Clinical Trials for Progressive Differentiated Thyroid Cancer: Patient Selection,Study Design,and Recent Advances

The targets in cell signaling and angiogenesis

In 80% of the papillary thyroid carcinomas (PTCs), activating mutations have been found in genes encoding signaling molecules upstream of the mitogen-activated protein (MAP) kinase pathway and are believed to be the initiating event. This includes RET/PTC rearrangements, RAS and BRAF point mutations, with no overlap between these mutations. Consistent with the “oncogene addiction” hypothesis that implies continued dependence of the tumor upon such activated signaling, inhibition of these mutant kinases may lead to either tumor stabilization or regression (25). In follicular carcinomas, RAS mutations are found in 20–35% of the cases and PPARγ-PAX8 rearrangements in about 30% of the cases; RET/PTC rearrangements and BRAF mutations have not been found. In addition to the MAP kinase pathway, the phosphatidylinositol 3-kinase pathway may also be activated in PTCs and more frequently in follicular carcinomas (26,27). In poorly differentiated carcinoma, BRAF mutation is rarely found (28). P53 mutations are virtually nonexistent in DTCs, but are present in poorly differentiated and in anaplastic thyroid carcinomas (29).

Other molecular abnormalities have been found in these tumors and are believed to be the secondary events. Overexpression of other tyrosine kinase receptors may be observed in thyroid cancer cells, including fibroblast growth factor (FGF ), epidermal growth factor (EGF ), hepatocyte growth factor (c-Met), and insulin and insulin-growth factor 1 receptors. Acquisition of additional mutations and gene amplifications that activate the phosphatidylinositol 3-kinase signaling pathway may also be a common event during the progression of these malignancies to poorly and anaplastic thyroid carcinoma (30 –32).

Angiogenesis serves a critical role in the development of these hypervascularized tumors. Various vascular endothelial growth factors (VEGFs) and VEGF receptors (VEGFR-1 [Flt-1] and VEGFR-2 [Flk-1 and KDR]) are often overexpressed in thyroid cancer tissues, both in tumor cells and supporting vascular endothelium, and they also trigger the MAP kinase signaling pathway (33 –35). In experimental models, anti-VEGF therapy blocks the growth of DTC (36).

Interferences with signal transduction pathways

The antitumor activity of kinase inhibitors is being studied in DTC patients (37,38). These compounds generally compete with the binding of ATP to the catalytic domain of the kinase, and by doing so inhibit autophosphorylation and activation of the kinase and prevent the activation of the downstream proteins. A single compound may inhibit several kinases, including the VEGFR-2 that contributes to an antiangiogenic effect, and also may block other kinases including RET, BRAF, EGFR, platelet-derived growth factor receptor, FGF receptor, c-MET, and c-KIT (Table 2).

Available results from phase I and II trials with motesanib (39), sorafenib (40 –42), axitinib (43), sunitinib (44), and pazopanib (45) have clearly confirmed the clinical benefits of these compounds, with partial response observed in 8–32% of the patients and long-term stable disease in at least half. Although response criteria in these contemporary trials differ markedly from those evaluating cytotoxic chemotherapy, antitumor efficacy of these agents is likely greater than that of the earlier chemotherapies (Table 2). However, no complete response was achieved, and benefits on survival have not yet been demonstrated. Comparison of the outcome among these compounds is at the present time not possible.

Toxicities differ from those observed with cytotoxic chemotherapy, but may still be significant, including fatigue, hypertension, anorexia, diarrhea, and skin toxicities. These side effects may lead to dose reduction and even to the withdrawal of the drug and should be minimized because benefits may require long-term treatment over months or even years. Also, the serum TSH levels should be regularly monitored as they may increase during treatment with tyrosine kinase inhibitors (46). Of course any increase should lead to an increase in the daily levothyroxine treatment dose.

Given the commercial availability of sorafenib and sunitinib, these agents have entered clinical use for those patients with progressive, radioiodine-refractory disease who are not suitable candidates for clinical trials (47,48).

