Morphopathogenesis of Adult Acquired Cholesteatoma

Background and Objectives. The aim of this study was to compare the distribution of proliferation markers (Ki-67, NF-κβ), tissue-remodeling factors (MMP-2, MMP-9, TIMP-2, TIMP-4), vascular endothelial growth factor (VEGF), interleukins (IL-1 and IL-10), human beta defensins (HβD-2 and HβD-4) and Sonic hedgehog gene protein in cholesteatoma and control skin. Methods. Nineteen patient cholesteatoma tissues and seven control skin materials from cadavers were included in the study and stained immunohistochemically. Results. Statistically discernible differences were found between the following: the Ki-67 in the matrix and the Ki-67 in the skin epithelium (p = 0.000); the Ki-67 in the perimatrix and the Ki-67 in the connective tissue (p = 0.010); the NF-κβ in the cholesteatoma matrix and the NF-κβ in the epithelium (p = 0.001); the MMP-9 in the matrix and the MMP-9 in the epithelium (p = 0.008); the HβD-2 in the perimatrix and the HβD-2 in the connective tissue (p = 0.004); and the Shh in the cholesteatoma’s perimatrix and the Shh in the skin’s connective tissue (p = 0.000). Conclusion. The elevation of Ki-67 and NF-κβ suggests the induction of cellular proliferation in the cholesteatoma. Intercorrelations between VEGF, NF-κβ and TIMP-2 induce neo-angiogenesis in adult cholesteatoma. The similarity in the expression of IL-1 and IL-10 suggests the dysregulation of the local immune status in cholesteatoma. The overexpression of the Sonic hedgehog gene protein in the cholesteatoma proves the selective local stimulation of perimatrix development.


