Cell and gene therapies (CGTs) are expanding possibilities for patients with genetic diseases that have limited or no effective treatment options. Genome editing is now enabling direct and potentially durable modification of disease-relevant DNA sequences in somatic cells. However, the same features that make genome editing powerful also create distinct safety questions that differ from those associated with conventional therapies.
Regulatory agencies are increasingly focused on a risk-based characterization of intended editing outcomes, unintended editing, and broader genomic integrity consequences. In the United States, the FDA’s January 2024 guidance document entitled Human Gene Therapy Products Incorporating Human Genome Editing established a product-specific framework for evaluating these risks.1 More recently, FDA issued draft guidance focused on next-generation sequencing (NGS) approaches for safety assessment,2 reflecting the growing importance of sequencing strategy, sample selection, bioinformatics, and transparent data interpretation for investigational new drug (IND) and biologics license application (BLA) packages.
As editing platforms, cell types, delivery technologies, and therapeutic indications continue to diversify, developers must design genomic safety programs that are scientifically justified and fit for purpose for the product’s mechanism and clinical context. The objective is to build a coherent body of evidence that addresses the relevant risks, defines the limitations of each method, and supports a reasoned safety assessment.
From Historical Lessons to Modern Genome Editing
The need for rigorous genomic safety assessment in CGTs emerged from experience with integrating viral vectors, which can provide durable therapeutic gene expression but may also cause insertional mutagenesis by disrupting genes or altering regulation near an integration site. Early gammaretroviral gene therapy trials – including studies for X-linked severe combined immunodeficiency – showed that vector integration can, in rare cases, promote clonal expansion and hematologic malignancy.3 Modern, self-inactivating lentiviral vectors are designed to reduce this risk through improved vector architecture and different integration preferences. However, their risk is not zero. Reports of clonal expansion and hematologic malignancies in some lentiviral vector programs underscore that safety depends on the specific vector design and disease context. These cases do not diminish the substantial clinical benefit of vector therapies but demonstrate the importance of safety assessment.
Modern genome editing systems introduce a new set of potential risks. CRISPR nucleases generate double-strand DNA breaks at the intended site but, in some circumstances, can also cause breaks at genomic sites with partial guide RNA complementarity. Base editors can introduce nucleotide substitutions without creating a double-strand break but may create guide-dependent off-target DNA edits, bystander edits within the editing window, or off-target RNA activity depending on the editor design. Similarly, prime editors, recombinases, transposases, integrases, epigenetic editors, and other emerging systems each have distinct mechanisms of action and potential risk modes. As a result, genomic safety should not be treated as synonymous with off-target analysis alone.
A comprehensive assessment needs to consider multiple factors, including the spectrum and frequency of intended on-target editing outcomes, unintended editing at candidate off-target sites, large structural rearrangements, vector-related integration, the influence of patient-specific or population-level genetic variation on editing performance and off-target susceptibility. This broader framing is particularly important as the field moves beyond a limited set of CRISPR-Cas9 applications toward increasingly diverse ex vivo, in vivo, multiplexed, and individualized editing strategies.
A Risk-Based Regulatory Approach
The regulatory landscape is evolving alongside these technologies and is currently far from a universal set of fixed requirements. FDA’s January 2024 guidance describes considerations for preclinical assessment, product characterization, clinical development, and long-term follow-up. The agency’s 2026 draft guidance on NGS-based safety assessment adds further details regarding sequencing approaches that may be used to assess on-target outcomes, off-target editing, and genomic integrity in support of clinical development. Together, these documents reinforce several important principles:
- Genomic safety assessment should begin early in therapeutic development.
- The evidence package should be specific to the editor mechanism of action and clinical context.
- Orthogonal methods can be valuable when their combined strengths address the blind spots of a single technique.
- Assay limitations, sensitivity, specificity, sampling rationale, bioinformatics parameters, and decision criteria should be clearly explained.
The central task is to identify the plausible risks, determine which questions are critical to decision making, and select the methods that can address those questions with sufficient sensitivity and relevance. Casgevy provides an instructive example of a product-specific approach rather than a universal assay template. Its development program incorporated computational analyses, experimental off-target nomination approaches – including GUIDE-seq – and targeted sequencing-based evaluation of nominated sites in donor and patient material.4 The Casgevy example also raised awareness that rare genetic variants could alter off-target susceptibility in an individual, even where off-target editing was not detected in the samples evaluated during development.
Building a Robust Genomic Safety Strategy
A genomic safety strategy should be staged, risk-based, and designed early in development. The goal is not to apply every available assay, but to select complementary methods that address the most plausible risks associated with the editing modality and therapy. Assessment typically begins with characterization of the intended on-target outcome – including editing efficiency and editing spectrum – as well as unintended outcomes such as indels, bystander edits, large deletions, inversions, or other complex repair products.5 Targeted deep sequencing can provide sensitive evaluation at defined sites, while long-read or capture-based approaches may be useful when the editing mechanism or target locus creates a credible risk of larger structural changes.
Off-target evaluation generally combines computational prediction with one or more experimental nomination approaches, followed by targeted confirmation at candidate sites. Computational methods help to prioritize sites based on sequence similarity and genomic context, but may not fully reflect chromatin accessibility, cellular biology, or genetic variation. Experimental methods – such as GUIDE-seq, CHANGE-seq, and ONE-seq – can provide complementary perspectives, with the appropriate approach depending on the editor and biological context. Candidate sites should then be assessed using sufficiently sensitive sequencing methods in biologically relevant materials.
