RET Fusion as a Novel Driver of Medullary Thyroid Carcinoma

Abstract

Introduction:

Oncogenic RET tyrosine kinase gene fusions and activating mutations have recently been identified in lung cancers, prompting initiation of targeted therapy trials in this disease. Although RET point mutation has been identified as a driver of tumorigenesis in medullary thyroid carcinoma (MTC), no fusions have been described to date.

Objective:

We evaluated the role of RET fusion as an oncogenic driver in MTC.

Methods:

We describe a patient who died from aggressive sporadic MTC < 10 months after diagnosis. Her tumor was evaluated by means of next-generation sequencing, including an intronic capture strategy.

Results:

A reciprocal translocation involving RET intron 12 was identified. The fusion was validated using a targeted break apart fluorescence in situ hybridization probe, and RNA sequencing confirmed the existence of an in-frame fusion transcript joining MYH13 exon 35 with RET exon 12. Ectopic expression of fusion product in a murine Ba/F3 cell reporter model established strong oncogenicity. Three tyrosine kinase inhibitors currently used to treat MTC in clinical practice blocked tumorigenic cell growth.

Conclusion:

This finding represents the report of a novel RET fusion, the first of its kind described in MTC. The finding of this potential novel oncogenic mechanism has clear implications for sporadic MTC, which in the majority of cases has no driver mutation identified. The presence of a RET fusion also provides a plausible target for RET tyrosine kinase inhibitor therapies.

Activating point mutations occurring in the RET gene have been identified as the oncogenic driver in all cases of hereditary medullary thyroid carcinoma (MTC) and approximately 40% of sporadic MTC cases (1). Mutations of RAS genes have recently risen as second drivers found in 10–15% of MTC cases (2). Still, approximately half of sporadic MTC tumors have no identified mechanism of tumorigenesis, representing a gap in knowledge. With the recent development of targeted therapies in MTC, there is an obvious need to further define patient-specific mutations to better personalize treatment. This necessity is especially urgent in patients presenting with aggressive systemic disease in which the role of effective targeted therapy is essential, such as the patient described below. Oncogenic fusions involving RET have been described in well-differentiated thyroid cancer and non-small cell lung cancer. RET fusion oncoprotein overexpression is thought to lead to aberrant kinase dimerization and downstream RAS/MAPK pathway activation (3). Although a RET gene translocation mechanism has not been described to date in MTC, given the known importance of RET activation in this tumor, such a mechanism is reasonable to investigate.

Clinical Presentation

A 46-year-old otherwise healthy Caucasian female was incidentally found to have multiple hepatic lesions during a laparoscopic cholecystectomy for symptomatic cholelithiasis. Intraoperative liver biopsy identified a carcinoma of unknown origin that was immunohistochemically positive for calcitonin (CTN), carcinoembryonic antigen (CEA), TTF-1, synaptophysin, and CK7 and negative for CK20, CDX-2, and MART-1. On further staging evaluation, she was found to have a 2.3-cm left thyroid mass that, on fine-needle aspiration, was consistent with MTC. She also had radiological evidence of left lateral neck metastasis and multiple, bilobar liver metastasis, the largest of which was 2.6 cm within segment IV. The remainder of her imaging studies (brain magnetic resonance imaging, chest computed tomography, and bone scan) were without evidence of metastasis. Despite this substantial volume of disease, her baseline serum CTN was only 223 pg/mL (reference range ≤5 pg/mL), and CEA was 43 ng/mL (reference range 0.0–3.0 ng/mL). She underwent germline testing of RET exons 10, 11, and 13–16, which failed to identify any known pathogenic mutation, suggesting that her MTC was sporadic and not hereditary in origin. Additionally, the patient had no family history of MTC, pheochromocytoma, primary hyperparathyroidism, or neuromas to suggest a hereditary syndrome. A total thyroidectomy and compartment-focused lymph node dissections of level VI and left levels IIA, IIB, III, and IV were performed, with a final pathological diagnosis of MTC, stage IVC (T3N1bM1). The cervical disease consisted of a unifocal left thyroid lobe 3.0-cm tumor with extrathyroidal extension, lymphovascular invasion, and perineural invasion present. Three of 91 lymph nodes were positive for metastatic MTC with extranodal extension: one of 26 in the central neck, and two of 65 in the left lateral neck, in levels II and III. Immunohistochemical evaluation confirmed the diagnosis, showing that the tumor cells were positive for CTN, CEA, TTF-1, CK19, synaptophysin, and chromogranin. Two months after surgery, the patient's CTN and CEA had risen to 641 pg/mL and 74 ng/mL, respectively, prompting reimaging that depicted no evidence of disease in the neck but progressing liver metastasis and development of bone metastases in the pelvis and spine. Because of significant progressive disease, the patient was treated with sunitinib (no Food and Drug Administration [FDA] approved drugs for progressive MTC existed at that time). Before sunitinib initiation, CTN was 1019 pg/mL, and CEA was 81.4 ng/mL; after one cycle of treatment (sunitinib 50 mg daily for 4 wk, followed by 2 wk off), tumor markers declined (CTN, 678 pg/mL; CEA, 59 ng/mL), suggesting an initial response to therapy. However, after her second cycle of sunitinib, there was radiographic evidence of progression in bone and liver lesions by Response Evaluation Criteria in Solid Tumors (RECIST), as well as rising tumor markers (CTN, 1018 pg/mL; CEA, 72 ng/mL). At this point, sunitinib was discontinued. Before initiation of second-line therapy, the patient's performance status declined, and she died from progressive metastatic disease less than 10 months from initial diagnosis.

