Apocrine-Eccrine Carcinomas: Molecular and Immunohistochemical Analyses

Apocrine-eccrine carcinomas are rare and associated with poor prognosis. Currently there is no uniform treatment guideline. Chemotherapeutic drugs that selectively target cancer-promoting pathways may complement conventional therapeutic approaches. However, studies on genetic alterations and EGFR and Her2 status of apocrine-eccrine carcinomas are few in number. In addition, hormonal studies have not been comprehensive and performed only on certain subsets of apocrine-eccrine carcinomas. To investigate whether apocrine-eccrine carcinomas express hormonal receptors or possess activation of oncogenic pathways that can be targeted by available chemotherapeutic agent we performed immunohistochemistry for AR, PR, ER, EGFR, and HER2 expression; fluorescence in situ hybridization (FISH) for EGFR and ERBB2 gene amplification; and molecular analyses for recurrent mutations in 15 cancer genes including AKT-1, EGFR, PIK3CA, and TP53 on 54 cases of apocrine-eccrine carcinomas. They include 10 apocrine carcinomas, 7 eccrine carcinomas, 9 aggressive digital papillary adenocarcinomas, 10 hidradenocarcinomas, 11 porocarcinomas, 1 adenoid cystic carcinoma, 4 malignant chondroid syringomas, 1 malignant spiradenoma, and 1 malignant cylindroma. AR, ER, PR, EGFR and HER2 expression was seen in 36% (19/53), 27% (14/51), 16% (8/51), 85% (44/52) and 12% (6/52), respectively. Polysomy or trisomy of EGFR was detected by FISH in 30% (14/46). Mutations of AKT-1, PIK3CA, and TP53 were detected in 1, 3, and 7 cases, respectively (11/47, 23%). Additional investigation regarding the potential treatment of rare cases of apocrine-eccrine carcinomas with PI3K/Akt/mTOR pathway inhibitors, currently in clinical testing, may be of clinical interest.


Introduction
In a recent World Health Organization (WHO) classification of cutaneous appendageal carcinomas; apocrine-eccrine, follicular, and sebaceous carcinomas were the three main categories cited in the consensus classification after taking into account the clinical, histologic, and molecular genetic features [1]. Apocrine-eccrine carcinomas are rare and associated with poor prognosis [2][3][4]. Three of nine cases of clear cell eccrine carcinomas reported by Wong et al. [4] developed metastases despite local excision, radiation, and chemotherapy. In the largest series of 69 cases of porocarcinomas by Robson et al. [3], 17% and 19% experienced local recurrence and lymph node metastases, respectively. In recent Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute data from 1978 through 2005, the incidence rate of apocrine-eccrine carcinomas was reported to be 2.6 per 1 million person-years [2]. The five-year relative survival rates for apocrine-eccrine carcinomas were 99% for localized, 94% for regional and 51% for distant disease [2].
Due to the frequent tendency for recurrence (50%) and potential to metastasize (14%), amputation of the digit is often the treatment for aggressive digital papillary adenocarcinoma [5]. For the remaining apocrine-eccrine carcinomas, wide surgical excision is also the treatment of choice. Currently there is no uniform guideline concerning the treatment for various types of apocrine-eccrine carcinomas, especially for those with metastases. Treatment success for metastatic disease has been documented only in isolated case reports of metastasizing hidradenocarcinomas and eccrine carcinoma [6][7][8].
Targeted therapy may be a potential treatment option in patients whose tumors are characterized by a relevant oncogene mutation [4,9,10]. In a growing number of tumor types including breast, colorectal and lung cancer, selective agents that target critical cancer-promoting pathways are now the treatment of choice for those patients carrying the genetic changes recognized by the drugs [4,9,10]. Members of the ERBB receptor tyrosine kinase family, including epidermal growth factor receptor (EGFR), HER2, HER3 and HER4, also present possible targeted therapeutic options [11]. ERBB2 (v-erb-b2 erythroblastic leukemia viral oncogene homolog 2) gene amplification and response to trastuzumab were documented in a case of metastasizing hidradenocarcinoma [7]. The membranous expression of these markers has therapeutic implications and second-generation epidermal growth factor receptor tyrosine kinase inhibitors such as HKI-272, XL647, and BIBW2992 have dual activity, inhibiting both EGFR and HER2 receptors [12,13]. Targeting the hormone receptor pathway may also represent one potential therapeutic approach [14].
We have previously reported mutations of PIK3CA and TP53 in 2 and 3 tumors, respectively, in a series of 14 metastasizing apocrine-eccrine carcinomas [16]. Only one case of aggressive digital papillary adenocarcinoma was included in that study. In addition, subtypes such as adenoid cystic carcinoma, malignant chondroid syringoma, malignant spiradenoma, and malignant cylindroma were not included in our previous series [16]. To investigate whether apocrine-eccrine carcinomas express hormonal receptors or possess activation of oncogenic pathways that could be targeted by available chemotherapeutic agent we performed immunohistochemistry for AR, ER, PR, EGFR, and HER2 expression; fluorescence in situ hybridization (FISH) for EGFR and ERBB2 gene amplification; and single base extension genotyping [24] for recurrent mutations in 15 cancer genes including AKT-1, EGFR, PIK3CA, and TP53 on an expanded series of 54 cases of apocrine-eccrine carcinomas ( Table 1).

