CRISPR Gene Therapy for Sickle Cell and Thalassemia Shows Strong Efficacy, but Long-Term Safety Questions Remain

A systematic review of 21 clinical and preclinical studies has concluded that CRISPR/Cas9 gene-editing therapies for transfusion-dependent beta-thalassemia and sickle cell disease deliver striking short-term clinical benefits, with the approved therapy exagamglogene autotemcel achieving transfusion independence in 91 percent of beta-thalassemia patients and freedom from vaso-occlusive crises in 97 percent of sickle cell patients. But the review, published in Gene Therapy, also warns that these numbers rest on follow-up data covering only roughly one and a half to two years, leaving open the possibility that rare genotoxic consequences could emerge decades after treatment.

The analysis, conducted by Abeer M. Al-Subaie and J. Francis Borgio following the PRISMA 2020 reporting guidelines, screened more than 22,000 raw records from PubMed and Google Scholar published between 2014 and 2026. After removing duplicates, retracted studies, secondary literature and work on non-CRISPR modalities such as base editing, prime editing and lentiviral vectors, the authors narrowed the field to 21 studies that reported foetal haemoglobin induction, transfusion independence and safety outcomes for standard CRISPR/Cas9 interventions. The authors focused on conventional CRISPR/Cas9 because it constitutes a mature, clinical-stage benchmark supported by extensive human data, in contrast to newer editing platforms still in early-phase evaluation.

The therapeutic principle behind these treatments targets the +58 BCL11A erythroid-specific enhancer. Disrupting this regulatory element reactivates foetal haemoglobin, the oxygen-carrying protein normally produced in the fetus that is switched off after birth. In patients with sickle cell disease or beta-thalassemia, mutations in the HBB gene cripple adult haemoglobin production, forcing lifelong transfusions or risky donor stem cell transplants. By editing the patient’s own haematopoietic stem and progenitor cells to raise foetal haemoglobin, the therapy sidesteps both problems without requiring a donor.

The clinical evidence centres on exagamglogene autotemcel, known as exa-cel. In the phase 3 CLIMB THAL-111 trial, 32 of 35 transfusion-dependent beta-thalassemia patients, or 91 percent, achieved transfusion independence for 12 months. In the CLIMB SCD-121 trial, 29 of 30 sickle cell patients, or 97 percent, remained free of vaso-occlusive crises for 12 months. Mean foetal haemoglobin levels rose to as high as 11.9 grams per decilitre in treated patients. Early paediatric experience reinforced these findings: in a first-in-child trial, two children with transfusion-dependent beta-thalassemia showed stable multilineage engraftment and transfusion independence lasting more than 18 months, with total haemoglobin climbing to 15.0 and 14.0 grams per decilitre respectively, backed by allelic editing rates of 85.46 to 89.48 percent in the bone marrow.

Quality-of-life data added a further dimension. Secondary analyses of the CLIMB trials found that both adult and adolescent beta-thalassemia patients reported clinically meaningful improvements on standardized instruments including the EQ-5D-5L, FACT-G and PedsQL, with benefits sustained up to 48 months after treatment. Parallel assessments in sickle cell cohorts showed that successful haematological improvement and elimination of vaso-occlusive crises correlated with durable gains across physical, emotional and social domains. The review also notes that economic modelling suggests substantial reductions in the lifetime burden of chronic transfusion and associated healthcare costs.

On the safety side, the review draws a sharp distinction between the gene-editing step itself and the myeloablative conditioning that must precede it. Severe adverse events recorded in the trials, including febrile neutropenia in up to 54 percent of patients, stomatitis in up to 55 percent and thrombocytopenia in up to 48 percent, were attributed primarily to busulfan-based conditioning rather than to the CRISPR/Cas9 editing process. Within the CLIMB clinical programme, the authors report no cases of treatment-related mortality, secondary malignancy or graft failure. Yet they caution that gene-editing product safety and conditioning toxicity cannot be fully separated, because conditioning is embedded in the therapeutic procedure and edited clones may interact with conditioning-induced stress pathways during long-term repopulation.

