Managing cholesteatomas with labyrinthine fistula

Abstract

Introduction:

A retrospective study (2011 to 2018) was conducted to evaluate the management of cholesteatomas with labyrinthine fistulae (LFs), clinical characteristics and postoperative hearing outcomes in a hospital.

Methods:

Demographic data of patients with primary middle ear mastoidectomies for cholesteatoma were extracted. Preoperative high-resolution computed tomography (HRCT) temporal bone and intraoperative findings, and hearing levels preoperatively and postoperatively were evaluated.

Results:

Of the middle ear cholesteatomas, 15.6% (n = 14) of ears were complicated by LF. HRCT scans showed 92.9% sensitivity and 94.7% specificity in the identification of LFs. Intraoperative findings of LFs include stapes erosion (78.6%), malleus erosion (78.6%), incus erosion (92.9%), dehiscence of tegmen tympani (28.6%) and tympanic facial canal (64.3%). Compared to the non-LF group, the LF group showed significantly higher incidence of stapes erosion (P < 0.001), tegmen tympani dehiscence (P = 0.016) and semicircular canal dehiscence (P < 0.001). Matrix was removed completely in 85.7% (n = 12) and was left behind in 14.3% (n = 2) of ears. Also, 21.5% (n = 3) had preoperative dead ears. Postoperative hearing results had a mean follow-up time of 2.1 (standard deviation 1.5, range 0.14–4.84) years. In the matrix removal group (n = 9), 77.9% had unchanged hearing levels, 11.1% showed improvement and 11.1% showed decrease in hearing levels. The matrix preservation group (n = 2) had deteriorated hearing levels.

Conclusion:

Preservation of hearing in LFs is possible with cautious matrix removal. Despite matrix preservation to preserve hearing in large LFs, our patients’ hearing deteriorated postoperatively. Longer follow-up of hearing with matrix preservation may show poorer hearing outcomes.

INTRODUCTION

Labyrinthine fistulae (LFs) are formed from erosion of the bony labyrinth and occur as a complication in 2.7%–21%[1,2] of cholesteatomas. Cholesteatomas are a result of proliferation of keratinised squamous epithelium and debris produced from the matrix. The propensity to invade into surrounding structures including bone destruction, potentially causes adverse auditory and vestibular effects. The early identification of LF allows for better preoperative planning of surgical technique to remove cholesteatomas and reduce potential vestibulocochlear sequelae.

The diagnosis of LF is confirmed intraoperatively, whereby dehiscence of semicircular canal, cochlear and vestibule can be clearly visualised. A history of hearing loss, aural fullness, dizziness without vertigo or disequilibrium may support the diagnosis of LF, but these symptoms could also be present in chronic suppurative otitis media. High-resolution computed tomography (HRCT) temporal bone scans are the mainstay preoperative investigation in the evaluation of LF. However, in the literature, HRCT has shown a variable detection rate of 53.8%–85%,[3,4] making preoperative diagnosis difficult.

Surgical management of LF has been debated across various institutions. Firstly, the two-staged procedure of leaving the cholesteatoma matrix overlying the fistula and exteriorisation of the cavity is followed by a second intervention to remove the matrix based on postoperative hearing deterioration and vestibular symptoms. This was traditionally done to avoid opening the labyrinth to reduce potential vestibulocochlear sequelae.[5] Secondly, the one-stage procedure of complete removal of cholesteatoma and simultaneous fistula repair has been increasingly used due to its association with lower risk of recurrence and reduced infections from the remaining nidus in the former.[6] Although classification systems considering the size of fistula and hearing status have been proposed to guide operative techniques, to date, there is no standardisation. A systematic review including 28 studies from 1970 to 2014 found that hearing preservation rates for either matrix removal or exteriorisation were high and there was no difference in hearing preservation rates for either technique.[7]

Consequently, in this study, we reported our local experience in the management of LF in middle ear cholesteatomas, over an 8-year period, at a single tertiary hospital. We aimed to discuss the clinical characteristics, and the management and outcomes of LF.

