Hearing Loss, Autosomal Recessive 106 (DFNB106) — Comprehensive Research Report

1. Disease Information

Overview. Autosomal recessive deafness-106 (DFNB106) is a rare, genetically defined form of nonsyndromic sensorineural hearing loss caused by biallelic pathogenic variants in EPS8L2 (EPS8-Like protein 2), a gene encoding an actin-binding stereocilia protein of cochlear hair cells. Unlike most autosomal recessive nonsyndromic hearing loss (ARNSHL), which is typically congenital, stable, and profound, DFNB106 is distinctive for being prelingual-to-childhood onset and progressive, evolving from mild/moderate loss at low frequencies toward severe-to-profound loss, particularly at high frequencies, over years (Dahmani et al. 2015, PMID:26282398; Owrang et al. 2026, PMID:41514136).

Key identifiers: - OMIM (phenotype): #617637 — DEAFNESS, AUTOSOMAL RECESSIVE 106; DFNB106 - OMIM (gene): 614988 — EPS8-LIKE PROTEIN 2; EPS8L2 - Gene locus: 11p15.13 (NCBI Gene ID 64787; HGNC:21296; UniProt Q9H6S3) - MONDO: MONDO:0033198 (confirmed via ClinVar cross-reference) - MedGen: C4539954 - ClinVar example record: RCV000499522 (NM_022772.4(EPS8L2):c.1014del, p.Ser339fs) — Pathogenic, 1-star review - Synonyms:* DFNB106; Deafness, autosomal recessive 106; EPS8L2-related hearing loss

Source of information. Nearly all information available is derived from a small number of published family case series/cohort reports (whole-exome sequencing of individual pedigrees), not large-scale EHR-aggregated data — reflecting the extreme rarity of this specific locus. As of the most recent (2026) synthesis, only 8 families / ~14 affected individuals worldwide have been reported in the literature (Owrang et al. 2026, PMID:41514136).

2. Etiology

Disease causal factor: Purely genetic — biallelic (homozygous or compound heterozygous) loss-of-function or spliceogenic variants in EPS8L2. There is no known environmental, infectious, or acquired contributor; this is a monogenic Mendelian disorder.

Genetic risk factors

• Causal gene: EPS8L2 (11p15.13). All reported pathogenic alleles predict truncation, nonsense-mediated decay, or aberrant splicing leading to frameshift — i.e., loss-of-function as the shared mechanism.
• Consanguinity as an enabling risk factor: Several reported families are consanguineous (Algerian second-cousin union; Iranian and Pakistani consanguineous pedigrees), consistent with autosomal recessive transmission and increasing homozygosity risk for a rare allele (Dahmani et al. 2015; Wang et al. 2017, PMID:28281779).
• Founder effect: Two unrelated Iranian families in the 2026 report share a 4.17 Mb run of homozygosity (overlapping interval Chr11:g.197337–4367675, GRCh38) around a recurrent EPS8L2 missense/spliceogenic variant, suggestive of a regional founder allele rather than independent mutation events (Owrang et al. 2026, PMID:41514136).
• Gene constraint: gnomAD v4.0 reports EPS8L2 pLI = 0 and LOEUF = 1.17, indicating the gene is not strongly constrained against heterozygous loss-of-function in the general population — consistent with a purely recessive disease mechanism where carriers are unaffected.

Protective factors

None reported. No modifier alleles, protective variants, or environmental protective exposures have been described for this ultra-rare condition.

Gene-environment interactions

None documented; no evidence of environmental modulation of onset or severity has been reported.

3. Phenotypes

All phenotypic data reported to date are auditory — no extra-auditory features are described in any published case, consistent with the "nonsyndromic" designation.

