Real-world practice patterns and outcomes for RAI-refractory differentiated thyroid cancer

Background The optimal timing for initiating multi-kinase inhibitors (MKIs) in patients with radioactive iodine-refractory (RAI-R) differentiated thyroid cancer (DTC) remains unclear. Thus, we evaluated the real-world practice patterns and outcomes in asymptomatic patients with progressive RAI-R DTC (≥1 lesion ≥1 cm in diameter) in the USA (US population) and outside the USA (non-US population). Methods In this prospective, non-interventional, open-label study, eligible patients were chosen by treating physicians to receive MKI therapy (cohort 1) or undergo active surveillance (cohort 2) at study entry. Cohort 2 patients were allowed to transition to MKI therapy later. The primary endpoint was time to symptomatic progression (TTSP) from study entry. Data were compared descriptively. When endpoints were inestimable, 36-month rates were calculated. Results Of the 647 patients, 478 underwent active surveillance (cohort 2) and 169 received MKI treatment (cohort 1). Patients underwent surveillance at a higher rate in the US (92.6%) vs the non-US (66.9%) populations. Half of US and non-US patients who qualified for MKI treatment had initial American Thyroid Association (ATA) low-to-intermediate-risk disease. In cohort 2, the 36-month TTSP rates from study entry were 65.6% and 66.5% in the US and non-US populations, respectively. Cohort 2 patients treated later demonstrated 36-month TTSP rates of 30.8% and 55.8% in the US and non-US populations, respectively. Conclusions Active surveillance is a viable option for asymptomatic patients with progressive RAI-R DTC. However, early intervention with MKI therapy may be more suitable for others. Further research is needed to identify patients who are optimal for active surveillance. Registration NCT02303444.


Introduction
Differentiated thyroid cancer (DTC) accounts for nearly 90% of all thyroid cancer cases (1).The standard initial treatment is surgery followed by radioactive iodine (RAI) or observation; however, approximately 5-15% of patients become refractory to RAI therapy (2).The prognosis for these patients is poor, and the 10-year survival rate for RAI-refractory (RAI-R) thyroid cancer is 10% (3).
Multi-kinase inhibitors (MKIs) are the standard of care for RAI-R DTC.While MKIs are not curative, they can improve progression-free survival (PFS) (4,5) and prolong the duration of response (6).Careful consideration must be given to selecting patients for MKI therapy, including the timing of treatment initiation.MKIs can significantly lower a patient's quality of life (4,5,7,8), and RAI-R patients can be asymptomatic, progression-free, or have slowly progressive disease for several years (8).Therefore, patients with stable or indolent disease should be considered for active surveillance.Real-world data suggest that delaying MKI therapy until rapid disease progression may result in a suboptimal clinical benefit (9).However, there is no consensus on when to initiate MKI therapy.
To elucidate the optimal timing for administering MKIs, we aimed to evaluate the real-world practice patterns and long-term outcomes of asymptomatic patients with progressive RAI-R DTC in the USA (US population).Our study investigated two groups of patients: those under active surveillance and those who received multi-kinase inhibitor (MKI) therapy at the time of study entry based on the treating physician's discretion.Furthermore, a qualitative comparison was made between the findings in the US group and those from outside of the USA (non-US population).

Study design
The RIFTOS MKI study (NCT02303444) was a noninterventional, prospective, open-label study conducted in 19 countries across 92 sites.Data collection occurred from April 2015 to July 2020.US data was kept separate from non-US data to be analyzed separately.At study entry, eligible patients chosen for MKI therapy by their treating physician were enrolled in cohort 1, while patients chosen for initial surveillance by their treating physician were enrolled in cohort 2. Cohort 2 patients were allowed to transition to MKI therapy later.
The study adhered to the guidelines and regulations of the European Medicines Authority, the US Food and Drug Administration, and local laws and regulations.In the USA, the protocol was reviewed and approved by the Western Institutional Review Board (WIRB).Participating sites were required to subject their informed consent forms for site specific approval.When required, 77 sites used their local IRB.All patients provided written informed consent before enrollment.

Patient selection
Eligible patients had asymptomatic RAI-R DTC and showed radiological progression, with at least one lesion ≥1 cm.RAI-R status was defined based on previously reported criteria (10).Patients who were previously treated with MKIs, enrolled in clinical trials, or in hospice care were excluded.A minimum life expectancy of six months was required.Survival information was collected even if patients joined an interventional clinical trial during the observation period.