METHODS

A retrospective case review via the electronic surgical logbook was done for mastoidectomies performed between July 2011 and July 2018 by two senior surgeons. Patients with primary cholesteatomas were included, while patients with congenital cholesteatomas or with incomplete data were excluded. Ethics approval was obtained from the institutional review board (IRB 2019/2791) before commencement of the study.

Data extracted comprised preoperative demographics, HRCT 0.5-mm thickness temporal bone scan results, intraoperative findings and hearing results. Diagnostic accuracy and correlations of noncontrast HRCT scans with intraoperative findings were analysed for (a) erosion of structures — ossicular chain, tegmen tympani, tympanic facial canal and semicircular canal; (b) location of squames; and (c) ossicular chain encasement. Surgical approach for all LFs involved canal wall down mastoidectomy. The technique of matrix preservation or removal was dependent on intraoperative fistula size and patient’s preoperative hearing. Matrix removal was done for small fistulae, defined as <4 mm, while matrix preservation was done for large fistulae, defined as ≥4 mm, in an ear with good hearing, that is, bone threshold <40 dB. The material used to repair LF was also noted. To enable optimal exposure and clearance of cholesteatoma, all cases underwent canal wall down mastoidectomy. Matrix removal and fistula repair were done after cholesteatoma had been completely removed in other parts of the ear before middle ear reconstruction. Extra care was taken not to apply direct suction over the fistula. Hearing outcome measures analysed were average air conduction (AC), bone conduction (BC) and air–bone gap pure tone threshold (PTA), performed at 500 Hz, 1 kHz and 2 kHz for all patients pre- and postoperatively. A change in BC PTA of 10 dB or more was considered clinically significant.

Statistical analysis was conducted using IBM SPSS Statistics version 25 (IBM Corp, Armonk, NY, USA). Numerical variables were described as mean (standard deviation, range). Normality of data was tested for by using Shapiro–Wilk test. Cholesteatomas that were categorised into ‘LF’ and ‘non-LF’ groups showed a normal distribution. Comparison of categorical variables was performed using the Pearson chi-squared test. A P value < 0.05 was considered statistically significant. However, given the very small sample sizes involved, meaningful statistical conclusions cannot be made.

RESULTS

A total of 90 ears underwent mastoidectomies for primary middle ear cholesteatomas during the 8-year period. Of these, 5.6% (n = 5) had bilateral cholesteatomas. Also, 15.6% (n = 14) of ears in 13 patients were found to have LFs intraoperatively, with an equal number of left- and right-sided LFs. Of the patients, 54.4% (n = 49) were male and 69.2% (n = 9) were Chinese. Mean age of diagnosis was 52 (SD 19.6, range 8–78) years. Presenting complaints of ears with LFs were hearing loss in 85.7% (n = 12), otorrhoea in 78.6% (n = 11), giddiness in 28.6% (n = 4), tinnitus in 7.1% (n = 1), facial palsy in 7.1% (n = 1) and otalgia in 7.1% (n = 1). Moreover, 64.3% (n = 9) had both hearing loss and otalgia.

Preoperative HRCT accurately identified 13 of the 14 LFs diagnosed intraoperatively, with a false positive of four ears and false negative of one ear. This reflected a sensitivity of 92.9% and a specificity of 94.7% in the identification of LFs through semicircular canal dehiscence. All fistulae were located in lateral Semicircular Canal (SCC) and 7.1% (n = 1) were located in superior SCC.

