Papillary thyroid cancer (PTC) is quite rare and has generally a good prognosis: approximately 80% of all cases are cured after radical surgery and radioiodine ablative treatment (RAI). Up to 18% of patients are diagnosed with locoregional recurrence within the first 5 years after initial treatment (1), which is usually cured by a second RAI. Thirty-three to 55% of cases diagnosed with recurrence (2) and approximately 5% of all cases are diagnosed with RAI-refractory (RAI-R) disease (3), which is associated with a significantly poorer outcome (5-year survival <50%).

The European Thyroid Association (ETA) Guidelines state that the choice to continue with RAI therapy in a patient without RAI uptake, with a mixed pattern of uptake or progression despite RAI uptake “can only be addressed in a multidisciplinary team (MDT), weighing the benefits and disadvantages of continuing RAI eventually combined with local treatments” (3). If RAI-R disease is limited to a single lesion or more than one lesion within the same organ, the possibility of performing a local treatment should be considered, according to the ETA (3). Since so far, no systemic therapy option for RAI-R disease is curative, treatment should be started only in the setting of rapid metastatic progress according to recist criteria, taking into account patients’ age and comorbidities.

Chemotherapy plays no significant role in the systemic treatment of advanced differentiated thyroid cancer (DTC). Medical treatment is currently based on tyrosine kinase inhibitors (TKIs) as targeted molecular treatment based on studies showing that cellular dedifferentiation resulting in unresponsiveness to RAI therapy correlate with the degree of MAPK activation (4). The kinase inhibitors Sorafenib and Lenvatinib have been shown to significantly improve progression-free survival (PFS) rates in advanced RAI-R DTCs (4) and have consequently been approved by the Food and Drug Administration (FDA) and the European Medical Agency (EMA) for treatment of advanced RAI-R DTCs (5). An increase in overall patient survival, however, has been described only for Lenvatinib selectively for the subgroup of cases >65 years old (5–7).

There are, however, several drawbacks to PTC treatment with TKIs. The medication needs to be taken for the rest of the patient’s life and has a wide spectrum of adverse side effects with a significant impact on quality of life. Further, some patients develop a resistance to treatment (8). For this reason, it is generally recommended to carefully choose the starting point of treatment, with particular emphasis on the patient’s general condition and quality of life.

Based on in silico data mining, we have previously reported that drugs already approved by the FDA for different cancer types show potential to be used in other cancer entities (9). Here, we investigate whether any of the 129 currently FDA-approved drugs act on targets also commonly found in papillary thyroid cancer based on genetic mutations reported by the “The Cancer Genome Atlas (TCGA)”.

All data were obtained from open access databases and referenced accordingly. The study was conducted in accordance with the provisions of the Declaration of Helsinki and local laws.

Current targeted therapy options for radioiodine-refractory PTC are based on data from placebo-controlled randomized trials in the absence of sufficient clinical trial data comparing single agents or combinations (25). The current options encompass either mutation-specific or antiangiogenic multitargeted kinase inhibitors (aaMKIs). They frequently produce objective response, but no general prolongation of overall survival, and they have significant toxicities (4–7).

In this study we attempted to find new targetable options based on genes commonly altered in PTC. We analyzed the TCGA thyroid cancer study for commonly affected genes and found a high mutation rate in only one single point mutation: the BRAF V600 mutation. This mutation is seen in 68% of classic papillary thyroid cancer and in 93% of tall-cell variant patients. Given the predominance of BRAF mutations in RAI-R DTC (26, 27), it is possible that classic papillary thyroid cancer and tall-cell variant more often develop to refractory disease. The second highest frequency of mutation hotspots occurred in the RAS-Pathway. NRAS, HRAS, and KRAS mutations, which are seen in 30–50% of follicular thyroid cancer (FTC) (28), were detected in 8% of classic and in 33% of follicular variant PTCs. BRAF and RAS mutation are mutually exclusive in DTC (19). An association between the type of mutation (BRAF or RAS) and the metastasis pathway typical for papillary and follicular thyroid cancer (lymphogenic or hematogenic) has also been described (29). Although the RAS-Pathway displayed the second highest frequency mutation hotspots, there are no FDA-approved drugs targeting the specific mutations found in NRAS, HRAS, and KRAS for PTC. At this time, it is unclear if drugs targeting NRAS, HRAS, or KRAS (BI1701963 targeting KRAS and Tipifarnib targeting HRAS) will have any relevance for PTC in the future. Ho et al. (30) reported stable disease in 9 of 11 HRAS-mutated thyroid cancer patients treated with Tipifarnib; however, they focus on head and neck squamous cell carcinoma in their report and the specific histology of the thyroid cancer patients is not included. Given the predominance of RAS mutations in RAI avid as opposed to RAI-refractory DTC reported in the literature (31), this pathway might be less relevant for the treatment of RAI-R disease. Other potential targetable genetic alterations were found in less than 10% cases: the RET gene fusion was found in 8%, NTRK1 and 3 gene fusion in 3%, and other mutations in less than 2% of classic papillary thyroid cancers.

