A Phase II Redifferentiation Trial with Dabrafenib-Trametinib and ¹³¹I in Metastatic Radioactive Iodine Refractory BRAF p.V600E-Mutated Differentiated Thyroid Cancer

This article is featured in Highlights of This Issue, p. 2337

Distant metastases from differentiated thyroid cancer (DTC) with radioiodine uptake are treated every 6 to 12 months with high activities of ¹³¹I. This provides long-term benefits, in particular in young patients with a well-differentiated histotype, with small distant metastases disclosing high ¹³¹I uptake and no fluorodesoxyglucose (¹⁸FDG) uptake (1). However, two thirds of metastatic patients become radioiodine refractory (RAIR) during their disease course because of a loss of differentiation of cancer cells with an impaired iodine metabolism (1, 2). First-line treatment for progressive RAIR disease is based, in the absence of NRTK or RET fusion, on lenvatinib or sorafenib which both improve the progression-free survival (PFS) in comparison with placebo. Treatment though is prolonged with a high rate of toxicity that leads to a decrease in quality of life (QoL) and frequently to a decrease in drug dosage or even in its withdrawal (3–5).

Tumor BRAF p.V600E mutation is responsible, through the activation of the mitogen-activated protein (MAP) kinase pathway, for the loss of expression of proteins involved in the iodine metabolism (6). In experimental models of BRAF p.V600E-mutated tumors, the inhibition of the MAP kinase pathway is able to revert this dedifferentiated phenotype (7). In patients with RAIR DTC, a pioneer study showed that the inhibition of the MAP kinase pathway with selumetinib, a MEK inhibitor, used during 6 weeks induced the reappearance of significant ¹³¹I tumor uptake in 44% of the BRAF p.V600E-mutated DTC patients who were then treated with ¹³¹I, resulting in partial tumor responses in 11% of the patients who were then treated with ¹³¹I (8). Re-induction of ¹³¹I uptake was also obtained with BRAF inhibitors such as vemurafenib or dabrafenib with tumor response following ¹³¹I treatment occurring in 25% and 20% of the patients, respectively (9, 10). Recently, a combination of vemurafenib and the anti-ErbB3 mAb CDX-3379 re-induced ¹³¹I uptake in 5 of 6 patients and after ¹³¹I treatment a partial response (PR) at 6 months was observed in 2 (11). Finally, in 6 patients treated for with dabrafenib and trametinib (a MEK inhibitor), ¹³¹I uptake was re-induced in 33% of the cases with a 17% PR rate, 3 weeks after ¹³¹I treatment (12). The best drug combination for ¹³¹I uptake re-induction and the length of benefits are however largely debated.

A profound inhibition of the MAP kinase pathway achieved with a combination of an anti-BRAF and of an anti-MEK was more effective to induce redifferentiation and tumor response in BRAF p.V600E-mutated tumors in a murine model (13).

In the current phase II trial in patients with metastatic BRAF p.V600E-mutated DTC, we evaluated the efficacy and safety of a combination of dabrafenib and trametinib before the administration of a high ¹³¹I activity following recombinant human (rh)TSH injections.

MERAIODE is a multicentric prospective phase II trial in patients with RAIR DTC, with two independent cohorts, one for BRAF p.V600E and one for RAS mutated DTC. This latest cohort is ongoing. This investigator-initiated trial is sponsored by Gustave Roussy. It was conducted within the French Endocan-TuThyRef network, in accordance with the protocol approved by ethics committees and national authorities and in accordance with the Declaration of Helsinki. All patients gave their written informed consent.

Patients above the age of 18 years had a thyroid carcinoma of follicular origin with a BRAF p.V600E mutation (controlled in a central laboratory) that was considered RAIR, based on the presence of distant metastases without ¹³¹I uptake on a post-therapy radioactive whole-body scan (WBS) and/or of distant metastases disclosing RECIST progression within 12 months after a therapeutic ¹³¹I administration. Patients had measurable disease, no lesion larger than 3 cm, and RECIST progression within 18 months prior to enrollment. Representativeness of Study Participants are discussed in the Supplementary Table S1. Inclusion and exclusion criteria are detailed in the Supplementary Table S2.

