Received:February 19, 2001Accepted:May 19, 2001

Stavros HatzopoulosUniversity of FerraraDepartment of Audiology and Centre of Bioacoustics203 Corso Giovecca, Ferrara 44100, Italy

Original Article

Audiology 2001; 40:253–264

Cisplatin Ototoxicity in the Sprague Dawley

Rat Evaluated by Distortion Product

Otoacoustic Emissions

Ototoxicidad por Cisplatino en la rata Sprague Dawley evaluada

mediante productos de distorsión de las emisiones otoacústicas

Stavros Hatzopoulos*M. Di Stefano*K. C. M. Campbell†

D. Falgione‡

D. Ricci §

M. Rosignoli*M. Finesso$

A. Albertin*M. Previati §

S. Capitani §

A. Martini*

*Department of Audiologyand Centre of Bioacoustics,University of Ferrara, Italy

†Department of Surgery, Audiology,Southern Illinois UniversitySchool of Medicine, Springfield,Illinois, USA

‡Department of Biology, Sectionof General Physiology,University of Ferrara, Italy

§Department of Morphology andEmbryology, Section ofHuman Anatomy, Universityof Ferrara, Italy

$Fidia Research Laboratories,Abano Terme, Padova, Italy

Key WordsDistortion product otoacousticemissions, cisplatin, ototoxicity

Introduction

When the cochlea is stimulated simultaneously withtwo pure tones at frequencies f1 and f2 (with f2 > f1), itgenerates additional frequency components called inter-modulation distortion tones. These are not present in theoriginal stimuli and are viewed as products of two-toneinteractions caused by the non-linearity of the cochlea.1

The frequencies of these distortion tones or productsare called primaries f1 and f2. For two given primaries,numerous families of distortion products (distortion pro-duct otoacoustic emissions (DPOAEs)) exist, but in themammalian ear the cubic distortion product 2f1–f2

is the most prominent one.1,2 There is evidence thatthe 2f1–f2 product is generated by the outer hair cells

(OHCs),2–5 and that it is sensitive even to small cochlearalterations.6, 7

The methods available for DPOAE testing combinestimulus and frequency parameters. In the DP-grammodality, the DPOAE responses are recorded using fixedintensities of the tonal stimuli (L1, L2) and a fixed fre-quency ratio (f2/f1) at increasing f2 frequency values(1.5, 2.0, 2.5, 3.0 kHz). In the input–output curve (I/O)modality, the DPOAE responses are recorded at a fixed f2

frequency and frequency ratio, while the level of the tonalstimuli (L1 and L2) is increased in systematic steps (30, 35,40 dB SPL). For the I/O curve estimation the primarylevels (L1 and L2) can be equal, but often in clinicalpractice unequal levels are employed (e.g. L2�L1 + 8 dBSPL).

AbstractThe present study has evaluated the use ofdistortion product otoacoustic emission (DPOAE)responses in the detection of cisplatin-inducedototoxicity in a Sprague Dawley rat animalmodel. The cisplatin was administered as a 16mg/kg, dose introduced by a slow 30-minintraperitoneal infusion. Data from three DP-gram protocols, DPOAE input–output responsesat 8 kHz, and auditory brainstem responses(ABRs) at 8, 12 and 16 kHz were collected beforeand 72 h after treatment. The post-treatmentABRs at 16 kHz showed the greatest meanthreshold shift of 33.6 dB. The post-treatmentDP-gram data showed significant reduction of thesignal to noise ratios in the majority of thefrequencies tested, across all tested protocols. Thedata suggest that the most sensitive DPOAEprocedure for the early detection of the cisplatin-induced ototoxic damage is the DPOAE I/Oprotocol. Morphological analyses indicated thatthe inner hair cells remained intact, while severaltypes of alterations were observed in the arrange-ment of the stereocilia in the outer hair cells.

SumarioEste estudio evaluó el uso de los productos dedistorsión de las emisiones otoacústicas (DPOAE)en la detección de daño por ototoxicidad inducidapor Cisplatino en el modelo animal, rata SpragueDawley. Se administró Cisplatino a 16 mg/kg porinfusión intraperitoneal lenta durante 30 minutos.Se colectaron resultados de tres protocolos deDPgramas, respuestas de entrada/salida (I/O) a 8kHz y respuestas de tallo cerebral (ABRs) a 8, 12y 16 kHz; antes y 72 hrs. después del tratamiento.Las respuestas ABR a 16 kHz post-tratamientomostraron el mayor cambio de umbral, enpromedio de 33.6 dB. El DPgrama post-trata-miento mostró una reducción significativa de ladiferencia señal/ruido en la mayoría de lasfrecuencias examinadas, de todos los protocolosrealizados. Los datos sugieren que el procedi-miento más sensible para detectar el daño porototoxicidad inducida por Cisplatino es la funciónI/O DPOAE. El análisis morfológico indicó quelas células ciliadas internas quedaron intactas,mientras que varios tipos de alteraciones seobservaron en la disposición de los estereociliosde las células ciliadas externas.

254 Audiology, Volume 40 Number 5

Cisplatin is a very powerful antineoplastic drug used inthe treatment of various types of cancer and presentsmultiple dose-dependent side-effects,8 including nephro-toxicity and ototoxicity. While in clinical practice theincidence of nephrotoxicity is decreased by use of diureticagents and hydration,9 the ototoxicity and the subsequenthearing loss are unaffected by these treatments.10 Variousstudies have shown that the hearing loss induced affectsmainly the high frequencies, both in humans11 and inanimals.8,10,12–14 It has been demonstrated that there is ahigh correlation between lesions in the inner ear andcisplatin dosage.15 The mechanism by which cisplatindamages the cochlea is not fully understood, but datafrom histological analyses have suggested that: (1) theOHCs of the organ of Corti and the stria vascularis areinvolved;12,15–20 and (2) there is a decrease in the level ofcochlear glutathione (GSH) and formation of reactiveoxygen species.21–24

To assess the ototoxicity of cisplatin in an animalmodel, it is common to use electrophysiological tech-niques such as electrocochleography (ECochG)14,25,26 andauditory brainstem responses (ABRs).19,21 These tech-niques record global responses from the cochlea or fromthe afferent auditory nerve fibres, but do not allowa direct evaluation of the subtle alterations of thefunctional status of the OHCs, which occur after theadministration of cisplatin. For the latter it is assumedthat the DPOAEs, which are considered to be responsesmediated by the OHCs, might provide a useful tool fordetecting any induced functional changes of the cochlearstructure.

