Gene therapy and gene editing
Gene therapy adds a working gene to a patient's cells, and gene editing changes the faulty DNA where it lies. Both depend on carrying genetic material into the right cells, both carry risks of their own, and both moved slowly from the first trial in 1990 to a growing list of approvals, under regulators who ask for particular evidence before approval and for follow-up long after it.
What gene therapy and gene editing are
Both technologies start from the central dogma: a gene's DNA is transcribed into RNA, processed into messenger RNA and translated into protein. A mutation in the DNA can yield a nonfunctional or dysfunctional protein, and so a disease. Most drugs target proteins, and for genetic disorders most available drugs target the faulty protein. Therapeutics can also target RNA or DNA, and that is where both technologies work, getting around the mutations that produce nonfunctional or dysfunctional proteins. Targeting DNA to manage disease suits two situations. Monogenic disorders, caused by a defect in a single gene, number in the thousands, among them cystic fibrosis, thalassaemia and Huntington disease, and together affect millions; some are very rare, and most have no cure and no treatment. Cancer is caused by mutations in somatic cells that let them grow uncontrollably, and restoring the mutated genes' function could control tumour growth and metastasis.
Gene therapy introduces new DNA to the cells, and gene editing fixes the faulty DNA; both have the potential to treat or cure human genetic disorders. Gene therapy uses a vector to carry DNA or RNA coding for a functional protein into target cells, where it augments the defective gene without removing or modifying the mutated sequence, so as to alter gene expression and prevent, halt or reverse a pathologic process. Gene editing changes DNA in its native location, altering the organism's DNA sequence rather than augmenting gene expression. CAR T-cell therapy is a type of gene therapy, and CRISPR is a gene editing tool.
Other definitions draw gene therapy more widely, some with editing inside it. In one, gene therapy is the long and arduous search for ways to change identified disease-causing mutations, so as eventually to treat or cure the diseases they cause. Its basic premise is to override the effects of a mutation, either by providing a correct copy of the gene to the correct target tissue, essentially ignoring or overriding the faulty copy, or by targeting the mutation directly and correcting its expression with editing. In another, gene therapy is the manipulation of genes or their expression inside cells to treat a disease or stop its progress, by replacing a mutated gene with a healthy copy, turning a mutated gene off or a healthy one on, or correcting the mutated gene. In a third, it is the modification of genes within a person's cells to target the underlying causes of disease, by replacing faulty genes, deactivating malfunctioning ones or introducing new genes to fight illness. Gene therapy offers the potential to treat and even cure genetic disorders, and it promises to transform treatment for conditions once considered incurable.
Gene editing is the technology researchers use to remove and replace mutated genes and to make the genome easier to manipulate. It hones gene therapy, and is in some ways more exacting than most gene therapy approaches because it gives researchers a precise way to target the problem area in specific genes. If successful, this easily adaptable technology could transform disease treatment and possibly offer a cure. By 2017 researchers had been studying gene editing methods in laboratories for well over a decade.
Delivery into the cell
Both gene therapy and gene editing need a delivery mechanism, such as a viral vector, to carry the therapeutic elements to target cells. A vector is something, such as a bacteriophage or a plasmid, that transfers genetic material into a cell. Vectors evolved over millennia to carry genetic material into cells, and that capacity has been harnessed to transfer genes into cells in a dish or in animal or human tissue, sometimes into the genome itself, at designated or nonselected sites. Viruses are often used as vectors because of this natural ability, and the virus is modified so that it can no longer cause infectious disease. A vector is most often engineered from a virus made nonpathogenic by removal of its normal viral genes.
