Lateral Tympanoplasty for Total or Near-Total Perforation: Prognostic Factors

Transcription

1 The Laryngoscope Lippincott Williams & Wilkins, Inc The American Laryngological, Rhinological and Otological Society, Inc. Lateral Tympanoplasty for Total or Near-Total Perforation: Prognostic Factors Simon I. Angeli, MD; Jessica L. Kulak; Jose Guzmán, MD AQ: 1 Objective: To identify prognostic factors affecting outcome in lateral tympanoplasty for total or neartotal tympanic membrane perforation. Study Design: Retrospective case series. Methods: Patients were those presenting with total or near-total tympanic membrane perforation undergoing lateral tympanoplasty from 1999 to We systematically collected demographic, clinical, audiologic, and outcome information. Student t test was used to determine group differences. Logistic regression analysis was used to examine the relationship between success of grafting (dependent variable) and the independent variables. Multiple regression analysis was used to examine the relationship between postoperative air-bone gap (ABG) and independent variables. Results: There were seventy-seven cases (58 primary and 19 revision cases) with average follow-up of 17 months. Successful tympanic membrane grafting occurred in 91% of cases. None of the independent variables studied was predictive of the success of graft incorporation (P >.05). The mean preoperative ABG was db and improved to a postoperative ABG of db (P <.001). Smaller preoperative ABG and normal malleus handle were associated with smaller postoperative ABG. In revision cases, mastoidectomy was associated with better functional results. Conclusions: Successful grafting of near-total and total tympanic membrane perforations occurred in 91% of the cases and was independent of demographic, disease, and technical variables. Disease variables (preoperative ABG and status of malleus handle) had a greater prognostic value on postoperative ABG than other variables. In revision tympanoplasty, mastoidectomy is associated with a better functional outcome. Key Words: From the Department of Otolaryngology (S.I.A.), University of Miami Miller School of Medicine, Miami, Florida, U.S.A, the University of Miami Miller School of Medicine (J.L.K.), Miami, Florida, U.S.A, and Hospital Militar (J.G.), Bogotá, Colombia. Editor s Note: This manuscript was accepted for publication May 31, Accepted for Oral Presentation at the 2006 Southern Section Meeting of The Triological Society, Naples, Florida, U.S.A., January 13 14, Send correspondence to Dr. Simon I. Angeli, Department of Otolaryngology, University of Miami Miller School of Medicine, 1666 N.W. 10th Avenue, Suite 312, Miami, FL sangeli@med.miami.edu DOI: /01.mlg Eardrum, tympanic membrane reconstruction, prognosis, regression analysis, conductive HL. Laryngoscope, 116:, 2006 INTRODUCTION Large tympanic membrane perforations present a surgical challenge. Large perforations tend to be more symptomatic (otorrhea and hearing loss [HL]) and are associated with greater degrees of middle ear disease, including cholesteatoma. It has been reported that the size of the perforation is a prognostic factor, and poorer results are obtained with large versus small perforations. 1 3 In cases of large perforations, the incorporation of the graft is compromised by several factors including limitations in surgical exposure, vascularity, and support for the donor material. There is some agreement that techniques that address these limitations provide a better chance of closure. The lateral or overlay technique as popularized by Sheehy and Anderson 4 effectively addresses these limitations and has been used by many centers with consistent good results. Alternative techniques have been described and include underlay grafting, rotational flaps, mediolateral grafting, and use of various grafting material including connective tissue autografts, homografts, and allografts. 5 8 The choice of technique has largely been the surgeon s preference, and controlled studies comparing outcomes among different techniques are lacking. Instead, it is customary to compare results among published case series. Another approach is the analysis of prognostic factors to identify those factors that have a significant relevance on the outcome; because some of these factors are technique related, this type of analysis assesses the relative influence of technical variables in interaction with demographic and disease variables. We have used the classic lateral tympanoplasty with fascia graft to treat total or near-total perforations for several years. Because we have prospectively and systematically collected the clinical information and have used a uniform management protocol, these cases are suitable for outcome analysis. Demographic, technical, and disease variables probably have an effect on the outcome in different ways. Most authors agree that the extent of disease is perhaps the most critical factor influencing outcome. 2 However, with few exceptions, 2,9,10 most studies analyze single factors separately. The purpose of this study is to 1

