Increased Progression-Free Survival with Cabozantinib Versus Placebo in Patients with Radioiodine-Refractory Differentiated Thyroid Cancer Irrespective of Prior Vascular Endothelial Growth Factor Receptor-Targeted Therapy and Tumor Histology: A Subgroup Analysis of the COSMIC-311 Study

Background: Lenvatinib and sorafenib are standard of care first-line treatments for advanced, radioiodine-refractory (RAIR) differentiated thyroid cancer (DTC). However, most patients eventually become treatment-resistant and require additional therapies. The phase 3 COSMIC-311 study investigated cabozantinib in patients with RAIR DTC who progressed on lenvatinib, sorafenib, or both and showed that cabozantinib provided substantial clinical benefit. Presented in this study is an analysis of COSMIC-311 based on prior therapy and histology. Methods: Patients were randomized 2:1 (stratification: prior lenvatinib [yes/no]; age [≤65, >65 years]) to oral cabozantinib (60 mg tablet/day) or matched placebo. Eligible patients received 1–2 prior vascular endothelial growth factor receptor-targeting tyrosine kinase inhibitors for DTC (lenvatinib or sorafenib required), had a confirmed DTC diagnosis, and were refractory to or ineligible for radioiodine therapy. For this analysis, progression-free survival (PFS) and objective response rate (ORR) per Response Evaluation Criteria in Solid Tumors version 1.1 by a blinded independent radiology committee were evaluated by prior therapy (lenvatinib only, sorafenib only, both) and histology (papillary, follicular, oncocytic, poorly differentiated). Results: Two hundred fifty-eight patients were randomized (170 cabozantinib/88 placebo) who previously received sorafenib only (n = 96), lenvatinib only (n = 102), or both (n = 60). The median follow-up was 10.1 months. The median PFS (months) with cabozantinib/placebo was 16.6/3.2 (sorafenib only: hazard ratio [HR] 0.13 [95% confidence interval, CI, 0.06–0.26]), 5.8/1.9 (lenvatinib only: HR 0.28 [95% CI 0.16–0.48]), and 7.6/1.9 (both: HR 0.27 [95% CI 0.13–0.54]). The ORR with cabozantinib/placebo was 21%/0% (sorafenib only), 4%/0% (lenvatinib only), and 8%/0% (both). Disease histology consisted of 150 papillary and 113 follicular, including 43 oncocytic and 36 poorly differentiated. The median PFS (months) with cabozantinib/placebo was 9.2/1.9 (papillary: HR 0.27 [95% CI 0.17–0.43]), 11.2/2.5 (follicular: HR 0.18 [95% CI 0.10–0.31]), 11.2/2.5 (oncocytic: HR 0.06 [95% CI 0.02–0.21]), and 7.4/1.8 (poorly differentiated: HR 0.18 [95% CI 0.08–0.43]). The ORR with cabozantinib/placebo was 15%/0% (papillary), 8%/0% (follicular), 11%/0% (oncocytic), and 9%/0% (poorly differentiated). Safety outcomes evaluated were consistent with those previously observed for the overall population. Conclusions: Results indicate that cabozantinib benefits patients with RAIR DTC, regardless of prior lenvatinib or sorafenib treatments or histology. Clinical Trial Registration Number: NCT03690388.

Background: Lenvatinib and sorafenib are standard of care first-line treatments for advanced, radioiodinerefractory (RAIR) differentiated thyroid cancer (DTC).However, most patients eventually become treatmentresistant and require additional therapies.The phase 3 COSMIC-311 study investigated cabozantinib in patients with RAIR DTC who progressed on lenvatinib, sorafenib, or both and showed that cabozantinib provided substantial clinical benefit.Presented in this study is an analysis of COSMIC-311 based on prior therapy and histology.Methods: Patients were randomized 2:1 (stratification: prior lenvatinib [yes/no]; age [£65, >65 years]) to oral cabozantinib (60 mg tablet/day) or matched placebo.Eligible patients received 1-2 prior vascular endothelial growth factor receptor-targeting tyrosine kinase inhibitors for DTC (lenvatinib or sorafenib required), had a confirmed DTC diagnosis, and were refractory to or ineligible for radioiodine therapy.For this analysis, progression-free survival (PFS) and objective response rate (ORR) per Response Evaluation Criteria in Solid Tumors version 1.1 by a blinded independent radiology committee were evaluated by prior therapy (lenvatinib only, sorafenib only, both) and histology (papillary, follicular, oncocytic, poorly differentiated).