Materials and Methods
This study has been approved by the Massachusetts General Hospital institutional review board (IRB No. 2011-P-2489). Since the study is limited to the use of excess human material and health related information, written consent was exempted by IRB. Archival materials of all cutaneous apocrine-eccrine carcinomas including apocrine carcinoma, eccrine carcinoma, aggressive digital papillary adenocarcinoma, hidradenocarcinoma, malignant spiradenoma, porocarcinoma, adenoid cystic carcinoma, malignant chondroid syringoma, and malignant cylindroma diagnosed between 1987 and 2011 were retrieved from the pathology files of the Massachusetts General Hospital, Boston, MA. In addition, the personal consultation files of MCM from 2006-2011 was searched. Age, gender, tumor site, tumor size and clinical follow-up information (such as local recurrence or metastasis) were extracted from the patients' medical records. All patient data were de-identified. The histologic sections of all cases were reexamined and the diagnoses were confirmed.

Mutational Analysis and EGFR and ERBB2 Fluorescence in situ Hybridization (FISH)
A SNaPshotH genotyping assay recently developed by our group was performed on 50 tumors with available archival materials [24]. This assay consists of multiplexed PCR followed by a singlebase extension reaction and uses the commercially available SNaPshot platform (Applied Biosystems). The original tumor genotyping panel described by Dias-Santagata et al [24], was expanded to include three additional assays (AKT1.49, testing for the AKT1 E17K mutation; and IDH1.394 and IDH1.395, testing for hotspot mutations in IDH1, which affect codon R132). The full panel is outlined in Table 1 and tests for common mutations in 15 cancer genes. SNaPshotH genotyping was performed using previously described conditions [24], and included the following additional primers for AKT1 and IDH1 EGFR gene copy number was assessed in 49 tumors by fluorescence in situ hybridization (FISH) as previously published [26]. FISH failed in 3 tested cases. Gene amplification and polysomy were defined per criteria outlined by Cappuzzo et al [27]. ERBB2 gene copy number was also assessed in 6 cases with 2+ HER2 protein overexpression. Polysomy 7 was defined as three or more CEP signals per cell.

Statistical Analysis
For patients with follow-up data available, metastasis-free survival time was calculated as time from diagnosis to identification of metastatic disease in the lymph nodes or distant organs. Patients who did not develop metastases were considered censored at the time of most recent follow-up report. Log-rank tests were used to compare the distribution of metastasis-free survival between subgroups defined by positivity for mutational status (defined as non-wild type for one or more genes tested); EGFR FISH; AR, ER, PR, EGFR, and HER2 expression. Kaplan-Meier plots were created to visually assess the differences in metastasisfree survival among subgroups. The statistical association of AR expression in apocrine versus eccrine carcinoma was analyzed by Fisher's Exact Test. A two-tailed p value of less than 0.05 was considered to be statistically significant.

