Recent Advances on Immunohistochemistry and Molecular Biology for the Diagnosis of Adnexal Sweat Gland Tumors

Simple Summary Cutaneous sweat gland tumors form an extremely diverse and heterogeneous group of neoplasms that show histological differentiation to the sweat apparatus. Due to their rarity, wide diagnostic range, and significant morphological overlap between entities, their accurate diagnosis remains challenging for pathologists. Until recently, little was known about the molecular pathogenesis of adnexal tumors. Recent findings have revealed a wide range of gene fusions and other oncogenic factors that can be used for diagnostic purposes and, for some, can be detected by immunohistochemistry. Among other organs containing exocrine glands, such as salivary glands, breasts, and bronchi, most of these biomarkers have been reported in homologous neoplasms that share morphological features with their cutaneous counterparts. This review aims to describe these recent molecular and immunohistochemical biomarkers in the field of sweat gland tumors. Abstract Cutaneous sweat gland tumors are a subset of adnexal neoplasms that derive or differentiate into the sweat apparatus. Their great diversity, rarity, and complex terminology make their pathological diagnosis challenging. Recent findings have revealed a wide spectrum of oncogenic drivers, several of which are of diagnostic interest for pathologists. Most of these molecular alterations are represented by gene fusions, which are shared with other homologous neoplasms occurring in organs containing exocrine glands, such as salivary and breast glands, which show similarities to the sweat apparatus. This review aims to provide a synthesis of the most recent immunohistochemical and molecular markers used for the diagnosis of sweat gland tumors and to highlight their relationship with similar tumors in other organs. It will cover adenoid cystic carcinoma (NFIB, MYB, and MYBL1 fusion), cutaneous mixed tumor (PLAG1 fusion), cylindroma and spiradenoma and their carcinomas thereof (NF-κB activation through CYLD inactivation or ALKP1 hotspot mutation), hidradenoma and hidradenocarcinoma (MAML2 fusion), myoepithelioma (EWSR1 and FUS fusion), poroma and porocarcinoma (YAP1, MAML2, and NUTM1 fusion), secretory carcinoma (ETV6, NTRK3 fusion), tubular adenoma and syringo-cystadenoma papilliferum (HRAS and BRAF activating mutations). Sweat gland tumors for which there are no known molecular abnormalities will also be briefly discussed, as well as potential future developments.


Introduction
Benign and malignant cutaneous adnexal tumors are rare [1]. They span a wide variety of different diagnoses, classified according to their apocrine, eccrine, follicular, and sebaceous differentiation. Similarly, these tumors have a wide range of prognoses, including benign tumors, low-grade locally recurring tumors, and aggressive carcinomas prone to metastases.
An update of the nosology of tumors of cutaneous appendages has been proposed by the fourth edition of the WHO classification of cutaneous neoplasms [2]. At the time of its publication, many clinicopathological entities lacked a specific and robust diagnostic hallmark, but recent discoveries have revealed new biomarkers with high diagnostic value. Many of these markers are related to specific molecular alterations demonstrated in these neoplasms.
Most skin carcinomas, especially basal and squamous cell carcinomas, exhibit an ultraviolet mutational signature and a high tumor mutational burden, which are considered key oncogenic events in their initiation and progression [3,4]. On the contrary, adnexal tumors, which occur primarily deep in the skin, are less prone to UV exposure. Adnexal tumors are, therefore, prone to stochastic but recurrent genetic alterations, such as missense mutations and oncogenic gene fusions. Interestingly, many of these molecular alterations can also be observed in other organs containing exocrine glands, i.e., salivary glands, lacrimal glands, lung, pancreas, and breast. Therefore, cutaneous adnexal tumors are frequently analogous to extracutaneous tumors with similar morphological, phenotypic, and molecular characteristics but with different, sometimes confusing, terminologies. For example, in salivary glands, the term "cylindroma" has been used as a synonym of adenoid cystic carcinoma and, while adenoid cystic carcinoma cases of the skin are analogous to their salivary counterparts, with recurrent fusion involving the MYB gene, skin cylindromas represent a distinct tumor entity with frequent CYLD mutation. Furthermore, some of these biomarkers are related to a whole group of adnexal neoplasms since they are detected in benign and malignant cases, while some evidence suggests a multistep progression from a benign precursor with a specific molecular background to their malignant counterpart. Histopathological classifications attempted to separate the wide variety of adnexal neoplasms based on their differentiation to specific parts of the skin appendages, including apocrine, eccrine, follicular, sebaceous, and multilineage differentiation. The cell of origin remains the subject of debate and investigation for many neoplasms. The difficulty in determining the cell of origin is related in part to the biology of many fusion-driven sweat gland tumors. Indeed, chimeric fusion proteins confer a new, often aberrant, histological phenotype that lacks a physiological cellular counterpart to correlate with for classification purposes. Hopefully, most of these gene fusions are highly specific in each morphological context, being mostly restricted to a single subtype of tumor.
Finally, sweat gland tumors lacking known recurrent molecular abnormalities will be briefly discussed in the last paragraph of this review. Related terminology: The designation adenoid cystic carcinoma is used regardless of the location considered. Confusingly, the term 'cylindroma' has historically been used to refer to salivary gland adenoid cystic carcinoma.
Discordant or controversial data: Contrary to the high specificity reported by numerous studies [5,6,12,16,19,20], MYB::NFIB fusions have been reported in tumors diagnosed as wild-type CYLD cutaneous cylindroma in a single report [28], but this finding was not further confirmed [29]. Since cylindroma and adenoid cystic carcinoma show significant histopathological overlap, the possibility of MYB fusion in bona fide cutaneous cylindroma is highly questionable.

