ROS1 Rearrangement in Thyroid Cancer

Introduction

Subsets of thyroid cancer have been shown to harbor recurrent gene fusions involving BRAF, RET, PPARG, NTRK1, NTRK3, ALK, RAF1, FGFR2, and THADA (1 –3). Previous large-scale genotyping studies in papillary thyroid carcinoma provided insights into the risk categories associated with these gene fusions (2 –4). Specifically, these studies showed that the molecular subgroup of thyroid cancer harboring gene fusions includes patients with a younger age at diagnosis and a lower risk of recurrence, suggesting that thyroid cancers with gene fusions belong mostly to the low- and intermediate risk categories (2). Additionally, insight into the molecular alterations in thyroid cancer has led to the utilization of targeted therapies, such as the multikinase inhibitor cabozantinib, for patients with advanced disease and trials are in progress (5). However, most genotyping data have been performed on the more common histologic variants of thyroid cancer, including classical papillary, follicular, and tall-cell variant thyroid cancer (1,5), and the presence of novel gene fusions in less common histotypes remains to be characterized.

Patient

A 24-year-old female athlete initially presented to her primary care physician for an enlarging “lump” in her left anterior neck present for approximately one year. She was otherwise healthy with a family history only notable for a great aunt with thyroid cancer. Computed tomography (CT) imaging of her neck showed a 3.5 cm vascular enhancing mass in the anterior left neck, inferior to the hyoid bone and extending over the left thyroid cartilage. An ultrasound scan showed an otherwise normal thyroid gland, and a follow-up magnetic resonance angiogram of her neck showed a 2.5 cm hypervascular mass just left of the midline, concerning for a venous malformation or a group of matted lymph nodes. The patient underwent an open biopsy at another institution, with the operative report significant for adherence of the mass to the underlying thyroid cartilage.

Histopathological examination of the biopsy specimen, reviewed at the authors’ institution, showed a 1.6 cm focus of papillary thyroid carcinoma with solid and follicular architecture without presence of frank nodal elements. On presentation, she was noted to have a 3 cm surgical scar overlying the left thyroid lamina with underlying induration but no palpable surrounding cervical lymphadenopathy. Evaluation with flexible fiberoptic laryngoscopy showed normal vocal cord motion and a normal-appearing larynx. Thyroid function tests were normal. CT and ultrasound imaging obtained approximately one month after her initial surgery showed a 3.3 cm irregular, lobulated, hypervascular mass arising from the pyramidal lobe of the thyroid gland and extending through the left paramedian cricothyroid membrane under the anterior one-third of the left vocal cord (Fig. 1A).

OPEN IN VIEWER

FIG. 1.

ROS1-rearranged thyroid cancer. (A) Axial computed tomography (CT) image demonstrating involvement of the tumor through the cricothyroid membrane. (B) Wedge resection of the left thyroid cartilage for laryngeal invasion with demonstration of the underlying musculature of the left true vocal cord (dashed square). Midline is depicted with identification of the thyroid cartilage notch (asterisk) and midline cricoid cartilage (arrow). (C) Histologic image detailing solid nests of cells separated by fibrous septa (hematoxylin and eosin [H&E], 40×). (D) High-power histologic image detailing the classic nuclear features of papillary thyroid carcinoma and colloid production (H&E, 40×). (E) Fusion sequence along with genomic exon structure of the fusion and predicted CCDC30-ROS1 fusion protein. (F) Fluorescence in situ hybridization of interphase tumor nuclei showing isolated 3′ (red)-probe signals in the tumor cells.

The patient underwent a total thyroidectomy with central and left lateral neck dissection with intraoperative nerve monitoring. At the time of surgery, it was confirmed that an irregular mass was arising from the pyramidal lobe of the thyroid with focal invasion into the lower margin of the left thyroid lamina and the upper margin of the left cricothyroid membrane. The mass was inseparable from the overlying strap musculature, necessitating an en bloc resection, including the overlying strap musculature, and a wedge resection of the left thyroid lamina and cricothyroid membrane (Fig. 1B). The laryngeal defect resulted in a unique view of the lateral surface of the anterior left vocal cord during intraoperative functional assessment. Using electrical stimulation of the vagus nerve, it was possible to visualize directly the endolaryngeal contractions as intraoperative evidence of normal vocal function (Supplementary Video S1; Supplementary Data are available online at www.liebertpub.com/thy). The patient did well postoperatively and was discharged on the day after surgery with no airway or vocal disturbances.