Introduction
3][4][5] While surgery and radioiodine therapy are effective treatments for most patients with DTC, 3 up to 15% of patients develop radioiodinerefractory (RAIR) disease and require systemic treatment. 2The standard of care first-line treatment is a single-agent multitarget tyrosine kinase inhibitor (TKI), such as sorafenib or lenvatinib. 6These treatments are initially effective in controlling the disease in most patients; however, a majority will eventually experience treatment resistance and disease progression. 7,8he clinical benefit of later-line treatments in RAIR DTC had not been established before the phase 3 COSMIC-311 study (NCT03690388).0][11] The study met the primary endpoint of progression-free survival (PFS) at interim analysis (median follow-up [range]: 6.2 [0.1-17.7]months); cabozantinib significantly prolonged PFS compared with placebo (median not reached vs. 1.9 months; hazard ratio [HR] 0.22 [96% confidence interval, CI, 0.13-0.36];p < 0.0001). 10afety was manageable and consistent with the known safety profile of cabozantinib. 10,12,13Based on these findings, cabozantinib was approved in the United States to treat locally advanced or metastatic DTC that progressed after prior VEGFRtargeted therapy in adult and pediatric patients aged ‡12 years. 14abozantinib has also received approval in the European Union to treat adult patients with locally advanced or metastatic DTC that is refractory to or not eligible for radioiodine and has progressed during or after prior systemic therapy. 15OSMIC-311 demonstrated the benefit of cabozantinib as second-or third-line treatment after sorafenib, lenvatinib, or both. 10,11COSMIC-311 was the first phase 3 study in RAIR DTC to include patients with prior lenvatinib treatment.A more detailed analysis of the benefit of cabozantinib based on prior treatment would interest physicians deciding on subsequent treatment for patients who progressed on first-or second-line VEGFR-targeted treatments.
3][4][5] For decisions on selecting the second and further lines of therapy, it is important to understand the sensitivity of specific histologies to the therapy of choice.Additional analysis of clinically relevant subpopulations of patients with various DTC histologies, such as follicular and papillary along with aggressive oncocytic and poorly differentiated subtypes, will provide valuable insights on outcomes in different DTC histologies.An understanding of cabozantinib efficacy and safety in different histologies may aid physicians determining the appropriate course of action.
Presented in this report are the efficacy and safety outcomes of COSMIC-311 at the extended data cut based on sorafenib and lenvatinib treatment history (sorafenib only, lenvatinib only, both sorafenib and lenvatinib) along with DTC histologies (papillary and follicular) and subtypes (oncocytic and poorly differentiated).

Study design and participants
COSMIC-311 was a multicenter, randomized, doubleblind, phase 3 trial (NCT03690388).Details of the study design have been previously published. 10Eligible patients were ‡16 years of age with a confirmed diagnosis of DTC (papillary or follicular and histological subtypes), measurable disease according to Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1, Eastern Cooperative Oncology Group performance status of 0 or 1, and adequate organ and marrow function.Patients were refractory or deemed ineligible for treatment with iodine-131.Patients must have received prior treatment with lenvatinib and/or sorafenib (receipt of up to two prior VEGFR-TKIs allowed) and experienced radiographic progression per RECIST version 1.1 during or following treatment with a VEGFR-TKI.Patients must have been receiving thyroxine suppression therapy with serum thyrotropin levels <0.50 mIU/L.Key exclusion criteria included uncontrolled significant illness, prior treatment with a selective BRAF inhibitor, and prior treatment with more than one immune checkpoint inhibitor or more than one systemic chemotherapy.
The study protocol was approved by the institutional review board or ethics committee at each center, and the trial was conducted in accordance with Good Clinical Practice, including the International Conference on Harmonisation and the Declaration of Helsinki.All patients provided written informed consent.(yes or no) and age (£65 or >65 years).Patients received a 60 mg tablet of cabozantinib or a matched placebo, which was self-administered orally once per day.All patients received best supportive care, and adverse events (AEs) were managed with dose modification (dose interruptions or reductions) and supportive care.Dose interruptions were allowed for up to 8 weeks or longer with sponsor approval.Dose reductions were from 60 to 40 mg daily and then to 20 mg daily.Patients were treated until disease progression by RECIST version 1.1 or unacceptable toxicity.Treatment beyond disease progression was allowed if patients experienced clinical benefit in the opinion of the investigator.Importantly, patients in the placebo arm who experienced disease progression per blinded independent radiology committee (BIRC) were permitted to crossover and receive open-label cabozantinib.

