Multicenter Randomized Double-Blind Phase III Trial of Donafenib in Progressive Radioactive Iodine-Refractory Differentiated Thyroid Cancer

The incidence of thyroid cancer in China has seen an increase, with more than 221,093 new cases reported, comprising more than one-third of the global total (1, 2). According to Surveillance, Epidemiology and End Results (SEER) database, patients with thyroid cancer usually carry a favorable prognosis with the 5-year survival rate of 98.5% in the United States, while the counterpart in China is only 84.3% (3, 4), indicating a substantial life-threatening situation among Chinese patients. Differentiated thyroid cancer (DTC) accounts for the majority (over 90%) of thyroid cancer (5), most of which could achieve favorable response after standard treatments, including surgery, selective radioactive iodine (RAI) treatment and thyrotropin (TSH) suppression. More than one-third of high-risk patients with DTC with locally advanced disease or distant metastasis eventually develop resistance to RAI therapy, which is identified as RAI-refractory DTC (RAIR-DTC; refs. 6–8). The 10-year survival rate of patients with RAIR-DTC with distant metastasis is around 10%, rendering RAIR-DTC a major clinical concern (8–10).

In the past decades, several signaling pathways and activating mutations have been identified as being involved in thyroid tumorigenesis, including the MAPK pathway and proangiogenic factors such as vascular endothelial growth factor (VEGF) and platelet-derived growth factor (PDGF; refs. 11–13). These findings have spurred the development of tyrosine kinase inhibitors (TKI) in managing RAIR-DTC, including sorafenib and lenvatinib. On the basis of findings from the DECISION and SELECT studies, sorafenib and lenvatinib are recommended as the first-line therapy, with lenvatinib preferred (14, 15). Of note, in 2017, the incremental cost-effective ratios (ICER) for sorafenib and lenvatinib were $64,067 per quality-adjusted life-year (QALY) and $25,275 per QALY, respectively. These ratios were significantly higher than China's per capita gross domestic product (GDP) of $8,817 in the same year. Consequently, these costs far exceeded the financial means of most Chinese patients, making them unaffordable (ref. 16; https://data.worldbank.org/indicator/NY.GDP.PCAP.CD?locations=CN.pdf). Meanwhile, the incidence of treatment-related adverse events (TRAE) of grade 3 or higher for lenvatinib was 87.4% among Chinese patients (17), while that for sorafenib was 41.8% from global patients in DECISION study, but not yet available for those enrolled Chinese patients. The concerning safety profile such as hand–foot syndrome (HFS), hypertension, or proteinuria, becomes the major worry among Chinese patients, which prevented both of them from wide clinical application (14, 15).

In 2016, with the above-mentioned clinical setting, the exploratory phase II/III clinical trials of donafenib, a new domestically developed agent, were approved in China. This approval came at a time when there were no TKI available for Chinese patients with RAIR-DTC. It was only in 2017 and 2020 that sorafenib and lenvatinib, respectively, received their final approvals. Donafenib was explored as a modified form of sorafenib with a trideuterated N-methyl group, potentially enhancing molecular stability with an improved pharmacokinetic profile. It has been demonstrated to inhibit the activity of multiple receptor tyrosine kinases such as VEGF receptor, PDGF receptor, and various Raf kinases, thereby suppressing tumor cell proliferation and angiogenesis. In a phase III study for hepatocellular carcinoma (HCC), donafenib was superior to sorafenib in improving overall survival (OS) and had favorable safety in patients with advanced HCC (18). In its phase II dose-exploratory study for RAIR-DTC, the 300-mg regimen (300 mg twice daily) appeared to be more clinically beneficial than the 200-mg regimen (200 mg twice daily) in terms of prolonged progression-free survival (PFS; 14.98 vs. 9.44 months; P = 0.351), favorable objective response rate (ORR; 13.3% vs. 12.5%; P = 1.000), and acceptable safety profile (19). On the basis of promising results, the sequential phase III clinical trial was continued to assess PFS among patients with RAIR-DTC who received donafenib with an initial dose of 300 mg twice daily over placebo.

