Lesion-Based Evaluation Predicts Treatment Response to Lenvatinib for Radioactive Iodine-Refractory Differentiated Thyroid Cancer: A Korean Multicenter Retrospective Study

Abstract

Background:

Lenvatinib, a tyrosine kinase inhibitor (TKI) recently approved for treating radioactive iodine-refractory differentiated thyroid cancer, has been shown to delay disease progression and provide meaningful benefit for overall survival (OS). However, there is no predictive marker for response to lenvatinib before initiating treatment. We comprehensively analyzed clinical and radiological parameters to predict response to lenvatinib using lesion-based assessments.

Methods:

Medical records were collected from 67 patients treated with lenvatinib in 11 referral hospitals across Korea from June 2015 to December 2017. Up to 96 measurable lesions, defined as per Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1, were evaluated serially until progressive disease (PD) occurred, and tumor doubling time (TDT) was calculated based on changes between historical computed tomography (CT) scans and baseline CT scans performed at treatment initiation.

Results:

Excluding patients with anaplastic thyroid cancer, no thyroidectomy, nontarget lesions only, or treatment periods of <1 month, 57 patients were analyzed, of whom 7 (12.2%) were TKI-naive. The median progression-free survival was 5.1 months (95% confidence interval [CI], 4.4–9.5), the median OS was 19.3 months (95% CI 12.4–not reached), the mean duration of response was 6.0 ± 4.4 months, and the objective response rate was 38%. In lesion-based assessments, 31 lesions (32.2%) with significant tumor shrinkage (complete remission or partial response) were significantly associated with shorter TDT (<12 months; p = 0.02). Patients with rapidly PD with a shorter initial TDT (<6 months) were more likely to respond to lenvatinib (p = 0.03). Patients exposed to lenvatinib at an average of ≥16 mg per day, or who were TKI-naive before treatment with lenvatinib, had a lower risk of progression; however, the risk reduction did not reach statistical significance (daily dosage p = 0.07, TKI exposure p = 0.09).

Conclusions:

TDT calculations at the beginning of treatment and lesion-based tumor assessment may help identify potential responders to lenvatinib therapy and predict therapeutic responses.

Introduction

Radioactive iodine (RAI) therapy is recommended for the treatment of differentiated thyroid cancer (DTC) to reduce recurrence and improve overall survival (OS) in patients with moderate or high risk of recurrence (1,2). However, 5–15% of DTC patients have RAI-refractory differentiated thyroid cancer (RR-DTC) and are resistant to RAI therapy (3 –5). Newly approved multiple tyrosine kinase inhibitors (TKIs) have been shown to achieve good responses in RR-DTC patients(6,7). Lenvatinib, a recently approved TKI for RR-DTC, can delay disease progression and improve OS in older patients (2). However, frequent adverse events of TKIs such as hand/foot skin reactions, hypertension, or severe proteinuria can complicate the use of these drugs in RR-DTC (8 –10). Moreover, there is evidence to suggest that patients of Asian ethnicity may experience a higher incidence of adverse events following treatment with TKIs (11,12).

Candidates for TKI treatment should be carefully selected based on both potential responsiveness to the treatment and susceptibility to serious adverse events. There have been many attempts to predict responses to TKIs or adverse events in various cancers, including thyroid cancer (13 –16); however, little is known about predictive markers before treatment initiation. Post hoc analysis of the SELECT trial showed that treatment-emergent hypertension was predictive of a good response to lenvatinib (17). In contrast, the presence of symptomatic disease was closely related with poor progression-free survival (PFS) and OS in Japanese patients with RR-DTC (18). An analysis of the timing of common adverse events in lenvatinib-treated patients in the SELECT trial (19) showed that the occurrence of diarrhea, together with baseline Eastern Cooperative Oncology Group (ECOG) performance status and histology, was a predictive factor for OS. However, there is limited evidence of biomarkers that predict adverse events or response to lenvatinib in real-world settings.

As for the assessment of treatment response, Response Evaluation Criteria In Solid Tumors (RECIST) is traditionally used to evaluate tumor responses based on objective and mathematical criteria (20). Lymph nodes are classified separately, and their responses are assessed using different criteria from those related to nonlymph node target lesions: larger size cutoff (1.5 cm in lymph nodes vs. 1 cm in nonlymph nodes) and different axis (short-axis diameter in lymph nodes vs. long-axis diameter in nonlymph nodes). Using two different axes for lymph nodes and nonlymph nodes in the response assessment is somewhat confusing and may be altered by the arbitrary judgment of the investigators. This makes it difficult to maintain consistency in tumor assessment. In addition, unified measurements using long axes to calculate tumor growth rates can also make the assessment simpler.

Recently, the doubling time of serum thyroglobulin or tumor volume (tumor doubling time [TDT]), reflecting the tumor growth rates, has been shown to predict the responsiveness or aggressiveness of thyroid cancer (21 –23). A recent study reported that TDT, produced by an average value of two representative lesions, is a useful marker for predicting responsiveness (22). However, a mixed response often occurs during TKI treatment, for which the cause remains unclear, and lymph nodes are not included in the analysis.