Mutational Analysis and FISH
The molecular results are summarized in Table 4. We have recently developed a multiplexed tumor genotyping clinical assay that uses the SNaPshot platform from Applied Biosystems [24]. This assay performs well with archived tissue and tests for recurrent mutations in 15 cancer genes, including potentially actionable targets such as BRAF, EGFR, KRAS, PIK3CA, and TP53 ( Table 1). The genes included in this panel were selected based on their clinical significance and on the availability of therapeutic agents (either FDA-approved or under clinical testing) targeting these cancer pathways [24]. Due to poor quality DNA, the assay completely failed in 3 cases. In 5 cases, only a portion of the assay failed (see Appendix S1). SNaPshot genotyping identified somatic mutations in 23% (n = 11) of 47 cases (Figure 3). Activating . AKT-1 (E17K) mutation was detected in one hidradenocarcinoma. Metastasis-free survival was not significantly associated with mutational status (p = 0.12, Figure 4).
FISH for EGFR gene amplification was successful in 46/50 cases. Either polysomy ( Figure 2C) or trisomy was detected in 14/46 (30%) cases. Of these 14 cases, 4 (9%) exhibited high polysomy of EGFR (defined as $4 copies in $40% of cells by Capuzzzo et al [27]). Although 84% of cases overexpressed EGFR, only 30% exhibited polysomy or trisomy and no cases showed gene amplification. ERBB2 gene amplification was not detected in any of the 6 cases with 2+ HER2 protein expression. Thus, there appears to be no correlation between IHC and FISH in evaluating both EGFR and ERBB2. There was no significant difference in the distribution of metastasis-free survival time comparing patients with detected polysomy/trisomy in EGFR FISH compared to those without (p = 0.6, Figure 5).

