Safety Assessment of Genome Editing in Human Gene Therapy Products Using Next-Generation Sequencing; Draft Guidance for Industry

The commenter, representing Simon Fraser University's Computational Biology Lab, proposes four specific quantitative reporting elements to be added to the draft guidance on genome editing safety assessments. They argue that the current guidance lacks specific statistical forms for "completeness" and "sensitivity," and they provide a methodology to account for off-target candidate universe completeness, structural-variant detection floors, sequencing-depth adequacy, and bioinformatics reproducibility.

The Alliance for Regenerative Medicine (ARM) supports the FDA's draft guidance on the safety assessment of genome editing in human gene therapy products. They recommend refining the guidance to ensure it remains risk-based, modality-specific, and proportionate to the biological risk of different editing technologies.

Dr. Kay C. Wiese, representing Simon Fraser University's Computational Biology Lab, proposes four specific quantitative reporting elements to be added to the draft guidance. The author argues for more rigorous disclosure of off-target nomination completeness, structural-variant detection floors, sequencing-depth adequacy, and bioinformatics reproducibility to address current qualitative gaps in the draft.

Diane Castillo, an independent scientist, argues that the proposed guidance is too permissive and fails to account for the stochastic and unpredictable nature of genome editing. She advocates for requiring single-cell long-read whole genome sequencing (scLRWGS) to detect large structural variants and off-target events that current short-read and targeted methods miss.

The American Society of Gene & Cell Therapy (ASGCT) supports the draft guidance on NGS-based safety assessment of human genome editing products. They request additional clarity on assay performance principles, emerging editing modalities, and specific parameters for off-target and chromosomal integrity analysis to ensure consistent implementation across the industry.

SeQure, a contract research organization, supports the draft guidance and suggests specific improvements regarding minimum sensitivity for off-target confirmation, the inclusion of optical genome mapping for structural variations, and the recommendation of insertion site analysis for certain payloads. They also advocate for high-throughput biochemical assays to account for human genetic variation and editor-specific nomination strategies.

A group of scientists and researchers from various biotechnology and technology development companies (including Pacific Biosciences, Praxis Molecular Solutions, Catalent, Sangamo Therapeutics, and Form Bio) expressed broad support for the FDA's draft guidance on NGS-based genome editing safety assessments. They recommend specific technical enhancements, such as establishing a limit of detection (LoD) for assays, requiring explicit bioinformatics version control, and encouraging validation against reference datasets.

PhRMA, representing leading biopharmaceutical companies, supports the FDA's draft guidance on genome editing safety assessments. They advocate for a risk-based, "fit-for-purpose" regulatory framework that provides more clarity on analytical thresholds, off-target site filtering, and specific requirements for different editing modalities to ensure regulatory predictability and innovation.

The commenter provides specific technical feedback on the draft guidance, requesting clarification on FDA expectations for assurance levels when no off-target edits are detected. They also suggest that the guidance should include methods to normalize against chromosomal abnormalities that occur independently of genome editing methods.

Researchers from Boston Children's Hospital and Massachusetts General Hospital support the draft guidance but argue that the sections on experimental design and statistical analysis are underdeveloped. They recommend specific improvements regarding the distinction between observed and unobserved editing rates, the use of prospective power analyses, and the reporting of statistical measures of uncertainty.