Introduction
The worldwide incidence of acquired adult cholesteatoma is from 9 to 12.6 cases per 100,000 adults annually [1]. Histologically, this benign tumor is composed of three parts. The innermost part contains anucleate epithelial cells, which form a cystic layer. The second part is the hyperproliferative squamous epithelial layer-the matrix. The outer part, the perimatrix, is a granulation tissue rich in different inflammatory cells [2]. Although cholesteatoma is benign, it acts destructively towards the surrounding tissue in the temporal bone [3]. However, the complex etiopathogenesis of cholesteatoma is uncertain.
Since cholesteatoma is constantly proliferating, the Ki-67 is one of the most frequently used markers to detect proliferation in cholesteatoma tissue [4]. Recent studies suggest that Ki-67 could be a prognostic factor for cholesteatoma's destructiveness in the middle ear. However, this marker cannot predict the recurrence of cholesteatoma [5]. Another cell factor that causes keratinocyte proliferation in cholesteatoma is the nuclear factor kappa beta (NF-κβ) [6]. It was shown that NF-κβ prevents epithelial cells in the cholesteatoma matrix from entering apoptosis and, therefore, that the growth and expansion of the cholesteatoma is supported by NF-κβ via this mechanism as well [7]. Additionally, Hamajima et al. [7] demonstrated that Ki-67 and NF-κβ act together to induce the proliferation process in the matrix.
In the patient group, 24 patients participated in this study, 11 male and 13 female (ages varied from 19 to 75 years, mean age 36, 37 years). Inclusion criterion was acquired adult cholesteatoma. Five patients were excluded from the study. The exclusion reasons were incomplete cholesteatoma material (cholesteatoma without matrix and/or perimatrix), which was invalid for immunohistochemical analysis.
In the control group, ten deep external meatal skin-tissue samples were obtained from 10 different cadavers (ages ranging from 35 to 70 years) in a collection from the Institute of Anatomy and Anthropology. The use of the cadaver material was approved by the Ethical Committee of the Riga Stradin , š University (29 October 2022; 2-PĒK-4/475/2022). Inclusion criterion was adults with no chronic ear or skin diseases. Three control-group skin samples were excluded because of insufficient skin material, which was invalid for immunohistochemical analysis.
Next step included rinsing of tissue samples in wash buffer (TRIS; T0083, Diapath S.p.A., Martinengo, Italy) two times for 5 min, followed by placing them in a microwave oven for up to 20 min in boiling EDTA buffer (T0103, Diapath S.p.A., Martinengo, Italy) and then cooling them down to 65 • C (~20 min). The specimen was placed in a TRIS wash buffer and blocking with 3% peroxidase block (925B-02, Cell Marque, Rocklin, CA, USA) was performed for 10 min. All antibodies used in research were diluted with Antibody Diluent (938B-05, Cell Marque, Rocklin, CA, USA).
The HiDef DetectionTM HRP polymer system (954D-30, Cell Marque, Rocklin, CA, USA) was used for the antibodies of mouse or rabbit origin. Slides were rinsed 5 times (5 min each) with TRIS buffer solution. Next, HiDef DetectionTM reaction amplifier (954D-31, Cell Marque, Rocklin, CA, USA) was applied for 10 min at room temperature. After this processing, the preparations were rinsed five times (for five minutes each time) in distilled water. After rinsing, HRP chromogen (used with DAB Buffer) (957D-30, Cell Marque, Rocklin, CA, USA) was used for 3-5 min. Chromogen was made fresh for each application. Subsequently, slides were rinsed 5 times with TRIS buffer solution. Next, slides were placed in a slide basket and immersed in filtered hematoxylin for 30-60 s. After staining with hematoxyline, the micro-preparations were rinsed in distilled water five times and dehydrated in alcohols (at 95% and 100% for 3 min), after which they were immersed in 3 containers with Xylene (5 min each), dried and covered with glue Pertex ® (00801-EX, HistoLab, Västra Frölunda, Sweden) glue. Positive controls in accordance with the companies guidelines and negative controls (Supplementary File, Figure S1) with exclusion of primary antibody were developed.
The slides were analyzed by light microscopy by two independent morphologists using semi-quantitative method. The results were evaluated by grading the appearance of positively stained cells in the visual field; multiple sections for each sample were scored. Structures in the visual field were labeled as follows: 0 = no positive structures, 0/+ = occasional positive structures, + = few positive structures, +/++ = small-to-moderate number of positive structures, ++ = moderate number of positive structures, ++/+++ = moderate-to-numerous positive structures, +++ = numerous positive structures, +++/++++ = numerous-to-abundant structures, ++++ = an abundance of positive structures in the visual field [29] (Supplementary File, Figure S2).
For a visual illustration, a Leica DC 300F digital camera and image-processing-andanalysis software, Image-Pro Plus (Media Cybernetics, Inc., Rockville, MD, USA) were used.
Spearman's rank correlation coefficient was used to determine correlations between factors, where r = 0-0.2 was assumed as a very weak correlation, r = 0.2-0.4 a weak correlation, r = 0.4-0.6 a moderate correlation, r = 0.6-0.8 a strong correlation and r = 0.8-1.0 a very strong correlation. The Mann-Whitney U test was used to analyze the control group versus the patient data. The level of significance for tests was chosen as 5% (p-value < 0.05).

Description of the Tissue
In the routine histological examination with hematoxylin and eosin (H-E), all three parts of the cholesteatoma were visualized. The outer part, the perimatrix, was composed of different inflammatory cells, including lymphocytes, leukocytes, macrophages, as well as fibrocytes, collagen fibers and blood vessels. The middle part was hyperproliferative epithelium, known as the matrix. The inner part, the cystic layer, was an anucleate keratin mass ( Figure 1a). The control tissue was deep-external-ear-canal skin that presented an intact stratified squamous epithelium and connective tissue without inflammation ( Figure 1b).

IHC Findings of Proliferation Markers
The Ki-67-positive cells varied from no (0) positive cells to few-to-moderate (+/++) positive cells in cholesteatoma, but there were mostly no Ki-67 positive cells in the control group.
The appearance of the NF-κβ-containing cells was graded from no (0) to numerous (+++) positive cells in the cholesteatoma and from no (0) to moderate (++; Figure 2a

IHC Findings on the Angiogenetic Factor
The vascular endothelial growth factor in the cholesteatoma presented a variance from no (0) to numerous (+++) positive cells. A similar distribution was seen in the control group (Figure 3a,b).

