Less than 60 minutes after his latest report was published online, Luigi Naldini, MD, PhD, called me to discuss his article, “Genotoxic effects of base and prime editing in human hematopoietic stem cells,” published in Nature Biotechnology. Despite the stark title, Naldini doesn’t give the impression that the use of base editors (BEs) and prime editors (PEs) is catastrophic and should be abandoned. Naldini explained that his team’s results require a more nuanced interpretation, in which there is both good news and bad news.

The bad news is that the research from the veritable San Raffaele Telethon Institute for Gene Therapy in Milan, Italy, shows that “state-of-the-art” BEs and PEs are not truly free of causing DNA double-stranded breaks (DSBs) in human hematopoietic stem and progenitor cells (HSPCs). These precision editing complexes, created by David Liu’s group in a series of important papers from 2016 to 2019, do not require a DNA break for the edit to take effect. However, the detection of translocations and the subsequent unfavorable cellular outcomes by Naldini’s group demonstrate that these editors can also have a number of side effects.

Of perhaps greater concern, Naldini, who is the director of the San Raffaele Telethon Institute for Gene Therapy, says that his group also detected a genome-wide mutagenic load. The mutations caused by BEs and PEs may be completely neutral or not, but he worries that this genome-wide mutagenic load triggers DNA repair mechanisms that may introduce another DNA break.

“We have to be thinking about [genotoxicity],” Naldini told GEN. “This has been a concern that has been raised before but has not really been proven.”

And there is good news, Naldini said. The new research shows the incredible efficiency of the editors in HSPCs, which is maintained long-term in repopulated cells from multiple hematopoietic lineages. He offers that, to a certain extent, the expression and exposure of these editors can be optimized to alleviate genotoxicity.

Hold your position

Reached for comment, David R. Liu, PhD, professor of chemistry at Harvard University and a core member of the Broad Institute, is neither surprised nor concerned when I ask him to comment on Naldini’s Nature Biotechnology report. In addition to sending me 22 references for articles on BEs and PEs vs. nucleases, he lays out some key points that challenge some of the findings.

First, Liu said, it was already known that BEs and PEs can generate intermediates that cause DSBs.

“Like virtually all systems that make covalent changes to DNA, they can generate intermediates that lead to double-strand breaks, as we reported in our original base editor and prime editor papers [studied led by former Liu postdocs Alexis Komor (2016), Nicole Gaudelli (2017), and Andrew Anzalone (2019)], and as many subsequent papers on base editors and prime editors from many labs have also reported,” Liu said.

What’s more, BEs and PEs have lower DSB frequencies (as measured by indel formation) compared to nucleases, typically by one or two orders of magnitude! That is a very substantial benefit if the gene editing application seeks to minimize the consequences of DSBs. In contrast, 100% of nuclease-mediated editing outcomes arise from DSBs because that’s how nucleases edit DNA—they are enzymes that catalyze DSBs. Additionally, nucleases generate uncontrolled mixtures of indels and do not convert a starting sequence to an edited sequence of your choosing. 

“Currently, there are only three classes of robust, programmable mammalian cell gene editing agents: nucleases, BEs, and PEs,” said Liu. “If you wish to make a precise change at a target site in a genome and a high priority is to minimize the consequences of double-strand DNA breaks, this study confirms what the papers cited above previously reported—that BEs and PEs are a good choice.”

Finally, Liu doesn’t really support the claim that the editors in question are “state-of-the-art.” Notably, BE4max was first reported more than five years ago, in 2018, only two years after Liu’s group reported the very first base editor, and is actually multiple generations older than current base editors. This is important when evaluating the claim in the paper that BE4max impairs long-term engraftment of HSPCs. As for prime editors, they test PE3, which PE4-6 has supplanted.

In defense of the authors, some of the experiments lasted 15 weeks, and Nature Biotechnology received the paper in November 2022. Liu’s group’s paper on PE6 was published in Cell one week before the paper from Naldini’s group, on August 31, 2023. At some point, they very much could have been using the “state-of-the-art” editors, and whatever genotoxic concerns were detected in the research from Naldini’s lab should be applied to the true “state-of-the-art” editors if they have not been evaluated with the same or similar tests.

The need for testing

Professor at the UMass Chan Medical School Erik Sontheimer, PhD, who has published several articles on the optimization of BEs and PEs, lands somewhere in between Naldini and Liu. 

“It is a valuable study, not because the whole concept of genotoxic effects is new, which I think it’s not,” said Sontheimer, who is also on the editorial advisory board for GEN Biotechnology. “We have known about the potential for [genotoxic effects], but I think we don’t know everything about the degree to which we need to worry about it or not. So, I think figuring out how much of a problem it might actually be and how important it is to mitigate it and what would be the best way to mitigate it still has value, and this article would be a useful contribution in that regard.”