In this report, we describe the results of a retrospective, real-world study that compared the lesion-based response with lenvatinib treatment in patients with RR-DTC across several organs. We also analyzed clinical and radiological parameters to identify potential biomarkers to predict responses to lenvatinib.

Materials and Methods

Patients

This is a multicenter, retrospective study involving 11 referral hospitals in Korea, conducted from June 2015 to December 2017. The study was approved by the institutional review board of the participating institutes (No. NCC2017-0162). Patient- and tumor-specific data were extracted from the medical records and computed tomography (CT) scans of 67 RR-DTC patients treated with lenvatinib. The starting dose of lenvatinib was 24 mg, and the dose was reduced if intolerable adverse events occurred.

Lesion-based assessments

RR-DTC was defined as a disease that was unresponsive to RAI therapy and showed no RAI uptake on post-therapeutic or diagnostic RAI imaging following thyrotropin hormone stimulation and an iodine-restricted diet (24).

Up to 96 measurable target lesions were evaluated serially until progressive disease (PD), according to RECIST version 1.1, was recorded (20). In brief, one or two representative lesions were determined as target lesions, and the sum of the longest diameter of nonlymphatic tissue or shortest diameter in lymphatic tissue was used in the assessment of treatment response (22). Target lesions were defined as soft-tissue metastatic lesions with the longest measurable diameter of at least 1 cm in at least one dimension (RECIST version 1.1). If a lymph node was included in the target lesion, its tumor response was assessed by short-axis diameter as well as long axis, and the results from both assessments were compared. This criterion was introduced because it is often difficult to distinguish lymph nodes from nonlymph nodes, especially from the chest, although the assessment of treatment response could be different when target lesions were classified as lymph nodes and when classified as nonlymph nodes. In addition, lymph node lesions often grow extensively in one direction; thus, the longest diameter of those lesions becomes larger, while the short axis sometimes does not change, which does not truly reflect the dynamics of tumor growth. As for bone metastasis, huge bone masses protruding into adjacent tissues were determined as target lesion according to RECIST criteria.

PFS, OS, best overall response (BOR), objective response rate (ORR, defined as complete remission [CR] or partial response [PR]), time to objective response, and duration of response (DoR) were evaluated.

To assess the growth rate of each tumor, TDT was calculated from a historical CT scan and a baseline CT scan performed before the first dose of lenvatinib at each institution. The tumor volume was calculated as described previously (22,25); tumor volume = π/6 × longest diameter × smallest diameter². After that, assuming that changes in tumor volume are exponential, a regression line, log y = log a + bx (x: years after the baseline radiologic image, y: tumor volume), was calculated by nonlinear square regression. The TDT of tumor volume was defined as (log2)/b. Each tumor response assessment was performed independently by two investigators (M.J.K, E.K.L).

Statistical methods

Numerical data are described with median, mean, standard deviation, quartiles, and percentages, as appropriate.

BOR was defined as the best PR or CR according to RECIST from the start of treatment until disease progression. PFS and OS were evaluated using Kaplan–Meier estimates with 95% confidence intervals (CIs). PFS was defined as the time elapsed between treatment initiation and tumor progression, death, or date of the last follow-up; OS was calculated from treatment initiation to death or date of the last follow-up in patients lost to follow-up.

Lesion-based assessments compared the characteristics of responsive and nonresponsive lesions. Lesions that exhibited a PR were categorized as responsive lesions; lesions that exhibited stable disease (SD) or PD were categorized as nonresponsive lesions.

For patient-based assessments, patients who experienced PD were categorized as nonresponders, and patients who did not experience progression were classified as responders.

Categorical data were analyzed using Pearson's chi-square test, Wilcoxon's rank-sum test, or Fisher's exact test; means were compared using a two-sample t-test. A p-value of ≤0.05 was considered statistically significant.

All statistical analyses were performed using SAS version 9.4 (SAS, Inc., Cary, NC).

Results

Patient demographics and clinical characteristics

The patient demographics and clinical characteristics are summarized in Table 1. The medical records and CT scans of 67 patients were assessed for eligibility. Excluding patients with anaplastic thyroid cancer (n = 1), no thyroidectomy (n = 2), an adverse event that occurred during a treatment period of <1 month (n = 2), and no target lesions (n = 5), the records of 57 patients were analyzed (Supplementary Fig. S1). Of these, 34 records (59.4%) were of female patients. The median (range) age of patients at enrollment was 67.4 (39.8–85.6) years with a mean follow-up period (±standard deviation) since initiation of lenvatinib treatment of 8.6 (7.2) months.

Table 1.

