RPS 2607 - Prostate cancer staging and management: where are we today?

Author Block: M. Boschheidgen, L. Schimmöller, T. Ullrich, C. Arsov, G. Antoch, H. Wittsack; Düsseldorf/DE
Purpose or Learning Objective: To determine the capability of gadolinium-free arterial spin labelling (ASL) sequence as contrast-free, non-invasive alternative perfusion imaging method to differentiate prostate cancer (PCA) from benign prostate tissue compared to conventional DCE MRI.
Methods or Background: Thirty men with histologically confirmed PCA were included in this prospectively enrolled single-centre cohort study. All patients received multiparametric MRI (T2, DWI, DCE) at 3T with additional ASL of the PCA lesion. The primary endpoint was the differentiability of PCA versus normal prostate tissue in ASL in comparison to DCE. Secondary objectives were differences in signal intensities (SI), contrast ratios (CR), and differences in the attenuation pattern of peripheral (PZ) and transition zone (TZ) PCA.
Results or Findings: In both ASL and DCE, the average SI of PCA areas differed significantly from SI in reference areas in the TZ and PZ (p<0,01, respectively). ASL had significantly higher CR discerning PCA and benign tissue in PZ and TZ (PZ=5.2; TZ=6.5) compared to DCE (PZ=1.6; TZ=1.4) (p<0.01, respectively). In subjective evaluation, ASL could visualise PCA in 28 patients, compared to 29 in DCE.
Conclusion: ASL had significantly higher contrast-ratios discerning PCA and benign tissue in PZ and TZ compared to DCE and visual discrimination of PCA does not differ significantly between the two sequences. As perfusion gadolinium-based contrast media is seen more critical in the last few years, ASL seems to be a promising alternative to DCE in PCA detection.
Limitations: Single-centre design. Small sample size. Single slice sequence. Long acquisition time.
Ethics committee approval: The study was approved by the local ethics committee.
Funding for this study: No funding was received for this study.

Author Block: N. Sushentsev¹, I. Caglič¹, A. Colarieti², A. Warren¹, B. Lamb¹, N. Shah¹, T. Barrett¹; ¹Cambridge/UK, ²Milan/IT
Purpose or Learning Objective: To retrospectively determine the prevalence and diagnostic performance of the capsular enhancement sign (CES) on dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) for the detection of prostate cancer (PCa) extracapsular extension (ECE).
Methods or Background: This retrospective study included patients who underwent DCE-MRI prior to radical prostatectomy. CES was defined as an area of asymmetrical early hyperenhancement on DCE-MRI that was adjacent to a peripheral zone tumour, matched or exceeded the tumour circumferential diameter, and persisted beyond the washout of contrast within the adjacent tumour. Two expert uro-radiologists evaluated the presence of CES on DCE-MRI, independently and then in consensus, with the interobserver agreement calculated using a bias-adjusted and prevalence-adjusted kappa (PABAK). CES diagnostic performance for prediction of ECE was assessed using sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).
Results or Findings: The study included 146 patients, of whom 91/146 (62%) harboured ECE on final surgical pathology. Following the initial review of the images, Reader 1 called 12/146 (8%) CES-positive cases, while Reader 2 reported 14/146 (10%) CES-positive cases, and a total of 15/146 (10%) lesions were subsequently identified as CES-positive following a consensus read. PABAK for CES between the two readers was high at 0.90. All consensus determined CES-positive lesions represented pathological ≥T3a disease, with the overall prevalence of CES among tumours with confirmed ECE being 15/91 (17%). Hence, whilst showing 100% specificity and PPV for ECE detection, CES had sensitivity, NPV, and accuracy of 16.5%, 41.29%, and 47.38%, respectively.
Conclusion: The presence of CES on DCE-MRI is highly predictive for the presence of ECE and may improve local staging in the small but significant percentage of patients in whom it is demonstrated.
Limitations: Not applicable.
Ethics committee approval: This study was funded by the NREC East of England.
Funding for this study: Not applicable.