The main objective of this investigation was theemployment of a robust animal (rat) model in the earlydetection of cisplatin ototoxicity. Analytically:

1. We have chosen to evaluate DPOAE-based procedures(ABRs were considered as the gold standard) of earlycisplatin ototoxicity detection, with a rat model becausethe rat is currently the most commonly used animal forstudying cisplatin ototoxicity. Such an animal modeldemonstrates marked hearing loss after the admini-stration of high dosages of cisplatin with a very lowmortality rate.19,27 In addition, its high resistance tomiddle ear infections and lower sensitivity to multipledoses of anaesthesia28 favour the choice of the rat modelfor an ototoxic study, rather than other models usingthe guinea pig or the chinchilla. A number of Americanstudies19,21,29 have used the Wistar rat strain for theearly detection of cisplatin ototoxicity. Because theSprague Dawley (SD) strain is very popular in Euro-pean biomedical research and in order to providecomparable ototoxicity data between rat strains, wehave investigated the performance of an (SD) rat modelusing a cisplatin ototoxicity dosage of 16 mg/kg infusedover 30 min.

2. Although the cubic distortion products have beenemployed as a method for the early detection ofototoxicity in various animal models,27,30,31 until nowno study has compared DP-grams and I/O functionsto determine which is the more sensitive and specificmethod for using DPOAEs in the early detection ofototoxicity.

The emphasis of the present work is placed mainly on aDPOAE analysis, since previous papers by some of theauthors19 have investigated in detail the ABR alterationsafter cisplatin administration.

Materials and methods

ChemicalsThe cisplatin solution (Platinex, concentration 0.5

mg/ml in normal saline) was purchased from Bristol-Myers Squibb, Italy. For the animal anaesthesia, an equal-volume combination of ketamine hydrochloride (Ketavet,100 mg/ml), xylazine (Rompun, 20 mg/ml) and saline wasused in dosages of 1 ml/kg of body weight. The anaestheticwas administered in two consecutive phases. In phase one,the animal received an intraperitoneal dose, and upon thefirst signs of muscular relaxation an additional half-dosevolume was administered subcutaneously (phase 2).

AnimalsTwenty male SD rats (weight 145–170 g) obtained from

Charles River Italy were divided into two groups. GroupG1 (15 rats) was treated with 16.0 mg/kg of cisplatin.Group G2 (five rats) received an equivalent volume ofsaline and was used as a control. The cisplatin wasadministered by an intraperitoneal slow infusion (post-anaesthesia) of about 30 min using a micro-pump fromHarvard Apparatus. The ABR and DPOAE responseswere recorded during anaesthesia, before and 72 h afterthe cisplatin administration.

The animals were treated according to the Italianguidelines DL 116/92 with reference to EEC directiven. 86-609.

Distortion product otoacoustic emissions recordingsThe distortion product otoacoustic emissions were

recorded in a sound-treated cabin with the VIRTUAL 330equipment. The frequency bandwidth of the DPOAEresponses was set to 4.0–8.0 kHz (referenced to f2) and 12points were sampled per octave. The primary tone ratiof2/f1 was set to 1.21. Each record was the average of 32responses with a noise tolerance of 10 dB SPL. Theresponses were evoked by a DPOAE protocol usingunequal primary tone stimulus intensities, i.e. L1 > L2.Such protocols are generally considered a better choicefor the identification of cochlear dysfunction.32–36 Theprotocols used were: high level (L1�75 and L2�65);

255Hatzopoulos/Di Stefano/Campbell/Falgione/Ricci/Rosignoli/Finesso/Albertin/Previati/Capitani/Martini

Cisplatin Ototoxicity in the SpragueDawley Rat Evaluated by DPOAE

medium level (L1�65 and L2�57) and low level (L1�55and L2�49 dB SPL) .

The I/O data were collected at f2�8.0 kHz, which is thehighest frequency allowed by the VIRTUAL 330 software.The responses were evoked by stimuli of 30–75 dB SPL in5-dB increments, using an asymmetrical stimulus intensityscheme of L2�L1�8 dB SPL. A DPOAE response wasconsidered to be present when the DPOAE amplitude wasat least 3 dB SPL above the noise level. To identify thestimulus level corresponding to the DPOAE threshold inthe I/O measurements, we used the following criterion,37

according to which three consecutive DPOAE responses(corresponding to three consecutive stimulus levels) mustbe in an increasing sequence (i.e. the correspondingDPOAE amplitudes must be also be increasing) beforethe primary-tone level which elicited the third DPOAEresponse is taken as the detection threshold. The noisefloor level in the VIRTUAL 330 is measured at 70–80 Hzbelow and above the DP 2f1–f2 frequency.

In order to record a DPOAE response from theanaesthetised rat, the animal was placed on a stereotaxicdevice supporting a plastic funnel (total length: 22 mm;large inner diameter, 4.5 mm; small inner diameter 2.5mm) which was inserted into the rat’s external acousticmeatus. Once inserted inside the funnel, the DPOAEprobe was positioned firmly by multiple layers of para-film.

Auditory brainstem responsesThe ABRs were recorded by three stainless steel needle

electrodes placed subdermally over the vertex (positive),ipsilateral mastoid (negative) and dorsum area (com-mon/reference). The responses were generated by tone-pipstimuli with a decreasing intensity from 100 to 40 dB SPLin steps of 10 dB, at 8, 12 and 16 kHz (10-ms plateau, 1.0-ms rise–fall time for 8-kHz burst and 0.5-ms rise–fall timefor 12- and 16-kHz bursts), amplified 20,000 times andfiltered from 20 to 5,000 Hz. The stimulus level wasmeasured in free field 4 cm from the sound transducerby a probe microphone (Brùel and Kjaer 4186). Thisprocedure resulted in a reliable calibration for all testfrequencies up to 16 kHz.

