Nuclear Protein in Testis (NUT) Carcinoma: A Comprehensive Immunohistochemical Analysis of 57 Cases With Consideration of Interpretation and Pitfall Recognition

The study was approved by the institutional review board at Mayo Clinic, Rochester, Minnesota.

Fifty-seven cases were identified from years 2012–2022 by performing a natural language search in the laboratory information system of Mayo Clinic, Rochester, for cases with a diagnosis of NUT carcinoma. Inclusion criteria were either positive NUT immunostaining (43 cases), NUTM1 gene rearrangement by FISH (3 cases), or both (11 cases) at the time of diagnosis, and areas of typical cohesive epithelioid morphology. This cohort included 10 in-house and 47 external consultation cases.

Pathology reports for all cases were reviewed. Patients’ age at diagnosis, sex, primary site of tumor, NUTM1 FISH results, and the IHC profile of all cases were recorded.

Available hematoxylin-eosin–stained slides and IHC-stained slides were reviewed. The IHC profile of all cases was recorded, including documentation of the following immunostains: NUT, keratins (cytokeratin [CK] AE1/AE3, CAM 5.2, OSCAR keratin, CK7, CK20), squamous cell markers (p40, p63, CK5/6, desmoglein 3 [DSG3]), neuroendocrine (NE) markers (synaptophysin, chromogranin, CD56, INSM1), thyroid transcription factor-1 (TTF-1) (clones SPT24 and 8G7G3/1), S100, desmin, SALL4, Ki-67, and PD-L1 (programmed death ligand-1). The results were recorded as positive or negative, and the extent of positivity was documented as focal, patchy, or diffuse.

The 10 in-house cases were stained for NUT (clone C52B1, Cell Signaling), multiple keratins (CK AE1/AE3 [clone AE1&AE3, Dako], CAM 5.2 [clone CAM 5.2, Sigma], OSCAR keratin [clone OSCAR, Covance], CK7 [clone OV-TL12/30, Dako], and CK20 [clone Ks20.8, Dako]), multiple squamous cell markers (p40 [clone BC28, Biocare], p63 [clone BC4A4, Biocare], CK5 [clone 123C3, Invitrogen], and DSG3 [clone EP306, Cell Marque]), multiple NE markers (synaptophysin [clone 27G12, Leica/Novocastra), chromogranin [clone LK2H10, Ventana], CD56 [clone 123C3, Dako], and INSM1 [clone A-8, Santa Cruz]), TTF-1 (clones SPT24 [clone SPT24, Leica/Novocastra] and 8G7G3/1 [clone 8G7G3/1, Dako]), S100 (polyclonal, Dako), desmin (clone DER11, Leica/Novocastra), SALL4 (clone 6E3, Ventana), Ki-67 (clone Mib-1, Dako), and PD-L1 (clone 22C3, Dako) when they were not performed at the time of diagnosis. The results were semiquantitatively recorded as described above.

Fifty-seven cases of NUT carcinoma were included in the study. Patients’ age ranged from 2 to 83 years (mean, 46.1 years). No sex predilection was identified. Primary sites for NUT carcinomas included the thoracic/mediastinal region (38 cases, 67%), H&N/sinonasal region (13 cases, 23%), and other sites (6 cases, 11%) including colon, pancreas, and pelvic soft tissue. The demographic data are presented in Table 1.

NUT carcinomas exhibited a wide spectrum of morphology, varying from sheets of small round blue cells to tumors with prominent abrupt keratinization (Figure 1, A and B). The diagnosis was established by positive NUT immunostaining (43 cases), NUTM1 gene rearrangement by FISH (3 cases), or both (11 cases). Three cases diagnosed by NUT IHC alone had subsequent NGS studies, showing MXD4::NUTM1 gene fusions in 2 cases and BRD4::NUTM1 gene fusion in 1 case, which was negative for NUTM1 gene rearrangement by FISH. For the 3 cases where the diagnosis was established by NUTM1 gene rearrangement by FISH, NUT IHC was performed on 2 available cases, and both demonstrated positive staining.