2 use regression models to assess the prognostic value of demographic, technical, and disease factors influencing the outcome of the lateral tympanoplasty when used specifically in a population of patients with total or near-total perforations of the tympanic membrane. Knowledge of the prognostic value of these variables may help clinicians formulate effective surgical plans and patients make informed decisions and realistic expectations. PATIENTS AND METHODS Study Design From September 1999 to December 2004, we kept an electronic database of 236 consecutive cases of tympanoplasty surgery cases performed by one of the authors (S.I.A.) at the University of Miami Ear Institute. We searched this database to identify patients who met the inclusion criteria for this study 1): diagnosis of total or near-total tympanic membrane perforation, 2) lateral tympanoplasty surgery with fascia graft, 3) diagnosis of chronic suppurative otitis media with and without cholesteatoma, and 4) minimum follow-up of 6 months. Total perforation is defined as complete absence of tympanic membrane remnant and annular ligament, and near-total perforation is defined as total membrane perforation with a preserved annular ligament. If both ears of a patient met the inclusion criteria, each ear was treated separately for analysis. We selected 81 cases that included 61 primary and 20 revision surgeries. During the study period, we followed a uniform management plan including demographic and clinical data collection, selection of patients for surgery, surgical technique, and postoperative care. Demographic and clinical data included age, sex, diagnosis (chronic suppurative otitis media with or without cholesteatoma), prior otologic surgery, details of surgical technique, intraoperative findings (middle ear mucosa status, ossicular chain status, and reconstruction), postoperative findings (graft incorporation, middle ear aeration, and complications), hearing, and duration of follow-up. The main outcome measures were success of graft incorporation and postoperative hearing function. Successful graft incorporation is defined as an intact neotympanum with an aerated middle ear at the last postoperative follow-up examination. Atelectatic grafts were classified as failure. In addition to reperforation and atelectasis, we specifically ascertained for complications related to the ear canal (ear canal epithelitis, ear canal synechia, and anterior angle blunting), tympanic membrane (lateralization of neotympanum, eardrum epithelitis, epithelial cysts), and change in hearing. Each patient underwent audiologic evaluation preoperatively and at determined intervals after surgery: 3, 6, and 12 months and yearly thereafter. Audiologic evaluation included pure-tone audiometry, including air-conduction and bone-conduction testing, and speech audiometry by speech reception threshold and word discrimination scores testing. Masking was used when appropriate. The audiologic data were analyzed and reported according to accepted guidelines. 11 The pure-tone average (PTA) for bone- and airconduction of 0.5, 1, 2, and 4 khz was calculated. Using values obtained at each time interval, we subtracted bone-conduction PTA from the air-conduction PTA to calculate the air-bone gap (ABG). We also calculated the three-frequency (1, 2, 4 khz) bone-conduction PTA preoperatively and at the postoperative intervals to evaluate for operative damage to hearing. We used the preoperative and the most recent postoperative audiogram for the analysis of the ABG closure. Surgical Technique and Postoperative Care The lateral tympanoplasty technique used in all of the cases was based on the one described by Sheehy and Anderson. 