Abstract

Objective:

Aminoglycoside antibiotics remain essential for treating serious neonatal and pediatric infections, yet carry a well-documented risk of permanent auditory and vestibular toxicity. This review examines the pharmacological mechanisms of ototoxicity in pediatric populations, identifies those at highest risk, and assesses current prevention, monitoring, and management strategies.

Methods:

PubMed, EMBASE, Cochrane Library, Web of Science, and CINAHL were searched for relevant literature from 2000 to 2025. The primary focus was pediatric populations, though mechanistic and translational pharmacology work from other age groups was included.

Results:

The ototoxicity pathway is increasingly well characterized: aminoglycosides accumulate in cochlear hair cells via mechanoelectrical transduction channels, disrupt mitochondrial function, trigger oxidative stress, and cause cell death through apoptotic, necroptotic, and ferroptotic mechanisms. Susceptibility varies substantially. Patients carrying the MT-RNR1 m.1555A>G pharmacogenomic variant face a markedly elevated risk of profound hearing loss even with a single standard course. Preterm neonates are at higher risk than older children owing to renal immaturity, altered volume of distribution, and incomplete blood-labyrinth barrier development. Co-administration of loop diuretics and vancomycin further amplifies ototoxic risk. Extended-interval dosing is associated with equivalent efficacy and reduced nephrotoxicity, with a non-significant trend toward lower ototoxicity in pooled analyses. Point-of-care genetic screening allows identification of high-risk patients before the first dose, though debate continues over universal versus targeted implementation. Model-informed dosing approaches, including Bayesian forecasting and AUC-targeted monitoring, offer individualized pharmacokinetic optimization but remain underutilized. Antimicrobial stewardship and minimizing concomitant ototoxin exposure are complementary strategies. When ototoxicity occurs, early audiological and vestibular identification enables timely intervention through hearing aid fitting, cochlear implantation, vestibular rehabilitation, and family-centered support, though vestibular ototoxicity remains widely under-recognized in pediatric populations.

Conclusion:

Evidence-based interventions to reduce aminoglycoside ototoxicity in children exist, including pharmacogenomic screening and dosing optimization, as well as structured monitoring and rehabilitation. However, a persistent gap remains between available evidence and routine clinical implementation. Key research priorities include pediatric otoprotective trials, validated cochlear injury biomarkers, and implementation strategies for diverse healthcare settings. Given the permanence of aminoglycoside-induced ototoxic injury and its downstream effects on speech, language, and developmental outcomes, closing this gap represents an urgent clinical priority.

Introduction

Thirty-four million children worldwide have hearing loss, with an estimated 60% of these cases potentially preventable (WHO, 2024). Aminoglycosides account for a meaningful fraction of preventable childhood hearing loss (Lanvers-Kaminsky and Ciarimboli, 2017; Diepstraten et al., 2021; Saunders et al., 2009). Aminoglycosides remain a cornerstone of empirical therapy and are among the most frequently prescribed antibiotics for suspected serious gram-negative infections in newborns (Ting et al., 2016; Cantey et al., 2015; Tamma et al., 2012). Yet they damage the inner ear in ways that cannot be undone (Forge and Schacht, 2000; Huth et al., 2011). Clinicians face this trade-off daily. The consequences are not trivial: unlike aminoglycoside nephrotoxicity, which typically resolves after drug discontinuation, ototoxicity produces permanent sensorineural hearing loss (Diepstraten et al., 2021; Schacht et al., 2012; Forge and Li, 2000). The scale of exposure compounds the concern, and the patients receiving aminoglycosides are precisely those most vulnerable to their toxic effects. Preterm and critically ill neonates are at particular risk, for reasons discussed in the Neonatal Vulnerability section below (Zimmerman and Lahav, 2013; Suzuki et al., 1998).

