Base Editing

Base editing chemically converts a single DNA base into another — A→G (adenine base editor, ABE) or C→T (cytosine base editor, CBE) — without creating a double-strand break. Invented in David Liu's lab (the same lineage as prime editing), it fuses a catalytically impaired Cas to a deaminase enzyme, so the editor "rewrites" one letter at a precise position rather than cutting and relying on the cell's repair machinery. Because there is no double-strand break, base editing avoids the indels and chromosomal rearrangements that cut-based CRISPR-Cas9 can produce — the reason it is widely framed as a "safer CRISPR." With two base editors you can make four of the possible single-base conversions, which covers a large fraction of known pathogenic point mutations.

By 2026 the modality has broken through in the clinic across multiple indications, almost all using lipid nanoparticle (LNP) delivery to the liver for in vivo correction (Beam, Verve) plus ex vivo programs (sickle cell). The contrast with Intellia's in vivo knockout approach is instructive: base editing corrects a disease-causing letter, whereas a Cas9 knockout simply inactivates a gene. Both now have human data.

Key Claims

• First-ever genetic correction in AATD patients — Beam BEAM-302 (in vivo ABE, LNP-to-liver) corrects the PiZ (Z-AAT) mutation. Single IV doses (15–60 mg) gave dose-dependent, durable increases in total AAT and decreases in Z-AAT; the highest-dose cohort reached AAT levels sufficient to halt disease progression. Evidence: moderate (Beam clinical 2026)
• >90% reduction in disease protein, sustained ≥36 weeks (sickle cell) — Beam BEAM-101 (ex vivo) eliminates the need for busulfan conditioning; a treatment-related death has been reported and is under monitoring. Evidence: moderate (Beam clinical 2026)
• FDA clears BEAM-304 IND for PKU (Jun 2026) — In vivo ABE correcting PAH mutations (first the R408W variant); preclinical mouse data showed robust liver editing and normalized plasma phenylalanine. Evidence: strong (regulatory action) (Beam clinical 2026)
• Up to 69% LDL reduction (cardiovascular) — Verve VERVE-102 (PCSK9 base editing) Phase 1b, dose-dependent, no serious adverse events. Evidence: moderate (Beam clinical 2026)
• No double-strand break = safer profile — When VERVE-101 showed thrombocytopenia, it was attributed to the LNP vehicle, not the editing machinery. Evidence: moderate (Beam clinical 2026)
• Base editing scales to functional genomics — Multiplexed in vivo base editing maps gene–variant–context interactions at organism scale, bridging in vitro variant screens to whole-organism biology. Evidence: strong (preprint) (Multiplexed base editing)

Benchmarks & Data

Open Questions

• Durability of in vivo base edits over years, and whether liver turnover erodes the correction.
• Long-term off-target and bystander-editing safety as doses escalate.
• Whether the sickle-cell safety signal is editor-related or conditioning/disease-related.
• Pricing and reimbursement for one-time corrective therapies.

Related Concepts

• Prime Editing — The search-and-replace sibling (same Liu lineage); handles insertions/deletions base editing cannot.
• CRISPR Clinical Translation — Where base editing sits in the broader clinical pipeline, alongside Intellia's in vivo knockout.
• Gene Therapy Delivery — LNP-to-liver is the dominant in vivo delivery route for these programs.

Backlinks

Pages that reference this concept:

• Beam Base Editing Clinical 2026
• Multiplexed In Vivo Base Editing
• Beam Therapeutics

Changelog

• 2026-06-24 — Initial compilation from the Beam 2026 clinical pipeline (BEAM-302/304/101) plus Verve cardiovascular programs and the multiplexed in vivo base-editing preprint. Created to separate base editing from prime editing as a distinct, now clinically-validated modality.