Characteristics: - Age of onset: Ranges from presymptomatic at birth (passing newborn hearing screening) to recognition around 4 years (most common), preschool age, or prelingual in some kindreds. The 2026 study documents the first pre-symptomatic case: a German child who passed newborn screening but developed measurable loss by ~23 months, confirming that normal hearing at birth is compatible with the genotype (Owrang et al. 2026, PMID:41514136). - Severity: Moderate-to-severe/profound, variable across families and even within the same variant. - Progression: Documented in the majority (5/6) of individuals with serial audiograms — an unusual feature for ARNSHL, most of which (e.g., GJB2-related) is congenital and stable. - Speech outcomes: Speech discrimination scores of 70–96% reported in the two most recent Iranian probands, correlating with moderate rather than profound loss at last follow-up.

Quality of life: Not formally measured with standardized instruments (EQ-5D/SF-36) in any report; clinical narrative notes affected children use hearing aids, and delayed language/speech development was observed in at least one case ("unclear language" and non-response to sound from age 2, Family 1 proband).

4. Genetic/Molecular Information

Causal gene and variant catalog

All known pathogenic EPS8L2 variants (NM_022772.4 transcript; 21 exons, 715-aa protein):

Variant classification: All reported variants are pathogenic or likely pathogenic per ACMG/AMP criteria (e.g., the spliceogenic missense classified PM2_P, PM3_M, PVS1_S). All converge mechanistically on loss of function — via frameshift/premature termination, nonsense-mediated decay, or exon skipping.

Population frequency: Pathogenic EPS8L2 alleles are essentially absent from population databases — the recurrent c.767C>G founder variant was found in only 1 carrier among ~40,000 exomes (UCL Queen Square database) and was absent from gnomAD, TopMed, and All of Us (PMID:41514136), underscoring extreme rarity.

Somatic vs. germline: All variants are germline; no somatic mosaicism reported.

Functional consequence: Uniformly loss-of-function — truncated/absent protein (frameshift/nonsense with NMD) or exon-skipping-induced frameshift (spliceogenic missense). No gain-of-function or dominant-negative alleles have been described.

Modifier genes

None identified to date.

Epigenetic information

Not investigated/reported for this gene-disease relationship.

Chromosomal abnormalities

None reported; disease is caused by small intragenic variants, not structural/copy-number changes.

5. Environmental Information

No environmental factors, lifestyle factors, or infectious triggers have been implicated. DFNB106 is a purely monogenic disorder with no reported gene-environment modulation.

6. Mechanism / Pathophysiology

Protein structure and localization

EPS8L2 (EPS8 Signaling Adaptor L2) is a 715-amino-acid, F-actin-binding member of the EPS8 protein family (EPS8, EPS8L1, EPS8L2, EPS8L3), containing a phosphotyrosine-binding (PTB) domain, an SH3 domain, an effector domain, and (per the 2026 paper) a SAM/pointed (SAM/PNT) domain near the C-terminus. In cochlear and vestibular hair cells, EPS8L2 localizes specifically to the tips of the shorter and intermediate rows of stereocilia, distinguishing it from EPS8, which localizes to and elongates the tallest stereocilia row (Dahmani et al. 2015, PMID:26282398; Furness et al. 2013 PNAS, PMID:23918390).

Causal chain (cellular process → tissue → organism)

1. Molecular: Loss-of-function EPS8L2 variant → absent or truncated actin-binding adaptor protein at stereocilia tips.
2. Cellular: Failure to maintain (not initially build) the actin core of the short/intermediate stereocilia rows in mature hair cells → gradual disorganization and shortening of these rows, with variable width abnormalities.
3. Tissue: Progressive deterioration of the hair bundle "staircase" architecture required for mechanotransduction in cochlear (and vestibular) hair cells of the organ of Corti.
4. Organism: Progressive, typically postlingual-onset sensorineural hearing loss, sloping toward high frequencies (or U-shaped in the spliceogenic-variant cases), because mechanotransduction becomes progressively impaired as the affected stereocilia rows degrade with age/use rather than failing to form in the first place.

This gives DFNB106 a mechanistically distinct signature versus most ARNSHL: EPS8 loss causes early, static, profound deafness (failure to elongate stereocilia in the first place — DFNB102), whereas EPS8L2 loss causes late-onset, progressive deafness (failure to maintain mature stereocilia) — a "maintenance vs. morphogenesis" dichotomy within one paralog family (Furness et al. 2013; [EPS8/DFNB102 review, PMID not extracted, PMC9837036]).