Most of the existing literature addresses adults, focuses narrowly on the mechanism, or ignores the developmental pharmacology that makes pediatric patients different (Diepstraten et al., 2021). Where pediatric data do exist, they are fragmented across studies that vary widely in monitoring intensity, diagnostic thresholds, and population characteristics, making synthesis difficult and clinical translation uncertain (Lanvers-Kaminsky and Ciarimboli, 2017; Diepstraten et al., 2021; Garinis et al., 2017a; Jacqz-Aigrain et al., 2012). This review brings these threads together, aiming to provide practical, age-specific guidance for clinicians managing aminoglycoside therapy in neonates and young children, integrating mechanistic, pharmacokinetic, and clinical evidence across developmental stages -- distinctions that adult-derived guidelines consistently overlook (Table 1).

Evidence categoryTypes of studiesEvidence level*StrengthsLimitationsMechanistic studiesIn vitro cochlear models, animal studies, and molecular biology investigationsPreclinicalProvides a detailed understanding of cellular pathways and temporal progressionLimited direct translation to clinical settings; species differencesGenetic studiesMitochondrial DNA analyses, genome-wide association studies, and pharmacogenetic investigationsPreclinical/Observational clinicalIdentifies specific risk factors; allows targeted screeningVariable penetrance; incomplete understanding of gene–environment interactionsClinical observational studiesCohort studies, case-control studies, retrospective analysesObservational clinicalDocuments real-world incidence, risk factors, and outcomes in pediatric populationsConfounding variables; heterogeneity in definitions and outcome measuresInterventional researchClinical trials of protective agents, monitoring protocols, and dosing strategiesRandomized clinical/Observational clinicalProvides direct evidence for preventive approachesLimited randomized controlled trials in pediatric populations due to ethical constraints; ototoxicity endpoints are inconsistently measured across studiesGuidelines and expert opinionConsensus statements, professional society guidelinesGuideline consensusSynthesizes evidence into practical recommendationsMay lag behind the most recent evidence; varies by region

Characterization of evidence included in this narrative review.

*

Evidence Level categories: Preclinical = in vitro and/or animal model data; Observational clinical = cohort, case-control, or retrospective human studies; Randomized clinical = randomized controlled trials; Guideline consensus = professional society consensus statements or guidelines. Where a category spans multiple evidence levels, both are listed.

BackgroundEpidemiology and clinical burden

Reported ototoxicity rates range from 2% to 63% across pediatric populations, reflecting differences in monitoring intensity, diagnostic thresholds, follow-up duration, and population risk profiles (Lanvers-Kaminsky and Ciarimboli, 2017; Diepstraten et al., 2021; Garinis et al., 2017a). Genetic susceptibility, particularly the MT-RNR1 m.1555A>G variant, substantially elevates individual risk even at standard therapeutic doses (McDermott et al., 2022a; McDermott et al., 2022b), as discussed in the Genetic Susceptibility section below. In the United States, the CDC National Vital Statistics System, which collects data from state birth certificate records, indicates that approximately 9% of newborns are admitted to neonatal intensive care units annually (8.7%–9.6% between 2016 and 2021) (Martin and Osterman, 2025), with rates varying by maternal age and other factors (Gamber et al., 2024). Aminoglycosides, especially gentamicin, are among the most commonly prescribed antibiotics in NICUs (Ting et al., 2016; Cantey et al., 2015) due to their efficacy against gram-negative pathogens and their synergistic activity with β-lactams (Tamma et al., 2012).