Trials aimed at restoring radioiodine uptake

Loss of differentiated functions (especially radioiodine uptake) is frequently observed during progression of thyroid carcinoma, often due to epigenetic alterations (49); BRAF mutation may specifically induce hypermethylation of promoters of several functional thyroid genes, leading to the loss of expression. An original treatment procedure is to increase the expression of the sodium iodide symporter and then to treat the patient with radioiodine. Despite promising preclinical models, retinoic acids, rexinoids, the PPARγ agonist rosiglitazone, and the histone deacetylase inhibitor depsipeptide did not produce any clinically relevant effects on tumor progression or patient outcome (50 –53); the BRAF inhibitor sorafenib did not induce any uptake in the nonfunctioning metastases (42). Many metabolic defects are present in the thyroid cancer tissues, and mere reexpression of the sodium iodide symporter gene may not be sufficient to induce a significant tumor uptake and tumor retention of radioiodine (49). Further, these tumors are refractory to radioiodine treatment and considered radioresistant, and low radiation doses will have no beneficial effects.

Baseline work-up and assessment of tumor response

Patients with radioiodine refractory disease for whom clinical trials of investigational therapies are contemplated must be accurately characterized concerning all clinical prognostic indicators, including age, performance status, disease extent and location, and progression rate. The importance of this latter parameter cannot be overemphasized, as many patients with metastatic DTC can be asymptomatically stable for long periods of time, and in such patients, the benefits of novel therapies may be largely outweighed by drug toxicities and rigors of clinical trial participation (2).

Imaging should emphasize identification of all clinically relevant sites of disease, including those tumors large enough to be serially assessed to determine response to therapy as well as those that might require additional localized intervention before systemic treatment, such as radiotherapy for brain metastases. However, lesions may be multiple and difficult to image. Lung metastases are frequently miliary and may be associated with mediastinal lymph node metastases not readily appreciated without contrast enhancement; brain metastases and other rare distant sites of disease involvement may be asymptomatic. Diagnostic procedures should include a spiral computed tomography scan of the neck, chest, and abdomen, with standardized injection of contrast medium, bone scintigraphy, and a computed tomography scan or magnetic imaging resonance (MRI) of the brain. Because slowly growing bone metastases are often difficult to view on bone scintigraphy (54), MRI of the spine and pelvis should be considered. A baseline ¹⁸FDG PET scan may complete the work-up to aid disease localization as well as prognostication, but widespread acceptance of PET imaging to measure tumor response in trials is lacking (54 –56). In the absence of treatment, standardized imaging is repeated every 6–12 months, and progression rate is assessed using RECIST (17,18) (Table 3). Patients with measurable lesions and documented progression in a given time interval (between 6 and 15 months) should be considered candidates for systemic treatment and therefore potential clinical trial participants.

The histopathology of the primary tumor should be accurately reviewed, which in a trial should preferably be confirmed by a central pathologist because this is frequently difficult and not reproducible (57,58). Histological subtype should be clearly characterized, including papillary (classical form and variants), follicular, Hürthle cell, and poorly differentiated carcinomas. Immunohistochemistry should exclude any other type of tumors. Patients with DTC should be evaluated separately from those with anaplastic and MTC for purposes of trial organization. In the future, molecular profiling will help to define more appropriately the homogeneous series of patients and may permit personalized selection of patients based on the matching molecular pathophysiology to drug mechanism of action rather than histology.

The clinical trial population must be well defined. This may decrease the number of patients eligible for study enrollment, but the estimates of efficacy and clinically relevant improvement in response, used to determine sample sizes, also become more precise. In reported studies, a major limitation is that a substantial proportion of patients (33–50%) had stable disease of varying duration as the best response. Given the indolent natural history of many of these tumors, a report of stable disease is of limited value. Therefore, only patients with documented progressive disease should be included, permitting a smaller sample size and a shorter time for follow-up to demonstrate a difference in response to therapy between treatment groups. Eventually, development of more effective therapies with fewer side effects will lead to a broader array of patient populations appropriate to include in trials, including those with stable or smaller extent of disease, and even studies of adjuvant treatments to reduce recurrence in those patients currently considered disease free.

