Prospective Cohort Study of Ototoxicity in Persons with Cystic Fibrosis following a Single Course of Intravenous Tobramycin

Abstract

Introduction

Aminoglycoside (AG) antibiotics, such as tobramycin, are known to be ototoxic but important clinically due to their bactericidal efficacy. Persons with cystic fibrosis (CF) are at risk for AG-induced ototoxicity due to the repeated use of intravenous (IV) tobramycin for the treatment of pulmonary exacerbations. While it is well-established that ototoxic hearing loss is highly prevalent in this clinical population, the progression of hearing loss over time remains unclear. Cumulative IV-AG dosing has been associated with a higher risk of ototoxic hearing loss, yet some individuals lose substantial hearing after a single IV-AG treatment, while others never seem to lose hearing.

Methods

31 persons with CF (18 on IV tobramycin, 13 controls) were enrolled in an observational study. Pure-tone hearing thresholds (0.25–16 kHz) were measured at baseline (pre-treatment) and at follow-up for each subject. A hearing shift was determined using various metrics, and outcomes were compared to characterize changes in hearing bilaterally for both study groups.

Results

Comparison of pure-tone threshold shifts between baseline and follow up audiograms following either a course of IV tobramycin (n = 18) or no intervening therapy (n = 13) demonstrated significant (p < 0.05) threshold shifts in all continuous metrics tested.

Conclusion

A single course of IV tobramycin causes ototoxic hearing loss in some people with CF, which supports the need for routine ototoxicity monitoring and management in this clinical population. These findings also suggest that people with CF are a suitable population for clinical trials examining ototherapeutics in single IV-tobramycin treatment episodes.

Introduction

It is well-documented that parenteral aminoglycoside (AG) antibiotics, cause hearing loss and loss of vestibular function in some patients, ,,, . AG-associated hearing loss (i.e., ototoxicity) initially affects hair cells in the basal end of the cochlea, resulting in high frequency (> 8 kHz) sensorineural hearing loss. With additional dosing, injury extends to the apical region of the cochlea affecting one’s ability to process low-frequency sounds. Although it has not been possible to predict which patients will suffer hearing loss, the best predictor of the cumulative loss of hearing is the cumulative dose of AGs . This has been borne out for conditions like CF or drug resistant tuberculosis where AGs are given in repeated courses , . For these patient populations, there is an urgent need to identify new drugs that can protect hearing against the cumulative effects of AGs.

For over 40 years, there has been an on-going debate about the incidence and severity of AG induced hearing loss in persons with CF. Previous papers were summarized by Zettner & Gleser (2018) to show that much of the variation was methodological. That paper also examined data from 165 adult CF persons (mean age= 30 yr.) who were treated at the University of California-San Diego (UCSD) CF clinic. Out of 103 persons treated with at least one 8 – 30 day course of IV tobramycin, 64% had hearing loss as defined by a hearing threshold of > 25 decibel hearing level (dB HL) at one or more frequencies (between 0.25 to 16 kHz) in either ear. Of the persons treated with 10 or more courses of IV tobramycin, the percent with hearing loss increased to over 75%. Other studies confirm the high rate of hearing loss in patients with CF. Garinis (2017) reported that 56% of teens and adults with CF (N=81 persons) exhibited sensorineural hearing loss. Pediatric persons with CF, ages 6+ years, exhibit a similarly high prevalence of hearing loss of up to 52% .

Protection against ototoxicity is now seen as a sizable market and the biotechnology industry is responding. However, to date, the United States Food & Drug Administration has not approved a pharmaceutical agent for otoprotection. As new otoprotection drugs prepare to go into clinical trials, it is important to know the natural history of the observable changes in hearing in order to select the appropriate clinical population and sufficiently power the study to establish hearing protection. If the ototoxic effect of a single course of AG therapy were large (deafness) and the toxicity apparent in all participants, then a protectant can be shown to be efficacious with a small number of patients to power the study. However, the magnitude of the hearing effect and the individual variation that occur from a single course of IV-AG therapy are not well documented, and most of the available data are from retrospective studies (re: Zettner & Gleser, 2018) ). Therefore, it is unclear how large of a clinical trial is necessary to show efficacy of even a perfect otoprotectant medication.

