Intercellular adhesion proteins are poorly characterized in acantholytic squamous cell carcinoma (ASCC), a more aggressive tumor than nonacantholytic invasive well-differentiated squamous cell carcinoma (SCC) of the skin. In this study we compared expression of Dsg3, E-cadherin, and syndecan-1 in ASCC and SCC. Immunohistochemical detection of Dsg3, E-cadherin, and syndecan-1 in 22 ASCCs and 22 SCCs was graded on a semiquantitative scale for intensity of staining (SI) and degree of circumferential staining (CS) about the cell membrane. Results were assessed by means of conditional logistic regression and χ2 analysis. Dsg3 and E-cadherin expression (SI, CS) was significantly decreased (P < .05) in ASCC compared with SCC, whereas staining for syndecan-1 was similar in the 2 tumor types. Differences in expression of adhesion markers between ASCC and SCC may contribute to the development of acantholysis in ASCC and its more aggressive biologic behavior.
Cutaneous squamous cell carcinoma (SCC) is one of the most common malignancies, responsible for most nonmelanoma skin cancer deaths in the United States.1 Acantholytic squamous cell carcinoma (ASCC), a subtype described by Lever in 1947,2 is characterized by proliferation of keratinocytes that form gland-like structures with variably sized, discohesive, acantholytic cells. ASCC expresses cytokeratins but not glandular and endothelial cell markers, which distinguishes it from eccrine and endothelial neoplasms.3–5 ASCC has been classified as an intermediate risk variant of SCC, with estimated metastasis rates of 3% to 10%4 and mortality rates of 3% to 19%.5,6 Whether other variables, such as tumor thickness and histopathologic grade, influence mortality data has been called into question in a recent literature review.7
Although acantholysis in pemphigus and inherited conditions such as Hailey-Hailey disease is well-studied,8 the mechanism of acantholysis in squamous neoplasia is incompletely characterized. Dysregulation and altered expression of intercellular adhesion molecules have been studied in SCC,9–13 but the literature offers limited insight into these abnormalities in ASCC.10,14 Using a semiquantitative scale of immunohistochemical staining levels, Bayer-Garner and Smoller10 found reduced expression of E-cadherin and syndecan-1 in ASCC; however, Moreno-Maldonado and colleagues14 found no such decrease in E-cadherin expression. Altered expression of E-cadherin, normally localized to the adherens junction in all epithelia,11 has been demonstrated in cutaneous SCC and SCC of the head and neck region.10,12,15–17 Moreover, aberrant E-cadherin expression may correlate with prognosis in some tumors,16–18 including Merkel cell carcinoma, where it was associated with decreased recurrence-free survival.19 The transmembrane heparin sulfate proteoglycan and signaling coreceptor syndecan-1 appears to be important in maintaining an epithelial phenotype through regulation of actin and E-cadherin.20–22 Decreased expression of syndecan-1 has been demonstrated in multiple malignancies including hepatocellular, head and neck, and laryngeal carcinomas. In addition, reduced syndecan-1 expression has been shown to correlate with reduced cell-cell adhesion in multiple myeloma cell lines20,23–26 and with a dedifferentiated phenotype of SCC.20 Furthermore, loss of syndecan-1 expression may also be associated with loss of E-cadherin expression.21
Another intercellular adhesion protein, desmoglein 3 (Dsg3), is a desmosomal cadherin expressed solely in stratified squamous epithelia. Several studies have examined desmoglein expression in malignancy,22,27–29 including cutaneous SCC.9 Interestingly, a patient who developed lymph node metastases of a cutaneous SCC that expressed desmoglein 1 (Dsg1) later acquired pemphigus foliaceus, an acantholytic immunobullous disease associated with autoantibodies directed against Dsg1.30 Although Dsg3 expression is increased in SCC of the lung and upper respiratory tract,29 the role of this particular protein in cutaneous ASCC has not been evaluated. In a patient with pemphigus vulgaris (associated with anti-Dsg3 autoantibodies) and SCC, we observed intratumoral acantholysis as well as suprabasilar acantholysis of non-neoplastic surrounding epidermis that was clinically uninvolved by pemphigus (unpublished observations, 2012). This case raises the possibility that ASCC may be associated with altered Dsg3 expression.
Because Dsg3, E-cadherin, and syndecan-1 each may play a role in tumor biology, we sought to study their expression in ASCC compared with invasive, well-differentiated nonacantholytic SCC.