In 2019 retroviral and adenoviral vectors were the ones most commonly used, although other viral vectors can be used, and the choice of vector depends on the purpose of the gene therapy. An integrating vector, such as a retroviral one, puts the DNA into the genome so that it passes to daughter cells, which suits stem cells, at the risk of unwanted expression if it is inserted in the wrong place; a non-integrating one, such as an adenoviral vector, leaves the DNA in the nucleus, where it is transient and serves cells that no longer divide. Adeno-associated viral vectors are preferred in vivo for their good safety profile. By 2023 the vectors most used in gene therapy and for CRISPR-Cas9 were adeno-associated viral vectors in vivo and lentiviral vectors ex vivo, and adeno-associated ones were the commonest choice, because their low immunogenicity reduces the likelihood of an inflammatory response. The CRISPR-Cas9 system exceeds their carrying capacity, so special versions are constructed for it. The same 2023 account also holds that lentiviral vectors, despite their larger carrying capacity, are not often used, because they can integrate randomly into the genome.
Delivery is either in vivo, with the vector given directly to the patient, or ex vivo, with the patient's cells removed, given the gene and returned. An in vivo example is injection into retinal cells for a genetic eye disease; in an ex vivo one, a patient's white blood cells are removed, given the functional gene in a vector, screened to confirm the transfer and infused back. CAR T therapies modify cells outside the body, whereas Spark Therapeutics, a publicly traded gene therapy company spun off from the Children's Hospital of Philadelphia, infuses the virus directly into patients, where it finds its target cells and delivers its genetic payload. Ex vivo delivery is more efficient and less likely to provoke an immune response, but it is not possible for every cell type — brain and retinal cells among them, although some stem cell research could allow those to be treated.
Risks of gene therapy
Delivery that is both efficient and cell-specific is often hard to achieve: some disorders need the gene delivered to millions of cells, and it can reach the wrong type of cell. The patient's immune system views the vector as foreign and may react to it, and the reaction can cause serious, even life-threatening, illness; ex vivo delivery can avoid it. An integrating vector lands at random sites and may disrupt critical genes, such as tumour suppressors or the genes that control cell growth, potentially causing cancer, a risk called insertional mutagenesis. Excessive T-cell activation is a further risk. Commercial viability is a pitfall of its own: gene therapy is expensive and often for rare disorders affecting a small population, and approved gene therapies cost on the order of hundreds of thousands of dollars for a single dose. By 2023 new vectors were being researched: baculovirus, which has a large carrying capacity and does not integrate into the genome, may be a safer alternative, and lipid-based nanocarriers and polymer-based nanoparticles were considered promising tools for delivering CRISPR-Cas9.
Editing tools before CRISPR
A nuclease is an enzyme that cleaves the chains of nucleotides in nucleic acids into smaller units, cleaving being cutting or splicing a molecule by breaking a particular chemical bond. Research in the 1980s showed that mammalian cells can take foreign DNA into their genomes by homologous recombination, the exchange of material between two DNA strands with very similar sequences, but at a low rate. Meganucleases, enzymes that recognise a 14–40 base-pair stretch, raised the rate by cutting both strands, but their specificity is a drawback, and the breaks are repaired by non-homologous end joining, which is error-prone and can delete or insert sequence. Zinc finger nucleases and transcription activator-like effector nucleases are the editing tools that came before CRISPR. Each zinc-finger module recognises three base pairs, so modules can be assembled for higher binding specificity and more efficient homology-directed repair, and each transcription activator-like effector nuclease recognises a single base pair, so several can be put together for greater specificity. Both raised editing efficiency, but targeting different parts of the genome meant cloning and expressing new sets of proteins, which proved a challenge. Both are proteins that must be engineered to cut DNA, and they are time-consuming, have low efficiency and have off-target effects.