Molecular pathway

EPS8L2 acts through actin cytoskeleton dynamics regulation — bundling/capping activity at stereocilia tips, and (based on EPS8-family biology) potential modulation of Rac/SOS1-linked signaling relevant to membrane/cytoskeletal remodeling (GeneCards annotation).

Suggested GO terms: - GO:0051017 — actin filament bundle assembly - GO:0032420 — stereocilium - GO:0003785 — actin monomer binding - GO:0060088 — auditory receptor cell stereocilium organization

Suggested CL terms: - CL:0000601 — inner hair cell of Corti's organ (or cochlear outer/inner hair cell, CL:0000589)

Suggested UBERON terms: - UBERON:0001844 — cochlea / organ of Corti - UBERON:0009865 — stereocilium bundle (if available) or UBERON:0002106 (spiral organ)

Molecular profiling / advanced technologies

• Zebrafish whole-mount in situ hybridization (WISH): eps8l2 is expressed in the otic vesicle — specifically the presumptive utricular and saccular maculae — and in the pronephric duct; later stages show broader otic-vesicle expression plus scattered spinal cord expression (Owrang et al. 2026, PMID:41514136).
• Minigene/splicing assay: For the c.767C>G missense variant, RT-PCR of a minigene construct demonstrated complete exon 9 skipping, converting a presumed benign missense change into a functional null allele — directly confirming spliceogenicity by wet-lab assay rather than in silico prediction alone (only 2/6 splice-prediction algorithms had flagged it).
• No transcriptomic, proteomic, or single-cell datasets specific to human EPS8L2-deficient tissue have been published (human inner-ear tissue is not accessible for biopsy).

7. Anatomical Structures Affected

• Organ level: Inner ear only (cochlea and, to a lesser functional extent demonstrated in mouse, the vestibular system) — no other organ system involvement reported (nonsyndromic).
• Tissue/cell level: Cochlear and vestibular hair cells (CL:0000601/CL:0000602), specifically their stereocilia bundles.
• Subcellular level: Stereocilia actin core/tip complex — the specific site of EPS8L2 localization (GO:0005884 actin filament; GO:0032420 stereocilium).
• Localization: Bilateral, symmetric involvement in all reported human cases (no unilateral or asymmetric presentations documented).

8. Temporal Development

• Onset: Ranges from prelingual/infantile to early childhood (~4–7 years most typical); one documented presymptomatic case with normal newborn hearing screening followed by measurable loss detected by ~23 months.
• Onset pattern: Insidious/gradual rather than acute.
• Progression: Progressive in the majority of longitudinally followed cases (5/6) — an atypical feature for ARNSHL. Progression can continue into adulthood (one individual "in her fourth decade" with prior progression now described as stable over the last decade of follow-up), suggesting eventual plateauing in at least some individuals.
• Disease course pattern: Chronic, generally progressive-then-potentially-stabilizing; not episodic or relapsing-remitting.
• Critical period: The 2026 paper explicitly frames early infancy/toddlerhood as a "therapeutic window" — before stereocilia degeneration is complete — for any future molecular intervention, based on parallel work in EPS8 (DFNB102) mouse gene-therapy models (see Treatment section).