After enrollment, the baseline work-up included a brain MRI, a neck-chest-abdomino-pelvic CT scan without contrast, a ¹⁸FDG PET/CT and a ¹³¹I diagnostic WBS (dc1-WBS) performed 2 days after the administration of 185 MBq (5 mCi) of ¹³¹I following rhTSH (0.9 mg intra-muscularly for 2 consecutive days with ¹³¹I administration 1 day after the second injection). Urinary iodine measurement ruled out any iodine overload.

At day 1 (D1), patients started treatment with dabrafenib (150 mg twice a day) and trametinib (2 mg once a day) for 42 days. Between D28 and D35, patients underwent a neck and chest CT-scan without contrast and a second ¹³¹I dc-WBS (dc2-WBS) using the same procedure as for the dc1-WBS. At D35 ±2 days, and whatever the results of the dc2-WBS, a fixed ¹³¹I activity of 5500 MBq (150 mCi) was empirically administered after rhTSH injections with a posttreatment WBS 5 days after ¹³¹I administration. The dabrafenib-trametinib treatment was maintained until posttreatment WBS was performed.

Radiologic response was assessed according to RECIST version 1.1 at 1, 3, 6, 12, and 18 months after treatment initiation or until tumor progression (14). Except at baseline and at 1 month, contrast was used for the CT-scans.

In case of PR at 6 or 12 months and according to the investigator assessment, patients could undergo a second course of dabrafenib-trametinib-¹³¹I (5500 MBq (150 mCi) after rhTSH), following the same procedures, but brain MRI, dc1–2-WBS and ¹⁸FDG PET/CT were not performed. Following the second treatment course, similar radiologic evaluations were performed.

The trial was conducted based on local radiologic assessments. The independent central review of the CT scans, WBS and PET/CT was performed at the end of the study. For RECIST evaluation, target lesions defined on the baseline CT scan of the first treatment course were used for the evaluation of the second treatment course. Metabolic ¹⁸FDG-response was assessed at 6 months according to PERCIST (15). Patterns of abnormal ¹³¹I uptake on the WBSs were compared in terms of number of metastatic organs with detectable ¹³¹I uptake.

Levels of serum thyroglobulin (Tg), and Tg antibodies were measured on site on l-thyroxine treatment at baseline, 3 and 6 months after treatment initiation. PR in serum Tg levels was defined as a decrease from baseline > 50% in the absence of anti-Tg antibodies.

Safety was monitored at D14, 28, 35, 42, and at months 3, 6, 12, and 18, according the NCI-CTC AE version 4.0.

QoL was evaluated with the EORTC QoL Questionnaire C30 (QLQ-C30) at inclusion, at D14, 28 and 42, and at 3 and 6 months of the first treatment course.

The main endpoint was the objective response rate (ORR) at 6 months determined by a central review according to RECIST version 1.1.

Secondary endpoints included: (i) ORR (central and local reviews) at 1 and 3 months, (ii) ORR (central and local reviews) at 12 and 18 months for patients treated once, (iii) ORR (central and local reviews) at 1, 3, 6, 12, and 18 months after the initiation of the second treatment course for patients treated twice, (iv) ¹⁸FDG metabolic response (central review) at 6 months, (v) PFS (central and local RECIST evaluation), (vi) changes in serum Tg levels, (vii) predictive role on the 6-months ORR of the intensity of ¹⁸FDG uptake at baseline, of the ¹³¹I uptake on the dc2-WBS and of the 1-month ORR, (viii) safety and (ix) QoL evolution during treatment.

The study design was a one-stage Fleming design. For patients with a BRAF p.V600E mutation, the treatment of reference (¹³¹I alone) was considered ineffective and the treatment under evaluation (dabrafenib-trametinib-¹³¹I) would be considered not effective if the ORR < 3% (H0: p≤p0) and of interest if the ORR is ≥ 25% (H1: p ≥p1). This chosen ORR was lower than the 65% ORR reported with lenvatinib alone (16), assuming a lower efficacy due to a shorter duration of treatment.