Genomic Profiling of the Tumor

Given the aggressive nature of this disease, its unique presentation of minimal cervical adenopathy in the setting of extensive distant metastasis, and incomplete response to targeted therapy, we performed genome profiling in an attempt to uncover potential underlying causes of tumorigenesis. Targeted next-generation sequencing (NGS) was carried out in a Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory (Foundation Medicine, Inc) on formalin-fixed, paraffin-embedded primary MTC tumor tissue (20 × 15 mm). Hematoxylin and eosin staining confirmed 80% tissue cellularity with 80% tumor cells. DNA extraction, library preparation, sequencing, and data analysis have been described in detail by the recent Meric-Bernstam et al publication (4). The genomic library was prepared from captured DNA for 3230 exons in 182 cancer-related genes plus 37 introns from 14 genes often rearranged in cancer.

No somatic mutations, novel truncations of tumor suppressor genes, gene amplifications, or likely homozygous gene deletions were observed among the 182 genes queried, including RET and RAS. Within the 14 genes with described intronic rearrangements, a reciprocal translocation involving RET intron 12 was identified. The allelic sequences spanning the breakpoints were obtained using a local breakpoint assembly approach (5). Detailed analysis derived the two translocation products as 5′MYH13-chr17:10210830 3′RET+chr10:43611045 and 5′RET+chr10:43611152 3′MYH13-chr17:10219396 (GRch37/hg19). The 5′MYH13 3′RET product is schematically depicted in Figure 1. The specific involvement of RET intron 12 was validated using a targeted break apart fluorescence in situ hybridization probe (CymoGen Dx LLC) tested in a CLIA-certified lab (Figure 2). Sequencing of RNA derived from frozen tumor confirmed the existence of an in-frame fusion transcript joining MYH13 exon 35 with RET exon 12 (Figure 1). The reciprocal fusion transcript was also observed (data not shown).

Figure 1.

The MHY13-RET kinase fusion is caused by a translocation of chromosome 17 and chromosome 10 joining the first 35 exons of MHY13 with exon 12–20 of RET. The in-frame junction of the fusion transcription was confirmed with Sanger sequencing of cDNA generated from patient tumor.

Figure 2.

RET rearrangement detected by fluorescence in situ hybridization in the sporadic MTC of a patient who died of disease at age 46 years. The RET (10q11.21) break apart probe was from Cymogen DX, with the centromeric side labeled in orange and the telomeric side labeled in green. Cells without rearrangement show overlap with the generation of a yellow signal, whereas cells with rearrangement (arrows) show isolated red and green signals.

