An Israeli HMO biobank sequenced 1,038 patients to hunt for deafness genes

Researchers performed whole-exome sequencing on 1,038 hearing-impaired patients enrolled in an Israeli health-maintenance organisation's biobank, and used it to solve 15% of cases and to flag new candidate genes for hearing loss [s1]. The study is a working example of a model that Israel's HMO structure makes possible: genomic data linked at scale to routine clinical records [s1].

The problem the study addresses

Hundreds of genes are implicated in hearing loss, yet up to half of inherited cases remain genetically unsolved [s1]. That gap has become more consequential as gene therapies for specific forms of deafness move toward trials — because eligibility for those trials requires a genetic diagnosis in the first place [s1]. A patient whose hearing loss cannot be traced to a specific variant is a patient who cannot be matched to a targeted therapy [s1].

The study set out to close some of that gap using existing infrastructure rather than a new prospective cohort. Whole-exome sequencing was performed on DNA from 1,038 hearing-impaired patients in the Maccabi Research and Innovation Center Tipa Biobank, with clinical data extracted from electronic medical records [s1].

What the records could and could not provide

The design exposes both the promise and the limit of biobank-plus-records research. Audiograms — the hearing tests that quantify the loss — were available for all cases [s1]. But data on age of onset, family history and mode of inheritance were mostly unavailable [s1].

That absence is not a footnote. Inheritance pattern is one of the most useful pieces of information for interpreting a genetic variant: whether a condition is dominant or recessive changes how a candidate variant is weighed. The researchers had to build an analysis strategy that could work despite incomplete and heterogeneous records — high-throughput annotation, filtering and prioritisation of variants across more than 1,000 patients, designed to accommodate the gaps [s1]. The engineering effort here is largely a response to the messiness of real-world clinical data.

The results

Using that approach, 15% of cases were solved or potentially solved through known or novel variants in established deafness genes [s1]. A further 3% of cases carried homozygous variants in novel candidate genes — genes not previously established as causing hearing loss [s1]. Homozygous means both copies carried the variant, which in a recessive condition is the expected pattern for an affected person and strengthens the case that the gene is responsible [s1].

The authors did not stop at statistical association. They performed functional characterisation of promising candidate genes to validate a role in the ear [s1]. That step is what separates a candidate gene generated by a database filter from one with biological evidence behind it.

Why the delivery model matters

The health-technology story is the source of the data. Israel's health-maintenance organisations hold longitudinal electronic records for large, stable member populations, and pairing those records with a biobank turns routine care into a research substrate [s1]. The same structure has been used elsewhere in Israeli health services for other clinical questions. What this study adds is a demonstration that the approach can expand the catalogue of disease genes, not merely audit existing practice [s1].

The limits

This is a single-centre study built on one biobank, and its diagnostic yield reflects that specific cohort and analysis pipeline [s1]. A 15% solve rate is a meaningful gain against a backdrop where half of inherited cases go unexplained, but it also means most cases in the sample remained unsolved [s1]. The novel candidate genes are candidates: functional work supports them, but confirmation as established disease genes requires replication in other populations [s1]. And the missing inheritance and family-history data — a limitation the authors are explicit about — constrains how confidently any single variant can be interpreted from records alone [s1].

The broader claim the study supports is narrower than a cure: that linking genomic and clinical data at scale can widen the set of patients who receive a genetic diagnosis, which is the gate to the therapies now in development [s1].