Thyroid cancers, while relatively uncommon at an annual incidence of 34,000 cases, have been increasing in incidence for reasons that remain unclear at a rate of 3% per year. 1, 40, 41
Known risk factors include prior radiation exposure, reduced iodine intake, lymphocytic thyroiditis, and a family history of thyroid cancer. 42
In addition, those exposed to nuclear fall-out as a result of nuclear disasters are known to have a higher risk of papillary thyroid cancer. 43
The rising incidence of thyroid cancer seems to be mainly due to increased rates of papillary thyroid cancer, as opposed to the other histologies, and appears driven in part by an increase in the detection of small cancers via ultrasound and other imaging technologies. 41
There are two distinct cellular origins. Papillary, follicular and anaplastic thyroid cancers arise from the follicular cells, while medullary cancers arise from the parafollicular C-cells. The majority of thyroid cancers are the differentiated histologies (DTC), with papillary thyroid cancer (PTC) accounting for 80%, follicular cancer/Hurthle cell variant (FTC) accounting for 15%, and anaplastic (ATC) accounting for 2% of diagnoses. The mainstay of treatment for thyroid malignancies is surgical resection. The differentiated thyroid cancers are often amenable to adjuvant treatment for cure with radioactive I-131, and this modality is also the initial preferred treatment for metastatic disease that is iodine-avid. This group of thyroid cancers is also sensitive to TSH stimulation and they produce thyroglobulin. In contrast, medullary thyroid cancers (MTC) do not have any of these features.
Doxorubicin is currently the only FDA-approved systemic agent for the treatment of advanced, incurable thyroid cancer. Doxorubicin has been shown to induce apoptosis in thyroid cancer cell lines. 44, 45
While the clinical experience with doxorubicin in thyroid cancer has spanned decades, in practice there here been little enthusiasm to use it as a routine first line option. Historically, numerous small phase 2 studies of doxorubicin with sample sizes ranging from two to nineteen patients have shown response rates ranging from 22–90%. 46–52
It is widely believed that the small patient numbers and varying criteria for assessing response, especially among the older studies which pre-dated spiral CT scans as well as consensus criteria for response assessment such as RECIST, exaggerated the effectiveness of this agent. Even a small phase 2 study of the combination of cisplatin and doxorubicin only resulted in a response rate of 9%. 53
Doxorubicin has been studied in two relatively contemporary trials. In one, seventeen patients were treated with doxorubicin in combination with interferon alpha. 54
Only one patient had a partial response and ten had stable disease, with a median time to progression (TTP) of 5.9 months. In another study, doxorubicin monotherapy (either given weekly or once every three weeks) was administered. 55
Among the patients with papillary or follicular cancer, there was a PR rate of 5% with 42% of patients showing SD. Among patients with medullary thyroid cancer, the rates of PR and SD were both 11%. Thus, while doxorubicin has single agent activity, there is an obvious need for a more effective, less toxic therapy.
Molecular Pathogenesis of Thyroid Cancer
Iodine-refractory thyroid cancer arises as a result of tumor cell de-differentiation and accounts for about 2–5% of all thyroid cancers. High risk features for developing eventual iodine refractory disease include tumor necrosis, extrathyroidal extension, older age, male gender, and high grade histology. The majority of deaths from differentiated thyroid cancer occur in patients with iodine refractory disease. In the past treatment options for this unfortunate group of patients have included surgery or external beam radiation for localized disease in the neck and upper thorax, doxorubicin for systemic disease, or referral for experimental therapy, usually phase 1 trials.
Novel molecular therapies are having a potentially dramatic impact on the course of incurable, iodine-refractory thyroid cancer as well as medullary thyroid cancer and are likely to change our treatment paradigms. An understanding of the pathogenesis of thyroid cancer is necessary in order to understand why responses are occurring and also to determine how best to utilize the multi-kinase inhibitors that are currently under evaluation in the clinic. This involves an understanding of the initiating genetic lesions responsible for these diseases, as well as those transformation events that lead to the progressive de-differentiation that result in undifferentiated and anaplastic cancers.
