Effect of tympanic perforations on the detection of distortion-product otoacoustic emissions.
The detection of distortion-product otoacoustic emissions (DPOAEs) depends on the viability of the ear's conduction apparatus. However, tympanic membrane perforations and other conductive disorders have not been fully investigated with regard to the examination of DPOAEs. Using the guinea pig model, we made perforations of different sizes and loci on the tympanic membrane and collected DPOAE data for frequencies between 2,193 and 5,508 Hz for each condition. We found that small perforations, up to 25% of the area of the tympanic membrane, still allow us to detect emissions at the specified frequencies. However, perforations of 50% and larger, as well as those accompanied by traumatic perilymph fistulas and ossicular disarticulations, severely interfered with the detection of DPOAEs. We discuss the clinical relevance of these findings with respect to the potential uses of DPOAEs.
Otoacoustic emissions have been the subject of considerable interest [1-5] since they were first reported by Kemp in 1979.  Otoacoustic emissions are the result of an active process in the organ of Corti,  and they can be applied as a useful, objective, and noninvasive test to evaluate the cochlear condition.  The ability to detect otoacoustic emissions is very dependent on the viability of the ear's conduction apparatus.  Therefore, the reliability of otoacoustic emission measurements in the presence of common middle ear conductive disorders, such as tympanic membrane perforations, otosclerosis, and serous otitis media, requires further study.
The effects of myringotomy and pressure-equalizing tubes on evoked otoacoustic emissions have been previously alluded to. [7,8] Studies have found that children with pressure-equalizing tubes exhibit measurable but lower levels of evoked otoacoustic emissions than those in normally hearing children. Therefore, the presence of a small, artificial perforation of the tympanic membrane does not preclude the testing of otoacoustic emissions. [7,8] The effects of larger perforations have not been systematically investigated.
Our study was conducted in order to fully investigate the effects of tympanic membrane perforations on a particular type of evoked otoacoustic emissions: distortion-product otoacoustic emissions (DPOAEs). Using the guinea pig as a controlled model, we made perforations of various sizes to determine how much perforation would still allow for the detection of DPOAEs. Then we evaluated the locus of these perforations for their effect on DPOAEs. In order to simulate tympanic membrane perforations as they occur in traumatic cases, we also created perforations accompanied by perilymph fistulas and ossicular disarticulations to study their impact on DPOAE detection.
Materials and methods
Subjects and surgical preparation. We performed experiments on 18 adult, Hartley albino guinea pigs weighing between 600 and 700 grams. All procedures were performed under sterile conditions and in accordance with our Institutional Animal Care and Use Committee's guidelines for animal research. Preoperative anesthesia, consisting of ketamine (35 mg/kg) and xylazine (2 mg/kg), was administered intramuscularly. The animals were placed on a heating pad and maintained on a mixture of [O.sub.2] and 1 to 2% isoflurane. Heart rate and oxygen saturation by pulse oximeter  were monitored throughout the procedure.
A postauricular incision approximately 5 cm in length was made, and the ear was reflected anteriorly and held by retractors as in a standard tympanoplasty procedure in order to obtain maximum exposure of the tympanic membrane. The surgeon took care to ensure that the tympanic membrane remained free of blood, cartilage, and other debris that might affect testing.
Distortion-product measurements. Otoacoustic measurements were obtained with a Grason-Stadler GS1 60 DPOAE system using the conventional 2f1-f2 method. Two primary tones (f1 and f2) generated by separated synthesizers were delivered to emitters contained within an ear probe. DPOAEs were recorded by a microphone inside the probe and processed by a spectrum analyzer. The frequencies and sound pressure levels (dB SPL) of the primary tones used for testing are shown in the table. Geometric means of f2/f1 were fixed at 1.2.
Procedure and experimental design. Following surgical exposure of the tympanic membrane, an otoacoustic probe was held securely in the ear canal to record baseline DPOAE levels from the intact tympanic membrane. A check for a proper acoustic seal was performed prior to each test to ensure that outside noise did not contaminate the results. DPOAE amplitudes and their associated noise floors were recorded as "DP-grams" and saved for later analysis.
