MRI perfusion of solitary parotid lesions : a prospective study

Abstract

Background: Parotid tumours (PTs) have a variety of pathological subtypes, each requiring distinct surgical approaches. Nevertheless, the current preoperative diagnostic methods often fall short in providing accurate PT diagnoses. Aim: Can DCE-MRI help predict histology in patients with focal parotid lesions? Design: A prospective study recruiting patients with known focal lesions of the parotid gland who are being worked up for surgery. A dynamic contrast-enhanced MRI (DCE-MRI) scan is performed, after which post-processing of data is used to obtain MR perfusion quantitative parameters (Ktrans, Kep and Ve) obtained by drawing a region of interest (ROI) over the PT. In this study 2 individual sets ROIs were drawn by two authors. The first set of ROIs vary in size, whereas the second are a fixed range ROI size limited to 10-20mm². The information obtained from DCE-MRI is compared to tissue histology and cytology, to assess the diagnostic accuracy of the DCE-MRI. The intra and inter-observer correlation of the different ROI methods is also assessed. Ethical standards were adhered to, with informed consent obtained from all participants. The data collection and analysis process spanned approximately 10 months. Methods: In this prospective study, 15 patients with histologically or cytologically proven PTs who underwent DCE-MRI were enrolled including 7 pleomorphic adenomas (PAs) and 8 Warthin’s tumours (WTs). Quantitative parameters of DCE-MRI (Ktrans, Kep, and Ve) of lesions were calculated and analysed. A second set of analysis was done to assess the intra and inter-observer correlation when using different ROI techniques. Results: Statistically significant (p < 0.05) inter-group differences were found between PA and WT for Kep and Ve. Correlation with histological results using the Mann-Whitney U test, where Ktrans, Kep and Ve were found to have p-values of 0.132, 0.008 and 0.003 respectively. Inter-observer correlation for the first set of variable size ROIs for the different MRI parameters were: Ktrans 0.590, Kep 0.622 and Ve 0.848, and for the second fixed range set Ktrans 0.945, Kep 0.901 and Ve 0.964. The intra-observer results for author 1 for the different MRI parameters were: Ktrans 0.358, Kep 0.718 and Ve 0.436. The intra-observer results for author 2: Ktrans 0.950, Kep 0.581 and Ve 0.762. Conclusions: In summary, DCE-MRI perfusion parameters aid in the differentiation of PA and WT. This technique demonstrates reliability and reproducibility. Standardised ROI sizes with fixed ranges minimise the impact of radiologist experience on perfusion parameter calculations, resulting in better interobserver agreement. However, when evaluating the reproducibility of perfusion parameters obtained by the same reader using different ROI techniques, radiologist experience significantly affects parameter calculations.