Comparing Audiometric Comparing Audiometric Threshold Shift Definitions Threshold Shift Definitions for for OtotoxicityOtotoxicity Detection Detection

Jane Gordon, Dawn KonradJane Gordon, Dawn Konrad--Martin, Martin, Kelly Reavis, Debra Wilmington, Kelly Reavis, Debra Wilmington,

Gene Bratt, & Stephen FaustiGene Bratt, & Stephen Fausti

National Center for National Center for Rehabilitative Auditory ResearchRehabilitative Auditory Research

NCRARNCRARPortland VA Medical CenterPortland VA Medical Center

Portland, OregonPortland, Oregon

Work supported by VA Rehabilitation R&D Service (grants: C3213R; C4697R; and C4447K)

OutlineOutline

Learning ObjectivesLearning ObjectivesBackgroundBackgroundStudy ObjectivesStudy ObjectivesMethodsMethodsResultsResultsConclusions & Clinical ImplicationsConclusions & Clinical ImplicationsQuestions & AnswersQuestions & Answers

Learning ObjectivesLearning Objectives

Describe rationale for early detection of Describe rationale for early detection of ototoxicototoxic hearing losshearing lossOutline benefits of using Clinical Decision Outline benefits of using Clinical Decision Theory for examining test performanceTheory for examining test performanceDiscuss implications of this study for Discuss implications of this study for various various ototoxicityototoxicity monitoring methodsmonitoring methods–– Criteria for hearing threshold shiftsCriteria for hearing threshold shifts–– TestTest--frequency step sizesfrequency step sizes–– Testing above 8 kHzTesting above 8 kHz

Patients may not notice hearing Patients may not notice hearing change until speech frequencies change until speech frequencies affected affected

Serum levels are not good indicators Serum levels are not good indicators cochleototoxicitycochleototoxicity

The only way to know if a person is The only way to know if a person is losing their hearing is direct losing their hearing is direct assessment of auditory functionassessment of auditory function

BackgroundBackground

Attributes of Screening TestAttributes of Screening TestHigh sensitivity (sensitivity rate=hit rate)High sensitivity (sensitivity rate=hit rate)

–– High hit rate in exposed ears High hit rate in exposed ears

High specificityHigh specificity (1(1--specificity=false positive rate)specificity=false positive rate)

– Low false alarm rate in unexposed ears

Most tests are imperfect, so that some clinical Most tests are imperfect, so that some clinical errors are made. There is a trade off between errors are made. There is a trade off between sensitivity and specificitysensitivity and specificity

– Make cutoff more stringent, reduce false positives, but also reduce hit rate

– Make cutoff more lax, increases hit rate, but also increases false positives

Time efficiencyTime efficiency

OtotoxicityOtotoxicity MonitoringMonitoring

Benefits of Benefits of OtotoxicityOtotoxicity MonitoringMonitoringEarly detection may prevent hearing Early detection may prevent hearing damage that requires rehabilitationdamage that requires rehabilitation

If change observed, treatment If change observed, treatment modification can prevent further hearing modification can prevent further hearing loss; if no change observed, continued loss; if no change observed, continued treatment warrantedtreatment warranted

OtotoxicityOtotoxicity monitoring program monitoring program –– educates patients, care givers and physicians educates patients, care givers and physicians

about about ototoxicototoxic symptoms (hearing loss, symptoms (hearing loss, tinnitus, and balance problems)tinnitus, and balance problems)

–– raises awareness of synergistic effects of raises awareness of synergistic effects of toxins and noisetoxins and noise

–– ensures ensures audiologyaudiology work up and rehabilitation work up and rehabilitation plan are implemented if and when appropriate plan are implemented if and when appropriate

Sensitive Range of Sensitive Range of OtotoxicityOtotoxicity(SRO) Principle(SRO) Principle

Most initial changes seen in a limited Most initial changes seen in a limited range near the highest audible frequencyrange near the highest audible frequency

Range for each individual is unique and Range for each individual is unique and specific to their hearing configuration specific to their hearing configuration

Frequencies with thresholds > 100 dB SPL Frequencies with thresholds > 100 dB SPL usually remain unchangedusually remain unchanged

A sensitive range for ototoxicity (SRO) can A sensitive range for ototoxicity (SRO) can be defined as a one octave range near the be defined as a one octave range near the upperupper--frequency limit of hearing (i.e., frequency limit of hearing (i.e., frequency with a threshold frequency with a threshold <<100 dB SPL)100 dB SPL)

-100

102030405060708090

100110120

0.5 1 2 3 4 6 8 9 10 11.2 12.5 14 16 18 20

Thre

shol

d (d

B SP

L)