IHC Findings on the Angiogenetic Factor
The vascular endothelial growth factor in the cholesteatoma presented a variance from no (0) to numerous (+++) positive cells. A similar distribution was seen in the control group (Figure 3a,b).

IHC Findings on the Tissue-Remodeling Factors
A range from no (0) to numerous-to-abundant (+++/++++) of MMP-2-containing cells was detected in the patient group, but in the control group, a range from no (0) to moderate-to-numerous (++/+++) MMP-2-containing cells was found.

IHC Findings on the Tissue-Remodeling Factors
A range from no (0) to numerous-to-abundant (+++/++++) of MMP-2-containing cells was detected in the patient group, but in the control group, a range from no (0) to moderate-to-numerous (++/+++) MMP-2-containing cells was found.

IHC Findings of Shh Gene Protein
In the patient group, the Shh findings demonstrated a range from occasional (0/+) to abundant (++++) positive cells, but in the control group, the range was from none (0) to numerous-to-abundant (+++/++++; Figure 8a,b).

Statistical Analysis
To determine the difference between the groups, we used a Mann-Whitney U test.  The distribution of the tissue-defensin-HβD-2-containing cells was graded from none (0) to moderate-to-numerous (++/+++) and the HβD-4-positive cells displayed a variance from no (0) to numerous (+++) positive cells. However, in the control group, the number of HβD-2-and HβD-4-positive cells ranged from none (0) to moderate (++) (Figure 7a-d).

IHC Findings of Shh Gene Protein
In the patient group, the Shh findings demonstrated a range from occasional (0/+) to abundant (++++) positive cells, but in the control group, the range was from none (0) to numerous-to-abundant (+++/++++; Figure 8a

Statistical Analysis
To determine the difference between the groups, we used a Mann-Whitney U test. There were statistically significant differences between the following: the Ki-67 in the matrix and the Ki-67 in the skin epithelium (p = 0.000); the Ki-67 in the perimatrix and the Ki-67 in the connective tissue (p = 0.010); the NF-κβ in the matrix and the NF-κβ in the control epithelium (p = 0.001); the MMP-9 in the matrix and the MMP-9 in the epithelium (p = 0.008); the HβD-2 in the perimatrix and the HβD-2 in the control connective tissue (p = 0.004); and the Shh in the perimatrix and the Shh in the control connective tissue (p = 0.000) ( Table 2). Table 2. Statistically significant differences between patient and control groups.
In the patient group, there were moderate correlations between the following: the MMP-2 and the TIMP-2 in the matrix (r = 0.482, p = 0.037); the MMP-9 and the TIMP-  Table 3).