Comparison of CT and intraoperative findings of between LF and non-LF groups is shown in Table 1. In LF group, intraoperative dehiscence was observed — 28.6% (n = 4) facial canal, 64.3% (n = 9) tegmen tympani. Ossicular erosion occurred in 78.6% (n = 11) stapes, 78.6% (n = 11) incus and 92.9% (n = 13) malleus. Intraoperative squames were found, 92.9% (n = 13) attic, 92.9% (n = 13) middle ear and 71.4% (n = 10) mastoid. Ossicular chain encasement occurred in 71.4% stapes (n = 10), 92.9% (n = 13) incus and 78.6% malleus (n = 11). Compared to the non-LF group, we found a significantly higher incidence of stapes erosion (P < 0.001) and dehiscence in tegmen tympani (P < 0.001) in LF group [Table 1].

Table 1.

CT and intraoperative findings of cholesteatomas between ‘LF’ and ‘non-LF’ groups.

Characteristic	n (%)			P
	Total (n=90)	LF (n=14)	Non-LF (n=76)
CT findings
Ossicular chain erosion
Stapes	31 (34)	8 (57)	19 (31)	0.068
Incus	64 (71)	13 (93)	52 (85)	0.450
Malleus	78 (87)	12 (86)	43 (71)	0.245
Dehiscence
Tegmen	24 (27)	7 (50)	14 (23)	0.042*
Facial canal	41 (46)	7 (50)	29 (48)	0.028*
Semicircular canal	13 (1)	13 (93)	4 (7)	<0.001*
Intraoperative findings
Ossicular chain erosion
Stapes	27 (36)	11 (79)	16 (26)	<0.001*
Incus	67 (89)	13 (93)	54 (89)	0.245
Malleus	53 (71)	11 (79)	42 (69)	0.471
Tegmen dehiscence	8 (11)	4 (29)	4 (7)	0.016*
Facial canal dehiscence	34 (45)	9 (64)	25 (41)	0.114
Semicircular canal dehiscence	14 (19)	14 (100)	0 (0)	<0.001*
Location of squames
Attic	68 (91)	13 (93)	55 (90)	0.755
Eustachian tube	3 (4)	0 (0)	3 (5)	0.397
Middle ear	51 (68)	13 (93)	38 (62)	0.027*
Mastoid	49 (65)	10 (71)	39 (64)	0.595
Encasement of ossicular chain
Stapes	42 (56)	10 (71)	32 (53)	0.197
Incus	66 (88)	13 (93)	53 (87)	0.535
Malleus	53 (71)	11 (79)	42 (69)	0.471

*P<0.05 was considered statistically significant. CT: computed tomography, LFs: labyrinthine fistulae

All cholesteatomas with LFs underwent canal wall down mastoidectomy and primary mastoid obliteration with local tissue flaps (middle temporal artery and inferior musculoperiosteal flaps). Middle ear reconstruction utilised cartilage placed onto remnant stapes suprastructure if present (type 3 tympanoplasty) or stapes footplate if the stapes suprastructure was absent (type 4 tympanoplasty). Surgeon 1 performed complete matrix removal in 85.7% (n = 12) of LFs, by leaving the matrix and cholesteatoma over the SCC till the last stage of operation, before delicately removing the matrix. This was followed by sealing the fistula with bone wax in 50.0% (n = 6) and temporalis fascia in 8.3% (n = 1). The size of the fistula in these cases was less than 4 mm. Surgeon 2 performed matrix preservation in the remaining 14.3% (n = 2) SCC fistulae. These two patients had large fistulae of the lateral semicircular canal, measuring 6 and 10 mm, respectively, with preoperatively normal hearing. All the fistulae had breached the endosteum of the semicircular canal.

Recurrence of cholesteatoma occurred in one patient (7.1%), requiring revision mastoidectomy. The time to recurrence was 5 years from the primary surgery. The initial disease was very extensive, with cholesteatoma occupying the entire epitympanum, mastoid and sinus tympani. This patient’s tympanic membrane was severely retracted onto the promontory. The cholesteatoma recurred in the epitympanum and mastoid.

