Epithelial elements in superficial angiomyxomas: mimicry of adnexal development, and mesenchymal‐to‐epithelial transition

Abstract

Aims

A subset of superficial angiomyxomas contains epithelial elements. Rarely, especially in Carney's complex patients, arborizing adnexal proliferations are present, with some authors proposing that they could be authentic adnexal neoplasms. Our aim was to determine if the epithelial elements are trapped adnexal epithelium, epithelial outgrowths triggered by myxoid tumour cells or genuinely neoplastic.

Methods and results

We studied the clinicopathological and immunophenotypic features (immunohistochemistry and double immunofluorescence) of 43 superficial angiomyxomas, including one from a Carney's complex patient. We found hair follicle placode/germ formation in 7 cases; elongated and/or branched eccrine elements in 4 cases and solid/cystic infundibular‐like hyperplastic epithelium in 10 cases. S100A4‐positive mesenchymal condensates surrounded follicular and eccrine epithelial buds close to myxoid tumour cells, but only the latter showed variable PRKAR1A loss. In the Carney's complex case, prominent eccrine duct branching was found mimicking fibrofolliculoma, with vimentin‐positive cell aggregates connecting and integrating with the pre‐existing epithelium. By double immunofluorescence, they showed cytokeratin and E‐cadherin immunoreactivity while still expressing vimentin as evidence of mesenchymal‐to‐epithelial transition.

Conclusions

Our results suggest that adnexal epithelial elements reproduce their embryogenesis through the involvement of S100A4‐positive mesenchymal niches in both follicular and eccrine elements. These niches are close to myxoid cells with PRKAR1A loss, whereas the complex epithelial structures retain PRKAR1A, which suggests that the former induce non‐neoplastic growth of the latter. Finally, we provide evidence for the role of mesenchymal‐to‐epithelial transition in the branching of the eccrine duct epithelium in the Carney's complex case, probably secondary to protein kinase A activation.

Abbreviations

CK

cytokeratin

E‐cadh

E‐cadherin

MET

mesenchymal–epithelial transition

PKA

protein kinase A

PRKAR1A

proteinkinase cAMP‐dependent type I regulatory subunit alpha

SAM

superficial angiomyxomas

Vim

vimentin

α‐SMA

α‐smooth muscle antigen

Introduction

Epithelial elements in superficial myxomatous tumours were first described by Lund, ¹ and Headington called them trichogenic myxomas. ² Allen reported 30 cases of what he called ‘superficial angiomyxomas’ (SAM), also known as ‘cutaneous myxomas’, and nine of them had epithelial elements such as epidermoid cysts, linear strands of epidermal cells or trichofolliculoma‐like epidermoid cysts, admixed with angiomyxoid tissue. ³ Calonje et al. reported 39 cases and found in eight primary lesions and three recurrences an epithelial component that the authors considered most likely derived from entrapped hyperplastic adnexal structures. ⁴ In a recent series of 54 cases, epithelial elements including follicular induction, basaloid proliferations and epithelial inclusions were found in 19 cases (35%). ⁵

Carney initially proposed that Carney's complex was a genetically determined malfunction of mesenchymal cells that proliferate and produce excessive amounts of proteoglycan. ⁶ This hypothesis was confirmed and Carney's complex was the first disease to be associated with mutations in the protein kinase A (PKA) holoenzyme. ⁷ Therefore, Carney's complex is a model of dysregulation of the cAMP/PKA signalling. In more than 70% of the index cases, it is due to inactivating heterozygous mutations of PRKAR1A, encoding the regulatory subunit 1α (R1a) of PKA, which is ubiquitously expressed. ⁸ Importantly, most mutations are not expressed due to nonsense mediated decay, provoking PRKAR1A haploinsufficiency, but PKA activity is enhanced due to disruption of the normal holoenzyme stoichiometry. ⁹ In this regard, PRKAR1A loss has also been reported in 53%–80% of SAM from patients without Carney's complex. ¹⁰ , ¹¹

The goal of our study was to determine if the epithelial elements are just entrapped adnexal epithelium, epithelial proliferations induced by myxoid tumour cells or even a new truly neoplastic growth, as proposed in some unusually complex epithelial adnexal proliferations, particularly found in Carney's complex patients.