Next, a myringotomy was performed in the antero- or posterosuperior quadrant of the tympanic membrane. After repeat testing, four subsequent perforations of increasing size (25, 50, 75, and 100% of the area of the membrane) were made (figure 1). Otoacoustic measurements were taken after each successive perforation. The final perforation resulted in the complete separation of the tympanic membrane from the malleus handle and effectively represented an intact ossicular chain. Care was taken to prevent ossicular injury during the enlargements of the perforations.
A similar protocol was followed to measure DPOAEs in the presence of traumatic perilymph fistulas and Ossicular disarticulations. In order to simulate a traumatic perilymph fistula, a grade I perforation was created in the posterosuperior quadrant of the tympanic membrane. This perforation allowed for a direct visualization of the round window. A perilymph fistula was surgically created by lacerating the round window with a fine-gauge needle. Again, otoacoustic measurements were taken after each manipulation.
To simulate a traumatic ossicular disarticulation, a grade I perforation was made in the posterosuperior quadrant of the tympanic membrane, which allowed for the introduction of an instrument to disarticulate the ossicular chain. Otoacoustic measurements were obtained after each manipulation. Postmortem dissections were performed on each ear under a microscope to confirm the perilymph fistulas and ossicular disarticulations.
A Zeiss operating microscope with an attached videocamera was used to record the intact and perforated conditions; a microruler was placed alongside the tympanic membrane, the perforation, and the malleus handle for calibration. This photographic documentation later provided assistance in estimating the size of each perforation.
Data analysis and control conditions. We examined the effects of perforation, traumatic perilymph fistula, and ossicular disarticulation on DPOAEs with a repeated-measure analysis-of-variance statistic. In order to establish whether DPOAEs were detectable in each perforation condition, we adopted a criterion threshold of 3 dB above the noise floor. 
Measurements taken from the intact tympanic membranes served as the control for each ear. We developed an exclusion criteria to control for ears that did not generate DPOAEs at baseline: Each ear tested must have had generated a baseline DPOAE above the criterion threshold at four of five frequencies tested after the animals had been anesthetized and surgical exposure of the intact tympanic membrane had been achieved.
Complete data sets, including DPOAE measurements at all perforation conditions, were obtained for 12 ears. Five additional ears provided DPOAE data concerning traumatic perilymph fistulas, and four ears provided DPOAE data on ossicular disarticulations. Among the reasons that the other ears were excluded from the study were a failure to establish baseline readings, respiratory failure of the animal while under anesthesia, or an error in the size of the perforation. In the case of ossicular disarticulation, any evidence of incomplete disarticulation at the postmortem dissection mandated exclusion from the study.
Effects of operating room noise. Normal intraoperative noise must be considered when recording DPOAE measurements. To demonstrate the effects of operating room noise on DPOAE measurements, two subjects were administered ketamine, moved to a sound-attenuated booth, and subjected to DPOAE testing from 500 to 8,000 Hz. The same animals were then moved to the operating room and testing was repeated. The deleterious effects of normal operating room noise on DPOAE measurements were evident at frequencies lower than 2,000 Hz. The equipment we used in this study did not test at frequencies below 2,000 Hz, the point at which low-frequency background noise in the operating room overcame the noise-rejection ability of the recording system.  For this reason, testing on all animals was restricted to the frequencies above 2,000 Hz (table), the point where signal-to-noise relationships were appropriate for the intraoperative measurements of DPOAEs.
The size of the perforation was also inversely related to the level of the noise floors: The larger the perforation, the lower the noise floor.
Myringotomy. DPOAEs in the myringotomy condition did not differ significantly from DPOAEs in the intact condition (figure 2). Mean emissions were generated at amplitudes well above the criterion threshold for detection (figure 3).
25% Perforation. At each frequency tested, the DPOAEs in the 25% perforation condition were significantly lower than those in the intact membranes (p[less than]0.05) (figure 2). However, the reduction in mean DPOAE amplitude was not large enough to prevent the detection of otoacoustic emissions at levels above the criterion threshold for detection (figure 3).