This was a multicenter, randomized, double-blind, placebo-controlled, phase III trial, aimed to evaluate the efficacy and safety of donafenib for locally advanced or metastatic RAIR-DTC based on the best supportive treatment (DIRECTION). The patients enrolled in this study met the following key criteria: (i) age 18 years or older, (ii) locally advanced or metastatic RAIR-DTC (confirmed by histologic type: papillary, follicular, Hürthle cell, or poorly differentiated cancer), (iii) center-confirmed progression within the past 14 months according to Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST 1.1), prior TKI treatments were allowed, but should be abstained for at least 4 weeks as a washout period, (iv) at least one measurable lesion identified by CT or MRI according to RECIST 1.1, (v) expected survival time longer than 12 weeks, (vi) adequate organ function, and Eastern Cooperative Oncology Group (ECOG) performance status score 0–2. RAIR-DTC was characterized by the presence of at least one measurable lesion that did not show uptake of RAI on diagnostic scans or after ¹³¹I treatment, or at least one measurable lesion that progressed within 14 months after the last ¹³¹I treatment with a dose of 3.7 ∼ 7.4 GBq (100 ∼ 200 mCi) despite the RAI avidity, or the cumulative ¹³¹I activity of at least 22 GBq (600 mCi) for RAI treatment (20, 21). Identification of RAIR-DTC was based on the conditions of successful thyroid ablation, TSH > 30 mIU/L, and no interference of exogenous iodine (following low-iodine dietary restrictions and avoiding iodine contamination from IV contrast agents before RAI therapy).

Patients were randomized by the Central Randomization System-Interactive Web Response System (IWRS) in a ratio of 2:1 to receive donafenib (300 mg twice daily) or matching placebo in 28-day cycles. Study drugs were distributed to patients by investigators who were blind to treatment assignments through the exclusive drug number specified by IWRS. Randomization was stratified with distant metastasis (M0 vs. M1) and TKI treatment history (yes vs. no) using a stratified blocked randomization method.

Safety was assessed every 4 weeks and efficacy evaluation according to RECIST 1.1 was conducted every 8 weeks until progression or intolerable toxicity. The study drug interruption and dose reduction (up to an adjustment of four doses) were allowed when unacceptable adverse events (AE) related to the study drug occurred. The dose reduction scheme is to sequentially reduce to 200 mg twice daily, 300 mg once daily, 200 mg once daily, and 200 mg once every other day. When grade 3 or above AE occurred and didn't recover to grade 1 or less within 2 weeks after drug interruption, the test drug should be discontinued permanently.

The treatment regimen would be unmasked when disease progression was assessed and confirmed by an independent review committee (IRC). According to the trial protocol, since there was no approved TKI available at the time of approval, an open-label period was included. During this period, patients were allowed to continue receiving the treatment as determined beneficial by the investigators. In the open-label period, patients in the donafenib and placebo groups could receive donafenib until the second progression or intolerable toxicity.

The primary endpoint was PFS defined as the time from randomization to disease progression documented first and assessed by the IRC or to death from any cause, whichever occurred first. Secondary endpoints included ORR [defined as the number of patients achieving an overall best response of complete response (CR) or partial response (PR) divided by the total number of patients], OS (defined as time from randomization to death due to any cause), time to progression (TTP, defined as time from randomization to disease progression), disease control rate [DCR, defined as the number of patients achieving an overall best response of CR, PR or stable disease (SD) for ≥ 6 weeks(or ≥ 6 months in post hoc analysis) divided by the total number of patients], the changes of thyroglobulin (Tg) and antithyroglobulin antibody (TgAb). Biochemical response defined as a decreased Tg level ≥ 25%, biochemical progression defined as an increased Tg level ≥ 25%, and Tg II and anti-Tg by Roche Diagnostics GmbH were used to measure serum Tg and TgAb (22). AE were evaluated according to National Cancer Institute (NCI) Common Terminology Criteria for Adverse Events version 4.03.

