A clinical-radiological score to diagnose Hashimoto’s thyroiditis: a proposal

Introduction. Early diagnosis of Hashimoto’s thyroiditis (HT) may be difficult. The heterogeneity of criteria used to diagnose HT may prevent strong conclusions from being drawn in studies focusing on clinical aspects of HT. The aim of this study is to design a simple score to diagnose chronic autoimmune thyroiditis. Methods. 1021 consecutive patients that were advised to undergo total thyroidectomy at an University thyroid referral center. Given a dichotomous outcome, a set of 2 demographic (sex and age), 3 biochemical (overt or subclinical hypothyroidism, positive anti-Tg Ab, and positive anti-TPO Ab) and 4 imaging covariates (hypoechogenic parenchyma, heterogeneous thyroid echopattern, color-Doppler pattern and estimated thyroid volume) was analyzed. Results. Analysis showed that anti-TPO Ab [area under the curve (AUC) under the ROC curve, 0.67], and anti-Tg Ab (0.63) were univariate predictors of the diagnosis of HT, which is largely recognized. Combined covariates were then tested using stepwise logistic regression. A model to predict the final diagnosis was calculated by using multivariate logistic regression analysis. The final model included anti-TPO Ab, anti-Tg Ab and thyroid vascularity (AUC 0.72). A second scoring system was developed to diagnose HT, with the addition of heterogeneous echopattern and goiter (AUC 0.76). Conclusions. A simple scoring system for the early diagnosis of HT could easily be applied in clinical practice and research. The better proposed score has been shown to have an overall low degree of sensitivity and speci?city, but higher than the single predictors alone (sensitivity 45.5% and specificity 89.0%, with a cutoff value of 1.7). International multicenter studies can recruit a higher number of patients and provide a sufficient amount of data to integrate all features of HT into a consensus diagnostic score.