Squamous and Neuroendocrine Specific Immunohistochemical Markers in Head and Neck Squamous Cell Carcinoma: A Tissue Microarray Study

Abstract

The performance characteristics of neuroendocrine-specific and squamous-specific immunohistochemical markers in head and neck squamous cell carcinomas (SCC), in particular in oropharyngeal tumors in this era of human papillomavirus (HPV)-induced cases, are not well-established. The differential diagnosis for poorly differentiated SCCs, for nonkeratinizing oropharyngeal SCCs, and for other specific SCC variants such as basaloid SCC and undifferentiated (or lymphoepithelial-like) carcinomas includes neuroendocrine carcinomas. Given that neuroendocrine carcinomas of the head and neck are aggressive regardless of HPV status, separating them from SCC is critically important. In this study, we examined the neuroendocrine markers CD56, synaptophysin, and chromogranin-A along with the squamous markers p40 and cytokeratin 5/6 in a large tissue microarray cohort of oral, oropharyngeal, laryngeal, and hypopharyngeal SCCs with known HPV results by RNA in situ hybridization for the oropharyngeal tumors. Results were stratified by site and specific SCC variant. The neuroendocrine stains were rarely expressed in SCC (<1% overall) with CD56 the least, and chromogranin-A the most, specific markers. Further, p40 and cytokeratin 5/6 were very consistently expressed in all head and neck SCC (>98% overall), including very strong, consistent staining in oropharyngeal HPV-related nonkeratinizing SCC. Undifferentiated (or lymphoepithelial-like) carcinomas of the oropharynx are more frequently p40 or cytokeratin 5/6 negative or show only weak or focal expression. In summary, markers of neuroendocrine and squamous differentiation show very high specificity and sensitivity, respectively, across the different types of head and neck SCC.

Electronic supplementary material

The online version of this article (doi:10.1007/s12105-017-0825-y) contains supplementary material, which is available to authorized users.

Introduction

Relatively specific immunohistochemical markers of squamous and neuroendocrine differentiation have existed for many years. Their performance is well established in lung non-small cell carcinomas due to the critical need to distinguish squamous cell carcinomas (SCC), adenocarcinomas, and neuroendocrine carcinomas [1–6], particularly in small transbronchial and transthoracic needle biopsy specimens. The performance of these markers in head and neck SCC is not as well studied, and mostly consists of small studies of basaloid and other poorly differentiated carcinomas and their differential diagnosis [7, 8]. For conventional head and neck SCCs, this issue is probably understudied because the diagnosis of these tumors is usually quite easy just on hematoxylin and eosin stained slides, even in small specimens. There is one systematic investigation of neuroendocrine markers in head and neck SCC, by Schartinger et al. [9], which found 41 and 18% of tumors to express chromogranin-A and synaptophysin, respectively, albeit in patchy distribution. This contradicts significant personal experience with commercially-available, highly specific antibodies. Further, there is a particular paucity of data in oropharyngeal SCC. In the human papillomavirus (HPV) era, most oropharyngeal SCCs have a nonkeratinizing appearance [10, 11], and the differential diagnosis with neuroendocrine and other blue cell tumors is more difficult. Recent reports of HPV-related small [12–14] and large cell neuroendocrine carcinomas of the oropharynx [15] with poor prognoses underscore the importance of separating these aggressive tumors from SCC and highlight the need to know the performance of squamous and neuroendocrine markers in oropharyngeal SCC versus neuroendocrine carcinomas. Anecdotally, oropharyngeal small and large cell neuroendocrine carcinomas can be very difficult to distinguish from SCC by H&E morphology alone. Adding to the difficulty is the fact that small and large cell neuroendocrine carcinomas are often mixed with a nonkeratinizing SCC component [12–15]. HPV and p16 status cannot be used to support a tumor as SCC because neuroendocrine carcinomas of the oropharynx are also usually HPV positive. p63 is a squamous specific marker but is not as specific as its newer isoform p40 [1, 2, 4, 5]. Antibodies to the latter, which is one of the many isoforms of p63, have been little investigated in head and neck SCC [16].

