Microscopically identified regional lymph node involvement is the most critical independent unfavorable prognostic marker in MCC patients [16,17,18]. Santamaria-Barria et al. [19] revealed that one-third of early-stage patients (stages I and II) had micrometastases in sentinel lymph nodes (in 9 out of 27 patients (33%)) and 39% (7/18) of SLNB-negative patients developed recurrence. These clinical findings indicate that more effective histopathologic methods of detection of MCC nodal micrometastases are needed.

MCC metastasis detection requires the implementation of a uniform, highly reproducible protocol, which combines histologic examination followed by IHC stains according to a standardized algorithm [1, 20]. Our multi-institutional and comprehensive study analyses compare different IHC markers in detecting MCC metastases. We showed that a panel of three antibodies (SATB2, CKAE1/AE3, and synaptophysin) is the most helpful in routine pathological assessment. We recommend revising the MCC metastasis diagnostic guidelines since there is no one precise IHC protocol [20]. We highlight the importance of using optimal IHC supporting tools as finding even a single MCC metastatic cell changes the clinical staging and influence on treatment modification.

According to our results, the most promising antibody is SATB2, belonging to the SATB family proteins, which are fundamental regulators of gene expression and modifiers of chromatin structure [21]. We found that SATB2 was positive in all nodal metastases; nearly 90% of cases demonstrated positivity in ≥ 75% of tumoral cells. The stain interpretation was simple since its nuclear expression remained uniformly moderate to strong. In previous studies, SATB2 was described as an effective marker in the differential diagnosis of MCC and extracutaneous neuroendocrine carcinomas [10, 22, 23]. In skin, Fukuhara et al. [24] showed that normal Merkel cells present SATB2 reactivity, and their results confirmed the diagnostic utility of SATB2 in distinguishing MCC from other skin neoplasms. Kervarrec et al. [22] showed a high diagnostic accuracy of immunohistochemistry for SATB2 in detecting Merkel cell carcinoma versus extracutaneous neuroendocrine carcinomas with high specificity (98%) and a positive likelihood ratio of 36.6.

INSM1 is the second recently described protein tested in MCC detection [7]. Lilo et al. [11] showed that INSM1 stained > 75% of tumor nuclei in 89% of MCC, and it was influential in the distinction between MCC and other primary cutaneous neoplasms but not discriminating MCC from metastatic neuroendocrine carcinomas of extracutaneous origin. In our study, all metastases were INSM1 positive, with 73% cases with cut-off ≥ 75% stained nuclei. Surprisingly, there was no 100% INSM1 positive MCC, a significant limitation of its usage in SLNB evaluation; an unexpected observation was entirely negative mitotically active tumoral cells.

Cytokeratins have been recommended for MCC metastasis detection over the years [25, 26]. From various cytokeratins, CK20 with characteristic perinuclear dot-like positivity is the most specific [27,28,29,30]. Tetzlaff et al. [31] demonstrated the total percentage of CK20-positive vs. CK20-negative MCC as 87.4% vs. 12.6%, respectively. In our study, the IHC with the highest completely negative IHC was CK20 and comprised 7.8% of cases; moreover, 22.6% displayed positivity in < 25% of cells. Interpreting single dot-like reactions in lymph nodes is challenging; additionally, dendritic cells might express cytokeratins, but identifying spindle shape on closer inspection is supportive. We showed a superior diagnostic value of CKAE1/AE3 over the CK20, which was indicated by (1) a significantly higher percentage of tumoral cells with moderate to strong expression (83% vs. 59% of cases for CKAE1/AE3 vs. CK20, respectively), and (2) the total percentage of 100% positive cases, which constituted 32% for CKAE1/AE3 and 5% for CK20. We propose that wide-spectrum cytokeratin (CKAE1/AE3) should be maintained as the second antibody of choice (following the SATB2) for identifying MCC metastases.