The inner ear is one of the three anatomical parts of the ear, consisting of the cochlea and the vestibular apparatus (semicircular canals, utricle, and saccule), which together constitute the membranous labyrinth, a fluid-filled (endolymph) system of sensory structures (Bruss and Shohet, 2026). Within the cochlea, the organ of Corti contains two functionally distinct populations of sensory cells: a single row of inner hair cells, which serve as the primary afferent transducers of sound, and three rows of outer hair cells, which amplify basilar membrane motion through electromotility (Forge and Schacht, 2000; Schacht et al., 2012). The vestibular apparatus contains separate mechanosensory receptors, Type I and Type II vestibular hair cells, which detect angular and linear acceleration (Schacht et al., 2012). The anatomical basis for aminoglycoside ototoxicity centers on the cochlea and vestibular organs (Figure 1) (Forge and Schacht, 2000; Schacht et al., 2012). Outer hair cells are especially vulnerable, and damage proceeds in a characteristic base-to-apex gradient, affecting high-frequency hearing first (Huth et al., 2011). This tonotopic pattern reflects both increased expression of MET channels in basal hair cells (Pan et al., 2018; Alharazneh et al., 2011) and their higher metabolic demands (Sha and Schacht, 1999). Type I vestibular hair cells show similar susceptibility (Forge and Li, 2000). Mammalian cochlear hair cells cannot regenerate (Forge and Li, 2000), and damage may progress even after treatment cessation due to persistent drug retention (Hailey et al., 2017) and ongoing oxidative stress (Sha and Schacht, 1999). Gentamicin primarily causes vestibular damage, while amikacin, kanamycin, and neomycin mainly affect the cochlea (Forge and Schacht, 2000; Huth et al., 2011; Selimoglu, 2007). These agent-specific toxicity patterns continue to inform drug selection in clinical practice (Forge and Schacht, 2000; McDermott et al., 2022b). Economic analyses estimate lifetime costs exceeding $1 million per affected individual when accounting for educational support, lost productivity, and healthcare utilization (Mohr et al., 2000). More recent modeling confirms that the societal economic burden remains substantial, with annual US costs estimated at $37 billion (Cejas et al., 2024).

Anatomy of the Inner Ear and Organ of Corti Relevant to Aminoglycoside Ototoxicity. (A) The membranous labyrinth (purple) is housed within the bony labyrinth (pale blue) and comprises the cochlea, semicircular canals, utricle, saccule, and endolymphatic sac. The cochlea transduces sound-induced vibrations into neural signals transmitted via the cochlear nerve; the semicircular canals and otolithic organs (utricle and saccule) subserve vestibular function. Both the endolymph and perilymph compartments accumulate aminoglycosides following systemic administration. (B) Cross-section through one cochlear turn showing the three fluid-filled scalae (scala vestibuli, scala media, scala tympani), Reissner’s membrane, the tectorial membrane, the basilar membrane, and the organ of Corti containing the mechanosensory hair cells. A single row of IHCs and three rows of OHCs are arranged along the full length of the basilar membrane. MET channels at the tips of the stereocilia admit K+ and Ca2+ upon deflection, depolarizing the hair cell and initiating afferent signaling. The unrolled basilar membrane schematic (center) illustrates the tonotopic arrangement of three OHC rows and one IHC row (amber, ×1) from the high-frequency cochlear base to the low-frequency apex, with audiometric frequency positions mapped to cochlear location using the Greenwood function. Colour coding of the OHC rows reflects cumulative aminoglycoside-induced hair cell loss by exposure zone: dark red (>8 kHz; basal outer hair cells, lost at low cumulative doses), orange (4–8 kHz; moderate cumulative doses), and amber (≤2 kHz; high cumulative doses, with onset of IHC involvement). The representative audiogram (lower left) illustrates the characteristic high-frequency-first pattern of aminoglycoside-induced hearing threshold elevation compared with normal hearing. OHC loss in mammals is irreversible. Aminoglycoside-induced vestibulotoxicity additionally affects hair cells of the crista ampullaris at the base of each semicircular canal. Abbreviations: IHC, inner hair cell; MET, mechanoelectrical transduction; OHC, outer hair cell; SCC, semicircular canal. Created in BioRender. Bitner-Glindzicz et al. (2009) https://BioRender.com/d58o4vz.

Neonatal Vulnerability

Age makes a significant difference. Neonates are especially vulnerable due to renal immaturity, altered volume of distribution, and incomplete blood-labyrinth barrier development, which together increase both drug accumulation and cochlear susceptibility. NICU infants have approximately 10-fold higher rates of hearing loss compared to well-baby nursery populations (2%–4% vs. 0.1%–0.3%), with prematurity and low birth weight consistently identified as independent risk factors (Zimmerman and Lahav, 2013; Al-Ani, 2023). The pharmacokinetic and physiological basis for this vulnerability is discussed in detail in the Age-Related Vulnerabilities section below.