This study investigated the performance of the most common immunohistochemical markers of squamous and neuroendocrine differentiation in a large tissue microarray cohort of oral, oropharyngeal, laryngeal, and hypopharyngeal SCCs with established HPV status by RNA in situ hybridization for oropharyngeal cases. Marker reactivity in some of the uncommon specific variants of head and neck SCC such as basaloid, undifferentiated (lymphoepithelial-like), and adenosquamous carcinoma was also investigated.

Materials and Methods

With approval from the Human Research Protection Office of Washington University in St. Louis, newly diagnosed cases of primary, untreated oropharyngeal, oral cavity, laryngeal, and hypopharyngeal SCC were identified from 1997 to 2008 from the databases of Radiation Oncology and Otolaryngology Head and Neck Surgery. Clinical follow up information was obtained from clinician databases and from the electronic medical records, including survival data, smoking, and other clinical variables. All cases were reviewed by one study pathologist (JSL) for confirmation. The oropharyngeal SCC were typed histologically according to our established system as keratinizing, nonkeratinizing with maturation, and nonkeratinizing SCC [10, 11, 17]. Conventional (or keratinizing) SCC were graded by standard descriptions as well, moderately, or poorly differentiated. The SCC variants that were on the arrays were typed according to World Health Organization definitions and established features from defining studies [18–21]. Other pathologic features were obtained by report review from the Copath database. All patients were treated either with primary surgery with or without postoperative radiation and chemotherapy or were treated definitively with radiation therapy with or without chemotherapy. All radiation therapy was intensity modulated. The paraffin blocks were obtained from department files, and a tissue microarray constructed. According to the amount of available biopsied or resected tumor tissue, duplicate 2 mm punches (or if only limited amounts of tumor tissue present, 0.6 mm punches) were taken from each case. Since most of the cases (>75%) were treated with primary surgery, the majority of cases on the array had the larger (2 mm) punches.

Immunohistochemistry

Immunohistochemistry was performed on 5 micron sections of formalin-fixed paraffin embedded sections for CD56, synaptophysin, and chromogranin-A on a Leica Bond Max automated stainer according to manufacturer’s instructions. All steps besides dehydration, clearing and coverslipping were performed on the Bond Max. The Bond Polymer Refine detection system was used for visualization. Slides were the dehydrated, cleared and coverslipped. Immunohistochemistry was also performed on 5 micron sections of formalin-fixed, paraffin-embedded sections for p40 and cytokeratin 5/6 on a Ventana BenchMark ULTRA automated stainer and an ultraView detection system according to manufacturer’s instructions (Ventana Medical Systems Inc. Tucson, AZ). Deparaffinization and antigen retrieval (i-view detection system; Ventana) were carried out as an automated program of the above-mentioned stainer. Appropriate positive controls were used on all runs. Details of the various antibodies are presented in Table 1. Immunohistochemical stains were interpreted by one study pathologist (JSL) for distribution in quartiles (1 = 1–25% expression; 2 = 26–50%; 3 = 51–75%; 4 = 76–100%) and intensity (1 = weak; 2 = moderate; 3 = strong). Cases with less than 10% surface area of tumor across both punches on the array were excluded.

Table 1.