Materials and Methods

This study was approved by our Institutional Ethics Committee on 21 July 2021 (reference number: 2021/198). Written informed consent was obtained from the patients.

To characterize the epithelial elements present in SAM, we studied the histological and immunohistochemical features of all consecutive cases of SAM received from 2000 to 2023, of which H&E slides and paraffin blocks were available, using antibodies for: adnexal epithelium (cytokeratins AE1‐AE3 and E‐cadherin as general markers; CK7 for the secretory cells of the eccrine coil; α‐smooth muscle antigen [α‐SMA] and calponin for myoepithelial cells of eccrine coils; CK14 for luminal cells of eccrine ducts ¹² , ¹³ ); mesenchymal stromal cells (vimentin), mesenchymal niches for adnexal development (S100A4); and PRKAR1A, mutated in most Carney's complex patients and commonly lost in SAM ¹⁰ , ¹¹ (Table S1). In addition, double immunofluorescence was performed for vimentin and CK AE1‐AE3 as well as for vimentin and E‐cadherin. Briefly, double immunohistochemistry was performed by consecutive incubations of two primary antibodies, each one followed by incubation with Alexa Fluor 488‐labelled goat anti‐mouse IgG1 and Alexa Fluor 633‐labelled goat anti‐mouse IgG1 (Table S1).

Results

The study included 43 cases. They were all characterized by a dermal and/or subcutaneous lobulated growth composed of stellate and spindle cells embedded in abundant mucin with irregularly distributed thin‐walled branching vessels and variable, not constant, presence of neutrophils.

Mutation analysis of peripheral blood samples to detect the characteristic germline mutations of Carney's complex was performed in two children. One was a 10‐year‐old girl with hirsutism and a Cushing phenotype who had previously been diagnosed with psammomatous melanotic schwannoma (malignant melanotic schwannian tumour). Carney's complex due to a PRKAR1A pathogenic mutation (c.658_659del; p.Asn220Cysfs*12) was confirmed. The other patient was a boy with a large, clinically unusual scrotal lesion. However, no genetic findings indicative of Carney's complex were identified.

The main clinicopathological and immunohistochemical features are shown in Table 1. We found adnexal epithelial elements in 20 cases (46.5%), including immature hair follicles (seven cases) (Figure 1A–D), solid or cystic infundibular‐like epidermoid elements (Figure 1E,F) in 10 cases, and immature eccrine elements in four cases (Figure 1H–K). Immature hair follicles recapitulated the embryonic developmental stages of the hair follicle, especially the placode and follicular germ stages ¹⁴ (Figure 1C,D). In two cases, aberrant sebaceous glands were associated with the follicular elements (Figure 1B). The epithelial elements were multifocal, except in two cases where only a single focus was present. Mature hair follicles were excluded as we could not rule out their presence prior to SAM development. In all cases, there was a cellular condensate of vimentin‐ and S100A4‐positive spindle stromal cells (Figure 1G–I) surrounding the placodes and hair follicle germs, and in close contact with SAM myxoid cells, the latter showing variable PRKAR1A loss in most cases (Table 1, Figure 1J,K). No loss of PRKAR1A was found in any of the hair follicle structures.

Table 1.

Clinicopathological and immunohistochemical findings

Clinicopathological features	n (%)	S100A4+ stromal niches	PRKAR1A loss		Vim and E‐Cadh Vim and CK AE1‐AE3
			Myxoid cells	Epithelial cells	Co‐expression
Gender
Male	26 (60.46)
Female	17 (39.54)
Age rank (mean)	3–75 (47)
Tumour sites
Limbs	14 (32.55)
Trunk	14 (32.55)
Head and neck	11 (25.58)
Genital area	2 (4.65)
Unknown	2 (4.65)
Tumour size in mm (median/mean)	4–33 (8/11)
Adnexal epithelial elements
Immature hair follicles	7	7	7	0	0
Epidermoid (infundibular‐like)	10	8	7	0	0
Aberrant sebaceous glands	2	1	1	0	0
Immature eccrine ducts (CK14+, CK7− and α‐SMA−)	4
Elongation	2	2	2	0	0
Branching	2	2	2	0	2

CK, cytokeratin; E‐Cadh, E‐cadherin; PRKAR1A, protein kinase cAMP‐dependent type I regulatory subunit alpha; Vim, vimentin; α‐SMA, alpha‐smooth muscle antigen.

