top of page
SpecialeInvest_WhiteLogo.png

Democratizing Gene Therapy

  • Writer: Ahammad Shibil
    Ahammad Shibil
  • 5 days ago
  • 7 min read

Before we get into the biology, look at what pharma spent money on in the last twelve months.

Acquirer

Target

Deal Value

What they were buying

Bristol Myers Squibb

RayzeBio

$4.1B

Targeted radionuclide delivery in vivo — payload finds tumour, no cell extraction

Novo Nordisk

Cardior Pharmaceuticals

€1.03B

RNA therapeutics platform — non-viral in vivo cardiac delivery

Sanofi

Inhibrx AATD Program

$2.2B

In vivo protein replacement — same indication now drawing three gene editors

AstraZeneca

Amolyt Pharma

$1.05B

Peptide-based in vivo endocrine targeting — expanding delivery repertoire

Roche

Carmot Therapeutics

$2.7B

Receptor-targeted systemic peptide delivery — non-surgical, whole-body payload

J&J · Pfizer

Multiple pipeline moves

Undisclosed

Non-viral delivery infrastructure — the capability gap is visible from the outside


The first generation: real cures, impossible scale

The original cell and gene therapy medicines were genuine scientific achievements. CAR-T — chimeric antigen receptor T cell therapy — was the landmark. Extract a patient's T cells. Engineer them outside the body to express a synthetic receptor that recognises a tumour antigen. Expand them to therapeutic numbers. Reinfuse. The engineered T cells find and destroy cancer cells with a specificity that no small molecule approaches.


The results in specific blood cancers were extraordinary. Novartis' Kymriah showed 81% complete remission in paediatric relapsed/refractory B-cell ALL — a patient population where standard treatment had essentially nothing to offer. Gilead's Yescarta hit 83% overall response in relapsed large B cell lymphoma. For the right patient, at the right time, first-generation CAR-T is curative.


~25,000 patients treated with approved CAR-T products globally since 2017. Six FDA-approved products. Two-year survival rates in previously incurable haematological cancers now routinely exceeding 40–50%. Beyond CAR-T: lentiviral gene transfer to haematopoietic stem cells — bluebird bio's betibeglogene autotemcel demonstrated that monogenic inherited diseases could be corrected at the DNA level. The biology worked but scale did not.


The cost floor — before any institutional margin — is roughly $100,000–$150,000 in manufacturing alone. List prices in the US sit at $400,000–$500,000 per treatment. Each patient is a bespoke manufacturing run. Because of India's position of operating at cheaper cost, we are able to do it at 10x cheaper, but it's still capped by manufacturing and capability.


The second generation: edit in place

In vivo (inside the body) genetic medicine collapses the manufacturing architecture entirely. Deliver the editing machinery directly into the body. Let it find the target tissue, make the change, and get cleared. The drug product becomes a formulated nanoparticle suspension — manufactured in bulk, quality-characterised, stored, and shipped like a standard biologic.


Three technologies had to mature before this was possible, and they did so in sequence. Lipid nanoparticle chemistry spent decades improving RNA delivery and then got tested at scale in COVID mRNA vaccines. Base editing and prime editing then emerged as precision genome modification tools that operate without double-strand DNA breaks, removing the genotoxicity ceiling that made nuclease-based editing clinically risky. Finally, GalNAc conjugation and ionisable lipid engineering began extending LNP tissue tropism beyond the liver, opening targets previously unreachable by non-viral means.


The acquisitions happening now are large pharma buying the third of those three capabilities — the delivery engineering they don't have internally. The biology was already solved. The delivery is the gap.


Three programmes that showed it works in humans

The clinical proof arrived across three programmes in the last eighteen months. All three target the liver — not because liver disease is the only opportunity, but because current LNP chemistry naturally routes there. That's a starting point, not a ceiling.


Beam Therapeutics' BEAM-302 targets alpha-1 antitrypsin deficiency, a liver disease caused by a single misplaced letter in the genome. BEAM-302 delivers a base editor via LNPs that corrects that letter directly in hepatocytes — no cutting, no gene addition. Phase 1/2 data from March 2025: 91% corrected protein at Day 28 in the 60mg cohort, 79% reduction in the toxic mutant form. Side effects were minor and self-resolving.


Verve Therapeutics' VERVE-102: A single infusion reduced LDL cholesterol by a mean of 53% — up to 69% in some patients — with zero serious adverse events across 14 participants. If that holds over time, it's a permanent replacement for a daily statin.


Intellia Therapeutics' nex-z showed 87% knockdown of a disease-causing protein, sustained at three years after one dose. Intellia's approach uses a DNA-cutting mechanism that base editors avoid — and that distinction in safety profile is now shaping every programme that comes after it.

Three programmes. Two that are working cleanly, one that is working but with a safety flag that matters. Enough signal to say in vivo editing in humans is real.


The real bottlenecks

Two constraints define where the field currently is and where it is not.

The delivery problem: Getting out of the liver requires specificity engineered directly into the particle. A comprehensive 2025 review in Small Methods (Wu et al.) maps these non-liver targeting strategies systematically: polymer-lipid hybrids, organ-tropic LNP libraries, antibody-conjugated particles for muscle, lung, and CNS access. The approaches are technically credible. Clinical translation outside the liver is still preclinical for most programmes. Lung and muscle are probably two to three years from first-in-human. CNS is further.


