In Vivo CRISPR Explained: Editing Genes Inside the Body

You’ve read about CRISPR gene editing. Casgevy, the sickle cell cure, made headlines. Vertex paid $8 billion for it. But here’s what nobody explains: Casgevy is ex vivo editing. They extract your stem cells, ship them to a lab, edit them over weeks, then reinfuse them. It costs over $2 million per patient and has treated fewer than 100 patients in 18 months. That’s the bottleneck.

Now a second approach is emerging: in vivo CRISPR. Instead of removing cells, you deliver the editing machinery directly into the body with an IV infusion. The edit happens inside the patient. If this works at scale, it changes the economics of gene therapy from bespoke manufacturing to something closer to a drug.

The problem

Ex vivo editing works, but it’s expensive and slow. Each patient’s cells are their own batch. You need apheresis centers, clean rooms, trained technicians, and weeks of processing. The capacity doesn’t scale. Casgevy, the approved Casgevy ex vivo sickle cell CRISPR therapy, has demonstrated that the science is real but the delivery is the constraint. If gene editing is going to reach more than a few thousand patients, it needs a delivery method that doesn’t require a lab per patient.

The solution

In vivo CRISPR delivers the editing components directly to the target tissue using a carrier, typically a lipid nanoparticle (LNP) or adeno-associated virus (AAV) vector. The edit happens inside the body. No cell extraction, no lab processing, no reinfusion. One infusion, one lifetime edit.

Step 1: Understand the difference between ex vivo and in vivo

Ex vivo means “outside the body.” Cells are removed, genetically modified in a lab, and returned to the patient. Casgevy and CAR-T therapies like Yescarta and Carvykti are ex vivo. The editing is precise because you control the lab environment, but the manufacturing is autologous, meaning each patient is a separate batch. See the AAV vs lentivirus vectors explainer for how the viral vector side works.

In vivo means “inside the body.” The CRISPR components are packaged into a delivery vehicle, injected into the bloodstream, and the edit happens in the target cells while they’re still in the patient. No extraction, no lab, no reinfusion. The theoretical advantage is manufacturing at scale: one production run treats thousands of patients, not one.

Step 2: Learn the delivery vehicles

The delivery problem is the whole game. CRISPR components, a Cas9 nuclease and a guide RNA, need to reach the right cells without editing the wrong ones. Two vehicles carry them:

Lipid nanoparticles (LNPs) are tiny fat bubbles that protect the mRNA and guide RNA in the bloodstream and ferry them into cells. LNPs naturally accumulate in the liver, which makes them ideal for liver-targeted diseases. Intellia’s NTLA-2001, the first in vivo CRISPR therapy tested in humans, uses an LNP to deliver CRISPR-Cas9 to liver cells to knock out the TTR gene in transthyretin amyloidosis. In a Phase 1 trial, a single infusion produced a 93% mean reduction in serum TTR at the highest dose, and the reduction was sustained at 12 months. That’s a one-time edit replacing chronic treatment. The NEJM published the data.

AAV vectors are modified viruses that deliver DNA encoding the CRISPR components. Different AAV serotypes target different tissues: AAV9 crosses the blood-brain barrier for neurological diseases, AAV8 targets the liver, AAVrh74 targets muscle (used by Sarepta for Elevidys in Duchenne muscular dystrophy). AAV is better for tissues LNPs can’t reach, but it has a packaging size limit, and high doses carry immunological risk, as documented in the AAV gene therapy deaths covered in the CRISPR stocks analysis.

Step 3: Base editing and prime editing change the risk profile

The original CRISPR-Cas9 cuts both DNA strands, creating a break that the cell repairs imperfectly, usually destroying the target gene. That’s fine for knocking out a harmful gene but it’s blunt. Two newer approaches are quieter:

Base editors change a single DNA letter without cutting the double strand. A C becomes a T, or an A becomes a G. This is more precise and less likely to cause off-target damage, which matters when you’re editing inside a living person and can’t undo it. Verve Therapeutics is using adenine base editing in VERVE-102 to inactivate the PCSK9 gene in the liver for cardiovascular disease. In the Heart-2 Phase 1b trial, a single infusion produced dose-dependent LDL-C reductions up to 59% at the highest dose, with no serious adverse events. The implication is a one-time shot that could replace daily statins for a lifetime.