The stimuli were presented at the rate of 11/s usinga Motorola tweeter (SH # 13245) placed 4 cm from thetested (right) ear. Each recording was the averageof 250–1000 individual responses. All responses wererecorded by an EWP Biolab apparatus inside a sound-treated cabin. The criterion for the presence of an ABRwas a visual identification of wave III, according toBourre et al.38 Several recordings were acquired atthreshold level. No responses were present below thestimulus level of 40 dB SPL (pretreated animals), whichwas considered to be the threshold level for our experi-mental setup. Latencies were estimated for each ABRwave I and wave III.

Throughout the recordings, the body temperature ofeach animal was maintained at a 37 ± 0.5°C by a HarvardApparatus homeothermic blanket.

Morphological analysisAfter the electrophysiological and acoustical measure-

ments, a random number of 10 animals (4 controls and 6treated) was chosen for an exploratory scanning electronmicroscopy (SEM) histological investigation. Each ratwas perfused intra-aortically, under deep anaesthesia,with 2% glutharaldehyde in 0.1 M sodium cacodylate–HCl buffer, pH 7.2, for 1 h. The temporal bones wereremoved from the animals by exposing the otic capsule.The cochleas were fixed by local perfusion through theopened round and oval windows and holes were drilled atthe cochlear apex. For the post-fixation a 0.5% aqueoussolution of osmium tetroxide was used for 1 h andwashed again in 0.1 M sodium cacodylate–HCl buffer,pH 7.2. The specimens were dehydrated in a graded seriesof ethanol and subjected to critical-point drying (BalzersCPD 030). The dried samples were mounted on alumenstuds, sputter-coated with gold (EDWARDS Sputtercoating S 150) and then examined by SEM (CambridgeS-360, Cambridge Technology Inc., Cambridge, MA,USA).

Since the ABR and the DPOAE equipment testedrelatively low frequencies, in terms of the hearing acuityrange of the rat,39 the morphological observations refer tostructural data from the middle cochlear coil (turn).

Statistical methodsThe DPOAE 2f1–f2 recordings were analysed as signal-

to-noise ratios. The rationale for such a choice is thatbetween the pre- and post-recording sessions, the DPOAEamplitude and the noise floor level values varied. It wasdecided to use as a point of reference the noise levelduring each recording, so that the resulting signal-to-noise ratio could reflect in a more efficient way theinduced alterations in the DPOAE response.

The ABRs were mainly analysed in terms of thresholdshift and changes in the latency of waves I and III. Forthe DP-gram and ABR measurements, an analysis ofcovariance (ANCOVA) was performed with weight as acovariate.

The DPOAE and ABR post-treatment data sets werecompared to their corresponding pretreatment datasets using a Wilcoxon matched-pair test for dependentsamples. The data of the DPOAE I/O curves and theABR latencies, by stimulus level, were fitted to linearregression models (assuming that the probability of type Ierror was not high) in order to define possible changesbetween the slopes of pre- and post-treatment data. Thestatistical significance of the changes in the regressionslopes of the ABR latencies was evaluated by a Wilcoxonmatched-pair test.

256 Audiology, Volume 40 Number 5

The statistical analysis was carried out using SPSSversion 6.1 software. The criterion for statistical signi-ficance for all measures was p<0.05.

Results

DPOAE resultsDP-GRAM

All animals were alive 72 h after the cisplatin admini-stration, and only one rat (from group G1) died duringthe post-treatment anaesthesia. The pretreatment data ofthe G1 group were compared with those from the controlgroup G2 across all f2 frequencies in the DP-grams andacross all intensity levels of the I/O curves. No significantdifferences were found between the signal-to-noise ratiosof the DP-grams, suggesting that the pre-test data fromthe G1 and G2 groups were homogeneous. Because bothbaseline and post-treatment data were obtained, eachsubject served as its own control in addition to theseparate saline-injected control group. A correlationanalysis between the signal-to-noise ratios and weightindicated a significant correlation coefficient (Pearson r�0.40, 2-tailed p<0.001). Therefore, in the exploratoryANCOVA analysis of the DP-gram data in terms ofgroup, treatment, f2 frequency and protocol, the weightwas introduced as a covariate. The ANCOVA analysisindicated that the protocol, treatment, group and f2

frequency factors significantly affect the DPOAE signal-to-noise ratios. The model employed explained approxi-mately 67% of the variance. It was assumed that theunexplained 33% was caused by within-subject differencesand by variables related to the recording method(equipment, probe, cone, distance from tympanic mem-brane, etc.).

A comparison of the pre- and post-treatment DPOAEsignal-to-noise ratios showed a decrease of the post-treatment responses for the majority of the f2 frequenciestested, but the exact pattern was protocol-dependent.Figure 1 shows the means and standard deviations of thepre- and post-treatment signal-to-noise ratios from groupG1 across the three tested DP-gram protocols. The f2

frequencies showing significant pre–post-treatment differ-ences (p<0.05) are indicated with an asterisk.

The post-treatment data from the low-level protocol(L1/L2�55/49) showed significant changes across six(Figure 1, top graph) frequencies. The data showed thatthe significant effects were within the frequency range off2�5.34–6.73 kHz. For the medium level protocol (L1/L2

�65/57), it was observed that nine f2 frequencies werestatistically affected in the 4.49–8.0-kHz range, but thedistribution of the affected frequencies (effective band-width) was different from that of the low-level protocol.The DPOAE responses from the high-level protocol(L1/L2�75/65) shown in Figure 1 (bottom graph) weresignificantly affected at six f2 frequencies in the range

5.04–7.56 kHz. The common denominator in all thesemeasurements is the fact that a relatively large frequencyrange shows significant alterations after the admini-stration of cisplatin.

DP I/O CURVES

The analysis of the I/O curves in Figure 2, for the f2

frequency of 8 kHz showed two main trends: (1) the post-treatment signal-to-noise ratios presented a different slopepattern in comparison to the pretreatment data; (2) thepost-treatment data showed a significant decrease of theDPOAE signal-to-noise ratios in respect of the pre-treatment data.