Endpoints and assessments
The primary endpoint was the time to symptomatic progression (TTSP) from study entry.Symptomatic progression encompassed metastasis-related bone symptoms, respiratory symptoms, neurologic events, bleeding, discomfort or pain, declining general health, or reduced mobility or death from any cause.Secondary endpoints were PFS and overall survival (OS) from study entry of the overall population (cohorts 1 and 2 combined) and cohort 2 alone.TTSP, PFS, and OS were also assessed according to cohort 2 subgroups: age, American Thyroid Association (ATA) risk stratification after surgery (herein referenced as initial ATA risk), median sum of the longest diameter of target lesions prior RAI dose, and MKI treatment post enrollment.TTSP was met when patients developed symptoms attributed to the site of existing metastatic lesions.This required confirmation by radiological examination of Trial profile.(A) Excluded from the full analysis set based on final data checks.24 patients were excluded during data review because they were found to either meet exclusion criteria or did not meet the specified inclusion criteria.(B) Due to the limited enrollment of 13 patients in US cohort 1, data from this cohort could not be analyzed independently.(C) Among active surveillance patients, 41 out of 162 US patients (25%) and 124 out of 316 non-US patients (39%) later received MKI treatment.
European Thyroid Journal (2024) 13 e230039 https://doi.org/10.1530/ETJ-23-0039 the DTC lesion at the symptomatic location.Patients who remained asymptomatic until analysis were censored at their last evaluable assessment date.Those who joined an interventional trial during the study were censored at their last evaluable assessment prior to trial enrollment.Patients without any follow-up after study entry were censored on day 1.

Statistical analysis
Patient demographic, clinical, and treatment characteristics were analyzed descriptively, using counts, percentages for categorical variables, medians, and interquartile ranges (IQRs) for continuous variables.Bivariate analyses using propensity score matching were planned to balance the study cohorts; however, due to insufficient patient enrollment, the planned propensity score matching was not conducted.Consequently, no statistical comparisons between cohorts were performed.Instead, data were analyzed descriptively using Kaplan-Meier methods to estimate median PFS, OS, and TTSP along with their corresponding 95% CIs.Since the median was not reached for many parameters, the 36-month PFS, OS, and TTSP rates were calculated.
Additionally, we conducted a post hoc subgroup analysis, stratifying patients from cohort 2 based on age, initial ATA risk, sum of the longest target lesion(s) diameter, prior RAI dose, and MKI treatment status post enrollment.Data were not analyzed for US cohort 1 alone because only 13 patients were enrolled.

Patient sample
Between April 2015 and March 2020, 671 patients from 19 countries in North America, South America, Europe, Asia, and the Middle East were enrolled in the study.Twenty-four patients were excluded for not fulfilling the criteria, leaving 647 patients evaluable for the study (175 US and 472 non-US patients) (Fig. 1).At the data cutoff (July 24, 2020), 96 (54.9%)US and 277 (58.7%) non-US patients were still being observed.
In the US population, the median duration of observation from study entry was 32.7 months (range 0.2-57.4) in cohort 2 and 32.8 months (range 0.2-57.5) in the overall population (cohorts 1 and 2 combined).Non-US patients were observed for a similar duration.
In cohort 2, 25.3% US and 39.2% non-US active surveillance patients eventually received MKI therapy.Among them, 51.3% US and 45.9% non-US patients initiated therapy within 12 months of study entry (Table 2).
Because the sample size of US cohort 1 was insufficient for analysis, this manuscript focuses on data from the overall population and cohort 2. We compared US to non-US practices to investigate regional differences (11).

US overall population
Most DTC patients presented with papillary histology and distant metastases (Table 3).In the US overall population, the median sum of the longest diameter of target lesions was 20 mm, the median number of target lesions was 2, and the median number of nontarget lesions was one at study entry.According to initial ATA risk criteria, the percentage of low-tointermediate-risk patients in the study was 37.7% (66 low-to-intermediate-risk patients/175 patients in the overall population), of which 25.1% were low-risk and 12.6% were intermediate-risk patients (Table 3).A similar percentage of high-risk (38.9%) patients were enrolled, and ATA risk status was unknown in 23.4% of patients (Table 3).
Approximately 62.9% of patients progressed, and 33.1% developed new metastases during the study (Table 3).The median time from the initial DTC diagnosis to study entry was 89.1 months (IQR 34.2-147.6)(Table 3).The median time from the latest radiologic progression assessment to study entry was 0.7 months (IQR 0.2-1.9)(Table 3).