Functional Analysis of the RET Fusion Gene

To evaluate the functional impact of the fusion transcript, we employed the murine Ba/F3 cell reporter model (6). This bone marrow-derived pro-B cell line is dependent on IL-3 for proliferation. Oncogenic transformation allows IL-3-independent growth, thus representing a sensitive tool for measuring the effect of an introduced alteration on cell proliferation and survival (6). Ba/F3 cells were transduced with expressed protein control (mCherry), wild-type proteins (RET-WT, MYH13-WT), or the predicted fusion proteins (MYH13-RET, RET-MYH13). Previously established PIK3CA mutants served as oncogenic controls (PIK3CA_H1047R) (7). Transduced cells were incubated with an IL-3 concentration (0.5 pg/mL) to support Ba/F3 marginal growth for approximately 96 hours. Cell viability was measured at three time points (1, 1.5, and 2 wk). Figure 3A provides representative 1.5-week time point data. Observations were statistically reproduced at all time points in two independent experiments (data not shown). Ectopic expression of MYH13 or RET had no significant effect on cell viability compared to control. A similar finding was observed for the RET-MYH13 fusion construct. Consistent with other oncogenic RET fusions (8), overexpression of the kinase domain by the MYH13-RET exhibited strong “driver” activity. Indeed, proliferative activity was similar to the highly oncogenic PIK3CA (H1047R) construct (Figure 3A). Because our patient had an initial reduction in tumor markers after one cycle of sunitinib, we tested whether or not MYH13-RET activity was sensitive to RET-directed tyrosine kinase inhibitors (TKIs) (Figure 3, B–D). Dose response curves were derived for MYH13-RET Ba/F3 cells treated with sunitinib and the two FDA-approved TKIs for MTC (vandetanib and cabozantinib). All three drugs blocked IL-3-independent cell growth at concentrations similar to those reported for RET-PTC1 fusion (sunitinib IC₅₀ = 0.37 μm; vandetanib IC₅₀ = 0.41 μm; cabozantinib IC₅₀ = 0.06 μm) (9).

Figure 3.

A, MYH13-RET fusion transcript supports IL-3-independent Ba/F3 cell growth. Control and fusion constructs were transfected into Ba/F3 cells to assess the rescue cell viability in the absence of IL-3. Full-length reference sequences (WT) for RET and MYH13, along with the nonspecific protein mCherry were included as negative controls. To establish oncogenicity, constructs encoding both predicted fusion constructs, RET-MYH13 and MYH13-RET, were compared to the established positive control PIK3CA (H1047R). Data display the cell viability determined at 1.5 weeks for three independent transfections ± SD. Asterisks indicate significant difference from mCherry control (P < .001). B-D, Tyrosine kinase inhibitors targeting RET block MYH13-RET fusion-driven Ba/F3 cell growth. Cells were seeded in 96-well format at 1 × 10⁴ cells per well in medium with IL-3 or without IL-3. After overnight incubation, sunitinib (B), vandetanib (C), and cabozantinib (D) were added at indicated concentrations (0–10 μm; DMSO as vehicle). A final concentration of 0.1% DMSO was used in all wells. Data display the cell viability at 72 hours as mean ± SD normalized to 0 μm control (100%). Asterisks indicate significant difference from +IL-3 control (P < .005). Cell viability was measured using CellTiter-Glo assay (Promega) (A) or Prestoblue Cell viability Reagent (Life Technologies, Inc) (B–D) according to manufacturers' instruction.

Discussion

Although germline oncogenic point mutations in RET cause hereditary MTC and are responsible for approximately 40% of sporadic nonfamilial disease (10), the most recent COSMIC database figures fail to identify genetic drivers in nearly 50% of MTC (1). Therefore, additional unidentified mechanisms are responsible for MTC tumorigenesis. Unlike well-differentiated thyroid cancers and despite a shared anatomic location, no fusion oncogenes have yet been described in MTC. This report represents the first such observation to our knowledge. Our in vitro studies suggest that a MHY13-RET fusion leads to aberrant activation of RET kinase and should be considered as a potential new driver mutation of MTC with contemplation of targeted treatment with small-molecule kinase inhibitors.

In general, gene fusions occur through both intra- and interchromosomal translocations, uniting two coding regions within a single reading frame allowing expression of a novel protein. Although once associated almost solely with hematological malignancies, chromosomal rearrangements have become increasingly identified in solid tumors (11). Tyrosine kinase fusion genes, of which the RET rearrangement reported here is one, occur in approximately 5% of hematological, mesenchymal, and epithelial tumors (11).