The percentage of treated animals (group G1) showingan acceptable DPOAE response (> � 3 dB SPL) and anacceptable threshold is shown in Figures 3A, and 3B. Thepresented data can be summarised as follows: (1) thetreated animals gave responses, satisfying the set criteria,only at higher stimulus intensities, i.e. in the interval 65–75dB; (2) only 40% of the treated animals showed acceptableresponses in the stimulus range from 50 to 55 dB SPL; (3)at the lower level L1�30 dB, no treated animal showed anacceptable response. It should be noted, however, that twocontrol animals did not show acceptable signal-to-noiseratios at this stimulus level; and (4) the post-treatmentthreshold range, shown in Figure 3B, demonstrates valuesbetween 40 and 70 dB SPL with data from eight animals(57%) between 45 and 60 dB SPL. The distribution of thethreshold-shift values is shown in Figure 3C, whichindicates that the range of the threshold elevation isbetween 0 and 30 dB (mean 13.19 dB ± 2.6 dB).

To evaluate the slope differences between the pre- andthe post-treatment I/O data, a more detailed regressionanalysis was performed, and the results are shown inFigure 4. A regression control curve (solid line ± onestandard deviation) was calculated from pre-treatmentresponses. Interpolating the regression curve with the x-axis (assuming that in this segment the fit was linear), itwas possible to estimate the minimum amplitude levelthat could evoke a DPOAE response with a signal-to-noise ratio of 3 dB. The mean stimulus value for the pre-treated animals (all responses were pooled together) wasestimated to be L1�23 dB SPL. An exploratory clusteranalysis of the G1 post-treatment responses suggested thepresence of two subgroups in the data set, different fromthe pretreatment responses in terms of slope and signal-to-noise ratios. Each subgroup is represented in Figure 4by a separate regression curve. The curve of the firstsubgroup had a steeper slope (beginning approximatelyfrom the stimulus level of 45 dB SPL) than the referenceregression curve of the untreated animals. By usinginterpolation, the minimum mean stimulus level to elicita 3-dB response was estimated to be 37.3 dB, 14.3 dBhigher (37.3�23) than the pretreatment value. Usinginterpolation on the data from the second subgroup, it

257Hatzopoulos/Di Stefano/Campbell/Falgione/Ricci/Rosignoli/Finesso/Albertin/Previati/Capitani/Martini

Cisplatin Ototoxicity in the SpragueDawley Rat Evaluated by DPOAE

was found that the minimum mean stimulus to evoke a 3-dB response was equal to 56.3 dB, 33.3 dB higher (56.3�23) than the pre-treatment mean.

ABR resultsIn the post-treatment responses, the range of the

threshold shifts depended on the stimulus frequency. The

highest shift was observed at 16 kHz (mean 33.6 ± 3.6 dB)and the lowest at 8 kHz (mean 17.2 ± 2.4 dB). Thedistribution of threshold shift per tested frequency isshown in Figure 5.

The wave I and III latencies from the post-treatmentresponses evoked by the 8-, 12- and 16-kHz tone burstswere found to be significantly increased with respect

Figure 1. Mean signal-to-noise histograms of the DP-gram responses of theG1 group (treated with cisplatin). The responses were evoked by a low-levelprotocol L1/L2�55/49 (top graph), by a medium-level protocol L1/L2�65/57(middle graph), and by a high-level protocol L1/L2�75/65 (bottom graph). Themeans are expressed in dB and are plotted in the frequency range 4–8 kHz(reference to f2). The asterisks indicate statistically significant differences betweenpre- and post-treatment data. The bars over the histogram values correspond toone standard deviation.

258 Audiology, Volume 40 Number 5

to the pretreatment values in both treated groups. Aregression analysis was performed only on latency valuesof wave III to evaluate the presence of significant slopechanges by stimulation frequency. It was not possible toestimate regression curves for the wave I responses,because the number of post-treatment responses was toolow. The curves generated from the fit of the latency datato the regression models are shown in Figure 6. Thecomparison between pre- and post-treatment latenciesper group resulted in significant differences across allfrequencies. The post-treatment regression curves showedlarge slope differences at 8 kHz and small slope differ-ences at 12 kHz.

Relationship between DPOAE and ABR thresholdsTo evaluate the similarity of the electrophysiological

measurements with the acoustical measurements, wecompared the ABR and the DPOAE I/O data evoked by8-kHz stimuli. The ABR data were used as predictors ofthe DPOAE I/O values. Linear and quadratic regressionmodels were applied. The most significant relationshipwas observed between the DPOAE and ABR values at 8kHz. The quadratic model (r2�0.78), shown in Figure 7,outperformed the linear model (r2�0.58), but wasinfluenced by data from one animal. Removing those datavalues, the r2 of the quadratic model was reduced to 0.62.The relationship between the DPOAE thresholds at 8 kHzand the ABR thresholds at 12 and 16 kHz was notsignificant across all models, with r2 values ranging from0.36 to 0.03.

SEM findingsThe control cochleas, such as that shown in Figure 8,

presented a regular layout of three rows of OHCs. Their

Figure 2. Mean (+1 standard error mean) of the DPOAEsignal-to-noise ratios in the I/O curves at f2�8 kHz. The topand bottom curves show the pre- and post-treatment datarespectively. The x- and y-axes represent the amplitude of thestimulus in dB SPL and the signal-to-noise values in dBrespectively. The post-treatment data show a significant reductionof the DPOAE responses. This effect is most prominent at lowerstimulus intensities of 35–50 dB SPL.

Figure 3. Post-treatment DPOAE responses from group G1.(A) Distribution of the percentage of the post-treatment DPOAEI/O responses satisfying the 3-dB criterion (see the methodssection), at various stimulus intensities. The x-axis shows stimulusintensity in terms of L1. (B) Threshold distribution of the post-treatment DPOAE I/O responses satisfying the 3-point criterion.The x-axis shows DPOAE threshold in dB SPL and the y-axisthe number of cases per threshold range. (C) DPOAE thresholdshifts at f2�8 kHz. The x-axis shows the threshold shift in dBand the y-axis the number of cases per threshold shift range.