Endpoints and assessments
Endpoints presented in this report are PFS (time from randomization to the earlier of either radiographic progression by RECIST version 1.1 per BIRC or death from any cause), objective response rate (ORR; proportion of patients with confirmed complete or partial response by RECIST version 1.1 per BIRC), disease stabilization rate (DSR; proportion of patients achieving a confirmed complete or partial response or stable disease with a duration of at least 16 weeks), and safety and tolerability.While overall survival (OS) was included in COSMIC-311 as an additional endpoint, the study was not powered for OS.Since the primary endpoint was met, and since the study underwent protocolapproved crossover from placebo to cabozantinib, it has been determined that no meaningful results can be obtained by updated OS analyses comparing the cabozantinib arm with the placebo arm.
Efficacy endpoints are presented for prior therapy and histology subgroups and safety endpoints reported for prior therapy subgroups.Prior therapy subgroups were prior sorafenib and not prior lenvatinib (sorafenib only), prior lenvatinib and not prior sorafenib (lenvatinib only), and prior sorafenib and lenvatinib (sorafenib/lenvatinib).Histology subgroups were papillary, follicular, oncocytic (described as Hu ¨rthle cell in the primary analysis of the study), and poorly differentiated.Histology determination was performed by investigator.Efficacy and safety endpoints for the overall population have been previously reported. 10,11umor response and progression were determined by magnetic resonance imaging or computed tomography at baseline, every 8 weeks for 12 months after randomization, and every 12 weeks thereafter.Safety was assessed every 2 weeks until week 9, every 4 weeks thereafter, and 30 days after treatment discontinuation.AEs were evaluated by investigators with severity graded based on the National Cancer Institute Common Terminology Criteria for Adverse Events, version 5.0.COSMIC-311 PRIOR THERAPY AND HISTOLOGY ANALYSIS

Statistical analyses
Statistical analysis details have been previously published. 10Efficacy endpoints in this current analysis were assessed in all randomized patients (intent-to-treat population).Safety and tolerability were evaluated in all patients who received at least one dose of the trial regimen (safety population).All the subgroup analyses included are descriptive.The Kaplan-Meier method estimated medians and associated CIs for time-to-event endpoints, and HR was estimated using a Cox proportional hazards model (unstratified).Assessment of efficacy and safety endpoints based on prior sorafenib or lenvatinib treatment and DTC histology was post hoc, except for PFS and ORR, which were prespecified for the prior therapy and papillary/follicular histology subgroups.Categorical and continuous data were summarized with descriptive statistics.Safety and efficacy were monitored by an independent data monitoring committee through the primary analysis.All analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC).

Patients
At a median follow-up of 10.1 months (range 0.2-23.4months), 258 patients had been randomized to cabozantinib (n = 170) or placebo (n = 88).Patient disposition data have been previously published. 11There were 96, 102, and 60 patients who received prior sorafenib only, prior lenvatinib only, and prior lenvatinib and sorafenib, respectively.Of these patients, 150 had papillary histology and 113 had follicular histology; 5 of these patients had both papillary and follicular histology and were included in both histology subgroups for analyses.The histological variants of oncocytic and poorly differentiated represented 17% (n = 43) and 14% (n = 36), respectively.One oncocytic patient had both papillary and follicular histologies.Baseline demographics and clinical characteristics were generally balanced between the treatment groups for the respective prior sorafenib/ lenvatinib therapy and histology subgroups (Supplementary Tables S1 and S2).

