Measuring serum Tg and performing a diagnostic whole body scan (DxWBS) has become the standard for follow-up of patients with differentiated thyroid carcinoma. The primary aim of this study was to determine whether recombinant human TSH (rhTSH)-stimulated Tg alone is sufficiently sensitive to identify residual cancer in patients with no clinical evidence of disease and undetectable or very low serum Tg levels during thyroid hormone (TH) therapy. A secondary aim was to investigate the frequency of tumor in such patients. One hundred and seven consecutive patients, aged 10.9-85.3 yr (median, 36.3), at the time of initial surgery who had Tg levels on TH therapy that were undetectable (95% < or =0.5 ng/ml) or low (4% 0.6 ng/ml, 1% 1.0 ng/ml) and who underwent rhTSH-stimulated testing 10 months to 35 yr (median, 3.5 yr) after initial thyroidectomy and (131)I ablation were retrospectively studied. Many (50%) were at high risk of tumor recurrence, and 5 had distant metastases during the course of their disease. In response to rhTSH, Tg ranged from 0.5 or less to 17.9 ng/ml, remaining at 0.5 ng/ml or less in 68 (64%) patients and increasing to levels between 0.6 and 2 ng/ml in 19 (18%) others and to levels higher than 2 ng/ml in 20 (19%) patients. Eleven patients (10%), all of whom had rhTSH-stimulated serum Tg levels above 2 ng/ml, were found to have persistent tumor in lung (4 patients), lymph nodes (5 patients, 3 with cervical central compartment, 1 bilateral cervical, and 1 with mediastinal nodes) identified by fine needle cytology, surgical pathology, posttherapy whole body scans, or computed tomography and, in two patients, with high serum Tg values alone (4.6 and 7.0 ng/ml after rhTSH and, respectively, 28.5 and 70.6 ng/ml after TH withdrawal), although in neither could the tumor site be identified. Thirteen patients (12%) were treated with surgery or (131)I, and in some cases both, as a result of the rhTSH studies; 10 had tumor, 1 had residual uptake in the thyroid bed visible on rhTSH-stimulated diagnostic whole body scan (DxWBS), and 2 had high serum Tg levels, presumably originating from a tumor site that could not be identified. A patient's tumor status, even in retrospect, usually was not predictable on the basis of Tg during TH therapy or tumor node metastasis status: among patients found to have tumor after rhTSH, serum Tg during TH therapy was 0.5 ng/ml or less in 55% and 0.6 ng/ml in 36%, and tumor node metastasis status was T2N1 or less in 82%. In no case did the rhTSH-stimulated DxWBS show the site of persistent tumor. There were correlations between visible thyroid bed uptake on DxWBS and quantitated (131)I uptake (r(2) = 0.11; P = 0.001), between DxWBS and rhTSH-stimulated Tg (r(2) = 0.54; P = 0.001), and between rhTSH-stimulated Tg and (131)I uptake (r(2) = 0.66; P = 0.0001). There was no statistically significant difference (P = 0.4) in bed (131)I uptake in patients with rhTSH-stimulated serum Tg levels of 0.5 ng/ml or less compared with that in subjects with higher rhTSH-Tg levels. An rhTSH-stimulated Tg level greater than 2 ng/ml had a sensitivity of 100%, a negative predictive value of 100%, and a false positive rate of 9%. The rhTSH Tg had a substantially better performance than the other studies; the false negative rates were 64% for Tg higher than 0.5 ng/ml on TH therapy, 73% for rhTSH-stimulated DxWBS showing uptake, and zero for an rhTSH-stimulated Tg more than 2 ng/ml. In conclusion, of 107 patients who were clinically free of disease, 10% had persistent tumor (4 with pulmonary metastases and 5 with regional disease) that was only identified with an rhTSH-stimulated serum Tg level greater than 2 ng/ml. This study shows that tumor amenable to early therapy may be found when rhTSH-stimulated serum Tg rises above 2 ng/ml without performing a DxWBS, which merely provides data concerning the completeness of thyroid ablation, but not persistent tumor. An elevated rhTSH-stimulated Tg greater than 2 ng/ml warrants further study.