- Email: [email protected]
Gerbilline cholesteatoma development Part III. Increased proliferation index of basal keratinocytes of the tympanic membrane and external ear canal
Otolaryngology–Head and Neck Surgery (2006) 135, 116-123
ORIGINAL RESEARCH
Gerbilline cholesteatoma development Part III. Increased proliferation index of basal keratinocytes of the tympanic membrane and external ear canal Steven P. Tinling, PhD, and Richard A. Chole, MD, PhD, Davis, California; and St. Louis, Missouri OBJECTIVE: To quantify the rate of basal cell division for keratinizing epithelium (KE) of the tympanic membrane (TM) and external ear canal (EAC) in spontaneous and induced gerbilline cholesteatomas. STUDY DESIGN AND SETTING: Cholesteatomas (3 spontaneous and 5 by induction) were labeled with tritiated thymidine for autoradiography and a KE proliferation index (PI) was determined. The PI was defined as the average number of labeled cells/mm overall and per anatomic region. RESULTS: For all regions combined, the PI was 27.3 in ears with cholesteatoma and 4.1 in normal ears (P ⬍ 0.0001). Additionally, there were significant regional differences in the PI in both normal ears and ears with cholesteatoma. CONCLUSION: The KE of cholesteatomas in gerbils proliferates at approximately 7 times the rate measured in control ears. SIGNIFICANCE: Hyperproliferation of keratinocytes is a causative factor in the development and progression of spontaneous and experimental cholesteatomas in gerbils. © 2006 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved.
in which an increase in keratin accumulation, regardless of etiology, leads to increased proliferation of basal keratinocytes, inflammation of the TM and/or middle ear (ME), and more keratin accumulation. Regional differences in proliferation rate may also affect the development, appearance, and progression of the cholesteatoma. The gerbil, Meriones unguiculatus, forms spontaneous cholesteatomas with age8 and is a model for the retraction pocket and basal cell hyperplasia theories.7,9 The gerbil represents an excellent model for investigation of the relationship of KE hyperproliferation to cholesteatoma formation. We induced gerbilline cholesteatoma by EAC ligation or unilateral eustachian tube cautery separately or in combination and compared their KE to control tissue and spontaneous cholesteatoma by autoradiography. We determined and compared their overall and regional rates of basal cell proliferation.
MATERIALS AND METHODS
T
here are 4 current hypotheses for the generation of primary acquired cholesteatoma:1 1) metaplasia theory,2 2) immigration,3-5 3) basal cell hyperplasia theory,3,6-8 and 4) retraction pocket theory.3,6,7 The metaplasia theory depends on the conversion of squamous middle ear epithelium to keratinizing epithelium (KE) and does not initially involve the tympanic membrane (TM) or external auditory canal (EAC). The other theories propose a cyclical process
Fourteen adult male (60 to 90 days old) Mongolian gerbils (Meriones unguiculatus) from Tumblebrook Farms (now Charles River Labs, West Brookfield, MA) with normal external auditory canals and TMs were reared on a 12-hour light/dark cycle, maintained between 68°F and 76°F, and given Ralston Purina Co. rodent laboratory chow #5001 and distilled water ad libitum. All procedures were performed in accordance with the U.S. Public Health Service Policy on
From the Department of Otolaryngology–Head and Neck Surgery, University of California, Davis, School of Medicine (Dr Tinling); and the Department of Otolaryngology–Head and Neck Surgery, Washington School of Medicine, Washington University, St. Louis (Dr Chole). Supported in part by grants from the Deafness Research Foundation and NIH Grant R01-NS21079.
Presented at the Twenty-Seventh Midwinter Research Meeting of the Association for Research in Otolaryngology, Daytona Beach, FL, February 21-26, 2004. Reprint requests: Steven P. Tinling, PhD, Otolaryngology Research Laboratory, 1515 Newton Ct. Rm. 209, Davis, CA 95616. E-mail address: [email protected].
0194-5998/$32.00 © 2006 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved. doi:10.1016/j.otohns.2005.12.025
Tinling and Chole
Gerbilline cholesteatoma development III. Increased . . .
Humane Care and Use of Laboratory Animals, the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and the Animal Care and the Animal Welfare Act (7 U.S.C. et seq). The specific animal use protocol was approved by the Institutional Animal Care and Use Committee of the University of California, Davis.
Cholesteatoma Induction Gerbils were anesthetized (intraperitoneally) with ketamine HCl (Ketaset, 50 g/g) and xylazine hydrochloride (Rompun, 12.5 g/g) and injected with atropine sulfate (0.2 mL) (subcutaneously) to dry secretions and prevent fluid aspiration. Cholesteatoma was induced by 3 methods: 1) ear canal ligation (CL), 2) eustachian tube obstruction (ETO), and 3) both ear canal ligation and eustachian tube obstruction (CL ⫹ ETO). In method one, the right cartilaginous EAC of 6 animals was ligated with 4-0 silk through a retroauricular incision.10 In method two, the left eustachian tube of 6 animals was cauterized through the soft palate.9 In method three, 2 gerbils received both unilateral ear canal ligation and unilateral eustachian tube cautery (ETO ⫹ CL) on the same side— one on the left and one on the right. For each group, the untreated ear was intended to represent the normal control. However, 3 control ears (1 in each group) developed spontaneous cholesteatomas. One animal each, from the ETO and CL groups, was sacrificed at 3 days, 1 week, 2 weeks, 1 month, and 3 months. The first animal in the ETO ⫹ CL group was sacrificed at 2 months and the second at 3 months. Cholesteatomas were graded as Stage 1-5.8
Histopathologic Preparation and Autoradiography One hour prior to sacrifice, gerbils were injected intraperitoneally with 4 Ci/g body weight of tritiated thymidine (ICN Pharmaceuticals, Inc., Costa Mesa, CA). Animals were sacrificed and prepared for histology as previously described.11 Each bulla was oriented in either the transverse or horizontal plane and serially divided into 4 to 5 slices approximately 1 to 2 mm thick using a miniature lathe. The most representative cut through the midline of the EAC was sectioned and analyzed. For samples in the horizontal plane, this required 2 sections per block. Because of the difficulty in orienting osmicated whole bullae in plastic, a midline cut was not present in some of the transverse samples. For these samples, 2 representative cuts (1 more anterior and 1 more posterior) were selected. These samples were treated as repeated measures for statistical analysis. Six serial sections from each block were cut and placed one per slide at a standardized length from the end. Slides were coated with Type NTB2 autoradiographic emulsion (Eastman Kodak Co., Rochester, NY), placed in light-tight boxes, and stored at ⫺20°C. Slide sets were removed and the emulsion developed at 3- to 6-month intervals until sufficient grain density for analysis was observed. This resulted in sufficient grain density for analysis (total incubation time 15 months) in 3 sets of slides for left ears and 2
117
sets for right ears (1 set lost during development). Marrow cavities were present in all samples and were used to qualitatively assess positive uptake of label. There was no difference noted in the degree of labeling in any of the samples. The slides were then stained with toluidine blue and basic fuchsin in a 40°C water bath for 1½ hours. Slides were rinsed with distilled water to clear emulsion of excess stain, dried, and coverslipped. Slides were coded in a random double-blind manner and assessed by light microscopy. Each slide was evaluated qualitatively for: 1) normal appearance, 2) effusion, and 3) cholesteatoma (stage 1-4). The following regions, if present, were identified on each slide: 1) superior lateral ear canal, 2) inferior lateral ear canal, 3) superior medial ear canal, 4) pars flaccida (PF), 5) pars tensa (PT), 6) the PT over the malleus, 7) anterior, and 8) posterior. Regions were digitally captured at 10⫻ as calibrated images and stored using a PixelFly digital camera (The Cooke Corp., Auburn Hills, MI). The length of the basal KE layer and the number of labeled basal keratinocytes for each region was measured using ImagePro Plus (Media Cybernetics, Silver Spring, MD) image analysis software. Data were tabulated and the proliferation index (PI), defined as the number of labeled keratinocytes per mm per region, was calculated.
