Molecular Profiling of Syringocystadenocarcinoma Papilliferum Reveals RAS-Activating Mutations

Eleven cases of SCACP were obtained from the surgical pathology files from 6 institutions that contained archival formalin-fixed, paraffin-embedded (FFPE) tissue, with approval by their Institutional Review Boards (Massachusetts General Hospital: 2019P003688, Boston). The diagnosis and adequacy for molecular testing underwent central review and were confirmed by at least 3 board-certified dermatopathologists. The diagnosis of SCACP was rendered when a carcinoma was seen arising in association with areas displaying features clearly reminiscent of SCAP, especially because an association with a known preexisting SCAP cannot always be confirmed. More specifically, the diagnosis required a directly associated and contiguous epithelial proliferation composed of tumor cells displaying features of malignancy, such as cytologic atypia, frequent mitotic figures, and/or necrosis, as previously described, accompanied by features akin to SCAP, such as a benign transitional squamoglandular epithelium, often forming papillae associated with lymphoplasmacytic inflammation.¹⁴  Additional clinical information, such as age, sex, tumor site, size, location and description, the presence of an associated nevus sebaceus when referenced, treatment (if any), and follow-up data, was obtained when available.

Each tumor was marked on an initial hematoxylin-eosin–stained slide, and the malignant component was microdissected from subsequent unstained slides. For cases in which the SCAP component could also be microdissected, microdissection was also performed. DNA was extracted from the FFPE tissue (Qiagen, Germantown, Maryland) and analyzed by next-generation sequencing for “hot spot” mutations in 50 cancer genes (Ampliseq Cancer Hotspot Panel v2, Thermo Fisher Scientific, Waltham, Massachusetts) as described previously.^15,16 

BRAF V600E mutation status was further investigated using a BRAF peptide nucleic acid clamp real-time polymerase chain reaction (PCR) as described previously.^16,17  The assay detects all known mutations between BRAF codons 598 and 602. Positive specimens were confirmed by BigDye Sanger sequencing of the peptide nucleic acid clamp PCR product (Thermo Fisher Scientific). This technique has a detection limit of less than 0.1% and is optimal for detecting mutations in small specimens with a low tumor percentage (<40%) that would be missed by ordinary Sanger sequencing methods.

Somatic mutations detected in SCACP were compared to previously established mutation signatures. The analysis considered each somatic mutation in the context of the surrounding 5′ and 3′ nucleotides for both the forward and reverse orientations. Mutational signature analysis associated with the following mutational mechanisms was included: (1) exposure to environmental mutagen, including tobacco smoke (C>A) and ultraviolet (UV) damage (C>T at dipyrimidine sites and CC>TT); (2) exposure to chemotherapeutic agents, such as temozolomide (TZ; C>T at CC or CT); (3) DNA repair defects, including mismatch repair (MMR) deficiency (indel and repeats, also C>T at GC or CG) and POLE mutations (C>A at TCT + C>T at TCG + T>G at TTT); and (4) activation of the apolipoprotein B mRNA editing catalytic polypeptide-like (APOBEC) family of nucleic acid editing enzymes (C>G and C>T with a 5′ thymine).¹⁸ 

A mouse monoclonal anti-BRAF V600E antibody (clone VE1, Spring Biosciences, Pleasanton, California) was used for immunohistochemistry on FFPE tumor sections, and the staining patterns were scored positive when unequivocal cytoplasmic staining was observed in the tumor cells. A rabbit monoclonal anti-P53 antibody (clone DO-7, Biocare Medical, Pacheco, California) was also used in a subset of cases with available material and was scored positive when unequivocal nuclear staining was observed in more than 90% of tumor cells. All the immunohistochemical slides were evaluated by at least 2 board-certified dermatopathologists.