As stated above, ototoxic effect of a single course of IV-AG treatment is not yet well documented. To shed light on this issue, Gleser and Zettner conducted a retrospective analysis (using data from the same source as in Zettner and Gleser to compare various metrics of hearing loss in persons with CF either treated with a single course of IV tobramycin (> 13 days), or examined over a comparable period of time without receiving tobramycin. In this study, 45% of persons who received a single course of IV tobramycin demonstrated a worsening of hearing thresholds using the accepted audiological standard defined by the American Speech Hearing Language Association (re: ASHA shift criteria ) compared to 23% of persons not receiving tobramycin between baseline and follow-up audiograms. This shift in hearing was considered ototoxic, given that persons did not have other known risk factors for hearing loss, and the loss of hearing occurred within a specific time frame tied to a course of treatment.

There are several limitations of using retrospective data, such as in the Gleser and Zettner study . The most significant limitation was the lack of control over intervals between when the baseline audiogram was taken in relation to the beginning of treatment, as well as the follow-up audiogram after the end of treatment. As a result, the time between audiograms was large and variable (49 – 2,500 days). Further, inhaled tobramycin and other potentially ototoxic medications were not examined. These are problems that can be minimized in a prospective study. Our research team conducted such a study in an established cohort of adult persons with CF at the Oregon Health & Science University (OHSU) to verify findings reported in Gleser & Zettner, 2018. We report on the effects of a single course of IV tobramycin to further understand the ototoxic effects of these treatments in a controlled, isolated therapeutic event.

Experimental Section

Participant Selection

This study was approved by the OHSU Institutional Review Board as study number #10808. The OHSU Adult and Pediatric CF Centers follow over 350 persons on a regular basis. 31 participants (17 females, 14 males) provided written informed consent or assent to participate in this study. The median age was 33.9 yrs. with a range of 16 to 57 yrs. (see Table 1). Persons were primarily recruited during a routine clinical CF care appointment, or during an inpatient stay at OHSU. Participants 15–17 years of age were required to have parental written consent to participate. Study participants who were scheduled for a course of IV tobramycin participated in the treatment group (TM) and persons who were not receiving IV-tobramycin treatment participated in the no treatment (noTM) group. Participants were excluded if they reported congenital hearing loss, had evidence of middle-ear dysfunction or conductive hearing loss in either ear or had a history of IV-tobramycin treatment within the previous 60 days. It is important to note that persons from both groups could have had exposure to IV tobramycin prior to study participation, and hearing loss at baseline was not a controlled variable.

Table 1:

Person Characteristics of noTM Group and TM group including gender, age at first audiogram, days between audiograms, HFBI (average of thresholds of 8.0–12.5 kHz frequencies), single threshold > 25 dB in either ear. (Computed values shown as mean ± standard deviations).

Parameter	noTM Group	TM Group	P value
Males	6 (54%)	8 (44%)	= 0.9
Females	7 (46%)	10 (56%)	= 0.9
Age at baseline (yrs.)	28.8 (± 7.7)	31.2 (± 11.6)	= 0.2
Age range	17 – 44	16 – 57
Median days between audiograms	63	64	= 0.9
HFBI (dB)	17.2 (± 21.1)	16.5 (± 21.7)	ns
Single threshold > 25 dB HL (%) 0.25 – 8kHZ	31%	11%	= 0.2
Single threshold > 25 dB HL (%) 0.25 – 16 kHz	46%	67%	= 0.3