With a unilateral risk α = 5% and a power of 95%, 20 patients had to be included: if ≥ 3 responses were observed, the evaluated treatment is considered effective.

Continuous variables are described as number of observations, mean value, stable disease (SD), and range. Categorical variables are summarized as counts and percentages.

Confidence intervals are given at 95%, except for the estimates of the ORR which are given with their 90% exact confidence intervals (corresponding to one-sided 5%) using binomial distribution, considering the small population size. Analyses of the main criteria were performed on the modified intent to treat population in evaluable patients, i.e., in patients who received at least one complete course of treatment with dabrafenib-trametinib-¹³¹I in the absence of iodine contamination. A descriptive exploratory analysis was done on patients who received a second treatment course. A post hoc analysis of the time without retreatment was performed.

Paraffin sections were prepared from tumor tissue blocks, according to the sandwich method. Library preparation, sequencing techniques, pipeline were performed with workflow Oncomine Tumor Mutation Load Assay (ThermoFisher Scientific, Courtaboeuf, France), covering 1.65 Mb among 409 genes. Sanger sequencing was performed for TERT promoter analysis (17). Gene expression (GE) were generated by targeted RNA sequencing using the HTG EdgeSeq technology and the Oncology Biomarker Panel (18). Details of samples analyses by NGS are available in Supplementary Table S3. GE raw data are available at ArrayExpress under the E-MTAB-12837 identifier.

The trial was supported by a grant from Institut National du Cancer (INCa). Genzyme provided thyrotropin alfa (rhTSH, Thyrogen); Novartis provided dabrafenib and trametinib. These companies did not participate in any aspect of the study design, data accrual, data analysis, or manuscript preparation and submission. The authors assume responsibility for the accuracy and completeness of the data and analyses.

The data generated in this study are not publicly available due to privacy concerns but are available upon reasonable request from the corresponding author.

From March 2018 to January 2020, 24 patients were included in eight centers (Table 1). Most patients were women (63%) with a mean age of 67 years, most had a tall cell papillary thyroid cancer (central review) and only 1 patient had received a systemic treatment. Among them, 21 patients were considered evaluable at 6 months for the efficacy of the first treatment course. The other 3 patients were not evaluable, one for psoas hematoma and consent withdrawal at D8 and two for iodine contrast administration prior ¹³¹I administration (Fig. 1). Based on the investigator RECIST assessment, 11 patients with PR at 6 or 12 months were retreated with a second course of dabrafenib-trametinib-¹³¹I. One patient did not complete this second treatment course because of the COVID-19 pandemic. Baseline median TSH level was 0 mUI/L.

dc-1-WBS was performed after baseline CT in all cases except one in whom it was performed 7 days before.

Abnormal ¹³¹I uptake was present at baseline (dc1-WBS) in a single metastatic site in one of the 21 patients (4.8%), on dc2-WBS in 11 of the 17 patients in whom it was performed (64.7%) and on the posttreatment-WBS in 20 of the 21 patients (95.2%). ¹³¹I uptake on posttreatment-WBS was located in lungs (20 patients), neck (thyroid bed or lymph nodes; 12 patients), mediastinal lymph nodes (6 patients) and bones (3 patients). Six patients without any metastatic uptake on dc2-WBS disclosed abnormal uptake on the posttreatment-WBS. The mean number of metastatic site with abnormal ¹³¹I uptake was 1.3 on dc2-WBS and 2.6 on the posttreatment-WBS and the number of metastatic site with abnormal uptake was higher in 8 patients on posttreatment-WBS compared with dc2-WBS. The one patient with abnormal ¹³¹I uptake at baseline, had abnormal uptake in the neck at baseline and in the lung and the neck on dc2-WBS.