How CRISPR edits a gene
A palindromic sequence of DNA or RNA reads the same in either direction, as CATGTAC does. CRISPR, for clustered regularly interspaced short palindromic repeats, is the third kind of genetic scissors used in genome editing. It is an adaptive immune system found in bacteria that protects them against phage and viral infection: short palindromic repeats in the bacterial genome are separated by spacers taken from phages and viruses that infected it before, and they guide a Cas effector protein to cut DNA at a specific sequence. The editing complex pairs a guide RNA designed to find the target sequence with a nuclease that cleaves it — Cas9 most commonly, as CRISPR/Cas9, with Cpf1 and modified versions of Cas9 also used or under investigation. CRISPR was identified in bacteria in the 1980s, first in Escherichia coli in 1987. The systems fall into two classes and six types, the most widely used being the type II Cas9 of Streptococcus pyogenes. Jennifer Doudna and Emmanuelle Charpentier received the 2020 Nobel Prize in Chemistry for their breakthrough work on CRISPR-Cas9.
Like the older tools, CRISPR changes a gene's structure by targeting and correcting specific mutations in the DNA sequence; it differs from them in its ease of use and, in some instances, its specificity. To target a new sequence only the guide RNA has to change, with no complex cloning or protein engineering. CRISPR-Cas9 lets scientists modify the DNA of living organisms selectively and precisely, and its discovery has changed the approaches to gene therapy. It is a relatively simple, inexpensive technique for removing and replacing specific gene sequences, introduced in 2015, and genetics has moved forward at an unprecedented pace since. Gene editing with it has proved fast, efficient and versatile, so it has many potential applications, medicine among them. The system has revolutionised biology, letting researchers study the functions of genes and their role in disease, the regulation of gene expression and epigenetic modification. Beyond human and animal disease research, it has been used in plant biotechnology and agriculture to produce crops that resist stress and give higher yields.
Once the DNA is cut, the mutated gene can be revised, by replacing one nucleotide with another, or removed, by cutting a segment out, and either knocks the dysfunctional gene out. Or it can be replaced from template DNA, which restores its function. The cut is a double-strand break, repaired either by non-homologous end joining, which introduces unwanted changes and serves revision or removal, or by homology-directed repair with donor DNA, which serves replacement. Research has therefore gone into raising the rate of homology-directed repair, by changing the Cas9 genes or inhibiting the non-homologous end-joining pathway. In 2017 CRISPR was showing great promise in guiding the targeting machinery to the desired location and removing the faulty gene, and in some cases in replacing it with a normal gene template, which was harder to do. A "dead" Cas9 that still targets but cannot cut lets CRISPR interference switch a gene on or off, or regulate it, reversibly.
Limits of CRISPR editing
Like a gene therapy, the CRISPR complex, with template DNA if needed, can be delivered in vivo or ex vivo, by viral or non-viral systems. The guide RNA–Cas9 machinery may be too large to package in viral vectors , because Cas9 variants found in nature are large proteins, and scientists have been looking for smaller Cas9 variants in other bacterial species. Lipid nanoparticles and other delivery methods were being explored as alternatives. In vitro, plasmids or RNAs carrying the CRISPR system are put into cells with transfection reagents, which cannot be done in vivo. A gene editing treatment needs a knockout efficiency of 60% to 95% to work, which can be difficult to achieve. On-target editing efficiency, the rate of modification at the intended sites, can be low. Editing can also leave two or more genetic lines in one organism (mosaicism), and it can change unintended sites (off-target editing). In 2019 research was under way to find alternatives to Cas9 or ways to improve its efficiency.
Off-target effects and their detection
Off-target effects are nonspecific, unintended genetic modifications that can arise during gene editing, as when the tool cuts not only at its target but at unintended sites with similar DNA sequences, and they are a major concern with CRISPR-based approaches. Editing specificity depends on the design of the guide RNA, and the risk can be controlled to some degree by that design and by the choice of nuclease, including engineered high-fidelity versions of Cas9. In 2019 regulators were expected to want evidence beyond prediction: in vitro and cell-based assays would likely be needed to show minimal off-target effects before gene editing products are approved , and health authorities would most likely require more sensitive methods in clinical trials, detecting off-target mutations in an unbiased, genome-wide way.