Literature search strategy

This narrative review is based on a structured literature search of PubMed, EMBASE, Cochrane Library, Web of Science, and CINAHL for English-language articles published between January 2000 and March 2025. Search terms included “aminoglycosides,” “ototoxicity,” “hearing loss,” “cochleotoxicity,” “vestibulotoxicity,” “vestibular,” “pediatric,” “neonate,” “infant,” “pharmacogenomics,” “MT-RNR1,” “mitochondrial,” and “therapeutic drug monitoring,” used in various combinations. Seminal publications predating the search window were included if they provided foundational mechanistic, pharmacological, or historical context that was not superseded by subsequent work.

All retrieved records were imported into EndNote reference management software (n = 273). Duplicate records were identified and removed using EndNote (n = 34). The remaining records (n = 239) were screened based on title and abstract, followed by iterative full-text review. We prioritized studies addressing pediatric populations and translational mechanistic research and examined the reference lists of included articles for additional relevant publications. Grey literature, conference abstracts, and non-English-language publications were not systematically included.

Studies were included based on relevance to predefined thematic areas, including mechanistic pathways of ototoxicity, pharmacokinetic and pharmacodynamic considerations, genetic susceptibility, and clinical outcomes. Studies were excluded if they did not address aminoglycoside-associated ototoxicity, lacked relevance to pediatric or neonatal populations, did not include clinically or mechanistically relevant outcomes, or were non-primary reports (e.g., editorials or conference abstracts without sufficient detail).

Age categories followed International Council for Harmonisation (ICH) E11 (R1) guidelines (International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use ICH, 2017): neonates (0–27 days), infants (28 days–23 months), children (2–11 years), and adolescents (12–18 years). We further stratified neonates by gestational age into preterm (<37 weeks) and term (≥37 weeks) subgroups to reflect the distinct pharmacokinetic and maturational considerations relevant to aminoglycoside disposition in this population.

This synthesis integrates mechanistic, clinical, and translational evidence without formal systematic review methodology or Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting. Screening and eligibility assessment were conducted iteratively rather than as discrete, prospectively recorded stages; accordingly, stage-specific counts are not reported. Given the breadth of the topic and clinical heterogeneity in study designs, ototoxicity definitions, and audiometric endpoints across the literature, formal meta-analysis was not attempted. A total of 132 studies were included in the final synthesis. An adapted literature search and study selection workflow is provided to enhance transparency (Figure 2).

Adapted literature search and study selection workflow. A total of 273 records were identified through database searching of PubMed, EMBASE, Cochrane Library, Web of Science, and CINAHL, and imported into EndNote reference management software. Duplicate records were removed (n = 34), and the remaining records (n = 239) were screened by title and abstract followed by iterative full-text review. Records were excluded if they did not address aminoglycoside-associated ototoxicity, lacked relevance to pediatric or neonatal populations, did not include clinically or mechanistically relevant outcomes, had insufficient methodological detail to support inclusion, or were non-primary reports (editorials or conference abstracts). A total of 132 studies were included in the final qualitative synthesis. Because this review was conducted as a structured narrative review, screening and eligibility assessment were not conducted as discrete, prospectively recorded stages; accordingly, stage-specific record counts are not reported and the diagram reflects the overall selection process rather than sequential PRISMA-compliant stages. Record counts are reported for transparency rather than as reproducible PRISMA-compliant outputs. Exclusion reasons are presented as categories without individual counts, consistent with the iterative methodology described in the Literature Search Strategy section. Figure adapted from the PRISMA 2020 framework (Page et al., 2021).

