Real-world treatment outcomes and clinicopathologic and molecular determinants of response to first-line lenvatinib in patients with advanced follicular cell-derived thyroid cancer

Abstract

Background

There is a paucity of real-world data regarding lenvatinib for locally-recurrent, metastatic and RAI-refractory thyroid cancer. Here we examined the efficacy of first-line lenvatinib in a genomically-characterized cohort and identified clinicopathological/molecular correlates of drug response.

Methods

Patients with advanced follicular cell-derived thyroid cancer who underwent NGS at Princess Margaret Cancer Centre and commenced first-line lenvatinib monotherapy between 2015-2023 were included. Kaplan-Meier method, log-rank tests and univariable/multivariable proportional hazard models were employed.

Results

In total, 77 patients were included (48% female, majority papillary (52%), poorly differentiated (17%) or invasive encapsulated follicular variant papillary (16%)). Most (79%) underwent total thyroidectomy and adjuvant RAI (median cumulative dose 231 mCi). At lenvatinib initiation, median age was 62.9 years, 68% were ECOG performance status ≥2, 81% had lung metastases, 53% had bone metastases and 8% had liver metastases. Most patients started with ≤14 mg of lenvatinib daily. Median time to treatment discontinuation was 33 months. Older age, ECOG ≥ 2, liver metastases and TP53 mutation(s) were associated with shorter time to treatment discontinuation; ECOG ≥ 2 and TP53 mutation(s) remained significant on multivariable analysis.

Conclusion

Our findings reinforce the clinical efficacy of lenvatinib in advanced thyroid cancer patients with heterogenous clinicopathologic/molecular features and highlight variables for future treatment stratification.

Introduction

More than 585,000 new cases of thyroid cancer are diagnosed worldwide annually, with over 43,000 associated deaths [1, 2]. Papillary, invasive encapsulated follicular variant papillary, follicular and oncocytic thyroid carcinomas are most prevalent, accounting for more than 90% of all thyroid cancer cases [3, 4]. The remainder of cases include poorly differentiated and differentiated high-grade follicular-derived carcinomas, as well as anaplastic follicular cell-derived thyroid carcinomas (ATC) [4, 5]. While most thyroid cancer patients are cured with surgery, with or without adjuvant radioactive iodine (RAI) and thyroid stimulating hormone (TSH) suppression, a subset of patients develop recurrent or metastatic (herein termed ‘advanced’) disease that becomes refractory to standard treatments [6, 7].

Management options for patients with advanced follicular cell-derived thyroid cancer are limited. These patients ultimately require systemic treatment. Although chemotherapeutic agents such as etoposide, cisplatin, paclitaxel or doxorubicin may be used, their efficacy is generally limited [8]. As such, systemic therapies for advanced thyroid cancers have been evolving over the last decade, guided by an enhanced understanding of the underlying molecular features of these tumours [9, 10]. Molecular alterations in the mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase (PI3K) pathways have been identified in follicular cell-derived thyroid cancer patients [11]. Most activating somatic mutations involve the BRAF and RAS genes, while alterations that signal in a Braf-like or Ras-like manner, such as RET fusions, are also observed [11]. Additionally, non-BRAF and non-RAS alterations include, but are not limited to, TP53 mutations and TERT promoter alterations [11]. At present, preferred systemic therapies generally consist of multitargeted tyrosine kinase inhibitors (TKI) [12]. Lenvatinib is one such multitargeted TKI, approved by the Food and Drug Administration (FDA) in 2015 for the treatment of patients with locally recurrent or metastatic, progressive RAI-refractory thyroid cancer [13, 14]. Approval was made on the basis of the randomized phase III SELECT (Study of Lenvatinib in Differentiated Cancer of the Thyroid) clinical trial, in which patients who received lenvatinib had improved objective response rates (ORR) as compared to the placebo arm (64.8% versus 1.5%, respectively, p < 0.001), as well as significantly longer progression free survival (PFS, 18.3 months versus 3.6 months) [13]. Ensuing subset analyses highlighted an overall survival (OS) benefit with the use of lenvatinib in patients older than 65 years and separately in patients with a lower tumour burden at baseline [15, 16].