Behavioral SRO

Example SROExample SRO

F

F-1

F-2

F-3

F-4F-5F-6

Sensitivity: SRO 1/6th OctaveSensitivity: SRO 1/6th OctaveTotalTotal

(Ears)(Ears)HitHit MissMiss

5454

226226

5959

4646

339339

207207

88

1919

995050

3636303303

Initial Change Initial Change in SROin SRO

AMGAMG 85%85%

CisplatinCisplatin 92%92%

CarboplatinCarboplatin 85%85%

TotalTotal 89%89%

Few studies have compared audiometric Few studies have compared audiometric methods and hearing loss change criteriamethods and hearing loss change criteriaMethods vary for monitoring hearing in Methods vary for monitoring hearing in patients receiving patients receiving ototoxicototoxic drugsdrugs– frequencies tested – frequency step sizes used– decision variables (number frequencies affected)– magnitude of threshold shifts

OtotoxicityOtotoxicity MonitoringMonitoring

0102030405060708090

100110120130

0.5 2 4 6.35 8 10 12.5 16 20

Frequency (kHz)

Thre

shol

d (d

B S

PL)

19 July Baseline17 August Monitor18 August Retest5 October 1 Month20 March 6 Month

Case Study: SRO 7.13 - 14 kHz

F

F-1F-2

F-3F-4F-5

F-6

Case StudyCase StudyConventional audiometric testing (<8 kHz) Conventional audiometric testing (<8 kHz) would have identified hearing change later would have identified hearing change later Hearing continued to decline after Hearing continued to decline after treatment ended (black line: 1 month posttreatment ended (black line: 1 month post--drug; and light blue line: 6 month postdrug; and light blue line: 6 month post--drug)drug)– ultimately, 20-dB change at 3 kHz; 40-dB

change at 4 kHz; 65-dB change at 8 kHz Greatest changes at the conventional Greatest changes at the conventional frequencies occurred after treatment frequencies occurred after treatment ended!ended!Information used to council patient about Information used to council patient about aural rehab options and necessity of follow aural rehab options and necessity of follow up with audiologistup with audiologist

BackgroundBackgroundChallenge is to quantify Challenge is to quantify ototoxicototoxic changes changes in a group of adultsin a group of adultsDifferent frequencies may be affected in a Different frequencies may be affected in a given person, so averaging across all given person, so averaging across all frequencies tested tends to wash out frequencies tested tends to wash out ototoxicototoxic changeschanges90% of changes occur within an octave of 90% of changes occur within an octave of highest audible frequency highest audible frequency (Fausti et al., (Fausti et al., 1993)1993)Therefore, can average threshold shifts Therefore, can average threshold shifts across individuals, normalizing to the across individuals, normalizing to the highest frequency able to be heard highest frequency able to be heard

Study ObjectivesStudy Objectives1.1. Examine ability of individuallyExamine ability of individually--tailored tailored

threshold monitoring to detect threshold monitoring to detect ototoxicototoxichearing loss using various (significant hearing loss using various (significant threshold shift) STS definitions threshold shift) STS definitions

2.2. Determine whether evidence supports Determine whether evidence supports use of ASHAuse of ASHA--recommended STS recommended STS definitionsdefinitions

3.3. Determine whether testing in 1/3Determine whether testing in 1/3-- or or 1/61/6--octave steps improves test octave steps improves test performance when compared to use of performance when compared to use of 1/21/2--octave stepsoctave steps

MethodsMethodsAll subjectsAll subjects– at least 3 audiograms, one audio ~1.5 months, one

audio ~6.5 months after initial dose– baseline within 2 days of first dose (some controls were

within 3 days)CisplatinCisplatin--exposed Groupexposed Group– used for hit or true positive (TP) rates– 78 ears of 41 patients receiving cisplatin– cumulative dose at least 350 mg – mean age 59.4 years (SD 10.2)Control Group Control Group – used for false positive (FP) rates– 53 ears of 28 hospitalized patients receiving non-

ototoxic antibiotics – mean age 56.0 (SD 10.5)

Example: Frequencies Tested in Example: Frequencies Tested in One SubjectOne Subject

Baseline Test Frequencies

(kHz)

0.50

1.

00

2.00

3.

00

4.00

6.

00

6.35

7.

13

8.00

9.