Statistical Analysis Statistical analysis was provided by the Statistical Laboratory, University of California, Davis. Quantitative data was analyzed by the General Linear Model for analysis of variance with log transformation of all ratios. Comparisons included: 1) overall effect, 2) effect by category (normal or cholesteatoma) at sacrifice, 3) effect by category vs region, and 4) effect by region vs region. All comparisons met requirements for normality and sufficient N for significance.
RESULTS Cholesteatoma formation occurred spontaneously or by induction in 8 of 28 ears. Within the ETO group, a cholesteatoma occurred in only 1 of 6 ears by induction and spontaneously in its contralateral control (Table 1). In the CL group cholesteatoma occurred in 3 of 6 ears by induction and spontaneously in 1 contralateral control. In the ETO ⫹ CL group cholesteatoma occurred by induction in 1 ear and spontaneously in the control ear of the other animal. All cholesteatomas were either stage 2 (N ⫽ 3) or stage 3 (N ⫽ 5) (Table 1). Regardless of method or time of sacrifice, none of the cholesteatomas was histologically distinguishable from any other with blind, randomized analysis. Additionally, only in the CL group was N greater than 1, so a comparison of groups based on PI means was not possible. Similarly, none of the induction failures was histologically distinguishable in blinded analysis from the normal controls by method or time of sacrifice. A comparison by two-way analysis of
118
Otolaryngology–Head and Neck Surgery, Vol 135, No 1, July 2006
Table 1 Cholesteatoma induction outcome and mean number of labeled cells per millimeter Result
Mean
Condition-orientation right ear
Result
Mean
ETO*-T† ETO-T ETO-HS ETO-T ETO-T ETO-HM
Normal Normal Normal Normal Normal Stage 3
2.87 8.67 4.57 2.77 4.08 37.06
Control-T Control-T Control-T Control-HM Control-T Control-T
Normal Normal Normal Normal Normal Stage 3
2.43 2.23 2.44 4.68 3.76 24.33
3 days 1 week 2 weeks
Control-T Control-T Control-HI
Normal Normal Normal
3.96 3.63 4.87
1 month 2 months
Control-T Control-HS Control-HI Control-HS Control-HI
Normal Normal Normal Stage 2 Normal
5.66 7.02 2.16 35.03 4.00
CL-HM CL-T CL-HI CL-HM CL-T CL-HS CL-HI CL-T
Normal Stage 2 Normal Normal Stage 2 Normal Normal Stage 3
2.29 53.25 1.28 3.15 17.52 12.36 4.02 22.79
Control-HM ETO ⫹ CL-T
Normal Normal
2.86 3.24
ETO ⫹ CL-T Control-T
Stage 3 Stage 3
16.25 22.15
Time 3 1 2 1 2 3
Condition*-orientation† left ear
days week weeks month months months
3 months 2 months 3 months
*ETO, eustachian tube obstruction by cautery of the palatal opening; CL, external auditory canal ligation; ETO ⫹ CL, eustachian tube obstruction and external auditory canal ligation to the same ear. †T, transverse (midline); HI, horizontal inferior; HM, horizontal medial; HS, horizontal superior.
variance of the PI, contrasting time of sacrifice vs treatment (ETO, ETO control, CL, and CL control), showed no statistical difference by group or time (Pgroup ⬎ 0.28 and Ptime ⬎ 0.63). Therefore, time of sacrifice and method of induction were omitted as statistical variables and PI data for cholesteatomas and normal specimens were respectively grouped together for subsequent comparisons.
Autoradiography Labeled nuclei of basal keratinocytes were readily distinguishable and background was minimal (Figs 1 and 2). There was an overall effect for the general linear model for analysis of variance (P ⬍ 0.0001) and there was no effect for left vs right (P ⬎ 0.20) or slide set (P ⬎ 0.82). For all sample averages combined, the mean and standard error of the PI for ears with cholesteatoma with matrix (C ⫹ matrix) was 27.27 ⫾ 4.10 and the PI for normal ears was 4.09 ⫾ 0.47. In the stage 2 spontaneous cholesteatoma (Table 1), the matrix did not extend into the inferior portion of the ear canal (Fig 2). Consequently, horizontal sections from the inferior portion of the canal from this animal, which did not include matrix, were ranked normal during blinded, random quantification. For statistical analysis, sections from this animal were subsequently coded C ⫹ matrix and cholesteatoma minus matrix (C ⫺ matrix). For all animals combined, the PI for the C ⫹ matrix significantly differed from the PI for the C ⫺ matrix animal (4.00 ⫾ 0.13) and normal animals (P ⬍ 0.0001 for both), but the PI for the C ⫺ matrix ear and normal individuals did not differ significantly from one another (P ⬎ 0.84). There was a significant effect overall of the PI for comparison of condition (cholesteatoma or normal) vs region
(superior lateral, inferior lateral, etc.) with P ⬍ 0.0001 (Fig 3). In all region-by-region comparisons, the PI for C ⫹ matrix was significantly different from normal with P ⬍ 0.0001. However, the PI for C ⫺ matrix was not significantly different in comparison with the corresponding normal region (P ⬎ 0.96). For specific region-by-region comparisons within normal samples (Fig 3): 1) the PI for the PT over the malleus (2.7 ⫾ 0.62) was not significantly different (P ⬎ 0.35) from the inferior PT (1.3 ⫾ 0.15); 2) the PI for the PF (5.7 ⫾ 0.76) was significantly greater (P ⬍ 0.001 for both) than the PI for either PT region. However, the PF was not significantly different (0.8 ⬎ P ⬎ 0.1) from superior medial (5.6 ⫾ 1.90), superior lateral (6.1 ⫾ 1.36), or inferior lateral (4.2 ⫾ 0.71) external ear canal; 3) there was no significant difference in the PI between superior medial and superior lateral EAC (P ⬎ 0.9); however, the PI for superior lateral was significantly increased compared to inferior lateral (P ⬍ 0.04); and 4) the PI for the posterior region (7.4 ⫾ 1.28) was significantly larger (P ⬍ 0.004) compared with the anterior region (4.3 ⫾ 0.74). For specific region-by-region comparisons within the cholesteatoma samples (Fig 3): 1) the PI for the inferior PT (14.5 ⫾ 2.62) was significantly smaller (P ⬍ 0.003) than all other regions; 2) the PI for the PT over the malleus (35.1 ⫾ 5.11) was significantly increased (P ⬍ 0.0004) over the inferior portion of the PT; 3) the PI for the PF (43.4 ⫾ 7.80) was not significantly different from the PT over the malleus (P ⬎ 0.096) or the superior medial (34.2 ⫾ 6.04) region (P ⬎ 0.24), but was significantly increased (P ⬍ 0.0034 for both) in comparison to superior (26.7 ⫾ 8.20) and inferior
Tinling and Chole
Gerbilline cholesteatoma development III. Increased . . .
