TG-IGF1R : A Novel Receptor Tyrosine Kinase Fusion Oncogene in Pediatric Thyroid Cancer

Introduction

Gene fusion is the main genetic mechanism of carcinogenesis in pediatric papillary thyroid cancer (PTC). Fusions of RET, NTRK3, NTRK1 and ALK and other receptor tyrosine kinases (RTK) with several partner genes have been identified in 60% to 70% of the cases. ^1,2 Fusion oncogenes involving RTKs trigger constitutive activation of MAPK and PI3K/AKT signaling pathways, resulting in increased cell proliferation and survival and suppressed differentiation. ³

Pediatric PTCs harboring RTK fusions are associated with increased invasive behavior and a decreased response to radioactive iodine (RAI) therapy. ^1,4 Preliminary clinical data suggest that selective kinase inhibitors can restore thyroid tumor differentiation and RAI avidity and consequently improve response to RAI therapy. ^{5 –8} Therefore, knowledge of the genetic alterations driving the disease is a valuable step to improve the prognosis of thyroid cancer patients and predict their responses to redifferentiation therapies followed by RAI treatment.

The spectrum of genetic alterations between pediatric and adult patients seems to be different, with gene fusions found in a higher percentage of pediatric tumors. ⁹ The genomic landscape of pediatric thyroid cancer is comparatively underexplored and comprehensive next-generation sequencing (NGS) studies can help identify novel genetic alterations driving the disease in children. Here, we identify a novel oncogenic fusion of the insulin-like growth factor 1 receptor gene (IGF1R) in thyroid cancer by whole-transcriptome sequencing. This finding expands the repertoire of targetable RTK fusions in pediatric PTC with potential clinical relevance in the prognostication and therapy for patients with tumors harboring this alteration.

Methods

Results

The patient was a 17-year-old female with no history of radiation exposure during childhood or medical history of thyroid disease. She was found to have a thyroid nodule on unrelated head and neck imaging, and on physical exam, there was a palpable left-sided thyroid mass. Thyroid ultrasound (US) revealed a unifocal solid, hypoechoic nodule in the left lobe with smooth margins, wider-than-tall shape, and no evidence of echogenic foci. The central neck lymph node compartments were not assessed, and there were no abnormal lymph nodes in the lateral neck on preoperative US. The patient underwent fine needle aspiration biopsy with the cytology classified as a follicular neoplasm (The Bethesda System for Reporting Thyroid Cytopathology IV).

A left lobectomy was performed with histopathology confirming an infiltrative follicular variant papillary thyroid carcinoma, 4 cm in greatest dimension, with capsular and extensive vascular invasion (more than four vessels) and no extrathyroidal extension. Completion thyroidectomy was then performed which revealed unremarkable thyroid. A diagnostic whole-body scan (DxWBS) showed 12.7% uptake in the neck with two distinct anterior foci, one consistent with a thyroid remnant and one consistent with a lymph node in level VI. The patient received 102 mCi (3,744MBq) of radioiodine with the posttherapy WBS showing the same distribution of the radioiodine as the DxWBS resulting in American Joint Committee on Cancer classification of T2N1aM0. TSH-stimulated thyroglobulin (TG) levels were 2.4 ng/ml at the time of RAI treatment (TSH 31.6 uIU/mL), and the anti-TG antibody level was below the lower limit of detection. Three months later, unstimulated Tg became undetectable (<0.1 ng/ml with TSH of 0.022 uIU/mL). Overall, the patient demonstrated an excellent response to treatment (no evidence of disease) and was in remission based on an undetectable Tg, and no evidence of anatomical disease by neck US 2 years after initial treatment, the date when the patient left our health care system.

