Keywords

1 Definition

NC (aka NUT midline carcinoma) is a poorly differentiated carcinoma with rearrangement of the NUTM1 (aka NUT) gene (French and den Bakker 2015).

2 Clinical Features

NC is possibly the most aggressive solid tumor known, having a median survival of 6.7–10.1 months (Bauer et al. 2012; Jung et al. 2019). The majority of tumors arise centrally near the airways in the head and neck (39%) and thorax (50%) (Bauer et al. 2012), giving the tumor’s original name NUT “midline” carcinoma; however, with the increasing diagnosis of this entity, cases are being reported in a broad range of non-midline sites, including kidney (Bishop et al. 2016; Sirohi et al. 2018; Zhu et al. 2019), thyroid (Landa et al. 2016), soft tissue (Dickson et al. 2018), salivary glands (Agaimy et al. 2018; Seim et al. 2017; Vulsteke et al. 2016; Ziai et al. 2010; den Bakker et al. 2009), pancreas (Shehata et al. 2010), and bladder (French et al. 2004). In fact, non-thoracic, non-head and neck primary sites comprise ~10% of NCs (Bauer et al. 2012; Chau et al. 2019). Radiologic (CT or MRI) features are not specific, typically revealing a large, invasive, heterogenous mass that often confluently involves local lymph nodes (Polsani et al. 2012; Sholl et al. 2015) (Fig. 22.1). Metastases are common at presentation (51%) (Bauer et al. 2012), including to bone and other solid organs (Sholl et al. 2015). The early presentation of hematogenous metastases and frequent origin within the lung or mediastinum can mimic that of small cell carcinoma; however, it progresses more rapidly and often effects younger individuals.

Fig. 22.1
figure 1

Typical appearance of NC on thoracic CT scan reveals a large, heterogenous mass with extensive local invasion and lymph node involvement

Full size image

3 Molecular Genetics

In all cases of NC, NUTM1 is rearranged in a chromosomal translocation, most commonly (~70%) (Bauer et al. 2012; Chau et al. 2014) with BRD4, forming a BRD4-NUTM1 fusion resulting from a t(15;19)(q14;p13.1). In almost every case, the breakpoint is upstream of exon 3 (transcript variant 1) of NUTM1, fusing this exon with the partner gene (Fig. 22.2). The most common alternative partner genes are BRD3 (~15%) (Chau et al. 2019), which is highly homologous to BRD4, and NSD3 (~6%) (Chau et al. 2019). Rare variant partners include ZNF532 and ZNF592. Interestingly, all of these encoded NUTM1 fusion partners (BRD3, BRD4, NSD3, ZNF532, ZNF592) interact with BRD4 and are critical components of the BRD4-NUT oncogenic complex (Alekseyenko et al. 2017). Thus, the fusion of any of these proteins to NUTM1 leads to its association with BRD4 and is sufficient to recapitulate the function of the canonical BRD4-NUT oncogenic complex.

Fig. 22.2
figure 2

Schematic of NUTM1 fusions reported in NC

Full size image

Molecular pathology utilizing next generation sequencing (NGS) has uncovered numerous novel NUTM1-fusions in a variety of tumors with uncertain relationship to NC. These NUTM1-fusion partners include CIC (Schaefer et al. 2018), BCORL1 (Dickson et al. 2018), MYXD1 (Dickson et al. 2018), MYXD4 (Dickson et al. 2018; Tamura et al. 2018), and MGA (Stevens et al. 2019; Diolaiti et al. 2018) in solid tumors with various histologies, including some bearing resemblance to NC, and others with spindle cell sarcoma morphology. Remarkably, NUTM1-fusions (BRD9 (Andersson et al. 2015), ACINI (Andersson et al. 2015; Hormann et al. 2019; Gu et al. 2016; Liu et al. 2016), SLC12A (Gu et al. 2016), ZNF618 (Gu et al. 2016), IKZF1 (Gu et al. 2016; Lilljebjorn et al. 2016), BPTF (Liu et al. 2017), CUX1, and IKZF1 (Hormann et al. 2019)) have also been discovered in a variety of leukemias, and CIC-NUTM1 was originally described as a variant of primitive neuroectodermal tumors (PNET) of the central nervous system (Sturm et al. 2016). The various NUTM1-fusions and corresponding pathology are summarized in Table 22.1.

