NSD3::NUTM1 fusion evidenced on RNA sequencing in poorly differentiated thyroid cancer: a report of two cases

Abstract

Background

Nuclear protein in testis (NUT) carcinomas are aggressive, poorly or undifferentiated cancers, generally arising from midline structures. This subtype of squamous cell carcinoma is rare and has a poor prognosis. NUT cancers are defined by NUTM1 fusions. Rearrangements of the NUTM1 gene have rarely been described in primary thyroid cancer and are mainly reported in patients ≤45 years old. NSD3::NUTM1 translocation is the most common NUTM1 fusion transcript reported in thyroid cancer. As they are very infrequent, NUTM1 fusions are not routinely sought in poorly differentiated thyroid cancer (PDTC) or anaplastic thyroid cancer (ATC).

Case presentation

We report two PDTC patients >65 years old with locally advanced disease and ¹⁸FDG-avid distant metastases. NSD3::NUTM1 translocation was evidenced in both patients by RNA sequencing using the next-generation sequencing panel of our institution.

Conclusion

We suggest including the search for NUTM1 fusions in the RNA sequencing panel for advanced and refractory thyroid cancers.

Established facts

• Nuclear protein in testis (NUT) carcinomas are rare, aggressive, poorly or undifferentiated cancers with a poor prognosis, generally arising from midline structures in young patients. NUT cancers are defined by NUTM1 fusions that can be evidenced by immunohistochemistry (IHC), fluorescence in situ hybridization, or RNA sequencing.

• NUTM1 fusions have rarely been described in primary thyroid cancer, usually in patients ≤45 years old. NSD3::NUTM1 translocation is the most common NUTM1 fusion transcript reported in thyroid cancer.

• As they are very infrequent, NUTM1 fusions are not routinely sought in PDTC or ATC.

Novel insights

• We report two patients >65 years old with locally invasive refractory PDTC and NSD3::NUTM1 translocation.

• NSD3::NUTM1 translocation might be more systematically sought in locally advanced PDTC/ATC with no point DNA mutations or other rearrangements.

• This strategy would better identify these patients, whose pathological diagnosis and clinical management may be challenging.

Introduction

Nuclear protein in testis (NUT) carcinomas (NC) are aggressive, poorly or undifferentiated cancers, generally arising from midline structures of the thorax or head and neck region (1). This subtype of squamous cell carcinoma (SCC) is rare and mainly occurs in young patients. NC have a very poor prognosis with a median survival of roughly 7 months (2, 3). NC are defined by NUTM1 fusions (1). Rearrangement of the NUTM1 gene involves fusion with the BRD4 gene in 70–80% of cases (3, 4) and less frequently with BRD3 or NSD3 fusion partners (5, 6). NUTM1 fusions can be evidenced using immunohistochemistry (IHC), fluorescence in situ hybridization, or RNA sequencing (2). To increase awareness of this uncommon disease, an international NC registry exists online (https://nc-registry.org/).

NUTM1-rearranged carcinomas rarely occur in the thyroid gland. Nineteen cases of thyroid cancers with NUTM1 fusion have been described, mainly in patients ≤45 years old (7, 8, 9, 10, 11, 12). NSD3::NUTM1 translocation is the most frequent NUTM1 fusion transcript reported in thyroid cancer (7, 9, 10, 11, 12), while BRD4::NUTM1 translocation is less frequent (8, 10).

Pathological diagnosis may be challenging. When squamous differentiation is observed, it can be difficult to distinguish NC from a squamoid anaplastic thyroid cancer (ATC) (10). In addition, NC of the thyroid can present without typical morphology, mimicking other primary thyroid malignancies, such as poorly differentiated thyroid cancer (PDTC) (10).

NC are genetically defined by a chromosomal rearrangement of the NUTM1 gene on chromosome 15 (13). NSD3 is a histone methyltransferase (4). The NSD3::NUTM1 fusion protein acts similarly to the BRD4::NUTM1 fusion protein that interacts with chromatin, leading to an oncogenic complex that drives the transcription of oncogenic genes, such as MYC and p63 (14, 15, 16, 17, 18). The NUT fusion oncoprotein appears to be the predominant oncogenic driver in NC (6, 19).

We report two PDTC patients with refractory disease, in whom next-generation sequencing (NGS) with our in-house RNA sequencing panel revealed NSD3::NUTM1 rearrangement.