Discussion
To prove the hyperproliferative activity of cholesteatoma cells compared to the control skin cells, we used Ki-67 and NF-κβ. Several authors presented the upregulation of Ki-67 in cholesteatoma [30][31][32]. Our results were similar and showed a statistically significant overexpression of Ki-67 in the matrix (p = 0.000) and perimatrix (p = 0.010) compared to the epithelium and connective tissue of the skin. However, controversies about this matter exist; for example, Kuczkowski et al. [33] presented a study in which the difference between the Ki-67 in cholesteatoma and in a control group was not statistically significant. Our results are substantiated by those of Heenen et al. [34], who showed limited Ki-67 activity in an unchanged epidermis and proved that these cells were in a non-proliferative state. In addition there is controversy as to whether Ki-67 overexpression is associated with the level of bone resorption. Hamed et al. [35], Juhász et al. [36] and Mallet et al. [37] concluded that cholesteatomas that cause more bone erosion have a higher expression of Ki-67 compared to those that cause less destruction. By contrast, Aslier et al. [2] did not observe a correlation between bone erosion and the expression of Ki-67.
Further, we observed a statistically discernible difference between the number of NF-κβ containing cells in the cholesteatoma matrix (p = 0.001) compared to the skin epithelium, but not in perimatrix (p = 0.055) compared to the connective tissues of the skin. However, the p-value is very close to being statistically significant, and it might be a tendency. Our results are supported by Byun et al. [6], who found increased levels of NF-κβ compared to retro-auricular skin. Our results show a moderate correlation between Ki-67 and NF-κβ (r = 0.538, p = 0.017). The explanation for these results is that Ki-67 and NF-κβ act through the same pathway (the inhibitor of the DNA binding protein 1 (Id1)→NF-κB→cyclin D1→Ki-67) to induce keratinocyte proliferation [7]. Therefore, we conclude that Ki-67 and NF-κβ can be used as proliferation markers in cholesteatoma to prove the existence of a pathologic proliferation stage in cholesteatoma cells compared to control skin.
Another characteristic pattern in acquired middle-ear cholesteatoma is a local osteolytic process in the temporal bone [38]. The remodeling factors MMP-2 and MMP-9 are associated with bone remodulation in the middle ear for patients with cholesteatoma [8,9]. In contrast to Morales et al. [39] and Olszewska et al. [40], who presented the overexpression of MMP-9 and MMP-2 in cholesteatoma in opposition to retro-auricular skin, we presented no statistically discernible differences between MMP-2 in the patient and the control groups in the soft tissues. These findings are similar to those of Banerjee [41], who did not find differences in the expression of MMP-2 in cholesteatoma and deep meatal skin. Additionally, our results presented statistically significant reduced relative numbers of MMP-9 positive cells in the matrix (p = 0.008) compared to the skin epithelium. Limited data exist on the role of decreased levels of MMP-9 in cholesteatoma or any other pathology in humans. However, Pozzi et al. [42] concluded that reduced levels of MMP-9 are associated with increased tumor angiogenesis. We can speculate as to the specific decreased-expression pattern of MMPs in cholesteatoma soft tissue, but the lack of studies on this topic prevents us from expanding the role of decreased MMP-9 in cholesteatoma tissue. Further and more specific studies are needed.
Furthermore, our results did not show differences between the relative numbers of TIMP-2 and TIMP-4 between both study groups. Nevertheless, the difference between the TIMP-2 in the matrix and that in the skin epithelium was nearly statistically discernible (p = 0.055) and was decreased in the cholesteatoma tissue, in contrast to the skin. Kaya et al. [12] also proved the downregulation of TIMP-2 in cholesteatoma compared to healthy tissue. We suggest that decreases in the activity of TIMP-2 might affect MMPs and cause an imbalance between MMPs and TIMPs. This imbalance can cause proteolysis in the extracellular matrix, which causes bone remodulation in cholesteatoma patients [8].
Neo-angiogenesis has a key role in cholesteatoma expansion [13]. However, our results did not present statistically discernible differences in the expression of VEGF between the patient and the control group. We suggest that increases in the expression of VEGF do not occur in developed blood vessels. It is known that TIMP-2 not only inhibits MMP-2 and MMP-9 but also inhibits neo-angiogenesis and VEGF directly as a separate function from inhibiting MMPs [43,44]. Our results showed a moderate positive correlation between the VEGF and the TIMP-2 (r = 0.581, p = 0.009) and decreased levels of TIMP-2 compared to the control skin. Therefore, we suggest that TIMP-2 intercorrelates with VEGF in cholesteatoma tissue and that reduced relative numbers of TIMP-2 mean reduced anti-angiogenetic properties in cholesteatoma tissue, which results in increased neo-angiogenesis. In addition, we found a positive correlation between the VEGF and the NF-κβ (r = 0.512, p = 0.025). Several authors proved that NF-κβ can regulate VEGF in cholesteatoma tissue [7,16]. These findings mark the complexity of angiogenesis in cholesteatoma.
We chose HβD-2, HβD-4, IL-1 and IL-10 to evaluate the inflammatory process in cholesteatoma. The results for the IL-1 and the IL-10 were not statistically discernible between the patient and control groups. These results are supported by those of Yetiser et al. [45] and Kuczkowski et al. [18], who also reported no difference between their patient and control groups for IL-1 and IL-10, respectively. Importantly, we found a very strong positive correlation between the IL-1 and IL-10 in the matrix and perimatrix (r = 0.820, p = 0.000) and, by contrast, opposition, a very strong negative correlation between the IL-1 and the IL-10 in the control group (r = −0.829, p = 0.021). These results might suggest that there is dysregulation between pro-and anti-inflammatory cytokines (IL-1 and IL-10) in cholesteatoma, which causes local inflammation in the middle ear. Moreover, our results showed a statistically significant overexpression of HβD-2 (p = 0.004) in the patient group, in contrast to the controls. However, the differences between both groups for HβD-4 were not statistically discernible. Similarly, the upregulation of HβD-2 in cholesteatoma versus the skin was shown by Song et al. [46] and Park et al. [24]. Furthermore, HβD-2 was more actively expressed in cholesteatoma tissue than in HβD-4. Interestingly, Song et al. [46] showed that HβD-2 is more expressed in cholesteatoma tissue than in HβD-3. Therefore, we might speculation that HβD-2 is most active in human beta defensins against bacterial infection, but more studies are needed to confirm this. Additionally, we found very strong positive correlations between HβD-2 and IL-1 (r = 0.822, p = 0.000), as well as strong positive correlations between HβD-2 and NF-κβ (r = 0.692, p = 0.001) and NF-κβ and IL-1 (r = 0.674, p = 0.002). These findings can be explained by the fact that IL-1 stimulates the production of HβD-2, and that this process is activated by NF-κβ [47,48]. Furthermore, Kanda et al. [49] presented positive correlations between IL-10 and HβD-2 and concluded that HβD-2 increases the production of IL-10 in T cells. We showed a similar correlation between HβD-2 and IL-10 (r = 0.663, p = 0.002) in the cholesteatoma perimatrix, where T cells predominate [50]. We did not find similar correlations between HβD-4 and IL-1, or between IL-10 and NF-κβ, as was the case with HβD-2. Therefore, we conclude that HβD-2 is a more potent antibacterial peptide in cholesteatoma than HβD-4.
Finally, our results demonstrated the statistically discernible overexpression of the Shh gene protein in the perimatrix (p = 0.000), in contrast to the control group. There are limited data available on Shh's role in cholesteatoma tissue. Our previous research, which compared children's cholesteatoma and deep meatal skin controls, showed similar findings to our current study [26]. However, it is known that Shh is responsible for developing the first pharyngeal arch, the pharyngeal endoderm, as well as regulating Fgf8 in the ectoderm from which the middle and external ear develop [51,52]. Furthermore, it has been proven that the loss of the Shh gene causes middle-and outer-ear pathologies [28]. We suggest that Shh is involved in the postnatal stimulation of endodermal/mesodermal tissue.
However, we understand the limitations of our study, namely the relatively small control group and the fact that tissue material were taken from cadavers. However, the ethical considerations mandated the use of this control group. Furthermore, standardized laboratory measurements (e.g., ELISA) could be useful in the evaluation of IHC-stained samples.