(A)

(B)

(C)

2 2

3 3

2 2

3 3

4

4

L

259Hatzopoulos/Di Stefano/Campbell/Falgione/Ricci/Rosignoli/Finesso/Albertin/Previati/Capitani/Martini

Cisplatin Ototoxicity in the SpragueDawley Rat Evaluated by DPOAE

typical V-shaped stereocilia arrangement is shown inFigure 8B under a higher magnification (22,500).

As expected, the inner hair cells (IHCs) from thecochleas of the treated groups appeared not to be affectedby the cisplatin administration, and their stereociliaseemed to maintain their typical structure and organi-sation. The OHCs were affected in several ways; in somesubjects, the type of damage of the OHC stereociliaappeared quite disordered, and it seemed that the OHCshad lost their V-shape. In the specimen shown in Figure 9,a complete fusion of stereocilia was observed eitherwithin single OHCs (Figure 9B) or between adjacentOHCs (Figure 9C), particularly the OHCs from the firstrow.

Discussion

Cisplatin ototoxicity has been investigated in manystudies, with various differences in concentration, dosage,modality of administration, and animal models. Schweitzeret al 40 used cisplatin in several dosage schedules (upto 15 mg/kg) administered subcutaneously in guineapigs; Amsallem and Andrieu-Guitrancourt41 administeredsingle or fractioned cisplatin dosages of 6–12 mg/kgintraperitoneally in the guinea pigs; Laurell8 found thatan intravenous dosage of 12 mg/kg may result in a highmortality rate but little ototoxic effect in SD rats. Saitoand Aran14 administered intramuscular cisplatin boli toguinea pigs in consecutive dosages to a cumulative dosage

Figure 4. Regression curves of the pre- and post-treatmentI/O data. The three fine-line traces in the graph depict the meanpretreatment signal-to-noise ratios ± one standard deviation.In the G1 group, the DPOAE post-treatment responses wereclustered into two subgroups, indicated as subgroup no. 1(triangular dotted line) and subgroup no. 2 (ball dotted line).The starting value of the y-axis is 3 dB.

Figure 5. Post-treatment ABR threshold shifts at 16, 12 and8 kHz. The x-axis shows the amount of elevation in dB, and they-axis the number of cases per threshold-shift range. The labelsinside the graphs show the mean threshold values ± the standarderror mean.

10.0 20.0 30.0 40.0 50.0 60.0

Threshold shift (dB)

Num

ber

of

case

s

6

5

4

3

2

1

0

1

3 3

1 1

16 kHz: 33.6±3.6 dB

10.0 20.0 30.0 40.0 50.0

Threshold shift (dB)

Num

ber

of

case

s

6

5

4

3

2

1

0

3 3

5

1

2

12 kHz: 27.1±3.5 dB

10.0 20.0 30.0 40.0

Threshold shift (dB)

Num

ber

of

case

s

8

6

4

2

0

7

5

1 1

8 kHz: 17.1±2.4 dB

5

260 Audiology, Volume 40 Number 5

of 15 mg/kg. Ravi and Somani21 administered cisplatin inWistar rats by slow infusion and established that thecisplatin dosage capable of elevating the ABR thresholdby approximately 30 dB was 16 mg/kg. These examplesshow the diversity of cisplatin experimentation, which isthe main reason why the data on cisplatin-induced oto-toxicity are not completely comparable. Ekborn et al42

showed that two different modes of intravenous cisplatinadministration (bolus versus slow infusion) in the guineapigs did not cause significantly different cochlear damage,but that the slow infusion resulted in less weight loss anda reduced mortality of the treated animals. Fillastre andRaguenez-Viotte43 recommended the use of slow cisplatininfusion rates in human patients as a means of reducingnephrotoxicity. The intraperitoneal slow infusion ofcisplatin, used in the present work, is very close toanalogous human treatments and is able to reduce themortality rates of the animals.

In the above papers, the investigation of cochlearfunction was conducted using exclusively ABR or ECochGresponses. There are only a few studies which haveemployed DPOAE protocols. McAlpine and Johnstone12

showed, by DPOAE recordings, that cisplatin ototoxicityaffects the mechanical process of the movement of thestereocilia in guinea pigs. Sie and Norton31 used DPOAEI/O curves and ABR to assess cisplatin ototoxicity in thegerbil. Shi and Martin30 also recorded DPOAE I/O curvesand ABR to detect cochlear damage in guinea pigs bygentamicin. The use of DPOAEs in a rat animal modelfor the early detection of ototoxicity has been presentedrecently,27,44 although it has been shown for some timein the literature that DPOAEs are easily detectable inrats.6,45,46

The present study evaluated the capacity of theDPOAEs to detect cisplatin ototoxicity in a rat animalmodel, using a drug-administration modality whichresembles the current clinical procedures. In order toverify which DPOAE recording procedure (DP-gram orI/O) could identify better the alterations of cochlearfunctional status, three DP protocols were tested at threedifferent stimulus intensities, and a single I/O curve at thehighest recordable frequency of the VIRTUAL 330apparatus. The reason for testing three protocols at threedifferent stimulus intensities is that the latter reflect theinvolvement of different cochlear mechanisms.

It was initially expected that the alteration of thesignal-to-noise ratios of the DPOAE responses wouldcorrespond to the higher f2 frequencies, since the ototoxiceffects of cisplatin occur initially at basal cochlearlocations.10,17,21,47,48,49 Within this context, it was expectedthat the responses from higher f2 frequencies wouldpresent more significant post–pretreatment differences. Incontradiction to this hypothesis, the statistical analysis ofthe DP-grams suggests that the ototoxic effects do notshow a particular selectivity in terms of f2 frequency (see

Figure 6. Regression curves of the wave III latencies fromthe pre- and post-treatment ABR recording sessions in thetreated group G1 from 8-, 12- and 16-kHz tone-pip stimuli.Scatter-plots of latency values per stimulus intensity are alsopresented.