9. Inheritance and Population

• Inheritance pattern: Autosomal recessive (both homozygous and compound heterozygous genotypes documented).
• Penetrance: Appears complete for biallelic loss-of-function genotypes, though age-dependent (phenotype may not be measurable at birth).
• Expressivity: Variable — age of onset, severity, rate of progression, and audiogram shape (sloping vs. U-shaped) differ even among carriers of the same variant (e.g., within the shared founder-variant Iranian families).
• Genetic anticipation: Not applicable/not reported (not a repeat-expansion disorder).
• Germline mosaicism: Not reported.
• Founder effects: A shared ~4.17 Mb homozygous haplotype around c.767C>G in two ostensibly unrelated Iranian families strongly suggests a regional founder allele.
• Consanguinity: A major contributing factor in most reported pedigrees (Algerian, Iranian, Pakistani).
• Carrier frequency: Cannot be reliably estimated — pathogenic alleles are essentially unseen in gnomAD/TopMed/All of Us; the disease is considered ultra-rare.
• Epidemiology: No formal prevalence/incidence estimate exists. Only 8 families (~14 affected individuals) have been published worldwide as of the January 2026 update (Owrang et al. 2026), spanning Algeria, Pakistan (≥2 families), Iran (≥3 families), Germany, and China (1 case). For context, ARNSHL overall accounts for ~80% of genetic nonsyndromic hearing loss, with GJB2/DFNB1 explaining roughly 16.9% globally (up to 27.1% in European cohorts) and >85 genes implicated overall in ARNSHL — DFNB106 is a very minor contributor to this landscape.
• Geographic/ethnic distribution: Cases reported across North Africa (Algeria), South Asia (Pakistan), the Middle East (Iran), Europe (Germany, with mixed Greek/German parentage), and East Asia (China) — no single predominant population, though consanguineous-marriage-practicing populations are overrepresented, as expected for rare recessive disease ascertainment.
• Sex ratio: No sex predilection reported (X-linked/mitochondrial mechanisms excluded; autosomal recessive).

10. Diagnostics

• Clinical/audiologic tests: Standard pure-tone audiometry (air and bone conduction across 125 Hz–8 kHz), speech reception threshold and speech discrimination testing. Serial audiometry over years is essential to document the characteristic progression.
• Newborn hearing screening: May be passed (normal) despite carrying a biallelic pathogenic EPS8L2 genotype — a critical diagnostic pitfall, since a normal newborn screen does not rule out DFNB106 given its progressive, sometimes-delayed onset.
• Genetic testing:
• Whole-exome sequencing (WES) has been the discovery and diagnostic method for essentially all reported cases (candidate-gene approach after excluding GJB2), reflecting how rare/unrecognized this gene is on standard hearing-loss panels.
• Targeted deafness gene panels may include EPS8L2 (e.g., listed on the Genomics England PanelApp "Monogenic hearing loss" panel) but coverage varies by lab.
• Autozygosity/homozygosity mapping has been used in consanguineous families to define candidate intervals and detect founder alleles.
• Minigene/splicing functional assays are recommended when a missense variant is discovered, given the demonstrated risk of missense variants (e.g., c.767C>G) acting as cryptic splice-disrupting alleles rather than simple amino-acid substitutions — routine in silico tools alone missed this in most predictors.
• Differential diagnosis: Other genetic causes of progressive, nonsyndromic, prelingual-to-childhood-onset sensorineural hearing loss (e.g., STRC-related, TMPRSS3-related, OTOF-related, other DFNB loci), and syndromic causes should be excluded via broader panel/exome testing and clinical evaluation (absence of other organ involvement supports "nonsyndromic" classification).
• Screening: No population-level screening program exists for EPS8L2 given its rarity; identification is via diagnostic exome/genome sequencing after clinical suspicion (progressive ARNSHL, especially with a family history or normal-but-later-failing newborn screen).

11. Outcome/Prognosis

• Mortality: None — DFNB106 causes isolated hearing loss with no reported effect on life expectancy or systemic morbidity.
• Morbidity: Communication/speech-language developmental impact from progressive hearing loss, particularly if diagnosis and intervention (hearing aids) are delayed; documented delayed speech/language recognition in at least one proband.
• Disease course: Progressive threshold elevation over years to decades in most individuals; in a minority, progression appears to plateau (documented stability over a decade in an adult in her 4th decade of life; near-stability over 10 years in one Iranian proband).
• Prognostic factors: Earlier recognition and hearing-aid fitting appear associated with better functional outcomes (standard audiology principle); no genotype-severity correlation has yet been established across the small number of reported variants, though the two individuals with the spliceogenic c.767C>G variant shared a distinctive U-shaped audiogram, hinting at a possible variant-specific pattern.
• General ARNSHL cochlear implant data (not DFNB106-specific, given no published case yet requiring implantation): Progressive vs. congenital profound hearing loss show no significant difference in cochlear implant outcomes; genetic etiology overall does not reduce implant benefit, with performance driven mainly by age at implantation and duration of hearing loss prior to intervention.