Despite the positive findings in the SELECT trial and the approval and availability of lenvatinib for the treatment of advanced thyroid cancers, there remains limited real-world data documenting treatment patterns and clinical outcomes of patients with advanced thyroid cancers treated with lenvatinib [17–19]. This includes a paucity of outcomes data on patients with more heterogenous demographic and clinicopathologic characteristics than those included in trial cohorts [13]. The Princess Margaret Cancer Centre (PMCC) is a centre of excellence for thyroid cancer, and patients are referred from a broad geographic distribution. The present study therefore examines the use of lenvatinib monotherapy in the upfront setting for advanced follicular cell-derived thyroid cancer patients treated at the Princess Margaret Cancer Centre between 2015 and 2023, with the objective of identifying predictive clinicopathologic features for efficacy. Our patient cohort is uniquely enhanced by genomic data, as all included patients underwent next-generation sequencing (NGS) analysis.

Methods

Study design and patient cohort

This study was approved by the University Health Network (UHN) Research Ethics Board (CAPCR 23-5561). This study was a retrospective review of patients with advanced (defined as unresectable local disease, recurrent disease or distantly metastatic disease) follicular cell-derived thyroid cancer who initiated lenvatinib monotherapy as first line treatment in routine clinical practice between 2015 and 2023. All patients had NGS performed as part of either the Integrated Molecular Profiling in Advanced Cancers (IMPACT; NCT01505400) or Ontario-wide Cancer TArgeted Nucleic Acid Evaluation (OCTANE; NCT02906943) clinical trials, or through the Access to Genetic Advanced Testing (AGATE) programme—an institutional NGS initiative for patients with advanced solid tumour cancers. Informed consent for participation in these studies was previously obtained for all patients [20, 21]. Patients were typically sent for NGS at the time of referral to Medical Oncology for consideration of targeted therapy, typically once they became RAI-refractory. The protocols for IMPACT and OCTANE have been previously described [20, 21]. In brief, these trials enroled patients with advanced solid tumour cancers and performed targeted panel sequencing on archival tissue samples [20, 21]. Electronic medical records of eligible patients were reviewed to collect data, including baseline patient demographics, clinicopathologic characteristics, treatment details and follow-up information.

Statistical analysis

Demographic data were analyzed with descriptive statistics. Continuous variables were reported as medians and ranges, while categorical variables were reported as numbers and percentages. Time to treatment discontinuation (TTD) and overall survival (OS) data were analyzed using the Kaplan-Meier method and log-rank tests were used to compare groups, where applicable. TTD was defined as the time to from the date lenvatinib was initiated, to the date lenvatinib was discontinued (if applicable). OS was defined as the time from the date lenvatinib was started to the time of death or last follow-up. Univariable proportional hazards models were performed to assess associations between clinicopathological variables and TTD or OS. Multivariable analysis was subsequently conducted, if applicable, by selecting significant variables (p < 0.05) from the univariable analysis. Finally, to interrogate potential differences between the extremes of our patient cohort, we compared the demographic, clinical and pathological features of “poor” versus “excellent” responders—defined as the bottom and top quartiles of the patient cohort, in terms of length of time on lenvatinib. Patients with ongoing treatment were excluded from the poor responders cohort. Differences between these cohorts were compared using chi square, Fisher’s exact or Wilcoxon rank sum tests. All statistical analyses were performed using R Statistical Software (version 4.4.1) [22]. A p value of <0.05 was considered statistically significant.

Results

Demographics and clinical characteristics

In total, 77 patients with advanced follicular cell-derived thyroid cancer who were initiated on first-line lenvatinib monotherapy and who were consented and profiled as part of IMPACT, OCTANE or AGATE at the Princess Margaret Cancer Centre between 2015 and 2023 were eligible for inclusion in this study. Baseline clinicopathologic characteristics are shown in Table 1. The median age at initial cancer diagnosis was 53.8 years. Most patients (79%, n = 61) underwent total thyroidectomy as upfront management. The majority of patients, 52% (n = 40), had papillary thyroid carcinoma, while 21% (n = 16) had invasive encapsulated follicular variant papillary thyroid carcinoma, 6% (n = 5) had follicular thyroid carcinoma, 4% (n = 3) had oncocytic thyroid carcinoma and 17% (n = 13) had poorly differentiated thyroid carcinoma. The median tumour size on resection was 4.5 cm. At least 25% (n = 18) of patients were found to have extrathyroidal extension. Of the 14 patients who underwent upfront hemithyroidectomy, 8 patients returned for a completion thyroidectomy at a later point in their disease course. The majority of patients (99%, n = 76) had at least one course of RAI. The median cumulative RAI dose was 231 mCi (range 0–900 mCi).

