Logistic regression analysis of repeated radioiodine therapy treatments for papillary thyroid carcinoma patients.

Abstract

Radioiodine ablation treatment (RAT) is administered to papillary thyroid carcinoma patients post thyroidectomy. Multivariable logistic regression analysis can be applied to predict treatment failure. In this study, we propose a logistic regression model (LRM) to estimate the probability of repeating the treatment more than one time. A retrospective review of the last 5 years of RAT data revealed that 30 patients had received the RAT more than one time. Various factors including age, sex, pretreatment serum thyroglobulin (Tg), thyroid-stimulating hormone (TSH) and administered activity were analyzed to predict RAT failure and therefore the necessity to repeat the treatment by administering additional doses of radioiodine. The administered activity, the patient age, the presence of distant lymph nodes on the whole-body radioiodine scan (WBS) and the level of Tg before the treatment were found to be the predictive variables. The following LRM is proposed: Y = 7.8295 - 0.0012 [Activity in (MBq) - 0.0541 (Age) - 34.3 (Lymph Nodes) - 0.0042 (Tg)]. The prediction accuracy of the LRM was assessed using receiver operating characteristic (ROC) curve by calculating the area under the curve (AUC). We found the AUC = 0.8972. Patients who are older in age, who receive higher administered radioiodine activity, have higher serum thyroglobulin levels and have lymph node uptake reported in their post-ablation WBS are more likely to have unsuccessful treatment outcome and will repeat the treatment. This LRM could help in adjusting RAT options in order to reduce the repeat rate.