A semi-automated software program to assess the impact of second reads in prostate MRI for equivocal lesions: results from a UK tertiary referral centre. Issue 162 (May 2023)

• Record Type: Journal Article
• Authors: Woernle, Alexandre; Dickinson, Louise; Lelie, Steven; Pendse, Doug; Heffernan Ho, Daniel; Ramachandran, Navin; Kirkham, Alex; Von Stempel, Conrad; Punwani, Shonit; Wah So, Chun; Withington, John; Grey, Alistair; Collins, Thomas; Maffei, Davide; Haider, Aiman; Freeman, Alex; Emberton, Mark; Piper, Jonathan W; Moore, Caroline M; Hines, John; Orczyk, Clément; Allen, Clare; Giganti, Francesco
• Is Part Of: European journal of radiology. Issue 162(2023)
• Publication Date: 2023-05
• Subjects: Prostate MRI -- Workflow -- Prostate Cancer -- Biopsy -- Second reads -- Software; csPCa Clinically Significant Prostate Cancer -- GS Gleason Score -- IQR Interquartile Range -- MCCL Maximum Cancer Core Length -- PI-RADS Prostate Imaging Reporting and Data System -- PSA Prostate Specific Antigen -- PZ Peripheral Zone -- TZ Transition Zone -- DRE Digital Rectal Examination; Medical radiology -- Periodicals; Radiology -- Periodicals; Radiologie médicale -- Périodiques; Medical radiology; Periodicals; 616.075705
• DOI: 10.1016/j.ejrad.2023.110796
• Languages: English
• ISSNs: 0720-048X

Abstract

Highlights: Semi-automated workflows are useful to contour lesions, calculate prostate volume, and create a structured report for accurate biopsy targeting. Second reads could be helpful to reduce the number of indeterminate lesions, which harboured clinically significant disease after targeted biopsy only in a minority of cases. Second reads using customised workflows could help defer biopsies, reducing the number of side effects. Abstract: Purpose: To investigate the utility of a prostate magnetic resonance imaging (MRI) second read using a semi-automated software program in the one-stop clinic, where patients undergo multiparametric MRI, review and biopsy planning in one visit. We looked at concordance between readers for patients with equivocal scans and the possibility for biopsy deferral in this group. Methods: We present data from 664 consecutive patients. Scans were reported by seven different expert genitourinary radiologists using dedicated software (MIM®) and a Likert scale. All scans were rescored by another expert genitourinary radiologist using a customised workflow for second reads that includes annotated biopsy contours for accurate visual targeting. The number of scans in which a biopsy could have been deferred using biopsy results and prostate specific antigen density was assessed. Gleason score ≥ 3 + 4 was considered clinically significant disease. Concordance between first and second reads for equivocal scans (Likert 3) was evaluated. Results: A total of 209/664 (31%) patients scored Likert 3 on first read, 128 of which (61%) were concordant after second read. 103/209 (49%) of patients with Likert 3 scans were biopsied, with clinically significant disease in 31 (30%) cases. Considering Likert 3 scans that were both downgraded and biopsied using the workflow-generated biopsy contours, 25/103 (24%) biopsies could have been deferred. Conclusions: Implementing a semi-automated workflow for accurate lesion contouring and targeting biopsies is helpful during the one-stop clinic. We observed a reduction of indeterminate scans after second reading and almost a quarter of biopsies could have been deferred, reducing the potential biopsy-related side effects.