The primary efficacy analysis was based on the intention-to-treat (ITT) population, including all randomized patients. ORR and DCR were analyzed on the basis of the interim-ITT (iITT) population, which was a subset of ITT population and defined as patients with at least one postbaseline tumor assessment or who discontinued double-blind treatment due to any reason. The safety analysis set included all randomized patients who received at least one dose of study medication.

For the primary endpoint of PFS, it was estimated that 121 PFS events were required in the final analysis to achieve 80% power to detect an hazard ratio (HR) of 0.58 (assumed on the basis of the results of donafenib phase II clinical trial in the locally advanced or metastatic RAIR-DTC and sorafenib DECISION trial) at a two-sided significance level (α) of 0.05 with a 2:1 ratio of allocation to the experimental group or control group. We was planned to perform the first and the second interim analyses when 55 (46% of total PFS events) and 81 PFS events (67% of total PFS events) were observed, respectively. A Lan-DeMets (O'Brien & Fleming) α spending function was used to control the overall type I error (false positive) probability. Assuming that PFS data maturity was 59%, approximately 204 patients were required (136 in the donafenib group and 68 in the placebo group). An independent data monitoring committee (IDMC) consisting of two medical experts and one statistical expert was formed to review the interim analysis results and make recommendations on whether to stop the trial when the efficacy criteria of early termination were met.

The median PFS for each treatment group and the associated 95% confidence intervals (CI) were estimated using the Kaplan–Meier method. A stratified log-rank test was used in the main PFS analysis to test if the difference between treatment groups was statistically significant. The HR with its 95% CI of PFS based on the comparison between treatment groups was estimated using a stratified Cox proportional hazards model. The stratification factors including history of prior TKI therapy (yes vs. no) and distant metastasis (yes vs. no) were prespecified for the analyses. Subgroup analyses were also performed for prespecified baseline variables to evaluate the consistency across different subgroups by means of unstratified Cox proportional hazards model. Differences in ORR and DCR were analyzed using the Cochran–Mantel–Haenszel test stratified by history of prior TKI therapy (yes vs. no). The descriptive summary was provided for all safety parameters in the safety analysis set. Statistical analysis was performed using SAS version 9.4.

The study was conducted in accordance with good clinical practice (GCP) guidelines and the Declaration of Helsinki. All patients provided written informed consent before screening. The study protocol and each revision were reviewed and approved by the ethics committee of all participating centers. The trial was registered under ClinicalTrials.gov as NCT03602495.

The data generated in this study are not publicly available due to information that could compromise patient privacy but are available upon reasonable request from the corresponding author.

From August 29, 2018 to February 28, 2021, a total of 191 patients in 34 clinical centers across China were randomized to receive donafenib (n = 128) or placebo (n = 63). The baseline characteristics of the patients were well balanced in these two groups (Table 1). At the time of data cutoff, the median follow-up time was 10.9 months (range, 0–30 months) in the donafenib group and 13.2 months (range, 1–29 months) in the placebo group, respectively. A total of 72 patients [donafenib, 35 (27.3%) and placebo 37 (58.7%)] ended blinded treatment due to disease progression, and three patients [donafenib, one (0.8%) and placebo, two (3.2%)] ended due to death, finally, 92 patients [donafenib, 72 (56.3%) and placebo, 20 (31.7%)] were continuing receiving blinded treatment, 26 patients (20.3%) with donafenib, and 39 patients (61.9%) with placebo received open-label donafenib (Fig. 1).

The primary endpoint of PFS was met at the time of the second interim analysis (February 2021) when 66 and 85 PFS events were observed, resulting in two-sided α of 0.0048 and 0.0135 being allocated to the two interim analyses, respectively. Significant improvement in PFS for donafenib was observed in comparison with placebo (median, 12.9 vs. 6.4 months; HR, 0.39; 95% CI, 0.25–0.61; P < 0.0001; Fig. 2A). The 12-month PFS rates were 54.1% (95% CI, 42.4–64.5) and 26.1% (95% CI, 14.1–39.9) in the donafenib and placebo group, respectively.