4 The basic steps were 1) transmeatal canal incisions and elevation of a 2 posterior-superiorly based vascular strip; 2) postauricular exposure and harvesting of the temporalis fascia graft; 3) removal of anterior ear canal skin and epithelial layer of the eardrum remnant; 4) canalplasty with drilling of the anterior canal bulge; 5) placement of the graft under the manubrium but over the tympanic annulus, if present, or over a groove created in the anterior ear canal just lateral to the tympanic sulcus if the annulus was absent or insufficient; 6) replacement of the anterior ear canal graft and the pedicled vascular strip; 7) ear canal packing with bovine collagen sponge impregnated in antibiotic otic solution; and 8) closure of the postauricular incision. If the manubrium was absent, a flap was cut in the upper edge of the graft and tucked under the lateral wall of the epitympanum. In cases with absent tympanic annulus, nylon strips were wedged over the fascia and anterior skin graft to recreate the anterior tympanomeatal angle. In some cases, a mastoidectomy or ossiculoplasty with prosthesis were performed in combination with the tympanoplasty. When the incus was absent, we used one of three reconstructive prostheses based on the availability of the ossicles 1) incus replacement prosthesis when the malleus handle was present; (2) partial ossicular reconstruction prosthesis (PORP) when the handle was short or absent and the stapes suprastructure was present; and 3) total ossicular reconstruction prosthesis (TORP) when the handle was short or absent and the stapes suprastructure was absent. Cartilage grafts were sometimes used to reconstruct a defect in the wall of the epitympanum and for interposition between prostheses (PORP and TORP) and the neotympanum. All of the patients received equal postoperative care. Otic drops were used beginning 1 week after surgery and continued until the healing was complete. The ear canal packing and nylon strips were removed at the 3 to 4 week postoperative visit. Patients were observed monthly until healing was completed and then at 6 month intervals. To assure uniform record keeping, a data sheet was designed to record details of the technique as well as intraoperative and postoperative findings. Complications related to faulty healing were managed as office procedures during the postoperative visits. Epithelial cysts were marsupialized. We performed aural cleaning with local debridement of areas with epithelitis and granulation tissue; antibiotic and corticosteroid otic solutions were applied directly or by means of sponge wicks. Statistical Analysis The independent variables of this study were sex (male 0, female 1), age, diagnosis (chronic otitis media 0, cholesteatoma 1), preoperative ABG, type of surgery (primary 0, revision 1), mastoidectomy (not performed 0, performed 1), status of malleus handle (normal 0, short or absent 1), status of stapes suprastructure (normal 0, absent 1), status of tympanic annulus (normal 0, absent 1), and status of middle ear mucosa (normal 0, inflamed or wet 1). The dependent variables were success of graft incorporation (success 0, failure 1) and postoperative ABG. Hearing and age were continuous variables, whereas nominal and ordinal variables were dummy coded. Univariate analysis of group differences in terms of success of graft incorporation and postoperative ABG were evaluated by chi-square and Student s t test, respectively. Multiple regression analysis was used to examine the relationship between the dependent variable, postoperative ABG, and the independent variables. The standardized regression coefficients ( coefficient) were calculated to determine a hierarchy among the variables in terms of the magnitude of their influence. Logistic regression analysis was used to examine the relationship between success of grafting and independent variables. The level of significance was.05. The calculations were performed with the JMP IN statistical software package (SAS Institute, Inc., Belmont, CA).

3 T1 T2 TABLE I. Demographic and Clinical Data of Lateral Tympanoplasty for Total and Near-Total Tympanic Membrane Perforations. Datum Value Number of ears 77 Number of patients 74 Age at surgery, yr, mean SD (range) (2 75) Sex ratio male:female 1.75:1 Type of surgery (n) Primary 58 Revision 19 Diagnosis (n) Chronic otitis media 52 Cholesteatoma 25 Air-bone gap, decibels, mean SD (range) Preoperative (11 62) Postoperative * (5 55) *Student s t test, P.001 (see text for details of test). n number of cases; SD standard deviation. TABLE II. Results for Logistic Regression Analysis of Success of Graft Incorporation in 77 Cases of Lateral Tympanoplasty. Variable Estimate SE Chi-Square P Intercept Malleus(0) PreABG Me mucosa(0) Age Sex(0) Diagnosis(0) Mastoidectomy(0) Stapes arch(0) Revision(0) Annulus(0) Model R , P (0) not performed; Me middle ear; SE standard error; R 2 coefficient of determination. RESULTS Three primary surgery patients and one revision surgery patient were lost to follow-up or had incomplete records; therefore, subsequent analysis involved a final population of 77 cases: 58 primary and 19 revision cases. The average follow-up was 17 (range, 6 45) months. Table I shows patients demographic and clinical data. Successful graft incorporation was achieved in 70 of 77 (91%) cases: 52 of 58 (90%) primary cases and 18 of 19 (95%) revision cases. Reperforation of the neotympanum occurred in four primary cases, and severe retraction (atelectasis) of the neotympanum occurred in two primary cases and one revision case. One case of atelectasis was caused by recurrent cholesteatoma in the middle ear. None of the independent variables studied by univariate or multivariate analysis significantly influenced the success of graft incorporation (Table II). Most surgical complications were related to healing problems of the ear canal skin and neotympanum. Among the complications, there were four cases of small epithelial cysts on the neotympanum, three cases of blunting of the anterior tympanomeatal angle, two cases of persistent middle ear fluid, one case of severe epithelitis (requiring more than 3 mo to heal), and one case of a 15 db PTA sensorineural HL. Results of the postoperative ABG revealed hearing improvement of at least 10 db in 78% (60 cases: 46 primary and 14 revision), worsening in 5% (4 cases: 3 primary and 1 revision), and same hearing in 17% (13 cases: 9 primary and 4 revision). Table III shows preoperative and postoperative ABG values tabulated according to categories. Twenty-seven percent of patients had a postoperative ABG of less than 10 db, and 73% of patients had a postoperative ABG of 20 db or less. One patient had a drop of the bone conduction PTA of 15 db. The mean preoperative ABG of the population of 77 patients was 29.8 (SD 10.2) db, and the postoperative ABG improved to a mean value of 16.5 (SD 10.7) db (t 10.44, P.001, confidence intervals 10.8, 15.9). Table IV shows hearing results for all the nominal variables and the results of the univariate analysis. Postoperative ABG outcomes were generally more favorable in primary than in revision cases (P.0068), in cases of chronic otitis media versus cholesteatoma (P.002), in cases with intact versus short or absent malleus handle (P.001), and in cases with intact versus absent tympanic annulus (P.04). Cases that required a mastoidectomy were associated with poorer hearing outcomes (P.014). Table V shows the relationship between the continuous variables (age and preoperative ABG) and postoperative ABG. Preoperative ABG had a positive correlation with postoperative ABG and accounted for 18% of the variation in postoperative ABG, whereas age showed no significant correlation with the hearing outcome. Multiple regression analyses were then performed to determine whether the independent variables and their interactions accounted for postoperative ABG variance (Tables VI and VII). Postoperative ABG was significantly dependent on the status of the malleus handle and the preoperative ABG, whereas postoperative ABG was independent of type of surgery, age, sex, diagnosis, mastoidectomy, stapes arch status, and tympanic annulus status. Patients with intact malleus handle had smaller postoperative ABG. Patients with greater preoperative ABG also had greater postoperative ABG. When grouping patients into primary and revision surgery, some differences in the individual weight of Air-Bone Gap (decibel) TABLE III. Postoperative Air-Bone Gap of 77 Cases of Lateral Tympanoplasty Tabulated in Bins. Preoperative Level, No. of Ears (%) Postoperative Level, No. of Ears (%) 10 0 (0) 21 (27) (16) 35 (46) (39) 11 (14) (46)* 10 (13) *Subject to rounding error. T3 T4 T5 T6 7 3

4 TABLE IV. Postoperative Air-Bone Gap (ABG) after Lateral Tympanoplasty for Total and Near-Total Perforations: Results of Univariate Analysis of Group Difference on Effect of Nominal Variables by Student s t Test. Variable variables emerged. In primary tympanoplasty, intact malleus handle (but not preoperative hearing) had an effect on postoperative ABG. In revision tympanoplasty, hearing outcome was significantly dependent not only on preoperative hearing but also on mastoidectomy: patients on whom a mastoidectomy was not performed had a greater postoperative ABG. DISCUSSION The most common outcome measures to define success in tympanoplasty are graft incorporation, recreation TABLE V. Results of Bivariate Analysis on Relationship of Continuous Variables (Preoperative ABG and age) on Postoperative Air-Bone Gap (ABG) in 77 Patients. Variable Mean ABG SD Correlation R 2 P Value Hearing Preop ABG 29.8 db * Postop ABG 16.5 db 11 Age (yr) *Significant difference (P.05). R 2 coefficient of determination. n Mean ABG (db) SD (db) P Value Type of surgery Primary Revision * Malleus handle Normal Short or absent * Stapes suprastructure Present Absent Middle ear mucosa Normal Inflamed or absent Diagnosis Chronic otitis media Cholesteatoma * Mastoidectomy None performed Performed * Sex Male Female