CRISPR

Summary

CRISPR enters this reference as a counterpoint, not as a protocol: it is the only place where the promise of genetic engineering became an agency-approved drug, and seeing the price of that in evidence calibrates the reading of everything else on the site. The tool is 14 years old and has one approved product, Casgevy, for sickle cell disease and beta-thalassemia. In the two phase 3 trials, editing worked in 100% of patients; 97% became free of vaso-occlusive crises and 91% became transfusion-independent. What limits the therapy is the myeloablative conditioning with busulfan that precedes the infusion — and the hardest number in the survey is about fertility: of the 17 women with more than one year of follow-up, 17 had ovarian failure. In the United Kingdom, NICE counts 1,794 eligible and projects 23 treatments in the first year. In Brazil there is no registration, no registration petition and not even a clinical trial: the sweep of Vertex's 541 petitions returned zero for advanced therapy.

Why CRISPR is in this reference

Everything else on this site describes compounds that circulate without approval, with community-sourced doses and vials of uncertain provenance. CRISPR is the opposite of that, and that is why the page is worth it: it is the only place where the promise of rewriting the genome became a drug approved by a regulatory agency, with a phase 3 trial, a package insert and a price.

What matters here is not the molecule — there is no dose, no reconstitution, no cycle. What matters is how much it cost, in evidence, to go from promise to patient. That number is the best yardstick available for reading the rest of this reference.

The short answer: fourteen years, 64,641 articles, one approved product — and a limit that is not the technology's, but that of the chemotherapy that has to come before it.

What it is, in verifiable terms

CRISPR-Cas is an adaptive immunity system of bacteria and archaea. It works in three steps: the bacterium stores a piece of the invading virus's DNA as a “spacer” in its own genome; transcribes that archive into guide RNAs; and uses a Cas nuclease directed by that RNA to cut the invader when it returns.

The finding that became a tool: in Streptococcus pyogenes, two paired RNAs direct Cas9 to cut both strands of the target DNA — the HNH domain cuts the complementary strand, the RuvC-like domain cuts the other. And the pair works when fused into a single chimeric RNA molecule. That fusion is what made the thing programmable: changing the target became changing an RNA sequence, not designing a new protein.

The part that is almost never told: the scissors are precise, the repair is not. The cell itself repairs the break, and the most common repair mode generates small insertions and deletions that switch the gene off. Writing a new sequence in its place is far less efficient. A good share of the safety problems further down this page comes from that second half, not the first.

Verified chronology

Each row below was checked in PubMed by the article identifier. Two things the popular version of the story usually erases appear here: 25 years passed between seeing and understanding — the repeats had been published since 1987 and nobody knew what they were — and the demonstration that it was an immune system came out of a yogurt manufacturer, solving an industrial problem of starter cultures killed by bacteriophage.

The real size of the field

PubMed search on September 4, 2026, with the term in the title or abstract. The last number in the table is what matters for calibrating any headline about CRISPR.

Reading these numbers

• Of 64,641 articles, 67 are clinical trials. That is 0.10%. The field is overwhelmingly preclinical — which is not a criticism, it is the age of the technology. But it is the correct denominator.
• The growth is real and recent. From 139 articles in 2012 to 7,738 in the current year alone. Publication volume, however, is not clinical evidence volume: the two curves do not move together.
• The next two generations already have their own literature — 2,251 base editing articles and 1,080 prime editing articles. They are not cosmetic refinement: they exist as a direct response to a safety problem, which is further down this page.

The only approved product

Only one made it here: Casgevy (exagamglogene autotemcel). It is a non-viral autologous cell therapy. The patient's own blood stem cells are harvested; the region that controls the BCL11A gene is edited outside the body with CRISPR-Cas9; and they are reinfused after myeloablative chemotherapy with busulfan.

The logic is indirect and elegant: BCL11A is what switches off fetal hemoglobin after birth. Switch off BCL11A and fetal hemoglobin comes back — and compensates for the defective adult hemoglobin. The diseased gene is not fixed; it is bypassed by switching a backup gene back on.