Cutaneous Mixed Tumor (Chondroid Syringoma)
General features: Cutaneous mixed tumor (chondroid syringoma, Figure 2) is a multiphenotypic epithelial, myoepithelial, and mesenchymal neoplasm.  The stroma is often fibrous, myxoid, chondroid, lipogenic, and osteogenic, imparting a distinctive morphology. Apocrine and eccrine-type cutaneous mixed tumors have been described. In addition to the sweat gland component, areas of follicular and/or matrical differentiation are frequent. Myoepithelioma is a related neoplasm lacking ductal differentiation (cf. myoepithelioma).
Related terminology: The homologous neoplasm in the salivary gland, breast, and lung is called pleomorphic adenoma.
Immunohistochemistry: Like adenoid cystic carcinoma, a cutaneous mixed tumor shows epithelial and myoepithelial differentiation but also a stromal component. The panel of antibodies to highlight its multiple cell populations includes SOX10, p63, PS100, calponin, SMA, EMA, c-KIT, CEA, and various cytokeratins [12][13][14]. Follicular differentiation stains with BerEP4 and PHDLA1. Although not evaluated in cutaneous cases, SOX10 is highly sensitive for the diagnosis of pleomorphic adenoma but lacks specificity [8,9]. A more specific marker is PLAG1 (clone 3B7), which is overexpressed in 87 to 100% of cutaneous mixed tumors, while rare cases show HMGA2 expression [30,31]. Since cutaneous mixed tumors of eccrine-type lack PLAG1 expression, it is unclear if cutaneous mixed tumors of apocrine and eccrine-type belong to the same spectrum until more studies are conducted [30]. PLAG1 is not expressed in soft tissue myoepithelioma lacking ductal structures [32], and has not been specifically tested in cutaneous myoepithelioma. Consistent with these findings, PLAG1 and HMGA2 show high sensitivity and specificity for the diagnosis of salivary pleomorphic adenoma and its related carcinoma [33][34][35]. The precise performance of PLAG1 and HMGA2 in skin cases requires larger studies.
Molecular biology: Recurrent rearrangements of PLAG1 are detected in 33% of cutaneous mixed tumors by FISH [31], which is a developmentally regulated zinc-finger protooncogene. PLAG1 activation through gene fusion typically occurs due to promoter swapping from various fusion partner genes that are ubiquitously expressed, such as CTNNB1, LIFR, and CHCHD7. This explains why pleomorphic adenomas of the salivary gland demonstrate a wide range of PLAG1 and HMGA2 fusions: CTNNB1::PLAG1, LIFR::PLAG1, CHCHD7::PLAG1, TCEA1::PLAG1, HMGA2::FHIT, HMGA2::NFIB, and HMGA2::WIF1. Although NDRG1::PLAG1 and TRPS1::PLAG1 fusions have been described in cutaneous cases [31], another study did not detect the fusions found in the salivary gland in cutaneous cases, suggesting different gene rearrangements or mechanisms [36]. To date, no HMGA2 fusion has been reported in cutaneous mixed tumors. More studies are required to define more precisely the genetic landscape of cutaneous mixed tumors and, in particular, their relationship with myoepithelioma when their myoepithelial component is predominant (cf. myoepithelioma).
Discordant or controversial data: A PHF1::TFE3 fusion transcript was detected in one case morphologically diagnosed as malignant chondroid syringoma [37]. However, it remains to be determined whether this tumor actually belonged to the mixed tumor spectrum or was in fact a superficial ossifying fibro-myxoid tumor.