Computer algorithms predict off-target sites quickly, cheaply and generally accurately, which makes them useful when designing the guide RNA and very good for basic research. Their predictions, however, depend on assumptions about how the nucleases work, and because algorithms vary, their results do not always agree. In vitro assays detect double-strand breaks in the sequencing data of cell-free genomic DNA after cleavage and can find mutations at frequencies below 0.1%, but cannot detect mutations that occur inside cells; cell-based techniques, equally sensitive, identify off-target sites in a specific cell type under particular experimental conditions.
The first trials and approvals
The first gene therapy trial, and the very first for an inherited disease, took place in the United States on 14 September 1990, in adenosine deaminase deficiency, a disorder caused by an enzyme deficiency. Gene therapy replaced the genes that make the enzyme in the patient, Ashanti DeSilva, leaving a healthy young woman who no longer needs treatment. The first trials followed years of preclinical research, and in 2017 gene therapy was still largely experimental. A gene therapy trial–related death in 1999 drastically slowed the field; increased monitoring of US gene therapy trials followed, and consequently the United States approved no gene therapy product until 2017. Corey Haas, a boy going blind from the rare hereditary disorder Leber congenital amaurosis (LCA2), received gene therapy in September 2008, and by 2010 he could see.
China and the European Union approved gene therapies before the United States did. Gendicine, the world's first commercial gene therapy, was approved in China in 2004 for head and neck squamous cell carcinoma in combination with radiotherapy. Glybera (alipogene tiparvovec), for hereditary lipoprotein lipase deficiency, was the first gene therapy approved in the European Union, by the European Commission in 2012. It was not approved in the United States, and its sponsor let the marketing authorisation expire in 2017 because the therapy was a commercial failure, costing more than €1 million a dose, and could not gain US approval. The European Commission approved GlaxoSmithKline's Strimvelis in 2016 for children with severe combined immunodeficiency due to adenosine deaminase deficiency, a treatment available in the United States and Canada in 2017 only to some patients, through a clinical trial at the University of California, Los Angeles.
The United States approved three gene therapies in rapid succession in late 2017, and the first two, Kymriah and Yescarta, are CAR T-cell therapies. Kymriah (tisagenlecleucel, Novartis), approved on 30 August 2017, was the first therapy based on gene transfer that the FDA approved. That autumn it was the only FDA-approved genetic therapy available in the United States. Neither Kymriah nor Strimvelis uses CRISPR. Late in 2017 an FDA advisory committee voted to recommend approval of Luxturna (voretigene neparvovec), Spark Therapeutics' lead candidate, as a potential one-time treatment for vision loss due to confirmed biallelic RPE65-mediated inherited retinal dystrophies, a group of rare blinding conditions caused by one of more than 220 different genes. If approved, it would be the first gene therapy authorised to treat an inherited disease in the United States. The FDA approved Luxturna in December 2017 for an inherited form of retinal dystrophy that can cause blindness, as the first in vivo gene therapy in the United States , and it is the first gene therapy that targets a disease caused by mutations in a specific gene.
By September 2023 the European Medicines Agency had approved 14 gene therapies, with indications from ultra-rare and rare genetic diseases to various cancers. With their years of approval, they are Imlygic (2015) for unresectable melanoma lesions; Strimvelis (2016) for severe combined immunodeficiency due to adenosine deaminase deficiency; Kymriah, Luxturna and Yescarta (2018); Libmeldy, Tecartus and Zolgensma (2020); Abecma (2021); Breyanzi, Roctavian and Upstaza (2022); and Carvykti and Hemgenix (2023).
Clinical trials over time
By autumn 2017 gene therapy had become a robust research environment. The NIH reported more than 300 recruiting, active and completed gene therapy trials. A ClinicalTrials.gov search for active and recruiting trials in the United States found 129, targeting cancers, blood disorders, heart and vascular disease and neurological diseases such as Alzheimer disease, and 187 once other countries were included. Spark had two active trials, in haemophilia B and choroideremia, and two recruiting, in haemophilia A and haemophilia B. The Children's Hospital of Philadelphia's Raymond G. Perelman Center for Cellular and Molecular Therapeutics had manufactured and released 40 clinical-grade products used in 20 studies from phase 1 to phase 3.