Tympanic annulus Present Absent * *Significant difference (P.05). SD standard deviation. of an aerated middle ear, disease control, and hearing results. In this study, we have shown that the lateral technique described here effectively addresses the largest possible eardrum defects resulting in an intact neotympanum and an aerated middle ear in 91% of cases. Assuming that all of the four patients lost to follow-up were failures, the overall success rate can be recalculated to 86%. This rate compares favorably with published studies of tympanoplasty for large eardrum defects. A study by Jung and Park 5 using a mediolateral graft method to repair subtotal and anterior perforations demonstrated a 97% rate of eardrum closure and reported complications in 5 cases of 100. In their study, perforations that were located anterior to the malleus and that were less than subtotal were included. Blokmanis and Archibald 6 reported a closure rate of 98% for near-total perforations using a modified House tympanoplasty technique. However, in their study, 14 ears of 49 showed poor aeration of the middle ear space, often with retraction of the intact eardrum. Other complications were not specifically reported. Fishman et al. 8 found an overall success rate of 92% (46/50) in a study evaluating the efficacy of total tympanic membrane reconstruction using either dermis allograft (84% success rate) or temporalis fascia (97% success rate) for the grafting material. Other authors have demonstrated that the size of the perforation is a variable of prognostic significance and that reported rates of eardrum closure in large perforations can be as low as 56%. 1,2 In this study, the success of graft incorporation was independent of demographic, technique, and disease variables. We found that age had no predictive value on graft incorporation. Other authors, however, have reported that outcome is dependent of age and that children had worse outcome than adults. 3,12 Moreover, in this and other studies, 3 graft incorporation was independent of the status of middle ear mucosa, whereas others 2,10,13 have reported that a wet ear is a poor prognostic factor. We speculate that differences in the relative weight of variables on the outcome across studies may occur in part be caused by differences in methodology (univariate vs. multivariate statistics), duration of follow-up, and patient population. Our case population is relatively more homogeneous (i.e., including only cases of large perforations, lateral tympanoplasty, uniform indications for mastoidectomy, and ossiculoplasty) than other series that include perforations of different sizes and varying surgical techniques. In terms of functional outcome, the status of the malleus handle and the preoperative ABG were significantly associated with postoperative ABG. The standardized coefficient values for both of these variables are similar (0.301 for malleus, for preoperative ABG), meaning that they had comparable influence in predicting subsequent postoperative ABG. Our regression model showed that 40% of the variation of postoperative ABG can be accounted for by controlling for all variables. One could intuitively say that the malleus handle is an important factor because it provides stability to the neotympanum and has a key role as an impedance matcher, but it is also possible that cases with absent malleus handle represent a more advanced stage of the disease. The same can be argued for preoperative ABG: worse functional results 4

5 TABLE VI. Results for Multiple Linear Regression Analysis of Postoperative Air-Bone Gap (ABG) in 77 Cases of Lateral Tympanoplasty. Variable Estimate Coefficient SE t Ratio P Intercept * Malleus(0) * Preop ABG * Age Me mucosa(0) Revision(0) Annulus(0) Mastoidectomy(0) Diagnosis(0) Sex(0) Stapes arch(0) Model coefficient of determination (R 2 ) 0.402, root mean square error (RMSE) , P.001. *Statistically significant at P.05. (0) not performed; Me middle ear; SE standard error. in cases of more advanced disease. This suggests that the pathologic condition is probably the most important factor influencing the functional outcome, certainly more important than demographic and technique factors. Albu et al. 2 and Black 9 have also shown the prognostic value of the malleus handle. In contrast, Goldenberg 14 and Brackmann et al. 15 reported that the presence of the malleus handle was not an important prognostic factor of functional outcome; unlike other series that use the malleus handle for ossicular reconstruction when this is available, in this latter series, the columnella techniques (TORP and PORP) are always used regardless of the status of the malleus handle. TABLE VII. Results for Multiple