00

10.0

0 11

.20

12.5

0 14

.00

16.0

0 18

.00

20.0

0

Baseline Hearing Thresholds

25

25

35

40

55

55

55

60

70

75

85

90

100

110

> 11

5 >

115

> 11

5

Comparison Test Frequencies 1/6-octave F-

6 F-

5 F-

4 F-

3 F-

2 F-

1 F

Comparison Test Frequencies 1/3-octave F-

3 F-

2 F-

1 F

Comparison Test Frequencies 1/2-octave F-

2 F-

1 F

Group Data in 1/6Group Data in 1/6thth--octavesoctaves

Mean data underestimate effect because some subjects have hearing loss at a given frequency, but others do not

MethodsMethodsHearing changes calculated for Hearing changes calculated for frequencies within an octave of highfrequencies within an octave of high--frequency hearing limit frequency hearing limit Compared initial audiogram to tests Compared initial audiogram to tests obtained 1.5 and 6.7 months later obtained 1.5 and 6.7 months later Determined if changes met STS criteriaDetermined if changes met STS criteria– based on magnitudes of positive threshold

shifts (worsening of thresholds by 5, 10, 15, or 20 dB)

– and numbers of frequencies affected (threshold shifts at a minimum of 1, 2, or 3 adjacent frequencies)

Test performance evaluated using CDTTest performance evaluated using CDT

Clinical Decision Theory (CDT)Clinical Decision Theory (CDT)Can go beyond asking Can go beyond asking ‘‘Does the test work or Does the test work or not?not?’’, to the more useful question, , to the more useful question, ‘‘How well How well does it work in this context?does it work in this context?’’Can estimate test performance for all values of Can estimate test performance for all values of the experimental variable, in this case dB of the experimental variable, in this case dB of hearing changehearing change

in this case 5, 10, 15, or 20 dB of threshold shift

Can compare various different decision variables Can compare various different decision variables in this case, numbers of frequencies affected (1, 2, 3 or more adjacent)

Note absence of true Note absence of true ““gold standardgold standard””Compared STS rates for Compared STS rates for cisplatincisplatin--exposed group exposed group ((““truetrue”” positive rates) to nonpositive rates) to non--exposed group exposed group (false positive rates) following Dobie 2005(false positive rates) following Dobie 2005– Underestimates true positive rates– Accurately estimates false positive rates – Accurately estimates rankings of competing STS criteria

MethodsMethodsCompared test performance 2 ways Compared test performance 2 ways

1. Examined receiver operating characteristic 1. Examined receiver operating characteristic (ROC) curves (ROC) curves – Plots of true positive rate or sensitivity as a function of

false positive rate or 1 minus specificity– Compared relative position of curves and areas under

the curves

2. Determined STS definitions with the highest 2. Determined STS definitions with the highest true positive rates for a fixed false positive rate true positive rates for a fixed false positive rate near 5%near 5%– Chosen to minimize the number of patients incorrectly

diagnosed with ototoxic hearing loss

ROC CurvesROC CurvesReceiver Operating Characteristic (ROC) Curves

Each point in ROC curve represents a particular criterion cutoff (dB shift)

Sensitivity (true positive rate) is plotted vs false positive rate

Criterion cutoffs range from stringent (20-dB change) to lax (5-dB change) as curve progresses from left to right

Relative position of curves and area beneath them is used to compare them

ROC CurvesROC Curves

AUCsAUCs

** **

*

Final TestFinal Test

2015105

Magnitude of Change (dB)

100

80

60

40

20

0

Percent %

2015105 2015105

100

80

60

40

20

0

100

80

60

40

20

0

Three or more frequenciesTwo or more frequenciesOne or more frequencies

1/2 Octave

1/3 Octave

1/6 Octave

Frequency Step-Size

False Positives (Control)True Positives (Cisplatin)

ASHA Criteria for Hearing ChangeASHA Criteria for Hearing Change

20 dB change at any 1 test frequency20 dB change at any 1 test frequency10 dB change at any 2 adjacent test 10 dB change at any 2 adjacent test frequenciesfrequenciesLoss of responses (as little as 5 dB Loss of responses (as little as 5 dB change) at 3 consecutive frequencies, change) at 3 consecutive frequencies, where responses were previously obtained where responses were previously obtained close to the limit of the audiometerclose to the limit of the audiometerChanges confirmed by repeat testingChanges confirmed by repeat testing

ResultsResultsFor the 1/2For the 1/2--octave step size used octave step size used clinically, best performance was achieved clinically, best performance was achieved for threshold losses of 15 dB at a single for threshold losses of 15 dB at a single frequency, a change present in 50% of frequency, a change present in 50% of ears by completion of ears by completion of cisplatincisplatin therapy. therapy. Use of smaller frequency step sizes Use of smaller frequency step sizes consistently increased test performance; consistently increased test performance; (improvements were significant when (improvements were significant when threshold shifts were required to occur at threshold shifts were required to occur at 2 or 3 adjacent frequencies)2 or 3 adjacent frequencies)Best overall test performance was Best overall test performance was obtained using 1/6obtained using 1/6--octave steps and a octave steps and a criterion cutcriterion cut--off of 10 dB at 2 or more off of 10 dB at 2 or more frequenciesfrequencies