On the research level, CRISPR had by then been used to edit nucleic acids in plants, microbes and some animals, mice among them, and to edit several genes at once to create small and large animal models. The geneticist George Church used it to splice genes from a frozen woolly mammoth into the DNA of Asian elephant skin cells, the first time mammoth genes had functioned since the animals became extinct, according to a 2015 popular science report. A team at the Children's Hospital of Philadelphia reported in January 2017 that CRISPR/Cas9 decreased expression of the mutant huntingtin gene in laboratory animals and in cells taken from patients with Huntington disease. Huntington disease is caused by a repeat expansion within the first exon of that gene and destroys brain cells while affecting other tissues. As of August 2017 there were no NIH-sponsored US trials applying CRISPR directly to patients, but five trials in China were listed as recruiting — four in advanced cancers, one in HIV — and Memorial Sloan-Kettering Cancer Center and Juno Therapeutics were recruiting for a phase 1 trial in which gene editing would make a CAR T therapy for chronic lymphocytic leukaemia more effective.
At the end of 2018 no therapy based on gene editing had been approved anywhere. One count of gene editing trials found 17 active or recruiting trials, which it called preclinical, most of them in China, and one actively recruiting in the United States. Another, dated 13 December 2018, found 20 interventional trials involving CRISPR registered on ClinicalTrials.gov, most of them in cancer and a few in monogenic disorders such as thalassaemia and sickle cell disease. Animal studies had used CRISPR in models of heart disease, Duchenne muscular dystrophy, phenylketonuria and hereditary tyrosinaemia type 1. CRISPR had been used to engineer cancer cells that secrete a protein triggering a death switch in other cancer cells, reducing tumour size and prolonging survival, and to engineer a skin patch secreting an enzyme that digests cocaine, though whether that bears on treating substance use disorders was unclear. Companies' preclinical pipelines targeted transthyretin amyloidosis, alpha-1 antitrypsin deficiency, primary hyperoxaluria type 1, sickle cell disease, beta thalassaemia, Duchenne muscular dystrophy, cystic fibrosis and acute myeloid leukaemia.
By 2023 more than 3800 gene therapy trials had been run or were ongoing. By then the use of CRISPR-Cas9 as a tool for gene therapy had gained momentum over five years, with 42 trials in progress, mostly phase 1 with a few phase 2/3, in monogenic diseases such as sickle cell disease, beta-thalassaemia and hereditary amyloidosis, in cancers and in infectious diseases including COVID-19. In 2023 Vertex and CRISPR Therapeutics submitted their ex vivo therapy developed using CRISPR-Cas9, exa-cel, for sickle cell disease and beta-thalassaemia for approval in the US and the EU, and the UK's Medicines and Healthcare products Regulatory Agency approved it in November 2023.
What the FDA and the EMA require before approval
Because gene therapy's path to clinical application is complex, often with significant technical challenges and ethical considerations, regulatory guidance and frameworks are crucial in governing the design, testing and application of gene therapy in human trials, to ensure the therapies are safe and effective before they reach the clinic. Regulators such as the FDA and the EMA assess the risks and benefits through nonclinical testing in laboratories and animals, phased clinical trials and postmarketing surveillance, to protect patients from risks such as unintended genetic changes or long-term adverse effects. In 2023 the efficacy and safety of the technology were still not clear, although its benefits and risks were understood, so the FDA, the EMA and other regulators had strict approval criteria and post-marketing surveillance requirements. Getting a gene therapy trial up and running is more complicated than for a conventional drug: gene transfer is riskier, and the vector, how much of it, where it goes and whether that can be directed, the time course, the level of gene expression, what happens if something goes awry, and toxicity all have to be determined.