Cylindroma and Spiradenoma
General features: Cylindroma and spiradenoma ( Figure 3) are related epithelial and myoepithelial skin neoplasms, associated with activation of the NF-κB pathway, which is triggered by inactivation of CYLD or ALPK1 mutations. Spiradenoma differs from cylindroma by its nodular architecture and lymphocytic infiltrate. On rare occasions, both neoplasms show a morphological and phenotypical overlap with adenoid cystic carcinoma. Most cases occur sporadically; however, some cases arise in the context of a predisposition syndrome known as Brooke-Spiegler syndrome. Multiple cylindromas, spiradenomas, trichoepitheliomas, and basal cell salivary gland tumors occur in this autosomic dominant disorder linked to CYLD mutations.
of CYLD [42]. Most cases of cylindroma carry germline or somatic CYLD mutations compared to approximately a third of cases of spiradenoma [29]. Disruptive mutations in DNMT3A have also been described in cylindroma [29]. ALPK1 mutations involve predominantly a hotspot in the alpha-kinase domain (p.V1092A) and are detected in 43% and 28% of spiradenomas and spiradenocarcinomas, respectively. Consistent with a model of progression and transformation from precursor spiradenoma, loss-of-function TP53 mutations are restricted to spiradenocarcinoma [43].  Related terminology: Homologous neoplasms in the salivary gland and breast are called basal cell adenoma (membranous variant of) and adenocarcinoma. Confusingly, the term 'cylindroma' has historically been used in the salivary gland to describe adenoid cystic carcinoma.
Molecular biology: Spiradenoma and cylindroma are related to oncogenic activation of the NF-κB pathway, triggered by mutually exclusive mutations of ALPK1 and CYLD [29]. A wide range of CYLD mutations affect coding and splicing, resulting in inactivation of CYLD [42]. Most cases of cylindroma carry germline or somatic CYLD mutations compared to approximately a third of cases of spiradenoma [29]. Disruptive mutations in DNMT3A have also been described in cylindroma [29]. ALPK1 mutations involve predominantly a hotspot in the alpha-kinase domain (p.V1092A) and are detected in 43% and 28% of spiradenomas and spiradenocarcinomas, respectively. Consistent with a model of progression and transformation from precursor spiradenoma, loss-of-function TP53 mutations are restricted to spiradenocarcinoma [43].
Discordant or controversial data: MYB fusions are highly specific and are considered the hallmark of ACC [5,6,12,16,19,20], and have not been detected in CYLD-defective cylindroma [44]. A single study has reported that tumors diagnosed as CYLD-proficient dermal cylindroma demonstrated MYB fusions [28], which was not confirmed in a subsequent large-scale study [29] (cf. adenoid cystic carcinoma). Similar to adenoid cystic carcinoma, CYLD-defective cylindromas, however, demonstrate upregulation and overexpression of MYB and its target genes [44], suggesting the involvement of alternative mechanisms than MYB fusion [29]. CYLD mutations have recently been described in a subset of high-risk HPV-positive head and neck squamous cell and anal carcinomas, which show histological features close to cylindroma and have been designated as CYLD-mutant cylindroma-like basaloid carcinoma [45,46].

Hidradenoma
General features: Hidradenoma ( Figure 4)  Discordant or controversial data: MYB fusions are highly specific and are considered the hallmark of ACC [5,6,12,16,19,20], and have not been detected in CYLD-defective cylindroma [44]. A single study has reported that tumors diagnosed as CYLD-proficient dermal cylindroma demonstrated MYB fusions [28], which was not confirmed in a subsequent large-scale study [29] (cf. adenoid cystic carcinoma). Similar to adenoid cystic carcinoma, CYLD-defective cylindromas, however, demonstrate upregulation and overexpression of MYB and its target genes [44], suggesting the involvement of alternative mechanisms than MYB fusion [29]. CYLD mutations have recently been described in a subset of high-risk HPV-positive head and neck squamous cell and anal carcinomas, which show histological features close to cylindroma and have been designated as CYLD-mutant cylindroma-like basaloid carcinoma [45,46].  Related terminology: The homologous neoplasm in the breast, salivary gland, lung, and pancreas is mucoepidermoid carcinoma.