Mechanisms of aminoglycoside-induced ototoxicityCellular entry and accumulation

Figure 3, Table 2 summarize the mechanistic cascade from initial drug uptake to terminal hair cell loss. Aminoglycosides enter outer hair cells principally through apical MET channels and, to a lesser extent, via megalin-mediated endocytosis (Huth et al., 2011; Pan et al., 2018; Alharazneh et al., 2011). Once internalized, the drug accumulates to concentrations several-fold higher than those in the surrounding endolymph within hours of systemic administration and persists for months owing to high-affinity phospholipid binding (Hailey et al., 2017). This intracellular retention initiates a broadly sequential, though temporally overlapping, cascade of injury: mitochondrial targeting and disruption of oxidative phosphorylation (Schacht et al., 2012; Bottger and Schacht, 2013); reactive oxygen species (ROS) generation through iron–aminoglycoside Fenton chemistry and NADPH oxidase 3 (NOX3)-dependent pathways (Schacht et al., 2012; Sha and Schacht, 1999; Mukherjea et al., 2011); a bidirectional amplification loop between oxidative damage and calcium dysregulation (Esterberg et al., 2013; Tabuchi et al., 2011); and activation of multiple cell death programs, including caspase-mediated apoptosis (Tabuchi et al., 2011; Cunningham et al., 2002), RIPK3/MLKL-dependent necroptosis (Ruhl et al., 2019), and iron-dependent ferroptosis (Zheng et al., 2020; Han et al., 2023). The approximate temporal progression from entry (∼0–6 h) through to execution of cell death programs (∼24–72 h) defines a narrowing but clinically meaningful window for intervention. Each stage is discussed in the subsections that follow.

Cellular mechanisms and proposed timeline of aminoglycoside ototoxicity in cochlear outer hair cells. The main schematic depicts a stylised outer hair cell with five temporally ordered mechanistic stages, each color-coded and labeled on the left margin with approximate time windows derived from in vitro cochlear explant and in vivo rodent models. (1) Entry (∼0–6 h): aminoglycosides (AGs) enter outer hair cells primarily through mechanoelectrical transduction (MET) channels at stereocilia tips and via a secondary endocytotic route; AG interaction with TRP channels produces altered ion conductance,* disrupting K+ and Ca2+ flux and impairing mitochondrial ribosomal protein synthesis. (2) Mitochondrial Targeting (∼2–12 h): AGs accumulate in mitochondria and bind the 12S rRNA of the mitochondrial ribosome, causing translational errors and impaired oxidative phosphorylation, with consequent early ROS generation. (3) Oxidative Stress (∼6–24 h): iron–AG complexes catalyze Fenton-type free radical production; NADPH oxidase activation amplifies the oxidative burden; lipid peroxidation produces membrane damage and nuclear DNA fragmentation. (4) Calcium Dysregulation (∼12–36 h): oxidative injury to the endoplasmic reticulum (ER) impairs Ca2+ sequestration, producing pathological cytosolic Ca2+ elevation; sustained Ca2+ overload triggers cytochrome c release,‡ committing the cell to intrinsic apoptosis. (5) Cell Death (∼24–72 h): hair cell demise proceeds via three converging pathways -- intrinsic apoptosis (casp9/casp3 activation), RIPK1/RIPK3/MLKL-dependent necroptosis, and iron-dependent ferroptosis driven by unrestricted lipid peroxidation. The inset (upper right) illustrates the MET channel entry mechanism at the stereocilium tip. Panel 1 (left) shows the resting/closed state: the tip link is intact, the MET channel is closed, and no cation flux or AG entry occurs. Panel 3 (right) shows the open/depolarized state: stereociliary bundle deflection tensions the tip link, mechanically gating the MET channel open; K+, Ca2+, and AGs enter through the open pore driven by the endocochlear potential, constituting the principal Stage 1 entry mechanism; the resulting Ca2+ influx also seeds the dysregulation cascade of Stage 4. Panel 2 of the original MET transduction diagram (mechanical displacement) was omitted as it describes the normal auditory stimulus rather than the drug entry mechanism. The dashed connector links the inset to the MET channel zone of the main schematic. The horizontal timeline bar (base of figure) depicts the five colour-matched stage segments across the ∼72-h progression. The solid green bar (∼0–24 h) demarcates the proposed intervention window within which preclinical otoprotective strategies demonstrate maximal efficacy; the pale green zone (∼24–72 h) reflects progressively declining otoprotective efficacy as cell death commitment becomes irreversible. All timeframes are approximate and should not be extrapolated to clinical contexts, in which pharmacokinetic variability, patient age, renal function, aminoglycoside compound, and concurrent ototoxic exposures substantially modify the ototoxicity timeline. All depicted interventions are preclinical and have not been validated in randomised clinical trials.* “Altered Ion Conductance” replaces the previously used term “K+ Channel Dysfunction” to reflect mechanistic evidence implicating TRP channels and non-selective cation conductances in addition to K+-selective pathways in early AG-mediated electrophysiological disruption. ‡ Cytochrome c release is positioned at the Stage 4→5 boundary to reflect the temporal sequence observed in vitro models, wherein mitochondrial outer membrane permeabilization follows sustained Ca2+ dysregulation. Abbreviations: AG, aminoglycoside; Ca2+, calcium ion; casp3, caspase-3; casp9, caspase-9; ER, endoplasmic reticulum; Fe–AG, iron–aminoglycoside complex; MET, mechanoelectrical transduction; MLKL, mixed lineage kinase domain-like protein; NADPH, nicotinamide adenine dinucleotide phosphate (reduced); OHC, outer hair cell; RIPK1, receptor-interacting protein kinase 1; RIPK3, receptor-interacting protein kinase 3; ROS, reactive oxygen species; TRP, transient receptor potential. Created in BioRender. Sherwin C. (2026) https://BioRender.com/265z8w8.