Challenges of identifying appropriate study endpoints

Demonstration of treatment benefits in cancer trials is ultimately based on showing that patients live longer and function and feel better when treated with a new therapy compared with an existing standard of care (59 –62). In the current setting, where there is no effective and approved therapy for metastatic thyroid cancer, the comparator standard of care is being interpreted as best supportive care with a placebo intervention. Statistically, this requires demonstrating significant improvement of the overall survival rate in a randomized trial (treatment vs. standard of care or placebo) and necessitates following treated patients until a sufficient number of deaths have occurred before primary analysis can be performed. For thyroid cancers with fulminant clinical courses, such as anaplastic carcinoma, this is a very reasonable endpoint. But for more indolently lethal disease like metastatic DTC, study feasibility may be severely compromised by the number of patients and duration of the study required, and new and effective therapies may never be tested and approved for patients with these diseases. As a surrogate endpoint, improvement in progression-free survival can be more readily assessed in smaller or shorter randomized trials. However, studies based on this primary endpoint are more susceptible to the introduction of bias, require strict adherence to patient evaluation schedules, and for a variety of reasons may not reliably predict improved overall duration or quality of survival (59 –62). A common modification of this study design incorporates a crossover, by which patients who reach the primary endpoint of progression in the standard of care (or placebo) arm have the option to crossover to the new treatment arm. This design may enhance patient acceptance of the possibility of randomization to the standard of care or placebo arm, thus facilitating study enrollment; on the other hand, concern can be raised that this is a somewhat coercive inducement to patient enrollment that presupposes that the investigational therapy is better, which in fact is the hypothesis being tested (62). An even less predictive surrogate endpoint of survival is tumor response in a single arm, non-randomized trial, based on the measurement of lesion size every 2 to 3 months to permit assessment by RECIST of a therapy's antitumor efficacy (57,58). It should be explained both to patients and care providers that a partial response or even a stabilization when it is long lasting may provide more benefits on survival than a complete response of short duration. Considerable debate persists regarding the optimal sequence and design of trials to permit the most cost-effective and rapid development of beneficial therapies (59,60).

Finally, improvement of overall survival may often be determined as a secondary endpoint in randomized trials whose primary endpoints may be based on the progression-free survival. A complex dilemma arises with this design, though, when the possibility of crossover is considered. Without ever allowing a crossover during the course of the trial, the distinction between the two study arms with regard to any exposure to the investigational treatment remains clear until completion of the study. But acceptance of the study may be hampered, and there may be an ethical question from permanently withholding a potentially more effective treatment from patients. However, if crossover is permitted at the time of patient progression, the secondary endpoint becomes a more narrow (and possibly useless) assessment of the effect of earlier compared with the later exposure to the investigational therapy on the overall survival (61).

Another approach to the challenge of interpreting prolonged stable disease during treatment is the randomized discontinuation study design. This study design may be offered to patients with stable disease during treatment and may provide insights on both the efficacy of the investigational drug and optimal duration of treatment.

Recently, multicenter phase II and phase III trials for novel-targeted therapies have shown that the accrual of a large number of thyroid carcinoma (either DTC or MTC) patients who fulfill strict inclusion criteria is possible in a short time period. Indeed, these rare patients should preferably be included in prospective trials because only such trials may provide reliable insights on drug efficacy. Even phase I trials that are testing the newest therapies, selected based on the relation between drug mechanism and thyroid cancer pathophysiology, should be considered for patients with progressive thyroid cancer, as these protocols may allow early identification of possibly effective drugs. These early phase studies can incorporate various biomarkers to try to predict drug efficacy as early as possible, such as serum VEGF, VEGFR-2, and basic FGF determinations in the case of antiangiogenic drugs, serum markers of apoptosis, circulating endothelial and tumor cells, and functional imaging using various techniques such as FDG PET scan, MRI, or ultrasonography with injection of contrast medium. However, the use of these markers has not yet been validated in clinical practice for thyroid cancer. The tumor marker, thyroglobulin, may be of interest, given its value in identifying the presence and volume of disease, but the predictive value of this biomarker may be reduced by the presence of circulating antithyroglobulin antibodies and variations in TSH levels.

Toxicity and health-related quality of life should also be critical endpoints in these frequently asymptomatic patients with a long life expectancy (9,63). Finally, these treatment modalities should be submitted to careful socioeconomic evaluations, particularly given the high cost of developing these targeted therapies and their eventual financial burden on healthcare systems.