Table 22.1 Spectrum of NUTM1-rearranged tumors
Full size table

Next-gen sequencing has also been key to characterizing exonic mutations other than the NUTM1-fusion. Surprisingly, available data indicate that there are no additional oncogenic driver mutations or inactivating mutations of tumor suppressor genes, suggesting that the NUTM1 fusion may be the sole driver of NCs (Lee et al. 2017; Stathis et al. 2016).

4 Histology/Immunohistochemistry (IHC)

NC has a characteristic, though not diagnostic, histologic appearance. It appears as an undifferentiated, malignant neoplasm, exhibiting a sheet-like growth pattern comprised of medium-sized, round- to oval-shaped cells. Its monomorphism is distinctive (Fig. 22.3a) and distinguishes it from garden variety poorly differentiated carcinomas, which are larger and exhibit more pleomorphism (Fig. 22.3b). Another common feature is the presence of clear spaces around cells and occasional “fried egg cells” that have abundant, clear cytoplasm (Fig. 22.3a). Consistent with its rapid growth, mitoses and single-cell or geographic necrosis are characteristic. The immune infiltrate cvaries, and is either neutrophilic or lymphocytic (Fig. 22.4). As a subtype of squamous carcinoma, NC often (33–40%) (Bauer et al. 2012; Chau et al. 2019) displays morphologic squamous differentiation, which is characteristically abrupt (Fig. 22.5), rather than displaying a gradient of poorly to well-differentiated cells as seen in conventional squamous cancers. This type of abrupt squamous differentiation can also be seen in basaloid HPV-associated squamous cancers of the head and neck (Table 22.2) (Chernock et al. 2010). In keeping with its squamous lineage, NCs are typically (~75%) (Tilson and Bishop 2014) positive for p63/p40 IHC and often also stain for CK5.

Fig. 22.3
figure 3

Typical histologic appearance of (a) NC compared with (b) non-NUT poorly differentiated carcinoma. (a) NC is characteristically monomorphic with clear spaces separating cells and occasional “fried egg” cells. (b) Garden variety poorly differentiated carcinoma, by contrast, displays larger cell with greater pleomorphism. Both images are 400× magnification and H&E stained

Full size image
Fig. 22.4
figure 4

NC showing a prominent neutrophilic infiltrate. Image is 400× magnification and H&E stained

Full size image
Fig. 22.5
figure 5

Abrupt keratinization is a characteristic feature of NC. Image is 400× magnification and H&E stained

Full size image
Table 22.2 Distinguishing features of tumors in the differential diagnosis of NC
Full size table

Unusual cases of NC, typically arising from salivary gland (den Bakker et al. 2009) or soft tissue (Dickson et al. 2018), can display morphologic features of myoepithelial differentiation, with rhabdoid-like cells and even formation of cartilage (Fig. 22.6); however, IHC markers do not support myoepithelial lineage; these tumors are negative for S100 and SMA and variably express p63 or GFAP (Dickson et al. 2018; Schaefer et al. 2018).

Fig. 22.6
figure 6

Rare case of NC displays chondroid differentiation (left). Image is 400× magnification and H&E stained

Full size image

Due to the significant morphologic and immunophenotypic overlap of NC with other poorly differentiated malignancies, the diagnosis is often missed or mistaken for these other entities, including most commonly poorly differentiated carcinoma, poorly differentiated squamous carcinoma (Evans et al. 2012; Stelow et al. 2008; Stelow and French 2009; Chute and Stelow 2010), Ewing’s sarcoma (Mertens et al. 2007), sino-nasal undifferentiated carcinoma (Stelow et al. 2008), and small cell carcinoma (Evans et al. 2012) (Table 22.2). Other entities that can be mistaken for NC include poorly differentiated transitional carcinoma of the genitourinary tract (Bishop et al. 2016; French et al. 2004), small-round-blue-cell tumors (blastomas) (Shehata et al. 2010), carcinoma ex pleomorphic adenoma (den Bakker et al. 2009), thymic carcinoma (Kubonishi et al. 1991; Toretsky et al. 2003; Petrini et al. 2012), and even thyroid carcinoma (Landa et al. 2016). The difficulty in distinguishing NC from these other entities by morphology alone has contributed to the vast underdiagnosis of NC (see Demographics/Prevalence, below).