In the United States gene therapy is regulated by the FDA's Center for Biologics Evaluation and Research, which oversees biological products including vaccines and cell and gene therapies, and new gene therapy products are approved through a Biologics License Application. The FDA's guidance takes a structured approach to nonclinical testing, manufacturing and trial design. Nonclinical studies, in vitro and in vivo, assess biodistribution, persistence and potential toxicity before trials begin. The chemistry, manufacturing and control sections of an investigational new drug application describe the product's composition, manufacturing process, quality control and stability, and manufacturers must demonstrate control of critical manufacturing steps for identity, quality, purity and potency throughout shelf life, and rigorous testing and documentation of raw and starting materials. As for all clinical trials, a gene therapy trial needs an appropriate population, a well-evidenced dosing regimen, suitable safety and efficacy endpoints, and on-trial and long-term safety monitoring. For the small populations of rare diseases, the FDA suggests adaptive designs and natural history data to support efficacy assessments, and clear criteria for patient selection and monitoring are crucial to assessing safety and efficacy accurately.
The EMA's quality guidelines set requirements for controlling the vector, the transgene and the final product; its nonclinical guidelines require extensive studies in relevant animal models including genotoxicity, immunogenicity and tumorigenicity; and its clinical guidelines encourage innovative trial designs, patient registries and real-world evidence to support marketing authorisation applications. It offers flexibility through scientific advice, early dialogue and adaptive pathways.
The primary challenge in nonclinical studies of gene therapy is assessing the products thoroughly for their biodistribution and potential off-target effects. The FDA recommends extensive biodistribution studies identifying target and non-target tissues and assessing potential off-target effects, including the risk of insertional mutagenesis, particularly in gonadal tissue to prevent germline integration. The EMA's guidance stresses the distribution, persistence and clearance of the product in target and non-target tissues, and like the FDA it stresses evaluating off-target effects and insertional mutagenesis, particularly in germline cells. In trials, the FDA asks for a safety monitoring strategy, for example a toxicity grading system and a toxicity management plan, especially for off-target editing and the side effects of on-target editing.
Another significant challenge is ensuring product consistency and safety through robust vector design and manufacturing controls. Variability in vector design or production can lead to inconsistent therapy outcomes. The FDA recommends vectors with high specificity and minimal off-target effects, using tissue-specific promoters and enhancers, and stringent manufacturing controls. The EMA stresses stringent quality control of purity, potency and stability, and advises vectors with well-characterised safety profiles together with advanced techniques to minimise off-target effects and insertional mutagenesis. For trial conduct, the FDA requires protocols that specify primary and secondary endpoints, selection criteria and a design able to answer the safety and efficacy questions, and it stresses clinically meaningful endpoints focused on patients' reported outcomes. It recommends that first-in-human trials of gene-editing products enrol patients with no other feasible or warranted option, who should be at a less advanced or moderate stage of disease because of the products' potential risks in severe disease. By emphasising rigorous nonclinical studies, detailed clinical trial protocols and long-term patient monitoring, the agencies address the scientific and ethical challenges of gene therapy, and their oversight helps mitigate risks such as off-target and delayed adverse effects. One 2024 overview of the guidance holds that the frameworks must adapt as gene therapy evolves, which may mean integrating real-world evidence, advanced computational models for safety assessment and closer collaboration between regulators, researchers and industry, and that future oversight may adopt more flexible pathways for emerging gene-editing technologies.
Expedited approval on limited data
Both agencies offer streamlined processes to fast-track gene therapies aimed at unmet medical needs. The FDA's include the Regenerative Medicine Advanced Therapy designation — intensive guidance, rolling review and early discussion of surrogate endpoints that could support accelerated approval — and the Breakthrough Therapy and Fast Track designations; the EMA's is its PRIME scheme.
