopsied transrectally by two experienced radiologists, who usually read at least 400 mpMRI per year, using the mpMRI-TRUS Fusion image guide feature of “Aplio™ 500 ultrasound system”, after antibiotic prophylaxis and a cleansing rectal enema, using a non-disposable biopsy gun (Medgun, Medax, Modena, Italy) with a disposable 18-gauge needle and a US platform (Canon-Toshiba Aplio 500™, Ōtawara, Kanto, Japan) with an end-fire TRUS probe as previously described (10 (link)).
The biopsy samples were analysed by two dedicated genitourinary pathologists who primarily identified the presence or absence of a neoplastic pathology on the samples. Each neoplastic lesion was graded from 1 to 5 according to the International Society of Urological Pathology (ISUP) Grade Group System (GGS) (12 (link)); lesions having a GGS ≥ 2 were considered to be clinically significant.

The mpMRI examinations were carried out using a 1.5T scanner (Signa HDxt; GE Healthcare, USA) and a pelvic phased-array surface coil combined with a disposable endorectal coil.
The multiparametric study of the prostate gland and seminal vesicles included Fast Relaxation Fast Spin Echo (FR-FSE) T2w, DWI (Diffusion Weighted Imaging) and Dynamic Contrast-Enhanced (DCE) sequences; the scan parameters have already been described in a previous study (9 (link)).
The DWI and DCE images were processed on an independent workstation with dedicated software (Functool, 4.5.5, GE Healthcare, USA) in order to obtain the ADC (Apparent Diffusion Coefficient) map and signal intensity-time (I-T) curve, respectively.
All the PI-RADS 3 lesions were catalogued according to location (peripheral or transition zone) and maximum diameter.
Subsequently, the following parameters were measured on axial FR-FSE T2w sequences as previously described (11 (link)): 1) TransPA, 2) TransCGA, 3) TransPZA, and 4) TransPAI (Figure 2).
The prostate maximum sectional areas on the axial FSE T2w sequences were selected as follows; when the bilateral prostate lobes were basically symmetrical, and the quasi-circular internal urethral sphincter could be seen in the middle of the prostate, the maximum section was that in which the sectional area became smaller when scanning upward or downward.
The contouring of the total prostate area (TransPA) and the central gland area (TransCGA) were manually defined on this selected image; the peripheral zone area (TransPZA) was subsequently calculated.