How big is the risk of off-target edits, and how well can we measure it?
The biggest evidence gap is knowing exactly where base editors make unintended edits—both in DNA and RNA—and whether those edits cause harm. A 2024 study on base editing for chronic granulomatous disease used a high-fidelity PAMless base editor (ABE8e-SpRY) in human hematopoietic stem cells and reported 'minimal off-target or bystander edits' after profiling potential off-target DNA edits, transcriptome-wide RNA edits, and chromosomal perturbations [3]. That sounds reassuring, but the same study notes that this was an investigational new drug-enabling study—meaning it was designed to support a first-in-human trial, not to settle the question definitively. The authors themselves call for further characterization. A 2024 review of base editing technology explicitly lists 'targeting specificity, safety, and efficacy' as ongoing challenges and discusses possible solutions, indicating the field still considers off-target effects a major hurdle [5]. The bottom line: early data from one well-controlled preclinical study look good, but we lack large-scale, long-term safety data in humans.
A related gap is the risk of RNA edits—changes to the messenger RNA that could alter protein production without changing the DNA. The chronic granulomatous disease study did check for transcriptome-wide RNA edits and found them minimal [3], but this is only one study in one cell type. The field needs more systematic data across different base editors, delivery methods, and target tissues before we can confidently say RNA off-target effects are not a problem.
Can we get base editors to the right cells and make the edit last?
Delivery is a major evidence gap, especially for in vivo (inside the body) therapies. The most advanced in vivo data here comes from a phase 1 trial of YOLT-101, a base editing therapy for heterozygous familial hypercholesterolemia delivered via lipid nanoparticles [4]. In the highest dose group (0.6 mg/kg, n=3), a single infusion produced sustained reductions in PCSK9 (74.4%) and LDL cholesterol (52.3%) at 24 weeks. That's impressive, but the sample is tiny, and the therapy targets the liver—a relatively easy organ to reach with lipid nanoparticles. For other tissues like bone marrow or the central nervous system, delivery remains a much bigger challenge. The review paper notes that delivery methods are a key area of ongoing research [5].
Durability is another open question. The familial hypercholesterolemia trial shows effects lasting at least 24 weeks [4], and the CAR T cell trial for leukemia reports ongoing remissions 3-36 months after transplant in 7 of 11 patients [1]. But both are early-stage trials with small numbers and limited follow-up. We don't yet know if base edits can last for years or decades, or if cells carrying the edits will be outcompeted by unedited cells over time. The chronic granulomatous disease study showed that edited alleles persisted after transplantation into immunodeficient mice [3], but mouse models don't perfectly predict human outcomes.
Can cancer cells evade base editing therapies, and how common is that?
Yes, cancer cells can escape, and we don't yet know how often this happens. In the phase 1 trial of base-edited CAR T cells for T-cell acute lymphoblastic leukemia, 2 of 11 patients had 'suspected CD7 negative leukemic escape'—meaning the leukemia cells stopped expressing the target protein CD7, making the CAR T cells ineffective [1]. This is a known problem with targeted therapies, but the rate here (2 out of 11) is a red flag. The trial was small, so the true escape rate could be higher or lower, but it's a clear evidence gap that needs larger studies.
A separate study on base-edited CAR T cells for acute myeloid leukemia (AML) tried to get ahead of this problem by targeting two antigens (CD70 and CD33) simultaneously and also knocking out three cytokines (GM-CSF, IL-3, and FLT3LG) that can support AML growth [2]. In lab tests, these triple-edited CAR T cells showed enhanced anti-leukemia potency and proliferation compared to non-edited controls. But the study is still in preclinical stages—the authors say 'in vivo experiments are ongoing'—so we don't yet know if this multi-target approach will actually prevent escape in patients. The evidence gap here is clear: we have promising lab data but no human data on whether multi-targeting solves the escape problem.
About These Sources
This answer is built on 5 peer-reviewed studies — published from 2024 to 2026, 5 from 2024 or later, 5 in Q1 journals — selected as the most relevant from 6 studies that passed quality screening, drawn from 60 papers retrieved from a database of over 500 million.
Sources used in this answer
Universal base-edited CAR7 T cells for T-cell acute lymphoblastic leukemia
In a phase 1 trial of base-edited CAR7 T cells for relapsed/refractory T-ALL, 7 of 11 patients were in remission 3-36 months after transplant, but 2 patients had suspected CD7-negative leukemic escape, showing target antigen loss can undermine therapy.
Triple base-edited dual targeting CD70/33 CAR T cells to avoid cytokine-mediated resistance in Acute Myeloid Leukemia
Preclinical study of triple base-edited CAR T cells (targeting CD70 and CD33, with knockout of GM-CSF, IL-3, and FLT3LG) showed enhanced anti-leukemia potency in vitro, but in vivo results are still pending.
High-fidelity PAMless base editing of hematopoietic stem cells to treat chronic granulomatous disease
Using a PAMless adenine base editor (ABE8e-SpRY) in human hematopoietic stem cells, up to 70% correction of a chronic granulomatous disease mutation was achieved with minimal off-target DNA or RNA edits, and edited alleles persisted after transplant into immunodeficient mice.
In vivo base editing gene therapy for heterozygous familial hypercholesterolemia: a phase 1 trial.
In a phase 1 trial of in vivo base editing for familial hypercholesterolemia (YOLT-101), the highest dose group (n=3) showed sustained reductions in PCSK9 (74.4%) and LDL cholesterol (52.3%) at 24 weeks, with no grade ≥3 adverse events.
Breaking genetic shackles: The advance of base editing in genetic disorder treatment
A 2024 review of base editing technology identifies targeting specificity, safety, and efficacy as ongoing challenges and discusses possible solutions to advance clinical translation.