Patient Demographics and Clinical Characteristics

Variables	N = 57
Variables	n (%)
Median age at thyroid cancer diagnosis, years (range)	56.2 (23.9–73.5)
Median age at distant metastasis diagnosis, years (range)	60.4 (23.5–76.9)
Median age at the time of enrollment, years (range)	67.4 (39.8–85.6)
Time from metastasis to lenvatinib initiation, years (mean ± standard deviation)	6.3 ± 4.6
Follow-up period since lenvatinib treatment, months (mean ± standard deviation)	8.6 ± 7.2
Sex
Female	34 (59.4)
Male	23 (40.4)
Histological subtypes^a
Papillary	40 (70.2)
Follicular	16 (28.1)
Poorly differentiated	3 (5.3)
Previous RAI^b	57 (100%)
Cumulative dosage (mCi)	680.1 ± 596.6
Previous TKI^a
Sorafenib	46 (80.7)
Other TKIs	5 (8.8)
Naive	7 (12.2)
Distant metastasis^a
Lung	43 (86.0)
Bone	15 (30.0)
Liver	2 (4.0)
Brain	2 (4.0)
Adrenal gland	2 (4.0)
Kidney	1 (2.0)

Patients are included in more than one category, if applicable.

Two patients were missing post-RAI scan data (serial Nos. 12, 15).

RAI, radioactive iodine; TKI, tyrosine kinase inhibitor.

Overall, 51 patients (89.5%) had received ≥1 TKI before lenvatinib; of these, 46 (80.7%) were treated with sorafenib. Seven patients (12.2%) were TKI-naive. The reasons to stop the previous TKI treatment were disease progression (72.7%), adverse events (24.6%), and noncompliance (3.6%). Distant metastases presented predominantly in the lung (43 patients) and/or bone (15 patients), and the mean (±standard deviation) time from distant metastasis diagnosis to lenvatinib initiation was 6.3 (4.6) years.

Treatment response assessed by RECIST criteria

The median PFS was 5.1 months (95% CI, 4.4–9.5) (Fig. 1A). The median OS was 19.3 months (95% CI, 12.4–not reached) (Fig. 1B). The mean DoR was 6.0 ± 4.4 months, and the ORR was 38.0% (Fig. 1C and Supplementary Table S1).

FIG. 1.

Treatment response of lenvatinib-treated patients in Korea. (A) PFS. Kaplan–Meier estimate of the duration of PFS. Censored data are indicated by the tick marks. For PFS analysis, patients who had never experienced disease progression or had died by the time of analysis were censored at the time of their last evaluable RECIST assessment. The bold lines represent the median PFS while the other lines the 95% CI of PFS. (B) OS. Kaplan–Meier estimate of the duration of OS. Censored data are indicated by the tick marks. For OS analysis, 42 patients (57%) were censored. The bold lines represent the median OS while the other lines the 95% CI of OS. (C) Time to and durability of response to lenvatinib. (D) PFS according to response to lenvatinib (any responsive lesion vs. nonresponsive lesion only) for each target lesion. Censored data are indicated by the tick marks. (E) PFS according to the number of progressive lesions to lenvatinib (0 vs. 1 or more). Censored data are indicated by the tick marks. CI, confidence interval; OS, overall survival; PFS, progression-free survival; RECIST, Response Evaluation Criteria in Solid Tumors.

When BOR to lenvatinib treatment was analyzed by RECIST, a PR was achieved in 19 patients (38.0%) and SD in 30 patients (60.0%) of the 50 patients who had subsequent CT scans available (Supplementary Table S1). For the same target lesions, when lymph nodes were measured again by long-axis diameter (modified RECIST), the response rates were the same as those assessed using RECIST criteria (Supplementary Table S1).

Drug exposure and safety

Patients received a mean (±standard deviation) daily dosage of lenvatinib of 16.0 (4.7) mg for a median of 148 days. Dose reductions and dose interruptions were required in 35 (61.4%) and 19 (33.3%) patients, respectively. Lenvatinib dosage was reduced after a mean (± standard deviation) treatment time of 2.4 (3.1) months (median 1.6 months).

The most common adverse events leading to dose modification were general weakness (42.9%), hypertension (34.3%), oral mucositis (31.4%), proteinuria (25.7%), and thrombocytopenia (11.4%) (Supplementary Table S2). In total, 19 patients (33.3%) permanently discontinued lenvatinib treatment, of whom 9 discontinued treatment because of adverse events and 10 because of PD. Of nine patients who stopped lenvatinib due to intolerance, two patients showed PR while seven had SD. Adverse events leading to permanent treatment discontinuation were hypertension and general weakness (two cases each) and hand/foot skin reaction, proteinuria, weight loss, infection, and QT prolongation (one case each).

Efficacy analysis: lesion-based assessment

In the lesion-based assessments, 31 lesions (32.2%) with significant tumor shrinkage (CR or PR) were classified as responsive lesions and the remaining 65 lesions (SD or PD) were classified as nonresponsive lesions.

The average TDT of responsive and nonresponsive lesions was 11.5 and 14.3 months, respectively (Table 2). Responsive lesions had shorter TDT (<12 months; 61.3% in responsive lesions, 36.9% in nonresponsive lesions, p = 0.02, Pearson's chi-square test; odds ratio 0.370).

Table 2.