Table 1.

Clinicopathologic and genomic characteristics of advanced follicular cell-derived thyroid cancer patients treated with first-line lenvatinib.

Characteristic	All patients (n = 77)
Age at initial diagnosis (years)
Median (range)	53.8 (19.3–80.7)
Sex, n (%)
Male	40 (52)
Female	37 (48)
Primary surgery, n (%)
Total thyroidectomy	61 (79)
Hemithyroidectomy	14 (18)
Other	2 (3)
Histological type, n (%)
Papillary	40 (52)
Invasive encapsulated follicular variant papillary	16 (21)
Follicular	5 (6)
Oncocytic	3 (4)
Poorly differentiated	13 (17)
Maximal tumour size (cm)
Median (range)	4.5 (0.2–10.5)
RAI, n (%)
Yes	76 (99)
No	1 (1)
Cumulative RAI dose (mCi)
Median (range)	231 (0–900)
NGS panel used, n (%)
IMPACT	3 (4)
OCTANE	67 (87)
AGATE	7 (9)
Mutation(s), n (%)
BRAF	24 (31)
HRAS	8 (10)
NRAS	24 (31)
RET	2 (3)
TERT promoter	20 (26)
TP53	3 (4)

Most patients in our cohort underwent genomic profiling with an OCTANE panel (87%, n = 67). IMPACT was used for 4% (n = 3) of patients, and AGATE was used for the remaining 9% (n = 7) of patients. BRAF and NRAS mutations were most common amongst our cohort, occurring in 31% (n = 24) patients each. TERT promoter mutations were identified in 26% of patients (n = 20). Other common mutations are shown in Table 1. Of this cohort, 12 patients did not have a detectable driver mutation among the genes tested for.

Lenvatinib treatment patterns

Clinical characteristics related to initiation of lenvatinib are shown in Table 2. The median time from initial cancer diagnosis to lenvatinib start was 6.8 years. At initiation of lenvatinib, the median patient age was 62.9 years. Most patients were Eastern Cooperative Oncology Group (ECOG) performance status 2 (62% of patients, n = 48), followed by ECOG performance status 1 (26%, n = 20). Lung metastases were present in 81% of patients (n = 62) at the time of lenvatinib initiation, while bone, distant nodal, brain and liver metastases were present in 53% (n = 41), 43% (n = 33), 13% (n = 10) and 8% (n = 6) of patients, respectively. Most patients in our cohort were initiated on a starting dose of 10 mg (27%, n = 21) or 14 mg (23%, n = 18) of lenvatinib daily [23, 24]. The recommended 24 mg starting dose was used for 22% (n = 17) of patients. Nearly half the cohort (49%, n = 38) remained on first-line lenvatinib at the time of data collection. Of those who discontinued treatment, the majority of discontinuations were due to disease progression or death (34%, n = 26). Of the 51% (n = 39) patients who discontinued lenvatinib, 26% (n = 20) commenced a second line of systemic therapy.

Table 2.

Clinical characteristics of patients with advanced follicular cell-derived thyroid cancer at lenvatinib initiation.