Postoperative hearing PTAs were recorded at a mean of 2.1 (SD 1.5, range 0.14–4.84) years. Preoperative severe to profound sensorineural hearing loss was found in 21.5% (n = 3) ears. The remaining 11 ears were evaluated for their pre- and postoperative hearing. Of these 11 ears with hearing, the mean pre- and postoperative BC thresholds were 33.5 dB (16.1, 13–68) and 42.1 dB (19.6, 13–72), respectively, and the AC thresholds were 67.5 (20.4, 47–113) dB and 77.3 (26.1, 33–110) dB, respectively. Of the 11 ears, nine ears had their matrix completely removed and LFs repaired, and two ears had matrix preservation with a modified radical mastoidectomy performed as the fistulae measured 6 and 10 mm, respectively. Both patients with large fistula had preoperative normal hearing.

The mean preoperative and postoperative AC and BC thresholds for matrix removal versus preservation are shown in Table 2. In the matrix removal group, the mean preoperative and postoperative BC thresholds were 33.1 (17.8, 13–68) dB and 38.8 (19.3, 13–68) dB, respectively. The matrix removal group (n = 9) had hearing levels with no change in 78% (n = 6). One patient had improved average bone threshold from 32 to 13 dB, possibly due to Carhart’s effect. Two patients had their average bone thresholds deteriorated by more than 10 dB from 47 to 68 dB and from 30 to 53 dB, respectively. In terms of hearing reconstruction for these two patients, the first patient had type 3 cartilage tympanoplasty and the second patient had type 4 tympanoplasty.

Table 2.

Postoperative hearing levels in matrix removal and preservation intraoperatively in the LF group.

Hearing levels	n (%)
	Matrix removal (n=9)	Matrix preservation (n=2)
Change in hearing levels
No change	6 (78)	0 (0)
Improved	1 (11)	0 (0)
Decreased	2 (11)	2 (100)
Mean hearing thresholdsa
Air
Preoperative	66.7 (21.3, 47-113)	71.0 (22.6, 55-87)
Postoperative	71.6 (25.2, 33-108)	108.5 (2.12, 107-110)
Bone
Preoperative	33.1 (17.8, 13-68)	35.0 (7.07, 30-40)
Postoperative	38.8 (19.3, 13-68)	68.5 (4.95, 65-72)

^aData presented as mean (standard deviation, range). LFs: labyrinthine fistulae

The two LF with matrix preservation had deteriorated hearing levels, after a follow-up time of 3.2 years and 2.7 years respectively. Increase in BC thresholds preoperatively to postoperatively was from 30 to 65 dB over 2.7 years for patient 1 and from 40 to 72 dB over 3.2 years for patient 2. Patient 1 possibly had sudden sensorineural hearing loss 4 months postoperatively and his hearing continued to deteriorate to a dead ear 8 months postoperatively, in spite of two courses of oral prednisolone. Magnetic resonance imaging (MRI) did not show any loss of T2 signal or enhancement of the inner ear. Patient 2 had type 4 cartilage tympanoplasty.

DISCUSSION

Our study found LFs in 15.6% of middle ear cholesteatomas. The evaluation of cholesteatomas preoperatively with HRCT yielded a high sensitivity (92.9%) and specificity (94.7%) for LF. A significantly higher incidence of tegmen tympani dehiscence, stapes erosion was seen in ears with LF. Matrix management technique in our centre is based on intraoperative fistula size and preoperative hearing. The two LF patients had their matrix preserved, as both patients had large fistulae measuring 6 and 10 mm, respectively, with preoperatively normal hearing. Opening the membranous labyrinth results in poorer postoperative bone thresholds,[8] hence we opted not to remove the matrix for large fistulae (>4 mm).