Conclusions
The elevation of Ki-67 and NF-κβ in cholesteatoma tissue suggests the induction of cellular proliferation in this tumor, with the significant involvement of NF-κβ in this process.
The decrease in degradation enzymes and the similarity in the expression of TIMPs might cause pathologic remodulation in cholesteatoma tissue. Intercorrelations between TIMP-2, NF-κβ and VEGF induce neo-angiogenesis in adult cholesteatoma.
The similarity in the expression of pro-and anti-inflammatory cytokines in cholesteatoma suggests the possible stagnation (dysregulation) of the local immune status, which was demonstrated by the strong stimulation of HβD-2 and the intercorrelation between IL-1, NF-κβ and HβD-2. The stimulation of the antibacterial activity of HβD-2 is also not excluded.
The overexpression of the Shh gene protein in cholesteatoma indicates the selective local stimulation of perimatix development, which probably connected to the influence of the endodermal gene protein.
Our study shed light on the complexity of cholesteatoma pathogenesis, as we presented how the aforementioned cell factors intercorrelate. For future studies, our aim is to compare pediatric-and adult-cholesteatoma material to determine the differences between the cell factors in these groups, as well as to increase the number of subjects in the groups.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10 .3390/medicina59020306/s1, Figure S1-a negative IHC control of cholesteatoma; Figure S2-pictures of each criterion of semi-quantitative method; Table S1-an additional statistical information. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The datasets used and/or analyzed during the current study are presented in the results section.