mASH1 is Highly Specific for Neuroendocrine Carcinomas: An Immunohistochemical Evaluation on Normal and Various Neoplastic Tissues

Formalin-fixed, paraffin-embedded tissue microarrays (TMAs) (in-house tissue microarrays and commercially purchased tissue microarrays, US Biomax, Rockland, Maryland) were used, consisting of normal tissues (n = 30), various neoplastic tissues (n = 1114), NSCLC (n = 268), and SCLC (n = 114). Tissue sections were cut at 4 to 5 μm and placed on positive-charged slides; then, they were deparaffinized in a xylene substitute, hydrated in a series of graded alcohols down to water, placed in a modified citrate buffer solution, and heated in a pressure cooker at 110°C for 15 minutes. The titer for mASH1 (for in vitro diagnostic use; 24B72D11.1, Biocare Medical, Concord, California) was optimized at 1:200, followed by a 2-step immunohistochemistry horseradish peroxidase micropolymer detection system and visualized with 3,3′-diaminobenzidine chromogen. A case was considered positive if 1% or more of tumor cells were positive for m/hASH1. Statistical analyses were performed using GraphPad Prism version 6.0 (GraphPad Software, La Jolla, California). P values were determined using the χ² test and confirmed with a Fisher exact test. Values of .05 or less were considered statistically significant.

Results are summarized in Tables 1 through 4. In normal tissues, m/hASH1 (nuclear staining) was sparsely expressed in the molecular cell layer, white matter, and granular cell layer of cerebellum (cerebrum was negative), C cells in thyroid, and epithelial cells in thymus (Table 1). All other normal tissues were negative, including adrenal gland, pancreas, and argentaffin cells found in the gastrointestinal tract.

In NSCLC, mASH1 stained 1.1% (1 of 93) of adenocarcinomas, 0.9% (1 of 111) of squamous cell carcinomas, and 0% (0 of 30) of large cell carcinomas, and LCNEC stained 66.7% (6 of 9) (Figure 1, A; Table 2). In SCLC, mASH1 stained 82.5% (94 of 114) (Figure 1, B; Table 3). In various other neoplastic tissues, m/hASH1 was expressed in thyroid medullary carcinomas (Figure 2, A), but not in thyroid papillary and follicular carcinomas, thymic carcinomas (Figure 2, B), or brain cancers (Figure 2, C and D; Table 4). m/hASH1 was expressed focally in a low number of infiltrating ductal breast carcinoma (Figure 3, A), ovarian serous adenocarcinoma, and pancreatic neuroendocrine tumors (Table 2). All typical carcinoids of various sites were negative (0 of 12); however, in lung atypical carcinoids (Figure 3, B), m/hASH1 was expressed in 42.9% (9 of 21).

Past studies have shown CD56, chromogranin A, and synaptophysin to be the most reliable immunohistochemical markers to detect neuroendocrine differentiation in tumors of various sites, including neuroendocrine lung tumors.^8–11  However, this panel of neuroendocrine markers cannot distinguish carcinoid from HGNEC. Proliferation rate (Ki-67) and mitotic index can aid in diagnosis, but this requires extra stains and time-consuming quantitative analysis. Also, in cases of small biopsies with limited tumor cells and/or frequently seen crush artifact, diagnosis can be particularly difficult.²  Therefore, a more specific marker for distinguishing carcinoid and HGNEC may be of upmost importance in questionable cases.

In this study, various normal and nonneuroendocrine neoplastic tissues were tested for sensitivity and specificity with mASH1. In normal tissues, m/hASH1 was observed only in cerebellum, normal thyroid C cells, and epithelial cells in thymus (Table 1). However, survey of entire normal brain and other normal tissues is needed to further confirm the lack of m/hASH1 expression, as small punched tissue microarray spots may not be representative.