Discussion
It is known that apocrine-eccrine carcinomas often express estrogen receptor and progesterone receptor [21,23]. Due to their similar morphology and possible embryologic origin, it is not surprising that ER, PR, and AR expression can be seen in apocrine-eccrine carcinomas as observed in mammary carcinomas [21][22][23]28,29]. ER expression was seen in 10/33 (30%) apocrineeccrine carcinomas in a series by Swanson et al [23] and the positive tumors included 8 eccrine carcinomas, 1 porocarcinoma, and 1 mucinous eccrine carcinoma. Twenty-one percent (9/42) and nineteen percent (8/42) of primary sweat gland carcinomas expressed ER and PR, respectively, in a series by Busam et al [21]. Similarly, in our series 33% (17/51) and 20% (10/51) of apocrineeccrine carcinomas expressed ER and PR, respectively. The expression of hormonal receptors could have treatment implications since hormonal modulation plays an important role in the prevention and treatment of breast carcinomas [14]. Choudhry et al [30] reported AR expression in normal apocrine as well as eccrine glands. This suggests that androgen may modulate the function of both apocrine and eccrine glands. The expression of steroid receptors in aggressive digital papillary adenocarcinomas has not been studied. Of interest, all of the primary cutaneous apocrine carcinomas (9/9, 100%) and a third of the aggressive digital papillary adenocarcinomas express AR in more than 10% of tumor cells in our current series. By contrast, none of the eccrine tumors (0/7, 0%) in our series exhibited AR expression. This suggests that AR-positivity is strongly correlated with pure apocrine carcinoma morphology (p,0.0001). This raises the potential of anti-androgen therapy for these subsets of apocrineeccrine carcinomas. Androgen deprivation therapy using bicalutamide has been reported to be beneficial in the treatment of metastasizing salivary duct carcinoma [31]. Strong (3+) overexpression of HER2 (3+) and gene amplification have been documented in one case of metastasizing hidradenocarcinoma [7]. These findings suggested that ERBB2 may be Table 3. Summary of immunohistochemical analyses in apocrine-eccrine carcinomas.  a relevant therapeutic target in rare cases of apocrine-eccrine carcinoma; however, we were unable to confirm this finding. None of the six apocrine-eccrine carcinomas in our series with (2+) HER2 overexpression demonstrated ERBB2 gene amplification. This is consistent with prior published results indicating that high level ERBB2 gene amplification is unlikely in the setting of 2+ HER2 overexpression [17]. Epidermal growth factor receptor (EGFR/erbB-1) belongs to a receptor family with tyrosine kinase activity whose gene is located on chromosome 7p12. The EGFR signaling that mediates proliferation, migration, invasion, and suppression of apoptosis, can be blocked by a growing number of inhibitor drugs. The role of EGFR inhibitor therapy in apocrine-eccrine carcinomas with protein overexpression remains unclear. Although the majority (85%, 44/52) of apocrine-eccrine carcinomas in our series demonstrated EGFR protein overexpression, only EGFR trisomy or polysomy (13/47, 28%) and no gene amplification were documented by FISH, most frequently noted in eccrine carcinoma, hidradenocarcinoma and porocarcinoma. This is not an unexpected finding; since similar results were noted in our recent study of hidradenocarcinomas [20]. In addition, other studies have shown that EGFR overexpression appears to be independent of EGFR mutation [32,33]. In lung and salivary gland carcinomas, high polysomy of EGFR if considered to be FISH-positive and the patients would receive treatment [34,35]. Further studies are warranted to determine whether similar practice may be applied for apocrine-eccrine carcinomas.
Tumor suppressor gene, TP53, located on the short arm of chromosome 17p13 has been implicated in the regulation of cell growth, DNA repair, and apoptosis. The TP53 gene is frequently (14-52%) altered in human breast carcinomas and is the most commonly mutated gene in human tumors [36][37][38]. TP53 mutations are usually clustered within the most conserved regions of exons 4, 5, 7, and 8 [39]. TP53 mutations have been previously described in apocrine-eccrine carcinomas ( Table 5) [15][16][17][18][19]. Takata et al [18] performed PCR based assays for loss of heterozygosity on chromosomes 3p, 5q, 9p, 9q, 13q, and 17p; yet found only 1 case with TP53 mutation by direct sequencing of exons 5 to 8. Biernat et al [15] employed single-stranded conformation polymorphism analysis followed by direct DNA sequencing and found 5 of 16 sweat gland carcinomas possessed TP53 mutations ( Table 5). Two of eleven hidradenocarcinomas exhibited TP53 mutations in the series by Kazakoz et al. [17] Similar to findings reported by Biernat et al [15] and Kazakov et al [17], we found TP53 mutations in codon 248 of exon 7 and codon 273 of exon 8 in our 7 positive cases ( Table 5). Interestingly, these same mutations have also been reported in breast carcinomas [40][41][42]. Although SnaPshot testing works well for oncogenes, which are typically mutated at only a few distinct loci, it is not comprehensive for tumor suppressors. It captures only a few of the numerous mutation events described for TP53, at loci that are not covered by the assay.
