The diagnostic performance of ultrasonography and computerized tomography in differentiating superficial from deep lobe parotid tumours.

Abstract

To validate and compare ultrasound (US) versus computerized tomography (CT) criteria in the localisation of superficial/deep lobe tumours of the parotid gland. This was a retrospective study of diagnostic tests performed from January 2008 to June 2017. We included adult patients who were referred for a neck ultrasonography examination due to parotid tumours, and who subsequently underwent parotid surgery. We assessed the location of parotid tumours, comparing the minimum fascia-tumour distance (MFTD) criterion on an US with eight CT criteria. We analysed receiver operating characteristic (ROC) curves of the MFTD for malignant, benign, and all parotid tumours, and compared the accuracy, sensitivity, and specificity of the optimal MFTD with those of CT anatomical criteria. A total of 166 parotid tumours were included. The mean (SD) MFTD in superficial lobe tumours was significantly shorter than that of deep lobe tumours (1.2 [0.7] vs 2.8 [1.9] mm, effect size: 1.84; 95% CI, 1.27-2.41). The areas under the ROC curve were 0.63 for malignant tumours and 0.88 for benign tumours. The optimal MFTD cut point was 2.4mm for the 154 benign parotid tumours, and the accuracy, sensitivity and specificity were 90%, 80% and 91%, respectively. For the 136 benign parotid tumours that underwent CT examination, three criteria had an accuracy of over 90% (FNline, tMasseter and Conn's arc), but the sensitivities were all below 50%. Minimum fascia-tumour distance is more feasible for benign tumours than for malignant tumours for the localisation of parotid tumours. For benign parotid tumours, US is enough to guide operations.