The expression pattern of keratins in NUT carcinomas is summarized in Table 2, including aggregate results and comparison across different locations. Broad-spectrum keratin stains were positive in a high percentage of NUT carcinomas: CK AE1/AE3 was positive in 84% (38 of 45) of cases, CAM 5.2 was positive in 88% (23 of 26) of cases, and OSCAR keratin was positive in 76% (22 of 29) of cases. Thus, positivity rates for CK AE1/AE3, CAM 5.2, and OSCAR were generally similar, with CAM 5.2 being the most sensitive. Thoracic NUT carcinomas showed a lower rate of keratin expression than those located in the H&N. In thoracic cases, CK AE1/AE3 was positive in 81% (25 of 31) of cases, CAM 5.2 was positive in 82% (14 of 17) of cases, and OSCAR was positive in 76% (16 of 21) of cases. In contrast, all H&N NUT carcinoma cases stained positively for CK AE1/AE3 (10 of 10) and CAM 5.2 (7 of 7), and most stained positively for OSCAR (80%, 4 of 5). NUT carcinoma from other sites showed similar but slightly lower broad-spectrum keratin expression.

Most H&N NUT carcinomas were positive for CK7 (86%, 6 of 7), but only about half of thoracic NUT carcinomas were positive for CK7 (55%, 11 of 20). Of 17 thoracic NUT carcinomas, only 1 (6%) was positive for CK20, while CK20 was positive in 2 of 5 H&N NUT carcinomas (40%). The 1 appendiceal NUT carcinoma was positive for both CK7 and CK20 (focal).

NUT carcinoma is a carcinoma, by definition, that usually correlates with expression of keratin. Among the keratin-positive cases in this study, the extent of expression of keratin ranged from focal to diffuse. For the 5 keratin stains analyzed, most cases (85%, 47 of 55) were positive for at least 1 keratin (Figure 2, A through C), indicating that 15% of NUT carcinoma cases could be negative for multiple keratin stains (Figure 2, D).

Most NUT carcinomas in this study showed positivity for at least 1 marker for squamous differentiation analyzed (79%, 34 of 43; Table 3). The commonly used squamous markers p40, p63, and CK5/6 were positive in 65% (24 of 37), 87% (20/23), and 71% (24 of 34) of the cases, respectively. p40, a sensitive and specific marker widely used for pathologic diagnosis of squamous cell carcinoma, showed the least sensitivity among the 4 markers tested in thoracic and H&N NUT carcinoma. When comparing p40 to p63, p40 showed a 22% lower (65% versus 87%) positivity rate in all cases combined. Interestingly, DSG3 showed the highest positivity rate, with 100% (11 of 11) of cases staining positively across all tumor locations.

DSG3 and p63 generally demonstrated diffuse and strong positive staining, whereas p40 and CK5/6 had a higher frequency of focal positive staining pattern. Different squamous markers could show different patterns and strength of staining within the same case. For example, 1 case displayed strong and diffuse DSG3 and p63 expression (Figure 3, A and B) with only patchy CK5/6 and focal p40 staining (Figure 3, C and D).

Small cell carcinoma and combined small cell and squamous cell carcinoma are top considerations in the differential diagnosis of NUT carcinoma, owing to morphologic similarity. In the cases analyzed, 35% (15 of 43) had at least 1 NE marker showing positivity by IHC (Table 4). Individually, synaptophysin was positive in 15% (7 of 39) and CD56 was positive in 18% (5 of 28) of all cases analyzed. Chromogranin was negative in all tested cases (0 of 33). Surprisingly, INSM1 showed a high positivity rate in NUT carcinomas (54%, 7 of 13 cases).

The extent of positivity among NE markers ranged from focal to patchy, with only 2 cases displaying diffuse positive staining for CD56. In addition, cases could show differential staining for NE markers, with 1 example case displaying patchy strong INSM1 expression (Figure 4, A), focal weak synaptophysin staining (Figure 4, B), and negative chromogranin and CD56 staining (Figure 4, C and D).

Many entities have morphologic overlap with NUT carcinoma, including poorly differentiated lung adenocarcinoma, melanoma, lymphoma, leukemia, germ cell tumor, thoracic SMARCA4-deficient undifferentiated tumor, SMARCB1-deficient neoplasms, and other small round blue cell tumors (ie, rhabdomyosarcoma and Ewing sarcoma). Results of various immunohistochemical markers that may be expressed in entities in this differential diagnosis are summarized in Table 5. TTF-1 was largely negative, with focal positivity in 26% (10 of 38) of cases. No notable performance difference was identified between clones SPT24 and 8G7G3/1. S100 showed patchy positivity in 3% (1 of 33) of cases, desmin was focally positive in 6% (2 of 33) of cases, and CD45 was positive in 8% (2 of 26) of cases. CD34, CD99, and SALL4 were positive in 15% (3 of 20), 21% (3 of 14), and 15% (3 of 20) of cases, respectively, across all tumor locations. BRG1 and INI1 were retained in all cases tested (22 and 28 cases, respectively). Ki-67 proliferative index was high (median, 60%; range, 20%–90%). PD-L1 was negative in all 10 tested cases.