Figure 1.

(A) Low magnification view of a SAM with hair follicle and sebaceous elements (A and B); induction of hair follicle placode and germ stages (C–F) and cystic and solid infundibular‐like epidermoid elements (E and F); there was a variably cellular condensate of vimentin‐positive spindle stromal cells (G and H), which were also S100A4‐positive (I) surrounding placodes and hair follicle germs and in close contact with tumour cells showing PRKAR1A loss (J and K). Elongation and branching of eccrine epithelial structures (L–P) with exceptional lumen formation (N). (A–F and L–P) H‐E staining; (G–K) immunoperoxidase with DAB chromogen.

Elongation (four cases) and/or branching (two cases) of eccrine epithelial structures was found (Figure 1 L–P), all of them showing immunostaining for ductal epithelial markers (CK14) (Figure 2A) but not for the secretory coil luminal epithelium (CK7) (Figure 2B) or myoepithelial markers. These eccrine ducts were mostly solid, with exceptionally little luminal formation (Figure 1N). In the occipital lesion from the Carney's complex patient, branching was highly complex with a labyrinthine appearance (Figure 1L–O) mimicking fibrofolliculoma. Surrounding the tips of these multiple eccrine branches, highly cellular condensates of S100A‐4‐positive cells were found (Figure 2C) as well as PRKAR1A loss in the spindle and stellate cells of the adjacent myxoid component, whereas the ductal epithelial component retained constant PRKAR1A immunostaining (Figure 2D).

Figure 2.

Immunohistochemical and double immunofluorescence findings of eccrine structures. CK14 immunoreactivity (A) and negative CK7 (B). S100A‐4‐positive mesenchymal niche (C). PRKAR1A loss in spindle tumour cells but retained in the ductal epithelium (D). Vimentin‐positive cells of the multiple mesenchymal niches forming cohesive cords connected to the ductal epithelium (E–H). Double immunofluorescence showing vimentin‐positive cells connected (I) an integrated (J and K) in the eccrine ductal branches, and showing co‐expression of CK and vimentin (J and K) as well as E‐cadherin and vimentin (L–O). (A–H) Immunoperoxidase with DAB chromogen; (I–K) double immunofluorescence: CK green and vimentin red; (L–O) E‐cadherin green and vimentin red.

At the distal part and on the lateral faces of the nascent branches of the eccrine ducts, vimentin‐positive cells aggregated to form cohesive, solid nests and cords that were connected and integrated with the pre‐existing epithelium by apposition (Figure 2E,H). Double immunofluorescence confirmed that these cohesive aggregates progressively acquired cytokeratin and E‐cadherin expression, as supported by the detection of a heterogeneous population within the peripheral part of the eccrine epithelial branches that co‐expressed vimentin and cytokeratins (Figure 2I,K) as well as vimentin and E‐cadherin (Figure 2L,O).

Discussion

Experimental models for skin appendage formation in birds have demonstrated that PKA activation leads to feather primordia formation by increasing mesenchymal dermal cell condensation. ¹⁵ , ¹⁶ Our finding of vimentin‐ and S100A4‐positive dermal spindle cell condensations surrounding the abnormal hair follicle placode and germs is in accordance with the recreation of hair follicle development in SAM. ¹⁴

