REEVALUATION OF SODIUM THIOSULFATE OTOPROTECTION USING THE CONSENSUS INTERNATIONAL SOCIETY OF PEDIATRIC ONCOLOGY OTOTOXICITY SCALE: A REPORT FROM THE CHILDREN’S ONCOLOGY GROUP STUDY ACCL0431

Abstract

In two randomized trials (ACCL0431, SIOPEL-6), sodium thiosulfate (STS) demonstrated efficacy in preventing cisplatin-induced hearing loss (CIHL). However, the measures used in those trials have been superseded by the consensus Society of International Paediatric Oncology (SIOP) Ototoxicity Scale. To provide benchmark data for STS efficacy when using this contemporary scale, we reanalyzed ACCL0431 hearing outcomes with the SIOP scale and using multiple timepoints. As compared to the control arm, STS significantly reduced CIHL when assessed by the SIOP scale across these different approaches. These results provide critical data to inform treatment discussions and support future potential trial designs comparing otoprotectants.

Introduction

Cisplatin is essential for treating many pediatric and adult malignancies. However, cisplatin-induced hearing loss (CIHL) is common, irreversible, and broadly impacts quality of life in childhood cancer survivors.^1–5 Recently, sodium thiosulfate (STS) was shown to protect against CIHL in two independent randomized controlled trials, the Children’s Oncology Group (COG) ACCL0431 and International Childhood Liver Tumour Strategy Group SIOPEL-6 trials.^6,7 These trials led to regulatory approval of STS for prevention of CIHL, the first drug for this indication. However, each trial used a different hearing endpoint to evaluate STS efficacy.^8,9 Discordance in ototoxicity prevalence by grading system,^10–12 such as was performed for these two pivotal trials, complicates inter-study comparisons and developing benchmarks for testing future agents. In 2010, the International Society of Pediatric Oncology (SIOP) Ototoxicity Scale was developed to overcome key limitations of available ototoxicity grading systems. This international, consensus scale incorporated the best features of existing systems in a cross-sectional, post-exposure approach with increased sensitivity to detect communicatively significant hearing loss. The SIOP scale has become central to harmonizing hearing endpoints across multinational cohorts.^10,13

We therefore conducted a re-evaluation of hearing outcomes on ACCL0431 to address a key knowledge gap: what is the extent of anticipated hearing protection from STS when measured against the SIOP Ototoxicity Scale? The answer to this question will facilitate interstudy comparisons of future otoprotectants with STS, inform planning of collaborative otoprotective trials, and support provider-family discussions concerning the risk of CIHL.

Methods

This was a secondary analysis of audiology data collected in ACCL0431. As previously described,⁶ ACCL0431 enrolled patients 1-18 years of age with any newly-diagnosed cisplatin-treated tumor and normal hearing. Patients were randomized 1:1 between STS rescue and standard of care (observation), stratified by age, prior cranial radiation, and duration of cisplatin infusion. All patients were treated with cisplatin according to their disease-specific regimen. STS (16 gm/m²) was infused over 15 minutes beginning 6 hours post-completion of each cisplatin dose. Following assessment of middle-ear function, age-appropriate audiometry was performed at 0.5, 1,2,3,4,6, and 8 KHz. Substitution of frequency-specific auditory brain response was permitted. Hearing was assessed at baseline, serially during treatment, at end of cisplatin and prior to autologous bone marrow transplantation (ABMT), where applicable, and a final protocol-specified evaluation approximately 12 months following completion of chemotherapy or ABMT.

For this analysis, central review of all hearing evaluations using the SIOP Ototoxicity Scale was performed by an audiologist investigator (KRK) blinded to randomized allocation. Ototoxicity grade was assigned by the better ear or by sound field, if ear-specific testing was not possible. If only one ear was evaluable, ototoxicity was determined by that ear. To establish various benchmarks for STS otoprotection using the SIOP Ototoxicity Scale, hearing endpoints were evaluated here using three approaches (1) hearing loss during the trial assessment period, and prior to ABMT where applicable, (2) hearing loss at end of cisplatin therapy and prior to ABMT (to replicate the ACCL0431 trial primary endpoint), and (3) hearing loss as a time-to-event analysis, censored at time of ABMT. Hearing thresholds of SIOP Grade≥2 and Grade≥1 were evaluated (Supplemental Table S1). For the first two approaches, prevalence of SIOP Grade ≥2 and Grade ≥1 CIHL were reported for patients treated with STS versus observation. We estimated the magnitude of the association between STS assignment and the CIHL endpoint using the odds ratio (OR); p values for the test of OR=1 and corresponding 95% confidence interval (95%CI) were derived using the Wald test for the parameter associated with the randomized treatment assignment from a logistic model.¹⁴ The logistic model was stratified according to the original randomization strata. For the third approach, time to event was defined as the interval from enrollment date to the date of first audiometry evaluation with SIOP Grade ≥2 CIHL, or if SIOP Grade ≥2 was not identified, the date of the final audiometry (and prior to ABMT where applicable) The distribution of time to CIHL event was estimated by the method of Kaplan and Meier. Homogeneity of risk for a CIHL event across randomized regimen was assessed using the logrank test. A p-value of <0.05 was considered indicative of a statistically significant relationship.