119
The KE in the normal PF was 4 to 6 cells thick with occasional labeled cells in the basal layer (Fig 4) and the LP contained numerous small capillaries and loose connective tissue. In contrast, the PF in ears with cholesteatoma had a significantly thickened layer of KE with numerous labeled cells and a thickened hyperemic LP. An identical change occurred in the PT when cholesteatoma matrix was present (Fig 4). In normal animals, the rate of cell division in the annular region was very low with 0 to 2 dividing cells per section (Fig 5A). In cholesteatoma samples in which matrix extended to the annular rim (Fig 5B), there was increased cell division, formation, and/or enlargement of ceruminous glands, and osteoclast activation along the bony wall similar to that seen in the upper portions of the canal (Fig 1B). Additionally, the periosteum on the middle ear side was hyperemic and inflamed. In cholesteatoma specimens, the bone of the lateral and medial wall of the EAC showed significant remodeling by osteoclasts with osteoblastic activation and presumed replication of precursors on the adjacent surface (Fig 6). A very interesting finding in one of the cholesteatoma specimens was the appearance of a perforation in the KE and subjacent dense fibrous layer of the KE next to the inferior annulus (Fig 7). The keratin mass has infiltrated the
Figure 1 (A) Normal EAC. The labeled cells present are not sufficiently dense to be seen at this low magnification. (B) EAC with cholesteatoma. The KE (thin arrow) and LP (thick arrow) are significantly thickened. Thin arrow: basal KE cells covered with silver grains. A, annulus; il, inferior lateral; m, malleus; pt, pars tensa; sl, superior lateral.
lateral (25.5 ⫾ 6.50) epithelium; and 4) there was no significant difference in the PI between superior medial, superior lateral, inferior lateral, anterior (46.9 ⫾ 12.60), or posterior (39.9 ⫾ 6.50) regions (all P ⬎ 0.051 to 0.97).
Histopathology Cholesteatoma specimens demonstrated qualitatively thickened KE and lamina propria (LP) in the TM and also the KE and subepithelium within the EAC when compared with normal tissue (Figs 1 and 5). A distinct line of radioactive labeled cells is present along the basal layer of the KE in ears with cholesteatoma (Fig 2). All cholesteatomas were stage 2 or 3, and all but one stage 2 completely filled the EAC. The one cholesteatoma that formed in the ETO group did not undergo retraction of the PF and could not be distinguished qualitatively in blinded analysis from the spontaneous or CL-induced cholesteatomas. In the spontaneous stage 2 cholesteatoma, which had not completely filled the inferior portion of the EAC, the KE in this region appeared normal (Fig 2) with only an occasionally labeled cell in the TM and the inferior lateral wall of the EAC.
Figure 2 Inferior lateral (il) portion of the EAC in which the cholesteatoma did not yet extend beyond the superior canal. Compare to Figs 1B and 4D. pt, pars tensa.
120
Otolaryngology–Head and Neck Surgery, Vol 135, No 1, July 2006
Figure 3 Mean proliferation index (PI) for basal keratinocytes in the ear by region. The PI was significantly increased for all regions (black bars) in cholesteatoma. eac, external ear canal; c, cochlea; m, malleus; me, middle ear; S, superior; L, lateral.
LP and has almost broken through the squamous epithelium lining of the middle ear surface.
DISCUSSION Several studies of human cholesteatoma tissue support the hypothesis that an increasing PI is associated with cholesteatoma development. Mallet et al12 demonstrated a significant association between hyperproliferation, middle ear inflammation, and a higher risk of recurrence in human cholesteatomas using a marker for hyperproliferation (monoclonal antibody MIB1). They demonstrated a significant correlation for a high PI with aggressive cholesteatoma in children. Huisman et al13 compared levels of the cellular markers Ki-67, p53, p21, and active caspase 3 for cell proliferation, cell-cycle arrest, and apoptosis in human cholesteatoma epithelium to paired control retro-auricular skin. The nuclear antigen, Ki-67, is associated with hyperproliferation and the proteins, p53 and p21, are associated with apoptosis and cell-cycle arrest while caspase 3 is an indicator of cell death. They demonstrated a significant increase for each of these markers in cholesteatoma except caspase 3. An increase in PI was supported by increased localization of Ki-67 in the basal KE of cholesteatomas compared to control skin. The co-localization in the basal KE of cholesteatomas of increased levels of p53 without an increase in caspase 3 supports the hypothesis that alteration of the normal cell
cycle without cell death contributes to the increase in the PI and resulting cholesteatoma development. Additionally, the increased expression of p21 in the suprabasal layers of cholesteatoma specimens suggests that the increased thickness of KE in cholesteatomas results from an increased rate of proliferation of basal KE combined with inhibition of keratinocyte transformation. An increase in the PI of KE in cholesteatoma has also been linked to upregulation of keratinocyte growth factor (KGF) and its receptor (KGFR). YamamotoFukuda et al14 demonstrated a positive correlation between increased expression of both KGF and KGFR with an increased expression of Ki-67 in human cholesteatomas compared to normal skin. In addition, they showed a significant correlation between KGF and KGFR expression and cholesteatoma recurrence. In this study, regional differences in the PI for the TM of normal animals were present prior to cholesteatoma formation. The PI in the PF was higher than the PI in the PT over the malleus, which was itself higher than PT proper. However, it is unlikely that an increase in the PI of the PF alone results in the majority of the keratin production in the initial phase of cholesteatoma formation as the ratio of the area of the PT to the PF in the gerbil is 7:1 (unpublished data). It is more likely that there is alteration in the normal clearance mechanism and rate of KE proliferation for both the TM and the EAC in those gerbils which form cholesteatomas. The increasing thickness of the KE of the PF, PT, and EAC leads to continued buildup of keratin, thus perpetuating the cycle (Fig 8). Sequestered bacterial flora and the resulting immu-
Tinling and Chole
Gerbilline cholesteatoma development III. Increased . . .
Figure 4 Normal and cholesteatoma TM. (A) Normal PF. Thin arrow: dividing cells covered by silver grains. Thick arrow: LP. (B) Cholesteatoma PF. ke, desquamating keratin. (C) Normal PT. Arrow: thin layers of desquamating keratin. (D) Cholesteatoma PT. Thick arrow: LP. Dense accumulation of keratin (k) in the EAC. The large separation of the keratin from the surface of the pars tensa (thin arrow) may represent a fixation artifact.
nologic response maintain the inflammatory component of this process. Several studies both human15 and gerbilline16-19 have investigated the theory of alterations in cytokeratin expression in the TM of human cholesteatoma as a possible mechanism for cholesteatoma formation. In a study of human PT and cholesteatoma compared to normal skin, Kakoi et al15 determined that the cytokeratin expression pattern suggested a hyperproliferative matrix. Studies by Kim et al18,19 noted significant changes in the expression patterns of cytokeratins in gerbilline cholesteatoma, which were variable in location and depended on the type of induction method. The hyperproliferation marker (CK13/16) was more pronounced in cholesteatomas induced by eustachian tube cautery than by ligation or propylene glycol application. They concluded that both a hyperproliferative process and migratory process coexist and contribute to the pathogenesis of gerbilline cholesteatoma. Kim and Chung16 concluded that
121
Figure 5 Annulus in normal and cholesteatoma ears. (A) Normal. (B) Cholesteatoma. eac, external ear canal; g, cerumen gland; me, middle ear. Scale applies to both.
cholesteatoma in ligated gerbils originates in EAC epithelium. However, they recognized that induction by ligation would force EAC epithelium into the middle ear. There-
Figure 6 Enlargement of lateral bulla wall (broad arrow in Fig 1B). Osteoclasts (thin arrows) in the LP (lp) eroding the lateral wall of the ear canal. There is significant osteoblastic activity and precursor replication and in the external periosteum (broad arrow). b, bone.