At 6 months following the first treatment course, there were no CR, 8 (38%) PR, 11 (52%) SD, and 2 (10%) PD (Table 2; Fig. 2; Supplementary Table S4; Supplementary Fig. S1; Supplementary Fig. S2). The median size change of the target lesions was −22% (mean = −30%; range, −79 to +46). The ORR at 1 and 3 months were 48 and 57%, respectively. The kinetics of tumor size (Fig. 2) revealed an early decrease that was already present at 1 month, before the ¹³¹I treatment.

The 1-month size changes, the uptake on dc2-WBS and the intensity of ¹⁸FDG uptake at baseline did not predict for the 6-months RECIST response (Table 3).

FDG PET/CT were performed under levothyroxine treatment in most case (17/20 at baseline and 20/20 at 6 months) and after rhTSH in 3 cases at baseline. PERCIST evaluation at 6 months, available in 20 patients was PR in 13 cases (65%), SD in 4 (20%), and PD in 3 (15%).

Among the 15 patients without serum anti-Tg antibodies, mean Tg level was 96.2 ± 223.0 ng/mL at baseline, 32.5 ± 52.2 ng/mL at 3 months and 23.9 ng/mL ± 40.3 ng/mL at 6 months. The 3- and 6-month biochemical PR rates were both 33% [95% confidence interval (CI), 12–62] with a 3 and 6-months median TSH level of 0 mUI/L.

Among the 10 patients who did not receive a second course of dabrafenib-trametinib-¹³¹I, 1 patient had a persistent PR at 12 and 18 months (Table 2).

Among the 10 evaluable patients for the second treatment course, 9 (90%) had abnormal ¹³¹I uptake on the second posttreatment WBS. The mean number of metastatic organ was 1.6 (range, 1–4) on the second posttreatment WBS and 2.1 (range, 1–4) on the first posttreatment WBS and the number of metastatic organs with ¹³¹I uptake was lower in 5 patients on the second posttreatment WBS compared with the first posttreatment WBS.

At 6 months, among the 10 evaluable patients, 6 had a PR at 6 months and 1 patient achieved a complete response from the first month of the second course onwards (Table 2; Fig. 2; Supplementary Table S4; Supplementary Fig. S1; Supplementary Fig. S2). The median size change of target lesions since the baseline of the first treatment course was −53% (mean = −50%; range, −71 to −18), but compared with the baseline of the second course, it was 0% (mean = −11%; range, −50 to +16). Among the 8 patients without serum Tg antibodies, 3 had a persistent biochemical response.

On January 1, 2022, the median follow-up time was 34 months (range, 0.5–46.2). The median PFS was not reached according to central evaluation and was 21.8 months with local assessment. The 12-month PFS rates, according to central and local evaluation, were 82.0% (95% CI, 58.8–92.8) and 86.3 (95% CI, 63.4–95.4), respectively. The 24-month PFS rates, according to central and local evaluation, were 67.5% (95% CI, 43.5–83.0) and 48.0% (95% CI, 25.9–67.2), respectively. One death due to disease progression occurred at 24 months. The median time without retreatment was 39.6 months 95% CI (23.6–45.7) with a 12 months and 24 months without retreatment rate of 9.5% and 28.6%, respectively. None of the patient developed dedifferentiation with anaplastic transformation or another primary malignancy during follow-up.

During the courses of dabrafenib-trametinib, 23/24 (96%) patients experienced at least one adverse event (AE), related or not to treatment. Most AEs were grade 1 (25%) and grade 2 (42%; Table 4). There were 9 grade 3 AEs in 6 patients and one grade 4. Treatment related AEs occurred in 21 (88%) patients.

Treatment was definitively stopped in 2 cases due to AE. In one case, it was stopped at D8 for a grade 3 psoas hematoma and the patient withdrew his consent form. In the other case, treatment was stopped at D37 for a grade 4 anicteric cholestasis diagnosed after ¹³¹I administration on that day.

Treatment was temporary withdrawn in 3 patients (2 days for a grade 2 fever, 5 days for a grade 2 thrombopenia and 3 days for a grade 2 fever, asthenia, and anorexia).