The case had been previously screened for genetic alterations using the CHOP Solid Tumor Panel, interrogating SNVs, indels, and CNAs in 238 cancer genes and fusion oncogenes associated with 110 cancer genes. ¹⁰ This screening identified six variants of uncertain significance (Supplementary Table S1) and loss of chromosome 22. None of the most prevalent alterations in thyroid cancer were found, including mutations in BRAF, N-H-KRAS, DICER1, and fusions of RET, NTRK1-3, ALK, BRAF, and FGFR1-3. Because no apparent driver alterations were identified by targeted sequencing, we performed paired-end RNA sequencing and generated ∼80 million reads. Reads were aligned to the human genome (hg19) using STAR and search for gene fusions was carried out using STAR-Fusion. ¹¹ After stringent filtering for artifactual and/or readthrough fusions (Supplementary Table S2), we detected one putative pathogenic fusion: TG-IGF1R.

TG-IGF1R is an interchromosomal translocation t(8;15)(q24;q26) juxtaposing exons 1 to 47 of TG to exons 14 to 21 of IGF1R. The chimeric transcript generates an in-frame fusion protein containing the cysteine-rich domains and cholinesterase‐like domain of TG, joined to the transmembrane and protein tyrosine kinase domains of IGF1R (Fig. 1A). IGF1R has two different mRNA isoforms: NM_00875.5 and NM_001291858.2. They have slightly different sequences of exon 14, the first exon of IGF1R included in the fusion. NM_00875.5 has the addition of three nucleotides, leading to one additional amino acid (Alanine) in the resulting protein (Figs. 1A and 1B). Both isoforms are in-frame and retain the transmembrane and kinase domain of IGF1R. The presence of the TG-IGF1R fusion was validated by RT-PCR using primers for amplification of a 214 bp PCR product, from exon 47 of TG to exon 14 of IGF1R followed by Sanger sequencing (Fig. 1C and Fig. 1D). Tumor-adjacent nonneoplastic thyroid tissue was negative for the fusion, confirming this as a somatic event (Fig. 1C).

OPEN IN VIEWER

Fig. 1.

(A). Schematic overview of the TG-IGF1R fusion oncogene identified with chromosome positions, exon boundaries, protein domains, and nucleotide and protein sequences spanning the fusion junction. (B). Integrative Genomics Viewer plot of the TG-IGF1R fusion breakpoint and RNA-seq split reads supporting the occurrence of the fusion. RNA-seq reads spanning the fusion junction shows the occurrence of two isoforms derived from the alternative IGF1R transcripts: NM_000875.5 and NM_001291858.2. (C). RT-PCR using primers for exon 47 of TG and exon 14 of IGF1R (black arrows denote position of PCR primers). A PCR product with the expected size of 214 bp was detected in the tumor sample but not in the matched normal. GAPDH was used as internal control. (D). Sanger sequencing of RT-PCR product confirms the fusion of exon 47 of TG with exon 14 of IGF1R in the tumor sample. Sequence trace displayed is derived from sequencing traces from both forward primer (TG sequencing) and reverse primer (IGF1R sequencing). Individual traces for forward and reverse primers are shown in Supplementary FIG. S1. ^TM: transmembrane; Fwd: forward; Rev: reverse; L: ladder; N: normal; T: tumor; NTC: no template control.

TG is a thyroid-specific gene highly expressed in thyrocytes. ¹³ Since the expression level of the fusion is driven by the promoter of TG, the tumor harboring the fusion showed increased levels of IGF1R mRNA when compared with nonneoplastic thyroid (Fig. 2A). The expression was exclusively high for exons 14 to 21, the 3′ region of the gene included in the fusion, and coding for the kinase domain (Fig. 2B). The imbalanced expression between 5′ and 3′ regions of the IGF1R gene was verified in the tumor and the matched normal sample by quantitative PCR using primers for exons 10–11 (5′) and 17–18 (3′) of IGF1R. The 3′ region of the gene coding for the kinase domain showed a 12-fold increase compared with the 5′ region coding for the extracellular domain. Both regions have low mRNA expression in the nonneoplastic adjacent thyroid tissue, and no significant difference in expression is observed between 5′ and 3′ regions in the matched nonneoplastic sample (Fig. 2C).

OPEN IN VIEWER

FIG. 2.