122
Otolaryngology–Head and Neck Surgery, Vol 135, No 1, July 2006
Figure 7 Perforation of the TM near the annulus. The KE and dense fibrous layer next to the annulus (a) are perforated (curved arrow) and there is infiltration of the LP (lp) by the keratin mass (m).
fore, their analysis concentrated on the lateral portion of the EAC and did not investigate differences between the EAC and the TM. In an immunohistochemical study of BrdU labeling of basal keratinocytes in ligated gerbil ears, Park et al20 reported an overall and regional increase similar to ours. The majority of labeling in the TM of normal animals occurred in the basal layer of the PF and PT over the malleus with regional differences in the degree of labeling in the EAC. They noted that labeling was higher in the superior regions of the canal with no anterior-to-posterior difference. For cholesteatoma tissue, Park et al20 did not report regional differences, but compared cholesteatomas grossly by stage with all stages significantly greater than normal. They concluded that increasing keratin in the EAC may also cause an increased mitotic activity in the surrounding epithelium. However, in our study, we measured regional PIs and they were significantly increased over normal in all regions. The effect of accumulated keratin on hyperproliferation, hypothesized by Park et al,20 was evident in our study. In
Figure 8
the stage 2 cholesteatoma, which had not yet extended inferiorly, the PI in the inferior canal was not significantly different from normal, nor were any sections without accumulated keratin recognized as histologically abnormal during blinded analysis, whereas labeling was significantly increased in all other samples in which cholesteatoma matrix was present in the inferior EAC. Tanaka et al21 examined normal and cholesteatoma tissue from humans for the presence of proliferating cell nuclear antigen (PCNA) and transforming growth factor-␣ (TGF-␣). They detected increased PCNA above the suprabasal layer with differences ascribed to subepidermal cell inflammation. Staining for TGF-␣ occurred principally in the granular and prickle cell layers of normal skin and in all layers within cholesteatomas. However, it is unlikely that cell division was occurring above the basal layer in this study. Diffuse staining in the upper regions was most likely the result of nonresorbed PCNA in the differentiating epithelial cells. In this study, we demonstrate that the rate of cell division in basal cell keratinocytes of the gerbil TM and EAC within developing cholesteatoma is significantly increased by approximately 7 times that of normal TM and EAC. There was no apparent difference in division rate among the different methods of cholesteatoma induction and they appeared histologically identical. There were distinct regional differences in the PI for basal keratinocytes both within and between normal and cholesteatoma tissues. The regions with the highest PI were the medial and lateral regions of the superior canal. This suggests that, during cholesteatoma formation in gerbils, this portion of the canal and TM is the first to respond to an induction stimulus. Our results, in conjunction with those of Park et al20 and Tanaka et al,21 provide support for the following hypothesis of a cholesteatoma development cycle in gerbils (Fig 8): 1) middle ear inflammation occurs in conjunction with (or may cause) a disruption of the clearance mechanism for desquamating keratin, resulting in accu-
Cholesteatoma development cycle.
Tinling and Chole
Gerbilline cholesteatoma development III. Increased . . .
mulation of keratin on the; 2) a disruption in normal cell-cycle regulation results in an increase in the PI for basal keratinocytes of the PF and PT; 3) increased PI without a concomitant increase in keratinocyte maturation leads to an increased thickness of the keratinizing epithelium (KE); 4) the increasing accumulation of keratin results in blockage of the opening to the external ear canal. The resulting superior-to-inferiorly-directed filling of the canal involves unaffected regions of the PT and EAC and continues the cycle. Continued production of trapped keratin results in distension of the TM and middle ear morbidity in gerbils. Statistical analysis was provided by Dr Mitchell Watnik of the Statistical Laboratory, University of California, Davis.
REFERENCES 1. Chole RA, Tinling SP. Epidermoid cholesteatoma. In: Friedmann I, Arnold W, editors. Pathology of the ear. New York: Churchill Livingstone; 1993. p. 752. 2. Sadé J, Babiacki A, Pinkus G. The metaplastic and congenital origin of cholesteatoma. Acta Otolaryngol 1983;96:119 –29. 3. Friedmann I. The comparative pathology of otitis media— experimental and human. J Laryngol Otol 1955;69:588 – 601. 4. Graham MD, Nelson RA, Kemink JL. Tympanic membrane cholesteatoma (keratoma). Am J Otol 1984;5:175–7. 5. Goycoolea MV, Heub MM, Muchow D, et al. The theory of the trigger, the bridge and the transmigration in the pathogenesis of acquired cholesteatoma. Acta Otolaryngol 1999;119:244 – 8. 6. Lim DJ, Saunders WH. Acquired cholesteatoma: Light and electron microscopic observations. Ann Otol Rhinol Laryngol 1972;81:2–12. 7. Chole RA, Tinling SP. Basal lamina breaks in the histogenesis of cholesteatoma. Laryngoscope 1985;95:270 –5.
123
8. Henry KR, Chole RA, McGinn MD. Age-related increase of spontaneous aural cholesteatoma in the Mongolian gerbil. Arch Otolaryngol 1983;109:19 –21. 9. Wolfman DE, Chole RA. Experimental retraction pocket cholesteatomas. Ann Otol Rhinol Laryngol 1986;95:639 – 44. 10. McGinn MD, Henry KR, Chole RA. Cholesteatoma: Experimental induction in the mongolian gerbil. Acta Otolaryngol 1982;93:61–7. 11. Aminpour S, Tinling SP, Brodie HA. Role of tumor necrosis factor-␣ sensorineural hearing loss after bacterial meningitis. Otol Neurotol 2005;26:602–9. 12. Mallet Y, Nouwen J, LecomteHoucke M, et al. Aggressiveness and quantification of epithelial proliferation of middle ear cholesteatoma by MIB1. Laryngoscope 2003;113:328 –31. 13. Huisman MA, Heer ED, Grote JJ. Cholesteatoma epithelium is characterized by increased expression of Ki-67, p53 and p21, with minimal apoptosis. Acta Otolaryngol 2003;123:377– 82. 14. Yamamoto-Fukuda T, Aoki D, Hishikawa Y, et al. Possible involvement of keratinocyte growth factor and its receptor in enhanced epithelial-cell proliferation and acquired recurrence of middle-ear cholesteatoma. Lab Invest 2003;83:123–36. 15. Kakoi H, Tamagawa Y, Kitamura K, et al. Cytokeratin expression patterns by one- and two-dimensional electrophoresis in pars flaccida cholesteatoma and pars tensa cholesteatoma. Acta Otolaryngol 1995;115:804 –10. 16. Kim CS, Chung JW. Morphologic and biologic changes of experimentally induced cholesteatoma in Mongolian gerbils with anticytokeratin and lectin study. Am J Otol 1999;20:13– 8. 17. Kim H-J, Tinling SP, Chole RA. Expression patterns of cytokeratins in retraction pocket cholesteatomas. Laryngoscope 2001;111:1032– 6. 18. Kim H-J, Tinling SP, Chole RA. Expression patterns of cytokeratins in cholesteatomas: Evidence of increased migration and proliferation. J Korean Med Sci 2002;17:381– 8. 19. Kim H-J, Tinling SP, Chole RA. Increased proliferation and migration of epithelium in advancing experimental cholesteatomas. Otol Neurotol 2002;23:840 – 4. 20. Park K, Chun Y-M, Park H-J, et al. Immunohistochemical study of cell proliferation using BrdU labeling on tympanic membrane, external auditory canal and induced cholesteatoma in Mongolian gerbils. Acta Otolaryngol 1999;119:874 –9. 21. Tanaka Y, Shiwa M, Kojima H, et al. A study on epidermal proliferation ability in cholesteatoma. Laryngoscope 1998;108:537– 42.