Case reports

Case 1

A 69-year-old man with goiter and a left thyroid nodule suggestive of malignancy (Bethesda V cytology) was scheduled for thyroid surgery in October 2020. Owing to local tumor extension evidenced during surgery, only a biopsy could initially be performed. Pathology showed PDTC with gross muscle invasion. A sternotomy was planned to treat mediastinal tumor extension. However, only partial left thyroid lobectomy was feasible (gross extension). Pathology revealed a 95 mm PDTC with a trabecular, insular solid component (70%), a trabecular papillary component (30%), and gross muscle invasion (Fig. 1, panel A). The tumor was scored pT4aNxMx. One month after surgery, ¹⁸fluorodeoxyglucose (¹⁸FDG) positron emission tomography with computed tomography (PET/CT) depicted a suspicious left laterotracheal focus consistent with residual extra-thyroidal extension. Left neck external beam radiation therapy was performed from March to May 2021 (VMAT 6 MV of 70 Gy in 35 fractions of 2 Gy, 5 fractions per week with a Varian Clinac Rapid’Arc® LINAC system). The patient was followed up by CT scans. While NGS did not demonstrate any point mutation, particularly on BRAF and RAS genes, RNA sequencing revealed a fusion transcript NSD3::NUTM1 (Fig. 2, panel A). For this report, immunostaining with an anti-NUT antibody was performed and was negative. In February 2022, CT scan showed a 5 mm right lung nodule which progressively increased to 8 mm in August 2022 and 10 mm in January 2023 (Fig. 3, panel A). The lung nodule was ¹⁸FDG-avid on PET/CT in September 2022 (SUVmax = 4.2). ¹⁸FDG uptake in the lung nodule increased in March 2023 (SUVmax = 6.7), and the nodule now measured 11 mm (Fig. 3, panel B). Serum Tg level was 10 ng/mL without thyroglobulin antibodies (TgAb). Lung biopsy in May 2023 confirmed metastasis from PDTC, which was successfully treated by radiofrequency ablation (RFA) in June 2023 (Fig. 3, panels C & D). Furthermore, because of the growth of a left kidney tumor with high ¹⁸FDG uptake, the patient underwent nephroureterectomy in December 2022, revealing a high-grade non-invasive papillary urothelial carcinoma. At the last follow-up visit (March 2025), there was no functional or structural imaging evidence of residual/recurrent disease, including PDTC or kidney cancer.

Figure 1.

Pathology features of patients 1 and 2. In patient 1, pathology revealed a 95 mm PDTC with a trabecular, insular solid component (70%) and a trabecular papillary component (30%). (A) Hematoxylin and eosin–stained sections (HES) from patient 1; magnification 20×). In patient 2, biopsy revealed a PDTC with an oncocytic component. (B) HES from patient 2; magnification 20×).

Figure 2.

Structure of fusion transcripts NSD3::NUTM1 for patient 1 (A) and patient 2 (B). Clear boxes at each end are the 5′ and 3′ untranslated regions. Black arrows represent the start codons, and the lollipops the stop codon. Bold boxes are the coding exons, and their shape illustrates the coding frame.

Figure 3.

Lung imaging in patient 1. In January 2023, CT scan showed a 10 mm right lung nodule (A, arrow) with high ¹⁸FDG uptake (SUVmax = 7) on PET/CT performed 2 months later (B, arrow). Nine months after RFA, CT scan showed that the lung nodule had decreased to 6 mm (C, arrow) without ¹⁸FDG uptake on PET/CT (D, arrow).