Comparison Between Responsive and Nonresponsive Lesions (Lesion-Based Assessment)

Variables	Responsive lesions	Nonresponsive lesions
Variables	N = 31	N = 65	p
Mean daily dosage, mg (standard deviation)	15.5 ± 5.1	15.1 ± 4.5	0.74^a
<16 mg	18 (58.1)	40 (61.5)	0.74^b
≥16 mg	13 (41.9)	25 (38.5)	0.74^b
Metastatic site^c
Lung	29 (93.5)	55 (84.6)	0.00^d
Bone	8 (25.8)	20 (30.8)
Brain	2 (6.5)	1 (1.5)
Adrenal gland	0 (0.0)	4 (6.2)
Lymph nodes (cervical, mediastinal, intraabdominal)	2 (6.5)	8 (12.3)
No. of metastatic sites
1	22 (71.0)	49 (75.4)	0.58^b
2	8 (25.8)	11 (16.9)
3	1 (3.2)	3 (4.6)
≥4	0 (0.0)	2 (3.1)
Previous TKI treatment (+)	25 (80.6)	59 (90.8)	0.16^b
Size of analyzed target lesion
≤2 cm	8 (25.8)	24 (36.9)	0.17^b
>2, ≤4 cm	12 (38.7)	29 (44.6)
>4 cm	11 (35.5)	12 (18.5)
Initial TDT	11.5 ± 12.7	14.3 ± 13.2	0.27^a
≥12 months	12 (38.7)	41 (63.1)	0.02^b
<12 months	19 (61.3)	24 (36.9)	0.02^b
RAI avidity	6/29 (20.7)	19/55 (34.5)	0.19^b
FDG uptake	13/22 (59.1)	33/44 (75.0)	0.18^b
AE
Hypertension	6 (19.4)	14 (21.5)	0.81^b
HFSR	3 (9.7)	9 (13.8)	0.56^b

Two patients were missing post-RAI scan data, and 19 patients were missing initial TDT data.

Two-sample t-test.

Pearson's chi-square test.

Patients are included in more than one category, if applicable.

Fisher's exact test.

AE, adverse event; FDG, fluorodeoxyglucose; HFSR, hand/foot skin reaction; TDT, tumor doubling time.

Metastatic sites were significantly different between responsive and nonresponsive lesions; lung and brain metastases were more frequent in responsive lesions, whereas lymph node and bone metastases were prevalent in nonresponsive lesions (p = 0.00, Fisher's exact test). However, there were only three brain metastases overall and the comparison between lung versus nonlung lesions demonstrated no difference. In contrast, the comparison between lung versus nonlung lesions demonstrated no difference.

Lesions that were responsive initially tended to maintain responsiveness (median PFS, 4.6 months in nonresponsive lesions [95% CI, 4.4–5.8], 6.5 months in responsive lesions [95% CI, 4.4–18.2], p = 0.04) by log-rank analysis (Fig. 1D). However, the number of metastasized organs, history of TKI exposure, median dose of lenvatinib (both the starting and daily dosage), fluorodeoxyglucose (FDG) uptake, RAI avidity, and adverse events (hypertension and hand/foot skin syndrome) did not correlate with responsiveness.

Efficacy analysis: patient-based assessment

For the patient-based assessments, 31 patients were categorized as responders and 19 patients as nonresponders (Table 3). The mean daily dosage was significantly higher in responders versus nonresponders (16.5 ± 4.5 mg vs. 13.1 ± 4.4 mg, p = 0.02, Fisher's exact test). A history of TKI therapy or the type and number of TKIs used before lenvatinib did not affect the response to lenvatinib treatment (data not shown).

Table 3.

Comparison Between Responders and Nonresponders (Patient-Based Assessments)

Variables	Responders	Nonresponders	p
Variables	N = 31	N = 19	p
Age (years)	55.4 ± 11.4	56.2 ± 9.8	0.80^a
Men, n (%)	20 (64.5)	11 (57.9)	0.64^b
Time from diagnosis (years)	10.6 ± 7.2	9.8 ± 5.9	0.81^c
Histology^d
Papillary	21 (67.7)	14 (73.7)	NA
Follicular	8 (25.8)	6 (31.6)
Poorly differentiated	2 (6.5)	1 (5.3)
Previous RAI dose (cumulative, mCi)	685.5 ± 701.0	580.7 ± 326.5	0.97^c
Previous TKI therapy (+)	25 (80.6)	18 (94.7)	0.16^b
Mean daily dosage, mg (standard deviation)	16.5 ± 4.5	13.1 ± 4.4	0.02^c
Metastatic site^d
Lung	29 (93.5)	18 (94.7)	NA
Bone	14 (45.2)	8 (42.1)
Brain	3 (9.7)	3 (15.8)
Adrenal gland	0 (0.0)	2 (10.5)
Lymph nodes (cervical, mediastinal, intra-abdominal)	5 (16.1)	2 (10.5)
No. of metastatic sites
1	14 (45.2)	11 (57.9)	0.14^e
2	14 (45.2)	4 (21.1)
3	3 (9.7)	2 (10.5)
≥4	0 (0.0)	2 (10.5)
Sum of two representative target lesions
≤2 cm	1 (3.2)	0 (0.0)	0.68^e
>2, ≤4 cm	5 (16.1)	4 (21.1)
>4 cm	25 (80.6)	15 (78.9)
RAI avidity (target 1)	8 (25.8)	6 (31.6)	0.66^b
RAI avidity (target 2)	6 (19.4)	6 (31.6)	0.50^b
RAI avidity (any)	8 (25.8)	6 (31.6)	0.66^b
FDG uptake (target 1)	17 (54.8)	12 (63.2)	0.56^e
FDG uptake (target 2)	14 (45.2)	13 (68.4)	0.11^b
FDG uptake (any)	17 (54.8)	14 (73.7)	0.18^e
Initial TDT (months)	12.2 ± 11.1	12.0 ± 13.0	0.44^c
Initial TDT <6 months (any lesion)	16 (84.2)	8 (50.0)	0.03^e