50% Perforation. The 50% perforation condition severely affected DPOAE results (figure 2). At each frequency tested, they were significantly lower than those in the intact condition (p[less than]0.05). Detectable emissions were no longer consistently generated at frequencies where they had been present in the intact condition (figure 3).
75 and 100% Perforations. The 75% perforation condition affected testing so severely that the DPOAEs did not differ significantly from their own noise floors (p[greater than]0.05) (figure 2). Results were similar in the 100% perforation condition, with the exception of a consistent 3,500-Hz emission. Mean DPOAEs were below the criterion threshold for all but this one frequency in the 100% condition (figure 3).
Locus of perforation. The locus of the myringotomy in either the antero- or posterosuperior quadrant of the tympanic membrane had no effect on the detection of emissions (p[greater than]0.05) (figure 4). Neither did the locus of the 25% perforation in either the anterior or posterior quadrant (p[greater than]0.05) (figure 4).
Traumatic perilymph fistula. The presence of a perilymph fistula in addition to a grade I perforation severely affected DPOAE measurements at the frequencies tested. Mean DPOAEs recorded in this condition did not differ significantly from the associated noise floor (p[greater than]0.05)
Traumatic ossicular disarticulation. Ossicular disarticulation also proved to be severely detrimental with respect to the detection of otoacoustic emissions. Mean DPOAEs recorded in the ossicular disarticulation condition did not differ significantly from the associated noise floor (p[greater than]0.05).
There were significant differences between the condition of the perforation alone and the condition of the perforation with an accompanying traumatic perilymph fistula (p[less than]0.05) or ossicular disarticulation (p[less than]0.01).
The detection of DPOAEs depends on several factors intrinsic to the auditory pathway. First, a functional cochlea with intact outer hair cells is required to generate DPOAEs. However, DPOAE detection is not independent of the ear's conduction apparatus. Because DPOAEs generated in the cochlea are conducted through the middle ear, their measurements can be contaminated by middle ear dysfunction. The forward acoustic stimulus used to evoke an emission can also be reduced by faulty middle ear transduction. It is likely that both factors play a role in altering DPOAE measurements.
It might not be possible to estimate the level of DPOAE attenuation imparted by a conductive impairment solely on the air-bone gap.  Studying the effects of conductive impairment on evoked otoacoustic emissions might be necessary to serve as a guide to their reliability. Several studies have investigated the effects of conductive impairments on detecting evoked otoacoustic emissions. Zurek et al showed in the chinchilla that dislocating the ossicular chain resulted in a reduction of DPOAE amplitude below the limits of the measuring system used in their study.  However, both Owens et al  and Amedee  found that the presence of a small artificial perforation, specifically one made by a pressure-equalizing tube, resulted in recordable but lower evoked emissions than those obtained from normally hearing children. It is clear that otoacoustic emission measurements must be carefully interpreted in the face of conductive impairments until the effects of those impairments on detection are fully elucidat ed.
Our results show that the presence of a tympanic membrane perforation does not necessarily preclude the detection of DPOAEs in an animal model. Perforation size had a direct relationship to DPOAE amplitude, as amplitudes became lower when perforations became larger. We found that DPOAEs could still be detected in the presence of perforations as large as 25% of the pars tensa at the specified frequencies in the guinea pig, unless they were accompanied by additional trauma in the form of perilymph fistula or ossicular discontinuity; in that case, DPOAEs were not detectable. Perforations larger than 50% severely interfered with DPOAE detection.
Intraoperative testing in the frequencies below 2,000 Hz was not possible with the DPOAE system we used in this study. This effect was most likely the result of the low-frequency noise in the operating room, which interfered with the processing system's ability to distinguish DPOAEs from ambient noise. This is supported by the fact that two animals had recordable emissions in the low frequencies when they were placed in a sound-attenuated booth, but not when they were in the operating room environment. These findings are consistent with intraoperative human studies, which found that despite the strictest approaches to noise rejection, DPOAEs below 2,500 Hz could not be reliably monitored because of the high levels of background noise. 