Comparison of US to non-US overall populations
Both US and non-US physicians enrolled a similar ratio of patients previously characterized after initial ATA risk as low-to-intermediate-risk and high-risk patients (non-US: 45.1% low-to-intermediate-risk patients; 49.2% high-risk patients); however, fewer low-risk (14.4%) and more high-risk patients (49.2%) were enrolled (Table 3) in the non-US population compared to the US population.ATA risk status was unknown in 5.7% of non-US patients.Additionally, non-US patients were enrolled in the study earlier than US patients, with a median time from the initial DTC diagnosis to study entry of 80.8 months (IQR 41.5-145.8)(Table 3).The median time from the latest radiologic progression assessment to study entry was slightly longer in the non-US (1.0, IQR 0.3-2.3)(Table 3).

Comparison of US and non-US cohort 2 (active surveillance) to US and non-US overall populations
No notable differences in baseline characteristics were observed between US cohort 2 and the US overall population or non-US cohort 2 vs the non-US overall population (Table 3).

US overall population
In the US overall population, nearly all patients were treated with RAI prior to study entry (97.7%) (Table 4).Details regarding RAI treatment prior to study entry in the US population are described in Table 4.

Comparison of US to non-US overall populations
Several RAI practice differences were observed between the US and non-US overall populations.
Slightly fewer patients received RAI prior to study entry (92.4%) in the non-US cohort (Table 4).Among patients who received RAI treatment prior to study entry, the median cumulative RAI dose received prior to study entry was higher (250 mCi, IQR 150-430), the median average RAI dose was lower (120 mCi, IQR 100-150), and the median number of RAI treatments was the same      (2, IQR 1-4) (Table 4).Non-US patients demonstrated a shorter median time from RAI-R classification to study entry (10.5 months, IQR 1.2-35.5)(Table 3).

Comparison of US and non-US cohort 2 (active surveillance) to US and non-US overall populations
Differences in RAI administration practices were observed in US cohort 2 vs the US overall population and non-US cohort 2 vs the non-US overall population.The details of these differences are presented in Table 4.

US overall population
In the US overall population, 29 (16.6%)deaths were reported, and approximately one-quarter of the patients (n = 43) experienced symptomatic progression.The median TTSP and OS from study entry was not reached; therefore, the 36-month rate was assessed.The 36-month TTSP, OS, and PFS rates from study entry were 65.8% , 80.4%, and 22.3%, respectively (Table 5 and Fig. 2).
Median PFS from study entry was 15.2 months (Table 5).

Comparison of US to non-US overall populations
Similar efficacy outcomes were observed in US and non-US overall populations (Table 5 and Fig. 2); however, 128 deaths (26.9%) were reported in the non-US overall population, and one-third of the patients (n = 150) experienced symptomatic progression.

Comparison of US and non-US cohort 2 (active surveillance) to US and non-US overall populations
The 36-month TTSP, PFS, and OS rates were comparable in US cohort 2 and the US overall population (Table 5 and Fig. 2) as well as the non-US cohort 2 and the non-US overall population (Table 5 and Fig. 2).

Cohort 2 (active surveillance) subgroup results
A subgroup analysis of cohort 2 revealed certain groups that have a poorer prognosis under active surveillance (Tables 6, 7, Supplementary Tables 1 and 2, see section on supplementary materials given at the end of this article), including patients aged ≥65 years, non-US patients with intermediate-to-high risk DTC, and US patients with low-to-intermediate-risk DTC.Regardless of the subgroup, cohort 2 patients in all regions achieved a 36-month OS rate from study entry of ≥70%.
The following are the efficacy outcomes of each subgroup in the US and non-US populations.7 and Supplementary Fig. 3).High-risk patients achieved the highest 36-month OS rate from study entry (note, CI overlap) (Table 7 and Supplementary Fig. 3).

Comparison of US to non-US cohort 2 (active surveillance)
The outcomes of non-US cohort 2 patients differed from those of US cohort 2. The low-risk group demonstrated the highest 36-month TTSP rate from study entry and the highest 36-month OS rate from study entry (note, CI overlap) (Table 7 and Supplementary Fig. 4).

Investigator-reported sum of the target lesion diameter US cohort 2 (active surveillance)
In US cohort 2, the 36-month TTSP and PFS rates from study entry were 21.3 and 10.1 percentage points lower, respectively, in patients with a higher investigatorreported sum of the target lesion diameter (≤ 24 mm vs > 24 mm) (Supplementary Table 1).However, the investigator-reported sum of the target lesion diameter did not correlate with OS (Supplementary Table 1).