Two patients were missing post-RAI scan data, and 19 patients were missing initial TDT data.

Wilcoxon's rank-sum test.

Fisher's exact test.

Two-sample t-test.

Patients are included in more than one category, if applicable.

Pearson's chi-square test.

Consistent with the lesion-based assessments, a shorter TDT (<6 months) correlated with overall responsiveness to lenvatinib treatment (patients with any lesion TDT <6 months, 16 [84.2%] responders vs. 8 [50.0%] nonresponders, p = 0.03) (Table 3). However, the presence of responsive lesions did not predict a longer PFS (data not shown). In contrast, the presence of a remarkably large lesion predicted shorter PFS with a mean PFS of 5.8 months (95% CI 4.6–not reached) in patients with SD and 4.3 months (95% CI 3.0–9.5) in patients with any lesions growing remarkably (hazard ratio [HR] 2.38 [95% CI, 1.05–5.41]; p = 0.03) (Fig. 1E).

Age, sex, histologic subtype, previous RAI exposure, RAI dose, number of organs with metastases or the type of the involved organ with distant metastases, the sum of the size of tumor target lesions, RAI avidity, and FDG uptake did not differ between the groups. The type and number of previous TKI drugs before lenvatinib did not significantly alter the efficacy of lenvatinib therapy. The best changes in the response of tumors according to TDT and the type and number of previous TKI treatment(s) are described in Supplementary Figure S2.

Survival analysis

Patients exposed to lenvatinib at an average dose of 16 mg or more per day, or patients who were TKI-naive before receiving lenvatinib, had a lower risk of progression compared with patients receiving a mean dose of <16 mg or patients who had previously received a TKI, respectively, but the risk reduction did not reach statistical significance (daily dosage, HR 0.38 [95% CI, 0.13–1.11; p = 0.07]; TKI exposure, HR 4.88 [95% CI, 0.66–36.18; p = 0.09]). Patients with only nonresponse lesions showed poor PFS (median PFS, 4.6 months [95% CI 4.4–5.8] in patients with nonresponsive lesions only vs. 6.5 months [95% CI 4.4–18.2] in patients with any responsive lesions) and a higher risk of progression (HR 1.83 [95% CI, 1.01–3.32; p = 0.04]) (Fig. 1D).

Discussion

This real-world retrospective study conducted in Korea sought to identify clinical and radiological parameters for the assessment of lesion-based changes that can predict the response to lenvatinib in RR-DTC. Our results show that rapidly PD with a shorter initial TDT (<12 months in lesion-based assessment and <6 months in patient-based assessment) was more likely to respond to lenvatinib. These results add further support to the prognostic value of TDT in determining clinical outcomes and treatment response to lenvatinib in patients with metastatic thyroid cancer as demonstrated by Sabra et al. (22). Specifically, our results indicate that the early selection and treatment of patients with rapidly PD or shorter TDT could improve their treatment response to lenvatinib. Conversely, appropriate patient selection can reduce the unnecessary exposure of patients with slow TDT to a toxic and expensive drug. After the initiation of lenvatinib, patients with any lesion growing markedly, despite lenvatinib treatment, showed shorter PFS (19,26).

In our study, the median PFS was 5.1 months, the median OS was 19.3 months, the DoR was 6.0 ± 4.4 months, and the ORR was 38.0%. Among patients with evaluable CT scans, the best response to lenvatinib treatment according to RECIST version 1.1 was a PR in 19 patients (38.0%) and SD in 30 patients (60.0%). These PFS and ORR results were somewhat inferior to those observed in the previous phase 2 and phase 3 studies (19,26). In a single-arm, open-label, global phase 2 study (NCT00784303) (26) in patients with RR-DTC, the PFS was 12.6 months (95% CI, 9.9–16.1), the median DoR was 12.7 months, and the ORR was 50% (95% CI, 37–63%) after ≥14 months of follow-up; but the median OS has not been reported. Similarly, in the phase 3 SELECT trial (NCT01321554) (19), PFS was 18.3 months (99% CI, 0.14–0.31) in the lenvatinib group and ORR was 64.8% after a median of 17.1 months of follow-up. These differences may be attributed to variations in patient characteristics outside of the strict eligibility criteria for a clinical trial.