The molecular events associated with the development of papillary thyroid cancer mainly appear to involve alterations of genes encoding effectors of the MAPK pathway. This typically includes non-overlapping activating mutations in one of the following four genes, which are detectable in 70% of papillary thyroid cancers: RET/PTC rearrangements, BRAF, NTRK1 (neutrotrophic tyrosine kinase receptor 1) rearrangements, or RAS. 56, 57
Ret and NTRK1 are both tyrosine kinase receptors, and Raf is a serine/threonine kinase. Numerous RET/PTC rearrangements have been identified in sporadic and radiation exposure related papillary thyroid cancers, with RET/PTC1 and RET/PTC3 being more common. 58
A somatic mutation in BRAF, (V600E), is one of the more common mutations identified (36–69%), while RAS mutations are more rare in papillary cancers and appear to be more common in follicular cancers. 57, 59
RAF mutations correlate with adverse clinical features, such as extrathyroidal invasion, lymph node metastases, advanced stage, risk of recurrence, and loss of I-131 avidity. 60, 61
In one series, RAF point mutations have been detected in 38% of papillary carcinomas, 13% of poorly differentiated carcinomas, and 10% of anaplastic carcinomas, but not in follicular or Hurthle cell malignancies. 61
In addition, BRAF mutation has been shown to correlate with low expression of the sodium iodide symporter (NIS), which could provide a molecular explanation of the de-differentiated, non-iodine-avid phenotype of these cancers. 62
Interestingly, the RET/PTC alteration was not associated with NIS expression. Inhibition of the MAPK pathway therefore becomes an obvious therapeutic approach for iodine refractory thyroid cancers. Preclinical data from cell lines with either BRAF, RAS or RET mutations interestingly indicate that only the BRAF mutation predicts for sensitivity to MEK inhibition, the downstream effector of all of these pathways. 63
Raf kinase inhibitors also inhibited the growth of thyroid cancer cells with BRAF or RET/PTC activating mutations. 64
Follicular thyroid carcinomas differ molecularly from papillary cancers and are characterized by RAS mutations and PAX8-PPARγ rearrangements (t(2;3)(q13;p25). The fusion of the thyroid transcription factor PAX8 and the steroid nuclear hormone receptor PPARγ has been detected in up to 50% of follicular thyroid cancers, but not follicular adenomas nor papillary thyroid cancers, and results in a distinct gene expression profile. 65, 66
The rearrangement likely inhibits cell differentiation while stimulating growth, functioning as a dominant negative inhibitor of the wild type PPARγ receptor, the latter probably serving as a tumor suppressor. 67
In vitro, PPARγ agonists led to reduced growth of follicular carcinoma tumor cells, and thus the clinical study of PPARγ modulators in follicular cancers, such as the thiazolidenediones (pioglitazone and rosiglitazone) is warranted. 68
RAS mutations and PAX8-PPARγ rearrangements are rarely found in the same tumor, suggesting two separate molecular pathogenic pathways for this disease. 69
Point mutations in H-RAS and N-RAS have been detected in follicular thyroid carcinomas. 70, 71
As RAS mutations are also seen in papillary carcinomas, it is possible that these mutations contribute to tumorigenesis in conjunction with other oncogenes, and are not specific to follicular carcinomas and are not serving as the primary instigators of malignancy.