Materials and Methods
The Mayo Clinic (Rochester, MN) institutional review board approved this study. To compare the immunohistochemical staining patterns of Dsg3, E-cadherin, and syndecan-1 in ASCC with those in SCC, we performed a retrospective case-control study of patients with invasive cutaneous SCCs diagnosed at our institution between January 1992 and March 2011. ASCCs and SCCs were identified on histopathologic review of H&E—stained sections obtained during each patient’s episode of care. The SCC controls were matched for age within one decade, gender, and tumor location. Tumors with features of keratoacanthoma or for which complete microscopic and demographic data were unavailable were excluded. Clinical and pathologic characteristics recorded included age, gender, tumor location, tumor thickness, tumor diameter, and the presence of perineural invasion.
Antibodies against Dsg3 (clone 5G11, diluted 1:200, Novus Biologicals, Littleton, CO), E-cadherin (clone 4A2C7, diluted 1:100, Invitrogen, Carlsbad, CA), and syndecan-1/CD-138 (clone MI15, diluted 1:50, DAKO, Carpinteria, CA) were used to stain paraffin-embedded sections. Expression of each antibody was evaluated as follows: for ASCCs, only the invasive, acantholytic portion was graded; for SCCs, invasive areas of the tumor were graded. Three authors (J.R.G., C.C.W., and J.S.L.) independently evaluated the staining patterns of each specimen using 2 separate criteria: (1) intensity of staining (SI) compared with adjacent non-neoplastic epidermis and (2) the degree to which the stain was circumferential (CS) about the cell membrane of tumor cells Table 1. For SI, a score of 3 was considered increased compared with adjacent non-neoplastic epidermis, 2 was equivalent to uninvolved epidermis, 1 was decreased, and 0 was absent. A CS score of 2 was assigned when staining of more than 75% of tumor cells was completely circumferential, a score of 1 when 25% to 75% of cells demonstrated complete CS, and 0 when less than 25% of cells demonstrated complete CS. Scores were compiled and the majority score was recorded. If a majority score was not reached, the case was reviewed in the presence of all observers and a consensus score recorded.
Data analysis was performed using Stata 10.0 software (Stata Corp, College Station, TX). Statistical significance was assessed using conditional logistic regression. For immunohistochemical staining of Dsg3, E-cadherin, and syndecan-1, the χ2 statistic of the likelihood ratio test was used to assess significance for all levels of the variable. P value less than .05 was considered statistically significant in all analyses. All statistical tests were 2-sided.
Expression of Dsg3, E-cadherin, and syndecan-1 was analyzed in 22 ASCCs and 22 SCCs Image 1 and Image 2. All patients were men and most tumors were located on the scalp, face, or neck Table 2. For ASCC, SI of Dsg3 (P = .0004) and E-cadherin (P = .02) was significantly decreased compared with SCC, but the decrease in SI of syndecan-1 was of borderline significance (P = .1) Table 3 and Figure 1. For ASCC, CS of Dsg3 (P = .0002) and E-cadherin (P = .03) was decreased compared with SCC, but not of syndecan-1 (P = .2) (Table 3) Figure 2.
We found that immunohistochemical expression of Dsg3 and E-cadherin was decreased in ASCC compared with SCC in most fields evaluated. Decrease of syndecan-1 expression was of borderline significance. The mechanism for these differences may be reduced gene expression, production of an altered molecule, or post-translational modifications. Decrease in adhesion molecules may contribute to an acantholytic phenotype in ASCC. Syndecan-1 expression was not significantly different between ASCC and SCC, perhaps because of tumor heterogeneity. Alternatively, syndecan-1 may play a lesser role in maintenance of tumor cell cohesion than do Dsg3 and E-cadherin. Our findings regarding E-cadherin are in keeping with those reported by Bayer-Garner and Smoller,10 who demonstrated a decrease in E-cadherin in ASCCs. In contrast, More-no-Maldonado and colleagues14 reported increased expression of E-cadherin in ASCC compared with SCC. We did not detect the decrease in syndecan-1 expression that Bayer-Garner and Smoller10 reported when using a dual-staining technique for both E-cadherin and syndecan-1.
Limitations of this study included the subjectivity and interobserver variability associated with the grading of immunohistochemical staining. Moreover, adjacent nonneoplastic epidermis was used for comparison of staining pattern and intensity; it is possible that some dysregulation of expression was present in the normal-appearing epidermis around the tumors.
Nevertheless, our results add to the literature on cell-cell adhesion molecules in neoplasia, specifically ASCC. We report the novel finding of reduced Dsg3 expression and validate prior reports of reduced E-cadherin expression in ASCC compared with SCC. These results provide evidence that lower expression of intercellular adhesion molecules may contribute to intratumoral acantholysis in cutaneous ASCC. We suggest that future studies are needed to investigate the mechanism by which altered intercellular adhesion protein expression might contribute to the more aggressive biologic behavior of these tumors and their prognoses.