261Hatzopoulos/Di Stefano/Campbell/Falgione/Ricci/Rosignoli/Finesso/Albertin/Previati/Capitani/Martini

Cisplatin Ototoxicity in the SpragueDawley Rat Evaluated by DPOAE

Figure 7. Scatter plot and regression data fit between the DPOAE and ABR threshold values at 8 kHz.The data in the upper panel refer to a quadratic model. The lower panel indicates the regression values,showing r2�0.78, with an F–Statistic (F stat) value equal to 19.99 and an error probability (Pr>F) equalto 0.0002. A strong influence on the r2 value comes from the last scatter-plot point (DPOAE�30, ABR�80dB SPL). Neglecting this data point, the model becomes very similar to the linear one, with r2�0.62.

Figure 8. Morphological data from an animal of the G2 group (control). (A) Low magnification of part basal–middle turn ofcochlea with three outer rows of hair cells (OHC I, II, III). �4,890; (B) High magnification of the stereocilia of the first OHC row.�22,500.

Figure 9. Morphological data from an animal of the G1 group. (A) Complete fusion of stereocilia is visible in the OHCs of thefirst row. �1310; (B) Higher magnification of the outlined cell in (A). �12,300; (C) Fusion of stereocilia between adjacent cells isvisible, in particular in the first row of the OHCs. �8190.

(A) (B)

262 Audiology, Volume 40 Number 5

Figure 1), since a wide range of frequencies, across allthree DP-gram protocols, showed significant pre–postsignal-to-noise alterations. These findings confirm alsothe results from a gerbil animal model treated with 12mg/kg of cisplatin.31 There are several hypotheses whichmight explain our findings: (1) the chosen DP-gramresolution (12 points per octave) was too high, resulting inan overlapping of the DPOAE responses––in this context,cochlear damage spanning a few hundred hertz couldaffect a large number of f2 tested frequencies; (2) the f2

frequencies tested correspond anatomically to the middlecochlear coil of the rat, and it is feasible that, over thisrange (4.0–8.0 kHz), the cochlear lesions are homo-geneous, affecting all frequencies.

The analysis of the DPOAE I/O curves at 8 kHzpresented several patterns. For stimulus intensities between60 and 75 dB SPL, the majority of the treated animalspresented acceptable DPOAE thresholds. For the lowerstimulation intensities of 35–55 dB SPL, the percentage ofacceptable responses was reduced significantly and, asshown in Figure 3, only 45–57.5 per cent of the animalsshowed acceptable signal-to-noise ratios. The range of theobserved DPOAE threshold shift in the treated animalswas 0–30 dB (mean 13.19 dB ± 2.6 dB). These threshold-shift values are larger than the 7.5-dB mean value of thegerbil model reported by Sie and Norton.31 In thiscontext, it is plausible that the observed threshold shift isrelated not only to the sensitivity of the SD rat strain, butto a number of other factors, such as the higher dose ofcisplatin used in the SD model (16 versus 12 mg/kg) andthe drug-administration modality (infusion in the ratmodel versus subcutaneous bolus in the gerbil model).

The I/O observations suggest that stimulation levelsbetween 45 and 55 dB SPL can be utilized in the charac-terization of an ototoxic insult. These findings are inagreement with the hypothesis that, in rodents, theDPOAE responses are composed of two components, an“active” element which dominates at low stimulus levels,and a “passive” element which predominates at highstimulus levels.34,35,50 These two components show differentvulnerability to insults, in that the DPOAEs elicited bylow-level stimuli are more vulnerable to administration ofototoxic drugs, acoustic trauma and asphyxia than thosegenerated by high-level stimuli.33

The ABRs were affected by the cisplatin admini-stration, as shown by the increase of the wave IIIlatencies. Comparing the wave I–III latency valuesbetween the pre- and post-treatment responses, no signi-ficant differences were found, suggesting that the centralnervous system was not involved in the observed cisplatinototoxicity. In terms of ABR threshold elevation, thepost-treatment ABRs from the 16-kHz stimulus showedthe largest mean threshold shift of 33.67 dB. The threshold-shift values reported in this study match the data pre-sented by Ravi et al21 for the high frequencies, although in

the present study the mean threshold elevation at 8 kHzwas not so pronounced (17 dB versus 27 dB in Ravi’sstudy).

The significant correlation between the DPOAE andABR thresholds at 8 kHz suggests that the acousticalmeasurements (DPOAEs) present significant informationabout the degree of the ototoxic insult. Considering thatthe acoustical measurements are 10 times faster than theelectrophysiological measurements a rat model testedwith the I/O modality might have widespread clinicalapplication in the assessment of cochlear alteration afteradministration of ototoxic substances or exposure to noise.

The level of cisplatin ototoxicity, as revealed by theelectrophysiological and acoustical measurements (i.e. therange of the DPOAE- and ABR-induced threshold shiftsshown in Figures 3 and 5), showed a large variabilityacross different animals, corresponding to different degreesof cochlear damage. This variability has been noted inprevious studies in human subjects51 and in animals.15,49

According to Laurell et al,52 the observed variabilitymight be caused by pharmacological and physiologicalfactors, inasmuch as a different permeability of theblood–perilymph barrier might increase the local concen-tration of cisplatin in the cochlea and in the striavascularis. In the present study, the variability of thecisplatin-induced ototoxicity was also verified by themorphological analyses of the cochleas from the treatedanimals. The data available suggested that even thecochleas from the same treatment group showed differentdegrees of OHC damage.

It has been suggested that the ototoxic effects ofcisplatin might be reduced by using various protectivecompounds, such as fosfomycin,53 D-methionine,19 derivedneurotrophic factors54 and melanocortins,13,55 both inanimal models and in vitro. The authors consider that therat model and the employment of DPOAE I/O curvesat low stimulus intensities can be useful tools in theevaluation of the efficacy of newly suggested protectivecompounds.

Summary

The SD rat animal model, treated with 16 mg/kg ofcisplatin in a 30-min slow intra-peritoneal infusion, hadan excellent survival rate 72 h after treatment. All treatedanimals demonstrated significant alterations of theDPOAE and ABR responses and significant shifts ofthe ABR and DPOAE thresholds. Our findings can besummarised as follows:

• A wideband of f2 frequencies, in the DP-grams, showedsignificant reductions of the corresponding DPOAEsignal-to-noise ratios. The data from the mid-levelDPOAE protocol showed the highest number of post-treatment f2 frequencies (9 of 13) affected.