Other real-world studies of lenvatinib have yielded variable results. In France, a retrospective analysis in 88 patients treated with lenvatinib showed a median PFS of 10 months (27), while in Italy, 6- and 12-month PFS rates of 63.6% and 54.6%, respectively, were found in a smaller cohort of 12 patients (28). Retrospective real-world studies on lenvatinib have also been reported in Japan (18,29) and the United States (10), again with variable results. A Japanese study by Sugino et al. found that the presence of a symptom was the only factor associated with worse PFS and OS rates (18). Another Japanese study, by Suzuki et al., examined PFS according to lenvatinib dose and total tumor diameter at baseline, and reported that a maximum tumor diameter at baseline of ≥30.8 mm was associated with significantly worse PFS and OS rates; however, a total lenvatinib dose of ≥14 mg within 12 weeks was a poor prognostic factor for PFS but not OS (29). In a retrospective study conducted in the United States, 50% of the 25 patients with RR-DTC who had received lenvatinib achieved an RECIST PR; median PFS and OS had not been reached at the time of publication (10).

It is also possible that the availability of lenvatinib in Korea may have affected the outcomes in the present study. In the period following the launch of lenvatinib when these study data were captured, sorafenib was the only agent reimbursed by the Korean government, meaning that the patients in this study may have received lenvatinib at a later stage than in other studies. In our study, almost 90% of patients had received prior treatment (80% with sorafenib), compared with only 67–68% in the French and Italian studies (27,28), suggesting that patients in our analysis showed a response to lenvatinib when administered as a second-line treatment. In addition, while treatment was insufficient to achieve PR, a relatively large number of patients maintained SD compared with other studies.

Previous studies have suggested that age, metastatic sites, presence of specific adverse events, symptomatic disease, and rate of tumor progression are determinants of response to lenvatinib (2,4,10,18,28,29). In our analysis, lung lesions showed better responsiveness to lenvatinib than nonlung lesions. Moreover, data from real-life practice in France showed that the response to treatment were not different between 20 and 24 mg lenvatinib (27). However, we found that the mean daily dosage was higher in responders than in nonresponders (16.5 mg in responders and 13.1 mg in nonresponders), but it was lower than the dose administered in the French study. These results suggest that maintaining an adequate lenvatinib dose is important to elicit a good response.

The rate of adverse events found with lenvatinib in our study was consistent with previously reported clinical (6,26) and real-world studies (10,18,27 –29). Dose reductions and dose interruptions were required in 35 (61.4%) and 19 (33.3%) patients, respectively, while 19 (33.3%) patients permanently discontinued treatment. A relatively higher rate of adverse events with lenvatinib, such as proteinuria and thrombocytopenia, has been reported in patients of Asian ethnicity (12), which may explain, in part, the high rate of those adverse events in the present study. The incidence of hypertension leading to dose modification or discontinuation was also substantial. However, this can be managed by monitoring and proactive management of hypertension (10).

When modifying lymph node measurements in RECIST, the tumor response using the modified version of RECIST was found to be consistent with the original RECIST version. This suggests that measurements based on criteria for nonlymph node target lesions can also be applied to lymph nodes. If the intrathoracic mass is difficult to distinguish from the lymph nodes on CT scans, it does not seem of relevance whether it represents lymph node(s) or another tissue for assessing treatment response.

This study has several limitations. As a retrospective study, it was subject to associated biases, including selection bias. The number of analyzed patients was not sufficient to demonstrate significant differences for PFS or OS during subgroup analysis particularly by dosage and prior exposure. The overall incidence of adverse events was not recorded; rather, the frequency and severity of adverse events were based on the assessments of the individual investigators relating to the dose modification and discontinuation rates. Despite these limitations, this study might provide several novel insights. First, the results derived from this study provide the actual response rate for lenvatinib in RR-DTC patients as second-line treatment in the real-world setting. Usually, the efficacy of chemotherapy as a secondary agent is less than when administered as a primary drug. However, because of the long expectation of metastatic thyroid cancer patients, many physicians and patients expect to be able to use more than one drug to control the disease. Moreover, serious or fatal adverse effects are less common with sorafenib. Thus, for patients with a risk of thromboembolic disease or airway problems, sorafenib, rather than lenvatinib, is usually recommended as a first-line treatment. It is therefore important to collect data on the efficacy of lenvatinib as second-line therapy. The second insight is the finding that rapid progress, even though it is usually associated with poor outcome, was associated with a better response to lenvatinib treatment (Supplementary Fig. S2). In addition, we could demonstrate the usefulness of volume TDT as a method to define disease progression in real-world practice. In our analysis, rapidly progressing disease is defined as tumor volume doubling within 12 months or less based on the lesion-based assessment, and 6 months or less based on the patient-based assessment.