The clinicopathologic characteristics are summarized in Figure 1. The cohort comprised 6 female and 5 male patients with ages ranging from 29 to 96 years (mean, 73.6 years). The neoplasms occurred on the head and neck (n = 8; 73%) and extremities (n = 3; 27%). The lesions were clinically described as a verrucous/exophytic plaque (n = 3), red papule or nodule (n = 4), or a cyst (n = 1), with or without ulceration or hemorrhage, measuring 0.6 to 3.2 cm (mean, 1.5 cm) in greatest dimension. After diagnosis, 8 lesions were surgically excised, 3 with Mohs micrographic surgery and 1 with an additional sentinel lymph node biopsy, which was negative for metastatic disease. Of the 7 patients with available follow-up, 6 were alive without evidence of disease and 1 died of other disease (range, 17–77 months; mean, 70.0).

Three neoplasms possibly arose secondarily in a nevus sebaceus: one each on the scalp, helix of the ear, and forearm. One patient (case 1) was given a clinical diagnosis of nevus sebaceus and reported the presence of the lesion since birth. Characteristic findings seen in a nevus sebaceus, such as epidermal acanthosis and papillomatosis, prominent and abnormally located sebaceous glands, and/or presence of numerous sweat glands, were identified histologically in these cases.

The malignant component of these cases was contiguous with areas clearly reminiscent of SCAP and showed variable morphologies; 4 cases showed at least carcinoma in situ (adenocarcinoma/SCACP in situ [SCACPIS], n = 3; squamous cell carcinoma [SCC], n = 1), and 7 cases were invasive, composed purely of SCC (n = 5) or mixed with variable components of adenocarcinoma and SCC (n = 2). Areas akin to SCAP were identified in all cases composed of benign transitional squamoglandular epithelium, often forming papillae associated with lymphoplasmacytic inflammation. In all cases there was an associated contiguous atypical epithelial proliferation that displayed features of malignancy, such as loss of polarity, architectural disarray, cytologic atypia, and scattered mitotic activity, including atypical mitoses and intraluminal necrosis.

The 3 cases of SCACPIS, a subtype of adenocarcinoma in situ, displayed an intact myoepithelial/basal cell layer, which was highlighted by a p63 stain (case 11; Figure 2). Pagetoid spread was noted in 1 case (case 11), seen as a proliferation of small clusters and single cells within the epidermis and adnexal epithelium. Additionally, the tumor often formed prominent cribriform and solid areas, reminiscent of mammary ductal carcinoma in situ. The fourth case was of SCC in situ, showing full-thickness dysplasia involving the squamous epithelial component within the SCAP-like area (case 10; Figure 3).

A total of 5 of the remaining 7 cases of invasive carcinoma were purely of SCC/carcinoma with squamous differentiation, which varied from well to poorly differentiated. One of the 2 cases of poorly differentiated SCC displayed a sarcomatoid or spindled morphology that formed pseudovascular spaces (case 1). Immunohistochemically, the tumor cells were highlighted by MNF116, pancytokeratin, and p63 and were negative for ERG and CD31. The second case of poorly differentiated SCC was keratin forming but composed mainly of syncytial sheets of epithelioid cells with vesicular nuclei and prominent nucleoli admixed with a prominent inflammatory infiltrate, reminiscent of a lymphoepithelioma-like carcinoma (case 8). The 2 mixed cases displayed features of both an SCC and adenocarcinoma, the latter of which was composed mainly of cords and islands of tumor cells that displayed cytologic atypia, scattered mitoses, and variably formed ductal/glandular structures.

The molecular findings are summarized in Figure 1. A total of 8 of the 10 cases (80%) with enough material for next-generation sequencing had 15 hot spot mutation(s) involving HRAS (n = 4; 27%), KRAS (n = 1; 7%), TP53 (n = 4; 27%), ATM (n = 2; 13%), BRAF (n = 1; 7%), FLT3 (n = 1; 7%), CDKN2A (n = 1; 7%), and PTEN (n = 1; 7%). The 2 mixed cases were not found to harbor hot spot mutations, and 1 invasive SCC case failed because of poor DNA quality and or quantity. A total of 5 RAS-activating mutations were identified, of which 4 were HRAS mutations (G13R, n = 3; G13D, n = 1) and occurred on the head and neck, 1 of which was co-mutated with BRAF (V600E), whereas the case with a KRAS mutation (G12D) occurred on the lower extremity. Three cases in which the SCAP-like component was also successfully sequenced revealed the same HRAS mutations as their malignant counterparts. TP53 mutations were the most frequent inactivating events in tumor suppressor genes, occurring in 4 of 10 cases (40%).