Antibodies

	Instrument	Antibody clone	Company	Dilution	Antigen retrieval	Incubation time (min)
CD56	Leica Bond Max	PA0191	Leica	Ready to use	Heat induced, epitope retrieval solution 1; 20 min	60
Synaptophysin	Leica Bond Max	PA0430	Leica	Ready to use	Heat induced, epitope retrieval solution 1; 20 min	30
Chromogranin-A	Leica Bond Max	PA0299	Leica	Ready to use	Heat induced, epitope retrieval solution 2; 20 min	60
p40	Ventana Benchmark Ultra	BC28	Biocare Medical	Ready to use	Heat induced, epitope retrieval; solution CC1; 36 min	32
Cytokeratin 5/6	Ventana Benchmark Ultra	D5/16 B4	Cell Marque	Ready to use	Heat induced, epitope retrieval; solution CC1; 64 min	32

HPV RNA In Situ Hybridization

In situ hybridization for high risk HPV E6/E7 mRNA was previously performed for other studies [22, 23] by hand using the RNAscope™ HPV kit (Advanced Cell Diagnostics, Inc., Hayward, CA) as described in more detail previously. The cocktail detected 13 high risk types (16, 18, 31, 33, 35, 45, 51, 52, 56, 58, 59, 68, and 70). The array (and corresponding control) slides were previously read by one study pathologists (JSL) and classified in a binary manner as either positive or negative. Positive cases had to have granular cytoplasmic and/or nuclear brown staining that was above any signal present on the DapB negative control slide. Cases where there was less than 10% surface area consisting of tumor averaged across the two microarray punches were excluded.

Results

There were 214 oropharyngeal and 82 non-oropharyngeal SCC cases on the arrays that had at least one result for the immunohistochemistry. The total numbers of informative cases varied between the markers due to variable core loss from the histologic sectioning of the multiple microarray blocks (17 total blocks needed for array construction). All results are presented in Supplemental Table 1. For the oropharyngeal cases, 168 (78.5%) were HPV RNA in situ hybridization positive. Results of the staining for all sites and subsites are presented in Table 2. Overall, for the neuroendocrine markers, 2.3% of the tumors were positive for CD56, 0.9% for synaptophysin, and none for chromogranin-A (Fig. 1). There was a slight difference by HPV status for the oropharyngeal cases, with 5 neuroendocrine marker positive tumors amongst the 172 informative cases overall (2.9%), 2 amongst the 137 HPV positive cases (1.5%), and 3 amongst the 33 informative HPV negative ones (10%; p = 0.051). This trend was actually towards more neuroendocrine marker expression in the HPV-negative oropharyngeal SCC cases.

Table 2.

Results of immunohistochemistry for all sites, subsites, and, for oropharyngeal SCC, by HPV RNA in situ hybridization status

	CD56	Synaptophysin	Chromogranin A	p40	Cytokeratin 5/6
Oropharynx	4/169 (2.4%)	1/170 (0.5%)	0/172 (0%)	195/197 (99.0%)	206/208 (99.0%)
HPV-positive	1/136 (0.7%)	1/137 (0.7%)	0/137 (0%)	156/157 (99.4%)	160/162 (98.8%)
HPV-negative	3/33 (9.1%)	0/33 (0%)	0/35 (0%)	39/40 (97.5%)	46/46 (100%)
Oral cavity	0/22 (0%)	0/24 (0%)	0/25 (0%)	30/31 (96.8%)	42/42 (100%)
Larynx	0/17 (0%)	0/15 (0%)	0/18 (0%)	25/25 (100%)	26/26 (100%)
Hypopharynx	1/6 (16.7%)	1/5 (20%)	0/7 (0%)	7/7 (100%)	10/10 (100%)
All sites	5/214 (2.3%)	2/214 (0.9%)	0/222 (0%)	257/260 (98.8%)	284/286 (99.3%)

Fig. 1.

Representative images of the cases of conventional squamous cell carcinoma positive for neuroendocrine markers. a Moderately differentiated squamous cell carcinoma, keratinizing type, of the oral cavity (H&E) with positive CD56 (b) and synaptophysin immunohistochemistry (c). d Nonkeratinizing squamous cell carcinoma of the oropharynx showing positive staining for CD56 (e) but negative for synaptophysin (f). (a, b, c, e, f—20× magnification; d—8× magnification)

Results for the oropharyngeal cases by histologic type are presented in Table 3. The neuroendocrine marker positive cases included both keratinizing and nonkeratinizing SCC (three and two cases, respectively). Thus, despite their blue cell appearance and lack of maturing squamous differentiation in most cases, nonkeratinizing SCC were not more likely to non-specifically express neuroendocrine markers than keratinizing SCC nor than the specific SCC variants. Results for the non-oropharyngeal cases by histologic type are presented in Table 4. For the poorly differentiated, keratinizing type, SCC, there were no cases positive for neuroendocrine markers, and the few SCC variants on the array were also all negative.

