Dynamic phenotyping of radioiodine-refractory differentiated thyroid cancer: thyroglobulin trajectories, dual-modal imaging, and genotype-defined treatment resistance.

BACKGROUND: Radioiodine-refractory differentiated thyroid cancer (RAIR-DTC) is a clinically aggressive disease state with limited treatment options. While individual biomarkers, imaging modalities, and genomic drivers have each been studied in isolation, the dynamic relationships among longitudinal thyroglobulin kinetics, evolving dual-modal imaging phenotypes, and underlying genotypes, together with their consequences for targeted therapy response, remain insufficiently characterized. METHODS: We retrospectively analyzed 424 patients with intermediate- to high-risk differentiated thyroid cancer. Longitudinal serial stimulated thyroglobulin (sTg) measurements, paired ¹³¹I whole-body scan and ¹8F-FDG PET/CT, and next-generation sequencing of postoperative tumor tissue were integrated. Imaging-defined phenotypes (¹³¹I-avid, ¹8F-FDG-avid, or dual-avid) were stratified across sTg levels and correlated with metastatic distribution, genotype, and response to first-line tyrosine kinase inhibitors (TKIs). RESULTS: Sixty-four patients (15.1%) progressed to RAIR-DTC. Median sTg rose from 18.7 ng/mL at initial radioiodine therapy to 302.7 ng/mL at RAIR-DTC diagnosis; pre-radioiodine sTg discriminated RAIR-DTC progression with an area under the curve of 0.911 (95% CI 0.856-0.967; optimal threshold 10.77 ng/mL). With increasing sTg, imaging phenotypes evolved non-binarily, shifting from mixed ¹³¹I/¹8F-FDG-avid lesions toward predominantly ¹8F-FDG-avid disease; metastatic co-occurrence network analysis identified cervical lymph nodes and bilateral lungs as central dissemination hubs. TERT promoter (TERTp) mutations were significantly enriched in RAIR-DTC (43.8% vs 5.3% in non-RAIR-DTC; p < 0.001) and clustered with BRAFV600E within a tightly connected oncogenic network; patients harboring TERTp mutations had markedly shorter progression-free survival (median 18.2 vs 47.6 months; p < 0.0001). Lesions in patients with BRAFV600E/TERTp co-mutation exhibited significantly higher metabolic activity (maximum standardized uptake value; p < 0.0001) and earlier radiographic progression on first-line TKIs (68.4% vs 23.5% within six months; p = 0.001). CONCLUSION: Coupling longitudinal sTg kinetics with evolving dual-modal imaging phenotypes delineates genotype-associated disease courses in RAIR-DTC, in which TERTp mutation, particularly when co-occurring with BRAFV600E, characterizes a metabolically aggressive, TKI-resistant subset. These findings support a dynamic phenotyping approach for refining surveillance and informing genotype-guided treatment decisions in RAIR-DTC.