Characteristic	All patients (n = 77)
Time from initial diagnosis to lenvatinib start (years)
Median (range)	6.8 (0.2–39.4)
Age at lenvatinib start (years)
Median (range)	62.9 (23.7–85.9)
ECOG at lenvatinib start, n (%)
0	4 (5)
1	20 (26)
2	48 (62)
3	4 (5)
4	1 (1)
Disease burden at lenvatinib start, n (%)
Local	1 (1)
Distant	49 (64)
Both	27 (35)
Site(s) of disease at lenvatinib start, n (%)
Lung	62 (81)
Bone	41 (53)
Lymph nodes (distant)	33 (43)
Brain	10 (13)
Pleura	6 (8)
Liver	6 (8)
Lenvatinib starting dose (mg), n (%)
4	1 (1)
10	21 (27)
14	18 (23)
18	1 (1)
20	15 (19)
24	17 (22)
Unknowna	4 (5)
Reason for discontinuation of lenvatinib, n (%)
Progression or death	26 (34)
Toxicity / intolerance	8 (10)
Unknown	5 (7)
Treatment ongoing	38 (49)
Further systemic lines, n (%)
Yes	20 (26)
No	19 (25)
Total systemic lines, n (%)
1	57 (74)
2	11 (14)
3	4 (5)
4	5 (6)

^aThese patients were on a lenvatinib dosing trial that required blinding of the starting dose.

Clinical outcomes

The median TTD was 33.02 months, and the median OS was 72.9 months (Fig. 1a, b). In both Kaplan-Meier analyses, the upper 95% confidence interval remained above 50% throughout the follow-up period. Univariable analysis revealed that older age at initial thyroid cancer diagnosis, presence of a TP53 mutation, older age at lenvatinib initiation, ECOG performance status ≥2 at lenvatinib initiation and the presence of liver metastases at lenvatinib initiation were each associated with a shorter TTD (Fig. 1c, d). On multivariable analysis of significant variables, the presence of a TP53 mutation was associated with a significantly higher hazard of treatment discontinuation (hazard ratio (HR) 5.238, confidence interval (CI) 95% 1.369–20.047, p = 0.016), as was ECOG performance status ≥2 at lenvatinib start (HR 2.424, 95% CI 1.116–5.263, p = 0.025). There were no significant findings on univariable analysis for OS.

Fig. 1. Outcomes of first-line lenvatinib monotherapy and association with TP53 mutational status and liver metastases.

a Time to lenvatinib treatment discontinuation of advanced follicular cell-derived thyroid cancer patients treated with first-line lenvatinib monotherapy. b Overall survival of advanced follicular cell-derived thyroid cancer patients treated with first-line lenvatinib monotherapy. c Time to treatment discontinuation of advanced follicular cell-derived thyroid cancer patients treated with first-line lenvatinib monotherapy according to TP53 mutational status (p < 0.001). d Time to treatment discontinuation of advanced follicular cell-derived thyroid cancer patients treated with first-line lenvatinib monotherapy according to the presence or absence of liver metastases (p = 0.028).

Poor versus excellent responders

Given the wide range in duration of lenvatinib treatment, we compared the clinicopathologic and genomic features of “poor” versus “excellent” responders in our cohort—defined as the bottom and top quartiles of the full patient cohort, in terms of length of time on lenvatinib (Table 3). Patients in the poor responders group were older at the time of initial thyroid cancer diagnosis, with a median age of 63.3 years, as compared to 53.8 years in the excellent responders group (p = 0.02). In both groups, papillary thyroid carcinoma was the most common histology. There were no statistically significant differences in terms of mutational frequency for common mutations between these groups, however TP53 mutations were noted to occur solely in the poor responders group, with 16% of patients having a TP53 mutation (p = 0.23). All patients with a TP53 mutation were in fact in the poor responders group. NRAS mutations were present in 42% of patients in the excellent responders group, compared to 16% in the poor responders group (p = 0.15), while TERT promotor mutations were present in 42% of patients in the poor responders group, compared to 26% in the excellent responders group (p = 0.49).

Table 3.

Clinicopathologic and genomic characteristics of advanced follicular cell-derived thyroid cancer patients treated with first-line lenvatinib, stratified as poor or excellent responders according to duration of time on lenvatinib.