The majority of NCs are carcinomas; however, a subset are so poorly differentiated that they lack epithelial markers (i.e., expression of keratin intermediate filaments). This has led to the misdiagnosis of leukemia or even lymphoma (unpublished observations), particularly when patient’s initial biopsy is from bone marrow. Moreover, a small subset of NUTM1-rearranged malignancies, approximately 5–10% (Dickson et al. 2018; Chau et al. 2019; Stevens et al. 2019), arise from soft tissue and show variable expression of epithelial markers, raising the possibility that some of these are of different lineage from typical NC.

5 Diagnosis

NC is defined molecularly by NUTM1-rearrangement and as such historically could only be diagnosed by direct demonstration of NUTM1 rearrangement, either by conventional cytogenetics, fluorescent in situ hybridization (FISH), or reverse-transcriptase PCR (RT-PCR). These methodologies are not widely available and thus hampered the diagnosis of NC during its early recognition. This changed in 2009 with the development of a NUT-specific antibody (Haack et al. 2009). Owing to the highly restricted expression of NUT protein to spermatids of the testis, expression of NUT is specific to NC and germ cell tumors (embryonal carcinoma, seminoma, and dysgerminoma) (Fig. 22.7); the rabbit anti-NUT monoclonal antibody clone C52B1 (Cell Signaling Technologies, Danvers, MA) exhibits a sensitivity of 87% and specificity of 100% when strict criteria are applied to interpretation: >50% of tumor nuclei positively stain (Haack et al. 2009). With this high specificity, positive NUT IHC is considered sufficient for the diagnosis of NC (French and den Bakker 2017).

Fig. 22.7
figure 7

Immunohistochemical staining of postmeiotic spermatids (a) and NC (b) nuclei using the anti-NUT, clone C52B1 antibody (Cell Signaling Technology, Inc.). The characteristic speckled nuclear staining pattern of BRD4-NUT can be seen in the inset (b). Images are 400× magnification

Full size image

Despite the high feasibility of diagnostic NUT IHC, many pathology laboratories still do not use it, due to the perceived rarity of NC; however, the emergence of NGS is having a large impact on detecting cases not originally considered by the pathologist/oncologist. Targeted exome NGS platforms such as those provided by Foundation Medicine (Mangray et al. 2018) or Oncopanel (Stathis et al. 2016; Wagle et al. 2012) have led to the discovery of NCs, but their sensitivity is hampered by limited coverage and the large breakpoint region of NUTM1 and BRD4. Rapid amplification of cDNA ends (RACE)-based NGS technology, however, is changing the landscape of NUTM1-rearranged tumors due to its unbiased use and high sensitivity. This test, which utilizes RNA from archival formalin-fixed paraffin-embedded material (FFPE), is able to detect the majority of NUTM1-rearrangements, because it can detect any fusion to NUTM1, including novel fusion partners, as long as the canonical exon 3 (transcript variant 1) of NUTM1 is involved, which it is in the majority of cases (Lee et al. 2017; Thompson-Wicking et al. 2013; Stirnweiss et al. 2015, 2017; French et al. 2003; Haruki et al. 2005). Companies that offer this assay include, but are not limited to, ArcherDx (Boulder, CO) (Dickson et al. 2018; Shiota et al. 2018), Caris Life Sciences (Phoenix, AZ) (Stevens et al. 2019), and Foundation Medicine (Cambridge, MA).