Accordingly, our in silico analysis predicted a high drug response for both the multikinase BRAF inhibitors (Regorafenib and Sorafenib) and the selective BRAF inhibitors (Dabrafenib, Vemurafenib, Cobimetinib), particularly in classic and tall-cell PTC ( Figure 2 ). These results may not be surprising. Sorafenib has been shown to significantly improve median progression free survival (PFS) compared to placebo in patients with RAI-refractory DTC (10.8 vs. 5.8 months, p < 0.0001) in the phase 3 DECISION trial (32), and it is one of the two currently recommended first-line treatment agents. Regorafenib is very similar to Sorafenib in its molecular structure, except for the addition of a fluorine to the center phenyl ring. Regorafenib shows increased anti-angiogenic activity over Sorafenib (33). Studies on medullary thyroid cancer are currently recruiting (NCT02657551). However, there is currently no data or studies on Regorafenib in radioiodine-refractory papillary thyroid cancer (33).

Interestingly, our study only predicted a low drug response for the other first-line agent Lenvatinib. Our in silico analysis predicted a drug response of Lenvatinib similar to that predicted for Apatinib, Pazopanib, Sunitinib, Cabozatinib, Vandetanib, Pralsetinib, and Selpercatinib. However, Lenvatinib has not only been shown to significantly improve progression-free survival (PFS) rates in advanced RAI-R DTCs similar to Sorafenib (4), but it also represents the only current targeted therapy, for which an increase in overall survival has been shown in the >65 year old subgroup (5–7). In the SELECT study, the benefit of Lenvatinib was independent of the BRAF and RAS mutational status (34). A possible reason is the wide range of tumor cell targets affected (VEGF, VEGFR1–3, FGFR1–4, PDGFRα, KIT, and RET) (35). Studies on Apatinib (NCT03048877), Pazopanib (NCT01813136), Sunitinib (NCT00519896), and Cabozantinib (NCT03690388) on radioiodine-refractory thyroid cancer are ongoing ( Table 1 ).

The V600 mutant specific inhibitors Vemurafenib (PLX4032) and Dabrafenib (GSK2118436) for which the in silico analysis predicts a high drug response and which have produced excellent response rates in melanoma studies and in hairy cell leukemia (41) are typically reserved for patients who show a contraindication for first-line therapy with multikinase inhibitors and are administered in off-label use. Although objective responses and a prolongation of progression-free survival in small phase 2 trials were seen, these results have not yet been demonstrated in larger randomized trials (40, 42–51) ( Table 1 ). Consequently, the inhibitors haven’t been approved by the FDA for thyroid cancer because toxicity seems to exceed the oncologic benefit. However, like Cobimetinib (GDC0973) (39), Vemurafenib and Dabrafenib have been reported to allow re-expression of genes responsible for iodine metabolism, restoring radioiodine uptake and permitting subsequent radioiodine treatment. In a trial with 10 BRAF V600 mutated PTC cases, 60% revealed new radioiodine uptake after 4 weeks of Dabrafenib therapy. Three months after radioiodine therapy in two of the six cases, a partial response could be observed (52). A similar partial response was also described for combined treatment with Cobimetinib and Vemurafenib in BRAF V600 mutated PTC cases by Iravani et al. (39). More data on the use of these inhibitors in PTC are critically needed, and “redifferentiation” is so far no clinical standard.

Given the fact that clinical trials with BRAF inhibitors produced various outcomes in different carcinomas, we performed a pan-cancer analysis for BRAF mutations ( Figures 3A, B ). A detailed knowledge and understanding of these results may also offer insights in the biology of PTC. The contrasting results of V600 mutant specific inhibitors in thyroid cancer compared to melanoma and hairy cell leukemia (53) could be due to different response and resistance mechanisms to BRAF inhibition dependent on cell types. A similar contrast was observed in clinical trials for colorectal and non-small-cell lung cancer compared to malignant melanoma (54). The mechanism responsible for a poor response could, for example, include the inhibition of the BCL2 pathway, a concurrent PI3KCA mutation and secondary alterations in the MAPK pathway (55). In colon carcinoma, for example, BRAF V600E inhibition has been shown to cause a rapid feedback activation of EGFR, which supports continued proliferation in the presence of BRAF V600E inhibition (56). Melanoma cells instead express low levels of EGFR and are therefore not subject to this feedback activation. Similarly co-treatment of thyroid cancer cell lines with an EGFR inhibitor increased antitumor efficacy and suppressed resistance to the BRAF V600E inhibition (57, 58).