Table 3.

Immunohistochemical results by histologic type in the oropharyngeal tumors

	CD56	Synaptophysin	Chromogranin A	p40	Cytokeratin 5/6
Keratinizing	3/34 (9%)	0/35 (0%)	0/36 (0%)	40/40 (100%)	46/47 (98%)
Nonkeratinizing	1/100 (1%)	1/100 (1%)	0/100 (0%)	117/117 (100%)	114/114 (100%)
Nonkeratinizing with maturation	0/44 (0%)	0/44 (0%)	0/44 (0%)	49/49 (100%)	54/54 (100%)
Basaloid	0/2 (0%)	0/2 (0%)	0/2 (0%)	1/1 (100%)	1/1 (100%)
Adenosquamous	0/2 (0%)	0/2 (0%)	0/2 (0%)	2/2 (100%)	2/2 (100%)
Undifferentiated	0/2 (0%)	0/2 (0%)	0/2 (0%)	1/3 (33%)	2/3 (67%)

Table 4.

Immunohistochemical results by histologic type in the non-oropharyngeal tumors

	CD56	Synaptophysin	Chromogranin A	p40	Cytokeratin 5/6
Keratinizing
Well differentiated	0/6 (0%)	0/6 (0%)	0/6 (0%)	8/8 (100%)	8/8 (100%)
Moderately differentiated	1/29 (3%)	1/27 (4%)	0/33 (0%)	38/39 (97%)	50/50 (100%)
Poorly differentiated	0/9 (0%)	0/8 (0%)	0/10 (0%)	12/12 (100%)	12/12 (100%)
Basaloid	0/1 (0%)	0/1 (0%)	0/2 (0%)	1/1 (100%)	1/1 (100%)
Adenosquamous	0/1 (0%)	0/2 (0%)	0/2 (0%)	3/3 (100%)	3/3 (100%)

For the squamous-specific markers (Fig. 2) p40 and cytokeratin 5/6, there was very high sensitivity, with >98.8% of all oropharyngeal and non-oropharyngeal cases positive for both markers. Sensitivity largely did not vary by histologic SCC type. In particular, despite appearing to be “poorly differentiated” with limited squamous maturation, all oropharyngeal nonkeratinizing SCC were very consistently positive for both markers. The one histologic variant that showed a difference in results was undifferentiated (lymphoepithelial-like) carcinoma. In this series, there were only three such cases, all oropharyngeal. Two of the three completely lacked both p40 and cytokeratin 5/6 expression (Fig. 3). None were positive for neuroendocrine markers.

Fig. 2.

Nonkeratinizing squamous cell carcinoma of the oropharynx (H&E) (a), showing strong and diffuse positivity for p40 (b) and cytokeratin 5/6 (c). (All images 10× magnification)

Fig. 3.

Immunohistochemistry results for undifferentiated (lymphoepithelial-like) carcinoma showing a case with lack of p40 and cytokeratin 5/6 staining. a Medium power view (20× magnification); b high power view (40× magnification); c negative p40 immunohistochemistry (20× magnification); d negative cytokeratin 5/6 immunohistochemistry (20× magnification)

Intensity of the expression in positive cases varied somewhat. For the two synaptophysin positive cases, the average fraction of staining was 1 (out of 4) and the intensity 1.5 (out of 3). For the five CD56 positive cases, the average fraction of staining was 2 (out of 4) and the intensity 2 (out of 3). For p40, the average fraction of cells was 3.9 (out of 4) and the intensity 2.9 (out of 3). For cytokeratin 5/6, staining was less intense and less complete than p40 with an average fraction of 3.85 (out of 4) and intensity of 2.7 (out of 3).