ORIGINAL RESEARCH
Gerbilline cholesteatoma development Part III. Increased proliferation index of basal keratinocytes of the tympanic membrane and external ear canal Steven P. Tinling, PhD, and Richard A. Chole, MD, PhD, Davis, California; and St. Louis, Missouri OBJECTIVE: To quantify the rate of basal cell division for keratinizing epithelium (KE) of the tympanic membrane (TM) and external ear canal (EAC) in spontaneous and induced gerbilline cholesteatomas. STUDY DESIGN AND SETTING: Cholesteatomas (3 spontaneous and 5 by induction) were labeled with tritiated thymidine for autoradiography and a KE proliferation index (PI) was determined. The PI was defined as the average number of labeled cells/mm overall and per anatomic region. RESULTS: For all regions combined, the PI was 27.3 in ears with cholesteatoma and 4.1 in normal ears (P ⬍ 0.0001). Additionally, there were significant regional differences in the PI in both normal ears and ears with cholesteatoma. CONCLUSION: The KE of cholesteatomas in gerbils proliferates at approximately 7 times the rate measured in control ears. SIGNIFICANCE: Hyperproliferation of keratinocytes is a causative factor in the development and progression of spontaneous and experimental cholesteatomas in gerbils. © 2006 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved.
in which an increase in keratin accumulation, regardless of etiology, leads to increased proliferation of basal keratinocytes, inflammation of the TM and/or middle ear (ME), and more keratin accumulation. Regional differences in proliferation rate may also affect the development, appearance, and progression of the cholesteatoma. The gerbil, Meriones unguiculatus, forms spontaneous cholesteatomas with age8 and is a model for the retraction pocket and basal cell hyperplasia theories.7,9 The gerbil represents an excellent model for investigation of the relationship of KE hyperproliferation to cholesteatoma formation. We induced gerbilline cholesteatoma by EAC ligation or unilateral eustachian tube cautery separately or in combination and compared their KE to control tissue and spontaneous cholesteatoma by autoradiography. We determined and compared their overall and regional rates of basal cell proliferation.
MATERIALS AND METHODS
T
here are 4 current hypotheses for the generation of primary acquired cholesteatoma:1 1) metaplasia theory,2 2) immigration,3-5 3) basal cell hyperplasia theory,3,6-8 and 4) retraction pocket theory.3,6,7 The metaplasia theory depends on the conversion of squamous middle ear epithelium to keratinizing epithelium (KE) and does not initially involve the tympanic membrane (TM) or external auditory canal (EAC). The other theories propose a cyclical process
Fourteen adult male (60 to 90 days old) Mongolian gerbils (Meriones unguiculatus) from Tumblebrook Farms (now Charles River Labs, West Brookfield, MA) with normal external auditory canals and TMs were reared on a 12-hour light/dark cycle, maintained between 68°F and 76°F, and given Ralston Purina Co. rodent laboratory chow #5001 and distilled water ad libitum. All procedures were performed in accordance with the U.S. Public Health Service Policy on
From the Department of Otolaryngology–Head and Neck Surgery, University of California, Davis, School of Medicine (Dr Tinling); and the Department of Otolaryngology–Head and Neck Surgery, Washington School of Medicine, Washington University, St. Louis (Dr Chole). Supported in part by grants from the Deafness Research Foundation and NIH Grant R01-NS21079.
Presented at the Twenty-Seventh Midwinter Research Meeting of the Association for Research in Otolaryngology, Daytona Beach, FL, February 21-26, 2004. Reprint requests: Steven P. Tinling, PhD, Otolaryngology Research Laboratory, 1515 Newton Ct. Rm. 209, Davis, CA 95616. E-mail address: [email protected].
0194-5998/$32.00 © 2006 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved. doi:10.1016/j.otohns.2005.12.025
Tinling and Chole
Gerbilline cholesteatoma development III. Increased . . .
Humane Care and Use of Laboratory Animals, the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and the Animal Care and the Animal Welfare Act (7 U.S.C. et seq). The specific animal use protocol was approved by the Institutional Animal Care and Use Committee of the University of California, Davis.
Cholesteatoma Induction Gerbils were anesthetized (intraperitoneally) with ketamine HCl (Ketaset, 50 g/g) and xylazine hydrochloride (Rompun, 12.5 g/g) and injected with atropine sulfate (0.2 mL) (subcutaneously) to dry secretions and prevent fluid aspiration. Cholesteatoma was induced by 3 methods: 1) ear canal ligation (CL), 2) eustachian tube obstruction (ETO), and 3) both ear canal ligation and eustachian tube obstruction (CL ⫹ ETO). In method one, the right cartilaginous EAC of 6 animals was ligated with 4-0 silk through a retroauricular incision.10 In method two, the left eustachian tube of 6 animals was cauterized through the soft palate.9 In method three, 2 gerbils received both unilateral ear canal ligation and unilateral eustachian tube cautery (ETO ⫹ CL) on the same side— one on the left and one on the right. For each group, the untreated ear was intended to represent the normal control. However, 3 control ears (1 in each group) developed spontaneous cholesteatomas. One animal each, from the ETO and CL groups, was sacrificed at 3 days, 1 week, 2 weeks, 1 month, and 3 months. The first animal in the ETO ⫹ CL group was sacrificed at 2 months and the second at 3 months. Cholesteatomas were graded as Stage 1-5.8
Histopathologic Preparation and Autoradiography One hour prior to sacrifice, gerbils were injected intraperitoneally with 4 Ci/g body weight of tritiated thymidine (ICN Pharmaceuticals, Inc., Costa Mesa, CA). Animals were sacrificed and prepared for histology as previously described.11 Each bulla was oriented in either the transverse or horizontal plane and serially divided into 4 to 5 slices approximately 1 to 2 mm thick using a miniature lathe. The most representative cut through the midline of the EAC was sectioned and analyzed. For samples in the horizontal plane, this required 2 sections per block. Because of the difficulty in orienting osmicated whole bullae in plastic, a midline cut was not present in some of the transverse samples. For these samples, 2 representative cuts (1 more anterior and 1 more posterior) were selected. These samples were treated as repeated measures for statistical analysis. Six serial sections from each block were cut and placed one per slide at a standardized length from the end. Slides were coated with Type NTB2 autoradiographic emulsion (Eastman Kodak Co., Rochester, NY), placed in light-tight boxes, and stored at ⫺20°C. Slide sets were removed and the emulsion developed at 3- to 6-month intervals until sufficient grain density for analysis was observed. This resulted in sufficient grain density for analysis (total incubation time 15 months) in 3 sets of slides for left ears and 2
117
sets for right ears (1 set lost during development). Marrow cavities were present in all samples and were used to qualitatively assess positive uptake of label. There was no difference noted in the degree of labeling in any of the samples. The slides were then stained with toluidine blue and basic fuchsin in a 40°C water bath for 1½ hours. Slides were rinsed with distilled water to clear emulsion of excess stain, dried, and coverslipped. Slides were coded in a random double-blind manner and assessed by light microscopy. Each slide was evaluated qualitatively for: 1) normal appearance, 2) effusion, and 3) cholesteatoma (stage 1-4). The following regions, if present, were identified on each slide: 1) superior lateral ear canal, 2) inferior lateral ear canal, 3) superior medial ear canal, 4) pars flaccida (PF), 5) pars tensa (PT), 6) the PT over the malleus, 7) anterior, and 8) posterior. Regions were digitally captured at 10⫻ as calibrated images and stored using a PixelFly digital camera (The Cooke Corp., Auburn Hills, MI). The length of the basal KE layer and the number of labeled basal keratinocytes for each region was measured using ImagePro Plus (Media Cybernetics, Silver Spring, MD) image analysis software. Data were tabulated and the proliferation index (PI), defined as the number of labeled keratinocytes per mm per region, was calculated.