Comparison of US to non-US cohort 2 (active surveillance)
No notable differences were observed in efficacy according to the sum of the target lesion diameter (Supplementary Table 1).

Prior RAI dose
Supplementary Table 2 presents the 36-month TTSP, PFS, and OS rates from study entry for US and non-US cohort 2 patients according to their previously received RAI dose (≤250 mCi vs >250 mCi).

Comparison of US to non-US cohort 2 (active surveillance)
Similar to US cohort 2, non-US cohort 2 patients who received MKI treatment later experienced worse outcomes than non-US patients who continued active surveillance.Median PFS was 9.9 months (95% CI: 8.7 to 12.6) in non-US patients treated later compared to 17.1 months (95% CI: 11.4 to 21.4) in non-US patients who continued active surveillance (Fig. 3 and Supplementary Table 3).The 36-month TTSP rate was 55.8% (95% CI: 45.0 to 65.3) in non-US patients treated later compared to 74.5% (95% CI: 66.7 to 80.8) in non-US patients who continued active surveillance (Fig. 3 and Supplementary Table 3).
Of interest, non-US patients treated later exhibited more favorable outcomes than their US counterparts.
In the US population, patients treated later experienced faster anatomic and symptomatic progression compared to non-US-treated patients (Fig. 3 and Supplementary Table 3).This progression became evident as early as 3 months post enrollment.In contrast, non-US cohort 2 patients, irrespective of treatment status, displayed a similar progression rate for approximately 9 months post enrollment (Fig. 3).Median PFS was 5.4 months (95% CI: 3.7 to 14.1) and 9.9 months (95% CI: 8.7 to 12.6) among US and non-US patients treated later, respectively (Supplementary Table 3).The 36-month TTSP rate was 30.8% (95% CI: 13.3 to 50.2) and 55.5% (95% CI: 45.0 to 65.3) among US and non-US patients treated later, respectively (Supplementary Table 3).

Discussion
Our study is the largest global observational study to provide real-world data regarding the treatment practices and long-term outcomes in asymptomatic patients with progressive RAI-R DTC.While the primary goal of this study was to determine when to initiate MKI treatment, we could not address this objective because of the slow accrual of events, with only 13 US patients receiving MKI treatment at study entry.
Cohort 2 (active surveillance) patients achieved a remarkable 36-month OS rate of ≥75% in the US and non-US, with 74.7% US and 60.8% non-US cohort 2 patients remaining untreated for the duration of the study.However, the cohort 2 analysis demonstrated that some patient populations may have less favorable OS with active surveillance, including patients aged ≥65 years, non-US patients with intermediate-to-high risk DTC, and US patients with low-to-intermediate-risk DTC.This finding is consistent with a prior analysis of the SELECT trial that demonstrated decreased overall survival in patients >65 years who were randomized to receive placebo with the option to cross-over to lenvatinib (12).Furthermore, we did not find investigator-reported sum of target lesion diameter (≤ 24 mm vs > 24 mm) to affect survival.It is important to note, however, that the measurement of the sum of the target lesion diameter does not accurately represent a patient's overall tumor burden, as they omit factors such as tumor volume and lesion count.Additional data are needed to fully understand the factors that contribute to improved outcomes in certain subgroups regarding the treatment practices.
The initial sharp fall-off of PFS and TTSP in US cohort 2 patients who received MKI treatment later suggests this group may have benefited from immediate MKI treatment.Patients who were treated later experienced poorer outcomes, perhaps due to difference in disease aggressiveness that may have been apparent to the investigator at the time of enrollment.Interestingly, outcomes varied by region.In the US, patients treated later experienced more rapid progression compared to their non-US counterparts.It is possible that the shorter differences in the mean imaging interval to detect disease progression prior to enrollment observed in the US vs the non-US cohort 2 (Fig. 3 and Supplementary Table 3) may have resulted in the inclusion of US patients with disease which was growing at a faster pace.Furthermore, there may be variations in shared decision-making between patients and physicians across geographic regions, leading to differences in time to treatment initiation.Thus, additional research is needed to identify patients who are best suited for active surveillance.
Differences in practice patterns related to active surveillance were seen between the US and non-US populations.US population patients were selected for active surveillance nearly 40% more frequently, despite meeting identical criteria for MKI treatment based on radiological progression of at least one lesion ≥1 cm.Additional data, not captured in this study, are needed to gain insight into the factors influencing physicians' treatment decisions, such as the rate of disease progression at study entry (9), tumor mutational status (8), and tumor volume.
Half of US and non-US patients who qualified for MKI treatment were initially classified with ATA lowto-intermediate-risk disease.Furthermore, the US population included 50% more patients who were initial ATA low-risk compared to non-US.This underscores the importance of closely monitoring patients, regardless of their initial ATA risk status.It is important to remember that initial ATA categorization was developed to determine risk of recurrence.Grani et al. reported that the initial ATA risk accurately predicted the outcomes of DTC patients 12 months after initial treatment (13).Initial ATA risk status may be a relevant short-term predictor of prognosis; however, its predictive reliability may diminish over time.