Hidradenoma
Immunohistochemistry: Hidradenoma does not demonstrate a specific phenotype due to the combination of different types of cells. They express strongly and diffusely p40, Related terminology: The homologous neoplasm in the breast, salivary gland, lung, and pancreas is mucoepidermoid carcinoma.
Immunohistochemistry: Hidradenoma does not demonstrate a specific phenotype due to the combination of different types of cells. They express strongly and diffusely p40, p63, CK5/6, and AE1/AE3, while S100 and SMA are usually negative [47,48]. Immunostaining for EMA and CEA highlights ductal differentiation when present. Data on SOX10 expression are currently lacking, but, in our experience, SOX10 is constantly negative. Higher Ki67 and PHH3 labeling index and p53 expression have been reported in malignant cases [49].
Molecular biology: Half to 75% of the cases of hidradenoma have CRTC1::MAML2, and, more rarely, CRTC3::MAML2 fusions [50]. Similarly, salivary gland mucoepidermoid carcinoma demonstrates 34-70% of CRTC1::MAML2, and, more rarely, CRTC3::MAML2 fusions [51][52][53]. Hidradenocarcinomas also demonstrate MAML2 rearrangement but with lower frequencies than their benign counterparts, while a subset carries TP53, AKT1, PI3KCA mutations, and Her2/neu amplification [54,55]. CREB-related transcriptional coactivator 1 (CRTC1) is located on chromosome 19 and activates cAMP-responsive element binding protein (CREB) signaling. Mastermind-like 2 (MAML2) is located on chromosome 11 and is associated with the Notch signaling pathway. EGFR overexpression has been found in the whole spectrum of cutaneous hiradenoma, namely hidradenocarcinoma, atypical hidradenoma, and hidradenoma [49]. However, a study investigating the amplification of EGFR in cutaneous cases lacked to demonstrate a clear correlation between the protein expression and the amplification of EGFR. Nonetheless, the EGFR signaling could still represent the mechanism of action of the CRTC1::MAML2 fusion protein since the CRTC1 portion of the chimeric protein acts on AREG, which is a ligand of EGFR. In addition, some cases of salivary mucoepidermoid carcinoma that lack CRTC1::MAML2 fusion rather demonstrate amplification and overexpression of EGFR and higher grade [56]. In this setting, therapies targeting the EGFR signaling axis are being investigated, such as Gefitinib [57]. Considering the current lack of data, the exact incidence of MAML2 gene fusion in hidradenoma and hidradenocarcinoma requires more studies.
Discordant or controversial data: EWSR1::POU5F1 fusion has been described in tumors diagnosed as hidradenoma in a single report [58] but has not been further confirmed [59]. A subsequent study has shown that this fusion defined a subset of myoepithelial tumors that can be mistaken for hidradenoma due to an extensive clear cell morphology [60] (cf. myoepithelioma).