Author Block: Y. KIM, S. H. Kim, H. PARK, T. Baek; Busan/KR
Purpose or Learning Objective: To compare the diagnostic performance in tumour detection and inter-observer agreement between acquired diffusion-weighted imaging (aDWI) b2000 and computed DWI (cDWI) b2000 for patients with prostate cancer (PCa).
Methods or Background: A total of 88 patients (mean age: 68.6 years, range: 47-82 years) who had been diagnosed with PCa by radical prostatectomy and undergone pre-operative 3.0-Tesla magnetic resonance imaging (3T-MRI) including DWI (b values, 0, 100, 1000, 2000 s/mm2) were included in this study. cDWI b2000 was made from aDWI b0, 100 and 1000 under a mono-exponential decay model. Two independent reviewers performed a 4-week-interval review of aDWI b2000 images and then cDWI b2000 in random order for each session. T2-weighted images were presented for both sessions. A region of interest was drawn for an index tumour on each dataset, and a PIRADS score based on PIRADS v2.1 was recorded. Topographic maps served as the reference standard. The McNemar test was performed to compare the sensitivities for tumour detection, and kappa statistics were used to evaluate the inter-observer agreement on the PIRADS score on each dataset.
Results or Findings: The study population consisted of Gleason score (GS) 6 (n=16), GS 7 (n=53), GS 8 (n=9) and GS 9 (n=10) patients. For both reviewers, the sensitivities of cDWI b2000 and aDWI b2000 for detection of PCa showed no significant difference (for reviewer 1, both 94% (83/88); for reviewer 2, both 90% (79/88), P = 1.000, respectively). The kappa values of cDWI b2000 and aDWI b2000 for the PIRADS scores were 0.422 (95% CI, 0.240-0.603) and 0.495 (95% CI, 0.308-0.683), respectively.
Conclusion: cDWI b2000 showed comparable diagnostic performance and sustained moderate inter-observer agreement with aDWI b2000 for detection of PCa.
Limitations: No limitations were identified.
Ethics committee approval: This study was approved by the Pertinent institutional review board.
Funding for this study: No funding was received for this study.

Author Block: A. M. Hötker¹, U. J. Mühlematter¹, S. Skawran¹, S. Ghafoor¹, I. A. Burger², M. Huellner¹, O. F. Donati¹; ¹Zurich/CH, ²Baden/CH
Purpose or Learning Objective: To predict the presence of PET-positive pelvic lymph nodes in prostate cancer using quantitative parameters of multiparametric MRI (mpMRI) and clinical information.
Methods or Background: This study included 35 patients with high suspicion for prostate cancer undergoing multiparametric prostate MRI and PSMA-PET/CT prior to MRI-guided biopsy. All MRI examinations were assessed by a radiologist, and the Apparent Diffusion Coefficient (ADC, mean and volume), capsular contact length, volume and maximal diameter on T2-weighted sequences and parameters of dynamic contrast-enhanced MRI (iAUC, kep, Ktrans, ve) were calculated for the index lesion. Clinical data was extracted from the hospital information system to calculate the Briganti 2018 nomogram scores. PET examinations were evaluated by two board-certified nuclear medicine physicians and served as the standard of reference.
Results or Findings: Quantitative imaging parameters of mpMRI mostly demonstrated mediocre to good performance in prediction of PET-positive nodes (AUCs, ADCmean: 0.74, ADCvol: 0.55, iAUC: 0.42, kep: 0.71, Ktrans: 0.64, ve: 0.37, T2capsular: 0.59, T2diameter: 0.58, T2vol: 0.55), while the Briganti 2018 nomogram (including maximum diameter of the index lesion) reached an AUC of 0.78 (95%-CI: 0.61-0.95). Quantitative MR parameter did not provide added value to the Briganti 2018 model alone.
Conclusion: The Briganti 2018 model, which includes clinical/pathological data and the maximal tumour length of the index lesion on prostate MRI, performed well in predicting PET-positive lymph nodes and may serve as a tool to stratify patients for primary staging using PSMA-PET.
Limitations: The relatively low number of patients. Retrospective study design.
Ethics committee approval: Approved by local IRB.
Funding for this study: Not applicable.

Author Block: N. Pötsch¹, E. Rainer¹, P. Clauser¹, G. Vatteroni², T. H. Helbich¹, P. A. Baltzer¹; ¹Vienna/AT, ²Milan/IT
Purpose or Learning Objective: To test the inter-reader agreement of the Prostate Imaging Quality (PI-QUAL) score for multiparametric prostate MRI and its impact on diagnostic performance.
Methods or Background: Prebioptic multiparametric (T2-weighted, DWI and DCE) prostate MRIs (mpMRI) of 50 patients undergoing transrectal ultrasound-guided MRI-fusion and systematic prostate biopsy were included. Two radiologists independently assigned a PI-QUAL score to each patient. PI-RADS categories were assigned in a lesion-based approach, dividing the prostate into six regions (left and right: base/midglandular/apex). Additionally, the diagnostic quality of each sequence was evaluated independently. Inter-reader agreement was calculated by using Cohen’s kappa and diagnostic performance was compared by the area under the ROC curve (AUC).
Results or Findings: In a total of 274 diagnostic areas, the malignancy rate was 62.2% (22.7% clinically significant prostate cancer ISUP ≥ 2). Inter-reader agreement for the diagnostic quality was only slight for T2w (kappa 0.19) and fair for DWI and DCE (kappa 0.23 and 0.29). For PI-QUAL as such, the inter-reader agreement was moderate (kappa 0.51). For PI-RADS category assignments, the inter-reader agreement was almost perfect (kappa 0.86). Overall diagnostic performance was significantly better in studies with a PI-QUAL score > 3 compared to a score ≤ 3 (P=0.03; AUC 0.805 and 0.839).
Conclusion: In conclusion, the diagnostic performance of mpMRI for the detection of prostate cancer does depend on MRI image quality. Though there is room for improvement regarding inter-reader reproducibility, PI-QUAL is a tool that provides value by predicting the accuracy of diagnostic mpMRI results.
Limitations: As all patients underwent MRI-US fusion biopsy due to suspicious mpMRI findings, the rate of malignancy is higher compared to routine clinical practice which may bias the outcomes in non-selected patients.
Ethics committee approval: IRB waived the need for informed consent.
Funding for this study: No funding was received for this study.

