Chemotherapeutic treatment with cisplatin is highly effective against many types of cancer. Unfortunately, cisplatin can produce cochlear damage resulting in irreversible hearing loss if treatment continues. Preventing or minimizing cochlear damage is critical, especially in a population of Veterans where pre-existing hearing losses are often present prior to drug treatment. At least 30% of Veterans become unable to provide reliable behavioral data during the course of their treatment. A rapid, sensitive and reliable non-behavioral method that does not require an alert and attentive patient would be indispensable for an ototoxicity monitoring program. During the current support period, we developed a multivariate ototoxicity risk assessment (ORA) model combining the pre-exposure (to medication) audiogram and cumulative cisplatin dose with fine resolution distortion-product otoacoustic emission (DPOAE) level changes observed at each chemotherapeutic visit. The ORA is a quick, sensitive and reliable detector of ototoxicity that can provide hearing information for those Veterans who are unable to take a hearing test. Before it can be used in the clinic, however, the ORA must be validated. Objectives: The first objective is to validate the ORA on a new and independent group of subjects. The second objective is to evaluate if the ORA can be improved even further by using multiple DPOAE frequencies (2f1-f2 and 2f2-f1) or, alternatively, by using stimulus-frequency otoacoustic emissions (SFOAEs). The third objective is to determine if OAEs can predict an individual's susceptibility for hearing change during future treatment with cisplatin. Objectives 2 and 3 can be accomplished using the validation procedures with only a small increase in testing time for the SFOAEs. Methods: Two hundred and forty-six subjects receiving cisplatin and 30 control subjects receiving non-ototoxic chemotherapy medications will be screened at the Portland VA Medical Center (PVAMC) during the study period. Behavioral audiometric thresholds at conventional and extended high frequencies and fine structure OAE level measures (DPOAEs and SFOAEs) with 1/48th-octave precision will be obtained bilaterally on each subject within 24hrs of starting their chemotherapy regimen, at each subsequent treatment and at one-month after cessation of treatment. Testing will be done on the chemotherapy unit at PVAMC Analysis: The ORA will be used to diagnose hearing change for each subject at each monitoring visit. Receiver Operating Characteristic curve analysis will be used to evaluate the accuracy of the ORA on this new sample to address aim #1.
Aim #2 will be addressed using leave-one-out cross-validation analysis of candidate ORAs using SFOAE and multiple DPOAE measures and compared to the original ORA using non-parametric statistical tests.
Aim #3 will be evaluated using discrete-time survival analysis to predict the risk of hearing change over the course of treatment as a function of baseline OAE measurements along with other patient factors. This proposal has three major strengths over previous attempts to use OAE testing for the early detection of ototoxicity. First, it is unique in its utilization of Clinical Decision Theory methodology for evaluating test performance and developing criteria for determining that a significant change in the OAE has occurred. Second, we use a multivariate approach by combining OAEs obtained over a range of frequencies with pre-cisplatin exposure hearing data and the cisplatin cumulative dose, in order to obtain the ototoxicity risk assessment. Finally, individual components of the ORA are weighted such that the linear combination of these components best distinguishes ears of cisplatin-treated subjects with ototoxic hearing change from those without. The final ORA diagnostic model will be selected from a set of competing models using cross validation procedures.
A comprehensive ototoxicity monitoring program using individualized behavioral and objective techniques will significantly reduce the number of Veterans who suffer permanent disabling hearing loss requiring costly rehabilitation. During the current support period, we developed a multivariate ototoxicity risk assessment (ORA) model combining the pre-exposure (to medication) audiogram and cumulative cisplatin dose with fine resolution distortion-product otoacoustic emission (DPOAE) level changes observed at each chemotherapeutic visit. This multivariate ORA model is highly accurate at detecting ototoxic hearing changes when they cannot be directly measured. Before it can be used in the clinic, however, the ORA model must be validated. Further improvements, such as may be accomplished through the combined use of 2f1-f2 and 2f2-f1 DPOAEs or stimulusfrequency otoacoustic emissions, are also desirable for testing patients with greater hearing loss.