Thibaut Pierre1, Francois Cornud2,3, Loïc Colléter4, Frédéric Beuvon5, Frantz Foissac6, Nicolas B Delongchamps7, Paul Legmann1. 1. Department of Radiology, Hôpital Cochin, 27 rue du Faubourg Saint-Jacques, 75014, Paris, France. 2. Department of Radiology, Hôpital Cochin, 27 rue du Faubourg Saint-Jacques, 75014, Paris, France. francois.cornud@imagerietourville.com. 3. IRM Paris 16, 46 rue Chardon Lagache, 75016, Paris, France. francois.cornud@imagerietourville.com. 4. Department of Radiology, Hôpital St Louis, 1 avenue Claude Vellefaux, 75010, Paris, France. 5. Department of Pathology, Hôpital Cochin, 27 rue du Faubourg Saint-Jacques, 75014, Paris, France. 6. Department of Biostatistics, Hôpital Tarnier, 89 rue d'Assas, 75006, Paris, France. 7. Department of Urology, Hôpital Cochin, 27 rue du Faubourg Saint-Jacques, 75014, Paris, France.
Abstract
PURPOSE: To compare inter-reader concordance and accuracy of qualitative diffusion-weighted (DW) PIRADSv2.0 score with those of quantitative DW-MRI for the diagnosis of peripheral zone prostate cancer. MATERIALS AND METHODS: Two radiologists independently assigned a DW-MRI-PIRADS score to 92 PZ-foci, in 74 patients (64.3±5.6 years old; median PSA level: 8 ng/ml, normal DRE in 70 men). A standardised ADCmean and nine ADC-derived parameters were measured, including ADCratios with the whole-prostate (WP-ADCratio) or the mirror-PZ (mirror-ADCratio) as reference areas. Surgical histology and MRI-TRUS fusion-biopsy were the reference for tumours and benign foci, respectively. Inter-reader agreement was assessed by the Cohen-kappa-coefficient and the intraclass correlation coefficient (ICC). Univariate-multivariate regressions determined the most predictive factor for cancer. RESULTS: Fifty lesions were malignant. Inter-reader concordance was fair for qualitative assessment, but excellent for quantitative assessment for all quantitative variables. At univariate analysis, ADCmean, WP-ADCratio and WL-ADCmean performed equally, but significantly better than the mirror-ADCratio (p<0.001). At multivariate analysis, the only independent variable significantly associated with malignancy was the whole-prostate-ADCratio. At a cut-off value of 0.68, sensitivity was 94-90 % and specificity was 60-38 % for readers 1 and 2, respectively. CONCLUSION: The whole-prostate-ADCratio improved the qualitative inter-reader concordance and characterisation of focal PZ-lesions. KEY POINTS: • Inter-reader concordance of DW PI-RADSv2.0 score for PZ lesions was only fair. • Using a standardised ADCmean measurement and derived DW-quantitative parameters, concordance was excellent. • The whole-prostate ADCratio performed significantly better than the mirror-ADCratio for cancer detection. • At a cut-off of 0.68, sensitivity values of WP-ADCratio were 94-90 %. • The whole-prostate ADCratio may circumvent variations of ADC metrics across centres.
PURPOSE: To compare inter-reader concordance and accuracy of qualitative diffusion-weighted (DW) PIRADSv2.0 score with those of quantitative DW-MRI for the diagnosis of peripheral zone prostate cancer. MATERIALS AND METHODS: Two radiologists independently assigned a DW-MRI-PIRADS score to 92 PZ-foci, in 74 patients (64.3±5.6 years old; median PSA level: 8 ng/ml, normal DRE in 70 men). A standardised ADCmean and nine ADC-derived parameters were measured, including ADCratios with the whole-prostate (WP-ADCratio) or the mirror-PZ (mirror-ADCratio) as reference areas. Surgical histology and MRI-TRUS fusion-biopsy were the reference for tumours and benign foci, respectively. Inter-reader agreement was assessed by the Cohen-kappa-coefficient and the intraclass correlation coefficient (ICC). Univariate-multivariate regressions determined the most predictive factor for cancer. RESULTS: Fifty lesions were malignant. Inter-reader concordance was fair for qualitative assessment, but excellent for quantitative assessment for all quantitative variables. At univariate analysis, ADCmean, WP-ADCratio and WL-ADCmean performed equally, but significantly better than the mirror-ADCratio (p<0.001). At multivariate analysis, the only independent variable significantly associated with malignancy was the whole-prostate-ADCratio. At a cut-off value of 0.68, sensitivity was 94-90 % and specificity was 60-38 % for readers 1 and 2, respectively. CONCLUSION: The whole-prostate-ADCratio improved the qualitative inter-reader concordance and characterisation of focal PZ-lesions. KEY POINTS: • Inter-reader concordance of DW PI-RADSv2.0 score for PZ lesions was only fair. • Using a standardised ADCmean measurement and derived DW-quantitative parameters, concordance was excellent. • The whole-prostate ADCratio performed significantly better than the mirror-ADCratio for cancer detection. • At a cut-off of 0.68, sensitivity values of WP-ADCratio were 94-90 %. • The whole-prostate ADCratio may circumvent variations of ADC metrics across centres.
Entities:
Keywords:
Adult; Diffusion magnetic resonance imaging/methods; Prostatic neoplasms/diagnostic imaging; Retrospective studies; Sensitivity and specificity
Authors: Olivio F Donati; Asim Afaq; Hebert Alberto Vargas; Yousef Mazaheri; Junting Zheng; Chaya S Moskowitz; Hedvig Hricak; Oguz Akin Journal: Clin Cancer Res Date: 2014-05-21 Impact factor: 12.531
Authors: Klaas N A Nagel; Martijn G Schouten; Thomas Hambrock; Geert J S Litjens; Caroline M A Hoeks; Bennie ten Haken; Jelle O Barentsz; Jurgen J Fütterer Journal: Radiology Date: 2013-01-17 Impact factor: 11.105
Authors: F Cornud; Gaby Khoury; Naim Bouazza; Frederic Beuvon; Michael Peyromaure; Thierry Flam; Marc Zerbib; Paul Legmann; Nicolas B Delongchamps Journal: J Urol Date: 2013-12-12 Impact factor: 7.450
Authors: Hashim U Ahmed; Ahmed El-Shater Bosaily; Louise C Brown; Rhian Gabe; Richard Kaplan; Mahesh K Parmar; Yolanda Collaco-Moraes; Katie Ward; Richard G Hindley; Alex Freeman; Alex P Kirkham; Robert Oldroyd; Chris Parker; Mark Emberton Journal: Lancet Date: 2017-01-20 Impact factor: 79.321