263Hatzopoulos/Di Stefano/Campbell/Falgione/Ricci/Rosignoli/Finesso/Albertin/Previati/Capitani/Martini

Cisplatin Ototoxicity in the SpragueDawley Rat Evaluated by DPOAE

• The DPOAE I/O curves, tested at 8 kHz, were shown tobe the most sensitive DPOAE procedure to detectototoxic damage caused by cisplatin administration.The signal-to-noise responses from the stimulationrange of L1�35–55 dB SPL provide direct evidence ofan ototoxic cochlear insult. The I/O DPOAE and ABRthresholds at 8 kHz were found to be significantlycorrelated.

• After cisplatin administration, the ABR thresholdswere elevated and the latency values for waves I and IIIwere significantly altered. The greatest threshold changewas observed at 16 kHz (33.6 ± 3.6 dB).

• Exploratory morphological analyses in the middlecochlear coil verified that cisplatin ototoxicity alters theorganisation of the OHC stereocilia but does not affectthe IHCs.

Acknowledgments

The authors wish to thank Fidia S.p.A. for use oflaboratory facilities. The authors also wish to thankanonymous reviewers, Dr Larry Hughes, Dr JoePetruccelli and Dr Silvano Prosser for comments onearlier versions of the manuscript.

References

1. Patuzzi R. Cochlear micromechanicsand macromechanics. In: Dallos P,Popper AN, Fay RR, eds. Springerhandbook of auditory research, Vol. 8:The cochlea. New York: SpringerVerlag, 1996: 186–257.

2. Horner KC, Lenoir M, Bock GR.Distortion product otoacoustic emis-sions in hearing-impaired mutant mice.J Acoust Soc Am 1985;78:1603–1611.

3. Brownell WE. Outer hair cell electro-motility and otoacoustic emissions. EarHear 1990;11:82–92.

4. Trautwein P, Hofstetter P, Wang J, SalviR, Nostrant A. Selective inner hair cellloss does not alter distortion productotoacoustic emissions. Hear Res 1996;96:71–82.

5. Hofstetter P, Ding D, Powers N, SalviRJ. Quantitative relationship of carbo-platin dose to magnitude of inner andouter hair cell loss and the reduction indistortion product otoacoustic emissionamplitude in chinchillas. Hear Res 1997;112:199–215.

6. Brown AM. Acoustic distortion fromthe rodent ears: a comparison of res-ponses from rats, guinea pigs andgerbils. Hear Res 1987;31:25–38.

7. Brown AM, Kemp DT. Intermodu-lation distortion in the cochlea: couldbasal vibration be the major cause ofround window CM distortion? HearRes 1985;19:191–198.

8. Laurell G. Ototoxicity of the anticancerdrug cisplatin. Thesis, KarolinskaHospital, Stockholm, 1991.

9. Hayes DM, Cvitkovic E, Golbey RB.High dose cis-platinum diamminedichloride: amelioration of renal toxicityby mannitol diuresis. Cancer 1977;39:1372–1381.

10. Schweitzer VG. Cisplatin-induced oto-toxicity: the effect of pigmentation andinhibitory agents. Laryngoscope 1993;103:1–52.

11. Fausti SA, Schechter MA, RappaportBZ, Frey RH, Mass RE. Early detection

of cisplatin ototoxicity. Cancer 1984;53:224–231.

12. McAlpine D, Johnstone BM. Theototoxic mechanism of cisplatin. HearRes 1990;47:191–204.

13. Hamers FPT, Klis SFL, Gispen WH,Smoorenburg GF. Application of aneuroprotective ACTH(4–9) analog toaffect cisplatin ototoxicity: an electro-cochleographic study in guinea pig. EurArch Otorhinolaryngol 1994;251:23–29.

14. Saito T, Aran JM. Comparative oto-toxicity of cisplatin during acute andchronic treatment. ORL 1994;56:315–320.

15. Hoeve LJ, Mertenz zur Borg IRAM,Rodenburg M, Brocaar MP, GroenBGS. Correlation between cis-platinumdosage and toxicity in a guinea pigmodel. Arch Otorhinolaryngol 1988;245:98–102.

16. Tange R, Vuzevski VD. Changes in thestria vascularis of the guinea pig due tocis-platinum. Arch Otorhinolaryngol1982;239:41–47.

17. Tange R, Conijin A, Van Zeig IL.Cortitoxic effects of cisplatin in theguinea pig. Arch Otorhinolaryngol 1982;237:17–26.

18. Laurell G, Bagger-Sjöbäck D.Degeneration of the organ of Cortifollowing intravenous administration ofcisplatin. Acta Otolaryngol 1991;111:891–898.

19. Campbell KCM, Rybak L, Meech P,Hughes L. D-Methionine providesexcellent protection from cisplatin oto-toxicity in the rat. Hear Res 1996;102:90–98.

20. Suzuki M, Kaga K. Effect of cisplatinon the basement membrane anionicsites in the ampulla, macula, and striavascularis of guinea pigs. Ann OtolRhinol Laryngol 1997;106:971–975.

21. Ravi R, Somani SM, Rybak LP.Mechanism of cisplatin ototoxicity:antioxidant system. Pharmacol Toxicol1995;76:386–394.

22. Clerici WJ, Hensley K, DiMartino DL,Butterfield A. Direct detection ofototoxicant-induced reactive oxygenspecies generation in cochlear explants.Hear Res 1996;98:116–124.

23. Ford MS, Nie Z, Whitworth C, RybakLP, Ramkumar V. Up-regulation ofadenosine receptors in the cochlea bycisplatin. Hear Res 1997;111:143–152.

24. Lautermann J, Crann SA, McLaren J,Schacht J. Glutathione-dependent anti-oxidant systems in the mammalianinner ear: effects of aging, ototoxic drugsand noise. Hear Res 1997;114:75–82.

25. Stengs CH, Klis SF, Huizing EH,Smoorenburg GF. Protective effects of aneurotrophic ACTH(4–9) analog oncisplatin ototoxicity in relation to thecisplatin dose: an electrocochleographicstudy in albino guinea pigs. Hear Res1998;124:108–117.

