Targeting non-canonical pathways as a strategy to modulate radioiodide uptake in thyroid cancer

The sodium iodide symporter (NIS) transports iodide and is critical for radioiodide (RAI) therapy of thyroid cancer. However, NIS activity is diminished in 25-50% of thyroid cancers, limiting adequate RAI uptake for effective therapeutic ablation. New approaches to identify key targetable processes that govern NIS function are urgently needed to enhance RAI uptake and diminish recurrence. Here, we utilised NIS in high-throughput screening (HTS) and undertook rigorous evaluation of lead drugs to identify key processes underpinning NIS function, as well as to construct a predictive risk model of recurrent thyroid cancer.

We utilised the mutated yellow fluorescent protein as a biosensor of iodide and screened 1200 drugs (95% FDA approved) in TPC1-NIS cells. NIS function was monitored by radioiodide (I125) uptake assays in human primary thyrocytes and thyroid cancer cells. The Cancer Genome Atlas (TCGA) data was appraised for proteostasis genes.

From our HTS ~8% drugs (98/1200) gave a meaningful increase in NIS function (SD>2). Of these, we identified drugs targeting multiple processes that regulate NIS, including transcriptional (formoterol), and proteasomal (disulfiram) mechanisms. Categorization revealed a high proportion of lead drugs (8/16) that modulate the proteostasis network, including VCP–a critical component of the proteasome system and autophagy. Dose-dependent effects were apparent across cancer cell models, as well as human primary thyrocytes, implying that proteostasis pathways are central to the innate control of NIS function. Exploiting mechanistic insights, several drugs–which target pathways separate to canonical MAPK regulation of NIS–gave robust increases in RAI uptake when combined (>5-fold, P<0.001). Appraisal of TCGA identified significant dysregulation of 13 proteostasis genes linked to recurrence in papillary thyroid cancer (PTC) (P<0.05). Critically, a predictive risk model showed a significantly worse prognosis for high-risk RAI-treated PTC [Hazard Ratio (HR)=35.9, 95%CI 4.8-267.4; P<0.001; n=137]. After controlling for multiple clinical factors, multivariate analysis showed that the 13 gene risk score was an independent predictor for recurrence (HR=4.6, 95%CI 2.4-8.7; P<0.001) in the entire cohort (n=438).

Collectively, we propose a new model for the targetable steps of intracellular processing of NIS, with translatable potential to address the lack of clinical options for patients treated with RAI who typically have poorer outcomes.