HeartVar: An LLM-Assisted Tool for Clinical Classification of Variants in Cardiovascular Disease Cohorts

Abstract

Manual clinical DNA variant classification is the bottleneck of every clinical and research rare disease workflow. The process typically requires a curator to assemble evidence from numerous databases, weigh 28 criteria, reconcile competing evidence, and produce a defensible case for the final classification. Additionally, the framework used to assess variants is not static and successive addenda have revised individual criteria. The most complete and current evidence aggregators available are commercial platforms, which can limit researcher access. We present HeartVar, an open-source web tool that automates the evidence-gathering and interpretation steps of variant classification associated with cardiovascular disease. Given a gene, a variant, and clinical context, HeartVar queries 20 public databases in parallel and assigns ACMG/AMP criteria through a hybrid rule-based/large language model (LLM) approach. Criteria that can be resolved from structured data are computed programmatically, and only those requiring interpretation of unstructured evidence are passed to the LLM. HeartVar returns a classification, point score, per-criterion breakdown, clinical-narrative summary, and database annotations. Benchmarking of 106 expert-curated ClinGen variants showed HeartVar outperformed other curation tools, assigning the correct ACMG tier in 72% of cases. HeartVar demonstrates that an LLM constrained by a domain-specific prompt and grounded in structured evidence can produce variant interpretations of first-pass quality for a cardiovascular disease cohort; however, it is not intended to replace manual assessment by a qualified variant curator. The tool is freely available to use and hosted at www.heartvar.victorchang.edu.au.