6 When to Consider the Diagnosis of NC

Because testing and diagnosis of NC is increasing, the disease spectrum is becoming broader, with cases arising from a large variety of non-midline organs including pancreas (Shehata et al. 2010), kidney (Bishop et al. 2016), thyroid (Landa et al. 2016), bladder (French et al. 2004), salivary glands (Ziai et al. 2010; den Bakker et al. 2009; Chau et al. 2014), bone (Mertens et al. 2007), and soft tissues (Dickson et al. 2018; Stevens et al. 2019). For this reason, we recommend performing NUT IHC to rule out NC in all poorly differentiated non-cutaneous carcinomas, with or without squamous differentiation, that have a monomorphic appearance. It is important to not exclude NC on the basis of some lineage-associated markers, such as those of neuroendocrine (chromogranin or synaptophysin), pulmonary (TTF-1), or stem cell (CD34), because NCs can exhibit positive staining for any of these (French et al. 2004; Tanaka et al. 2012; Raza et al. 2015; Bishop and Westra 2012). Moreover, advanced patient age or history of smoking should not be criteria that exclude NC, because it can affect patients of all ages, including elderly patients (Bauer et al. 2012; Stelow et al. 2008), and numerous patients with a smoking history have been diagnosed.

When should one not consider NC? Gland-forming NCs are extremely rare, if nonexistent; thus, NC need not be considered in the differential diagnosis of adenocarcinomas. Moreover, known viral etiology, such as HPV or EBV, has never been detected in NC and can be used as a basis to exclude NC. This being said, expression of the HPV-associated marker, p16, is frequently seen in NC (Salles et al. 2014) and should not be used to exclude it.

7 Demographics/Prevalence

NC affects patients of all, but predominantly young, ages, with a median age of 16–22 (range 01.–81.7 years) (Bauer et al. 2012; Chau et al. 2014) and with an equal predilection for males and females (Bauer et al. 2012; Chau et al. 2019). NC is not associated with smoking, but a history of smoking is not uncommon (unpublished observations). The prevalence of NC is not precisely known; however, a recent study using the Caris RACE-NGS platform to identify NUTM1-fusions in a cohort of 14,107 tumors revealed 9 NUTM1-fusion-positive tumors, suggesting a rough incidence of ~0.06% among all tumors (Stevens et al. 2019). Extrapolating this data based on the estimated 1.7 million new cases in the USA in 2019 (American Cancer Society) would suggest an incidence of over 1000 new cases of NC in the USA per year. This estimate confirms that NC is vastly underdiagnosed and that testing must increase.

8 Pathogenesis (BRD4-NUT Function)

NGS indicates that BRD4-NUT is the sole oncogenic driver of NC (Lee et al. 2017; Stathis et al. 2016). The fusion protein is known to bind to acetylated histones via the dual bromodomains of BRD4 (Grayson et al. 2014), which tether NUT, an unstructured protein, to chromatin. NUT recruits the histone acetyltransferase (HAT), p300, to BRD4-NUT (Alekseyenko et al. 2017; Reynoird et al. 2010) where it is presumed to acetylate the chromatin and recruit further BRD4-NUT in an iterative process that leads to massive contiguous regions of chromatin enriched with BRD4-NUT (Alekseyenko et al. 2015). These regions are termed megadomains and function essentially to upregulate transcription of underlying coding and noncoding genes (Alekseyenko et al. 2015). A critical target of BRD4-NUT in all NCs is MYC, whose upregulation drives growth and arrests NC cell differentiation (Grayson et al. 2014; Alekseyenko et al. 2015). Knockdown of either MYC or BRD4-NUT results in rapid differentiation of these cells, indicating that targeting either of these proteins could be effective strategies to treat this cancer (Grayson et al. 2014; French et al. 2008).