Linear Regression Analyses of Postoperative Air-Bone Gap (ABG) by Type of Surgery (Primary vs. Revision) in 77 Cases of Lateral Tympanoplasty. Variable Estimate SE t Ratio P Primary Tympanoplasty (revision 0) Model R , RMSE , P.001. Intercept Malleus(0) Revision Tympanoplasty (revision 1) Model R , RMSE , P.0458 Intercept Preop ABG Mastoidectomy(0) Only independent variables showing significance (P.05) are shown. (0) not performed; SE standard error; R 2 coefficient of determination; RMSE root mean square error. When grouping our cases by type of surgery, we found a different relative weight for some variables. In primary tympanoplasty, intact malleus handle (but not preoperative hearing) had an effect on postoperative ABG. In revision tympanoplasty, hearing outcome was significantly dependent not only on preoperative hearing but also on mastoidectomy: patients on whom a mastoidectomy was not performed were associated with a greater postoperative ABG. The positive sign in the parameter estimate for mastoidectomy(0) ( mastoidectomy not performed ) indicates that the predicted value of postoperative ABG increases by in patients without mastoidectomy, holding the values of all other variables in the regression equation constant. This latter finding is interesting and supports a role for mastoidectomy in revision cases. Mastoidectomy was performed to eradicate mucosal disease in the mastoid and for increasing the volume of the mastoid to enhance aeration. Although technique variables did not appear to influence functional outcomes when analyzing the entire population, technique variables may have a greater influence in revision tympanoplasty. Also note that the univariate analysis showed that cases with mastoidectomies had worse hearing results, in contrast with the result of the multivariate analysis for revision tympanoplasty. In the multivariate analysis, all of the variables and their interactions are accounted for, as opposed to the inherent limitation of univariate analysis, which is often confounded by the interrelationship among variables themselves. This clearly illustrates the usefulness of the multivariate analysis which, in this study, allowed us to recognize the favorable effect of mastoidectomy in revision tympanoplasty. In conclusion, successful grafting of near-total and total tympanic membrane perforations by the lateral technique occurred in 91% of the cases and was independent of demographic, disease, and technical variables. Preoperative ABG and status of malleus handle had a significant effect on postoperative ABG. In revision tympanoplasty, mastoidectomy is associated with a better functional outcome. 5

6 BIBLIOGRAPHY 1. Lee P, Kelly G, Mills RP. Myringoplasty: does size of the perforation matter? Clin Otolaryngol 2002;27: Albu S, Babighian G, Trabalzini F. Prognostic factors in tympanoplasty. Am J Otol 1998;19: Sade J, Berco E, Brown M, Weinberg J, Avraham S. Myringoplasty in children: short and long term results in a training program. J Laryngol Otol 1981;95: Sheehy JL, Anderson RG. Myringoplasty: a review of 472 cases. Ann Otol Rhinol Laryngol 1980;89: Jung T, Park SK. Mediolateral graft tympanoplasty for anterior or subtotal tympanic membrane perforation. Otolaryngol Head Neck Surg 2005;132: Blokmanis A, Archibald J. Modified house tympanoplasty for successful closure for near-total tympanic membrane perforations. J Otolaryngol 2004;33: Aidonis I, Robertson T, Sismanis A. Cartilage shield tympanoplasty: a reliable technique. Otol Neurotol 2005;26: Fishman A, Marrinan M, Huang T, Kanowitz S. Total tympanic membrane reconstruction: ALLODERM versus temporalis fascia. Otolaryngol Head Neck Surg 2005;132: Black B. Ossiculoplasty prognosis: the Spite method of assessment. Am J Otol 1992;13: Denoyelle F, Roger G, Chauvin P, Garabedian EN. Myringoplasty in children: predictive factors of outcome. Laryngoscope 1999;109: Committee on Hearing and Equilibrium. Committee on Hearing and Equilibrium guidelines for the evaluation of results of treatment of conductive hearing loss. Otolaryngol Head Neck Surg 1995;113: Podoshin L, Fradis M, Malatsakey S, Ben-David J. Type I tympanoplasty in children. Am J Otolaryngol 1996;17: Lau T, Tos M. Tympanoplasty in children: an analysis of late results. Am J Otol 1986;7: Goldenberg RA. Hydroxylapatite ossicular replacement prostheses: preliminary results. Laryngoscope 1990;100: Brackmann DE, Sheehy JL, Luxford WM. TORPs and PORPs tympanoplasty: a review of 1042 operations. Otolaryngol Head Neck Surg 1984;92:

7 JOBNAME: AUTHOR QUERIES PAGE: 1 SESS: 3 OUTPUT: Mon Jul 3 10:09: /rich2/zln lar/zln lar/zln00806/zln5185d06z AUTHOR QUERIES AUTHOR PLEASE ANSWER ALL QUERIES 1 1 Was this presented?