Microcystic Adnexal Carcinoma
General features: Microcystic adnexal carcinoma (MAC) predominantly affects the sun-exposed area of the head and neck and has a high recurrence rate, slow growth, and low metastatic potential. MAC is an asymmetric and deeply infiltrative tumor that can present three components arranged in a vertical gradient, some of which may be lacking: small keratinizing cysts on the surface, solid aggregates in the center of the lesion, and, finally, deep tubular structures. As the histology of MAC in the superficial part of the skin resembles the histology of syringoma and basal cell carcinoma, an accurate diagnosis of MAC can be difficult on superficial biopsy. No specific molecular or immunohistochemical hallmark has yet been discovered.
Related terminology: Microcystic adnexal carcinoma is specific to the skin and lacks a homologous neoplasm in other organs. Solid carcinoma, reported in a series of 14 cases, may represent a variant of microcystic adnexal carcinoma [61]. A low-grade variant has recently been described as microcystic adnexal adenoma [62]. Hamartomatous lesions similar to MAC but with benign behavior have recently been reported in MALTA syndrome related to germline MYH9 mutations [63].
Immunohistochemistry: Immunostains aim to differentiate between MAC, desmoplastic trichoepithelioma, basal cell carcinoma, and squamous cell carcinoma. In this setting, BerEP4, p63, and Ki67 can be used. BerEP4 shows negative or very infrequent staining in MAC, including the solid variant, while strong BerEP4 staining suggests basal cell carcinoma [61,[64][65][66][67][68]. Immunostaining for p63 highlights scattered positive cells in the upper part of the neoplasm, but the deepest part of the MAC can be negative. On the contrary, p63 expression is diffuse in basal and squamous cell carcinoma [69,70]. Low-proliferation, i.e., Ki67 < 5%, favors MAC over basal and squamous cell carcinoma [68,71]. The luminal cells of the ductal structures are stained with CK7 and CEA, but this does not discriminate between other sweat gland neoplasms with ductal differentiation. PHLDA1 does not help to differentiate MAC from DTE, CK20 is usually negative [72], and SOX10 is almost always negative [73].
Molecular biology: Analysis of mutation burden, UV signature, mutational status of 400 cancer-relevant genes, and copy number variations showed few recurrent somatic mutations and, in most cases, a lack of an ultraviolet signature, a counterintuitive finding regarding the preferential occurrence of MAC in chronically sun-exposed skin [12]. The same study reported inactivating TP53 mutations and frame-preserving indel mutations located in the kinase domain of JAK1, copy number alterations with gain of JAK2, MAF, MAFB, JUN, or FGFR1, and loss of CDKN2A. These data were consistent with another study in which sequencing indicated a mutation of TP53 with loss of CDKN2A and CDKN2B in metastatic MAC [74]. Another study reported that, unlike basal cell carcinoma, Hedgehog signaling was not significantly altered in MAC [75]. Considering its lack of UV signature and since no oncogenic mutation was identified in one third of the cases in the largest study to date, more studies are required, including the detection of gene fusions, to reveal the genomic landscape of MAC.