StageTimeframeKey processesMolecular markersIntervention windowEntry∼0–6 hMET channel permeation; endocytosis; TRP channel entryFluorescent AG accumulation; endocytic vesicle formationMET channel blockers (preclinical); endocytosis inhibitors (preclinical)Mitochondrial targeting∼2–12 hBinding to 12S rRNA; protein synthesis inhibition; initial ROS productionMitochondrial membrane depolarization/ultrastructural disruption; ↓ ATP production; ↑ superoxideAlternative ribosome binders (preclinical); mitochondrial protectants (preclinical)Oxidative amplification∼6–24 hIron–AG complex formation; lipid peroxidation; GSH depletion4-HNE adducts; protein carbonylation ↑; GSH:GSSG ratio ↓Antioxidants; iron chelators; GSH precursorsIon dysregulation∼12–36 h↑ intracellular Ca2+; altered membrane conductance and ion homeostasis; membrane permeabilizationCa2+ waves; membrane potential changes; phospholipid scramblingCa2+ chelators; ion channel modulatorsCell death execution∼24–72 hCytochrome c release; caspase activation; necroptosis activation; ferroptosis (iron-dependent lipid peroxidation)TUNEL-positive nuclei; activated caspase-3; RIPK3/MLKL translocation; lipid peroxidation markers (e.g., 4-HNE)Anti-apoptotic agents (preclinical); caspase inhibitors (preclinical); RIPK1 inhibitors (preclinical); ferroptosis inhibitors (preclinical)

Integrated pathway of aminoglycoside ototoxicity.

Timeframes represent approximate, overlapping windows derived primarily from experimental (in vitro and animal) models and should not be interpreted as precise clinical kinetics. Values vary by dose, compound, species, and developmental stage. Interventions marked “preclinical” have not been validated in human clinical trials. Molecular markers represent commonly reported mechanistic features rather than universal or sequentially obligatory events. Stages may overlap and vary according to dose, compound, species, and developmental context. Arrows indicate direction of change (↑ increase; ↓ decrease).

Abbreviations: AG, aminoglycoside; ATP, adenosine triphosphate; Ca2+, calcium ion; GSH, reduced glutathione; GSSG, oxidized glutathione (glutathione disulfide); GSH: GSSG, ratio of reduced to oxidized glutathione; 4-HNE, 4-hydroxynonenal; MET, mechanoelectrical transduction; MLKL, mixed lineage kinase domain-like protein; RIPK1, receptor-interacting protein kinase 1; RIPK3, receptor-interacting protein kinase 3; ROS, reactive oxygen species; TRP, transient receptor potential; TUNEL, terminal deoxynucleotidyl transferase dUTP nick end labeling.