In conclusion, TDT calculation at the beginning of treatment and lesion-based tumor assessment may help selecting patients who are likely to respond to lenvatinib therapy.

Footnotes

Acknowledgments

The authors acknowledge Emma Donadieu, MSc, and Marion Barnett, BSc, of Edanz Medical Writing for medical writing support.

Author Disclosure Statement

E.K.L., D.-J.L., D.Y.S., B.-C.A., B.H.K., M.-H.K., and Y.J.P. are advisers for Eisai Korea, Inc. S.-M.K. has received a research grant from Eisai Korea, Inc. All other authors have no conflicts of interest to declare.

Funding Information

No funding was received for this article.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

Supplementary Figure S1

Supplementary Figure S2

References

Haugen

. 2017. 2015 American Thyroid Association Management Guidelines for Adult Patients with Thyroid Nodules and Differentiated Thyroid Cancer: what is new and what has changed?. Cancer, 123:372–381.

Brose

, Worden

, Newbold

, Guo

, Hurria

. 2017. Effect of age on the efficacy and safety of lenvatinib in radioiodine-refractory differentiated thyroid cancer in the phase III SELECT trial. J Clin Oncol, 35:2692–2699.

Pacini

, Ito

, Luster

, Pitoia

, Robinson

, Wirth

. 2012. Radioactive iodine-refractory differentiated thyroid cancer: unmet needs and future directions. Expert Rev Endocrinol Metab, 7:541–554.

Berdelou

, Lamartina

, Klain

, Leboulleux

, Schlumberger

. 2018. Treatment of refractory thyroid cancer. Endocr Relat Cancer, 25:R209–R223.

Faugeras

, Pirson

, Donckier

, Michel

, Lemaire

, Vandervorst

, D'Hondt

. 2018. Refractory thyroid carcinoma: which systemic treatment to use?. Ther Adv Med Oncol, 10:1758834017752853.

Schlumberger

, Tahara

, Wirth

, Robinson

, Brose

, Elisei

, Habra

, Newbold

, Shah

, Hoff

, Gianoukakis

, Kiyota

, Taylor

, Kim

, Krzyzanowska

, Dutcus

, de las Heras

, Zhu

, Sherman

. 2015. Lenvatinib versus placebo in radioiodine-refractory thyroid cancer. N Engl J Med, 372:621–630.

Brose

, Cabanillas

, Cohen

, Wirth

, Riehl

, Yue

, Sherman

. 2016. Vemurafenib in patients with BRAF(V600E)-positive metastatic or unresectable papillary thyroid cancer refractory to radioactive iodine: a non-randomised, multicentre, open-label, phase 2 trial. Lancet Oncol, 17:1272–1282.

Gruber

, Colevas

. 2015. Differentiated thyroid cancer: focus on emerging treatments for radioactive iodine-refractory patients. Oncologist, 20:113–126.

Brose

, Frenette

, Keefe

, Stein

. 2014. Management of sorafenib-related adverse events: a clinician's perspective. Semin Oncol, 41(Suppl 2):S1–S16.

10.

Jasim

, Iniguez-Ariza

, Hilger

, Chintakuntlawar

, Ryder

, Morris

, 3rd, Bible

. 2017. Optimizing lenvatinib therapy in patients with metastatic radioactive iodine-resistant differentiated thyroid cancers. Endocr Pract, 23:1254–1261.

11.

Liu

, Fiocco

, Swen

, Guchelaar

. 2017. Assessment of ethnic differences in sunitinib outcome between Caucasian and Asian patients with metastatic renal cell carcinoma: a meta-analysis. Acta Oncol, 56:582–589.

12.

Kiyota

, Schlumberger

, Muro

, Ando

, Takahashi

, Kawai

, Wirth

, Robinson

, Sherman

, Suzuki

, Fujino

, Gupta

, Hayato

, Tahara

. 2015. Subgroup analysis of Japanese patients in a phase 3 study of lenvatinib in radioiodine-refractory differentiated thyroid cancer. Cancer Sci, 106:1714–1721.

13.

Garcia-Donas

, Beuselinck

, Inglada-Perez

, Grana

, Schoffski

, Wozniak

, Bechter

, Apellaniz-Ruiz

, Leandro-Garcia

, Esteban

, Castellano

, Gonzalez Del Alba

, Climent

, Hernando

, Arranz

, Morente

, Pisano

, Robledo

, Rodriguez-Antona

. 2016. Deep sequencing reveals microRNAs predictive of antiangiogenic drug response. JCI Insight, 1:e86051.

14.

Haddad

, Luo

, Krabbe

, Darwish

, Gayed

, Youssef

, Kapur

, Rakheja

, Lotan

, Sagalowsky

, Margulis

. 2017. Prognostic value of tissue-based biomarker signature in clear cell renal cell carcinoma. BJU Int, 119:741–747.

15.

Liu

, Wu

, Lv

, Yao

, Xiao

, Yuan

, Song

. 2013. Skin rash could predict the response to EGFR tyrosine kinase inhibitor and the prognosis for patients with non-small cell lung cancer: a systematic review and meta-analysis. PLoS One, 8:e55128.