In the prespecified subgroups analysis, all participants tended to benefit from donafenib in terms of PFS, regardless of clinicopathologic features (Fig. 3). Among those who received prior TKI treatments, accounting for 18.3% (35/191) of all enrolled patients (Supplementary Table S3), the median PFS was 11.0 months with donafenib versus 3.7 months with placebo (HR, 0.23; 95% CI, 0.09–0.61; Supplementary Table S7), and for the TKI-naïve subgroup was 18.3 months versus 7.4 months (HR, 0.45; 95% CI, 0.27–0.73) accordingly. Distant metastasis as a stratification factor was not further analyzed, due to only 2.6% of M0 patients in this study. In addition, during the open-label period, the median PFS was 5.8 months and 5.6 months for the patients with prior donafenib and placebo, respectively (Supplementary Table S6).

Donafenib was associated with improved ORR [based on iITT all confirmed PR (23.3%; 28/120) over placebo (1.7%; 1/58); P = 0.0002; Supplementary Table S2]. The median time to confirmed objective response for donafenib was 1.84 months (95% CI, 1.81–2.83). The median duration of response for patients with donafenib was 16.53 months (95% CI, 10.97–NE). Higher incidence of tumor shrinkage was observed in the donafenib over placebo (Fig. 2B). For the patients with prior TKI treatments, the ORR was 0 (95% CI, 0–14.8%) with donafenib versus 10% (95% CI, 0.3%–44.5%) with placebo (Supplementary Table S7). During the open-label donafenib treatment period, the ORR was 0 (95% CI, 0.0–13.2%) and 12.8% (95% CI, 4.3%–27.4%) for the patients with prior donafenib and placebo, respectively (Supplementary Table S6).

Improved DCRs for ≥ 6 weeks and ≥ 6 months from randomization were observed in the donafenib arm over the placebo arm with 93.3% (112/120) versus 79.3% (46/58; P = 0.0044) and 52.5% (63/120) versus 29.3% (17/58; P = 0.0032; Supplementary Table S2), respectively. The median TTP was 12.9 months with donafenib versus 7.3 months with placebo (HR, 0.381, 95% CI, 0.24–0.60; P < 0.0001). Twenty-four patients (donafenib, 13; 10.2% and placebo, 11; 17.5%) died before the data cutoff. Due to the limited number of OS events observed, the median overall OS has not yet been determined. This situation could be attributed to the study's crossover design, where more than 60% of patients initially receiving placebo switched to receiving donafenib. These patients demonstrated a subsequent PFS advantage when they crossed over to donafenib treatment (Fig. 2C). The 6-, 12-, and 18-month survival rates were 100%, 91.8%, and 88.3% in the donafenib group and 98.3%, 89.0%, and 73.5% in the placebo group, respectively.

Median serum Tg dropped from baseline rapidly in the first 8 weeks and remained at a low level over treatment in the donafenib group, but in the placebo group the median serum Tg increased from the baseline and remained upward (Fig. 2E). Early in cycle two, biochemical response was observed in 74.1% of patients in the donafenib group and 9.3% in the placebo group, while biochemical progression was observed in 4.7% of patients with donafenib and 34.9% of patients with placebo (Supplementary Fig. S1).

The median duration of treatment was 233.5 days (range, 4–910 days) among patients who received donafenib and 176.0 days (range, 21–872 days) among patients who received placebo. The average daily dose was 522 mg/day in the donafenib group, accounting for 87.0% of the initial intensity (600 mg/day). In the double-blind period, among patients receiving donafenib, 85 (66.41%) maintained the initial dose of 600 mg/day, while 29 (22.66%), nine (7.03%), and five (3.91%) modified it to 400 mg/day, 300 mg/day, 200 mg/day, respectively (Fig. 2D).