A
For the prior therapy subgroups, TEAEs occurred in ‡96% of patients receiving cabozantinib and ‡81% of patients receiving placebo (Table 4).Grade 3/4 TEAEs occurred in 63%, 57%, and 69% of patients receiving cabozantinib for the prior sorafenib only, lenvatinib only, and sorafenib/ lenvatinib subgroups, respectively, and 21%, 41%, and 19% of patients receiving placebo (Table 4).The most common Grade 3/4 TEAEs for the prior sorafenib only, lenvatinib only, and sorafenib/lenvatinib subgroups included hypertension (11%, 12%, and 13% with cabozantinib vs. 0%, 6%, and 0% with placebo), palmar-plantar erythrodysesthesia (10% [all] vs. 0% [all]), fatigue (3%, 12%, and 13% vs. 0% A B FIG. 5. Best change from baseline in sum of target lesions per BIRC with cabozantinib for patients with oncocytic and poorly differentiated DTC.Data are included for patients before or on the date of first progression and before or on the date of first systemic non-radiation therapy.Only ITT patients with at least one baseline and one post-baseline assessment are shown.If the percent change was in the interval 0 to 5 (including 0), then the percent change is shown as 5%; if the percent change is in the interval -5 to 0, then it is shown as -5% for visual clarity.The bars that are shown as 5% and -5% for visual clarity are indicated with an asterisk, and the bars that represent 0 are indicated with a 0.  [all]), and hypocalcemia (10%, 4%, and 10% vs. 3%, 3%, and 0%).There were no Grade 5 treatment-related AEs.Grade 5 TEAEs not related to DTC that occurred £30 days after the last dose of the study treatment were reported in four patients receiving cabozantinib and one patient receiving placebo. 11

Discussion
A previous limited subgroup analysis of the COSMIC-311 data showed that those receiving cabozantinib had a greater PFS benefit versus placebo irrespective of prior sorafenib or lenvatinib treatment or papillary or follicular histology. 11s these patients represent a population with poor prognosis and limited options, a more extensive evaluation of the efficacy and safety was warranted.
In this more detailed subgroup analysis of the COSMIC-311 trial, the PFS benefit associated with cabozantinib was maintained irrespective of treatment history for those patients who received prior sorafenib only, prior lenvatinib only, or both prior sorafenib/lenvatinib. Similarly, the PFS benefit associated with cabozantinib was maintained in all disease histology subgroups, including the more aggressive variants of oncocytic and poorly differentiated.The ORR and DSR were also greater with cabozantinib versus placebo in these subgroups, consistent with the overall population. 10,11The results of this study suggest that patients with a history of TKI treatments, regardless of histology, would benefit from cabozantinib treatment.
The safety profile of cabozantinib observed in the evaluated subgroups was comparable among the subgroups when evaluating AE frequency, as well as dose modifications and discontinuations for AE management.The safety profile in all evaluated subgroups was consistent with that reported for the overall population and in previous studies. 12,13here have been limited studies that evaluated a multitarget TKI in patients with RAIR DTC previously treated with sorafenib and/or lenvatinib or with histologies associated with a poor prognosis.7][18][19] COSMIC-311 is the first randomized phase 3 study that evaluated cabozantinib, a multitarget TKI, in these patient subgroups and further established cabozantinib as a subsequent therapy in the treatment of patients who progressed after sorafenib and/or lenvatinib treatment, showing clinical benefit in different DTC histologies.
In this analysis, cabozantinib showed clinical benefit regardless of prior sorafenib or lenvatinib treatment.The PFS and ORR with cabozantinib were numerically higher in patients who received prior sorafenib (without lenvatinib) compared with patients who received prior lenvatinib.Although differences in study design limit cross-trial comparisons, the ORR and PFS were lower with sorafenib in the DECISION study (ORR 12%, PFS 10.8 months) compared with lenvatinib in the SELECT study (ORR 65%, PFS 18.3 months). 7,8Sorafenib and lenvatinib have different kinase target profiles, which might contribute to the apparent differences in efficacy. 20,21Differential response to prior therapy may have contributed to the range of clinical benefit with cabozantinib across these prior therapy subgroups.Regardless, COSMIC-311 is the only study, to the best of our knowledge, to demonstrate the efficacy with a TKI after treatment with lenvatinib.Patients in all three prior therapy subgroups experienced tumor reduction and disease control, indicating that cabozantinib is a suitable option after progression on sorafenib, lenvatinib, or both.
It is notable that patients with oncocytic carcinoma, an aggressive subtype of DTC with a higher incidence of distant metastases and rapid progression, 4,22 experienced a benefit in PFS with cabozantinib compared with placebo, with a majority of these patients demonstrating a reduction in target lesions.Similarly, patients with poorly differentiated carcinoma had longer PFS with cabozantinib compared with placebo, with a majority achieving a reduction in target lesions.Poorly differentiated carcinoma is the leading cause of disease-related morbidity and death in non-anaplastic follicular cell-derived thyroid cancer and therefore urgently requires new treatments. 5A limitation of this subgroup analysis was the small sample size for certain subgroups.

