A Novel Sonographic Scoring Model in the Prediction of Major Salivary Gland Tumors.

Abstract

To create a sonographic scoring model in the prediction of major salivary gland tumors and to assess the utility of this predictive model. Retrospective case series, academic tertiary referral center. Two hundred fifty-nine patients who underwent ultrasound (US), US-guided needle biopsies, and subsequent operations were enrolled. These data were used to build a predictive scoring model and the model was validated by 10-fold cross-validation. We constructed a sonographic scoring model by multivariate logistic regression analysis: 2.08 × (boundary) + 1.75 × (regional lymphadenopathy) + 1.18 × (shape) + 1.45 × (posterior acoustic enhancement) + 2.4 × (calcification). The optimal cutoff score was 3, corresponding to 70.2% sensitivity, 93.9% specificity, and 89.6% overall accuracy. The mean areas under the receiver operating characteristic curve (c-statistic) in 10-fold cross-validation was 0.90. The constructed predictive scoring model is beneficial for patient counseling under US exam and feasible to provide us the guidance on which kind of needle biopsy should be performed in major salivary gland tumors. 3b Laryngoscope, 131:E157-E162, 2021.