Following the treatment initiation, the global health and all the functioning dimensions (except for emotional functioning) were impacted for 6 months (Supplementary Table S5; Supplementary Table S6; Supplementary Fig. S3). A continuous deterioration of scores was observed from baseline to the end of treatment (D42). The lowest QoL score occurred at the end of treatment with more than 40% of patients having a clinically significant QoL deterioration (≥ 5 points) in the global health, physical functioning and role functioning dimensions. This deterioration was reversible, as scores raised after the end of the treatment course until 6 months. Similar trends were observed for symptoms, with half of the patients presenting an increase ≥ 5% for fatigue, nausea/vomiting and diarrhea at the end of the treatment course.

Fourteen tissue samples were available for NGS genotyping. Three poor quality samples were excluded because of low sequencing QC metrix. The average coverage depth was 773X (399X to 930X) allowing good variant detection. Tumor mutational load ranked from 0.8 to 11.3 mutations per megabase. For all contributive samples, the presence of a BRAF p.V600E mutation was indeed confirmed. Seven samples harbored a TERT promoter (rs1242535815) NM_198253.3:c.-124C>T mutation. Nine samples had one to five additional variants, mainly of unknown significance (VUS). Three pathogenic mutations were detected in separate samples, in SMARCA4 (NM_001128849.3: c.1513_1514delGC; p.Ala505fs); TRIP11 (NM_004239.4:c.2611C>T; p.Arg871Ter) and HNF1A (NM_000545.6:c.864delG; p.Pro291GlnfsTer51). Seven of the 409 genes covered by Oncomine Tumor Mutation Load Assay were related to SWI/SNF pathway (ARID1A, ARID2, BCL11A, BCL11B, PBRM1, SMARCA4, SMARCB1). In addition to the SMARCA4 pathogenic mutation, four VUS (missense variants) were observed for 2 other patients, two in ARID1A, one in ARID2, and one in BCL11A. Detailed description of all curated variants is available in Supplementary Table S7.

The GE analysis of 2,559 transcripts from genes associated to cancer in thirteen samples did not show differential expression associated to tumor response.

The inhibition of the MAP kinase pathway for a short period of time restores ¹³¹I uptake in most metastatic BRAF p.V600E-mutated DTC patients and at 6 months following a rhTSH aided ¹³¹I treatment, a partial tumor response is observed in 38% of patients.

The rate of patients with metastatic ¹³¹I uptake on the post-therapy WBS is higher than previously reported with either vemurafenib or dabrafenib alone and similar to that reported with a combination of vemurafenib and Anti-ErbB3 mAb (10, 11). It was much higher than in a previous study on 6 patients that used dabrafenib and trametinib but during a shorter period of time (21 days vs. 45 days; ref. 12). Re-induction of ¹³¹I uptake by itself is however not sufficient to consider the treatment successful, and the theoretical advantage of the redifferentiation strategy over a long-term conventional treatment is to reach long-term tumor responses with a treatment of short duration. Here, we report a 1, 3 and 6-months PR rates of 48%, 57% and 38% in patients who were all RECIST progressive within 18 months prior to enrollment. Compared with long-term BRAF inhibitors this response rate is higher (vs. 27 to 39% with vemurafenib, 29% to 33% with dabrafenib alone and 35% with dabrafenib plus trametinib; refs. 19–21). Similarly, compared with long-term BRAF inhibitors the median PFS is longer (19). The patients included in the current study were however treated as first-line treatment, with a limited tumor volume (no lesion larger than 3 cm) and prior progression within 18 months whereas in trials with long-term BRAF inhibitors patients were treated as second or third line with prior progression within 12 to 14 months and some patients had larger tumor volume. The best tumor response rate is 57%, at 3 months, higher than previously reported (20%–25%) in a cohort of patients that were all RECIST progressive within 18 months prior to enrollment which was not the case in other redifferentiation trials. Given the difference in patients selection and their small numbers, any direct comparison of response rate with other trials cannot be done.