Aminoglycosides enter cochlear and vestibular hair cells primarily through MET channels located at stereocilia tips. These cation-selective channels, composed of transmembrane channel-like proteins TMC1 and TMC2, normally transduce mechanical stimuli into electrical signals (Pan et al., 2018). The polybasic nature of aminoglycosides facilitates permeation through these channels, with smaller molecules, such as gentamicin, showing higher entry rates than larger compounds, such as amikacin. This differential permeability has clinical implications for drug selection.

Counterintuitively, sound itself promotes drug uptake: when stereocilia deflect, MET channels open more widely, increasing aminoglycoside entry. Functional MET channels are required for aminoglycoside ototoxicity, and increased channel open probability during acoustic stimulation enhances drug entry (Alharazneh et al., 2011). While clinical outcome data are limited, experimental and observational evidence suggest that the ambient noise in NICUs may be enough to worsen drug entry (Zimmerman and Lahav, 2013; Brown, 2009). Additional uptake routes are also relevant. Transient receptor potential (TRP) channels, notably TRPV1 and TRPV4, are upregulated during inflammation, which may explain why concurrent infection seems to heighten the risk of ototoxicity (Cross et al., 2015).

Once inside, aminoglycosides accumulate within hair cells to concentrations that persist for months owing to high-affinity binding to phosphatidylinositol 4,5-bisphosphate (Huth et al., 2011; Hailey et al., 2017). Advanced imaging reveals that aminoglycosides rapidly access the cytosol of hair cells, where a proportion is subsequently sequestered into endosomes and lysosomes; disruption of this lysosomal trafficking potentiates hair cell death (Hailey et al., 2017). This prolonged retention explains why ototoxicity can persist even after treatment stops, a factor clinicians need to consider when deciding how long to monitor patients.

Mitochondrial dysfunction

Aminoglycosides preferentially localize to mitochondria by binding to the 12S ribosomal RNA, which is encoded by the mitochondrially encoded gene MT-RNR1. This binding disrupts mitochondrial protein synthesis, which is essential for oxidative phosphorylation (Hobbie et al., 2008). The structural similarity between bacterial 16S rRNA and mitochondrial 12S rRNA, reflecting evolutionary origins, underlies both therapeutic efficacy against bacteria and ototoxicity to host cells.

Structural studies using cryo-electron microscopy have mapped the binding sites of aminoglycosides on mitochondrial ribosomal subunits (Bottger and Schacht, 2013). Upon binding, aminoglycosides cause misreading of the genetic code through conformational changes that affect codon-anticodon recognition (Hobbie et al., 2008), premature termination of protein synthesis, and downstream inhibition of electron transport chain complexes I, III, and IV, which contain mitochondrially encoded subunits (Schacht et al., 2012; Rybak and Ramkumar, 2007).

The MT-RNR1 m.1555A>G (Figures 4A,B) variant substantially amplifies this mitochondrial vulnerability, as discussed in the Genetic Susceptibility section below. Aminoglycosides also disrupt mitochondrial dynamics by inhibiting fusion, leading to fragmented mitochondrial networks with compromised function (Esterberg et al., 2013). Live-cell imaging shows mitochondrial fragmentation beginning several hours before detectable hair cell death, suggesting a window for early intervention (Hailey et al., 2017; Esterberg et al., 2013).

