The search was conducted by an experienced librarian in January 2021 and updated in April 2022. The MEDLINE/PubMed, EMBASE, Web of Science, LILACS/VHL, Cochrane Library, and Brazilian Digital Library of Theses and Dissertations databases were used, with the descriptors shown in Table 1. Experts consulted for further article search.

Two independent reviewers evaluated the eligibility of the studies published by April 2022 without language or year of publication restrictions. Articles published in a language other than Portuguese or English were evaluated with the help of a translator. Disagreements were resolved by a third reviewer.

Studies could be included if they met the following criteria: 1) population: adults (>20 years old) of both sexes that underwent radiosurgery for vestibular schwannoma treatment; 2) exposure: cochlea radiation; 3) outcome: decline of hearing function, quantified objectively with audiometric assessment; 4) type of study: prospective or retrospective cohort studies.

Studies containing samples of participants treated with hypofractionated radiosurgery or undergoing prior conventional radiotherapy or drug treatment to maintain auditory thresholds were excluded. Studies involving patients with anacusis before treatment or with neurofibromatosis type 2 were also excluded.

Study selection was performed with the Rayyan application, with which two independent reviewers analyzed the titles and abstracts of all studies identified from the search results. Studies that did not meet the eligibility criteria were excluded. The complete text of the remaining studies was retrieved and reevaluated by the two reviewers, also independently and using the same application. Studies that were not published could be included only if they presented the necessary data in the abstract. Disagreements between the two reviewers during the selection stage were resolved by a third evaluator. The studies in this stage were sorted into three categories: 1) included studies; 2) excluded studies, and 3) studies that needed more information to assess their eligibility. The authors of the studies in the third category were contacted and asked to provide the necessary information; in the impossibility of obtaining the answers, the studies were excluded from the review.

The Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines were used to guide the elaboration of the study selection flowchart.27

The two main reviewers independently completed a custom data extraction form. It included the following information: study title, last name of the first author, publication year, the country where the study was conducted, study type, study population, participants’ characteristics (gender, age), tumor characteristics (size, laterality), follow-up time, treatment characteristics (therapeutic dose, cochlear dose, previous microsurgery) and audiological characteristics before and after radiosurgery (Gardner–Robertson classification; PTA).

The certainty of the evidence was evaluated by two independent reviewers using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE), according to which three factors can increase the certainty of the evidence of observational studies: magnitude of effect, dose-response gradient, and all potential confounding factors, which increase confidence in the estimated effect. On the other hand, the certainty of the evidence can be lowered by five factors: risk of bias, inconsistency, indirectness, imprecision, and publication bias.28–32

For the risk of bias assessment, the same reviewers conducted the analysis using the Newcastle–Ottawa Quality Assessment Scale (NOS) for cohort studies.

A descriptive approach was carried out with text and tables for data synthesis. The results compilation table were made by two independent blind reviewers and contains clinically relevant information on population, sex, age, tumor size and laterality, follow-up time, previous microsurgery performance, Gardner–Robertson and PTA classifications at the beginning of the cohort, radiation therapy dose, as well as information on mean cochlear dose in both patients groups with or without hearing loss and/or the number of patients exposed to a high cochlear radiation dose that progressed to loss of hearing function.

In total, 333 articles were identified in the database search. 166 duplicate records were excluded. After evaluating titles and abstracts based on eligibility criteria, 122 articles were excluded. Of the 45 remaining articles, 13 did not have available full texts. 32 papers were read in full, 25 of which were excluded for the following reasons: Age group included participants <20 years old (11 articles), treatment had fractionated radiotherapy and not just single-dose radiosurgery (3 articles), study exposure was not cochlear dose (2 articles), clinical treatment was performed after radiosurgery (3 articles), included participants with anacusis (1 article), included participants with cochlear nerve schwannoma (1 article), lacked information on the eligibility criteria and it was not possible to obtain a response from the lead author (3 articles did not report whether treatment was performed before radiosurgery and 1 article did not determine the age group of the population).

Finally, seven studies were included in the review11,13–16,25,33 (Fig. 1).

Figure 1.

Flowchart of article selection steps following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA), 2020. Source: created by the author using the PRISMA (2020).

(0.62MB).

Table 2 shows the characteristics of the included studies published in four countries from 2005 to 2019: six retrospective cohorts and one prospective cohort. The studies included patients aged 21–87 who underwent a radiation therapy dose of 9–14Gy, with an audiometric follow-up after radiosurgery from 0.77 to 141 months. Only two studies16,25 included patients with non-usable hearing at the beginning of the cohort.

Regarding exposure, five studies evaluated the mean cochlear radiation dose.13–16,33 Of these, two additionally assessed the maximum cochlear dose.14,16 Two studies analyzed exposure based only on maximum cochlear dose.11,25 The sample was dichotomized in four studies, based on exposure,13,15,25,33 with divergent cut-off points: mean cochlear dose of 3Gy (1 study), mean cochlear dose of 5Gy (1 study), mean cochlear dose of 4Gy (1 study) and maximum cochlear dose of 4Gy (1 study).