Importantly, off-target analysis alone may not adequately address all genomic safety concerns. Depending on the product, developers may also need to evaluate translocations, large structural variations, or other integrity outcomes. These considerations may be particularly important for editors that rely on double-stranded breaks, multiplex editing strategies, long-lived or proliferative cell populations, and products involving extensive ex vivo cell expansion.
A robust safety strategy should also consider the potential impact of human genetic variation. Variants in either the intended target region or candidate off-target sites can alter editing efficiency or create patient-specific susceptibility to unintended editing. Testing in multiple donors and patient-derived material, as well as variant-aware computational analyses, can help to determine whether the safety assessment is relevant to the intended patient population.
No single assay can fully characterize every genomic risk. The strongest strategy is therefore a scientifically justified package that explains the strengths and limitations of each method, integrates the results across assays, and defines the residual uncertainty in the context of the product’s anticipated clinical benefit.
Personalized Therapies
One of the most promising aspects of CGTs is their ability to address highly specific disease-causing variants, including variants affecting an individual patient or a very small patient population. These “N-of-1” or individualized therapies raise unique regulatory questions because conventional, randomized trials may be infeasible, and the relationship between prior platform knowledge and the new product must be carefully defined.
FDA’s draft plausible mechanism framework, issued in February 2026, is intended to help developers generate substantial evidence of effectiveness and safety for individualized therapies targeting specific genetic conditions with known biological causes when conventional development paradigms are impractical.6 The framework is particularly relevant for ultra-rare diseases, and may apply to genome editing, RNA-based products, and other modality classes.
Operational Implications for CGT Developers
The expanding scope of genomic safety assessment is creating important operational challenges for CGT developers. Developing, qualifying, and interpreting assays for off-target activity, on-target outcomes, structural variants, and broader genomic integrity requires specialized expertise in genome editing, NGS, bioinformatics, and regulatory science. Requirements are also likely to evolve during development, meaning that a strategy used for a previous therapy may not be sufficient for a future program.
Therapeutic developers can build these capabilities internally or engage with specialized assay providers. Internal teams offer direct control but require sustained investment. External partners can provide specialized platforms and scalable capacity. The most effective partners are not simply those offering a broad assay menu, but those that can help to develop a fit-for-purpose evidence package that is appropriate for where the therapy is in development, with transparent bioinformatics, defined assay limitations, and experience supporting regulatory interactions.
Conclusion
The regulatory landscape for CGTs is moving towards more explicit, risk-based assessment of intended editing outcomes, off-target activity, and genomic integrity. This evolution reflects both the promise of new editing technologies and the need to understand their potential consequences in clinically relevant biological settings. Early integration of genomic safety strategies, assay development, bioinformatics, and regulatory engagement can reduce late-stage uncertainty and support more efficient development. As gene editing technologies continue to advance, organizations that combine rigorous science with clear regulatory reasoning will be best positioned to translate innovative therapies into meaningful benefits for patients.
References
- U.S. Food and Drug Administration. Human Gene Therapy Products Incorporating Human Genome Editing: Guidance for Industry. January 2024. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/human-gene-therapy-products-incorporating-human-genome-editing
- U.S. Food and Drug Administration. Safety Assessment of Genome Editing in Human Gene Therapy Products Using Next-Generation Sequencing: Draft Guidance for Industry. 2026. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/safety-assessment-genome-editing-human-gene-therapy-products-using-next-generation-sequencing
- Hacein-Bey-Abina S, von Kalle C, Schmidt M, et al. Insertional oncogenesis in 4 patients after retrovirus-mediated gene therapy of SCID-X1. J Clin Invest. 2008;118(9):3132-3142. doi: 10.1172/JCI35700
- Vertex Pharmaceuticals Incorporated. Exagamglogene autotemcel (exa-cel) for the treatment of sickle cell disease in patients 12 years and older with recurrent vaso-occlusive crises: Sponsor briefing document, BLA 125787. Cellular, Tissue and Gene Therapies Advisory Committee Meeting, U.S. Food and Drug Administration; October 31, 2023. https://www.fda.gov/media/173415/download
- Kosicki M, Tomberg K, Bradley A. Repair of double-strand breaks induced by CRISPR-Cas9 leads to large deletions and complex rearrangements. Nat Biotechnol. 2018;36(8):765-771. doi: 10.1038/nbt.4192
- U.S. Food and Drug Administration. Considerations for the Use of the Plausible Mechanism Framework to Develop Individualized Therapies That Target Specific Genetic Conditions With Known Biological Cause: Draft Guidance for Industry. February 2026. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-plausible-mechanism-framework-develop-individualized-therapies-target-specific
About the Author
Huei-Mei Chen is Director of Gene Editing Sciences at MaxCyte, Inc., where she leads scientific and strategic initiatives supporting cell and gene therapy development. She is experienced in advancing therapeutic programs from early research through preclinical and IND stages, including AAV-based gene therapies, Cas9-edited cell therapies, and in vivo base editing programs. Her expertise includes genomic safety, NGS assay development, genetic toxicology, and regulatory strategy.