(A). Expression levels of IGF1R mRNA in nonneoplastic thyroids (NT, n = 5) and tumor positive for IGF1R fusion (77T). (B). Expression levels of IGF1R for exons 1–13 and exons 14–21 revealed imbalanced expression of the 5′ and 3′ regions of the gene in tumor positive for the fusion. (C) qRT-PCR analysis of IGF1R mRNA levels upstream (exons 10–11) and downstream (exons 17–18) the fusion breakpoint for TG-IGF1R + tumor and matched nonneoplastic thyroid. Tumor shows increased expression of the 3′ region (exons 17–18) of IGF1R but not the 5′ region (exons 10–11). Matched normal shows low expression of IGF1R for both regions, 5′ and 3′. (D). FFPM values for TG-IGF1R (NM000875.5), TG-IGF1R (NM001291858.2), and IGF1R-TG shows predominant expression of TG-IGF1R fusion over IGF1R-TG. (E). Immunostaining of papillary thyroid cancer positive for TG–IGF1R showing strong cytoplasmic immunoreactivity with anti-IGF1R antibody (phospho-Y1161) in the tumor and lack of staining of adjacent normal thyroid tissue (Magnification: x100; Scale bar = 200 μm). (F). Heatmap displaying increased expression of selected genes related to MAPK output in 5 nonneoplastic thyroid tissues vs. tumor positive for IGF1R fusion. Red: upregulated genes; Blue: downregulated genes. (G). qRT-PCR analysis of NIS mRNA levels for tumor harboring the IGF1R fusion (77T) and matched normal (77N). NT: nonneoplastic tissue; FFPM: Fusion fragments per million total RNA-seq fragments.

Fusion fragments per million total RNA-seq fragments (FFPM) are a normalized measure of the quantity of RNA-seq fragments supporting the fusion event and can be used as an estimate for the fusion expression level. ¹² Both isoforms of TG-IGF1R are highly expressed, but the FFPM values for isoform NM_00875.5 are higher than NM_001291858.2 (7.5 vs. 5.3). The reciprocal fusion IGF1R‐TG, juxtaposing exons 1–13 of IGF1R to exon 48 of TG, was also detected. However, the FFPM value was around 60 times lower than the average FFPM observed for TG-IGF1R isoforms (Fig. 2D).

To determine whether TG-IGF1R has increased kinase activity, we assessed phosphorylation levels of tyrosine residue 1161 by immunohistochemistry. We observed strong cytoplasmic immunostaining for phospho-IGF1R (Y1161) in the tumor cells, indicating autophosphorylation of the kinase domain (Fig. 2E). Adjacent nonneoplastic tissue showed negative to weak immunostaining, mostly nuclear. Consistent with the activation of the IGF1R pathway and its role on MAPK activation, we observed increased expression of genes of the MAPK transcriptional output (Fig. 2F), indicating increased pathway activity. Activation of the MAPK pathway is associated with diminished thyrocyte differentiation. ¹⁴ Therefore, we determined mRNA expression of 16 differentiation genes defined as the thyroid differentiation score in The Cancer Genome Atlas study in the tumor and nonneoplastic thyroid samples. Interestingly, we observed a striking reduction of SLC5A5 (NIS) expression compared with nonneoplastic thyroid, while the expression of other thyroid differentiation genes was fairly similar between samples, including transcription factors that are regulators of NIS mRNA expression (Supplementary Fig. S2). The reduction in NIS mRNA levels was further validated by quantitative PCR in the tumor sample harboring the fusion vs. matched nonneoplastic tissue (Fig. 2G). We confirmed a striking reduction of NIS expression in TG-IGF1R + tumor cells, which has important therapeutic implications for patients with thyroid cancer.

Discussion

IGF1R is a central member in the IGF family of signaling molecules, a group of proteins that serve as key regulators of energy metabolism and growth in both normal as well as neoplastic tissue. ¹⁵ However, although IGF1R-mediated proliferation is a hallmark of cancer, activating mutations and fusion oncogenes of the IGF1R gene are rarely reported in human cancer. In this study, we identified TG-IGF1R as a novel fusion oncogene in pediatric thyroid cancer through transcriptome sequencing. This is the first report of IGF1R fusion in thyroid cancer, but chimeric fusions involving IGF1R with different 5′ partner genes have been previously reported in single cases of inflammatory myofibroblastic tumor, gastrointestinal stromal tumor, lung, and breast cancer. ^{16 –18} Like the one found here, these fusions are predicted to be pathogenic as they are in-frame and retain the kinase domain of IGF1R.