The diagnosis of NUT carcinoma has increased over time owing to improved awareness of the entity and the increasingly widespread availability of NUT IHC. However, histologic challenges in the diagnosis of NUT carcinoma persist owing to morphologic overlap with many other entities. To aid in this often-difficult differential diagnosis, we comprehensively examined the immunoprofile of NUT carcinomas to identify potential diagnostic pitfalls.

Keratin expression in NUT carcinomas helps establish the epithelial lineage. However, previous studies have shown that keratin expression in NUT carcinomas can be variable. Sholl et al¹⁵  observed focal positivity in all cases. In contrast, Evans et al¹⁶  observed negative staining for pan-keratin cocktail, CAM 5.2, and/or CK AE1/AE3, with only multifocal CK7 expression in areas of keratinization. While most cases in our study were positive for at least 1 keratin, 15% of all cases and 19% of thoracic cases were negative for multiple keratins, presenting a potential diagnostic pitfall. The sensitivity of individual CK AE1/AE3, CAM 5.2, and OSCAR antibodies was generally similar in our cohort, ranging from 76% to 88%, with CAM 5.2 displaying the highest sensitivity in all sites. As a result, multiple keratin stains should be considered, particularly including CAM 5.2, when NUT carcinoma is suspected during diagnostic workup. Since expression of keratins was relatively lower in thoracic cases than in H&N cases, negative keratin staining would not rule out NUT carcinoma, especially in the thoracic region. CK7 and CK20 expression patterns do not appear to effectively suggest the primary site or region of tumor origin.

One case from the colon and 1 from the terminal ileum/right colon were revealed to have MXD4::NUTM1 fusion by NGS in our study, but both were also positive for keratin stains and had epithelioid morphology. Whether these represent true NUT carcinomas with uncommon fusion partners or an epithelioid variant of the recently reported NUTM1-rearranged (colorectal) sarcoma¹⁷  is an open discussion. We restricted the inclusion criteria for this study to “carcinoma,” based on how the cases were originally diagnosed and the confirmed epithelioid morphology upon re-review. The recently reported NUTM1-rearranged entities include neoplasms originating from different organs and encompass a broad spectrum of histomorphology features. Definitive classification of these neoplasms as carcinoma, sarcoma, or undifferentiated tumors currently remains very challenging in some cases. These entities may eventually be defined by the NUTM1 fusion partner observed in the tumor. Classification efforts could be greatly aided by future multicenter collaboration studies using NGS to better characterize this diverse set of tumors.

NUT carcinoma has been considered a unique subtype of squamous cell carcinoma, implying that the cell of origin is normal squamous cells. Markers for squamous differentiation are frequently used in thoracic and H&N neoplasms to facilitate the establishment of squamous origin/squamous differentiation. It has been reported that most NUT carcinomas show nuclear staining for p63/p40.¹  In our study, commonly used squamous markers displayed variable positivity in NUT carcinoma. p40, an isoform of p63 with relatively high sensitivity and specificity for squamous differentiation, which is favored by many practices, showed a notably lower positivity rate than p63 overall (65% versus 87%) in both thoracic and H&N cases. A similar finding was reported by Matsuda et al¹⁸  in a cohort of 7 cases. DSG3 is a sensitive and specific marker for squamous cell carcinomas, with a reported sensitivity of 88% and specificity of 98%.¹⁹  Interestingly, although only a small number of cases were stained, our study suggests that DSG3 is a reliable marker of squamous differentiation in NUT carcinomas, with the highest positive rate (100%, 11 of 11) among all squamous markers analyzed. These findings suggest that the use of a p40 antibody alone may create a diagnostic pitfall and negative p40 staining has limited implication to rule out NUT carcinoma. In this context, adding p63 and/or DSG3 to the panel of immunostains may help establish squamous differentiation and aid in the diagnosis of NUT carcinoma.