UV-associated mutations defined as C>T transitions or CC>TT tandem substitutions (or reverse complement G>A, GG>AA) at dipyrimidine sites^19,20  were identified in 4 cases involving TP53 (n = 4), CDKN2A (n = 1), ATM (n = 2), HRAS (n = 1), and KRAS (n = 1). These 4 SCACPs were identified in patients who were all older than 80 years, occurred on sun-exposed sites, and all displayed histopathologic evidence of photodamage in the form of microscopic solar elastosis.

To rule out a different mutation mechanism, the somatic mutations were compared to additional mutation signatures associated with C>T transitions (ie, MMR defect, POLE mutation, APOBEC activation, TZ therapy). The CC>TT tandem substitution in 3 variants is considered unique to UV damage. Of the remaining 6 somatic variants with a UV-damage signature, 2 overlap with the APOBEC and TZ therapy mutation signatures (Figure 1). The other 4 variants show overlap with 1 additional mutation signature, 3 with TZ therapy and 1 with MMR deficiency. Because none of the patients received TZ therapy, UV damage remains the most likely mechanism behind at least 6 mutations (ie, cannot exclude APOBEC or MMR deficiency in 3 variants).

Immunohistochemically, 2 SCACPIS cases were diffusely and strongly positive for the BRAF (VE1) antibody (cases 3 and 5; Figure 4, A and B), of which 1 was confirmed to harbor a BRAF V600E mutation by both next-generation sequencing and real-time PCR (Table 1). The 3 SCACPs with a TP53 mutation and tissue remaining for immunohistochemistry were all found to display overexpression of p53 (case 4; Figure 4, C and D).

Activating mutations in the RAS/MAPK signaling pathway have been identified in both SCAPs and nevus sebaceus and are thought to contribute to their pathogenesis.^8,9  A RAS gene alteration has been reported in up to 38% of SCAPs; more specifically, it has been identified in up to 30% of sporadic and 100% of secondary tumors tested.^8,21  Most of these mutations involve HRAS (86%) and, less frequently, KRAS (14%).^7,8,10,21  Mutations in BRAF V600E have been detected in up to 67% of cases and appear to be mutually exclusive of RAS alterations.^7,8,10,21  Supporting this idea, immunohistochemistry for BRAF V600E protein expression has revealed positive staining in up to 64% of sporadic SCAPs,¹¹  but none of the secondary SCAPs were reported to harbor a BRAF mutation.^8,21  Similarly, instances of nevus sebaceus have been found to harbor postzygotic mosaic mutations in HRAS and KRAS in up to 95% and 5% of cases, respectively, one of which was co-mutated with BRAF (non-V600E mutation).^9,12,13  The prevalence of RAS mutations in nevus sebaceus may rationalize why BRAF mutations are uncommon in secondary SCAPs.

To our knowledge, this represents the largest cohort and the first study to analyze 50 cancer gene hot spot mutations in a series of SCACPs to date. In our literature review, comparable studies are limited to 2 prior case reports of an SCACP in which genomic profiling was performed, 1 of which revealed a MAP2K1 alteration that is predicted to activate RAS-MAPK signaling and 2 verrucous carcinomas arising in association with SCAP that were found to harbor a V600E BRAF mutation by PCR and immunohistochemistry.^22–24  As in benign SCAPs and nevus sebaceus, our data show that SCACPs also harbor activating mutations in the RAS-MAPK pathway, with a RAS gene alteration occurring in 50% of cases, of which most (80%) involve HRAS and a smaller subset (20%) occurring in KRAS. Three of these cases harboring HRAS mutations were also noted to harbor the same mutation in the background SCAP-like areas. The hot spot HRAS mutation identified in SCAPs and most cases of nevus sebaceus involves G13R; in SCACP they similarly occur in G13R and G13D.^8,9,12,13  The presence of analogous RAS mutations suggests these are likely early oncogenic events and further supports that these tumors are related and may arise as a result of malignant transformation from their benign counterpart.

