The role of α-smooth muscle actin in confirming the microinvasion of laryngeal squamous cell carcinoma.

BACKGROUND
The diagnosis of microinvasive laryngeal squamous cell carcinoma (LSCC) is not always straightforward and sometimes can be very challenge in daily clinical practice. The focus lies in the confirmation of microinvasion. Cancer-associated fibroblasts (CAFs), as the major element of reactive tumor stroma, are believed to participate actively in the growth and invasion of tumor cells.


OBJECTIVES
To evaluate the diagnostic role of α-smooth muscle actin (α-SMA) labelling CAFs in microinvasive LSCC.


METHODS
A total of 81 laryngeal biopsy specimens were retrieved, including 41 cases of microinvasive LSCC with depth of invasion no more than 3 mm, 20 laryngeal squamous intraepithelial lesion (SIL), and 20 benign pseudoepitheliomatous hyperplasia (PEH). All cases were stained for immunohistochemistry, using antibody against the α-SMA antigen. The correlation between the presence of CAFs in microinvasive LSCC and tumor histological characteristics was investigated.


RESULTS
Immunoreactivity of α-SMA was detected in twenty-nine microinvasive LSCC (29/41, 71%), while no reactivity was observed in laryngeal SIL (0/20, 0%), and rarely in PEH (2/20, 10%). The α-SMA expression pattern in stroma of microinvasive LSCC was significantly different from that of SIL (χ2￼ = 26.966, p = 0.000) and PEH (χ2￼ = 19.838, p = 0.000). In addition, there seemed to be a certain correlation between the histological characteristics of microinvasive LSCC and the presence of interstitial CAFs.


CONCLUSIONS
This study highlights the practical role of utilizing α-SMA in the pathological diagnosis of microinvasive LSCC, with emphasis on variable histomorphologic features of microinvasive LSCC.


Introduction
As a common malignant otorhinolaryngologic neoplasm, laryngeal cancer accounts for 90% of malignancies occurring in head and neck region [1,2]. Of such, squamous cell carcinoma is the most common neoplasm accounting for more than 95% of all the laryngeal carcinoma [1,3]. The prognosis of advanced laryngeal cancer is currently not optimal [4]. However, the diagnosis of mLSCC is not always straightforward, particularly in the circumstances with differentials such as SIL with inverted growth pattern and PEH.
MLSCC is de ned as LSCC that have penetrated the basement membrane and in ltrated into the super cial compartment of the lamina propria [1,3]. Currently, the con rmation for microinvasion of LSCC is mainly based on constellate histomorphologic features, including tumor cells penetrating basement membrane of the surface epithelium, the downward growth of tumor islands, cords, or isolated tumor cells, with associated desmoplastic stromal changes, excessive deposition of the extracellular matrix, and neovascularization. However, the distinct basement membrane border is usually not clear, the invasive carcinomas may be extremely well-differentiated, and there are no clear-cut criteria for tumorrelated desmoplasia.
Previous studies on the pathological diagnosis of LSCC always focus on the tumor cells [5,6] , we sought to explore the diagnosis and differential diagnosis of mLSCC from tumor stroma. Fibroblasts activated in the tumor microenvironment, named as myo broblasts, peritumoral broblasts or cancer-associated broblasts (CAFs), has been found typically expressing α-SMA and sharing some characteristics of broblasts and smooth muscle cells [7]. As a major component of tumor stroma, CAFs are believed to play a major role in tumor growth, invasion, and metastasis [8][9][10].
Our hypothesis is that α-SMA can be used as an ancillary biomarker to help with rendering de nite diagnosis of mLSCC. In our study, we sought to investigate the presence and expression pattern of α-SMA positive broblasts in mLSCC with comparison with H&E in-situ histomorphology, and compare with the results from laryngeal SIL and benign PEH. Additional 5 biopsy of normal adult laryngeal mucosa used as negative controls. All cases were independently reviewed by Three senior pathologists (YB, CY and HL), and any cases with diagnostic disconcordance were excluded. A representative section was selected for immunohistochemical staining.

Materials And
Clinical data collected includes age, gender, and tumor laterality.
Based on the depth of invasion (DOI), the 41 samples of mLSCC were divided into three subgroups.
Group 1-1 DOI ≤1 mm, Group 1-2 DOI more than 1mm but ≤2 mm, Group 1-3 DOI more than 2 mm but ≤3 mm. The depths of the invasion were measured from the basement membrane of the epithelium to the deepest point of invasion of the tumor by using microscope rulers.

Histological Evaluation and Immunohistochemical staining
Formalin-xed para n-embedded tissues were sectioned (3 μm), and stained for immunohistochemistry, using antibody against the α-SMA antigen (clone UMAB237, OriGene) and Peroxidase-Streptavidin method. Positive and negative controls gave appropriate results for each procedure. The presence of brownish-yellow cytoplasmic staining was considered to be a positive result. Three-tier system has been used to quantitative evaluate the immunoreactivities. More than 50% of broblasts positive for α-SMA were considered as diffusely positive, between 10-50% of positive broblasts were considered as focal positive, and less than 10% positive broblasts were rare or scattered positive. Positive endothelial cells were counted as positive internal controls and excluded from the analysis.