Providing neuroendocrine origin of metastasis is required for MCC diagnosis confirmation. Kervarrec et al. [22] showed that NF is the most specific for MCC among all neuroendocrine markers. Most published NF images focus on dot-like reactions; our study proved that NF was uniformly expressed in 77.45% of cases presented as the scattered, dot-like reaction in 1–25% of MCC cells. In our opinion, this result and the fact there were no cases with high expression pattern (≥ 75% positive cells) disqualifies NF as a marker for SLNB assessment. By contrast, synaptophysin was a stable, positive IHC neuroendocrine marker in all cases, with no negative metastatic MCC lymph nodes. Although chromogranin A showed a higher number of 100% positive cases compared to synaptophysin (27.45% and 16.67%, respectively), it is not recommended for the histopathologic lymph node assessment because of its heterogeneity, presence of negative metastatic lymph nodes, and focally significant unspecified background of IHC reaction.

Importantly, a questionable aspect of all recommendations is the methodology of IHC interpretation. We used a highly reproducible 5-tier scoring system. On the contrary, implementing the H-score method in everyday practice is complex and characterized by the highest deviation among investigators [9, 32]. On the other hand, some simplified protocols are limited to reporting the lack of positive stain, or positive status was evaluated only on low microscopic magnification [33]. From our experience, we consider that it might lead to misinterpretation; the population of reactive lymphocytes may show weak to moderate SATB2 expression, which might be confused with MCC metastases. Careful comparison with the cell morphology at higher magnification (200 × or 400 ×) and/or the re-evaluation of the case with additional IHC (CKAE1/AE3 or synaptophysin) is helpful. In addition, a single MCC cell qualifies a node as metastatic, so there is an evident need to use higher magnification.

We acknowledge that there are some challenging pitfalls in MCC SLNB diagnostics. The nonspecific patterns observed for SATB2 and pan-keratin for detecting MCC would not be problematic in most cases. However, in the context of single metastatic cells or minute clusters, the interpretive difficulty could emerge; some pathologists might encounter a mildly enlarged and moderately SATB2-positive nucleus that is not clearly tumor vs. lymphoid or puncta of keratin that could be either a paranuclear dot or tangential cut of dendritic labeling. Such nonspecific patterns also increase the chance that single cells are missed as the pathologist examines across the lymph node whole sections. For this reason, pan-keratin has not been a stand-alone immunohistochemical marker for SLNB evaluation in MCC. By contrast, CK20 is highly specific, allowing for much greater confidence in detecting and diagnosing single-cell metastases. These practical considerations account for the frequent use of combined pan-keratin and CK20 stains for evaluating MCC SLNB, despite the known limitations of CK20. We recognize this weakness and recommend synaptophysin as a more specific marker, which has low but not zero cross-reactivity with normal populations in lymph nodes.

The strengths of our study are the following: (1) the comparison of well-established IHC markers (CKAE1/AE3, CK20, synaptophysin, chromogranin A, and NF) with recently available new antibodies (SATB2 and INSM1) for MCC metastasis identification; (2) the high number of evaluated cases; (3) providing assessment intertumoral heterogeneity in multiple metastases; (4) emphasis on the practical aspects by implementing a user-friendly, semiquantitative, reproducible scoring system that pathologists could apply in routine diagnostics. We are aware that the limitations include (1) tissue microarray instead of whole slide evaluation and (2) lack of comparison with primary and metastasic MCC. To minimize weaknesses, we used at least two samples (cores) from one metastasis and analyzed the intratumoral heterogeneity whenever possible (multiple metastases cases). According to the literature review, the immunoprofile of MCC primary showed strong convergence with metastatic [31].

In conclusion, we have shown that SATB2 would be helpful in MCC as the first-line marker in the SLNB/LNB histopathologic assessment. We also suggest including CKAE1/AE3 and synaptophysin, which should be performed sequentially after SATB2. These are widely available antibodies that can be easily implemented; the pathological, semiquantitative interpretation is also reproducible and not particularly time-demanding. Since no specific IHC marker exists, the neuroendocrine metastases originating from other sites should be excluded.