Characteristic	Poor responders (n = 19)	Excellent responders (n = 19)	Significance (p value)
Age at initial diagnosis (years)
Median (range)	63.3 (32.5–80.7)	53.8 (35.9–69.2)	0.02
Sex, n (%)			0.74
Male	13 (68)	11 (58)
Female	6 (32)	8 (42)
Primary surgery, n (%)			>0.999
Total thyroidectomy	17 (89)	18 (95)
Hemithyroidectomy	2 (11)	0 (0)
Other	0 (0)	1 (5)
Histological type, n (%)			0.18
Papillary	11 (58)	13 (68)
Invasive encapsulated follicular variant papillary	5 (26)	1 (5)
Follicular	0 (0)	2 (11)
Oncocytic	1 (5)	0 (0)
Poorly differentiated	2 (11)	3 (16)
Maximal tumour size (cm)			0.03
Median (range)	5.0 (0.2–8.0)	6.4 (2.5–10.5)
RAI, n (%)			>0.999
Yes	19 (100)	19 (100)
No	0 (0)	0 (0)
Cumulative RAI dose (mCi)			0.42
Median (range)	230 (100–600)	231 (100–755)
Mutation(s), n (%)
BRAF	8 (42)	8 (42)	>0.999
HRAS	1 (5)	1 (5)	>0.999
NRAS	3 (16)	8 (42)	0.15
RET	0 (0)	1 (5)	>0.999
TERT promoter	8 (42)	5 (26)	0.49
TP53	3 (16)	0 (0)	0.23
Time from initial diagnosis to lenvatinib start (years)
Median (range)	6.3 (0.6–14.9)	8.3 (0.2–16.9)	0.32
Age at lenvatinib start (years)			0.06
Median (range)	71.4 (39.5–85.9)	62.0 (38.7–85.8)
ECOG at lenvatinib start, n (%)			0.006
1	1 (5)	8 (42)
2	15 (79)	11 (58)
3	3 (16)	0 (0)
Disease burden at lenvatinib start, n (%)			0.32
Local	0 (0)	0 (0)
Distant	13 (68)	9 (47)
Both	6 (32)	10 (53)
Site(s) of disease at lenvatinib start, n (%)
Lung	19 (100)	16 (84)	0.23
Lymph nodes (distant)	13 (68)	11 (58)	0.74
Bone	10 (53)	8 (42)	0.75
Brain	4 (21)	1 (5)	0.34
Pleura	2 (11)	1 (5)	>0.999
Liver	5 (26)	1 (5)	0.18
Lenvatinib starting dose (mg), n (%)			0.11
4	0 (0)	1 (5)
10	7 (37)	3 (16)
14	4 (21)	6 (32)
20	4 (21)	4 (21)
24	4 (21)	1 (5)
Unknown	0 (0)	4 (21)
Time to treatment discontinuation (months)
Median (range)	3.0 (1–10)	55.0 (51–70)	0.007
Reason for discontinuation of lenvatinib, n (%)			0.34
Progression or death	8 (42)	3 (16)
Toxicity/intolerance	8 (42)	0 (0)
Unknown	3 (16)	0 (0)
Treatment ongoing	0 (0)	16 (84)

With regards to clinical characteristics related to lenvatinib initiation, there was no difference in the time from initial diagnosis to lenvatinib start (median time 6.3 years for poor responders versus 8.3 years for excellent responders, p = 0.32). Poor responders were older at lenvatinib start (median age of 71.4 years, versus 62.0 years for excellent responders), though not significantly so (p = 0.06). Poor responders had a higher ECOG performance status at the time of lenvatinib initiation (p = 0.006). All patients in the poor responders group had lung metastases at lenvatinib initiation, and more patients were noted to have brain, liver and pleural metastases than in the excellent responders group (21% versus 5%, 26% versus 5%, and 11% versus 5%, respectively), though these differences were not statistically significant. Median TTD from the date of lenvatinib initiation was 3 months for poor responders, compared to 55 months for excellent responders.

Discussion

This retrospective study examined treatment patterns, clinical outcomes and correlates of response of genomically-characterized, advanced follicular cell-derived thyroid cancer patients who received first-line lenvatinib monotherapy. This is one of the largest real-world patient cohorts, and included a broad representation of patients. The study period spanned nine years, enabling sufficient evaluation of outcomes. To our knowledge, this is also the first study examining lenvatinib in a patient cohort that has undergone NGS.