Statistical Analysis
Chi-square statistics test was used to compare the expression pattern of α-SMA in stromal broblasts. Statistical software SPSS 23.0 was used for statistical analysis. A p-value of less than 0.05 was considered statistically signi cant.

Clinical features
Among the 41 cases of mLSCC , 39 cases were men (39/41, 95.1%) and 2 cases were women (2/41, 4.9%). Patients' age ranged from 44 to 81 years (mean 60 years). There were 35 cases occurred in the vocal cords (35/41, 85.4%), of which 20 cases in the right vocal cords, and 15 cases in the left vocal cords. The remaining 6 cases occurred in other parts of the larynx, including 3 cases at anterior joint, 2 cases at right pseudo-vocal cord, and 1 case at right arytenoid zone. ( Table 1) 3

.2 Histologic Features and Immunohistochemical results
Grossly, 36 of 41 mLSCC were at/non exophytic type, and 5 cases were exophytic papillary or verrucous type. Microscopically, 29 cases showed an in ltrative growth pattern, which were characterized by small irregular tumor nests, cords and/ or single tumor cells with poorly de ned in ltrating borders, while 12 cases showed an expansive growth pattern, which were characterized by large tumor islands with wellde ned pushing borders. Varying degrees of desmoplastic stromal reaction were detected in 21 cases. 7 cases showed densely chronic in ammatory in ltrates with no de nite desmoplastic reaction identi ed, and 13 cases showed no distinct stromal changes. ( Table 2 ) Of  Table 2. Of the 29 α-SMA positive cases of mLSCC, there were 4 cases in Group 1-1, 9 cases in Group 1-2, and 16 cases in Group 1-3. The positive rate in Group 1-2 and 1-3 (DOI greater than 1 mm) were signi cantly higher than that in Group 1-1 (DOI≤1 mm) ( = 4.230, p=0.040) ( Table 3). OF the 12 cases of α-SMA negative mLSCC, half of them showed densely chronic in ammatory in ltrates (Fig. 3). And the remaining 6 negative cases showed no distinct stromal changes, of which 5 cases belonged to Group 1-1 (DOI≤1 mm) and 1 case belonged to Group 1-2 (1 mm DOI ≤2 mm ). In H&E stained sections, the stroma corresponding to α-SMA positive areas in mLSCC were rich in collagen with increased stromal cellularity. The desmoplastic stromal cells /CAFs appeared around the epithelial tumor nests, arranging in streaming patterns.