12. Treatment

No disease-specific or FDA-approved therapy exists for DFNB106. Management is symptomatic/supportive, following standard practice for progressive sensorineural hearing loss:

• Hearing aids (NCIT:C15302 is for physical therapy; the relevant NCIT concept is closer to "Hearing Aid" under assistive devices) — used by affected children as soon as loss is documented; one German proband fitted with aids from young childhood, his presymptomatic sibling scheduled for first aids at 2 years 8 months once threshold elevation was confirmed.
• Cochlear implantation — not yet reported for a confirmed DFNB106 case in the literature, but by extrapolation from general ARNSHL cochlear-implant literature, would be expected to be effective if hearing loss progresses to severe/profound levels, with outcomes driven by implantation timing rather than genotype per se.
• Genetic counseling (NCIT:C15240) — recommended given the autosomal recessive inheritance, ~25% recurrence risk per pregnancy for carrier couples, and the discovery of a Fars-region Iranian founder allele relevant to regional carrier screening.
• Speech and audiologic rehabilitation (NCIT:C159273 speech therapy) — standard adjunct for children with progressive hearing loss.

Experimental/preclinical therapeutic direction ("therapeutic window")

The 2026 Owrang et al. paper explicitly discusses gene-therapy precedent from the sister gene EPS8 (DFNB102): a preclinical AAV-mediated (Anc80L65 serotype) EPS8 gene-replacement study in Eps8 knockout mice rescued hair-bundle structure but not functional hearing, and rescue efficacy dropped sharply if AAV delivery was delayed past postnatal day 1–2 ("when delayed to postnatal day 3, almost no recovery was observed"). The authors argue this underscores the urgency of early molecular diagnosis for EPS8L2, since its later, more gradual onset (compared to EPS8/DFNB102's early profound congenital deafness) may in principle offer a longer intervention window for a future EPS8L2-directed gene therapy — though no such therapy has yet been developed or tested for EPS8L2 itself. This is framed as a rationale for early genetic testing/referral, not as an available treatment.

No clinical trials (NCT) specific to EPS8L2/DFNB106 were identified in the search.

13. Prevention

• Primary prevention: Not applicable in the biological sense (no modifiable risk factor); the only "primary prevention" lever is reproductive genetic counseling for known-carrier couples (e.g., preimplantation genetic diagnosis or prenatal testing where a familial variant is known), particularly relevant in consanguineous unions or populations carrying an identified founder allele (e.g., the Fars, Iran region).
• Secondary prevention/early detection: Universal newborn hearing screening remains valuable but is explicitly shown to be insufficient alone for DFNB106, since affected infants can pass screening and only develop measurable loss in the second year of life — supporting a case for genetic newborn screening or expanded gene panels in at-risk families (e.g., known familial variant, consanguinity, family history of progressive childhood-onset ARNSHL) to enable close audiologic monitoring even after a normal screen.
• Tertiary prevention: Early hearing-aid fitting and speech-language intervention to minimize the developmental impact of progressive threshold elevation.
• Genetic/carrier screening: Expanded carrier screening panels in consanguineous or founder populations could, in principle, include EPS8L2, though it is not yet part of standard commercial hearing-loss carrier panels given its rarity.