Statistical Analysis Statistical analysis was provided by the Statistical Laboratory, University of California, Davis. Quantitative data was analyzed by the General Linear Model for analysis of variance with log transformation of all ratios. Comparisons included: 1) overall effect, 2) effect by category (normal or cholesteatoma) at sacrifice, 3) effect by category vs region, and 4) effect by region vs region. All comparisons met requirements for normality and sufficient N for significance.
RESULTS Cholesteatoma formation occurred spontaneously or by induction in 8 of 28 ears. Within the ETO group, a cholesteatoma occurred in only 1 of 6 ears by induction and spontaneously in its contralateral control (Table 1). In the CL group cholesteatoma occurred in 3 of 6 ears by induction and spontaneously in 1 contralateral control. In the ETO ⫹ CL group cholesteatoma occurred by induction in 1 ear and spontaneously in the control ear of the other animal. All cholesteatomas were either stage 2 (N ⫽ 3) or stage 3 (N ⫽ 5) (Table 1). Regardless of method or time of sacrifice, none of the cholesteatomas was histologically distinguishable from any other with blind, randomized analysis. Additionally, only in the CL group was N greater than 1, so a comparison of groups based on PI means was not possible. Similarly, none of the induction failures was histologically distinguishable in blinded analysis from the normal controls by method or time of sacrifice. A comparison by two-way analysis of
118
Otolaryngology–Head and Neck Surgery, Vol 135, No 1, July 2006
Table 1 Cholesteatoma induction outcome and mean number of labeled cells per millimeter Result
Mean
Condition-orientation right ear
Result
Mean
ETO*-T† ETO-T ETO-HS ETO-T ETO-T ETO-HM
Normal Normal Normal Normal Normal Stage 3
2.87 8.67 4.57 2.77 4.08 37.06
Control-T Control-T Control-T Control-HM Control-T Control-T
Normal Normal Normal Normal Normal Stage 3
2.43 2.23 2.44 4.68 3.76 24.33
3 days 1 week 2 weeks
Control-T Control-T Control-HI
Normal Normal Normal
3.96 3.63 4.87
1 month 2 months
Control-T Control-HS Control-HI Control-HS Control-HI
Normal Normal Normal Stage 2 Normal
5.66 7.02 2.16 35.03 4.00
CL-HM CL-T CL-HI CL-HM CL-T CL-HS CL-HI CL-T
Normal Stage 2 Normal Normal Stage 2 Normal Normal Stage 3
2.29 53.25 1.28 3.15 17.52 12.36 4.02 22.79
Control-HM ETO ⫹ CL-T
Normal Normal
2.86 3.24
ETO ⫹ CL-T Control-T
Stage 3 Stage 3
16.25 22.15
Time 3 1 2 1 2 3
Condition*-orientation† left ear
days week weeks month months months
3 months 2 months 3 months
*ETO, eustachian tube obstruction by cautery of the palatal opening; CL, external auditory canal ligation; ETO ⫹ CL, eustachian tube obstruction and external auditory canal ligation to the same ear. †T, transverse (midline); HI, horizontal inferior; HM, horizontal medial; HS, horizontal superior.
variance of the PI, contrasting time of sacrifice vs treatment (ETO, ETO control, CL, and CL control), showed no statistical difference by group or time (Pgroup ⬎ 0.28 and Ptime ⬎ 0.63). Therefore, time of sacrifice and method of induction were omitted as statistical variables and PI data for cholesteatomas and normal specimens were respectively grouped together for subsequent comparisons.
Autoradiography Labeled nuclei of basal keratinocytes were readily distinguishable and background was minimal (Figs 1 and 2). There was an overall effect for the general linear model for analysis of variance (P ⬍ 0.0001) and there was no effect for left vs right (P ⬎ 0.20) or slide set (P ⬎ 0.82). For all sample averages combined, the mean and standard error of the PI for ears with cholesteatoma with matrix (C ⫹ matrix) was 27.27 ⫾ 4.10 and the PI for normal ears was 4.09 ⫾ 0.47. In the stage 2 spontaneous cholesteatoma (Table 1), the matrix did not extend into the inferior portion of the ear canal (Fig 2). Consequently, horizontal sections from the inferior portion of the canal from this animal, which did not include matrix, were ranked normal during blinded, random quantification. For statistical analysis, sections from this animal were subsequently coded C ⫹ matrix and cholesteatoma minus matrix (C ⫺ matrix). For all animals combined, the PI for the C ⫹ matrix significantly differed from the PI for the C ⫺ matrix animal (4.00 ⫾ 0.13) and normal animals (P ⬍ 0.0001 for both), but the PI for the C ⫺ matrix ear and normal individuals did not differ significantly from one another (P ⬎ 0.84). There was a significant effect overall of the PI for comparison of condition (cholesteatoma or normal) vs region
(superior lateral, inferior lateral, etc.) with P ⬍ 0.0001 (Fig 3). In all region-by-region comparisons, the PI for C ⫹ matrix was significantly different from normal with P ⬍ 0.0001. However, the PI for C ⫺ matrix was not significantly different in comparison with the corresponding normal region (P ⬎ 0.96). For specific region-by-region comparisons within normal samples (Fig 3): 1) the PI for the PT over the malleus (2.7 ⫾ 0.62) was not significantly different (P ⬎ 0.35) from the inferior PT (1.3 ⫾ 0.15); 2) the PI for the PF (5.7 ⫾ 0.76) was significantly greater (P ⬍ 0.001 for both) than the PI for either PT region. However, the PF was not significantly different (0.8 ⬎ P ⬎ 0.1) from superior medial (5.6 ⫾ 1.90), superior lateral (6.1 ⫾ 1.36), or inferior lateral (4.2 ⫾ 0.71) external ear canal; 3) there was no significant difference in the PI between superior medial and superior lateral EAC (P ⬎ 0.9); however, the PI for superior lateral was significantly increased compared to inferior lateral (P ⬍ 0.04); and 4) the PI for the posterior region (7.4 ⫾ 1.28) was significantly larger (P ⬍ 0.004) compared with the anterior region (4.3 ⫾ 0.74). For specific region-by-region comparisons within the cholesteatoma samples (Fig 3): 1) the PI for the inferior PT (14.5 ⫾ 2.62) was significantly smaller (P ⬍ 0.003) than all other regions; 2) the PI for the PT over the malleus (35.1 ⫾ 5.11) was significantly increased (P ⬍ 0.0004) over the inferior portion of the PT; 3) the PI for the PF (43.4 ⫾ 7.80) was not significantly different from the PT over the malleus (P ⬎ 0.096) or the superior medial (34.2 ⫾ 6.04) region (P ⬎ 0.24), but was significantly increased (P ⬍ 0.0034 for both) in comparison to superior (26.7 ⫾ 8.20) and inferior
Tinling and Chole
Gerbilline cholesteatoma development III. Increased . . .