While the genes discussed above have been implicated in the initial pathogenesis of thyroid cancers, other growth factor receptors appear to play a role in the progression and behavior of thyroid carcinomas. For instance, the vascular endothelial growth factor (VEGF) is detected at increased levels in papillary and follicular thyroid cancers compared to hyperplastic or benign thyroid tissue, and is associated with increased risk for recurrence and metastatic disease. 72–75
Targeting angiogenesis in this disease is thus another area of extensive research. The growth of tumors in anaplastic thyroid cancer mouse xenografts, for example, was curtailed by AZD2171, a potent inhibitor of the vascular endothelial growth factor receptors (VEGFR). 76
The EGFR (epidermal growth factor receptor) is also expressed and phosphorylated in thyroid cancer cell lines and tissue. 77
It is overexpressed in human thyroid cancer as well, and is associated with a worse prognosis. 78
EGFR activation may also result in activation of c-met signaling. 79
There are also provocative interactions between the EGFR and the RET/PTC fusion protein. It is known that the RET/PTC oncogene dimerizes, resulting in autophosphorylation of tyrosine kinase motifs and constitutive activation of downstream signaling. 80
Inhibition of EGFR decreases RET autophosphorylation, and the two proteins co-immunoprecipitate from cell lysates, and thus appear to form a complex. 81
In vitro, a role for EGFR activation in thyroid cancer is also supported by the finding that the EGFR and multi-kinase inhibitors, gefitinib, PKI1166 and AEE788 had growth inhibitory effects in cell lines with the RET activating mutation. AEE788 also has activity against VEGFR, and this compound also had inhibitory effects on ATC xenografts. 82
The agent ZD6474 (vandetanib), which targets RET, VEGFR and EGFR, is another attractive compound to explore in this setting, and in preclinical studies does limit the growth of thyroid cancer cells with the RET/PTC activating rearrangement. 83
Activation of Akt, a downstream effector of PI3kinase, has been observed in follicular and papillary thyroid cancer cells. 84
Inhibition of Akt did result in decreased cell proliferation and increased apoptosis in thyroid carcinoma cell lines in vitro. It is also well known that patients with Cowden’s syndrome, who have a loss of PTEN resulting in activation of the Akt pathway, are at increased risk of developing thyroid cancer. A mouse model of follicular thyroid adenoma has been generated by engineering a loss of PTEN in the thyroid follicular cells, but another genetic event is likely required for malignant transformation. 85
Anaplastic and undifferentiated thyroid cancers are aggressive malignancies that are not responsive to radioactive iodine or other systemic cytotoxic therapies. It is felt that originally differentiated thyroid cancers undergo additional molecular changes resulting in clonal evolution to a less differentiated variant. One such genetic event includes p53 mutations, which have been detected in anaplastic carcinoma cell lines, but not in the more differentiated histologies. 86, 87
In one series, evidence of p53 mutations was noted in cells which also harbored BRAF mutations, suggesting that both events are important to malignant transformation. 87
Mutations in the catalytic subunit of PI3K, PIK3CA, have also been observed in ATC cell lines. 88
Modification of the extracellular matrix likely influences malignant progression and metastatic potential as well. For instance, E-cadherin expression is low in recurrent, metastatic thyroid carcinomas, but is present in less advanced, well-differentiated cancers. 89, 90
β-catenin, which associates with cadherins, was found to have low expression as well, with increased nuclear localization of this molecule. 91
In the future, further delineation of the molecular events associated with the undifferentiated histologies will aid greatly in the development of more effective therapies for this aggressive and notoriously refractory group of carcinomas.
Medullary thyroid cancer (MTC) is characterized by activating mutations in the RET proto-oncogene with constitutive activity of this tyrosine kinase receptor. The majority, 75%, of MTCs are sporadic, with mutations in RET detected in up 25–66% of this population. 92
Most of these somatic mutations are in exon 16. In contrast, the remaining 25% of MTCs that are familial as part of the MEN2 (multiple endocrine neoplasia type 2) syndrome nearly all carry RET mutations, often in exons 10 or 11. 93, 94
Systemic Therapy for Metastatic Disease
Novel multi-kinase small molecule inhibitors are being actively studied in I-131 non-avid papillary and follicular carcinomas (DTC), as well as medullary thyroid cancers (MTC), with encouraging early results. lists these agents, the targets they act upon, and the subtypes of thyroid cancer in which they have been studied. It is clear from the Table that a shared mechanism of action for most of the agents is to inhibit angiogenesis, with most of these molecules binding to VEGF or VEGF receptors. In addition, RET is an obvious target, given its role as an initiating factor in papillary and medullary thyroid cancer. summarizes the numerous phase 2 studies that have been conducted in recent years with these agents, several of which will be discussed further below.