It is interesting to note that approximately one-third of SCAPs harbor a RAS gene alteration, whereas the remaining two-thirds contain a BRAF V600E mutation.^7,8,10,12,21  However, only 1 BRAF gene alteration (7%) was detected in SCACP, which was co-mutated with HRAS in our study. When looking more closely at the variant allele frequencies, the HRAS mutation was present at a much higher frequency than the BRAF V600E mutation, implying the HRAS mutation was either gained or the BRAF mutation was subclonal in this tumor (case 5). Nevus sebaceus is associated with a relatively high incidence of forming secondary tumors, and RAS gene alterations, which have also been found in these secondary tumors, such as trichoblastomas, SCAPs and trichilemmomas, are likely the oncogenic driver.⁹  We speculate that, like nevus sebaceus, the subset of SCAPs that harbor RAS gene alterations infer a molecular predisposition to undergo malignant transformation. This hypothesis may explain why BRAF mutations were not prominent in SCACPs, despite being the most common molecular finding in SCAPs. Another possibility is the HRAS-mutated tumor cells comprise a greater population in comparison with those harboring a BRAF mutation, apportioning a greater risk for transformation.

We also identified a proportion of SCACPs that demonstrate a UV-mutation signature and occur in tumors located on sun-exposed sites from elderly patients who were all 80 years or older. TP53 mutations were the most frequent inactivating event in the tumor suppressor genes, of which most (3 of 4; 75%) demonstrated a UV signature. The ataxia telangiectasia mutated (ATM) gene, which is also involved in DNA repair, and CDKN2A, which induces cell cycle arrest, are additional tumor suppressor genes with inactivating mutations that were also found to harbor UV signatures. None of the cases with RAS gene alterations but lacking a tumor suppressor mutation demonstrated UV-associated mutations, suggesting that a subset of SCACPs arise in a UV-dependent pathway. These findings are similar to the multistep models of cancer development, where additional molecular alterations are necessary for malignant progression, and in this instance are predominantly UV-associated inactivating mutations of tumor suppressor genes.

Likewise, poromas and porocarcinomas, in addition to YAP1 fusions, have also been found to harbor activating HRAS mutations in 12% to 33% and 17% to 40% of cases, respectively.^25–28  This implies that comparable to SCACPs, a portion of porocarcinomas are likely a result of poromas undergoing malignant transformation.^25,26  Molecular profiling revealed TP53 inactivating events are present in 67% to 80% of porocarcinomas.^25,26  Additionally, UV-associated mutations were seen in 42% to 80% of porocarcinomas, occurring most frequently in TP53 but also found in APC, PTEN, ABL1, RB1, CDKN2A, and RET.^25,26  In our study, we similarly identified a UV signature in TP53, CDKN2A, and ATM in SCACPs.