Equipment

Testing was completed in a double-walled sound attenuated booth approved for clinical audiometric evaluation at the Oregon Hearing Research Center located at OHSU. A Grason Stadler Instrument (GSI) Tympstar V calibrated to ANSI standards (S3.39–1987 (R2012) was used for assessments of middle-ear status that included 226-Hz tympanometry. A GSI-61 audiometer calibrated to ANSI standards (S3.6–1996) was used to measure pure-tone air conduction (AC) thresholds using standard transducers provided with the audiometer: ER3A insert earphones for conventional frequencies (0.25–8.0 kHz) and Sennheiser HDA 200 circumaural headphones for extended high frequencies (EHF, 9.0–16.0 kHz). Clinical audiometric testing was done in dB HL based on the calibration of audiometers and transducers to meet current standards of clinical care. The ER3A insert earphone was coupled to the ear with a disposable foam eartip that fit securely in the person’s ear canal. Bone conduction (BC) thresholds were measured using a Model #B-71 bone oscillator to estimate the degree of conductive pathology via air-bone-gap (ABG) calculations and/or to confirm sensorineural hearing loss.

To comply with the CF Foundation’s clinical care guidelines and OHSU infection control requirements of CF persons, all participants were asked to wear respiratory masks and providers wore protective disposable gowns and gloves to prevent transfer of bacterial infection. Prior to and following each test session, the audiometric equipment and booth were wiped down using medical grade Cavicide wipe disinfectants. The research team scheduled appointments ~60 minutes apart to allow for complete disinfection and to further reduce the transfer of airborne bacteria between persons with CF.

Study Visit Criteria

All study participants were tested at a baseline and a follow-up post-treatment visit with the same tests described below in Audiometric Tests and Study Procedures. To be included in the study, participants in the TM group had baseline audiograms obtained prior to or no more than 3 days after the initiation of IV tobramycin. In order to isolate evidence of ototoxic hearing shifts to the targeted course of treatment in the TM group, participants recruited for both study groups did not receive IV tobramycin for a minimum of 60 days prior to the baseline hearing test (or initiation of the current course of IV tobramycin for the TM group). This pre-treatment criteria was chosen to minimize the potential for progressive hearing loss due to previous inhaled and/or IV AG therapies in both study groups, however this provision does not eliminate the possibility that a patient may experience on-going ototoxic hearing loss prior to study participation. Test visits were separated by a minimum of 30 days following the last dose of IV tobramycin in the TM group. This allowed time for progressive changes in hearing that may have developed post-treatment, although evidence shows that AGs may stay in cochlear hair cells for up to 6 months . For persons who received repeat audiograms for ototoxic monitoring as part of a clinical standard of care, only audiograms that met the above criteria were considered for data analyses in this study.

Audiometric Tests and Study Procedures

The following clinical tests of auditory function were conducted in a sound-booth using the following test procedures: 1) otoscopy to ensure that the ear canal was clear and free from drainage and that the tympanic membrane was visible 2) tympanometry with a 0.226-kHz probe tone to rule out middle ear dysfunction; 3) pure-tone air conduction audiometry at conventional frequencies 0.25, 0.5, 1.0, 2.0, 3.0, 4.0, 6.0 and 8.0 kHz; 4) pure-tone air conduction audiometry at EHFs: 9.0, 10.0, 11.2, 12.5, 14.0 and 16.0 kHz; and 5) pure-tone bone conduction audiometry at 0.5, 1.0, 2.0 and 4.0 kHz.

Tympanometry using a 0.226-kHz probe tone was conducted prior to pure-tone behavioral hearing assessment. This test was used to obtain information about the transfer of sound energy through the middle ear, called acoustic admittance. Results of this test specify how well the eardrum and middle-ear bones (i.e., ossicles) function compared to normative standards. A normal tympanogram with acoustic admittance between 0.3 – 1.7 mmhos would indicate that the outer ear (i.e., ear-canal) was relatively clear and the middle-ear functioned normally to a low-pitch tone. An abnormal result would indicate potential pathology of the outer ear or middle ear (e.g., fluid behind the eardrum). Ears with abnormal admittance were excluded from data analyses .