Current small molecule inhibitors in trials for thyroid cancer and their targets
Response rates in recent phase 2 studies in thyroid cancer
Motesanib is an oral inhibitor of the VEGF receptors 1, 2, and 3; PDGF (platelet derived growth factor); KIT; and Ret. An open label, single arm, multicenter phase 2 study was conducted in patients with locally advanced or metastatic differentiated thyroid cancers that were refractory to radioactive iodine therapy. 95
In total, 93 patients (61% PTC) were treated with 125mg of motesanib once daily. 83% had not received prior chemotherapy. Tumor genotyping was conducted, but was not performed in 52% of samples; 30% had the BRAF(V600E) mutation, and 18% had RAS mutations, but neither finding was clearly associated with response. The objective response rate was 14%, with stable disease observed in 67% of patients. The median duration of response was 32 weeks, with a median progression free survival (PFS) of 40 weeks. Grade 3 toxicities occurred in 55% of patients, including diarrhea, hypertension, fatigue, nausea and anorexia. Of note, 12 patients discontinued treatment due to adverse events and 5 developed cholecystitis.
Axitinib is also a potent oral anti-angiogenesis agent, with selective inhibition of VEGF receptors 1, 2, 3. An open label phase 2, single arm, multicenter study enrolled 60 patients. 96
All histologies were included (papillary-50%, follicular-25%, anaplastic-3% and medullary-18%) as long as the disease was not appropriate for treatment with I-131. Therapy was started at a dose of 5 mg orally twice daily, with an option to increase to 7 mg and 10 mg twice daily if there were minimal toxicities. Only 15% of patients had received prior chemotherapy. The objective response rate was 30%, with disease stability greater than 16 weeks in an additional 38%, and a median PFS of 18.1 months. Treatment benefit and response was noted in all histologies. Due to adverse events, 8 patients discontinued therapy. Grade 3 and 4 toxicities occurred in 19 patients (32%) and 3 patients, respectively, and included hypertension, diarrhea, proteinuria and fatigue. A decrease in soluble VEGFR levels was noted in response to treatment, but no conclusions could be made in regards to how this correlated with response.
Sunitinib, an oral tyrosine kinase inhibitor that inhibits RET, PDFGR and VEGFR is an attractive agent to study given its dual actions as an anti-angiogenic agent and a RET inhibitor. A recently presented phase 2 study treated 43 patients with all histologies of thyroid cancer (37 DTC, 6 MTC) with sunitinib at a dose of 50mg orally daily for 4 weeks, followed by 2 weeks off. 97
Response assessments were as follows: in DTC PR 13%, SD 68% and in MTC the SD rate was 83%. Grade 3–4 toxicities included neutropenia (26%), thrombocytopenia (16%), hypertension (16%), fatigue (14%), palmar-plantar erythrodysesthesia (14%), and gastrointestinal tract events (14%). Preliminary results from two other phase 2 studies with this agent were reported at the 2008 American Society of Clinical Oncology meeting. In the first, 17 patients have been enrolled so far with a PR in 1 and SD in 12. 98
In another phase 2 trial, a schedule of sunitinib 37.5 mg orally daily was used to treat DTC and MTC. After enrollment of 18 patients, preliminary results indicate 44% of patients had a PET response after 7 days, but further efficacy data are not mature. 99
Sorafenib, another multi-kinase inhibitor against Raf, VEGFR and PDGFR has also been studied in several phase 2 studies. One open label phase 2 study has treated 36 patients with iodine refractory disease (papillary 61%, follicular 28%, anaplastic 5.5%) and MTC (5.5%). 100
Seven patients had a PR, and an additional 20 had SD. Grade 3 toxicities reported to date include hypertension and palmar-plantar erythrodysesthesia. Early correlative studies show a decrease of pERK and pAKT in post-treatment tissue. 101
An additional study has been reported in which 18 patients (10 MTC, 8 DTC) were treated, but it is too early to draw conclusions about response. 102
Additional agents that have been evaluated include gefitinib and lenalidomide. In a phase 2 study of gefitinib, there were no objective responses noted in the 25 patients treated. Thus, while there are preclinical data regarding EGFR activation in thyroid cancer, EGFR inhibitors may have limited single agent activity. 103