Results

Characteristics of the ACCL0431 cohort have been described.⁶ Following repeat audiological central review, 121 of 125 (92%) patients were evaluable for hearing loss using the SIOP scale (Supplemental Table S2). Using the first approach, a lower incidence of Grade ≥2 CIHL occurred in the STS versus observation arm (3/58 [5.2%] versus 18/63 [28.6%], Figure 1A). With the second approach, in analyses restricted to the end of cisplatin therapy/prior to ABMT timepoint, similar findings were present for the proportion of patients with SIOP Grade ≥2 CIHL between trial arms (Figure 1B). After adjustment for stratification variables, the odds of developing SIOP Grade ≥2 was significantly lower for patients in the STS arm for both approaches (approach #1: OR 0.13, 95%CI 0.03-0.48, p=0.002; approach #2: OR 0.10, 95%CI 0.02-0.50, p=0.005). ; The same pattern was seen for SIOP Grade ≥1 CIHL (Supplemental Figures S1, S2). Using the third approach’s time-to-event analysis, patients randomized to STS were also at significantly reduced risk for developing SIOP Grade ≥2 hearing loss (Figure 2, p=0.0024.

Figure 1: Influence of STS on SIOP Grade ≥2 Hearing Loss.

Of SIOP-evaluable patients enrolled in the ACCL0431 trial, fewer patients developed SIOP Grade ≥2 hearing loss on the STS versus the observation arm when evaluated: (A) during the trial period (from stratified logistic regression, odds ratio 0.13, 95% confidence interval [CI] 0.03-0.48, p=0.002), and (B) at end of cisplatin therapy (odds ratio 0.10, 95%CI 0.02-0.50, p=0.005).

Figure 2: Time to SIOP Grade ≥2 Hearing Loss.

Kaplan-Meier analysis compared time to SIOP Grade ≥2 hearing loss between patients in the STS (dashed line) and observation/control (solid line) arms (log rank test p=0.0024)

Discussion

Reanalysis of hearing outcomes from the ACCL0431 trial demonstrated evidence for effective otoprotection from STS using the SIOP Ototoxicity Scale. As compared to the observation group, children treated with STS were approximately 90% less likely to develop Grade ≥2 CIHL during the trial period or at the end of cisplatin therapy. Similar findings were seen in the time-to-event analysis. These data are consistent with the primary ACCL0431 analysis demonstrating efficacy for STS otoprotection.⁶ As the SIOP scale is now a widely accepted ototoxicity grading system, these findings enable ACCL0431 results to provide a benchmark for evaluating efficacy of future otoprotectants and to inform otoprotection trial planning across different study designs for hearing endpoints.

Importantly, though use of the SIOP scale enabled only five more participants (4% of the randomized ACCL0431 cohort) to be evaluable for the ACCL0431 trial’s primary hearing endpoint, 121 of 125 participants (97% of the randomized cohort) were evaluable when hearing was assessed throughout the trial period and in the time-to-event analysis. Timing of hearing evaluation is a critical consideration for otoprotectant trial design. ACCL0431 required baseline audiology testing, but very young, sick children are often unable to complete baseline assessments, or require testing with different modalities, which may make them unevaluable at later time points under criteria requiring pretreatment comparisons. Similarly, cisplatin-treated tumors often involve intensive multiagent chemotherapy, and patients may be unevaluable for hearing at a specific timepoint due to logistical challenges or transient chemotherapy morbidity. Loss of such participants exacerbates the difficulties posed by small sample sizes in pediatric otoprotection trials. Future trial design considerations will need to address whether to include only a single post-treatment evaluation time-point or, as hearing loss is permanent, hearing loss during cisplatin therapy. It is also important to note that these considerations relate solely to hearing evaluations as an efficacy endpoint in trials; routine age-appropriate audiology monitoring should occur at diagnosis and throughout treatment, as is strongly recommended within consensus guidelines for monitoring ototoxicity.¹⁵

There are two salient limitations. First, the ACCL0431 audiology assessment approach was not designed specifically for the SIOP scale (i.e., prioritizing the grade-defining hearing frequencies during testing) resulting in several unevaluable patients. Trials designed for the SIOP scale would further minimize this source of data loss.¹⁶ Second, the time-to-event analysis relied upon serial measurements at regimen-defined time-points and CIHL may have occurred in the intervals between assessments. However, the short assessment intervals and large difference observed between treatment arms make a significant impact unlikely. In conclusion, this reanalysis using the SIOP Ototoxicity Scale validates the otoprotective efficacy of STS with this contemporary scale and establishes important benchmarks informing design considerations for future otoprotection trials.

Abbreviation List:

CIHL

Cisplatin-induced hearing loss

COG

Children’s Oncology Group

95%CI

95% Confidence Interval

OR

Odds ratio

SIOP

International Society of Pediatric Oncology

SIOPEL

International Childhood Liver Tumour Strategy Group

STS

Sodium thiosulfate