16.

Perez-Soler

, Saltz

. 2005. Cutaneous adverse effects with HER1/EGFR-targeted agents: is there a silver lining?. J Clin Oncol, 23:5235–5246.

17.

Wirth

, Tahara

, Robinson

, Francis

, Brose

, Habra

, Newbold

, Kiyota

, Dutcus

, Mathias

, Guo

, Sherman

, Schlumberger

. 2018. Treatment-emergent hypertension and efficacy in the phase 3 Study of (E7080) lenvatinib in differentiated cancer of the thyroid (SELECT). Cancer, 124:2365–2372.

18.

Sugino

, Nagahama

, Kitagawa

, Ohkuwa

, Uruno

, Matsuzu

, Suzuki

, Masaki

, Akaishi

, Hames

, Tomoda

, Ogimi

, Ito

. 2018. Clinical factors related to the efficacy of tyrosine kinase inhibitor therapy in radioactive iodine refractory recurrent differentiated thyroid cancer patients. Endocr J, 65:299–306.

19.

Haddad

, Schlumberger

, Wirth

, Sherman

, Shah

, Robinson

, Dutcus

, Teng

, Gianoukakis

, Sherman

. 2017. Incidence and timing of common adverse events in Lenvatinib-treated patients from the SELECT trial and their association with survival outcomes. Endocrine, 56:121–128.

20.

Eisenhauer

, Therasse

, Bogaerts

, Schwartz

, Sargent

, Ford

, Dancey

, Arbuck

, Gwyther

, Mooney

, Rubinstein

, Shankar

, Dodd

, Kaplan

, Lacombe

, Verweij

. 2009. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer, 45:228–247.

21.

Pacini

, Sabra

, Tuttle

. 2011. Clinical relevance of thyroglobulin doubling time in the management of patients with differentiated thyroid cancer. Thyroid, 21:691–692.

22.

Sabra

, Sherman

, Tuttle

. 2017. Tumor volume doubling time of pulmonary metastases predicts overall survival and can guide the initiation of multikinase inhibitor therapy in patients with metastatic, follicular cell-derived thyroid carcinoma. Cancer, 123:2955–2964.

23.

Tuttle

, Fagin

, Minkowitz

, Wong

, Roman

, Patel

, Untch

, Ganly

, Shaha

, Shah

, Pace

, Li

, Bach

, Lin

, Whiting

, Ghossein

, Landa

, Sabra

, Boucai

, Fish

, Morris

LGT

. 2017. Natural history and tumor volume kinetics of papillary thyroid cancers during active surveillance. JAMA Otolaryngol Head Neck Surg, 143:1015–1020.

24.

, Lee

, Kang

H-C

, Koh

, Kim

, Na

, Nam

K-H

, Park

, Bae

, Baek

S-K

, Baek

, Lee

B-J

, Chung

K-W

, Jung

Y-S

, Cheon

, Kim

, Chung

, Rho

Y-S

. 2016. 2016 Revised Korean Thyroid Association Management guidelines for patients with thyroid nodules and thyroid cancer. Int J Thyroidol, 9:59–126.

25.

Kim

, Kim

S-M

, Lee

, Hwangbo

, Lee

, Cho

, Park

, Lee

, Park

. 2019. Tumor doubling time predicts response to sorafenib in radioactive iodine-refractory differentiated thyroid cancer. Endocr J, 66:597–604.

26.

Cabanillas

, Schlumberger

, Jarzab

, Martins

, Pacini

, Robinson

, McCaffrey

, Shah

, Bodenner

, Topliss

, Andresen

, O'Brien

, Ren

, Funahashi

, Allison

, Elisei

, Newbold

, Licitra

, Sherman

, Ball

. 2015. A phase 2 trial of lenvatinib (E7080) in advanced, progressive, radioiodine-refractory, differentiated thyroid cancer: a clinical outcomes and biomarker assessment. Cancer, 121:2749–2756.

27.

Berdelou

, Borget

, Godbert

, Nguyen

, Garcia

, Chougnet

, Ferru

, Buffet

, Chabre

, Huillard

, Leboulleux

, Schlumberger

. 2017. Lenvatinib for the treatment of radioiodine-refractory thyroid cancer in real-life practice. Thyroid, 28:72–78.

28.

Nervo

, Gallo

, Sama

, Felicetti

, Alfano

, Migliore

, Marchisio

, Berardelli

, Arvat

, Piovesan

. 2018. Lenvatinib in advanced radioiodine-refractory thyroid cancer: a snapshot of real-life clinical practice. Anticancer Res, 38:1643–1649.

29.

Suzuki

, Kiyota

, Imamura

, Goto

, Suto

, Chayahara

, Toyoda

, Ito

, Miya

, Miyauchi

. 2018. Exploratory analysis of prognostic and predictive factors of lenvatinib for radioiodine-refractory differentiated thyroid cancer. Int J Radiat Oncol Biol Phys, 100:1386–1387.