Diagnostic accuracy of different magnetic resonance imaging sequences for detecting local tumor progression after radiofrequency ablation of hepatic malignancies. Issue 94 (September 2017)

• Record Type: Journal Article
• Authors: Weiss, Jakob; Rempp, Hansjoerg; Clasen, Stephan; Notohamiprodjo, Mike; Keßler, David-Emanuel; Pereira, Philippe L.; Blumenstock, Gunnar; Nikolaou, Konstantin; Hoffmann, Rüdiger
• Is Part Of: European journal of radiology. Issue 94(2017)
• Page Start: 85
• Page End: 92
• Publication Date: 2017-09
• Subjects: Diagnostic accuracy -- Magnetic resonance imaging -- Radiofrequency ablation -- Hepatic malignancies -- Local tumour progression -- Imaging pitfalls; Medical radiology -- Periodicals; Radiology -- Periodicals; Radiologie médicale -- Périodiques; Medical radiology; Periodicals; 616.075705
• DOI: 10.1016/j.ejrad.2017.06.009
• Languages: English
• ISSNs: 0720-048X

Abstract

Highlights: Unenhanced T1-FLASH-2D is an essential sequence for RFA follow-up imaging. Diagnostic confidence depends on the individual observers' experience. Cholestasis and vessels adjacent to ablation zone can cause false-positive rating. A hepatic abscess can mimic local recurrence in all sequences. Abstract: Objective: To evaluate the diagnostic accuracy of the individual sequences of a clinical routine liver MRI protocol for the detection of local tumour progression after radiofrequency (RF) ablation of hepatic malignancies. Material and methods: A cohort of 93 patients treated for 140 primary and secondary hepatic malignancies with RF ablation was assembled for this retrospective study. The cohort contained 31 cases of local tumour progression, which occurred 8.3 ± 6.2 months (range: 4.0–28.2 months) after treatment. All patients underwent clinical routine follow-up MRI at 1.5T including following sequences: unenhanced T1-weighted fast low angle shot (FLASH-2D), T2-weighted turbo-spin-echo sequence, contrast-enhanced (CE) T1-weighted volume-interpolated breath-hold examination (VIBE), diffusion-weighted imaging (DWI). Follow-up was 32.7 ± 22.5 months (range: 4.0–138.3 months). Two readers independently evaluated the individual sequences separately for signs of local tumour progression. Diagnostic confidence was rated on a 4-point scale. Inter-reader agreement was assessed with Coheńs kappa. Long-term follow-up and histological specimen served as standard of reference. Results: Both readers reached the highest sensitivity for detection of local tumour progression with unenhanced T1-FLASH 2D (88.2% and 94.1%, respectively) and the highest specificity with CE T1-VIBE (96.2% and 97.2%, respectively). Highest inter-reader agreement was reached with T1-FLASH-2D (kappa = 0.83). Typical pitfalls for false-positive diagnoses were focal cholestasis and vasculature adjacent to the ablation zone. Diagnostic confidence was highest with CE T1-VIBE for reader 1 and DWI for reader 2. Conclusion: Unenhanced T1-FLASH-2D is an essential sequence for follow-up imaging after tumour ablation with a high sensitivity for detection of local progression and a high inter-reader agreement.