The editing tool tradeoff: CRISPR-Cas9 makes double-strand breaks — efficient, broad, and carries genotoxicity risk as Intellia's data made concrete. Base editors change single nucleotides without cutting — cleaner safety profile, limited to transition mutations. Prime editors can write short sequences without breaks — most versatile, but large cargo size limits efficient LNP packaging. The field is not converging on one tool. A toolkit is being assembled, matched to mutation type and tissue target. Platform companies owning multiple editing modalities across this matrix have a durable advantage.


Flagship's stack — Tessera + Generate + Cellarity: Tessera Therapeutics built Gene Writing — writing therapeutic sequences into the genome rather than cutting, with a proprietary LNP system for in vivo delivery. Bill & Melinda Gates Foundation: $50M in December 2024 for in vivo sickle cell. Regeneron partnership December 2025 for TSRA-196 targeting AATD — the same indication Beam is pursuing with base editing and Intellia with Cas9. Three companies, three editing modalities, same disease. Within 24–36 months, head-to-head clinical data comparing fundamentally different genome editing approaches in the same indication — a technology tournament that hasn't happened before in genetic medicine. Flagship positioned Tessera for exactly this readout — paired with Generate Biomedicines at the protein design layer and Cellarity targeting cell-behaviour network states rather than individual mutations. The in vivo turn is a stack, not a single company.


In vivo CAR-T and the oncolytic parallel

Two adjacent bets belong in the same frame as LNP-mediated genome editing.

In vivo CAR-T — engineering T cells inside the patient rather than in a factory — is the most structurally ambitious version of the in vivo turn. Targeted LNPs with antibody conjugates (anti-CD3, anti-CD8) deliver CAR-encoding mRNA to T cell subsets in circulation, reprogramming them without ever leaving the body.


Oncolytic viruses represent a complementary logic. Instead of editing the patient's genome, engineer a virus to selectively replicate in and destroy tumour cells — a biological payload that amplifies itself at the target site. For hard-to-reach solid tumours where LNP delivery remains years away and where the immunosuppressive microenvironment limits both CAR-T and checkpoint inhibitors, oncolytic approaches offer a different route to the same underlying principle: a therapeutic that finds its target inside the body and acts there. The combination potential with immune checkpoint blockade has barely been explored clinically.


The structural shift — and what it means from India

The unit economics of first-generation cell therapy reflect the manufacturing architecture directly. Each patient is a run. The cost floor — leukapheresis, viral vector production, quality testing, logistics — is roughly $100,000–$150,000 before institutional margin. There is no path to manufacturing a bespoke autologous cell therapy for $10,000, regardless of how efficiently the process runs.


In vivo editing changes the manufacturing class of the product. An LNP formulation is a drug product. Per-patient manufacturing cost at scale looks more like a monoclonal antibody than a cell therapy: thousands of dollars rather than hundreds of thousands. Treatment happens in any hospital with an infusion chair and cold storage, not a specialist centre with a GMP clean room.


From India, the reimbursement debate that dominates Western discussions is the wrong frame entirely. The barrier is access, geography, and a price point that is structurally inaccessible. India demonstrated the proof of concept when ImmunoACT's NexCAR19 received CDSCO approval in October 2023 at ₹30–40 lakh (~$36,000–48,000), roughly a tenth of the US list price, by manufacturing locally through IIT Bombay and Tata Memorial Centre. That delta exists because the team built the infrastructure domestically — not because the therapy was fundamentally cheaper to make.


The world is going to push the cost of in vivo genetic medicine down independent of what US payers decide. LNP manufacturing is chemistry, not cell biology — it scales with standard pharma infrastructure and will follow the same commoditisation curve as monoclonal antibody biosimilars once the IP window closes. A therapy costing $5,000 to manufacture and administrable in a district hospital is a categorically different product from one costing $150,000 requiring a specialist centre. NexCAR19 at $40,000 proved the access equation bends when manufacturing moves onshore. In vivo at $5,000 per treatment is the eventual trajectory.


Where this ends up

The field's next three to five years will look like a tournament, not a progression.


AATD is already the benchmark. Beam, Tessera, and Intellia converging on the same hepatocyte target with three different editing modalities will produce head-to-head clinical data in the same disease. That readout will answer questions that no single-arm trial can: which editing mechanism delivers the best correction-to-safety ratio, which delivery architecture holds durability longest, which manufacturing platform transitions most cleanly to global scale. The Wu et al. 2025 roadmap in Small Methods is the honest map of where the delivery engineering sits for non-liver targets — credible strategies, mostly preclinical, and a two-to-five-year translation window representing the next investment cycle.


The body has always been treated as the problem — the environment drugs navigate around, the immune system that has to be suppressed, the physiology that constrains delivery. In vivo genetic medicine inverts that. The body becomes the manufacturing facility. The patient's cells execute the edit.


A therapy that runs on the patient's own cellular machinery is a genuinely different class of medicine. Every major pharma company buying in vivo delivery capability right now understands what that implies. The acquisitions are not optimism. They are recognition.


Sources & References

bottom of page