Myoepithelioma
General features: Myoepithelioma ( Figure 5) is a rare neoplasm that occurs on the skin, soft tissue, salivary gland, and breast that is composed exclusively of myoepithelial cells. Half of extracutaneous cases are associated with various rearrangements of EWSR1 or FUS [76]. Unlike mixed tumors, myoepitheliomas are defined by the lack of ductal structures. Half of cutaneous cases exhibit a distinctive skin-specific syncytial sheet-like pattern and harbor a EWSR1::PBX3 fusion. Malignant cases are extremely rare.
Related terminology: The term myoepithelioma is used regardless of the organ considered. Depending on the classification, myoepithelial tumors are considered mesenchymal.
Molecular biology: EWSR1 belongs to the TET family of proteins, a highly conserved group of multifunctional, RNA-binding proteins that includes fused in sarcoma (FUS). EWSR1 and FUS are related in both structure and function, are expressed in most cell and tissue types, and serve a multifunctional purpose and predominantly reside in the nucleus. Fusion proteins involving EWSR1 and FUS act as aberrant transcription factors (EWSR1/FUS providing the transcriptional activation domain and the other gene involved usually contributing the DNA-binding domain). While EWSR1 gene fusions are lacking in cutaneous mixed tumors, many recurrent gene fusions involving the EWSR1 and FUS genes have been reported in cutaneous and soft tissue myoepithelioma [60], with a proportion of 82% and 18%, respectively. EWSR1::PBX3 fusions are more frequent in syncytial myoepithelioma compared to other cutaneous myoepitheliomas [80][81][82]. In extracutaneous myoepitheliomas, EWSR1::POU5F1 and EWSR1::PBX3 are the two most frequent fusions [76], while FUS::KLF17, EWSR1::PBX1, EWSR1::ZNF444, EWSR1::KLF15, FUS::POU5F1, EWSR1::KLF17, EWSR1::FUS, EWSR1::ATF1, and EWSR1::VGLL1 are less frequent [60,76,[83][84][85][86][87][88]. POU5F1-rearranged myoepitheliomas show a distinctive clear cell appearance that can be mistaken for hidradenoma, and PBX1-rearranged myoepitheliomas show sclerosis [60]. Loss of SMARCB1 (INI1) is observed in a subset of malignant cases but has not been reported on the skin [89][90][91]. Discordant or controversial data: EWSR1::POU5F1 translocation is frequent in a subset of myoepithelial tumors with frequent extensive clear cell morphology that can be mistaken for hidradenoma [60]. In this context and before the description of this fusion in clear cell myoepithelial neoplasms, EWSR1::POU5F1 had been reported in tumors diagnosed as hidradenoma in a single report [58] but was not confirmed afterwards [59] (cf.: hidradenoma). PLAG1 fusions have been reported in myoepithelioma with ductal structures, and it is not yet clear whether these neoplasms are distinct from mixed tumors with a prominent myoepithelial component [92]. Discordant or controversial data: EWSR1::POU5F1 translocation is frequent in a subset of myoepithelial tumors with frequent extensive clear cell morphology that can be mistaken for hidradenoma [60]. In this context and before the description of this fusion in clear cell myoepithelial neoplasms, EWSR1::POU5F1 had been reported in tumors diagnosed as hidradenoma in a single report [58] but was not confirmed afterwards [59] (cf.: hidradenoma). PLAG1 fusions have been reported in myoepithelioma with ductal structures, and it is not yet clear whether these neoplasms are distinct from mixed tumors with a prominent myoepithelial component [92].
Molecular biology: Recurrent YAP1 rearrangements have been detected in 88% and 63% of cases of poroma and porocarcinoma, respectively [109], with subsequent reports detecting a YAP1::MAML2 fusion in porocarcinoma [110] and in parotid gland squamoid porocarcinoma [111]. Yes1 associated transcriptional regulator (YAP1) is a transcriptional coactivator whose activity is controlled by the Hippo signaling pathway, regulating homeostasis and regeneration. YAP1 is also involved in cancer initiation, aggressiveness, metastasis, and therapy resistance [112]. Other fusions include YAP1::NUTM1 and WWTR1::NUTM1. Consistent with immunohistochemical data, a single study reported that YAP1::NUTM1 fusion was more common in benign and malignant poroid hidradenoma than in poroma and porocarcinoma [95]. BRD3/4::NUTM1 fusion has been described in exceptional cases of primary or metastatic cutaneous adnexal carcinoma [98,99]. All the fusions harbor the Nterminal TEAD-binding domain of YAP1 or WWTR1 and the MAML2-or NUTM1-derived regions that interact with transcriptional coactivators CBP and p300 [109]. Porocarcinomas also carry mutations involving oncogenes and tumor suppressor genes, such as CDKN2A, EGFR, ERBB2, FGFR3, HRAS, KRAS, NRAS, PIK3CA, TP53, and RB1 [109,[113][114][115]. A recent cohort of morphologically diagnosed porocarcinoma reported a high mutational burden related to UV exposure, recurrent copy number alterations, including losses in chromosomal regions containing BRCA2, and deleterious alterations in multiple homologous recombination repair pathway components. Although gene fusions in this study were not tested to confirm diagnosis, these data encourage investigation of targeting the p53 axis, PARP inhibition, and immunotherapy in the treatment of porocarcinoma [116].

Secretory Carcinoma
General features: Cutaneous secretory carcinoma (Figure 7) is a rare cutaneous NTRKrearranged carcinoma that shows distinctive intracytoplasmic secretory vacuoles and bubbly extracellular eosinophilic secretions.
Related terminology: Secretory carcinoma occurs in the breast, salivary gland, lacrimal gland, lung, and skin. Cutaneous terminology includes 'primary cutaneous mammary analogue secretory carcinoma' (MASC) and 'mammary-type secretory carcinoma of the skin'.
Immunohistochemistry: NTRK fusions can be detected by immunohistochemistry using a panTRK antibody (clone EPR 17341) with various performances and staining patterns, with a sensitivity up to 100% for secretory carcinoma [117][118][119]. When fusion involves ETV6, TRK immunohistochemistry is expected to stain the nucleus, while the cytoplasm and/or membrane are stained when alternative fusion partners are involved [120]. Other positive stains include SOX10, MUC4, S100, Mammaglobin, and STAT5A, while TTF1, DOG1, and p63 are negative [8,121,122].