Discussion
In our study, mLSCC were most often occurred in men, mainly in elders. The vocal cord was the most common site with right predominancy. Grossly, mLSCC were mainly at with a few cases showing exophytic or verrucous papillary con gurations. Histomorphologically, the growth pattern at the invasive front of mLSCC is mainly in ltrative, but the expansive growth pattern are not uncommon. Three types of tumor associated stromal changes were observed in our study, typical desmoplastic stromal changes, densely chronic in ammatory in ltrates, and no distinct stromal changes assembling adjacent normal lamina propria. Among these, desmoplastic reaction is the most common feature observed in mLSCC.
As the major component of tumor microenvironment, CAFs are found to be present in a variety of tumors such as esophageal cancer, lung cancer, hepatocellular carcinoma, and kidney cancer. Recent studies showed that the detection of CAFs can aid in the diagnosis of microinvasive ductal carcinoma, endoscopically removed invasive colorectal adenocarcinoma, and assist in differentiating between pancreatic ductal adenocarcinoma and chronic pancreatitis [11][12][13] [14,15]. In our study, majority of mLSCC (29/41, 70.7%) showed positive reactivity of α-SMA on CAFs. In contrast, there was no or few immunoreactivity identi ed in laryngeal SIL and PEH (0/20, 0% ; 2/20, 10%). SIL and PEH are lesions that often need to be differentiated from mLSCC. Although it is said that SIL showed an intact basement membrane, but this is di cult to identify histologically. SIL with inverted growth pattern or tangentialing section can also be di cult to distinguish from invasive carcinoma. PEH, a not uncommon benign lesion occurring in the laryngeal mucosa, can be seen in chronic irritation, non-speci c in ammation, tumors such as granular cell tumor, laryngeal tuberculosis and fungal infections, or keratosis [16][17][18]. As the name PEH implies, the histomorphologic appearance assemble to invasive carcinoma and characterized with mucosal epithelial hyperplasia showing sub-epithelial extension in lamina propria. The irregular epithelial projections in PEH is di cult to distinguish from the invasive nests of well-differentiated squamous cell carcinoma. In our study, the α-SMA expression pattern in stromal broblasts of mLSCC were signi cantly different both from those of SIL and PEH. This support that, the detection of CAFs by immunohistochemistry against α-SMA plays a valuable role in con rmatory diagnosis of mLSCC, and would be used as a reliable marker for the diagnosis of mLSCC and differentiated from SIL and PEH.
The origin of CAFs has been under debate for a long time. Several hypotheses have been proposed for the origin of CAFs such as resident tissue broblasts, bone marrow-derived mesenchymal stem cells, or epithelial cells [19]. A commonly accepted theory for CAFs origin points to resident tissue broblasts, which are activated by TGF-β1 and converted into CAFs [20]. TGF-β1, mainly localized in exosomes released by cancer cells, can promote the proliferation and expression of CAF markers [21]. Then CAFs secrete various cytokines, chemokines and in ammatory mediators such as stromal cell-derived factor 1 (SDF-1/CXCL12) in uence the growth and invasion of tumor [22]. This can explain why CAFs can be widely detected by α-SMA in mLSCC but rarely in laryngeal SIL and PEH. As for the two cases of PEH with SMA positivity in the interstitium, we believe that this is due to interstitial myo broblasts transformation caused by brosis. Fibrosis in vascular vocal cord polyps seems not uncommon. As for the positive case of keratosis, the patient had undergone surgical resection months ago, and the brosis should be the repair for injury caused by the previous surgery. Different from the SMA positive broblasts in mLSCC appearing around the invasive epithelial tumor nests, positive cells in PEH can appear in lamina propria covered with at epithelium or in the interstitium away from epithelium. And broblasts surrounding the downward growing epithelial nests in PEH may be SMA negative. The different distribution characteristics of SMA -positive broblasts can be used to distinguish CAFs from myo broblasts in brosis. The driving force of myo broblast differentiation in brosis are mainly mechanical tension and TGF-β1 which is released from a variety of in ammatory cells and platelets in the microenvironment of damaged or brotic tissue [23].
Comparing the histomorphologic features with the results of immunohistochemical staining of mLSCC, it was found that α-SMA positive cases, especially diffusely positive cases, were concentrated in cases showed at/non-exophytic gross appearance, in ltrative growth pattern, or with desmoplastic stromal reaction in the stroma. While cases showed exophytic gross appearance, expansive growth pattern or dense lymphoplasmacytic in ltration in the stroma tend to be focal, scattered positive or negative.
Especially the interstitial dense lymphoplasmacytic in ltration, our study support that there is a negative correlation between this characteristics and the presence of CAFs in LSCC. Similar ndings were reported by Zidar N et al [24]. In our study, six of 7 cases of mLSCC with dense lymphoplasmacytic in ltration were stromal negative for α-SMA, and the remaining 1 case was focally positive. Furthermore, 5 of these 6 cases with α-SMA negative labelling showed expansive growth pattern at the invasive front. It has been reported that pushing borders and good cohesion of the deep invasive front, and high levels of tumorin ltrating lymphocytes are indicators of good prognosis for laryngeal squamous cell carcinoma [25,26]. This may be related to less or no CAFs formation in stroma under these conditions, for it is reported that overexpression of CAFs correlates with poor prognosis in some tumors [27,28]. However, due to small sample and not enough long term follow-up data available in current study, a larger scale study is needed to provide a further con rmatory.
In our cohort, a small percentage (12/41, 29.3%) of mLSCC showed no immunoreactivity identi ed with α-SMA in the stroma, 6 of which showed dense lymphoplasmacytic in ltration as we said above, and the remaining 6 showed no distinct stromal changes. While, study from Kojc N et al reported that all cases of LSCC contain α-SMA positive stromal cells [29]. This difference may arise from the variabilities of sampling tissue and cases selection in some degree. There was no information about depth of invasion of the LSCC mentioned in their study. In our cohort, we collected mLSCC with depth of invasion not greater than 3 mm. We speculate that it may be due to the poor stromal response when tumor invades super cially. In our study, we compared positivity rate of α-SMA between cases with DOI less than 1 mm and DOI greater than 1mm, we found that the higher positive frequency of α-SMA was observed in mLSCC with DOI greater than 1 mm (P<0.05) . This indicates CAF formation is gradually enhanced along with the depth of tumor invasion. In addition, of the 6 α-SMA negative mLSCC cases showed no distinct histological changes in the stroma, 5 cases belonged to Group 1-1 (DOI≤1 mm) and 1 case belonged to Group 1-2 (1 mm DOI ≤2 mm ). This most likely due to the fact that the depth of tumor invasion is too super cial, resulting in a too weak stromal response to be detected and the tumor morphological characteristics have not yet been shown, which further support our hypothesis.
In summary, mLSCC is predominantly at/non-exophytic grossly, and often histologically shows an in ltrative growth pattern. Varying degrees of desmoplastic stromal reaction is observed in most cases, but in a few cases it is replaced by dense lymphoplasmacytic in ltration in the stroma. Our study support that there is a negative correlation between dense lymphoplasmacytic in ltration in the stroma and the presence of CAFs in LSCC. Our study further supports α-SMA, as a surrogate marker for CAFs, plays a reliable role in con rmatory invasion for mLSCC.