9 Treatment of NC

Currently, there is no established treatment strategy for NC; however, for children, the Scandinavian Ewing SSG IX regimen has led to cure of NC in a small number (n = 3) of reported cases (Mertens et al. 2007; Storck et al. 2017). Thus, for pediatric patients with localized or disseminated NC, this regimen is often recommended; however, it is not effective in all patients. In addition, complete resection has been shown repeatedly to lead to significantly (p = 0.0003–0.01) improved overall survival (Bauer et al. 2012; Chau et al. 2014), and so whenever possible, surgical resection with clean margins should be performed as soon as possible after diagnosis. Often, however, the patient has disseminated or unresectable disease at the time of diagnosis, due to its rapid growth, and chemoradiation is the only available option. More often than not, chemoradiation leads to a transient response, followed by rapid progression and death; thus, there is an urgent need for novel approaches to treat NC.

10 New Treatment Strategies

NC cells are extremely sensitive to histone deacetylase (HDAC) inhibitors in cell culture experiments (Schwartz et al. 2011). The effect of HDAC inhibitors was originally ascribed to reversal of global chromatin hypoacetylation imparted by BRD4-NUT. HDAC inhibitors cause NC cells to differentiate and arrest proliferation at low doses, and activity has been seen in mouse models (Schwartz et al. 2011). Anecdotally, HDAC inhibitors, alone or in combination with chemotherapy, do show activity against NC in humans (Fig. 22.8a), though the response is transient and not curative (Schwartz et al. 2011; Maher et al. 2015).

Fig. 22.8
figure 8

Response of NC patients to targeted inhibitors. (a) PET/CT of a 10-year-old patient to single agent HDAC inhibitor, vorinostat. Recurrence at 10 weeks was due to treatment interruption secondary to gastrointestinal toxicity. (Reproduced from Schwartz BE, Hofer MD, Lemieux ME, DE, Cameron MJ, West NH, Agoston ES, Reynoird N, Khochbin S, Ince TA, Christie A, Janeway KA, Vargas SO, Perez-Atayde AR, Aster JC, Sallan SE, Kung AL, Bradner JE, French CA. Differentiation of NUT midline carcinoma by epigenomic reprogramming. Cancer Res 2011;71(7):2686–96. https://doi.org/10.1158/0008.5472.CAN-10-3513. (b) PET scan of patient treated with the BET inhibitor, MK-8628/OTX015 comparing baseline (left) with two cycles of treatment (right). (Reproduced from Stathis A, Zucca E, Bekradda M, Gomez-Roca C, Delord JP, de La Motte Rouge T, Uro-Coste E, de Braud F, Pelosi G, French CA. Clinical Response of Carcinomas Harboring the BRD4-NUT Oncoprotein to the Targeted Bromodomain Inhibitor OTX015/MK-8628. Cancer Discov 2016; 6(5):492–500. https://doi.org/10.1158/2159-8290.CD-151335

Full size image

A more precise, though still generally nonselective, approach to treating NC came about in 2010 using BET inhibitor compounds, typified by the molecule, JQ1, which are acetyl-lysine mimetic compounds that competitively inhibit the binding of BET (BRD2, BRD3, BRD4, BRDT) protein dual bromodomains to chromatin (Filippakopoulos et al. 2010; Filippakopoulos and Knapp 2012). BET inhibitors evict BET proteins, including BRD4-NUT, from chromatin, resulting in loss of function and differentiation of NC cells in vitro and in vivo (Grayson et al. 2014; Filippakopoulos et al. 2010). Multiple trials were conducted evaluating the efficacy of BET inhibitors in human cancers, including NC. On-target activity has been shown (Fig. 22.8b); however, dose-limiting toxicity has precluded cure with these drugs (Stathis et al. 2016; Lewin et al. 2018; O’Dwyer et al. 2016). Likely a combination strategy with BET and/or other targeted inhibitors will be required for the ultimate cure of this aggressive cancer. A recent CRISPR-based screen suggests that combination of BET inhibitors with a CDK4/6 inhibitor, such as palbociclib, is synergistic in inhibiting NC (Liao et al. 2018).

11 Conclusions

NC is a recently described predominantly pediatric and young adult cancer that remains poorly recognized and underdiagnosed. The distinctly poor prognosis and need for alternative approaches to treat NC provide a strong rationale to make the diagnosis. The longer this disease remains undiagnosed, the longer it will take to better understand it and explore novel treatment approaches.