Tubular Adenoma and Syringocystadenoma Papilliferum
General features: Tubular adenoma and syringocystadenoma papilliferum (SCAP, Figure 8) are related entities with an epithelial myoepithelial phenotype and tubulopapillary architecture [146]. Both are related to activation of mutations in the mitogen-activated protein kinase (MAPK) signaling pathway, affecting BRAF, KRAS, and HRAS. Both tubular adenoma and syringocystadenoma papilliferum can develop sporadically or more frequently on the genetic background of a cutaneous mosaic RASopathy, such as nevus sebaceus of Jadassohn [147,148].
Related terminology: The salivary and bronchus homologous neoplasm is called sialadenoma papilliferum.
Immunohistochemistry: Immunostains can be used successfully to detect cases with BRAF p.V600E mutation, with excellent sensitivity and specificity with the anti-BRAF p.V600E antibody (clone VE1) [149]. Epithelial myoepithelial differentiation of SCAP and tubular adenoma can be revealed by a panel including SMA, calponin, SOX10, and p63 to highlight the outer myoepithelial layer and luminal cytokeratins and EMA for the inner layer. Most cases are associated with plasma cells within the papillary axis, which can be stained with MUM1 and CD138.

Tubular Adenoma and Syringocystadenoma Papilliferum
General features: Tubular adenoma and syringocystadenoma papilliferum (SCAP, Figure 8) are related entities with an epithelial myoepithelial phenotype and tubulopapillary architecture [146]. Both are related to activation of mutations in the mitogen-activated protein kinase (MAPK) signaling pathway, affecting BRAF, KRAS, and HRAS. Both tubular adenoma and syringocystadenoma papilliferum can develop sporadically or more frequently on the genetic background of a cutaneous mosaic RASopathy, such as nevus sebaceus of Jadassohn [147,148].
Related terminology: The salivary and bronchus homologous neoplasm is called sialadenoma papilliferum.
Immunohistochemistry: Immunostains can be used successfully to detect cases with BRAF p.V600E mutation, with excellent sensitivity and specificity with the anti-BRAF p.V600E antibody (clone VE1) [149]. Epithelial myoepithelial differentiation of SCAP and tubular adenoma can be revealed by a panel including SMA, calponin, SOX10, and p63 to highlight the outer myoepithelial layer and luminal cytokeratins and EMA for the inner layer. Most cases are associated with plasma cells within the papillary axis, which can be stained with MUM1 and CD138. Molecular biology: BRAF p.V600E mutations are detected in 50-64% of syringocystadenomas papilliferum and 66% of tubular adenomas, respectively. Tumors that arise in nevus sebaceus almost constantly harbor HRAS or KRAS mutations [150]. Activating mutations of BRAF, HRAS, and KRAS increase cell growth, differentiation, and survival. In sporadic tumors, HRAS p.G13R mutations are found in 7-26% of cases, while KRAS p.G12D mutations are less frequent [151,152]. DNA of HPV type 16 and 68 have been detected in SCAP of the perianal area and buttock, without HPV-related morphology [153]. Verrucous proliferations associated with SCAP demonstrate BRAF mutation in both the glandular and contiguous hyperplastic squamous epithelium [149,154,155]. Sialadenoma papilliferum exhibits the same morphology and carries BRAF p.V600E mutation in most cases, while reports of RAS mutations are currently lacking [156,157]. No data have been published on malignant SCAP, but various mutations have been reported in a case report of metastatic adenocarcinoma with papillary architecture, presumably originating from a SCAP that included ARID2, CDKN2A, ERBB4, FAT1, FGF10, FGFR1, GNA11, KDM5C, MAGI2, NFKBIA, NOTCH1, PIK3CA, RAC1, RICTOR, SLIT2, TERT, TP53 mutations. More studies are needed to discover the molecular landscape of the malignant counterpart of SCAP. A case of digital papillary adenocarcinoma with BRAF p.V600E has been published but could represent a malignant form of tubular adenoma [158]. In this context, inhibitors of BRAF could be employed. Molecular biology: BRAF p.V600E mutations are detected in 50-64% of syringocystadenomas papilliferum and 66% of tubular adenomas, respectively. Tumors that arise in nevus sebaceus almost constantly harbor HRAS or KRAS mutations [150]. Activating mutations of BRAF, HRAS, and KRAS increase cell growth, differentiation, and survival. In sporadic tumors, HRAS p.G13R mutations are found in 7-26% of cases, while KRAS p.G12D mutations are less frequent [151,152]. DNA of HPV type 16 and 68 have been detected in SCAP of the perianal area and buttock, without HPV-related morphology [153]. Verrucous proliferations associated with SCAP demonstrate BRAF mutation in both the glandular and contiguous hyperplastic squamous epithelium [149,154,155]. Sialadenoma papilliferum exhibits the same morphology and carries BRAF p.V600E mutation in most cases, while reports of RAS mutations are currently lacking [156,157]. No data have been published on malignant SCAP, but various mutations have been reported in a case report of metastatic adenocarcinoma with papillary architecture, presumably originating from a SCAP that included ARID2, CDKN2A, ERBB4, FAT1, FGF10, FGFR1, GNA11, KDM5C, MAGI2, NFKBIA, NOTCH1, PIK3CA, RAC1, RICTOR, SLIT2, TERT, TP53 mutations. More studies are needed to discover the molecular landscape of the malignant counterpart of SCAP. A case of digital papillary adenocarcinoma with BRAF p.V600E has been published but could represent a malignant form of tubular adenoma [158]. In this context, inhibitors of BRAF could be employed.