Preclinical studies compiled in the review paint a more nuanced picture of genotoxic risk. Most validated platforms reported negligible off-target activity. Whole-genome sequencing of corrected clones from two patients found zero candidate mutations across 1,467 computationally predicted off-target sites, with no unintended alterations in tumour suppressor genes. Edited cells across models ranging from induced pluripotent stem cells to primary CD34-positive haematopoietic stem and progenitor cells typically maintained normal karyotypes, morphology and multilineage colony-forming capacity. But the review catalogues rare warning signs: high doses of ribonucleoprotein delivery reduced cell viability and produced, in isolated instances, intergenic off-target indels at 44 percent frequency at one site and a rare translocation between that site and the HBB locus. Single-nickase approaches were associated with unexpected large-scale on-target deletions and complex rearrangements. Low-level off-target editing at the highly homologous HBD locus appeared in two corrected clones, and rare intronic off-target editing of 0.45 to 1.0 percent was identified in donors carrying a specific variant, rs114518452-C.

Engraftment dynamics, the review’s second major theme, appear encouraging but incompletely characterized. Human haematopoietic stem and progenitor cells edited via CRISPR/Cas9 with rAAV6 delivery achieved long-term multilineage engraftment in immunodeficient mouse models for approximately 16 weeks. Selection-free ribonucleoprotein-mediated correction of the sickle mutation also engrafted successfully in NSG mice over a 16-week observation period, though a decline in in vivo homologous recombination rates was noted compared with preinjection levels. The authors emphasize that several foundational studies relied on induced pluripotent stem cells or immortalized erythroid lines, highlighting the need for future longitudinal in vivo assessments to fully characterize clonal dynamics over time.

To address these gaps, the authors propose a Multilayer Safety Risks Screening and Surveillance framework spanning five integrated layers. The first covers bench and preinfusion screening, using automated high-throughput off-target analysis tailored to individual patient genotypes to catch rare variants that could influence unintended editing. The second monitors mobilization and conditioning, tracking risks associated with G-CSF and plerixafor mobilization and busulfan-based conditioning such as vaso-occlusive crises or veno-occlusive disease. The third layer mandates real-time tracking of clonal composition immediately after transplantation, using single-cell sequencing to detect lineage skewing or unexpected indel expansion. The fourth establishes a lifelong registry for patients extending beyond 15 years, employing longitudinal peripheral blood chimaerism assessments and targeted amplicon sequencing. The fifth adds integrated patient support, tracking psychological well-being and quality of life alongside equitable access and community-based counselling.

The review situates this proposal against existing regulatory structures. The United States Food and Drug Administration emphasizes standardized pre-IND validation and 15-plus years of long-term follow-up through risk evaluation and mitigation strategies, while the European Medicines Agency prioritizes decentralized registry-driven pharmacovigilance under advanced therapy medical product guidelines alongside vector shedding surveillance. The International Council for Harmonization S12 guidelines provide harmonized recommendations for nonclinical biodistribution studies of gene therapy products. The authors argue their framework extends beyond these standards by integrating high-throughput sequencing and off-target profiling with multiomics clonal tracking to anticipate insertional mutagenesis and chromosomal aberrations before clinical manifestation. They acknowledge that current surveillance protocols frequently fail to detect complex structural variants, rare large deletions or balanced translocations occurring below next-generation sequencing limits of detection, and that the long-term immunogenicity of persistent bacterial Cas9 ribonucleoproteins and cumulative epigenetic drift during ex vivo cell expansion remain unresolved questions. The authors also point to a recent human trial involving base editing through viral delivery that revealed a fatal immune reaction and thrombotic microangiopathy as evidence that advanced genetic therapies carry profound systemic risks requiring strict risk stratification. Their central conclusion is measured: CRISPR/Cas9 therapeutics for beta-thalassemia and sickle cell disease constitute a treatment option within current clinical frameworks, but continuous, unbiased, lifelong genomic surveillance is critical to ensure their safety and scalability for global patient care.

Subject of Research: Systematic review of genotoxic safety, efficacy and long-term engraftment dynamics of CRISPR/Cas9 gene-editing therapeutics for beta-thalassemia and sickle cell disease

Article Title: Genotoxic safety and long-term engraftment dynamics of efficient translational CRISPR/Cas9 therapeutics for β-haemoglobinopathies from bench to clinical trials

Article References: Al-Subaie, A. M., & Borgio, J. F. (2026). Genotoxic safety and long-term engraftment dynamics of efficient translational CRISPR/Cas9 therapeutics for β-haemoglobinopathies from bench to clinical trials. Gene Therapy. https://doi.org/10.1038/s41434-026-00656-8

Image Credits: AI Generated

DOI: 10.1038/s41434-026-00656-8

Keywords: CRISPR/Cas9, gene editing, beta-thalassemia, sickle cell disease, exagamglogene autotemcel, BCL11A enhancer, foetal haemoglobin, off-target effects, busulfan conditioning, engraftment, genotoxic safety, long-term surveillance