14. Other Species / Natural Disease

This is one of the best-documented sections for DFNB106, given a notable naturally occurring canine model:

• Rhodesian Ridgeback dog — early-onset adult deafness: A genome-wide association study (23 affected vs. 162 control dogs) followed by Sanger sequencing identified a 12-bp in-frame deletion in EPS8L2, segregating in an autosomal recessive pattern; "all affected dogs were homozygous for the deletion" (Kawakami et al. 2022, PMID:35385474, PLoS ONE). This represents a spontaneous, naturally occurring veterinary phenocopy of human DFNB106 and is a strong translational/comparative model, since it arose without laboratory engineering and mirrors the human recessive, hair-cell-maintenance mechanism.
• Suggested NCBI Taxon: NCBITaxon:9615 (Canis lupus familiaris); breed identifiers via VBO would apply to "Rhodesian Ridgeback."
• Mouse (Mus musculus, engineered knockout): Eps8l2-null mice (targeted knockout, not naturally occurring) — see Model Organisms below.
• No other species' natural disease has been reported for EPS8L2 to date (unlike some hearing-loss genes with described feline or bovine natural mutants).
• Comparative biology: The EPS8 gene family (EPS8, EPS8L1, EPS8L2, EPS8L3) and its stereocilia-tip actin-regulatory role are evolutionarily conserved across mammals, underpinning why the mouse, dog, and human phenotypes converge on the same "progressive stereocilia maintenance failure" mechanism rather than divergent pathology.
• Zoonotic potential: Not applicable — this is a non-infectious monogenic disorder.

15. Model Organisms

Model limitations: The mouse knockout shows a somewhat different stereocilia-row pattern (tall row affected) than what would be predicted from EPS8L2's normal tip localization at short/intermediate rows in wild-type animals — this apparent paradox is discussed in the primary literature but not fully resolved, and represents an open question about full concordance between mouse structural findings and the exact human audiometric/histopathologic correlate (human temporal bone histopathology is unavailable). No iPSC-derived otic organoid model of EPS8L2 deficiency has yet been published. Preclinical AAV gene-therapy rescue data exist only for the paralogous gene EPS8 (DFNB102 mouse model), not yet for EPS8L2 itself, and even that related rescue restored structure without restoring functional hearing — an important caveat when extrapolating "therapeutic window" arguments to EPS8L2.

Summary of Key Citations

1. Dahmani M, et al. "EPS8L2 is a new causal gene for childhood onset autosomal recessive progressive hearing loss." Orphanet J Rare Dis. 2015. PMID:26282398 — founding report (Algerian family).
2. Furness DN, et al. "Progressive hearing loss and gradual deterioration of sensory hair bundles in the ears of mice lacking the actin-binding protein Eps8L2." PNAS. 2013;110(34):13898–13903. PMID:23918390 — mouse knockout, mechanistic basis.
3. Wang R, et al. "Molecular Analysis of Twelve Pakistani Families with Nonsyndromic or Syndromic Hearing Loss." Genet Test Mol Biomarkers. 2017. PMID:28281779 — second EPS8L2 family (Pakistan, c.737delC).
4. Kawakami T, et al. "Early onset adult deafness in the Rhodesian Ridgeback dog is associated with an in-frame deletion in the EPS8L2 gene." PLoS One. 2022. PMID:35385474 — canine natural model.
5. Owrang [lead author], et al. "Expansion of Molecular and Clinical Aspects of EPS8L2 (DFNB106)-Associated Hearing Loss Emphasizes a Potential Therapeutic Window." Mol Neurobiol. 2026 (online Jan 2026). PMID:41514136 — most comprehensive, current synthesis (4 new families, splicing functional assay, therapeutic-window discussion, zebrafish expression).
6. Gan H, et al. "Identification and phased de novo mutation of the EPS8L2 gene in a patient with progressive hearing loss." Medicine (Baltimore). 2026. PMID:41578500 — first compound-het case including a de novo allele.
7. OMIM #617637 (DFNB106) and *614988 (EPS8L2) — https://omim.org/entry/617637 ; https://omim.org/entry/614988
8. ClinVar RCV000499522 — MONDO:0033198 cross-reference.

Data gaps to flag explicitly for curation: No published human temporal-bone histopathology; no confirmed cochlear-implant outcome case specific to DFNB106; no iPSC/organoid model; no EPS8L2-specific gene therapy (only paralog EPS8 preclinical data exists); prevalence/incidence figures are not formally estimated (only cumulative case counts across published families).

Reference Validation

Checked with linkml-reference-validator 0.2.1.

All extracted references resolved successfully.