The expression of TG-IGF1R fusion is driven by the TG promoter, a thyroid-specific gene essential for thyroid hormone synthesis, and one of the most expressed transcripts in the gland. ¹³ Consequently, the chimeric fusion shows abnormally high mRNA expression when compared with nonneoplastic thyroid. TG has been previously identified as 5′ partner in other fusion oncogenes reported in thyroid cancer: TG-FGFR1, TG-THADA, and TG-PBF. ^2,19,20 The resulting fusion protein contains the cholinesterase‐like domain of TG, essential in TG trafficking for thyroid hormone synthesis, fused to the transmembrane and protein tyrosine kinase domains of IGF1R. The cholinesterase‐like domain of TG has been previously shown to have dimerization properties. ²¹ In addition, the oncoprotein lacks the autoinhibitory extracellular domain, which inhibits kinase activity in the absence of ligand. In conjunction, fusion overexpression, deletion of the extracellular autoinhibitory domain, and ability for dimerization act synergistically for robust ligand-independent constitutive activation of the IGF1R. Autophosphorylation of Tyr1161, Tyr1165, and Tyr1166 in the kinase activation loop is critical for kinase activity and biological function. ²² Indeed, tumor cells displayed robust immunostaining for phosphorylation of residue Tyr1161, consistent with autophosphorylation of the TG-IGF1R kinase domain. The increased expression of genes from the MAPK transcriptional output observed in the tumor further supports kinase activation.

Additional comprehensive NGS studies are necessary to understand the prevalence of these fusions in thyroid cancer. The lack of identification of these alterations in previous studies using RNA sequencing in pediatric and adult thyroid cancer indicates that IGF1R fusion is a rare event. ^2,23,24 However, we cannot exclude the possibility that this alteration is underreported in thyroid cancer as IGF1R fusion is not included in different targeted sequencing panels. In fact, this fusion oncogene was missed by our initial genetic screening using the CHOP CSTP panel and would not have been detected by thyroid-specific sequencing panels widely used in the clinical practice such as ThyroSeq v.3, ThyGeNext, and Afirma Xpression Atlas. ^{25 –27}

In addition to the alteration presented here, there is evidence supporting that the IGF1R pathway is indirectly activated in thyroid cancer. THADA fusions show strong overexpression of IGF2BP3 and posttranscriptional upregulation of IGF2, resulting in stimulation of IGF1R signaling in thyroid tumors. ¹⁹ Moreover, there is strong evidence indicating that ETV6-NTRK3 fusion requires IGF1R signaling for transformation. ²⁸ Most importantly, IGF1R inhibitors are available for the treatment of solid tumors, including monoclonal antibodies and selective kinase inhibitors. ²⁹ The use of these agents is mainly effective when the kinase target is constitutively activated by gene fusions or mutations and patients can be stratified through molecular profiling. The lack of predictive markers was one of the main causes for the disappointing results in previous clinical studies of IGF1R inhibitors. ³⁰ The discovery of TG-IGF1R fusion as a genetic driver in thyroid cancer, especially if proven to be a recurrent alteration, uncover an attractive predictive marker for selective incorporation of IGF1R inhibitors into clinical practice.

Overall, our data show for the first time a novel driver genetic alteration involving IGF1R in thyroid cancer. Additional studies are needed to better understand whether this is a recurrent alteration in pediatric and adult thyroid tumors, as well as delineation of the clinicopathological characteristics of these tumors. Even if a rare event, this finding supports on-going efforts to incorporate somatic genetic profiling to individualize care using targeted therapeutic agents. Therefore, the incorporation of IGF1R fusion screening in the next versions of thyroid-specific targeted NGS panels may be beneficial to thyroid cancer patients and should become part of currently available thyroid-specific sequencing panels.