When considering NUT carcinoma, one of the most important differential diagnoses is high-grade NE carcinoma, in particular small cell carcinoma. This is especially true in the thoracic region, which is a preferred primary site for both. Furthermore, both often have a central-predominant distribution. While NE markers are frequently included in the IHC workup, focal positivity for synaptophysin and chromogranin in NUT carcinoma has been previously reported.^15,20  INSM1 is a transcription factor of NE differentiation that has demonstrated superior sensitivity and specificity for the diagnosis of NE tumors of the thoracic cavity and in H&N region, compared to synaptophysin, chromogranin, and CD56.^21,22  Surprisingly, our cohort displayed remarkably higher INSM1 positivity than synaptophysin or chromogranin positivity, with staining in 54% of the NUT carcinomas (versus 15% and 0%, respectively). While thoracic and H&N NUT carcinoma showed similar rates of any NE marker positivity (36% and 45%, respectively), none of the 4 tested abdominal/pelvic tumors showed NE marker positivity. These results highlight another potential diagnostic pitfall, as NE immunostains should be interpreted with caution to differentiate NUT carcinoma from NE carcinomas, especially when including multiple NE markers or using INSM1.

TTF-1 is a transcription factor marker frequently used to evaluate tumors arising in the thoracic and H&N regions. Although thoracic carcinoma with TTF-1 expression, along with lack of squamous differentiation, suggests a diagnosis of primary lung adenocarcinoma, focal expression of TTF-1 in NUT carcinomas has been described before.^12,23  The study by Hung et al²³  even demonstrated a NUT carcinoma case with diffuse and strong expression of TTF-1 and only focal p40 staining, presenting as a challenge for thoracic cases. In our study, we observed TTF-1 positivity in 26% of cases (31% of thoracic cases), although most were only focally positive. Taken together, these data indicate that primitive-appearing carcinomas with TTF-1 expression still require NUT carcinoma workup.

Melanoma, rhabdomyosarcoma, and lymphoma often enter the differential diagnoses of NUT carcinoma. S100, desmin, and CD45 were rarely positive (all <10%) in our cohort. In general, those markers are helpful in ruling out such entities. However, CD34, CD99, and SALL4 were positive in a noticeable proportion of cases (15%–21%) and could present as diagnostic pitfalls. CD34 positivity has been reported in some NUT carcinomas and can be particularly problematic if the tumor metastasizes to the bone, mimicking bone marrow involvement by acute leukemia.²⁴  A portion of NUT carcinomas have been reported to be CD99 positive²⁵  and thus can mimic Ewing sarcoma. Focal staining with SALL4 has also been described in a subset of NUT carcinoma cases, especially in those harboring the rare ZNF532::NUTM1 fusion, where Agaimy et al²⁶  reported SALL4 staining in 43% (3 of 7) of cases. A smaller portion (15%) of cases demonstrated SALL4 positivity in our cohort, which could mimic a germ cell tumor.

SMARCA4-deficient tumors and SMARCB1-deficient tumors typically have undifferentiated and often rhabdoid morphology, overlapping with NUT carcinoma. We found that BRG1 was retained in all 22 tested cases, helping with the differential diagnosis of thoracic SMARCA4-deficient undifferentiated tumors, similar to previous reports.²⁷  INI1 has been shown to be retained in NUT carcinomas²⁸  and helps to rule out SMARCB1-deficient carcinomas in the differential diagnosis. In our study, INI1 was retained in all 28 cases tested. Both BRG1 and INI1 stains are helpful in differentiating NUT carcinoma from SMARCA4- and SMARCB1-deficient tumors.

Ki-67 staining of NUT carcinomas has been reported to often show a very high proliferative index.²⁹  We observed similar findings, with Ki-67 proliferation indexes ranging from 20% to 90%, with a mean of 60%. Although there have been reported cases of PD-L1 positivity in NUT carcinomas,³⁰  making patients eligible for anti–PD-1 or anti–PD-L1 therapies, all cases tested in our cohort were negative.

This study was limited by the inability to perform immunostains on all external consult cases of this rare entity. As a result, the sample sizes for some immunostains were limited. In addition, molecular testing was not available on all cases, which limits the ability of this study to comment on the specific fusion genes observed.

In summary, this is the largest study to date examining the comprehensive IHC profile of NUT carcinoma. Our study raises awareness of potential diagnostic pitfalls when considering this rare diagnosis. Overall, understanding the immunophenotypic spectrum of NUT carcinoma will benefit surgical pathologists in interpretation of the significance of observed IHC results and facilitate NUT carcinoma workup in the differential diagnosis of poorly differentiated carcinoma with primitive morphology.