A Single Injection Rewrote a Genetic Typo in 26 Patients. 94% of Their Protein Came Back Normal.

Ninety-four percent. That is the fraction of alpha-1 antitrypsin protein circulating in patients' blood that was the corrected, functional M-type form after a single intravenous infusion of BEAM-302, a base editing therapy developed by Beam Therapeutics. Data reported on March 25, 2026 from an ongoing Phase 1/2 trial, represent the first time a gene editing therapy has directly corrected a disease-causing point mutation inside the human body and restored near-normal protein production.

That distinction matters. Every in vivo CRISPR therapy that has reached human trials before this one worked by breaking genes. Intellia's NTLA-2001 for hereditary transthyretin amyloidosis uses Cas9 to cut the TTR gene and stop toxic protein production. Verve's VERVE-102 for cardiovascular disease uses base editing to disable PCSK9, reducing LDL cholesterol by 53% on average. Both are gene disruption: identify a harmful protein, break the gene that makes it, let the body clean up. It works. It is also the molecular equivalent of unplugging a faulty appliance.

BEAM-302 does something harder. It changes a single adenine to a guanine in the SERPINA1 gene, converting the disease-causing PiZ variant back to the functional PiM form. Not a cut. Not a deletion. A correction. Instead, it gets rewired.

The Numbers

Alpha-1 antitrypsin deficiency affects an estimated 80,000 to 100,000 people in the United States and more than 3 million globally with allele combinations associated with severe deficiency. In AATD, the PiZ mutation causes the liver to produce a misfolded version of AAT protein that both accumulates in hepatocytes (damaging the liver) and fails to reach the lungs (where functional AAT protects tissue from neutrophil elastase). Patients develop emphysema, cirrhosis, or both. Current standard of care is augmentation therapy: weekly intravenous infusions of AAT protein purified from donor blood plasma, at a cost of $127,537 per year.

Augmentation addresses the lung component only. It does not stop Z-AAT from accumulating in the liver. It does not fix the mutation. It is a protein patch applied indefinitely to a single-nucleotide problem.

Beam's Phase 1/2 trial enrolled 29 patients across multiple dose cohorts, with 26 receiving treatment as of the February 2026 data cutoff. At the selected 60 mg dose, the results were:

Metric	Before Treatment	After Treatment (60 mg)
Corrected M-AAT (% of total)	0%	~94%
Mutant Z-AAT reduction	Baseline	-84%
Total AAT level	Below 11 µM	16.1 µM (mean steady-state)
Patients above 11 µM threshold	0%	100% (sustained 12 months)
Serious adverse events	N/A	0

The 11 µM threshold is the established clinical benchmark. Below it, lungs lose protection. Above it, emphysema progression slows dramatically. Every patient in the 60 mg cohort cleared it and stayed there for the full observation period.

The Datapoint That Matters Most

One patient in the trial developed a respiratory infection during follow-up. Infections are routine clinical events. What happened next was not.

AAT levels surged from 15.9 µM to 29.5 µM. Corrected M-AAT composition held at 95%. Edited hepatocytes had responded to the body's inflammatory signals, specifically IL-6 and other acute-phase cytokines, exactly as a wild-type SERPINA1 gene would. AAT is an acute-phase reactant; healthy people produce more of it when fighting infection. This patient's liver, with its corrected gene, did the same thing.

That single observation tells you more about the quality of the edit than any protein assay can. Beam's base editor did not insert a transgene with a synthetic promoter running at a fixed rate. It corrected a single nucleotide in the endogenous gene, in its native chromatin context, with all regulatory elements intact. Enhancers, promoters, response elements for inflammatory signaling: everything upstream and downstream of the mutation site remained functional. It did not just produce the right protein. It produced the right amount of the right protein at the right time in response to the right stimulus.

That is what "cure" looks like at the molecular level.

Disruption vs. Correction: A Taxonomy of Gene Editing in the Liver

In vivo gene editing in the clinic, as of 2026, falls into two categories. That distinction has practical consequences that extend well beyond AATD.

Therapy	Target	Mechanism	Editing Type	Key Result
Intellia NTLA-2001	TTR (hATTR)	Cas9 double-strand break	Gene disruption	90% protein reduction, 3yr sustained
Verve VERVE-102	PCSK9 (cardiovascular)	Base editing (A-to-G)	Gene disruption	53% LDL-C reduction (mean)
AccurEdit	PCSK9 (cardiovascular)	Base editing	Gene disruption	~50% LDL-C reduction
Beam BEAM-302	SERPINA1 (AATD)	Base editing (A-to-G)	Gene correction	94% functional protein restored

Gene disruption works when the disease is caused by a protein that should not be there. Transthyretin amyloidosis: the liver makes a misfolded TTR protein that deposits in nerves and heart tissue. Break the gene, stop the protein, stop the disease. PCSK9 regulates LDL receptors; knock it out, LDL cholesterol drops. Both are gain-of-toxic-function problems where the solution is subtraction.

AATD is different. Patients need functional AAT protein. They just have the wrong version. Disrupting the SERPINA1 gene would make the disease worse. You cannot subtract your way to a cure when the problem is a missing function. You have to add the right nucleotide back.

Base editing makes this possible without the double-strand DNA breaks that conventional CRISPR-Cas9 requires. The BEAM-302 editor chemically converts adenine to guanine at the PiZ mutation site, changing codon 342 from glutamic acid (the disease variant) to lysine (the wild-type). No cut. No insertion. No deletion. A molecular Find-and-Replace on a single letter in a 3.2-billion-letter genome.