Preliminary results with the anti-angiogenic agent lenalidomide showed favorable activity in DTC, with a PR of 22% and SD in 44% of the first evaluable 18 patients. However, this is a small number of patients, with only short-term follow-up. Grade 3 toxicities were primarily hematologic. 104
Vandetanib, an oral agent that targets RET, VEGFR and EGFR, has been studied predominantly in hereditary MTC, which would harbor a RET mutation. In one study, vandetanib was administered at a dose of 100mg orally daily to 19 patients. 105
16% of patients had a confirmed PR and 63% had SD. Three patients withdrew due to adverse events, and 6 patients had grade 3–4 events. An earlier study had treated 30 patients with hereditary MTC with a higher dose of 300mg daily. A PR was noted in 17%, with SD in 50%.106
Finally, another interesting agent, XL184, is in the process of being studied for medullary thyroid cancer. This inhibitor of RET, MET and VEGFR2 was evaluated in a phase 1 study in advanced solid tumor patients. In this study, 22 patients with MTC were treated; of interest, this number included patients who had received prior tyrosine kinase inhibitor therapy with either motesanib, vandetanib or sorafenib. Nearly all the MTC patients treated had a treatment benefit with 47% PR and 53% SD. Grade 3 toxicities included low rates of nausea, diarrhea, mucositis and increased liver function tests. This drug is currently being studied further in an ongoing trial in patients with MTC.
Many of the agents that were discussed above are still being evaluated in ongoing studies for advanced thyroid malignancies. Additional agents, such as the MEK inhibitor AZD6244 and the anti-angiogenesis small molecule pazopanib, are in trials now. There are a number of obvious challenges. One is patient selection. We have reviewed the principal genetic lesions driving the development and evolution of thyroid cancer. What we do not know is how well any of these lesions predicts response to a given agent or class of agents. For instance, based on preclinical data, a BRAF mutation may predict for responsiveness to agents that target MEK, but this needs to be studied prospectively in clinical trials. As well, the mechanisms of resistance, primary and acquired, to these drugs need to be studied further in order to optimize their use.
A second challenge is be able to determine when a patient should begin treatment with one of these agents. Although less toxic than traditional cytotoxic therapies (including doxorubicin), these drugs nevertheless have a number of very significant toxicities. One must also consider that many patients might end up being treated for prolonged periods of time if not the rest of their lives. Clinical trials designs must therefore take into account that advanced thyroid cancer, even when destined to be fatal, is an indolent metastatic cancer compared to many malignancies. Stable disease is an obvious endpoint that has already emerged from the ongoing trials but the clinical significance of stable disease remains a matter of ongoing scientific and regulatory debate. With so many potentially active agents for a relatively rare disease, designing feasible (completable) phase 3 studies to eventually merit FDA registration will also require an evaluation of patient resources and the creation of effective, likely international, research consortia. An additional relevant question is whether or not these agents need to be or should be compared to current approved FDA drug, doxorubicin, knowing that this approval occurred many years ago in a different therapeutic and regulatory environment. An international registration study for axitinib for patients who had formally failed doxorubicin was planned, but accrual was poor due to the requirement for prior doxorubicin treatment that patients and physicians were hesitant to consider.
While the data to date are preliminary it is clear that we are in a new era in the treatment of advanced thyroid cancer. The new drugs will provide significant insight into the disease itself and likely alter the natural history of what had previously been an untreatable medical illness. The next steps are clear, namely to validate the clinical utility of these agents in well-designed clinical trials, and understand the determinants of clinical response and resistance.