Pure-tone hearing thresholds were then measured for conventional audiometric frequencies and EHFs. All testing was completed by a licensed clinical audiologist using the Hughson-Westlake threshold searching technique. Pulsed pure tones were presented for 1–2 seconds, with varying intervals between tone presentations. This procedure uses a 10 dB down, 5 dB up procedure to estimate the hearing threshold. Thresholds were defined as the lowest dB HL at which responses occurred in at least one-half of a series of ascending trials. The minimum number of responses needed to determine thresholds of hearing was two responses out of three presentations at a single level. A threshold of >25 dB HL at one or more audiometric frequencies was considered to be outside the normal hearing range. These criteria are standard of care and meet the current National Guidelines by ASHA for Classification of Mild to Profound Hearing Loss . In the event that the person did not perceive a sound at the maximum output of the audiometer (i.e., no response) hearing threshold was recorded as the maximum output at that frequency plus 5 dB.

Quantifying an Audiometric Shift in Hearing

The following metrics were utilized to evaluate the magnitude of audiometric shifts observed in hearing for the baseline to follow-up visits for both study groups. These are all continuous metrics which are much more robust in detecting the effect size in groups of subjects than the simple ototoxicity “yes/no”, ASHA shift criteria. The ASHA criteria for a shift or meaningful change in hearing is defined as: ≥20 dB pure-tone threshold shift at one frequency, ≥10 dB shift at two consecutive test frequencies, or threshold response shifting to “no response” at three consecutive test frequencies .

• The calculated threshold shift (TS), the difference in hearing threshold (in dB HL) between the follow up and baseline, for each ear and for each frequency band between 0.25 to 16 kHz. (If either of the audiograms had a missing measurement at a given frequency, the threshold shift was “undefined” and was not used in any of the averages or standard errors.)

• For each ear, the largest TS for any frequency, termed the Maximal Single Frequency Threshold Shift (MSFTS), and the highest average value of the TS of adjacent two frequencies, termed the Maximal Adjacent Frequency Threshold Shift (MAFTS) were identified and the value is recorded for the ear with the largest TS. These continuous variables are derived from the ASHA (1994) standard for identifying ototoxicity. A MSFTS ≥ 20 dB would meet ASHA’s first definition of ototoxicity. MAFTS and MSFTS can be used as hearing loss metrics to compare groups of persons without regard to an arbitrary cutoff level of “ototoxicity”.

• The High Frequency Index (HFI) is the average hearing threshold (dB HL) of the 8, 9, 10, 11.2 and 12.5 kHz frequencies. The High Frequency Baseline Index (HFBI) is the HFI of the baseline audiogram.

• The Average High Frequency Threshold Shift (AHFTS) is the difference between the follow-up HFI (average of the 8.0 – 12.5 kHz thresholds) and the baseline HF Index from the ear with the larger TS. This is a metric used to compare hearing loss in the TM group compared to the noTM group. Both HFI and AHFTS metrics did not include threshold information from the 14 and 16 kHz, because it’s common for patients to have no measurable responses at these frequencies, obscuring the effects of treatment.

Statistical Analysis

All data were analyzed in Excel Version 2007. The means, standard deviations, standard error of the mean were calculated for the study groups using a variety of metrics for comparison. The null hypothesis that the mean score of the two groups was equal versus the alternative hypothesis that the treatment caused greater hearing loss was assumed. A chi-square test was used to compare the counts of persons meeting the ASHA criteria for ototoxicity between the two groups. A standard non-paired, one-sided T-test with different sized groups and with different variance was used to compare continuous metrics mean scores with a probability that the difference was random being rejected at the p < 0.05 level.

Results

There were 31 persons enrolled in the study, with 18 in the TM group and 13 in the noTm group (see Table 1). The baseline audiogram of persons recruited for this study were obtained prior to the start of IV-tobramycin treatment or before the 3^rd dose (Day 3) of treatment. The follow-up audiogram was usually completed at the first clinic visit following the treatment, but at least 30 days following the completion of treatment. For the noTM group, the median time between audiograms was 63 days while the TM group median time between audiograms was 64 days.