Patients in our cohort were most commonly ECOG performance status 2 (62%), with the majority (81%) having lung metastases prior to initiating lenvatinib, followed by bone metastases (53%). In comparison, patients in the SELECT clinical trial who received lenvatinib were primarily ECOG 0–1 (95%), and only 39.8% of patients had bone metastases. Higher ECOG performance status in our cohort likely reflects local practice, including timing of referrals for consideration of systemic therapy, as well as oncologist experience in treating patients with poorer performance statuses. The median time to treatment discontinuation was 33 months in our cohort. This is longer than the median duration of lenvatinib treatment reported in the SELECT trial (13.8 months), despite lower starting dose in most of our patients [13]. While 24 mg daily is the initial dose of lenvatinib supported by literature and used in the SELECT study, most patients in our cohort were initiated on lower doses [23, 24]. A two-arm study previously compared the safety and efficacy of a lower starting dose (18 mg) to the standard 24 mg daily and found a lower ORR (57.3% ORR for 24 mg starting dose, versus 40.3% for 18 mg starting dose), with a similar frequency of adverse events in both arms [15]. In the present study, lower starting doses were initiated based on patient performance status, as well as institutional practice to re-evaluate patients in the short term (typically 1–2 weeks) after drug initiation and proceed with dose escalation if tolerated. The longer time to lenvatinib discontinuation in our cohort may also reflect the fact that lenvatinib was continued in our patients as long as there was clinical benefit, whereas on trial the drug would have been stopped based on Response Evaluation Criteria In Solid Tumours (RECIST) or equivalent evidence of progression [13, 17]. Additionally, given the relatively limited systemic options for advanced thyroid cancer patients, in cases of oligometastatic disease or oligoprogression, local therapies, such as radiotherapy, would be more readily considered by our centre in order to enable continuation of the same systemic line, here lenvatinib [25]. Additionally, in contrast to trial protocol, patients are typically imaged less frequently in the real world, meaning patients in a non-trial cohort may be less likely to have PFS events.

There are a small number of studies examining factors correlated with lenvatinib response and clinical outcomes in advanced thyroid cancer patients to date. One such study showed that lower baseline tumour burden was significantly associated with longer PFS, similar to our findings [26]. A post hoc analysis of SELECT by Kiyota et al., showed that the sum of diameters of target lesions is a possible as a prognostic marker of OS in patients receiving lenvatinib [16]. Our study identified older age (both at initial thyroid cancer diagnosis and at time of lenvatinib initiation) to be associated with shorter TTD. Additionally, and to our knowledge for the first time, the presence of liver metastases at lenvatinib initiation, and separately, the presence of a TP53 mutation were each associated with a significantly higher hazard of treatment discontinuation. TP53 mutations are seen across a subset of thyroid carcinomas and have been implicated as a mutation seen more frequently in more aggressive forms of follicular cell-derived thyroid carcinomas, including those with morphologic dedifferentiation, and in cases with co-occurring ATC components [9, 27]. In this study, all TP53 mutated patients were found in the poor responder cohort. Future studies of larger cohorts may further elucidate the strength of these correlations, and stratification of these patients for alternate targeted therapies, or perhaps as preferred chemotherapy candidates, may be considered. With regard to other genomic findings, similar to a prior analysis, TERT promoter alterations did not seem to interfere with response to lenvatinib in our cohort [28]. Our cohort interestingly included a number of patients with either NRAS or HRAS mutations, the frequency of which exceeded that of BRAF mutations, which are typically more common [11]. This may reflect the varied follicular cell-derived thyroid histologies included in our study, whereas poorly-differentiated has been excluded from previous genomic studies [11]. Additionally, in our centre, patients with BRAF V600E mutations with relative or absolute contraindications to lenvatinib (due to risk of fistulization) are preferentially treated with BRAF/MEK inhibitors first-line and therefore are not included in the present study, which examined first-line use of lenvatinib.