26. Stengs CH, Klis SF, Huizing EH,Smoorenburg GF. Cisplatin ototoxicity.An electrophysiological dose–effectstudy in albino guinea pigs. Hear Res1998;124:99–107.

27. Hatzopoulos S, Di Stefano M, AlbertinA, Martini A. Evaluation of cisplatinototoxicity in a rat animal model. AnnNY Acad Sci 1999;884:211–225.

28. Martini A, Rubini R, Ferretti RG,et al. Comparative ototoxic potentialof hyaluronic acid and methylcellulose.Acta Otolaryngol 1992;112(2):278–283.

29. Rybak L, Husain K, Morris G,Whitworth C, Somani S. Effect ofprotective agents against cisplatin oto-toxicity. Am J Otol 2000;21:513–520.

30. Shi Y, Martin WH. ABR and DPOAEdetection of cochlear damage by genta-micin. J Basic Clin Physiol Pharmacol1997;8:141–155.

31. Sie KC, Norton S. Changes inotoacoustic emissions and auditorybrain stem response after cis-platinumexposure in gerbils. Otolaryngol HeadNeck Surg 1997;116:585–592.

264 Audiology, Volume 40 Number 5

32. Hauser R, Probst R. The influence ofsystematic primary-tone level variationL2–L1 on the acoustic distortion pro-duct emission 2f1–f2 in normal humanears. J Acoust Soc Am 1991;89:280–286.

33. Whitehead ML, Lonsbury-Martin BL,Martin GK. Evidence for two discretesources of 2f1–f2 distortion productotoacoustic emissions in rabbit. IIDifferential physiological vulnerability.J Acoust Soc Am 1992;92:2662–2682.

34. Mills DM, Norton SJ, Rubel EW.Vulnerability and adaptation of dis-tortion product otoacoustic emissionsto endocochlear potential variation.J Acoust Soc Am 1993;94:2108–2122.

35. Whitehead ML, McCoy MJ, Lonsbury-Martin BL, Martin GK. Dependence ofdistortion-product otoacoustic emissionson primary levels in normal and im-paired ears. I. Effects of decreasing L 2

below L1. J Acoust Soc Am 1995;97:2346–2358.

36. Whitehead ML, Stagner BB, McCoyMJ, Lonsbury-Martin BL, Martin GK.Dependence of distortion-product oto-acoustic emissions on primary levels innormal and impaired ears. II. Asym-metry in L1, L2 space. J Acoust Soc Am1995;97:2359–2377.

37. Lonsbury-Martin BL, McCoy MJ,Whitehead ML, Martin GK. Clinicaltesting of distortion-product otoacousticemissions. Ear Hear 1993;1:11–22.

38. Bourre JM, Durand G, Erre JP, AranJM. Changes in auditory brainstemevoked potentials in alpha-linolenicacid deficiency as a function of age in

in guinea pig with special reference totoxic effects in the stria vascularis.Laryngoscope 1988;98:871– 885.

48. Khvoles R, Freeman S, Sohmer H.Transient evoked otoacoustic emissionscan be recorded in the rat. Hear Res1996;97:120–126.

49. Laurell G, Engström B. The ototoxiceffect of cisplatin on guinea pig inrelation to dosage. Hear Res 1989;38:27–34.

50. Shera CA, Guinan JJ Jr. Evokedotoacoustic emissions arise by twofundamentally different mechanisms: ataxonomy for mammalian OAEs. JAcoust Soc Am 1999;105:782–795.

51. Laurell G, Jungnelius U. High-dosecisplatin treatment. Hearing loss andplasma concentrations. Laryngoscope1990;100:724–734.

52. Laurell G, Teixeira M, Sterkers O,Ferrary E. Paracellular transport pro-perties of inner ear barriers do notaccount for cisplatin toxicity in the rat.Hear Res 1997;110:135–140.

53. Ohtani I, Ohtsuki T, Aikawa T, AnzaiT, Ouchi J, Saito T. Reduction ofcisplatin ototoxicity by fosfomycin inanimal model. ORL 1985;47:229–235.

54. Gabaizadeh R, Staecker H, Liu W, et al.Protection of both auditory hair cellsand auditory neurons from cisplatininduced damage. Acta Otolaryngol 1997;117:232–238.

55. Heijmen PS, Klis SFL, De GrootJCMJ, Smoorenburg GF. Cisplatin oto-toxicity and the possibly protective effectof α-melanocyte stimulating hormone.Hear Res 1999;128:27–39.

rats. Audiology 1999;38:13–18. 39. Müller M. Frequency representation in

the rat cochlea. Hear Res 1991;51:247–254.

40. Schweitzer VG, Rarey KE, Dolan DF,Abrams G, Litterst CJ, Sheridan C.Ototoxicity of cisplatin vs. platinumanalogs CBOCA (JM-8) and CHIP(JM-9). Otolaryngol Head Neck Surg1986;94(4):458–70.

41. Amsallem P, Andrieu-Guitrancourt J.Étude experimentale de l’ototoxicité ducisplatinum. Ann OtoLaryngol 1985;102:365–372.

42. Ekborn A, Laurell G, Anderson A,Walin I, Eksborg S, Ehrsson H.Cisplatin induced hearing loss. Influenceof mode of administration in the guineapig. Hear Res 2000;140:38–44.

43. Fillastre JP, Raguenez-Viotte G.Cisplatin nephrotoxicity. Toxicol Lett1989;46:163–175.

44. Sochalingam R, Freeman S, ChernyTL, Sohmer H. Effect of high-dosecisplatin on auditory brainstem res-ponses and otoacoustic emissions inlaboratory animals. Am J Otol 2000;21:521–527.

45. Henley CM III, Owings MH, StagnerBB, Martin GK, Lonsbury-Martin BL.Postnatal development of 2f1–f2 oto-acoustic emissions in pigmented rat.Hear Res 1989;43:141–148.

46. Freeman S, Cherny L, Sohmer H.Thyroxine affects physiological andmorphological development of the ear.Hear Res 1996;97:19–29.

47. Khon S, Fradis M, Zidan J, Podoshin L,Robinson E, Nir I. Cisplatin ototoxicity