ResultsResultsCertain ASHACertain ASHA--recommended criteria recommended criteria performed wellperformed well– threshold shifts > 20 dB at 1 frequency – > 10-dB at 2 or more adjacent frequencies)For For ½½--octave step size, these STS octave step size, these STS definitions resulted in TP rates of 36% and definitions resulted in TP rates of 36% and 39%, respectively, at the final test date. 39%, respectively, at the final test date. 55--dB at three frequencies performed less dB at three frequencies performed less well due to increased FP ratewell due to increased FP rate

Conclusions & Clinical ImplicationsConclusions & Clinical Implications

Monitoring protocol that uses a tailored, Monitoring protocol that uses a tailored, patientpatient--specific one octave range of specific one octave range of frequencies is a clinically effective protocol frequencies is a clinically effective protocol for detecting early for detecting early ototoxicototoxic changes with changes with an acceptable false positive rate.an acceptable false positive rate.In general, using smaller frequency step In general, using smaller frequency step sizes increases false positives only sizes increases false positives only slightly, while increasing sensitivity slightly, while increasing sensitivity significantly. significantly.

High Frequency AudiometersHigh Frequency AudiometersInteracoustics AC40Interacoustics AC40–– HighHigh--frequency to 20 kHz, multifrequency to 20 kHz, multi--frequency, 1/6 octave frequency, 1/6 octave

(1 octave up to 1/24)(1 octave up to 1/24)–– With boost feature, 115With boost feature, 115--120 dB output120 dB output–– HL HL ≤≤ 8 kHz; SPL > 8 kHz8 kHz; SPL > 8 kHz–– Approximately $6200.Approximately $6200.DecosDecos Systems Systems AudiologyAudiology WorkstationWorkstation

HighHigh--frequency to 20 kHz, multifrequency to 20 kHz, multi--frequency, 1/6 octavefrequency, 1/6 octave120 dB HL output at all frequencies120 dB HL output at all frequencies--Computer basedComputer basedProgrammable option for testing in HL & SPL at all Programmable option for testing in HL & SPL at all frequenciesfrequenciesApproximately $12,800. Approximately $12,800.

ReferencesReferencesAmerican SpeechAmerican Speech--LanguageLanguage--Hearing Association (1994). Guidelines for the Hearing Association (1994). Guidelines for the

audiologicaudiologic management of individuals receiving management of individuals receiving cochleotoxiccochleotoxic drug therapy. drug therapy. ASHA,ASHA, 36: 1136: 11--19.19.

Dobie RA. Audiometric threshold shift definitions: Simulations aDobie RA. Audiometric threshold shift definitions: Simulations and nd suggestions. suggestions. Ear and HearEar and Hear 26: 6226: 62--77, 2005.77, 2005.

Fausti SA, Henry JA, Heidi I, Schaffer MA, Olson DJ, Frey RH, Fausti SA, Henry JA, Heidi I, Schaffer MA, Olson DJ, Frey RH, BagbyBagby GC. GC. HighHigh--frequency monitoring for early detection of frequency monitoring for early detection of CisplatinCisplatin ototoxicityototoxicity. . Arch Arch OtolaryngolOtolaryngol Head Neck Head Neck SurgSurg 119:661119:661--666, 1993.666, 1993.

Fausti SA, Helt WJ, Gordon JS, Reavis KM, Phillips DS, & KonradFausti SA, Helt WJ, Gordon JS, Reavis KM, Phillips DS, & Konrad--Martin D. Martin D. AudiologicAudiologic monitoring for monitoring for ototoxicityototoxicity and patient management, in and patient management, in Pharmacology and Pharmacology and OtotoxicityOtotoxicity for Audiologistsfor Audiologists (KC Campbell, ed.) 2006.(KC Campbell, ed.) 2006.

Fausti SA, Henry JA, Helt WJ, Phillips DS, Frey RH, Fausti SA, Henry JA, Helt WJ, Phillips DS, Frey RH, NoffsingerNoffsinger D, Larson VD, D, Larson VD, Fowler CG. An individualized, sensitive frequency range for earFowler CG. An individualized, sensitive frequency range for early ly detection of detection of ototoxicityototoxicity. . Ear and HearEar and Hear 20:49720:497--505, 1999.505, 1999.

Pasic TR, Dobie RA. Pasic TR, Dobie RA. CisCis--platinum platinum ototoxicityototoxicity in children. in children. LaryngoscopeLaryngoscope 101: 101: 985985--991, 1991.991, 1991.

Simpson TH, Simpson TH, SchwanSchwan SA, SA, RintelmannRintelmann WF. Audiometric test criteria in the WF. Audiometric test criteria in the detection of detection of cisplatincisplatin ototoxicityototoxicity. . J Am J Am AcadAcad AudiolAudiol 3: 1763: 176--185, 1992.185, 1992.