Improving the prediction of persistent and recurrent differentiated thyroid cancer using the American Thyroid Association 2015 risk stratification system.

Abstract

The 2015 American Thyroid Association risk stratification system (ATA RSS) is used in patients with differentiated thyroid carcinoma (DTC) to assess their risk of persistent/recurrent disease. Our aims were to validate the 2015 ATA RSS in a registry of DTC patients and to examine whether the addition of factors not included in it, such as pre-radioactive iodine therapy stimulated thyroglobulin (pre-RAI sTg), gender, and age could increase its predictive ability. We studied 403 patients with DTC, treated at a tertiary center from 1990 to 2018 and subjected to total thyroidectomy. All patients had received RAI therapy, except those with low-risk papillary microcarcinoma. Of our patients, 81.9% were women and 91.1% had papillary thyroid carcinoma. After a median follow-up of 5.0years, 53 cases of persistent and 21 cases of recurrent disease were recorded. The proportion of variance explained (PVE) regarding the outcome (presence or absence of recurrent/persistent disease) using the 2015 ATA RSS alone was 18.3% (persistence) and 16.9% (recurrence), increasing to 74.4% and 52.0%, respectively, when pre-RAI sTg was added to the logistic regression model. Gender and age were not associated with the disease outcome. In ROC analysis, pre-RAI sTg had a high predictive value for persistent (AUC 0.983, 95% CI 0.962-1.000) and recurrent disease (AUC 0.856, 95% CI 0.715-0.997). The optimal cut-offs and sensitivity, specificity, and positive and negative predictive value for pre-RAI sTg were the following: for persistence 12.75ng/ml, 100%, 90.5%, 64%, and 100%, and for recurrence 8.05ng/ml, 77.8%, 85.5%, 36.8%, and 97%. The 2015 ATA RSS displayed moderate performance in predicting recurrent/persistent disease in patients with DTC, which improved with the inclusion of pre-RAI sTg values; pre-RAI sTg was an independent predictor of the disease outcome, with high negative prognostic value.