Because the currently recommended first-line agents Lenvatinib and Sorafenib lack specificity for BRAF and the toxicity profile still represents a clinical hurdle, other treatment options would be useful in adjuvant setting after surgical management of radioiodine-refractory relapse. Since radioiodine-refractory papillary thyroid cancer is rare (3% of patients) and targeted therapy is currently initiated only after disease progression in a palliative setting, the present study was performed to select the most promising candidates among FDA drugs approved for solid cancer. The major limitation in this study is the fact that this is an in silico analysis. Clinical data would be critical to verify the results presented. Secondly, all data presented are based on the TCGA dataset and, thus, on primary papillary instead of radioiodine-refractory thyroid cancer. However, Sabra et al. (31) described BRAF mutations as predominant in RAI-R DTC and RAS mutations in radioiodine-avid DTC (19). For this reason, the results presented can possibly be extended to radioiodine-refractory thyroid cancer. It also needs to be mentioned that beside the histological subtype, other parameters such as pTNM stage, the patient’s age or co-morbidities must be considered for treatment stratification. In addition, spatial and temporal tumor heterogeneity, which are particularly relevant for the treatment of advanced thyroid cancer (59–61), are not sufficiently represented by the TCGA data. In the era of personalized medicine, the answer will likely require an ultra-deep sequence of the patient’s primary tumor in an attempt to identify relevant mutated genes. Appropriate drug combinations will help target the spatial and prevent the temporal heterogeneity of tumors, with the eradication of these clones (62). In addition, it is unclear how the data presented relates to pediatric patients, as the TCGA dataset does not cover patients <18 years. It is likely that the majority of pediatric PTC patients with high-risk tumors have a high genetic component and therefore druggable targets (63, 64).

Finally, several studies have demonstrated that TERT promoter mutations, particularly in association with BRAF V600E, correlate with negative clinical characteristics such as RAI- refractory disease (27, 65). TERT promoter mutations were reported in 36 (9.4%) of 384 informative tumors in the TCGA data (19). Further, for distant metastases, an enrichment in TERT mutations has been described (66). There are currently studies on telomerase inhibitors (INO5401, Telomelysin, and Imeltestat) on myeloid malignancy. These drugs might also play a role in the future for the treatment of RAI-R disease.

In summary our in silico analysis for drug prediction response based on alterations described in the TCGA data suggested BRAF inhibitors as the most promising target for most classic and tall-cell variants of thyroid cancer. The multikinase BRAF inhibitor Sorafenib is currently the first-line treatment for radioiodine-refractory thyroid cancer; however, for Regorafenib, clinical data for differentiated thyroid cancer are still lacking; and for the selective BRAF inhibitors (Dabrafenib, Vemurafenib, Cobimetinib), insufficient evidence has been collected for FDA approval in PTC treatment. Similar to colorectal and small lung cancer, resistance mechanisms might be responsible for the decreased efficacy of these medications. This needs to be considered in future studies. For Lenvatinib, which plays a major role in first-line targeted therapy of radioiodine-refractory thyroid cancer, our in silico analysis predicts a low response. Since targeted therapy does not lead to a significant improvement of overall survival in younger patients and is associated with high toxicity, alternative treatment options are urgently needed. These data suggest that resistance mechanisms to BRAF inhibition should be further investigated in order to ameliorate drug responses in the future.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

DH and HA designed the computational model and framework. DH, HA, and CC carried out the implementation. DH, BR, MH, MF, CB, HA, and CC contributed to the interpretation of the results. DH, BR, MH, MM, MF, CB, HA, and CC contributed critical feedback and helped shape the research, analysis, and manuscript. CC and DH wrote the first draft of the manuscript, and all authors critically revised the manuscript. All authors contributed to the article and approved the submitted version. All authors decided to submit this study and agreed to be accountable for all aspects of the work as recommended by the “International Committee of Medical Journal Editors” (ICMJE) authorship criteria.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.