Endocrine Mucin-Producing Sweat Gland Carcinoma
General features: Endocrine mucin-producing sweat gland carcinoma (EMPSGC, Figure 9) is a low-grade mucin-producing sweat gland carcinoma with a neuroendocrine phenotype and a predilection for the eyelids and periorbital skin [159,160].
Low-grade neuroendocrine carcinoma of the skin/primary cutaneous carcinoid tumors might represent a variant or a closely related entity [161].
Related terminology: The related homologous neoplasm is called solid papillary adenocarcinoma of the breast. Endocrine mucin-producing sweat gland carcinoma may represent a precursor of mucinous sweat gland carcinoma (neuroendocrine type) [159,160]. Low-grade neuroendocrine carcinoma of the skin/primary cutaneous carcinoid tumors might represent a variant or a closely related entity [161].

Main Clinical Settings to Use These Biomarkers and Their Limits
Although some of these biomarkers are directly linked to a specific diagnosis, other markers are associated with a whole spectrum. In fact, demonstrating a rearrangement of MYB, MYBL1, PLAG1, HMGA2, NTRK3, and a BRAF mutation in the appropriate morphological context is diagnostic of adenoid cystic carcinoma, cutaneous mixed tumor, secretory carcinoma, syringocystadenoma papilliferum, and tubular adenoma. On the other hand, the demonstration of fusion of YAP1, NUTM1, MAML2, EWSR1, FUS, and mutation of CYLD and ALPK1 points to the poroid, hidradenomatous, myoepithelial, cylindromatous, and spiradenomatous categories of tumors, and these alterations do not predict prognosis or malignancy, which is based on other characteristics, such as morphology and proliferation. Other alterations are used for diagnostic and theranostic purposes, such as the BRAF p.V600E mutation and the fusion of NTRK3.
Some cases still lack any detectable molecular or immunohistochemical hallmark, and there is not a single answer to this question: variation in the sensitivity of the molecular techniques used, unreported gene fusion, alternative molecular mechanisms, and misinterpretation of pathological findings may contribute to the variation in prevalence of observed anomalies. The exact prevalence of these molecular anomalies is still unknown given the lack of data and large studies that include bona fide cases.

Other Sweat Gland Tumors Lacking Recent Data
Numerous other sweat gland tumors and related carcinomas still lack extensive molecular or immunohistochemical data, notably histiocytoid carcinoma, papillary digital adenocarcinoma, polymorphous sweat gland carcinoma, squamoid eccrine ductal carcinoma, mucinous carcinoma, apocrine carcinoma, and adnexal adenocarcinomas without other specification (NOS). This lack of data is the consequence of the combination of their rarity and diversity, which prevents large-scale studies from focusing on their molecular background. Networks of pathologists with common and robust diagnostic criteria, collaborative work, and collections of these rare neoplasms are of utmost importance to improve their diagnosis and allow proper future research [1,170,171].

Conclusions
The molecular landscape of cutaneous adnexal tumors has changed dramatically in the past few years. The addition of robust biomarkers in an integrated molecular-pathological approach is more likely to positively impact the reproducibility of diagnoses. Due to their great variety, rarity, complex terminology, and overlapping microscopic and phenotypic features, the pathological diagnosis of rare cutaneous neoplasms is challenging. However, these novel molecular and immunohistochemical hallmarks are expected to improve overall diagnostic accuracy due to their high specificity when integrated into an appropriate clinical and pathological context. Ultimately, since the entities will be systematically validated with a specific hallmark, standardized diagnoses will pave the way for future studies investigating the prognosis and therapeutic management of these rare cutaneous tumors.