Conclusions
The results of this subgroup analysis show that cabozantinib extends PFS compared with placebo for patients with previously treated, RAIR DTC, regardless of their treatment history with sorafenib or lenvatinib or the histology of their disease.The improved PFS and response findings in oncocytic and poorly differentiated subtypes are encouraging.The acceptable safety profile of cabozantinib, along with its demonstrated efficacy, supports its use as a subsequent therapy for patients with RAIR DTC irrespective of treatment history or disease histology.

FIG. 2 .
FIG. 2.Best change from baseline in sum of target lesions per BIRC with cabozantinib based on prior sorafenib/lenvatinib treatment.Data are included for patients before or on the date of first progression and before or on the date of first systemic non-radiation therapy.Only ITT patients with at least one baseline and one post-baseline assessment are shown.If the percent change was in the interval 0 to 5 (including 0), then the percent change is shown as 5%; if the percent change is in the interval -5 to 0, then it is shown as -5% for visual clarity.The bars that are shown as 5% and -5% for visual clarity are indicated with an asterisk and the bars that represent 0 are indicated with a 0. ITT, intent-to-treat; SoD, sum of diameters.

FIG. 3 .
FIG.3.Kaplan-Meier estimates of PFS based on DTC histology.Disease progression was assessed with the use of Response Evaluation Criteria in Solid Tumors version 1.1 and was confirmed by blinded independent radiology committee.DTC, differentiated thyroid cancer.

BFIG. 4 .
FIG. 4.Best change from baseline in sum of target lesions per BIRC with cabozantinib based on DTC histology.Data are included for patients before or on the date of first progression and before or on the date of first systemic non-radiation therapy.Only ITT patients with at least one baseline and one post-baseline assessment are shown.If the percent change was in the interval 0 to 5 (including 0), then the percent change is shown as 5%; if the percent change is in the interval -5 to 0, then it is shown as -5% for visual clarity.The bars that are shown as 5% and -5% for visual clarity are indicated with an asterisk, and the bars which represent 0 are indicated with a 0.

Table 3 .
Treatment Exposure and Dose Modification Based on Prior Sorafenib/Lenvatinib Treatment (Intent-to-Treat Population) first dose level (40 mg) reduction due to AE, median (range)second dose level (20 mg) reduction due to AE, median (range)event; TEAE; treatment-emergent adverse event; TRAE, treatment-related adverse event.

Table 1 .
Response Outcomes Based on Prior Sorafenib/Lenvatinib Treatment Tumor response was assessed with the use of Response Evaluation Criteria in Solid Tumors version 1.1 and was confirmed by BIRC.
a Exploratory endpoint: complete or partial response or stable disease for ‡16 weeks.BIRC, blinded independent radiology committee; CI, confidence interval; NA, not applicable.

Table 2 .
Response Outcomes Based on Differentiated Thyroid Cancer Histology NATumor response was assessed with the use of Response Evaluation Criteria in Solid Tumors version 1.1 and was confirmed by BIRC. a Exploratory endpoint: complete or partial response or stable disease for ‡16 weeks.

Table 4 .
Treatment-Emergent Adverse Events Based on Prior Sorafenib/Lenvatinib Treatment (Safety Population)