It is interesting to note though that the median time without retreatment was 39.9 months which is a long interval of time during which patients can stay off systemic treatment including kinase inhibitors and enjoy a good QoL.

In many patients, a decreased tumor size was already significant at one month of dabrafenib-trametinib treatment and this clearly raises the question whether benefits from the treatment are related to the dabrafenib-trametinib treatment or to the addition of ¹³¹I to dabrafenib trametinib. Because of the long PFS, and because the 6 months response was independent from the 1-month tumor response, we believe that the treatment efficacy is not only due to dabrafenib-trametinib treatment during 6 weeks but also to its association with ^131II. A comparative randomized trial would be the only way to answer this specific question.

The high efficacy of the current combination might be related to a more complete inhibition of the MAP kinase pathway than that achieved with a single agent. It might also be related to the selection of patients with small tumor foci and a slow progression rate, which are two characteristics observed among the best responders to conventional ¹³¹I treatment (1). Our patients however had at least three negative predictive factors for response to conventional ¹³¹I treatment, namely age above 55 years, high ¹⁸FDG uptake and RECIST progression prior to enrollment. Length of treatment with dabrafenib-trametinib treatment might also have a high impact if we consider the rate of patients with ¹³¹I uptake.

Interestingly, the dc2-WBS was not predictive for the uptake on the posttreatment WBS and for treatment efficacy, and this might be related to several factors; first, a longer exposure (1 week) to dabrafenib-trametinib treatment. Second, the repetition of rhTSH stimulation before treatment might increase the ¹³¹I uptake. Finally, T-WBS is more sensitive than dc-WBS for the detection of foci with low uptake, but in these lesions only seen on T-WBS, the radiation dose delivered to tumor foci is expected to be low (22). Dc-WBS were analyzed in a binary fashion (abnormal or not) which might be not precise enough but has the advantage of being simple and was indeed the method used for other trials (10). These data suggest that pre-¹³¹I therapy dosimetry, when performed, should be interpreted with caution in redifferentiation protocols. Furthermore, the intensity of ¹⁸FDG uptake at baseline was not predictive for the 6-months tumor response. A lesion dosimetry analysis based on ¹³¹I scintigraphy after therapeutic administration of ¹³¹I is ongoing, with analysis of correlations between the delivered dose and tumor response and FDG uptake.

For the first time, patients who achieved a PR after the first course of dabrafenib-trametinib-¹³¹I had the possibility to undergo a second treatment course. The observed metastatic uptake was then lower that the first post-therapy uptake, and this decrease might be linked to a less efficient redifferentiation effect or to the eradication of the most differentiated cells after the first treatment course. This second treatment course led to a long lasting complete response in 1 case, but the mean decrease in target lesion sizes was modest and does not support a strategy of repeating these treatment courses over time.

Redifferentiation strategy offers the advantage of a short exposure to kinase inhibitors and the rate of AEs and of discontinuation was indeed lower than during long-term treatment with dabrafenib and trametinib for melanoma or lung cancer (23, 24). This rate is also lower than during treatment of RAIR DTC with lenvatinib or cabozantinib (16, 25). However, the occurrence of grade 3–4 AEs led to treatment discontinuation in 2 patients and temporary discontinuation in 3 patients. This, indeed, is not in favor of treating patients with asymptomatic DTC with nonprogressive small metastases. Of note, the study was not designed to address concerns for progression to ATC or secondary malignancies, but we did not observe any dedifferentiation with anaplastic transformation in a nearly 3-year median follow-up time.

TERT promoter mutations were found both in patients with or without tumor response. Interestingly, the patient with a mutation in SMARCA4, a gene involved in the SWI/SNF pathway known to be associated with tumor resistance to redifferentiation disclosed a tumor response (26).