Limitations
This study has limitations common to observational studies, including the lack of blinding and randomization and the heterogeneity of the patient population.Selection bias is an inherent limitation of noninterventional studies.Since the decision to prescribe MKIs is subject to physician and patient preference as well as regional treatment availability the overall population and the active surveillance cohort (cohort 2) may not be comparable at baseline.Additionally, this study homogenizes regional differences, which would affect the classification of a patient's risk status and practice patterns.
A further limitation is the insufficient collection of laboratory data to assess biochemical disease progression or adequate TSH suppression.Disease progression was assessed solely by the treating physician based on radiographic and clinical characteristics, without a centralized review board.

Conclusions
Despite regional differences, our large global observational study suggests that active surveillance is a viable option for asymptomatic patients with progressive RAI-R DTC, as evidenced by a 36-month OS rate of ≥75%.However, for a subset of US patients, initiation of systemic therapy at the beginning of the study may have been warranted.
Patients who experienced worse outcomes under active surveillance, included US and non-US patients aged ≥65 years, non-US patients with intermediate-to-high-risk DTC, and US patients with low-to-intermediate-risk DTC.
Owing to unavailable clinical information, we cannot conclude whether these populations should be treated with an MKI upfront or if regional treatment practices should be modified to achieve better outcomes.
Overall, despite regional differences, approximately 50% of US and non-US cohort 2 patients who later initiated MKI therapy did so within 12 months of study entry.
Half of US and non-US patients who qualified for MKI treatment were initially classified with ATA low-tointermediate-risk DTC, suggesting that initial ATA classification may not accurately predict long-term outcomes.
Future research should explore factors influencing treatment decisions to distinguish between those subgroups of patients most appropriate for active surveillance from those who will benefit most from early initiation of systemic therapy.

Figure 3 Kaplan-
Figure 3 Kaplan-Meier curves for cohort 2 (active surveillance), according to MKI treatment status post enrollment for time to symptomatic progression and progression-free survival from initial visit (US subjects n = 162 and non-US subjects n = 316).(A) US time to symptomatic progression.(B) US progressionfree survival.(C) Non-US time to symptomatic progression.(D) Non-US progression-free survival.

Table 1
Patient demographics and risk factors.Data are presented as median (IQR) or as n (%).
ECOG PS, Eastern Cooperative Oncology Group performance status.

Table 2 Time
to initiation of MKI therapy from initial visit for cohort 2 (active surveillance) patients who later received treatment.Data are presented as n (%). a Among active surveillance patients, 41 out of 162 US patients (25%) and 124 out of 316 non-US patients (39%) received MKI treatment; b Data regarding the time to initiation of any MKI from the initial visit were unavailable for 4 US and 6 non-US active surveillance patients who eventually received MKI treatment.MKI, multi-kinase inhibitor.

Table 3
Patient baseline clinical characteristics.Data are presented as median (IQR) or as n (%).
Multiple entries possible.Percentages may not total 100 due to rounding; b n = 127; c n = 407 ATA, American Thyroid Association; DTC, differentiated thyroid cancer; mCi, millicurie; RAI, radioactive iodine. a

Table 4
Prior and concomitant therapies.Data are presented as median (IQR) or as n (%).

Table 5
Kaplan-Meier estimates for the overall population and cohort 2. does not reflect the total number of deaths, as individuals with an unknown date of death were censored at their last known alive date.NE, not estimable; NR, not reached.wasthe lowest in the intermediate-risk group (low risk: 15.4%; intermediate risk: 11.1%; high risk: 14.7%) (Table aThe 'failed' number

Table 6
Kaplan-Meier estimates for cohort 2 (active surveillance), according to age group (years).

Table 7
Kaplan-Meier estimates for cohort 2 (active surveillance), according to initial ATA risk.