Figures 1 and 2 demonstrate that several patients in each of the noTM and the TM group had significant hearing loss for the “poorer” ear at baseline, particularly in the EHF range. There was one notable patient ear in the noTM group with a mild sloping to moderate sensorineural hearing loss for 0.25 to 16 kHz. There were also two patient ears in this group that exhibited normal hearing in the conventional range, except an unexplained “notch” in hearing within one frequency band that was in the borderline normal to mild hearing loss range. In comparison, the baseline audiograms for the TM group appeared to have a similar degree of variability in hearing thresholds across subject ears. Figure 3 demonstrates visually that the TM group had greater threshold shifts in the left ear than the noTM group, especially in the range of frequencies covering the AHFTS metric (i.e. 8 – 12.5 kHz). The right ear had similar results.

Figure 1.

Baseline pure-tone hearing thresholds for 0.25 to 16 kHz for the noTM group. All audiograms are plotted for the poorer ear as computed by the HFI (mean thresholds 8 – 12.5 kHz). Green dotted line= normal hearing cutoff at 25 dB HL.

Figure 2.

Baseline pure-tone hearing thresholds for 0.25 to 16 kHz for the TM group. All audiograms are plotted for the poorer ear as computed by the HFI (mean thresholds 8 – 12.5 kHz). Green dotted line= normal hearing cutoff at 25 dB HL.

Figure 3.

Mean change (+ SEM) in hearing thresholds in the left ear between baseline and follow-up visit of the left ear for the noTM group versus the TM group.

Our data in Table 1 (below) indicate that the study groups were not significantly different by gender, age at baseline audiogram, time between audiograms, or on various measures of hearing loss prior to the baseline audiogram. The prevalence of hearing loss at the baseline audiogram was ~46% for both study groups (threshold of > 25 dB HL in a single frequency band from 0.25 to 16 kHz). Both study groups had considerable hearing loss in the standard frequency range (re: 0.25 to 8 kHz) (31% for the noTM group; 11% for the TM group).

Table 2 presents the various metrics used to compare the TM group to the noTM group. Comparisons of the MSFTS, MAFTS and AHFTS were done using a student’s one-side t-test assuming unequal variance. All of the comparisons were statistically significant with the p < 0.05 even though there were only 31 total persons in the study. Compared to the retrospective Gleser & Zettner (2018) ,study, the standard deviations of all of the continuous metrics were much lower in this study, likely due to the much narrower range of dates between audiograms. The ASHA ototoxicity metric was compared using a Chi-square analysis. The ASHA ototoxicity metric was not significantly different between the TM and noTM groups (P>0.05), largely because there were three control patients that met the ASHA shift criteria that had two adjacent TS of exactly 10 dB. If the ASHA shift criteria was for two adjacent TS of >10 dB, there would have been only 13% of noTM patients meeting the criteria and the difference between groups would have been statistically significant. This points out the weakness of using the categorical ASHA shift criteria to compare groups of patients, as opposed to its intended use, i.e. to identify individual patients that might be at risk for ototoxicity.

Table 2:

Continuous metrics (MAFTS, MSFTS, and AHFTS) of hearing loss showing the mean ± standard deviations for each metric. The ASHA shift criteria was used to categorize persons as having ototoxic hearing shifts or no shift in hearing from baseline to follow-up test visits.

Group	MSFTS	MAFTS	AHFTS	ASHA shift in hearing
TM group	14.7 (± 7.5)	11.0 (± 6.6)	5.2 (± 5.0)	39%
noTM group	9.6 (± 3.8)	6.5 (± 2.4)	13 (± 2.4)	31%
t-test p value	0.01	< 0.01	< 0.005
Chi-square p-value				= 0.64

Discussion

Comparing this current prospective study of the effects of a single course of IV tobramycin to the retrospective study (Gleser & Zettner 2018), both studies found that the patients exposed to IV tobramycin suffered significantly more hearing loss as demonstrated by all of the continuous metrics. However, in this prospective study, likely because of the greatly decreased time between baseline and follow up audiograms, there were smaller threshold shifts in both control and treatment groups as well as smaller standard deviations for all metrics. The result of this is that the prospective study predicts the requirement for a much smaller study size to power a clinical trial of an otoprotectant medication.

