Only peak thyroglobulin concentration on day 1 and 3 of rhTSH-aided RAI adjuvant treatment has prognostic implications in differentiated thyroid cancer.

OBJECTIVE In patients with differentiated thyroid carcinoma (DTC), serum thyroglobulin levels measured at the time of remnant ablation after thyroid hormone withdrawal were shown to have prognostic value for disease-free status. We sought to evaluate serial thyroglobulin measurements at the time of recombinant human thyroid-stimulating hormone (rhTSH)-aided iodine 131 (131I) adjuvant treatment as prognostic markers of DTC. METHODS Six hundred-fifty patients with DTC given total/near-total thyroidectomy and adjuvant radioiodine post-rhTSH stimulation were evaluated. Thyroglobulin was measured on day 1 (Tg1; at the time of the first rhTSH injection), day 3 (Tg3; 1 day after the second, final rhTSH injection), and day 6 (Tg6; 3 days post-radioiodine administration). Treatment failure was defined as histopathologically confirmed locoregional recurrence, or radiologically-evident distant metastases (signs of disease on computer tomography (CT) or magnetic resonance imaging (MRI), or abnormal foci of radioiodine or [18F] fluorodeoxyglucose ([18F]FDG) uptake. RESULTS In univariate analysis, Tg1 (p < 0.001) and Tg3 (p < 0.001), but not Tg6, were significantly associated with structural recurrence. In multivariate analysis of the overall cohort, only Tg3 was independently associated with structural recurrence. In multivariate analysis of the subgroup (n = 561) with anti-Tg antibodies titers below the institutional cut-off, 115 IU/mL, Tg1 was an independent prognostic marker. Tg1 and Tg3 cutoffs to best predict structural recurrence were established at 0.7 ng/mL and 1.4 ng/mL, respectively. CONCLUSIONS Tg1 and Tg3, measurements made after rhTSH stimulation but before radioiodine treatment, independently predict a low risk of treatment failure in patients with DTC. Levels measured post-radioiodine application (e.g., Tg6) are highly variable, lack prognostic value, and hence can be omitted.