Case 2

A 73-year-old man with a history of childhood poliomyelitis and rheumatoid arthritis was referred for rapid growth of a cervical mass in August 2022. Neck US confirmed the presence of an enlarged thyroid mass with right internal jugular vein thrombosis. CT scan showed a retrosternal thyroid mass that extended into the superior mediastinum, infiltrating the brachiocephalic veins and the right internal jugular vein. Fine-needle aspiration cytology yielded a few necrotic cells, presumably of tumoral origin. The biopsy performed in September 2022 revealed a PDTC with an oncocytic component (Fig. 1, panel B). IHC for BRAF mutation was negative. In October 2022, ¹⁸FDG PET/CT showed that the thyroid tumor was highly hypermetabolic (SUVmax = 69) with cervical and retrosternal extension (Fig. 4, panels A & C). After discussion during the ENDOCAN-TUTHYREF (French Refractory Thyroid Tumors Network) tumor board, external beam radiation therapy was administered from November to December 2022 (VMAT 6MV of 70 Gy in 35 fractions of 2 Gy, with a simultaneous integrated boost of 73.50 Gy, 2.10 Gy per fraction, 5 fractions per week with a Varian Halcyon^® LINAC system). While NGS did not reveal any point mutation, particularly on BRAF and RAS genes, RNA sequencing revealed an NSD3::NUTM1 fusion (Fig. 2, panel B). IHC for NUT was retrospectively performed on two different biopsy specimens and the results were negative. In April 2023, ¹⁸FDG PET/CT showed a partial metabolic response (SUVmax = 34) (Fig. 4, panels B & D). ¹⁸FDG uptake was stable in October 2023 (SUVmax = 35). In April 2024, after a 1-year clinical and imaging follow-up, CT scan and PET/CT showed disease progression with the occurrence of new lesions with high ¹⁸FDG uptake, i.e. a retrosternal lymphadenopathy, a right lateral compartment cervical lymphadenopathy, and a bone focus on the sternum (Fig. 5). External radiation therapy was performed in August 2024 on the retrosternal lymphadenopathy and the sternum by a stereotactic technique to spare the already irradiated tissues at the cranial border of the target volume (50 Gy in 5 fractions with an Accuray CyberKnife^® robotic LINAC). In October 2024, the patient presented with tracheal invasion necessitating a tracheostomy. ¹⁸FDG PET/CT depicted local progression (Fig. 6, panel A). In the absence of other effective treatments, and despite the risk of bleeding, and with the patient’s agreement, systemic therapy with lenvatinib, a multikinase inhibitor, was started at 10 mg per day. Three months after the initiation of lenvatinib, PET/CT showed a decrease in the ¹⁸FDG-avid thyroid tumor located in the neck and upper thorax, suggesting a partial metabolic response (Fig. 6, panel B). However, the patient died a few days later from complications related to hemorrhage and sepsis.

Figure 4.

¹⁸FDG PET/CT imaging in patient 2 before and after external radiation therapy. In October 2022, pre-therapeutic PET/CT showed very high ¹⁸FDG uptake (SUVmax = 69) in the thyroid tumor with cervical and retrosternal extension (A, maximum intensity projection (MIP) image; C, fused axial image). Four months after the completion of intensity-modulated radiation therapy, PET/CT showed a decrease in the ¹⁸FDG-avid thyroid tumor located in the neck and upper thorax (SUVmax = 34), suggesting a partial metabolic response (B, MIP image; D, fused axial image).

Figure 5.

¹⁸FDG PET/CT imaging in patient 2 at progression. In April 2024, PET/CT revealed new ¹⁸FDG foci, i.e. bone uptake in the sternum (SUVmax = 21) (A, fused axial image), a 10 mm right lateral compartment cervical lymphadenopathy (SUVmax = 22) (B), and a 10 mm retrosternal lymphadenopathy (SUVmax = 15) (C). ¹⁸FDG uptake in the thyroid tumor (SUVmax = 55) (D) was increased as compared to the PET/CT performed 6 months earlier.

Figure 6.

¹⁸FDG PET/CT imaging in patient 2 showing partial metabolic response 3 months after initiation of systemic therapy. In October 2024, PET/CT showed high ¹⁸FDG uptake (SUVmax = 55) in the thyroid tumor (A, MIP image, blue arrow), with retrosternal lymphadenopathy (SUVmax = 13, lower red arrow), sternum metastasis (SUVmax = 11, green arrow), and another retrosternal lymphadenopathy (SUVmax = 8, upper red arrow). Lenvatinib was started immediately afterward. Three months after the initiation of lenvatinib, PET/CT showed a decrease (SUVmax = 17) in the ¹⁸FDG-avid thyroid tumor located in the neck and upper thorax (B, MIP image, blue arrow), and the two mediastinal lymphadenopathies had disappeared, suggesting a partial metabolic response. The focus of the sternum was stable (SUVmax = 11, green arrow).

Discussion

These two patients had thyroid cancer harboring the NSD3::NUTM1 fusion transcript, which is a very uncommon molecular alteration. Interestingly, both patients had several similar features.

First, they were >65 years old, whereas NC are usually reported in young patients. Of the 19 thyroid cancer patients with NUTM1 fusions previously reported, 17 (89%) were ≤45 years old (7, 8, 9, 10, 11, 12), and two were over 70 years (10, 11) (Supplementary Table 1 (see section on Supplementary materials given at the end of the article)).

Second, both patients had locally advanced thyroid cancer that prevented initial total thyroidectomy and necessitated external beam radiation therapy. This clinical presentation is common in thyroid cancer patients with NUTM1 fusions.