Genetic susceptibility to aminoglycoside ototoxicity: structural basis, nuclear modifiers, and population prevalence of the MT-RNR1 m.1555A>G variant. (A) Comparison of normal 12S rRNA (A1555, left) and mutated 12S rRNA (A1555G, right) at position 1,555 within helix 39. In the normal configuration, position 1,555 encodes adenine, producing a helix structure with limited hydrogen-bond donor capacity and low aminoglycoside binding affinity. The A→G substitution remodels helix 39 to structurally resemble the bacterial 16S rRNA decoding site (h44); guanine at position 1,555 introduces additional hydrogen-bond donors (N1 and N2 of guanine), enhancing aminoglycoside affinity (Hobbie et al., 2008; Guan et al., 2000; Guan, 2011). Hydrogen bonding changes are structurally inferred from comparative rRNA modelling and bacterial A-site homology; they have not been directly measured in human mitochondrial ribosomes. (B) Functional consequences of the structural change. Normal 12S rRNA (A1555) exhibits baseline aminoglycoside binding (∼1× affinity), preserving mitochondrial protein synthesis and function. Mutated 12S rRNA (A1555G) shows substantially increased aminoglycoside binding (∼5–10× affinity), resulting in impaired mitochondrial function even at therapeutic drug concentrations. Affinity values are schematic representations derived from differential ribosomal selectivity data rather than directly measured binding constants (Hobbie et al., 2008; Guan et al., 2000; Guan, 2011). (C) Proposed model of nuclear–mitochondrial interactions modulating penetrance of the m.1555A>G variant. Four nuclear modifier genes influence the phenotypic expression of MT-RNR1-associated ototoxicity: MTO1, GTPBP3, and TRMU participate in mitochondrial tRNA modification pathways; TFB1M methylates 12S rRNA at the mitochondrial ribosome (A10T variant of TFB1M reported to increase susceptibility approximately 2-fold) (Gao et al., 2017; Bykhovskaya et al., 2004). Aminoglycoside binding to the MT-RNR1-encoded 12S rRNA in the presence of these nuclear modifiers contributes to the variable and incomplete penetrance observed among carrier families. Carrier status does not predict clinical outcome; nuclear genetic background, environmental factors, and aminoglycoside exposure together influence penetrance. (D) General population carrier frequency of the m.1555A>G variant by population. European and US cohorts report consistent frequencies of approximately 0.19%–0.3%, while Chinese general population studies report a wider range of 0.14%–0.7%. The Taiwan estimate (0.1%) derives from a sensorineural hearing loss (SNHL) cohort rather than a general population screen and should be interpreted accordingly (Wu et al., 2007). The South African estimate (0.5%) is based on a small sample and should be interpreted with caution (Bardien et al., 2009). The European population estimate of approximately 1 in 520 is derived from the United Kingdom ALSPAC birth cohort (Bitner-Glindzicz et al., 2009). Among hearing-impaired populations, carrier frequency is substantially higher, particularly in East Asian cohorts, reflecting ascertainment bias. Other susceptibility variants are shown, including m.1494C>T (reported in East Asian populations), m.1095T>C (rarer, moderate susceptibility increase), and m.827A>G and m.961delT + C(n) (emerging evidence). Data sources: Population prevalence data from Bitner-Glindzicz et al. (2009), Vandebona et al. (2009), Gopel et al. (2014), Lu et al. (2010), Wu et al. (2007), Bardien et al. (2009), and Gaafar et al. (2024); prevalence table adapted from Usami & Nishio (GeneReviews) (Usami and Nishio, 1993). Hearing-impaired cohort frequencies from Guan et al. (2000) and Wu et al. (2007). Structural and binding concepts from Guan (2011), Guan et al. (2000), and Hobbie et al. (2008). Abbreviations: ALSPAC, Avon Longitudinal Study of Parents and Children; GTPBP3, GTP-binding protein 3; LBW, low birth weight; MT-RNR1, mitochondrially encoded 12S ribosomal RNA; MTO1, mitochondrial tRNA translation optimization 1; SNHL, sensorineural hearing loss; TFB1M, transcription factor B1, mitochondrial; TRMU, tRNA 5-methylaminomethyl-2-thiouridylate methyltransferase.

Oxidative stress cascades

Aminoglycosides generate reactive oxygen species through multiple converging mechanisms. Iron-aminoglycoside complexes catalyze Fenton reactions, converting hydrogen peroxide to highly reactive hydroxyl radicals that damage lipids, proteins, and DNA (Sha and Schacht, 1999). Simultaneously, drug binding activates NOX3, which is highly expressed in cochlear tissue, thereby generating superoxide anions (