On comparison of poor versus excellent responders to lenvatinib, we did not identify significant mutational differences, as our dataset is likely underpowered for this. As mentioned, all the TP53 mutated patients were noted to be in the poor responder cohort, and all patients in the poor responder cohort had lung metastases, as well as more frequently had distant nodal, bone, brain, liver and pleural metastases (though not significantly so). Poor responding patients were older at both the time of initial cancer diagnosis and the time of lenvatinib initiation. These observations collectively suggest that a higher disease burden may be associated with reduced efficacy of lenvatinib, as has been suggested by previous analyses [16, 26, 29, 30]. Along the same lines, an interesting question not completely examined in our study is the potential impact of early versus delayed initiation of lenvatinib, as it may be extrapolated that features such as older age and/or higher disease burden portend inferior outcomes.

Potential limitations of our study include, as mentioned, our cohort size. While thyroid cancer is the most common endocrine malignancy, most patients do well, managed with surgery alone, or with adjuvant RAI alone [1, 2, 6, 7]. Advanced follicular cell-derived thyroid cancer patients therefore represent a small cohort of thyroid cancer patients; further to this, our study included only patients who underwent first-line lenvatinib monotherapy, and who were genomically characterized on an NGS platform [20, 21]. Given the limited real-world literature on treatment patterns and outcomes of advanced thyroid cancer patients treated with lenvatinib, the present study nonetheless offers valuable insights into possible correlates of treatment response, or, equally importantly, poor response, that should be validated in a larger cohort and may inform the stratification of patients for lenvatinib monotherapy in the future. Additionally, while we captured pre-lenvatinib treatment data, as well as some post-discontinuation data, there are likely other variables impacting duration on lenvatinib and TTD that are beyond our current dataset. For example, use of radiotherapy for oligometastatic disease in the advanced thyroid cancer patient population is expanding and may in some cases enable prolongation of lenvatinib treatment, which is not captured in our dataset [31]. Finally, our study is inherently limited by its retrospective nature, relying on chart review and available documentation for data collection, which may introduce biases and limits the collection of certain data, including toxicity data.

Moving forwards, we are planning to validate and expand on our findings in a multicentre study with a larger patient cohort. This will enhance the generalizability of our findings, as well as allow a more in-depth investigation of questions raised in the present study, including the concept of early versus delayed initiation of lenvatinib, and whether this has an impact on clinical outcomes.

Conclusion

Advanced thyroid cancer is an aggressive and sometimes rapidly progressive disease, presenting a challenging clinical situation with relatively few validated treatment options [7, 32, 33]. Precise treatment selection for individual patients is critical to maximize efficacy outcomes. Our study focused on lenvatinib, an oral multitargeted TKI approved for advanced thyroid cancer. Using a large, heterogenous patient cohort we demonstrated excellent treatment outcomes, with many prolonged responses and prolonged survival. We also identified older age, poorer performance status, liver metastases and TP53 mutations to predict for shorter time to treatment discontinuation. These findings provide early evidence for the use of clinicopathological biomarkers to guide initiation of lenvatinib in advanced follicular cell-derived thyroid cancer patients, in order to achieve the best outcomes for these patients.

Author contributions

KMR, OM, AM, CCB and LXM conceptualized the study. KMR, TX, CCB and LXM performed data collection. KMR, YD and LXM analyzed the data. KMR, YD, TX, AE, DG, OM, AM, JL, MKK, CCB and LXM were involved in manuscript preparation and had final approval of the submitted version.

Data availability

Requests for data sharing may be addressed to the corresponding author.

Competing interests

LXM is a consultant for Eisai and Bristol Myers Squibb. OM has served as an advisory board member for Bayer regarding the multidisciplinary Canadian consensus on the multimodal management of high-risk and radioactive iodine-refractory thyroid carcinoma, and also for Eli Lilly Canada (facilitated directly via Precision RxDx) for participation in working group meetings on Recommendations on Actionable Biomarker Testing for Thyroid Cancer Management. MKK reports funding from Eisai, Exelixis, Ipsen, Eli Lilly and Novartis, and is on the Data Safety Monitoring Board for Novartis, and is an advisory board member for Ipsen. All other authors have declared no competing interests.

Ethics approval and consent to participate

This study was approved by the University Health Network (UHN) Research Ethics Board (CAPCR 23-5561). The study was performed in accordance with the Declaration of Helsinki.