Prime editors write new genetic sequences without cutting. They can insert or delete short stretches of DNA. This is the most precise editing tool, and it’s also the newest, with LNP delivery still being optimized.

For investors, the key point is that base and prime editing reduce the permanent-change risk that makes in vivo editing frightening. You’re making smaller, more controlled edits inside someone’s body.

Step 4: Know the companies

The in vivo CRISPR field is concentrated in a handful of companies, each with a different delivery and editing strategy:

• Intellia (NTLA): Lead in LNP-delivered Cas9 editing for liver diseases. NTLA-2001 is in Phase 3 (MAGNITUDE trial) for ATTR amyloidosis. NTLA-2002 targets angioedema. The Regeneron partnership funds development.
• Verve Therapeutics (VERV): Base editing for cardiovascular genes. VERVE-102 (PCSK9) in Phase 1b/2. Eli Lilly has an opt-in deal. If this works, the addressable market is cardiovascular disease, not rare disease. That’s millions of patients, not thousands.
• Beam Therapeutics (BEAM): Base editing platform across multiple targets. Partnership with Pfizer in sickle cell and rare diseases.
• Editas (EDIT): Pivoting from ex vivo to in vivo, focusing on eye and liver diseases.
• Prime Medicine (PRME): Prime editing platform, earliest stage but the most precise in theory.

Step 5: Understand the safety challenge

In vivo editing is permanent. If the edit goes wrong, you cannot extract the cells and redo it. The main risks:

• Off-target edits: The CRISPR machinery edits an unintended location in the genome. Preclinical screening catches most of these, but in vivo environments are harder to predict than cell cultures.
• Immune reactions: Cas9 originates from bacteria. Some patients have pre-existing immunity to Cas9 proteins, which can cause immune responses. AAV vectors also trigger immune responses, and you generally can’t redose with the same serotype.
• Permanent changes: Unlike a drug you can stop taking, a gene edit is for life. If the edit works, it’s a cure. If it doesn’t, there’s no undo.

The FDA has shown it takes these risks seriously. Sarepta’s Elevidys faced a clinical hold in 2025 after patient deaths from acute liver failure at high AAV doses. That’s a reminder that delivery safety, not editing precision, is the bottleneck.

Step 6: Frame the investment thesis

The reason in vivo CRISPR matters for investors is addressable market. Ex vivo editing is limited to rare blood disorders: sickle cell, beta-thalassemia, a handful of cancers. Each affects thousands of patients. The total market is capped by manufacturing capacity.

In vivo editing, if the delivery works, expands the addressable market from rare disease to common disease. Verve’s PCSK9 program targets cardiovascular disease, the leading cause of death globally. Intellia’s ATTR program targets a disease with 50,000+ US patients. Gene therapy catalysts through 2027 will start answering whether in vivo delivery can scale.

Common mistakes

• Confusing ex vivo and in vivo. They’re different technologies with different economics. A Casgevy competitor using ex vivo editing faces the same manufacturing bottleneck. An in vivo competitor doesn’t. Read the pipeline before assuming “CRISPR stock” means the same thing.
• Assuming in vivo is safer. Delivery risk is higher, not lower. The deaths at high AAV doses are real.
• Extrapolating Phase 1 data. Intellia’s 93% TTR reduction was in 6 patients. Verve’s LDL-C reduction was in 14. These are promising, not conclusive. Phase 2 to Phase 3 transition rates are roughly 30-45% across therapeutic areas. A positive Phase 1 is a hypothesis, not an approval signal.
• Ignoring cash runway. These are early-stage companies with no revenue. Verve’s cash funds operations into mid-2027. If the data slips, the dilution risk is real.

Final checklist

• Ex vivo (extract, edit in lab, reinfuse) vs. in vivo (edit inside the body with an injection)
• LNP delivery targets the liver. AAV delivery targets muscle, eye, brain, and other tissues.
• Base editing and prime editing are more precise than the original Cas9 scissors.
• Intellia (liver, Cas9), Verve (cardiovascular, base editing), Beam (platform, base editing), Editas (pivoting in vivo), Prime Medicine (prime editing)
• In vivo editing is permanent. Off-target edits and immune reactions are the core risks.
• The investment thesis is addressable market expansion from rare disease to common disease.

For the broader CRISPR stock field, see the CRISPR stocks analysis. For how ex vivo editing works in practice, read the Casgevy 18-month follow-up.