The efficacy numbers, and what they do not say

The two phase 3 trials are single-arm and open-label — there is no comparison group. That limits the reading and needs to be said before the numbers, not after.

The bottleneck is not CRISPR — it is busulfan

This is the most important part of the page, and the one that appears least in the headlines.

The edit worked in 100% of patients. In both trials, all engrafted. What limits the therapy's reach is not the editing step: it is the autologous transplant around it. The most frequent adverse events the FDA lists are mucositis and febrile neutropenia — chemotherapy, not genome editing. The labeling also carries a warning for neutrophil engraftment failure and for off-target editing.

Whoever receives Casgevy goes through a complete autologous transplant, with all the morbidity of one. That is what separates “functional cure” from “cure”.

And there is a cost that appears in no efficacy table. Busulfan is gonadotoxic — that is not a rare adverse effect, it is the drug working as expected. In a single-center experience with 40 patients undergoing gene therapy:

Seventeen of seventeen

• That number needs to be said with the same prominence as the 97%. A young patient who accepts Casgevy is, on the evidence published so far, trading vaso-occlusive crises for ovarian failure.
• Unless she first goes through fertility preservation — which is one more procedure, more cost, more time and one more specialized service that needs to exist in her city. The authors conclude exactly that: preservation is safe, feasible and should be considered beforehand.
• Sickle cell disease already compromises male fertility before any treatment. The median of 2 vials against 6, and of 7.1 million/mL against 80.4, is a difference between the two diseases, not an effect of the therapy. Whoever arrives for collection already arrives with less.

Why base editing and prime editing exist

Between 2018 and 2021, a series of results corrected the optimism of the first phase. All below were checked in PubMed.

The logic that links the four

The central safety problem of the CRISPR nuclease is not “the scissors cutting in the wrong place”. It is that cutting the double strand in the right place is already, by itself, a genotoxic event. Improving the guide's aim does not solve it.

That is exactly why base editing (2016) and prime editing (2019) exist: both were designed to edit without breaking the double strand. They are not convenience improvements — they are an engineering response to this block of literature.

And the question about secondary cancer still has no answer. No trial cited here has follow-up long enough to rule it out. The follow-up studies themselves state the timeline: Casgevy's long-term study has primary completion scheduled for 2039; a competitor's, for 2040. Until then, “no cancer occurred” means “no cancer within 48 months”, and nothing beyond that.

What changed in 2026: it left the transplant and went into the vein

Casgevy edits cells outside the body. The next generation edits inside — a lipid nanoparticle carrying the Cas9 messenger RNA and the guide to the liver, by intravenous infusion. No cell harvest, no chemotherapy, no transplant.

And that generation already has a placebo-controlled phase 3, which Casgevy never had.

The infant case, read for what it is

• It is not a demonstration of efficacy. It is n=1, with seven weeks of follow-up, and the authors themselves write that longer follow-up is needed.
• What it demonstrates is logistical and regulatory: a genome editing drug designed for a single person was manufactured, approved and infused within months. That is the precedent.
• If this repeats, the “one drug, one trial, one population” model stops describing the field. And with it, the logic of evidence this site uses to judge everything else stops holding.

The graveyard

Survey of ClinicalTrials.gov on September 4, 2026: 119 registered studies with CRISPR as the intervention. A sample of 60 was examined record by record. The pattern that appears is not random.

What the graveyard shows

• An entire block of CRISPR-edited allogeneic cell therapy was discontinued. It is not statistical chance: it is the area where the promise of the “off-the-shelf cell” ran into immune rejection and short cell persistence.
• What remains alive has another profile: monogenic diseases with a single, well-defined target, and editing in the liver by nanoparticle. Phase 3 underway for transthyretin amyloidosis and for hereditary angioedema.
• The new frontier is the riskiest. There is already a trial registered in 2026 of in vivo editing of the angiotensinogen gene to treat hypertension — a common, chronic disease with dozens of cheap existing treatments. A permanent genomic edit for a condition controlled with a daily pill entirely changes the calculus of acceptable risk.