Excluding brain, thymic, and medullary carcinomas, positive-staining cases in neoplasms were observed in only 6 of 910 cases (0.66%; Table 2). m/hASH1 has been shown to be highly expressed in thyroid medullary cancer, which is a neuroendocrine cancer derived from the thyroid C cells.¹²  In our results, m/hASH1 was expressed in 48.4% (15 of 31) and expression was also observed in 41.7% (5 of 12) of thymic carcinoma (Table 2). Thymic carcinoma with neuroendocrine differentiation has been previously reported; additionally, thymic epithelial cells have been shown to express neuroendocrine markers such as CD56, synaptophysin, and chromogranin.^13,14  In contrast, although soft tissue tumors, such as rhabdomyosarcoma, carcinosarcoma, leiomyosarcoma, and synovial sarcoma, have been shown to express the same neuroendocrine markers,^15–18  all 86 cases of bone and soft tissue tumors were negative for m/hASH1 expression.

In brain tumors, m/hASH1 was expressed in astrocytoma, glioblastoma, oligodendroglioma, and ependymoma tumors (Table 2). These data are supported by Somasundaram et al,¹⁹  as their study revealed that m/hASH1 was overexpressed in progressive astrocytoma and glioblastomas and levels were increased in higher grades. In our study, we also observed progressive staining percentages in grade I versus grades II through IV (Figure 2, B and C; Table 2). In total, m/hASH1 was expressed in 51.8% (72 of 139) of malignant brain cancers.

In breast cancers, only 3 cases (high grade, T4 stage) were focally positive for m/hASH1 (<10%; Table 2). Bogina et al²⁰  observed neuroendocrine differentiation in approximately 10% of breast cancers and showed significant association with T4 stage. Wachter et al²¹  also observed neuroendocrine differentiation in 20% of luminal B-like carcinomas using current World Health Organization criteria (at least 50% of tumor cells positive for synaptophysin or chromogranin A). Both studies used classic neuroendocrine markers, but did not use mASH1. Therefore, in tumors of unknown origin, breast cancer must be considered, especially when using chromogranin and synaptophysin as diagnostic markers.

Overall, m/hASH1 expression was highly specific to neuroendocrine tumors, with the exception of pancreatic neuroendocrine tumors, where merely 4.3% of specimens were positive (Table 2), and typical carcinoids in lung, which were all negative (0 of 4; Table 3). m/hASH1 expression was especially specific to SCLC versus nonneuroendocrine NSCLC (P < .001; Tables 3 and 4). Although NSCLC atypical carcinoids were positive (42.9%; cases had >10% positive cells) for mASH1, they stained less frequently than LCNEC (66.7%, P < .001) (Table 3) and SCLC, consistent with previous reports.^3,6  Each LCNEC case was stained with synaptophysin and 9 of 9 cases were positive. Low expression of synaptophysin was observed in 3 of 9 positive cases of LCNEC that corresponded to the same mASH1-negative cases, and in 2 of those negative cases, a few tumor cells were positive for mASH1 (<1%). Our results are consistent with Hiroshima et al³  and Jiang et al,⁶  where m/hASH1 expression was observed in 58.9% (10 of 17) and 71.8% (56 of 78) of LCNECs, respectively, whereas expression was absent in classic large cell carcinoma.

mASH1 stained only 1.1% (1 of 93) and 0.9% (1 of 114) of lung adenocarcinoma and lung squamous cell carcinoma cases, respectively (Table 3). The lung adenocarcinoma was confirmed by immunohistochemistry to be TTF-1/napsin A positive and the lung squamous cell carcinoma was negative for TTF-1/napsin A and positive for p40. This kind of observation was also seen by Ionescu et al²² ; however, the authors concluded that NSCLC with neuroendocrine differentiation should not be a subclass distinct from the other NSCLCs because the disease-specific characteristics and overall patient survival were not influenced by neuroendocrine differentiation.

On a final note, TTF-1 has been used to stain the majority of lung adenocarcinomas and has also been shown to stain SCLC; however, multiple studies have shown TTF-1 expression in lung carcinoids and LCNEC.^21,23,24  Therefore, the ubiquity of TTF-1 expression in pulmonary carcinoids and HGNECs argues against its use in their subclassification. Ultimately, studies will be required for the rare types of neuroendocrine cancers not presented in this study. However, although not organ specific, mASH1 was highly specific for HGNECs versus carcinoids and other nonneuroendocrine neoplasms including NSCLC in the lung.