The issue of focal staining can be partially addressed by intercore staining variability. For the five SCC cases positive for neuroendocrine markers, CD56 and synaptophysin, two of them only had one core present on the array stained sections (due to core loss of the other from sectioning). The remaining three all had staining, albeit focal, on both cores. For p40 and cytokeratin 5/6, all of the cases that had both cores present on the stained slides had positive staining. The few focal staining cases also had positivity on both cores.

One very interesting and unexpected finding was of a single case of large cell neuroendocrine carcinoma (Fig. 4). This tumor was diagnosed in routine practice and even in re-review for TMA creation as nonkeratinizing SCC. However, in this study, it was the one tumor that was positive for both CD56 and synaptophysin, negative for cytokeratin 5/6, and only very focal positivity for p40 (1+; <25%). Re-review of the TMA H&E and review of the whole slides from the surgical resection specimen confirmed that there was no squamous differentiation and that the tumor had round nuclei with a subtle speckled chromatin appearance and a suggestion of palisading. It was very difficult (nearly impossible) to tell from SCC by H&E grounds alone. The tumor was strongly and diffusely positive for p16 and was positive for high risk HPV type 16 by RNA in situ hybridization. We checked patient follow up and interestingly, the patient was alive and disease free nearly 7 years after treatment with primary radiation and chemotherapy.

Fig. 4.

Immunohistochemistry results for the large cell neuroendocrine carcinoma case. a Low power view (8× magnification); b high power view (40× magnification); c positive synaptophysin immunohistochemistry (10× magnification); d positive CD56 immunohistochemistry (20× magnification); e very focally positive p40 immunohistochemistry (20× magnification); f negative cytokeratin 5/6 immunohistochemistry (20× magnification)

Discussion

Although p40 immunohistochemistry has been studied extensively in lung carcinomas because of its utility in the differential diagnosis of SCC and adenocarcinoma or other non-small cell carcinomas [1–3, 5, 24], it has not been thoroughly evaluated in head and neck SCC. While p40 is clearly very useful for lung carcinoma evaluation, as a sensitive marker for SCC that is more specific than p63 expression [1], the inclination to simply extrapolate from lung to head and neck tumors may be misguided. Actual head and neck specific data is needed in order to correctly interpret staining results in routine clinical practice.

Tacha et al. [16] evaluated p40 immunohistochemistry in a large series of normal and tumor tissues. They had 67 various head and neck carcinoma, mostly SCC, in their study. Of the 59 cases of head and neck SCC, 50 (84.7%) were positive, defined as >1% tumor cell expression. However, the actual specific extent of reactivity was not reported and further, they reported only one case of “pharynx SCC”, and did not provide any HPV assessment, tumor grading, nor histologic typing. Singh et al. [25] evaluated p40 and cytokeratin 5/6 in the differential diagnosis of sinonasal undifferentiated carcinoma, sinonasal “poorly differentiated” SCC, nasopharyngeal SCC, and olfactory neuroblastoma and reported that all ten sinonasal SCC were p40 and cytokeratin 5/6 positive. Bishop et al. [26] evaluated p40 expression in head and neck spindle cell carcinomas and found that only 54% were positive.