Price and access: the declared bottleneck is not money

Casgevy has a list price of £1,651,000 per treatment in the United Kingdom — the same for both indications, with a discount to the public health service whose size is declared commercial confidentiality — and of US$ 2.2 million in the United States. Before approval, the American health technology assessment institute had calculated a ceiling of US$ 1.35 to 2.05 million. The price came out above the top of that range.

But the number that really explains access is not that one. When NICE, in England, projects how many people will be treated, it does not cite cost as the reason for low uptake. It cites “the long hospital stay required for the process involved”.

This section is a summary. The full survey — who pays, under what conditions, the cheaper comparator that wins in 100% of simulations, what Brazil knows and does not know about its own patient population — is in Casgevy — price and access.

Three things this funnel reveals

• Casgevy is not an alternative to transplant — it is what is left for those who found no donor. The authorization itself requires that transplant be appropriate and that there be no matched family donor. The indication was designed inside the niche that transplant left empty.
• And there is a cheaper comparator that wins in 100% of simulations. A cost-effectiveness analysis published in 2026 compared standard of care, haploidentical transplant and gene therapy: the standard yields 14.3 quality-adjusted life years at US$ 1.22 million; haploidentical transplant yields 20.1 at US$ 1.15 million; gene therapy yields 22.1 at US$ 2.75 million. Haploidentical beat gene therapy in 10,000 of 10,000 Monte Carlo iterations. Gene therapy is clinically better and economically losing — and haploidentical is precisely the technique that dispenses with full matching, that is, it attacks the niche of the indication from within.
• The distance between where the patients are and where the therapy is spans orders of magnitude. 515,000 babies per year are born with sickle cell disease worldwide, mostly in sub-Saharan Africa and the Caribbean; there are 7.74 million people living with the disease and a mortality burden of 376,000 deaths per year, of which 81,100 in children under 5. The same cost-effectiveness study estimated the cost-effective price ceiling of gene therapy in Nigeria, India and Tanzania at US$ 4,200 to US$ 22,000. The list price is 520 times the floor of that range. No confidential discount closes that.

In Brazil: not a slow queue, a queue not started

Gene and cell therapy does not go into the drug registry: it is an Advanced Therapy Product (PTA), with separate registration, under RDC 505 of 2021. Searching “exagamglogene” in the open drug database returns zero — and that means nothing, because it is the wrong database.

The sweep that closes the question

The 12 petitions are from Novartis (3, granted), Gilead (2, granted), PTC, Janssen, Roche, BioMarin (cancelled), Bristol-Myers Squibb, Ferring and Ultragenyx — the last three still under review. None is from Vertex, the holder of Casgevy.

To rule out the hypothesis that the petition sits under another subject, the sweep was redone by CNPJ. Vertex Farmacêutica do Brasil Ltda. — identified by the CNPJ listed as holder of the registrations for Trikafta, Symdeko, Orkambi and Kalydeco, all cystic fibrosis drugs — has 541 petitions in the system.

What this means, and what it does not

• There is no registration application for Casgevy in Brazil. And there is not even a clinical trial application, nor compassionate use, nor expanded access. It is not a case of “it is in the queue and takes time”: it never entered the queue.
• The Portuguese-language sources claiming approval “in early 2024” are wrong. Now this can be said on positive evidence, not on absence of proof.
• Without registration, SUS cannot even evaluate it. Decree 7,646/2011, in article 15, requires that the incorporation request document the process with the number and validity of the registration. The Brazilian debate over the price of Casgevy is not stuck on price — it is stuck three steps earlier.
• And, if it ever unlocks, we will assess in the dark. The best official estimate of how many people have sickle cell disease in Brazil is from 2007 and ranges between 25 thousand and 50 thousand. The Ministry of Health's 2024 clinical protocol says verbatim that “recent data were not identified”. The United Kingdom published 1,794 eligible with the whole funnel open; here there is no denominator.
• What was not verified: whether any petition was filed by another legal entity on behalf of the product. The query is by CNPJ and by subject, and the public interface does not return trade names. A text search for “exagamglogene” would require the SEI Public Search, which asks for a CAPTCHA and was not used.