PIK3CA (phosphatidylinositol 3-kinase, catalytic, alpha polypeptide) mutations in these tumors have been reported by our group [16]. In breast carcinomas, the majority of mutations have been identified in the helical domain (exon 9, 37%) and in the kinase domain (exon 20, 63%) of PIK3CA [43]. All mutations were single-based substitutions [43]. Similarly we detected 2 types of mutations, c.1624G.A:p.Glu542Lys (in one hidradenocarcinoma and one porocarcinoma) and c.1633G.A:Glu545Lys (in one porocarcinoma), in exon 9 of PIK3CA. These 2 mutations are among the three most frequently reported mutations in breast cancer [43]. Mutations in codon 545 represent a mutational hotspot reported in ovarian and colorectal carcinomas as well [44,45]. The clustering of mutations within PIK3CA may prove useful for therapeutic purposes. The phosphatidylinositol 3 kinase (PIK3) signaling pathway is an important regulator of cell growth, proliferation, cell motility, angiogenesis, and survival, and it has been shown that PIK3CA is the most frequently mutated gene in breast cancer [46,47]. It is thought that in breast cancer, oncogenic mutations in PIK3CA or low levels of PTEN expression may confer resistance to treatment with trastuzumab, a monoclonal antibody that targets the HER2/Neu receptor [48]. ERBB2 amplification and PIK3CA mutation were validated as biomarkers for sensitivity to the single-agent phosphoinositide 3-kinase (PIK3) inhibitor, GDC-0941, in breast cancer models [49]. Other studies have shown that cancers with PIK3CA mutations were sensitive to single-agent PI3K inhibitors and dual PI3K-mammalian target of rapamycin (mTOR) inhibitors [50,51].
AKT-1 mutation has not been previously described in cutaneous appendageal carcinomas. In carcinomas, the PI3K/AKT pathway is well characterized [46,52]; and the PI3K/Akt/mTOR pathway has been shown to be a target for cancer therapy including breast carcinoma [53]. While mutations in the PI3 kinase gene (PIK3CA) are common in breast carcinoma, somatic mutations in AKTs are rare [54,55]. Similarly we found only 1 of 47 cases exhibiting AKT-1 mutation.
The presence of either high polysomy, low polysomy or trisomy of EGFR does not appear to correlate with metastatic disease (p = 0.6, Figure 5). Five of the 8 cases (63%) with detected mutations had metastatic disease, while 9 of the 28 cases (32%) with no mutations detected developed metastases; however, there was no significant difference in metastasis-free survival (p = 0.12, Figure 4). Thus, there appears to be no correlation between the presence of mutations and metastatic disease. Apocrine-eccrine carcinomas and breast carcinomas are analogous tumors often with similar histology. It is interesting that in this study we detected mutations in PIK3CA and TP53 in a subset of apocrine-eccrine carcinomas, a therapeutically-   relevant finding since both of these genes are known to be frequently mutated in breast carcinomas. Currently there is no uniform guideline concerning the treatment for metastatic apocrine-eccrine carcinomas and treatment success has been documented only in isolated case reports [6][7][8]. A variety of chemotherapeutic agents have been used with varying degrees of responsiveness. Combinations of cyclophosphamide, bleomycin, cisplatin, and 5-fluorouracil; interferon-alpha, interleukin-2, sunitinib, and tamoxifen have been reported with some degree of response [6,8,56,57]. Remission was achieved for 16 months with paclitaxel and carboplatin in a case of apocrine carcinoma with lymph node, lung, and bone metastasis [58]. Hikada et al [59] reported a case of metastatic apocrine carcinoma responded to treatment with HER-2 inhibitors. Radiation therapy has been used in selected cases of metastatic porocarcinoma [60].
In this study we used a SNaPshot platform, previously reported by our group [24], to screen relevant cancer genes with available targeted therapeutic agents in rare tumors. The platform is cost effective due to two main reasons. First, SNaPshot testing uses thermocyclers (PCR machines) and capillary electrophoresis DNA sequencers, which are instruments that already exist in any typical clinical molecular laboratory. Thus, there is not need for an upfront investment in expensive equipment or bioinformatics personnel. Secondly, because SNaPshot tests for hotspot mutations in a multiplexed fashion, it only uses a fraction of tumor tissue required for probing the same number of exons using Sanger sequencing.
In summary, we report mutations in AKT-1, PIK3CA and TP53 in a subset of apocrine-eccrine carcinomas including eccrine carcinoma, aggressive digital papillary adenocarcinoma, hidradenocarcinoma, and porocarcinoma. There is strong correlation of AR expression in the apocrine carcinoma subtype, raising the potential for anti-androgen therapy. The role of EGFR inhibitor therapy in apocrine-eccrine carcinomas with protein overexpression remains unclear. The lack of correlation between the protein expression and polysomy/gene amplification suggests that molecular mechanisms other than gene amplification may play a role in EGFR overexpression in adnexal carcinomas. Based on our findings, targeted therapy including PI3K/Akt/mTOR pathway inhibitors, which is currently in clinical testing, may be potential treatment options for rare cases of apocrine-eccrine carcinomas. It will be interesting to determine whether these results will translate into real therapeutic response in followup clinical trial studies.  Table 5. Summary of TP53 mutations in apocrine-eccrine carcinomas [15][16][17][18][19].

Case
Histologic type TP53 mutation Reference

Supporting Information
Appendix S1 Appendix of failed cases.