The European Medicines Agency recommended on 14 November 2025 that Waskyra (etuvetidigene autotemcel), from Fondazione Telethon ETS, be authorised as the first gene therapy for Wiskott-Aldrich syndrome, for people aged 6 months and older with a mutation in the WAS gene. The syndrome is a rare inherited disease caused by abnormalities in the gene for the WAS protein found in blood cells and certain immune cells. Patients with a compatible donor are treated with haematopoietic stem cell transplantation, but most have none, and for them the need is unmet. The therapy is made from the patient's own CD34+ stem cells, collected from the blood and genetically modified in the laboratory to produce a functional WAS protein. Transplanted back after a conditioning regimen, the cells migrate to the bone marrow and make healthy blood and immune cells that produce a functional WAS protein, helping to relieve the symptoms of the disease. Waskyra is given once, by infusion. The FDA approved it in December 2025 as the first cell-based gene therapy for the syndrome, one of two cell and gene therapies it approved that month for rare, life-threatening diseases.
The evidence came from 27 patients: a main single-arm trial in 10 children aged 1 to 9, supported by another trial and an expanded access programme, including compassionate use and hospital exemption, covering 17 patients aged 1 to 35. The annualised rate of severe infections fell from 2.0 events in the 12 months before treatment to 0.15 in the first to second year after it and 0.12 in the second to third, and that of moderate and severe bleeding episodes from 2.0 to 0.16 in the second to third year. The most common side effects came from the procedures and medicines needed to receive the therapy — the conditioning regimen and pre-treatment, infections related to the infusion device and bleeding at the catheter site. The Committee for Advanced Therapies found the benefits outweighed the possible risks for patients needing a stem cell transplant with no suitable donor, and the human medicines committee agreed and recommended approval. The medicine had orphan designation and was supported through an agency pilot offering enhanced support to academic and non-profit developers of advanced therapies addressing unmet medical needs.
The main drawbacks in developing a cell or gene therapy product are the limited sample size, the product's long-term safety risk profile and, in some cases, the long duration of the clinical trials. Because of the unmet medical need, these products often gain approval through an expedited pathway on limited clinical data, as Waskyra did. Collecting post-approval data is therefore imperative to ascertain long-term efficacy or durability and to monitor long-term safety, including unknown adverse effects and mortality due to the underlying disease or the treatment.
Long-term follow-up
Cell and gene therapies offer the potential for durable or even curative benefit through permanent or long-acting biological modification, and that promise has brought heightened regulatory scrutiny, extending oversight of some gene therapy products well beyond conventional clinical trial timelines. The FDA's 2020 final guidance established long-term follow-up as an industry standard for high-risk gene therapy products, with follow-up of about 5 to 15 years depending on the therapy's biological characteristics and product-specific risk factors, to detect delayed adverse events that may not be observable during active clinical development. The follow-up runs as an extension of the parent study or under a dedicated protocol separate from it. Sponsors are required to prepare a follow-up protocol with visit schedules, sampling plans, monitoring tests and case histories, and they must report serious adverse events and unexpected suspected adverse reactions and submit periodic safety reports to the FDA. The guidance sizes the follow-up by risk, through five core questions answered in order: genome editing, ex vivo-only use of the vector, persistence in preclinical studies, integration or other genomic alteration, and potential for latency and reactivation. Preclinical data are encouraged for assessing delayed safety risks where the complete safety profile from clinical experience is lacking. The durations it recommends are up to 5 years for non-integrating adeno-associated virus vectors; up to 15 for herpes virus vectors able to establish latency, microbial vectors known to persist and genome-editing products; and 15 for integrating vectors such as gammaretroviral and lentiviral vectors and transposons.
The EMA tailors follow-up to the product: for viral vectors with a potential for chromosomal integration or latency, assessments before treatment and at 3, 6 and 12 months for at least 5 years, then annually until the data show no risk worth following, with separate recommendations for non-integrating viral vectors, plasmids and non-viral vectors. It also requires a detailed risk management plan in a clinical trial or marketing authorisation application covering pre- and post-marketing surveillance, and any change to the product's risk profile during the trial or after marketing should prompt a revision of the follow-up plan.