Although 2 cases in our study were found to be diffusely and strongly positive for the VE1 BRAF immunostain, only 1 case was confirmed to harbor a BRAF mutation by molecular testing. Although the overall sensitivity and specificity of BRAF V600E mutation-specific antibodies are fairly robust, 71% to 100% and 62% to 100%, respectively,²⁹  false-positive results have been reported.^29,30  A number of hypotheses have been proposed, such as variability in the reagents and antigen retrieval methods, low tumor burden that may be adequate for immunohistochemistry but insufficient for molecular testing, and varying degrees of scoring, resulting in different interpretations of what is considered to be positive.^29–31  Tumors harboring non-V600E BRAF mutations have also been found to be reactive for the VE1 antibody, which may be due to a resultant homologous conformational protein change that can still bind the antibody.^29–31  It is difficult to discern why there was a false-positive result in our cohort; there were no BRAF mutations detected by molecular testing (next-generation sequencing and PCR), and there was a large quantity of BRAF VE1-positive tumor present in the section (Figure 4, B). It is possible that a BRAF mutation that is not a part of the hot spot mutation panel, a mutation at the BRAF primer binding site, BRAF gene amplification, or a BRAF fusion may account for this discrepancy.³² 

Overall, adnexal carcinomas are rare and are typically associated with poor outcomes.^33,34  However, the prognosis of SCACPs has been conflicting, with some reports implying it is a low-grade malignancy with a favorable outcome,^35,36  whereas in a more recent series of 10 cases, 30% (3 of 10) of patients died of disease, all with distant metastases.¹  To our knowledge, a total of 88 SCACPs have been reported in the literature, including our cohort of cases, of which 11 (12.5%) of these patients with provided follow-up were found to have locoregional and/or distant metastases (Table 2).^* Of the 9 metastatic cases with follow-up data, 3 (33%) died of disease.¹  Zhang et al also noted that tumor size (≥5 cm) may be an indicator of a poorer prognosis.¹  Only 4 cases (45%) that provided a tumor size were 5 cm or greater, and, more interestingly, 2 cases measured only 2 cm in greatest dimension. In our series of 11 cases, none were found to have a local recurrence or metastasis at 70 months, although follow-up was limited to 7 cases with available clinical information and all tumors were less than 5 cm in greatest dimension.

Currently, there is no uniform guideline regarding treatment, and surgical excision remains the primary mode of therapy, with some performing Mohs micrographic surgery and lymph node biopsy/dissection.^1,63  In patients with metastatic disease, a few were additionally treated with chemotherapy and/or radiation.^{1,22,35,38,39}  A recent case report described next-generation sequencing on a lymph node metastasis from an SCACP of the back in a patient who was deemed not to be a surgical candidate. A total of 18 genomic alterations were identified (The FoundationOne assay; 315 genes), and despite the identification of actionable targets, the patient declined systemic therapy.²²  Similar to the findings in our series, molecular profiling of this metastatic SCACP identified a CDKN2A and TP53 mutation.²²  CDK 4/6 inhibitors, such as Palbociclib, have been shown to provide some clinical benefit in patients with tumors harboring CDKN2A alterations.^75–77  Blocking cell cycle checkpoint regulation through Wee-1 inhibitors, such as adavosertib, has also been shown to improve sensitivity to certain chemotherapy agents and may represent a new therapeutic target for TP53 mutant tumors.^77–80 

Additionally, because SCACPs are in part RAS-dependent tumors, they may be responsive to therapies targeting RAS and their downstream effectors, including BRAF, PI3K, and MEK/MAPK.^10,81  Pan-RAS inhibitors are currently under investigation that are not mutant selective but appear to be associated with higher levels of toxicity.⁸¹  Other newer agents, such as SHP2 inhibitors, which block a nonreceptor protein tyrosine phosphatase required for MAPK pathway activation, and combination therapy, such as with RAF and MEK inhibitors, are also currently in clinical trials.⁸¹ 

A limitation of our study is the small cohort size and its retrospective nature, which is principally due to the rarity of this entity. We also observed 2 cases without mutations, a finding that may represent our limited testing platform.

In summary, our study reveals that SCACPs share overlapping genetic features with SCAP in the form of RAS-activating mutations, which supports that at least a subset may arise from malignant transformation. Additionally, TP53 and UV-induced mutagenesis also plays a role in the development of a subset of SCACPs. Finally, our results support a rationale for therapeutically targeting a portion of SCACPs, particularly in those few cases that develop locoregional and distant disease.