Does prime editing have enough human evidence to justify the hype?

How well does prime editing actually work in human cells?

In human cells grown in the lab, prime editing can be remarkably efficient. One study using optimized all-in-one constructs achieved editing rates up to 95% in HEK293T cells, with an average of 67% across multiple sites [5]. That means for many targets, nearly every cell can be edited. Another study in human organoids—miniature organs grown from patient stem cells—corrected common cancer mutations with efficiencies up to 25% in colon organoids and up to 97% in liver organoids, and also repaired the cystic fibrosis mutation CFTR-F508del [4]. These are not just random edits: they are disease-relevant corrections in human tissue models.

However, efficiency varies hugely depending on the cell type and the specific edit. The same study that hit 95% in HEK293T cells found that efficiency in K562 and HeLa cells remained low [5]. A separate analysis of over 92,000 guide RNAs showed that editing rates can range from near zero to over 70%, and a deep-learning model (PRIDICT) was needed to predict which guides would work well [6]. So while prime editing can be extremely efficient, it is not yet plug-and-play for every human cell type or every genetic change.

What are the main barriers to using prime editing in patients?

The biggest barrier is delivery. Prime editors are large proteins, and getting them into the right cells in a living body is hard. The most successful in vivo study to date used a high dose of adenoviral vector (7 × 10^14 viral particles per kilogram) to deliver prime editors to mouse livers, achieving only about 11% correction of a phenylketonuria mutation—enough to lower blood toxin levels but not a cure [2]. That same study also noted immune responses to the viral vector and the editor protein, which would be a problem for repeated dosing in humans.

A second barrier is the DNA repair system itself. Multiple studies show that the cell's mismatch repair (MMR) machinery actively blocks prime editing, reducing efficiency by 2- to 17-fold depending on the cell line and edit type [3][9]. One clever workaround is to transiently inhibit MMR, which boosted editing efficiency by an average of 7.7-fold in some cell types [3]. Another approach is to engineer the pegRNA (the guide RNA that carries the edit) to be more stable—adding structured RNA tails improved efficiency 3- to 4-fold in several human cell lines [8]. These are real solutions, but they add complexity to an already intricate system.

Does prime editing cause unwanted side effects?

Prime editing was designed to avoid the double-strand DNA breaks that cause large unintended mutations in older CRISPR methods, and the evidence largely supports that promise. Whole-genome sequencing of prime-edited human organoids found no detectable off-target edits [4]. Another study compared prime editing to standard CRISPR and found that unintended indels (small insertions or deletions) were only slightly more frequent with prime editing, and off-target editing at predicted sites was comparable between the two systems [1].

But prime editing is not error-free. Some studies report unintended insertions, deletions, and even incorporation of parts of the pegRNA scaffold into the genome [7]. The mismatch repair system, when active, can also create unwanted byproducts [3]. The rate of these errors varies by target and cell type, and the field is actively working to minimize them. For now, the safety profile looks good compared to older methods, but it is not perfect.

About These Sources

This answer is built on 9 peer-reviewed studies — published from 2021 to 2023, 8 in Q1 journals, collectively cited 2,395 times — selected as the most relevant from 15 studies that passed quality screening, drawn from 57 papers retrieved from a database of over 500 million.

Sources used in this answer

Related research questions