In our study, venting the middle ear through a dry perforation resulted in a decrease in noise floors in the intact condition. The noise floors continued to decrease as the size of the perforation increased. This is not surprising considering the fact that perforating the tympanic membrane allows access to the middle ear space, which increases the volume and therefore decreases the sound pressure level of the closed system from which the noise floor was measured. Owens et al found that pressure-equalizing tubes, although they allowed access to the middle ear space and therefore increased the volume from which the noise floor was measured, resulted in higher noise floors than those seen in normally hearing children. 
Studies have shown that intraoperative monitoring with brainstem auditory-evoked responses is possible during surgery for conductive hearing loss, and intraoperative changes in those responses might be useful in predicting postoperative results.  It is possible that intraoperative DPOAE testing could be of value during surgery for conductive hearing loss in a similar manner. Our study shows that intraoperative DPOAE testing is possible during the manipulation of the tympanic membrane, and that it is acutely sensitive to varying degrees of conductive impairment induced surgically. This might have significance for otologic surgery. Further studies are warranted to assess the possibility of DPOAE monitoring during ossicular reconstruction and stapedectomy.
From the Division of Otolaryngology, Department of Surgery, University of Mississippi Medical Center, Jackson.
(1.) Whitehead ML, Lonsbury-Martin BL, Martin GK. Evidence fortwo discrete sources of 2f1-f2 distortion-product otoacoustic emission in rabbit: I. Differential dependence on stimulus parameters. J Acoust Soc Am 1992;91:1587-607.
(2.) Gaskill SA, Brown AM. The behavior of the acoustic distortion product, 2f1-f2, from the human ear and its relation to auditory sensitivity. J Acoust Soc Am 1990;88:821-39.
(3.) Lonsbury-Martin BL, Martin GK. The clinical utility of distortion-product otoacoustic emissions. Ear Bear 1990;l1:144-54.
(4.) Martin GK, Ohlms LA, Franklin DJ, et al. Distortion product emissions in humans. III. Influence of sensorineural hearing loss. Ann Otol Rhinol Laryngol Suppl 1990;147:30-42.
(5.) Harris FP, Lonsbury-Martin BL, Stagner BB, et al. Acoustic distortion products in humans: Systematic changes in amplitudes as a function of f2/f1 ratio. J Acoust Soc Am 1989;85: 220-9.
(6.) Kemp DT. Evidence of mechanical nonlinearity and frequency selective wave amplification in the cochlea. Arch Otorhinolaryngol 1979;224:37-45.
(7.) Owens JJ, McCoy MJ, Lonsbury-Martin BL, Martin GK. Otoacoustic emissions in children with normal ears, middle ear dysfunction, and ventilating tubes. Am J Otol 1993;14: 34-40.
(8.) Amedee RG. The effects of chronic otitis media with effusion on the measurement of transiently evoked otoacoustic emissions. Laryngoscope 1995;105:589-95.
(9.) Avan P, Loth D, Menguy C, Teyssou M. Evoked otoacoustic emissions in guinea pig: Basic characteristics. Hear Res 1990;44:151-60.
(10.) Telischi FF, Widick MP, Lonsbury-Martin BL, McCoy MJ. Monitoring cochlear function intraopcratively using distortion product otoacoustic emissions. Am I Otol 1995;16:597-608.
(11.) Sininger YS. Clinical applications of otoacoustic emissions. In: Myers EN, ed. Advances in Otolaryngology-Head and Neck Surgery. Vol 7. St. Louis: Mosby Year-Book, 1993: 247-69.
(12.) Zurek PM, Clark WW, Kim DO. The behavior of acoustic distortion products in the ear canals of chinchillas with normal or damaged ears. J Acoust Soc Am 1982;72:774-80.
(13.) Selesnick SH, Victor JD, Tikoo RK, Eisenman DJ. Predictive value of intraoperative brainstem auditory evoked responses in surgery for conductive hearing losses. Am J Otol 1997; 18:2-9.