119
The KE in the normal PF was 4 to 6 cells thick with occasional labeled cells in the basal layer (Fig 4) and the LP contained numerous small capillaries and loose connective tissue. In contrast, the PF in ears with cholesteatoma had a significantly thickened layer of KE with numerous labeled cells and a thickened hyperemic LP. An identical change occurred in the PT when cholesteatoma matrix was present (Fig 4). In normal animals, the rate of cell division in the annular region was very low with 0 to 2 dividing cells per section (Fig 5A). In cholesteatoma samples in which matrix extended to the annular rim (Fig 5B), there was increased cell division, formation, and/or enlargement of ceruminous glands, and osteoclast activation along the bony wall similar to that seen in the upper portions of the canal (Fig 1B). Additionally, the periosteum on the middle ear side was hyperemic and inflamed. In cholesteatoma specimens, the bone of the lateral and medial wall of the EAC showed significant remodeling by osteoclasts with osteoblastic activation and presumed replication of precursors on the adjacent surface (Fig 6). A very interesting finding in one of the cholesteatoma specimens was the appearance of a perforation in the KE and subjacent dense fibrous layer of the KE next to the inferior annulus (Fig 7). The keratin mass has infiltrated the
Figure 1 (A) Normal EAC. The labeled cells present are not sufficiently dense to be seen at this low magnification. (B) EAC with cholesteatoma. The KE (thin arrow) and LP (thick arrow) are significantly thickened. Thin arrow: basal KE cells covered with silver grains. A, annulus; il, inferior lateral; m, malleus; pt, pars tensa; sl, superior lateral.
lateral (25.5 ⫾ 6.50) epithelium; and 4) there was no significant difference in the PI between superior medial, superior lateral, inferior lateral, anterior (46.9 ⫾ 12.60), or posterior (39.9 ⫾ 6.50) regions (all P ⬎ 0.051 to 0.97).
Histopathology Cholesteatoma specimens demonstrated qualitatively thickened KE and lamina propria (LP) in the TM and also the KE and subepithelium within the EAC when compared with normal tissue (Figs 1 and 5). A distinct line of radioactive labeled cells is present along the basal layer of the KE in ears with cholesteatoma (Fig 2). All cholesteatomas were stage 2 or 3, and all but one stage 2 completely filled the EAC. The one cholesteatoma that formed in the ETO group did not undergo retraction of the PF and could not be distinguished qualitatively in blinded analysis from the spontaneous or CL-induced cholesteatomas. In the spontaneous stage 2 cholesteatoma, which had not completely filled the inferior portion of the EAC, the KE in this region appeared normal (Fig 2) with only an occasionally labeled cell in the TM and the inferior lateral wall of the EAC.
Figure 2 Inferior lateral (il) portion of the EAC in which the cholesteatoma did not yet extend beyond the superior canal. Compare to Figs 1B and 4D. pt, pars tensa.
120
Otolaryngology–Head and Neck Surgery, Vol 135, No 1, July 2006
Figure 3 Mean proliferation index (PI) for basal keratinocytes in the ear by region. The PI was significantly increased for all regions (black bars) in cholesteatoma. eac, external ear canal; c, cochlea; m, malleus; me, middle ear; S, superior; L, lateral.
LP and has almost broken through the squamous epithelium lining of the middle ear surface.
DISCUSSION Several studies of human cholesteatoma tissue support the hypothesis that an increasing PI is associated with cholesteatoma development. Mallet et al12 demonstrated a significant association between hyperproliferation, middle ear inflammation, and a higher risk of recurrence in human cholesteatomas using a marker for hyperproliferation (monoclonal antibody MIB1). They demonstrated a significant correlation for a high PI with aggressive cholesteatoma in children. Huisman et al13 compared levels of the cellular markers Ki-67, p53, p21, and active caspase 3 for cell proliferation, cell-cycle arrest, and apoptosis in human cholesteatoma epithelium to paired control retro-auricular skin. The nuclear antigen, Ki-67, is associated with hyperproliferation and the proteins, p53 and p21, are associated with apoptosis and cell-cycle arrest while caspase 3 is an indicator of cell death. They demonstrated a significant increase for each of these markers in cholesteatoma except caspase 3. An increase in PI was supported by increased localization of Ki-67 in the basal KE of cholesteatomas compared to control skin. The co-localization in the basal KE of cholesteatomas of increased levels of p53 without an increase in caspase 3 supports the hypothesis that alteration of the normal cell
cycle without cell death contributes to the increase in the PI and resulting cholesteatoma development. Additionally, the increased expression of p21 in the suprabasal layers of cholesteatoma specimens suggests that the increased thickness of KE in cholesteatomas results from an increased rate of proliferation of basal KE combined with inhibition of keratinocyte transformation. An increase in the PI of KE in cholesteatoma has also been linked to upregulation of keratinocyte growth factor (KGF) and its receptor (KGFR). YamamotoFukuda et al14 demonstrated a positive correlation between increased expression of both KGF and KGFR with an increased expression of Ki-67 in human cholesteatomas compared to normal skin. In addition, they showed a significant correlation between KGF and KGFR expression and cholesteatoma recurrence. In this study, regional differences in the PI for the TM of normal animals were present prior to cholesteatoma formation. The PI in the PF was higher than the PI in the PT over the malleus, which was itself higher than PT proper. However, it is unlikely that an increase in the PI of the PF alone results in the majority of the keratin production in the initial phase of cholesteatoma formation as the ratio of the area of the PT to the PF in the gerbil is 7:1 (unpublished data). It is more likely that there is alteration in the normal clearance mechanism and rate of KE proliferation for both the TM and the EAC in those gerbils which form cholesteatomas. The increasing thickness of the KE of the PF, PT, and EAC leads to continued buildup of keratin, thus perpetuating the cycle (Fig 8). Sequestered bacterial flora and the resulting immu-
Tinling and Chole
Gerbilline cholesteatoma development III. Increased . . .
Figure 4 Normal and cholesteatoma TM. (A) Normal PF. Thin arrow: dividing cells covered by silver grains. Thick arrow: LP. (B) Cholesteatoma PF. ke, desquamating keratin. (C) Normal PT. Arrow: thin layers of desquamating keratin. (D) Cholesteatoma PT. Thick arrow: LP. Dense accumulation of keratin (k) in the EAC. The large separation of the keratin from the surface of the pars tensa (thin arrow) may represent a fixation artifact.
nologic response maintain the inflammatory component of this process. Several studies both human15 and gerbilline16-19 have investigated the theory of alterations in cytokeratin expression in the TM of human cholesteatoma as a possible mechanism for cholesteatoma formation. In a study of human PT and cholesteatoma compared to normal skin, Kakoi et al15 determined that the cytokeratin expression pattern suggested a hyperproliferative matrix. Studies by Kim et al18,19 noted significant changes in the expression patterns of cytokeratins in gerbilline cholesteatoma, which were variable in location and depended on the type of induction method. The hyperproliferation marker (CK13/16) was more pronounced in cholesteatomas induced by eustachian tube cautery than by ligation or propylene glycol application. They concluded that both a hyperproliferative process and migratory process coexist and contribute to the pathogenesis of gerbilline cholesteatoma. Kim and Chung16 concluded that
121
Figure 5 Annulus in normal and cholesteatoma ears. (A) Normal. (B) Cholesteatoma. eac, external ear canal; g, cerumen gland; me, middle ear. Scale applies to both.
cholesteatoma in ligated gerbils originates in EAC epithelium. However, they recognized that induction by ligation would force EAC epithelium into the middle ear. There-
Figure 6 Enlargement of lateral bulla wall (broad arrow in Fig 1B). Osteoclasts (thin arrows) in the LP (lp) eroding the lateral wall of the ear canal. There is significant osteoblastic activity and precursor replication and in the external periosteum (broad arrow). b, bone.