The limitations of this study are the small number of enrolled patients, the central imaging review performed at the end of the study with treatment decisions based on local assessment which could have introduced a bias, and also the absence of tumor biopsies under treatment prior to ¹³¹I administration to describe changes related to dabrafenib-trametinib treatment. Redifferentiation strategies are indeed difficult trials, because of the timelines of short treatment duration and constraints linked to radionuclide therapy, leading to heterogeneity with missing exams such as the Dc2-WBS (27). Furthermore, only a prospective clinical trial comparing redifferentiation with a standard of care, either active surveillance or anti angiogenic or BRAF inhibitors will be able to definitively prove the benefits of redifferentiation.

It is, however, the first prospective study that used an association of dabrafenib-trametinib treatment with ¹³¹I treatment administration planned for all patients and not based on the dc2-WBS findings. It is also the first study to provide long-term follow-up and to offer the possibility of a second course of dabrafenib-trametinib-¹³¹I treatment.

The association of dabrafenib and trametinib in BRAF p.V600E-mutated DTC is effective for restoring ¹³¹I RAI uptake and the administration of high activity of ¹³¹I is followed by a significant rate of long-lasting tumor responses. These favorable results might apply to selected patients with BRAF p.V600E-mutated DTC with relatively small metastases and relatively slow progression rate. Whether patients with other clinical characteristics might also benefit from this protocol needs to be confirmed by further studies.

S. Leboulleux reports nonfinancial support from Novartis and Sanofi Genzyme during the conduct of the study as well as personal fees from Lilly, Eisai, and Ipsen outside the submitted work. L. Lamartina reports personal fees from Eisai, Lilly, Ipsen, and Roche and other support from AAA Novartis outside the submitted work. I. Borget reports grants from French Ministry of Health through the INCa (PHRC2015) and nonfinancial support from Novartis and Sanofi Genzyme during the conduct of the study as well as personal fees from Roche, BMS, Novartis, and MSD and other support from Gilead outside the submitted work. No disclosures were reported by the other authors.

S. Leboulleux: Conceptualization, resources, supervision, funding acquisition, validation, investigation, visualization, writing–original draft, project administration. C. Do Cao: Resources, investigation, writing–review and editing. S. Zerdoud: Resources, investigation, writing–review and editing. M. Attard: Resources, formal analysis, investigation, writing–review and editing. C. Bournaud: Resources, investigation, writing–review and editing. L. Lacroix: Resources, investigation. D. Benisvy: Resources, investigation, writing–review and editing. D. Taïeb: Resources, supervision, investigation, writing–review and editing. S. Bardet: Resources, investigation, writing–review and editing. M. Terroir-Cassou-Mounat: Resources, investigation, writing–review and editing. N. Anizan: Resources, investigation, writing–review and editing. E. Bouvier-Morel: Data curation, software, investigation. L. Lamartina: Resources, investigation, writing–review and editing. G. Lion: Resources, supervision, investigation, writing–review and editing. S. Betrian: Resources, investigation, writing–review and editing. C. Sajous: Resources, investigation, writing–review and editing. A. Schiazza: Resources, investigation, writing–review and editing. M.-E. Garcia: Resources, investigation, writing–review and editing. R. Ciappuccini: Resources, investigation, writing–review and editing. M. Schlumberger: Conceptualization, resources, investigation, visualization, writing–review and editing. A. Al Ghuzlan: Resources, investigation, writing–review and editing. Y. Godbert: Resources, investigation, writing–review and editing. I. Borget: Conceptualization, resources, data curation, software, formal analysis, supervision, validation, visualization, methodology, writing–original draft, project administration.

The authors would like to thank the patients, their families, all research staff, and investigators involved in this study and also Catherine Richon and Patrick Saulnier from BMO unit in AMMICa platform UAR3655/US23 for their technical assistance for Tissue genotyping and GE analysis and also Bastien Job from Bioinformatic unit in AMMICa platform UAR3655/US23 for GE bioinformatics analysis.

MERAIODE was financed by the French Ministry of Health, through the INCa (PHRC2015). Dabrafenib and Trametinib were provided by Novartis; rhTSH was provided by Sanofi Genzyme (NCT 03244956).

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