Patients in gene therapy trials must fully understand the risks, benefits and experimental nature of the trials and the available alternatives. They should be asked to give informed consent to long-term follow-up, which must describe its purpose, expected duration, necessary assessments or procedures, and any adverse reactions that may relate to the product.
Keeping patients in rigid, trial-centric follow-up for up to 15 years, often after they have benefited, proved hard, with attrition, participant burden and strain on sponsors and sites, and regulators acknowledged that meeting the safety expectations needed new approaches. In September 2025 the FDA released draft guidance on post-approval methods to capture safety and efficacy data for cell and gene therapy products, to help sponsors meet the 5- to 15-year follow-up requirements while balancing data integrity and practical feasibility. For real-world data, it recommends established sources such as registries, insurance claims and electronic health records, robust governance of data integrity and confidentiality, data-processing software compliant with 21 CFR Part 11, and early consultation with the FDA on the choice of data. It recognises that health records, medical claims and vital statistics are not usually designed for clinical outcomes, and sponsors must assess their limits, such as missing rare-disease specifics, and the feasibility of the selected data. Registries offer curated longitudinal data, including laboratory, genetic, imaging and patient-reported data, at a risk of enrolment bias, and after approval they serve to assess long-term durability, collect developmental milestones in children, survey for malignancy and capture fertility and pregnancy outcomes. Decentralised collection is adapted from decentralised trial models, and IRB review and informed consent still apply. For children, sponsors should consider potential lifetime monitoring, since they are often followed into adulthood, and need to manage their transition from paediatric to adult care and re-consent them on reaching adulthood.
In November 2025 the Multi-Regional Clinical Trials Center's working group of patients, advocates, sponsors, academics, contract organisations and oversight bodies released a draft toolkit, for comment until December 2025, built on regional regulators' follow-up guidance and covering investigational and approved gene therapies, including genetically modified cell therapies such as CAR T cells. Among cell therapies, long-term follow-up applies only to those that involve genetic modification. The toolkit's central theme is the Goldilocks concept, a balance between what and how much data is enough and the feasibility and sustainability of the studies. Its considerations and recommendations fall into nine subsections, covering defining purpose and scope with safety primary; collecting only must-have data with patients' perspectives; building flexibility for 5 to 15 years of change; consent timing, re-consent and renewal; retention through engagement, balanced visit frequency, decentralised options and pre-specified dropout analyses; signal detection with a data monitoring committee and CIOMS-based assessment; registration, timely publication including lay results and shared central repositories with COMET outcome sets; operational roles, sites and technology; and the responsibilities of patients, local providers and sponsors. Together the FDA's updates and the toolkit emphasise flexibility, fit-for-purpose strategy, patient-centricity and the integration of real-world data and decentralised elements, to meet safety objectives without compromising feasibility.
Promise and hype
In 2017 many stories about gene technology breakthroughs appeared in the national press, and some exaggerated its accomplishments and future. By 2019 targeting DNA through gene therapy or gene editing to manage disease had become a hot topic. The use of both was on the rise, and it was important for medical writers to understand the science of both, the ethical concerns over their use and the regulations governing them.
In 2017 many genetic diseases — Huntington disease and other neurodegenerative diseases, haemophilia, sickle cell anaemia, blindness, cancers of all types and autoimmune diseases — had no cure; gene therapies offered promise but had for the most part not reached clinical practice; and apart from Kymriah, with its targeted patient group, it was difficult to predict when they would become a common tool. The transition from laboratory models to patient care remained arduous and challenging for gene therapy and its editing tools.
At the end of 2018 both approaches were promising but likely to be expensive, and off-target effects had to be minimised before CRISPR-based therapies could reach patients. In spite of their problems, both offered potential solutions for a variety of diseases. Gene editing can be controversial, but it has the potential to greatly alter the medical landscape of the world.