One might argue that losing an average of 3.9 dB compared to controls with a course of IV tobramycin in the 8 – 12.5 kHz frequency range is not much hearing loss. However, adult persons with CF on the average get one course of IV tobramycin / year and many get 2 – 3 courses. The cumulative effect of this exposure is borne out by the fact that, in this study, CF patients in each group already had lost an average of over 16 dB hearing across this frequency range as measured by their baseline HFBI metric. In fact, 6 of the 31 persons had an HBFI > 40 dB, indicating that they had very little hearing remaining in the high frequency range.

Our results are in line with previous studies that reported on the importance of EHF testing in identifying ototoxic hearing shifts, ,. In this investigation, our reported prevalence of ototoxicity would have been considerably lower if we had limited test frequencies to the conventional hearing range of 0.25 to 8 kHz (See Table 1: 31% in the noTM group and 11% in the TM group). The continuous metrics used in this study provide a sensitive estimate of ototoxic shifts in hearing, and are considered comparable to the commonly used categorical metric called the sensitive range of ototoxicity (SRO) , which ultimately depends on the ASHA shift criteria for estimating a meaningful change in hearing. The SRO range is typically determined on an individual basis, and includes the highest frequency with a threshold ≤100 dB SPL and the next four lower frequencies. The range is often similar to the AHFTS frequency range in the current study, consisting of five frequencies, generally separated by 1/6 octaves, e.g., 8, 9, 10, 11.2, and 12.5 kHz. Thus, the combination of metrics in the current study are considered comprehensive for identifying meaningful hearing shifts in patients, including shifts in the SRO region.

There are some potential limitations with the current study, which may limit our understanding of cumulative AG exposure and the progressive effects on hearing within our study time frame. As previously discussed, we modeled our study design and cohort to closely resemble Gleser & Zettner (2018), with the intent to determine if their results were replicable in a prospective study design. We actually created a more stringent criteria of no IV-AGs for < 60 days prior to baseline for all study patients, to further reduce the progressive effects of previous IV-AG therapy on hearing thresholds during the controlled study time. Although this provision does not guarantee that there is no on-going loss of hearing from previous inhaled and/or IV-AG exposure, such an effect would not be expected to differ between test groups, as the rule was applied to every potential baseline audiogram without regard to whether the patient would fall into the noTM or TM cohort. Although patients likely have a varied history of cumulative AG dosing, it’s difficult to predict who may develop a late progressive shift in hearing given our current understanding of predictive modeling of ototoxic hearing loss.

Life expectancy of persons with CF has dramatically improved over the last 30 years, primarily due a better understanding of the disease process, advancements in airway clearance techniques and treatments of respiratory infections, as well as other CF-related complications. Now there are several new CF transmembrane conductance regulator modulator therapies which function at the cellular level to change the underlying cause of the disease and will continue to extend the lives of persons with CF. However, there is of yet no evidence that the rate and intensity of pulmonary exacerbations necessitating IV tobramycin treatment will decrease, strengthening the argument that an increased number of persons with CF will experience hearing loss over their extended lifetimes. Thus, finding a clinically useful medication which prevents the ototoxic effects of AGs remains an important goal. Furthermore, it is equally as crucial to improve our current ototoxicity monitoring and management programs by using optimal metrics (e.g., EHF) for identifying ototoxic hearing loss in at-risk patients.

Highlights.

• A single course of IV tobramycin results in a statistically significant loss of hearing threshold in adults with CF.

• Because of the cumulative effect of hearing loss, 46 – 67% of CF patients already demonstrate hearing loss in the EHF by average age 29 years old.

• The ASHA shift measurement provides a good screening mechanism for an individual patient, but may not be an ideal metric for comparing ototoxic hearing loss across groups of patients.

• Ototherapeutic treatments would be useful for protecting the hearing of patients with cystic fibrosis treated with aminoglycosides.