Author Block: H. Engel, B. Oerther, M. Reisert, A. Sigle, E. Kellner, F. Bamberg, M. Benndorf; Freiburg im Breisgau/DE
Purpose or Learning Objective: To analyse whether low PI-RADS categories of transition zone (TZ) lesions can be downgraded based on mean apparent diffusion coefficients (mADC) without risking false-negative results.
Methods or Background: This retrospective cohort study consists of consecutive patients with TZ lesions in multiparametric prostate MRI and subsequent MRI-ultrasound-fusion-biopsy between 05/2017-04/2020. Patients with known prostate cancer (PCa) are excluded. All lesions are scored by two blinded radiologists according to PI-RADSv2.1 guidelines. To determine mADC, lesions are manually segmented. Regression and ROC analyses are performed to determine the diagnostic performance of PI-RADSv2.1 categories and mADC.
Results or Findings: Among 85 patients with 98 TZ lesions, 33 (33.7%) are PCa and 65 (66.3%) are benign. 24 (72.7%) of the 33 PCa lesions are clinically significant (csPCa, ISUP-grade>1). mADC for PCa are significantly lower than for benign lesions (894 vs 1.091µm2/s, p<0.001). AUC values from ROC analysis with csPCa as outcome variable are 0.916 for PI-RADSv2.1 and 0.806 for mADC. Compared to PI-RADSv2.1 alone, a combination with an mADC cut-off of 950µm2/s for TZ lesions ≤ PI-RADS 3 improves the negative predictive value (0.95 vs 1.00). Among 58 TZ lesions ≤ PI-RADS 3, only 8 (13.8%) have mADC below 950µm2/s, 3 (37.5%) PCa and 1 (12.5%) csPCa.
Conclusion: The key question after a prostate MRI is whether a biopsy is indicated: cancer detection rates for PI-RADS 1+2 are very low while being too high for PI-RADS 3 to abandon biopsies completely. Thus, further parameters which allow avoiding unnecessary biopsies are desirable. Our data indicate that by applying an mADC cut-off for TZ lesions ≤ PI-RADS 3 most biopsies could be avoided without overlooking prostate cancer.
Limitations: mADC can differ between vendors/algorithms. External validation of our findings is warranted before clinical use.
Ethics committee approval: The ethics committee approval was obtained.
Funding for this study: No funding was received for this study.

Author Block: A. M. Hötker, A. Tiessen, D. M. Raffaele, E. Konukoglu, O. F. Donati; Zurich/CH
Purpose or Learning Objective: To develop and validate an artificial intelligence algorithm to decide on the necessity of dynamic contrast-enhanced sequences (DCE) in prostate MRI.
Methods or Background: A convolutional neural network (CNN) was developed on 300 prostate MRI examinations. The consensus of two expert readers on the necessity of DCE acted as the reference standard. The CNN was validated in a separate cohort of 100 prostate MRI examinations from the same vendor and 31 examinations from a different vendor. Sensitivity/specificity were calculated using ROC curve analysis and results were compared to decisions made by a radiology technician.
Results or Findings: The CNN reached a sensitivity of 94.4% and specificity of 68.8% (AUC: 0.88) for the necessity of DCE, correctly assigning 44%/34% of patients to a biparametric/multiparametric protocol. In 2% of all patients, the CNN incorrectly decided on omitting DCE. With a technician reaching a sensitivity of 63.9% and specificity of 89.1%, the use of the CNN would allow for an increase in sensitivity of 30.5%. The CNN achieved an AUC of 0.73 in a set of examinations from a different vendor.
Conclusion: The CNN would have correctly assigned 78% of patients to a biparametric or multiparametric protocol, with only 2% of all patients requiring re-examination to add DCE sequences. Integrating this CNN in clinical routine could render the requirement for on-table monitoring obsolete by performing contrast-enhanced MRI only when needed.
Limitations: The decision rendered by the neural network was dichotomous. The performance of the AI could be improved by defining a range of probability values in which it's unsure and prompt the technician to call a radiologist for this particular examination.
Ethics committee approval: This study was approved by the institutional review board and the requirement for study-specific informed consent was waived.
Funding for this study: Not applicable.