122
Otolaryngology–Head and Neck Surgery, Vol 135, No 1, July 2006
Figure 7 Perforation of the TM near the annulus. The KE and dense fibrous layer next to the annulus (a) are perforated (curved arrow) and there is infiltration of the LP (lp) by the keratin mass (m).
fore, their analysis concentrated on the lateral portion of the EAC and did not investigate differences between the EAC and the TM. In an immunohistochemical study of BrdU labeling of basal keratinocytes in ligated gerbil ears, Park et al20 reported an overall and regional increase similar to ours. The majority of labeling in the TM of normal animals occurred in the basal layer of the PF and PT over the malleus with regional differences in the degree of labeling in the EAC. They noted that labeling was higher in the superior regions of the canal with no anterior-to-posterior difference. For cholesteatoma tissue, Park et al20 did not report regional differences, but compared cholesteatomas grossly by stage with all stages significantly greater than normal. They concluded that increasing keratin in the EAC may also cause an increased mitotic activity in the surrounding epithelium. However, in our study, we measured regional PIs and they were significantly increased over normal in all regions. The effect of accumulated keratin on hyperproliferation, hypothesized by Park et al,20 was evident in our study. In
Figure 8
the stage 2 cholesteatoma, which had not yet extended inferiorly, the PI in the inferior canal was not significantly different from normal, nor were any sections without accumulated keratin recognized as histologically abnormal during blinded analysis, whereas labeling was significantly increased in all other samples in which cholesteatoma matrix was present in the inferior EAC. Tanaka et al21 examined normal and cholesteatoma tissue from humans for the presence of proliferating cell nuclear antigen (PCNA) and transforming growth factor-␣ (TGF-␣). They detected increased PCNA above the suprabasal layer with differences ascribed to subepidermal cell inflammation. Staining for TGF-␣ occurred principally in the granular and prickle cell layers of normal skin and in all layers within cholesteatomas. However, it is unlikely that cell division was occurring above the basal layer in this study. Diffuse staining in the upper regions was most likely the result of nonresorbed PCNA in the differentiating epithelial cells. In this study, we demonstrate that the rate of cell division in basal cell keratinocytes of the gerbil TM and EAC within developing cholesteatoma is significantly increased by approximately 7 times that of normal TM and EAC. There was no apparent difference in division rate among the different methods of cholesteatoma induction and they appeared histologically identical. There were distinct regional differences in the PI for basal keratinocytes both within and between normal and cholesteatoma tissues. The regions with the highest PI were the medial and lateral regions of the superior canal. This suggests that, during cholesteatoma formation in gerbils, this portion of the canal and TM is the first to respond to an induction stimulus. Our results, in conjunction with those of Park et al20 and Tanaka et al,21 provide support for the following hypothesis of a cholesteatoma development cycle in gerbils (Fig 8): 1) middle ear inflammation occurs in conjunction with (or may cause) a disruption of the clearance mechanism for desquamating keratin, resulting in accu-
Cholesteatoma development cycle.
Tinling and Chole
Gerbilline cholesteatoma development III. Increased . . .
mulation of keratin on the; 2) a disruption in normal cell-cycle regulation results in an increase in the PI for basal keratinocytes of the PF and PT; 3) increased PI without a concomitant increase in keratinocyte maturation leads to an increased thickness of the keratinizing epithelium (KE); 4) the increasing accumulation of keratin results in blockage of the opening to the external ear canal. The resulting superior-to-inferiorly-directed filling of the canal involves unaffected regions of the PT and EAC and continues the cycle. Continued production of trapped keratin results in distension of the TM and middle ear morbidity in gerbils. Statistical analysis was provided by Dr Mitchell Watnik of the Statistical Laboratory, University of California, Davis.
REFERENCES 1. Chole RA, Tinling SP. Epidermoid cholesteatoma. In: Friedmann I, Arnold W, editors. Pathology of the ear. New York: Churchill Livingstone; 1993. p. 752. 2. Sadé J, Babiacki A, Pinkus G. The metaplastic and congenital origin of cholesteatoma. Acta Otolaryngol 1983;96:119 –29. 3. Friedmann I. The comparative pathology of otitis media— experimental and human. J Laryngol Otol 1955;69:588 – 601. 4. Graham MD, Nelson RA, Kemink JL. Tympanic membrane cholesteatoma (keratoma). Am J Otol 1984;5:175–7. 5. Goycoolea MV, Heub MM, Muchow D, et al. The theory of the trigger, the bridge and the transmigration in the pathogenesis of acquired cholesteatoma. Acta Otolaryngol 1999;119:244 – 8. 6. Lim DJ, Saunders WH. Acquired cholesteatoma: Light and electron microscopic observations. Ann Otol Rhinol Laryngol 1972;81:2–12. 7. Chole RA, Tinling SP. Basal lamina breaks in the histogenesis of cholesteatoma. Laryngoscope 1985;95:270 –5.
123
8. Henry KR, Chole RA, McGinn MD. Age-related increase of spontaneous aural cholesteatoma in the Mongolian gerbil. Arch Otolaryngol 1983;109:19 –21. 9. Wolfman DE, Chole RA. Experimental retraction pocket cholesteatomas. Ann Otol Rhinol Laryngol 1986;95:639 – 44. 10. McGinn MD, Henry KR, Chole RA. Cholesteatoma: Experimental induction in the mongolian gerbil. Acta Otolaryngol 1982;93:61–7. 11. Aminpour S, Tinling SP, Brodie HA. Role of tumor necrosis factor-␣ sensorineural hearing loss after bacterial meningitis. Otol Neurotol 2005;26:602–9. 12. Mallet Y, Nouwen J, LecomteHoucke M, et al. Aggressiveness and quantification of epithelial proliferation of middle ear cholesteatoma by MIB1. Laryngoscope 2003;113:328 –31. 13. Huisman MA, Heer ED, Grote JJ. Cholesteatoma epithelium is characterized by increased expression of Ki-67, p53 and p21, with minimal apoptosis. Acta Otolaryngol 2003;123:377– 82. 14. Yamamoto-Fukuda T, Aoki D, Hishikawa Y, et al. Possible involvement of keratinocyte growth factor and its receptor in enhanced epithelial-cell proliferation and acquired recurrence of middle-ear cholesteatoma. Lab Invest 2003;83:123–36. 15. Kakoi H, Tamagawa Y, Kitamura K, et al. Cytokeratin expression patterns by one- and two-dimensional electrophoresis in pars flaccida cholesteatoma and pars tensa cholesteatoma. Acta Otolaryngol 1995;115:804 –10. 16. Kim CS, Chung JW. Morphologic and biologic changes of experimentally induced cholesteatoma in Mongolian gerbils with anticytokeratin and lectin study. Am J Otol 1999;20:13– 8. 17. Kim H-J, Tinling SP, Chole RA. Expression patterns of cytokeratins in retraction pocket cholesteatomas. Laryngoscope 2001;111:1032– 6. 18. Kim H-J, Tinling SP, Chole RA. Expression patterns of cytokeratins in cholesteatomas: Evidence of increased migration and proliferation. J Korean Med Sci 2002;17:381– 8. 19. Kim H-J, Tinling SP, Chole RA. Increased proliferation and migration of epithelium in advancing experimental cholesteatomas. Otol Neurotol 2002;23:840 – 4. 20. Park K, Chun Y-M, Park H-J, et al. Immunohistochemical study of cell proliferation using BrdU labeling on tympanic membrane, external auditory canal and induced cholesteatoma in Mongolian gerbils. Acta Otolaryngol 1999;119:874 –9. 21. Tanaka Y, Shiwa M, Kojima H, et al. A study on epidermal proliferation ability in cholesteatoma. Laryngoscope 1998;108:537– 42.