Third, both had PDTC, a histology observed in six (32%) of the 19 thyroid cancer patients with NUTM1 fusions (Supplementary Table 1) (10, 11). The other pathological diagnoses were NC (n = 6), SCC (n = 3), SCC/ATC (n = 1), ATC (n = 1), papillary thyroid cancer (n = 1), and undifferentiated cancer (n = 1) (Supplementary Table 1) (7, 8, 9, 10). Furthermore, pathological diagnosis may be difficult (20). Barletta et al. reported that about 40% of the patients in their cohort presented with histological features that were not typical of NC (10). When thyrocyte differentiation is present in non-squamous carcinoma, the diagnosis of NUTM1-rearranged thyroid carcinoma may be even more challenging. Therefore, the present cases suggest the need to search for NUTM1 fusions in PDTC without somatic mutations, particularly when a locally aggressive presentation is observed.

Fourth, IHC for NUT was negative in both patients. Haack et al. showed that using a cutoff of ≥50%, NUT IHC had a sensitivity of 87% (21). Hence, if NC is suspected but IHC for NUT is negative, NGS should be performed. Moreover, NGS testing is informative even when IHC is positive to identify which fusion partner is involved, because non-thoracic NC with NSD3::NUTM1 and BRD3::NUTM1 fusions are associated with better survival than non-thoracic NC with BRD4::NUTM1 fusions (3).

Fifth, NSD3::NUTM1 transcript fusion was revealed by RNA sequencing in both patients, as NUTM1 fusions are sought in the NGS panel of our institution. The list of the fusion genes included in our NGS panel is presented in Supplementary Table 2. RNA sequencing was performed because no somatic point mutations were evidenced (i.e. in our in-house NGS panel, BRAF, KRAS, HRAS, NRAS, and RET). According to Barletta et al., NC-like thyroid tumors present with typical NC morphology, whereas non-NC-like thyroid tumors are defined by the presence of specific histologic features without classical NC morphology and are generally positive for PAX8 and TTF-1 (10). Although NC-like and non-NC-like thyroid tumors might arise from different cell progenitors, somatic point mutations are absent in NC, and NUTM1 rearrangement is usually the single event that leads to neoplastic transformation (19). Since IHC can be negative, as in the present cases, and non-NC-like thyroid tumors can be observed, we suggest implementing NUTM1 transcript fusions in the RNA sequencing panel for thyroid cancers. Our policy was to include all actionable molecular alterations in the NGS panel, including BRD3, BRD4, and NUTM1. Indeed, targeted bromodomain and extraterminal protein inhibitor (BET inhibitor) therapy exists for patients with NUTM1 fusions (22). This makes it possible to highlight NSD3::NUTM1 transcript fusion in these PDTC patients.

Finally, patient 1 is alive 4 years after the initial diagnosis and has not to date received any systemic therapies such as tyrosine kinase inhibitors. This is in line with previous observations reporting that i) NUTM1-rearranged cancers arising in the thyroid gland have a better prognosis than other non-thyroid NC, and ii) patients with non-NC-like thyroid tumors have a better outcome than patients with NC-like thyroid tumors (10). Conversely, patient 2 received lenvatinib, which was initiated 2 years after initial management because of disease progression. As immunotherapy with PD-1/PD-L1 checkpoint inhibitors was not possible because programmed cell death-ligand 1 (PD-L1) expression was negative, and in the absence of any clinical trial, no other treatment option was possible. While the first ¹⁸FDG PET/CT assessment after systemic therapy initiation showed a partial metabolic response, infectious bleeding complications led to the patient’s death. BET inhibitors offer therapeutic perspectives (23, 24, 25, 26, 27). Molibresib, a BET inhibitor, was administered in two patients with thyroid NC but was rapidly withdrawn because of adverse effects (7, 9). Other BET inhibitors need to be assessed in NUTM1-rearranged thyroid cancer. Given these rapid-onset adverse effects and the lack of data with BET inhibitor therapy in thyroid cancer with NUTM1 fusions, our observation suggests that lenvatinib might be used as a first-line treatment before considering other systemic therapies.

In conclusion, these two cases show that locally advanced thyroid cancer with NUTM1 fusions is likely not exceptional. We recommend the more systematic search for this molecular alteration in patients with PDTC/ATC, and more generally refractory thyroid cancers without point mutations, by including the NSD3::NUTM1 fusion transcript in the RNA sequencing panel. This would allow better identification of these patients, whose pathological diagnosis and clinical management may be challenging.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding

This research did not receive any specific grant from any funding agency in the public, commercial, or not-for-profit sector.

Patient consent

Written informed consent was obtained from both patients for publication of the submitted article and accompanying images.