Moreover, it was assumed that, in contrast to hair follicle development, no dermal niche is present in eccrine gland development. However, recent work has demonstrated that there is in fact a dermal niche of S100A4+ stromal cells which guide eccrine gland development and disappears in mature adult glands. ¹⁷ Therefore, our finding of the presence of numerous S100A4‐positive cells oriented to the branching buds of eccrine ducts supports the concept that they are replicating the eccrine gland development in each one of the branches. In this regard, it is interesting to note that S100A4‐positive cells are also involved in normal mammary gland development ¹⁸ which, in contrast to the eccrine gland, has a normal branching architecture. We were indeed surprised by the presence of branched eccrine ducts in two cases. The case from the Carney's complex patient exhibited the highest complexity and mimicked fibrofolliculoma. Our findings of aggregation and apposition of vimentin‐positive stromal cells, and progressive acquisition of cytokeratin and E‐cadherin expression and a polygonal shape, while many of them still maintaining vimentin co‐expression, suggest the involvement of a mesenchymal–epithelial transition (MET) in eccrine duct branching in SAM.

Similar to the findings of Hafeez et al. ¹¹ we observed loss of PRKAR1A in myxoid spindle and stellate cells in most cases, including those with eccrine duct branching. In fact, the case with the most prominent eccrine duct branching had a germline PRKAR1A pathogenic mutation which is predicted to give rise to a truncated protein of 230 amino acids in place of the normal 381 amino acid protein. In this regard, it has been demonstrated that inactivating PRKAR1A mutations increase intracellular cAMP levels, which in turn lead to a hyperfunctional state of PKA and epigenetic reprogramming through the activation of the histone demethylase PHF2, a PKA substrate, promoting the epithelial differentiation of cells in the cAMP‐induced MET. ¹⁹ , ²⁰ MET is common during embryogenesis and organogenesis where the motile mesenchymal cells are converted to polarized epithelial cells. ²¹ Indeed, during morphogenesis, MET initiates with the formation of a sheet of epithelial cells that adhere to adjacent cells at apical–lateral borders through cadherin junctions. ²¹ Regarding our finding of the orientation of numerous S100A4 cells to the nascent eccrine buds, we should not forget that PKA has also an important role in cellular migration, ²² and that hypoxia has also been shown to induce MET in S100A4‐positive cancer‐associated fibroblasts. ²³

The PRKAR1A mutation in Carney's complex not only predisposes patients to primary myxomatous lesions, but also to adnexal epithelial proliferative growths, sometimes mimicking primary adnexal neoplasia such as trichofolliculoma, fibrofolliculoma or trichodiscoma associated with the myxomatous elements. ²⁴ , ²⁵ , ²⁶ In this regard, several authors believe that some adnexal lesions in Carney's complex patients are closer to authentic adnexal neoplasms with myxoid stroma than to myxomas with epithelial components. ²⁴ , ²⁵ , ²⁶ Our finding of fibrofolliculoma‐like labyrinthine branched eccrine ductal epithelial structures in a patient with Carney's complex could also raise the possibility of a true adnexal neoplasm. However, the fact that similar to Hafeez et al. ¹¹ we found loss of PRKAR1A in myxoid spindle and stellate cells in most cases but never in hair follicles or eccrine epithelial structures, suggests that PRKAR1A‐deficient myxoid cells induce non‐neoplastic growth of hair follicles and/or eccrine structures via mesenchymal niches, thereby mimicking normal adnexal development. Furthermore, our findings reveal that MET plays a role in the complex branching of eccrine ducts. However, we found no evidence to suggest that MET is involved in the growth of hair follicle epithelial elements.

In conclusion, we have shown that the formation of S100A4‐positive mesenchymal niches is associated with the induction of non‐neoplastic follicular and eccrine epithelial elements in SAM. Moreover, we provide morphological and immunophenotypic evidence suggesting the involvement of the PRKAR1A loss in myxoid tumour cells in the aberrant branching of the eccrine ductal epithelium within SAM through mesenchymal‐to‐epithelial transition.

Authors contributions

C.M. performed study concept and design; L.T. provided clinical information; C.M., L.T., E.A., P.M., M.E. and M.O. performed material preparation, data collection, and analysis and interpretation of data. C.M. wrote the manuscript. All authors read and approved the final manuscript.

Funding information

The authors received no specific funding for this work.

Conflicts of interest

The authors declare that they have no conflicts of interest.

Supporting information

Table S1. The antibodies used in the immunohistochemical study.

Data availability statement

Data are available on request from the authors.