The first fusion kinases discovered in solid tumors involved the RET and NTRK1 genes in well-differentiated epithelial-derived thyroid cancer (12). RET is not expressed in normal thyroid follicular cells, but the apposition of the 3′ kinase domain of RET to a 5′ partner of another gene and its promoter leads to fusion oncoprotein overexpression, aberrant kinase dimerization, and downstream RAS/MAPK pathway activation (3). The genomic rearrangement found in fusions is thought to be influenced by the spatial proximity of chromosomal regions in the nucleus, cellular stress, inappropriate DNA repair or recombination, and DNA sequence and chromatin features (13).

More recently, oncogenic fusions of RET with KIF5B (kinesin family member 5B) were identified in non-small cell lung cancer (NSCLC) tumors of both radiation-exposed and -naive patients (14). There is a higher prevalence of RET fusions in NSCLC among never or light smokers and younger cohorts, and this fusion occurs more frequently in poorly differentiated tumors (15). The identification of RET fusions has also led to revised targeting strategies beyond the more commonly observed oncogenic drivers (ALK, EGFR, KRAS, ROS1). Clinical trials are under way using a variety of different multikinase TKIs with RET activity, including vandetanib, cabozantinib, and ponatinib (16, 17).

The RET proto-oncogene has long been recognized as an oncogenic driver, either through creation of oncogenic fusions or activating mutations. To date, only activating RET point mutations have been identified in MTC. The success of the FDA-approved TKIs, vandetanib and cabozantinib, is believed to be in large part due to these agents' inhibition of RET and various other tyrosine kinases including VEGFR. Notably, responses are also observed in the absence of known RET point mutations. Thus, evaluation of additional mechanisms of tumorigenesis, including RET fusion, is warranted and is strongly supported by the treatment response observed in MYH13-RET-driven Ba/F3 cells.

Although the reciprocal fusion transcript was also observed, the data presented here, in addition to the existing literature, suggest that the 5′MYH13-RET3′ is the likely oncogenic form. Several common features define tyrosine kinase fusions in cancer (11), of which this fusion meets all criteria: 1) the portion of the tyrosine kinase gene that is involved in the fusion encodes the intact kinase domain (and includes a GXGXXG motif that is essential for activity [18]); 2) breakpoint is often conserved across the various fusions (for RET, the breakpoint is frequently found in intron 12 in lung adenocarcinoma [19], Spitz melanoma [20], and papillary thyroid cancer [PTC] [21]); and 3) each tyrosine kinase has numerous fusion partners, even within the same disease. Also, oncogenic translocations are typically the primary somatic alteration found and as such are the clear driver of tumorigenesis. This patient was one of 15 individuals with clinically aggressive MTCs evaluated by NGS for genetic drivers in our series. Of the remaining 14 patients, all had activating point mutations in either RET, HRAS, or KRAS; none had RET fusions (our unpublished data). Additionally, this patient was relatively young for MTC at the presentation of her sporadic disease, analogous to phenotypic findings in NSCLC and PTC patients with RET fusions, and she had aggressive disease, similar to NSCLC. She had no previous radiation exposure to her head and neck region, which although known to be an increased risk factor for fusion-associated carcinomas, is by no means a necessity. In PTC, for example, whereas 50–85% of radiation-associated tumors harbor a RET/PTC rearrangement, 10–40% of sporadic tumors in patients with no radiation history also have this driver alteration (22). It is also important to note that the fusion found in our study would not have been detected had an intronic capture strategy not been employed. This may explain the lack of reports of similar fusions in the limited number of MTC cases examined to date by NGS.

Summary

This case report describes a novel finding of an oncogenic RET fusion in MTC. The MHY13-RET fusion should potentially be considered as a new driver mutation of MTC because it is mutually exclusive from point mutations in RET and RAS. The existence of oncogenic RET in both a fusion and point mutation form in MTC is analogous to the identification of tumorigenic BRAF fusions and point mutations as drivers in PTC (23, 24). Given that we do not have identified driver mutations in most sporadic MTCs, further investigation of this novel mechanism is warranted, especially in the setting of known available targeted RET therapies. Going forward, we will evaluate for RET fusions in phenotypically aggressive sporadic MTCs that lack RET, HRAS, and KRAS; point mutations. If RET fusion is determined to be worth pursuing, using the CLIA-certified RET break apart assay, as is being utilized currently for NSCLC, will be another reasonable method of detection.