Even if it’s not news that these things can happen, Sontheimer, who is also a co-founder of Intellia, told GEN that there are still things to learn, especially in therapeutically meaningful environments like primary HSPCs, about how much of an issue it actually is, how important it is to address, and what the degree of the issue is that has to be addressed in the first place. The study has value for really looking at the extent to which these things are a problem in a somewhat deeper and more quantitative way than has been done before. While the concepts that are raised are already out there, there is still plenty to learn about genotoxicity in a clinically meaningful setting.

So, what does all of this mean?

For the time being, Naldini thinks this research is at best a benefit-risk assessment, saying that there is potential for genotoxic events. The point of this research is not to cast a negative light on base editing and prime editing and is by no means a call to stop using BEs and PEs, but to provide greater resolution into their genotoxicity.

“I don’t think this should be a call to stop any clinical development,” said Naldini. “It’s to raise attention to what could be some potential adverse effect that we have to know about. We can monitor in the long term and potentially also improve the editors themselves along the lines that we have now indicated.”

What he is most concerned about is that while some of these adverse effects are rare, others aren’t easily captured.

“To a certain extent, large deletions and things that happen with a [DNA] break may be there, and it may be difficult at the moment to really quantify,” said Naldini. “You can design [a test] in the laboratory pipeline for improving the tools, but whether you can really do a simple [clinical] test on a [genotoxic] product is not that easy at the moment.”

Currently, there aren’t any in vivo tests for translocations, let alone genome-wide mutational analysis—the cell always has to be killed. Without a reporter, which isn’t used in human clinical trials, the sample needs to be consumed or a biopsy taken to understand what some of those potential adverse effects might be. So, no matter what cell gets tested, even if it’s from a clonal population, it’s always a proxy that can be incongruent with the cell that’s going into the body. But what Naldini’s group did to get a readout on the mutagenic load was to generate hematopoietic progeny that were sampled without having to destroy the edited HSPCs themselves.

After all, to paraphrase something Naldini said, no tool is perfect.


There is data, although minimal, demonstrating the benefit of using BEs in the clinical setting.

Perhaps the most notable case is that of 13-year-old Alyssa from Leicester, who relapsed with T-cell acute lymphoblastic leukemia (T-ALL) after allogeneic stem-cell transplantation in 2021. She subsequently became the first patient to receive an infusion of base-edited CAR7 (BE-CAR7), which a group of researchers at UCL under the direction of Professor Waseem Qasim had created and developed. In just 28 days, Alyssa had molecular remission and then received a reduced-intensity (nonmyeloablative) allogeneic stem-cell transplant from her original donor, with successful immunologic reconstitution and ongoing leukemic remission. Alyssa’s bone marrow was found to be in morphologic remission with no minimal residual disease, and she was discharged 52 days after stem-cell transplantation.

Two other patients have since been treated with BE-CAR7 cells from the same bank, which showed potent activity. Although fatal fungal complications developed in one patient, the other patient underwent allogeneic stem-cell transplantation while in remission. Serious adverse events included cytokine release syndrome, multilineage cytopenia, and opportunistic infections.

Others have since followed. Beam Therapeutics, which Liu co-founded with Feng Zhang, PhD, and J. Keith Joung, PhD, announced on September 5, 2023, that the first patient with relapsed, refractory T-ALL and T-cell lymphoblastic lymphoma (T-LL) will receive doses of BEAM-201, a quadruplex base-edited allogeneic CAR-T cell investigational therapy. In November 2022, Beam Therapeutics began enrolling participants in their U.S.-based Phase I/II trial for BEAM-101, which uses base editing in autologous HSPCs to target sickle cell disease and beta-thalassemia. They’ve said that they may have data as early as 2024.

And BEs have been applied to other cells besides HSPCs, notably in VERVE-101, a first-in-class in vivo liver base editing program. On November 7, 2022, Verve Therapeutics put out a press release saying that the FDA had placed a hold on the investigational new drug (IND) application to conduct a clinical trial evaluating VERVE-101 in patients with heterozygous familial hypercholesterolemia (HeFH). In the same press release, Verve Therapeutics announced that it had recently completed a first round of dosing in three patients in New Zealand and the U.K. and reported no drug-related adverse effects. An independent Data Safety Monitoring Board reviewed safety data from the first cohort and recommended that the trial proceed to a second, higher dosage.

PEs, however, have yet to make it into the clinic. Liu’s Prime Medicine, which raised about $315 million in Series A/B and $200 million in an IPO in October 2022, announced at JPM 2023 that it had multiple programs advancing toward development candidates, with its first IND filing potentially as early as 2024.

According to EvaluatePharma, up to 21 cell therapies and 31 gene therapies are expected to be launched in 2024. So, over the next couple of years, we’ll see how many healthy chickens hatch.

Previous articleLights, Camera: Actio Biosciences Launches to Bridge Common and Rare Disease Treatments
Next articleHow Biomanufacturing Can Transform and Reshape the Construction Industry