Labyrinthine fistula is a known complication of cholesteatomas, most common in the lateral SCC;[9,10] superior SCC is affected in 6% of cases and posterior SCC in 2% of cases.[11] Patients with LF may be asymptomatic. Classic clinical signs such as fistula sign may only have a sensitivity of 21.7%–60%,[12,13] with a false-negative rate of up to 54%.[14] Also, there is lack of uniformity in performing the test. The evaluation of cholesteatomas for LF from literature yields a sensitivity of 53.8%–100%[3,15] and specificity of 90%–100%.[3,16] High-resolution computed tomography temporal bone has shown good sensitivity and specificity locally in detection of cholesteatomas and their extents.[17,18]

While some centres have used the Dornhoffer and Milewski’s classification to stage fistulae based on bony defects and size of fistulae to determine the surgical method, the fistula size and grade do not correlate well with surgical outcomes.[7] Surgeon 1’s technique of matrix removal was applied to all LFs, all of which were less than 4 mm. Sealing of fistulae included bone wax and temporalis fascia, which is consistent with literature that also reports the use of bone wax, temporalis fascia, bone dust, tisseel/tissucol, fibrin glue, Spongiostan and autogenous tissues.[7,11]

Postoperative deterioration of AC has been reported in 2.8%–26.9% of the cases,[7,9] which is attributed to the removal of cholesteatomas and granulation tissue that act as conduits for conducting sound into the ear. Postoperative deterioration in sensorineural hearing loss occurs in 23% to 50% of the cases that underwent complete matrix removal and 1% to 44% of the cases for matrix left in situ.[6,7] Unlike the review by Lim et al.,[7] we observed significantly better postoperative hearing outcomes in our patients with complete matrix removal than the two patients with matrix preserved in our study. Patients with their matrix removed either had same or worse BC thresholds. Two patients’ hearing deteriorated (BC from 30 to 53 and from 47 to 68), while the rest remained the same. We postulate that the remnant cholesteatoma may have caused chemical labyrinthitis, resulting in worsened sensorineural hearing. If hearing deteriorates over time due to labyrinthitis, steroids can be considered to improve hearing outcomes. This is also supported by increasing preference towards complete matrix removal for disease clearance, since residual disease from matrix preservation may potentially worsen hearing outcomes.[11,19] However, we acknowledged that two patients with matrix preservation is a small sample size and a larger number of patients may be needed for better comparison.

A greater deterioration of long-term hearing outcomes had also been reported postoperatively, despite initial improvements within the first year.[3] Reddy et al. observed hearing preservation post-labyrinthectomy and proposed that other methods, such as occlusion of the ductus reuniens with granulation tissue or bone dust, may be sufficient to achieve that outcome, despite the absence of fluid-filled vestibule post-labyrinthectomy.[20] Most recently, the “Sandwich technique” (temporal fascia–bone pate–temporal fascia) by Bartochowska et al reported hearing preservation and antivertiginous effect.[21]

Recurrence of cholesteatomas only occurred in one patient in this study, requiring revision surgery. A meta-analysis by Tomlin et al. reported a cholesteatoma recurrence rate of up to 5%–17% in canal wall down mastoidectomy and 9%–70% in canal wall up mastoidectomy.[22] With only two patients with matrix preserved, we cannot compare the hearing outcomes with the larger group of 12 patients with matrix removed. Additionally, there was variability in when the post-operative audiograms were done till. Follow-up time postoperatively was not uniform as some patients were unfortunately lost to follow-up. Other confounding variables for sensorineural hearing loss such as age, previous otitis media and use of ototoxic drugs have not been adjusted for in the study.

In conclusion, we presented our experience in management of cholesteatomas with labyrinthine fistulae, based on intraoperative fistula size and patient’s pre-operative hearing. Preservation of hearing in labyrinthine fistulae is possible with cautious matrix removal and immediate repair of the fistula. Despite matrix preservation to preserve hearing in two patients with large labyrinthine fistulae, both patients’ hearing deteriorated postoperatively. A larger sample size of patients with matrix preservation over large fistulae would be needed to determine if this group of patients is susceptible to poorer post-operative hearing.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.