Surgical pathology revealed a 3.7 cm solid-variant papillary thyroid carcinoma invasive into skeletal muscle and into the cricothyroid membrane (pT4a). Microscopically, the tumor was largely formed by solid nests of cells separated by streaming fibrous septa (Fig. 1C) with focal cytoplasmic clearing, nodular invasion, and rare true papillae. However, focal areas of the tumor demonstrated classic follicular differentiation with colloid production (Fig. 1D). Surgical margins were negative. An additional 2 mm focus of tumor with the same morphologic features was found in the right thyroid lobe. Metastases were seen in 6/14 central compartment nodes and 2/6 left lateral compartment nodes with extranodal extension and perinodal lymphatic involvement (pN1b). BRAF^V600E -specific immunohistochemistry was negative. The patient was staged according to her age (<45 years) as stage I. The solid-variant histotype and locally aggressive behavior triggered oncologic genotyping. Genotyping was conducted by extracting total nucleic acids from formalin-fixed paraffin embedded tumor tissue and performing massive parallel sequencing. Two assays were employed: one for mutation hotspots in 39 cancer genes (targeted genotyping) and the other for detection of gene fusions employing targeted anchored multiplex polymerase chain reaction assessing gene fusion transcripts involving 52 genes (6) (for details, see Supplementary Tables S1 and S2).

Targeted genotyping did not reveal a panel-specific point mutation. However, gene fusion assessment demonstrated a fusion involving ROS1. Mapping of the fusion and sequence analysis identified CCDC30 as the ROS1 fusion partner (Fig. 1E; fusion constructed with IBS) (7). Sequence-based prediction of the fusion product revealed exons 1–10 of the coiled-coil domain containing 30 (CCDC30) gene fused to the N-terminal ROS1 kinase domain, which included exons 36–43. This rearrangement creates CCDC30 as the postulated driver of ROS1. Sequence and domain analysis revealed that the coiled-coil domain of CCDC30 was present in the fusion that is significant, as these motifs have been shown to be involved in ligand-independent dimerization (8). The ROS1 fusion was also assessed by fluorescence in situ hybridization of interphase tumor nuclei in a paraffin-embedded tumor tissue section. Briefly, in addition to a normal merged red and green signal (unrearranged allele), the tumor cells showed isolated 3′ (red)-probe signals, confirming rearrangement of the ROS1 locus by an orthogonal, tumor cell specific method (Fig. 1F).

Review of the clinical database demonstrated no prior ROS1 gene fusion in 71 thyroid carcinomas assessed by the fusion assay (1/72 = 1.4%; status February 12, 2016). In addition, all currently reported ROS1 fusions in >7000 samples were reviewed (The Cancer Genome Atlas), and no prior reports of ROS1-CCDC30, ROS1 fusions, or ROS1 aberrations in thyroid cancer were identified in either database.

The patient was re-evaluated on postoperative day 14 and was found to be asymptomatic with a normal respiratory and vocal function. Flexible fiberoptic laryngoscopy demonstrated a normal larynx with fully mobile bilateral vocal cords and no edema or deformation. Subsequently, she underwent multidisciplinary evaluation with specialists from endocrinology, oncology, and radiation oncology. The histopathology and stage of disease justified application of 150 mCi of thyrogen-stimulated radioactive ¹³¹I without external beam radiation. A post-treatment whole-body iodine scan with single-photon emission computed tomography/CT demonstrated minimal iodine-avid tissue in the thyroid bed and a question of focal iodine uptake in the chest. However, a follow-up positron emission tomography/CT was negative for any evidence of metastatic disease.

Discussion

Herein, the first case of a ROS1 rearrangement in a solid-variant papillary thyroid carcinoma with a locally aggressive presentation is reported. Although the direct biological, clinical, or therapeutic implications in this patient are not known (i.e., follow-up is limited, and she remains stage I due to her age), this case with a novel gene fusion including ROS1 is considered notable for several reasons.

First, the data indicate that the overall frequency of ROS1 rearrangements in thyroid cancer could be as high as 1% because the lack of ROS1 abnormalities in other studies may be directly related to challenges in detecting novel gene fusions (i.e., next-generation sequencing technology, bioinformatics, analytical pipelines). Furthermore, most genotyping data are derived from more common histologic variants (5). Second, the case argues that genotyping of less common thyroid cancer histotypes, such as the solid variant, which comprises <3% of all papillary thyroid carcinomas (9), may reveal additional actionable findings (10,11). Third, the fusion partner (CCDC30) has not been reported in any of the databases that were assessed, and it remains to be determined whether larger numbers of thyroid cancer fusions will also have gene fusions harboring coiled-coil domains (see above). Fourth, from a diagnostic perspective, this case indicates that ROS1 fusions should not be confused with a lung metastasis but may occur in primary thyroid cancer. Fifth, fusions involving ROS1 may carry therapeutic implications. In lung cancer, the detection of ROS1 rearrangements has direct treatment implications by means of choice of kinase inhibitor (i.e., crizotinib) (12). While ROS1 rearrangement is considered an indicator of better overall survival in lung cancer (13), direct extrapolation to the comparatively more benign course of thyroid cancer is not possible. However, in light of the locally aggressive tumor with nodal metastases present in this patient, the utmost diligence in active surveillance is currently being pursued. In contrast to this case, which has the high-risk feature of macrosopic extrathyroidal extension (4), prior reports have suggested that gene fusions in thyroid cancer are associated with low- and intermediate-risk categories, and it is therefore speculated that ROS1 fusion positive cases may have a different disease biology. A review of additional patients with currently undetermined tumor genetics that share similar tumor histomorphology and clinical presentation is needed, especially in light of the availability of ROS1-targeted therapies.