Four studies used the Gardner–Robertson classification for outcome assessment, considering a decline to classes III, IV or V as loss of usable hearing.13–15,33 Two studies used the PTA difference ≥20dB11,16 as a criterion for the absence of hearing preservation. One study used the definition of Annual Hearing Decrease Rate (AHDR) to assess the outcome.25 The population was dichotomized based on the outcome of three studies, which evaluated the mean cochlear dose (mean or maximum) in groups with and without preservation of hearing function.

The summary of the risk of bias analysis for the cohort studies included in this review is shown in Table 3.

Most studies involved representative populations, except for one,11 which did not clearly describe the cohort selection. The unexposed cohort was drawn from the same community as the exposed in all studies, and all exposure records were safe, coming from the radiosurgery programming software. Not usable hearing was present at the beginning of the cohort in only two studies.16,25

It was not possible to judge the comparability between exposed and unexposed cohorts in any of the studies.

Four studies extracted the outcome from secure medical records with detailed audiometry.11,13,15,16 The others used databases to store and access information. Three studies performed an audiometric follow-up of more than 12 months11,14,15 and none of the studies described the adequacy of the follow-up of the cohorts.

It was only possible to adequately assess the certainty of evidence with GRADE for useful hearing loss measured by the Gardner–Robertson classification at exposure to a mean cochlear dose of >4Gy (Table 4). Only one study15 carried out this evaluation, with a total number of participants of 77, with 50 from the control group and 27 from the group exposed to high cochlear dose. Follow-up ranged from 24 to 80 months, and the relative risk for hearing loss was 1.44 (95% CI 0.6–3.44).

The risk of bias was serious due to the lack of adjustment for confounders. The outcome was measured in a single study, so inconsistency assessment was not possible, which was therefore categorized as severe. There was no indirect evidence or publication bias, but there was imprecision, which lowered the certainty of the evidence at two levels due to the low number of events and the confidence interval that crosses the line of statistical significance. None of the domains raised the level of evidence, and the certainty of evidence was very low.

There was no standardization regarding measuring exposure and outcome in the studies included in this review. Only the maximum cochlear dose was evaluated in two studies. Two studies assessed maximum cochlear dose only,11,25 three assessed mean cochlear dose only13,15,33 and two evaluated both.16,33 Additionally, only four studies defined a cut-off point for the characterization of “high dose”13,15,25,33 and there was a divergence between the adopted cut-off points (average cochlear doses of 3–5Gy and a maximum cochlear dose of 5Gy). Regarding hearing assessment, one study used the AHDR25 as a criterion for hearing decline, and two used changes in the PTA testing.11,16 The others used the drop in the Gardner–Robertson class from I or II to III, IV or V.

The studies did not provide sufficient information about the exposed and unexposed groups. Only one study15 reported the absolute number of participants in each cohort and the patients who showed the hearing decline outcome. For this study, the cut-off point adopted for the definition of “high dose” was a mean cochlear dose of 4Gy, and the relative risk for the outcome was 1.44 (95% CI 0.60‒3.44). Therefore, it was not possible to perform a meta-analysis.

As limitations of this review, it is essential to note that only observational studies were included, which usually leads to less certainty of evidence. However, this limitation stems from the authors’ proposal to assess exposure, so that is the most appropriate type of study found for that matter. Furthermore, every systematic review is susceptible to the methodological limitations of the original articles. The incomplete presentation of data from the cohorts is highlighted, which leads to publication bias since the data were hidden, especially in the analyses that did not find statistical significance, with small samples and without favorable results. It may also represent publication bias, the 13 records excluded for being only abstracts, without published article. Such abstracts did not have the necessary data for inclusion in this review.

In addition, the cochlear dose measurement methodology, as well as the definition of high dose, and the way of measuring auditory function were quite heterogeneous, making it impossible to carry out a meta-analysis.

The authors are unaware of a previous systematic review that evaluated cochlear radiation dose and decline in auditory function in patients with vestibular schwannoma undergoing radiosurgery. One strength of the study is the methodological rigor with which the review was conducted, in compliance with the MOOSE guidelines. The definition of strict eligibility criteria, taking care of defining clinically relevant outcomes based on objective audiometry data, reflects the effort to seek relevant results for patients in medical practice.

It is suggested that future studies adopt the mean cochlear dose to measure exposure, as it reflects the radiation that reaches the entire cochlear volume and not just a specific point in the cochlea. Determining a cut-off point from which there is a significant decline in hearing function is crucial for the clinical management of these patients and can potentially optimize the audiological prognosis in radiosurgery. Evaluating the outcome should also be standardized, and the Gardner–Robertson classification is considered a clinically relevant outcome, as it considers tonal thresholds and speech recognition.