TRAE occurred in 99.2% (127/128) of patients receiving donafenib and in 44.4% (28/63) of patients receiving placebo during the double-blind period (Table 2). Overall, TRAE of grade 3 or higher occurred in 43.8% of patients in the donafenib group. The most frequent TRAE in the donafenib group were HFS (84.4%, 108/128), alopecia (67.2%, 86/128), and diarrhea (63.3%, 81/128; Table 2). The TRAE of grade ≥ 3 (with incidence of > 5%) included hypertension (13.3%, 17/128) and HFS (12.5%, 16/128). Serious AE occurred in 22.7% (29/128) of patients with donafenib, of which 10 cases were treatment related. Correspondingly, serious AE occurred in 19.0% (12/63) of patients in the placebo group. Serious AE that occurred in 2% or more of patients with donafenib were lung inflammation (3.1%, 4/128) and abnormal liver function (2.3%, 3/128). Two treatment-emergent deaths occurred in the donafenib group and two in the placebo group, which were all attributable to underlying disease. (Supplementary Table S4, S5)

TRAE leading to dose reduction or interruption occurred in 42.2% (54/128) of patients taking donafenib and 1.6% (1/63) of patients taking placebo. The most common reasons for donafenib dose reduction or interruption were HFS (16/128, 12.5%), hypertension (14/128, 10.9%), and diarrhea (5/128, 3.9%). Treatment was discontinued in 6.3% (8/128) of patients taking donafenib. The most frequent causes of donafenib dose discontinuation were abnormal liver function (3/128, 2.3%) and rash (2/128, 1.6%).

In this study that enrolled the largest Chinese RAIR–DTC cohort, donafenib demonstrated clinical benefit with prolonged PFS and reduced disease progression or death by 61.1% over placebo (PFS 12.9 vs. 6.4 months, HR, 0.39). The ORR for donafenib was 23.3%; a longer OS trend was also observed with a 14.8% increase of 18-month survival rates (88.3% vs. 73.5%), even though the OS was not mature upon interim analysis. An exploratory subgroup analysis of PFS showed consistent improvement in all subgroups. Of note, among patients who never received prior TKI, donafenib exhibited a longer PFS over placebo (18.3 vs. 7.4 months), suggesting the earlier administration, the more prolonged PFS obtained in these TKI-naïve patients, which guarantees it as a first-line TKI option indicated for progressive RAIR–DTC patients. In addition, the median time to objective response of 1.84 months indicated the rapid tumor shrinkage by donafenib. In the DECISION study, no patients with prior TKI were allowed. However, in the DIRECTION study, 18.3% of patients with prior TKI including sorafenib, lenvatinib, apatinib, anlotinib, etc. were enrolled and continued to benefit from donafenib (HR, 0.227), indicating its potential as a second-line treatment option.

Tg is a sensitive and convenient biochemical responsive marker for DTC after total thyroidectomy and RAI therapy and might be a surveillance marker to comprehensively reflect the whole-body tumor burden. In this study, by using Tg evaluation, 74.1% of patients with biochemical response to donafenib were identified against 9.3% for placebo as early as two cycles after treatment. Whereas through RECIST evaluation, only 23.3% of patients with structural objective response to donafenib against 1.7% for placebo were identified, suggesting the biochemical response reflected by Tg might be a rapid and sensitive indicator comparing with the widely accepted RECIST evaluation. There remains an issue of debate regarding the exact comparison as well as association between biochemical and structural response to TKI in patients with RAIR–DTC, which awaits evidence from further investigation.

In this study, the TRAE spectrum was generally consistent with the known donafenib safety profile reported in the phase II trial for RAIR-DTC and the phase III trial for HCC (18, 19). Donafenib also exhibited promising safety particularly in the 13.3%, 1.6% and 0 occurrence rate of ≥ grade 3 hypertension, proteinuria, and death, respectively. The favorable safety profile probably lies in its refined pharmacokinetic feature, thus allowed the high adherence of average daily dose to its initial dose (87.0%) and added promise to its sustainable antitumor effect.