Our results help provide more robust data. p40 and cytokeratin 5/6 are expressed in the vast majority of head and neck SCC, including HPV-positive oropharyngeal SCC. Staining was usually extensive (>75% of cells) and strong in intensity, although more so for p40 than for cytokeratin 5/6. Even in the few cases where it was focal, staining was present on both cores suggesting that our TMA-based study sampled enough tumor to still catch focal or patchy positive tumors. This was also the case for the rare cases of SCC with neuroendocrine marker expression. The only SCC/SCC variant to lack expression was undifferentiated (lymphoepithelial-like) carcinoma of the oropharynx. Only one of three (33%) was positive for p40 and two of three (66.7%) for cytokeratin 5/6. Further, one of the two cytokeratin 5/6 cases had markedly reduced expression (1+; <25%; moderate intensity). All other histologic subtypes were consistently positive for both markers. This suggests that there truly is an “undifferentiated” carcinoma of the oropharynx and that it may or may not be related to transcriptionally-active high risk HPV.

Of particular note, although HPV-positive oropharyngeal SCCs often appear nonkeratinizing and therefore “poorly differentiated”, we know this is not actually the case. The new World Health Organization blue book [27] recognizes now that such tumors should not be graded nor a traditional differentiation status provided. Results from the current study support this concept. Although these tumors morphologically are composed predominantly or entirely of cells that do not look like they have maturing squamous differentiation, the tumor cells of nonkeratinizing SCC, on the basis of 100% positivity for p40 and cytokeratin 5/6 reactivity, may be highly differentiated. Although p40 is only a single isoform of p63, making it more specific for squamous differentiation, it still appears to retain high sensitivity in head and neck SCC.

Neuroendocrine immunohistochemistry was highly specific, with no chromogranin-A expressing SCC and only 0.9% synaptophysin, and 2.3% CD56 positive SCC. As practical experience suggests, CD56 is the least specific of the neuroendocrine markers and caution must be exercised in any tumor that is suspected to be neuroendocrine, but is positive only for CD56 and not for other neuroendocrine markers. True large cell neuroendocrine carcinomas of the oropharynx have been described recently, many related to transcriptionally-active high risk HPV [15]. These are morphologically similar to nonkeratinizing SCC and this diagnostic pitfall (calling such tumors SCC) has been highlighted by some investigators [15]. Indeed, one of the cases that was accumulated on this TMA of oropharyngeal SCC, after staining with the markers of the current study, actually is a large cell neuroendocrine carcinoma. It was excluded from the tables and analysis. Interestingly, it is high risk HPV RNA positive. The morphology of this tumor reveals just how similar to nonkeratinizing SCC these can look (Fig. 4) and highlights the need for a low threshold for immunohistochemistry if neuroendocrine differentiation is suspected. It also highlights the fact that p16 and/or high risk HPV testing cannot be used to support that a tumor is squamous because oropharyngeal neuroendocrine carcinomas are also usually HPV positive. Although this patient did not develop progressive disease after treatment, the literature on small and large cell neuroendocrine carcinomas of the oropharynx strongly support that they are more aggressive than SCC, even when high risk HPV-related [12–15].

Conclusion

In conclusion, this study better defines the expression of neuroendocrine and squamous specific markers by immunohistochemistry in head and neck SCC. Neuroendocrine markers are only rarely expressed by head and neck SCC, and the squamous specific markers p40 and cytokeratin 5/6 are very sensitive for squamous differentiation, a finding that does not appear to be decreased in nonkeratinizing oropharyngeal SCC. While these markers are clearly very useful in distinguishing SCC from aggressive neuroendocrine carcinomas, it is important to realize that their sensitivity and specificity are not absolute. A very small percentage of SCCs (including keratinizing and non-keratinizing/HPV-related types) express neuroendocrine markers, with CD56 staining the most commonly seen. Accordingly, as in other sites, staining with these markers by themselves is not sufficient for a diagnosis of neuroendocrine carcinoma. In addition, rare examples of SCC, especially the undifferentiated (lymphoepithelial-like) variant, lack p40 and cytokeratin 5/6 expression altogether. Indeed, squamous and neuroendocrine markers must be interpreted together with the morphologic features of a tumor to effectively distinguish SCC from small cell or large cell neuroendocrine carcinoma.

Electronic supplementary material

Below is the link to the electronic supplementary material.