A Cost Calculation Nobody Has Run

Augmentation therapy for AATD costs $127,537 per year in the United States. That figure comes from a 2023 analysis of insurance claims data published in the Journal of the COPD Foundation. Sixty-six percent of the cost is the augmentation product itself; the rest is physician visits, hospitalizations, and other medications.

AATD is typically diagnosed between ages 30 and 50. With augmentation, life expectancy extends into the 60s and 70s. Take a conservatively diagnosed 40-year-old patient who begins augmentation therapy and lives to 70:

$127,537 per year × 30 years = $3.83 million in lifetime direct medical costs.

For comparison, the most expensive approved gene therapies today range from $2.1 million (Zolgensma for spinal muscular atrophy) to $3.5 million (Hemgenix for hemophilia B). Casgevy costs $2.2 million. If BEAM-302 is priced in this range, the break-even against augmentation therapy occurs at 15.7 to 27.4 years, well within the treatment horizon for most patients.

But the cost calculation understates the case. Augmentation therapy only addresses lung disease. It does nothing for the hepatic accumulation of Z-AAT that causes cirrhosis in roughly 10% of AATD patients. BEAM-302 corrects the mutation, reducing Z-AAT by 84% and potentially addressing both organ systems with a single dose. A therapy that treats two disease manifestations for the price of one, delivered once instead of weekly for decades, is not merely cost-effective. It is a category shift.

The Safety Question

Across 26 patients receiving single doses of BEAM-302 up to 75 mg, no serious adverse events and no dose-limiting toxicities have been reported. Side effects consisted of mild to moderate infusion-related reactions and transient elevations in liver enzymes, both expected with lipid nanoparticle delivery to the liver.

Context is important here. Intellia's NTLA-2001, the most advanced in vivo CRISPR therapy, reported a severe adverse event in its Phase 3 trial in October 2025: a participant developed highly elevated liver enzymes. That trial had enrolled over 450 patients at that point. BEAM-302 has treated 26. Small cohorts rarely surface rare events. A clean safety profile is encouraging but not conclusive.

LNP delivery introduces its own considerations. Lipid nanoparticles trigger innate immune responses. Most in vivo gene editing trials see transient liver enzyme elevations. Repeat dosing may be complicated by anti-LNP antibodies, though BEAM-302 is designed as a single-dose therapy. Whether 94% correction persists as hepatocytes divide over years is an open question. Liver cells turn over slowly, with a half-life estimated at 200 to 300 days, but they do turn over. If edited cells are replaced by unedited progenitors, correction could fade.

What This Does Not Prove

Several boundaries constrain the interpretation. Twenty-six patients with a maximum of 12 months of follow-up is insufficient to declare AATD cured. The pivotal expansion cohort, targeting approximately 50 additional patients with 12 months of biomarker follow-up, will be the basis for an accelerated approval filing. That trial has not started.

No data on liver fibrosis progression has been reported. Reducing Z-AAT by 84% is biologically significant, but whether it halts or reverses hepatic damage requires longer follow-up and liver biopsy data that do not yet exist. The 11 µM AAT threshold is a surrogate marker validated for lung outcomes. Its predictive value for liver disease is less established.

The trial only enrolled patients with the PiZ mutation, which causes approximately 95% of severe AATD cases. Rare variants are not addressed. Pricing remains undisclosed. If BEAM-302 is priced above $3 million, the cost-effectiveness calculation narrows. If it requires repeat dosing, it collapses.

The Strongest Case Against This Being a Breakthrough

The most rigorous criticism comes from the gene therapy field itself: a 26-patient Phase 1/2 trial with 12 months of biomarker data has, historically, been a poor predictor of commercial success. Bluebird Bio's betibeglogene autotemcel (Zynteglo) showed striking Phase 1/2 results for beta-thalassemia, received conditional approval in Europe in 2019, and was commercially withdrawn by 2021 due to manufacturing costs and a leukemia signal. Solid Biosciences' Duchenne muscular dystrophy gene therapy produced striking biomarker improvements in early trials before serious cardiac adverse events emerged at larger scale.

Biomarker endpoints create a specific vulnerability. The FDA's accelerated approval pathway for BEAM-302 uses AAT blood levels as a surrogate. Surrogate endpoints have a mixed track record. The question is not whether BEAM-302 raises AAT levels. It clearly does. The question is whether 16.1 µM of AAT, 94% of which is corrected, produced by 60 mg of base editing machinery delivered once, translates to decades of protection against emphysema and cirrhosis. That question requires Phase 3 data and years of follow-up that do not exist yet.

The Bottom Line

For the first time, a gene editing therapy has corrected a disease-causing mutation inside living patients and restored near-normal protein function. BEAM-302 did not shut off a gene. It fixed a single DNA letter. The edited gene retained its regulatory architecture and responded to physiological signals the way a healthy gene would. If the pivotal trial confirms what 26 patients and 12 months of data suggest, AATD patients face a genuine choice: weekly infusions at $127,537 per year for decades, or a single IV dose that rewrites the underlying error. More broadly, BEAM-302 is the proof of concept for a therapeutic category that goes beyond gene disruption. There are roughly 7,000 known single-gene diseases, the majority caused by point mutations. If you can correct one nucleotide in one gene in one organ with one dose, the template applies to hundreds of them. Editing precision is here. Clinical validation is starting. The question, as always, is whether the system built to deliver it can keep up.