This article had several limitations. First, the primary goal of treatment for patients with advanced tumors is to prolong survival. However, due to the relatively long survival period of patients with thyroid cancer (median OS of 2.5–3.5 years), the primary endpoint of clinical studies on thyroid cancer is usually PFS instead of OS (such as DECISION and SELECT studies). At the same time, the cross over design of this study also influenced the accurate evaluation of OS benefit to some extent. Second, when the fixed phase II/III study was initiated, TKI drugs such as sorafenib and lenvatinib were not approved in China, so placebo control was selected. The comparison of donafenib with other TKI agents still requires additional real-world investigations. Finally, all patients enrolled in this study were Chinese, and a more geographically diverse study is needed to further extend our findings to other patient populations.

In conclusion, donafenib was well-tolerated, and demonstrated clinical benefit in terms of improved PFS, ORR, and DCR in patients with RAIR-DTC. The results suggest that donafenib could be a new treatment option for patients with RAIR-DTC.

No disclosures were reported.

Y. Lin: Conceptualization, resources, formal analysis, funding acquisition, methodology, writing–original draft, project administration, writing–review, and editing. S. Qin: Conceptualization, resources, formal analysis, writing–review, and editing. H. Yang: Formal analysis, project administration, writing–review, and editing. F. Shi: Formal analysis, project administration, writing–review, and editing. A. Yang: Formal analysis, project administration, writing–review, and editing. X. Han: Formal analysis, project administration, writing–review, and editing. B. Liu: Formal analysis, project administration, writing–review, and editing. Z. Li: Formal analysis, project administration, writing–review, and editing. Q. Ji: Formal analysis, project administration, writing–review, and editing. L. Tang: Formal analysis, project administration, writing–review, and editing. Z. Deng: Formal analysis, project administration, writing–review, and editing. Y. Ding: Formal analysis, project administration, writing–review, and editing. W. Fu: Formal analysis, project administration, writing–review, and editing. X. Xie: Formal analysis, project administration, writing–review, and editing. L. Li: Formal analysis, project administration, writing–review, and editing. X. He: Formal analysis, project administration, writing–review, and editing. Z. Lv: Formal analysis, project administration, writing–review, and editing. Q. Ma: Formal analysis, project administration, writing–review, and editing. Z. Shen: Formal analysis, project administration, writing–review, and editing. Z. Guo: Formal analysis, project administration, writing–review, and editing. Z. Chen: Formal analysis, project administration, writing–review, and editing. Y. Cui: Formal analysis, project administration, writing–review and editing. J. Tan: Formal analysis, project administration, writing–review and editing. Z.Gao: Formal analysis, project administration, writing–review and editing. S. Jing: Formal analysis, project administration, writing–review and editing. K. Lu: Formal analysis, project administration, writing–review, and editing. X. Luo: Formal analysis, project administration, writing–review, and editing. Y. Zhang: Formal analysis, project administration, writing–review, and editing. Y. Fang: Formal analysis, project administration, writing–review, and editing. Z. Li: Formal analysis, project administration, writing–review, and editing. Y. Cheng: Formal analysis, project administration, writing–review, and editing. S. Lei: Formal analysis, project administration, writing–review, and editing. S. Luan: Formal analysis, project administration, writing–review, and editing. G. Chen: Formal analysis, project administration, writing–review, and editing. G. Wang: Formal analysis, project administration, writing–review, and editing. L. Wu: Conceptualization, funding acquisition, writing–review, and editing. L. Liu: Visualization, methodology, writing–review, and editing.

This work received support from Zelgen Biopharmaceuticals Co., Ltd., the Project on Inter-Governmental International Scientific and Technological Innovation Cooperation in National Key Projects of Research and Development Plan (No. 2019YFE0106400), the Nonprofit Central Research Institute Fund of Chinese Academy of Medical Sciences (No. 2019XK320009), and the Chinese Academy of Medical Sciences (CAMS) Innovation Fund for Medical Sciences (CIFMS; No. 2020-I2M-2–003). We express our gratitude to all the patients, their families, the investigators, and the teams involved in this trial. Special thanks to Prof Shukui Qin, M.D., from PLA Cancer Centre, Nanjing Jinling Hospital, Nanjing, China, for providing professional advice on protocol design. We also acknowledge the valuable medical writing support from Zhuanzhuan Mu, Xin Zhang, and Yingjie Zhang, funded by Zelgen.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.