A distinction the public debate misses

Everything on this page is somatic editing: it alters the patient's own cells, is not heritable, dies with them. It is what is approved, what is in trials and what has a price.

What usually dominates the conversation about CRISPR is something else — heritable germline editing, which alters the genome passed on to descendants. In 2018 a Chinese researcher announced the birth of twin girls whose embryos had been edited, an episode that generated its own academic literature and led the World Health Organization to publish, in July 2021, a governance framework and a set of recommendations in nine areas, the result of more than two years of global consultation.

They are completely distinct ethical and legal regimes. Confusing the two is the main source of noise in the debate — and makes it look as if the approved therapy carries a moral problem it does not have.

What this page teaches the rest of the site

• Fourteen years, 64,641 articles, 67 clinical trials, 1 approved product. That is the real scale of a technology that worked. Any compound in this reference that promises a comparable effect with a fraction of that evidence is promising from nothing.
• The evidence that convinces an agency is of another order. Two phase 3 trials, 96 patients combined, follow-up of up to 48 months, engraftment documented in all, and even so the European approval came out conditional and the United Kingdom only cleared it with continued data collection. Compare with “doses reported by users on a forum”.
• The most serious adverse effect usually comes from the procedure, not the molecule. Here it is busulfan, not CRISPR. It serves as a lens for reading any report of harm: the question is always what else was being done at the same time.
• “No serious adverse events” has an expiration date. The long-term studies of this therapy end in 2039 and 2040. A compound that has existed for three years and “caused no problems in anyone” has said absolutely nothing.
• And approval in another country is not approval here. Three agencies approved Casgevy. In Brazil there is no registration, no petition, no trial. The distance between “it exists in the world” and “it exists legally for you” is exactly the subject of the rest of this reference.

References

1. Jinek M et al., 2012 — a programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. It is the Nobel paper. Science 337(6096):816-821.
2. Ishino Y et al., 1987 — sequence of the iap gene in E. coli. First published observation of the repeats, with no known function. J Bacteriol 169(12):5429-5433.
3. Mojica FJM et al., 2005 — the intervening sequences derive from foreign genetic elements. J Mol Evol 60(2):174-182.
4. Barrangou R et al., 2007 — CRISPR provides acquired resistance against viruses in prokaryotes. Work done at Danisco. Science 315(5819):1709-1712.
5. Deltcheva E et al., 2011 — CRISPR RNA maturation by trans-encoded small RNA and host RNase III. Nature 471(7340):602-607.
6. Cong L et al., 2013 — multiplex genome engineering using CRISPR/Cas systems. Science 339(6121):819-823.
7. Mali P et al., 2013 — RNA-guided human genome engineering via Cas9. Published in the same issue. Science 339(6121):823-826.
8. Komor AC et al., 2016 — programmable editing of a single base in genomic DNA without double-strand cleavage. Nature 533(7603):420-424.
9. Gaudelli NM et al., 2017 — programmable base editing of A•T to G•C without DNA cleavage. Nature 551(7681):464-471.
10. Anzalone AV et al., 2019 — prime editing: search-and-replace genome editing, without double-strand breaks or donor DNA. Nature 576(7785):149-157.
11. Frangoul H et al., 2024 — exagamglogene autotemcel for severe sickle cell disease. Phase 3, 44 patients. N Engl J Med 390(18):1649-1662.
12. Locatelli F et al., 2024 — exagamglogene autotemcel for transfusion-dependent beta-thalassemia. Phase 3, 52 patients. N Engl J Med 390(18):1663-1676.
13. Frangoul H et al., 2021 — CRISPR-Cas9 editing for sickle cell disease and beta-thalassemia. The first two patients. N Engl J Med 384(3):252-260.
14. Kosicki M et al., 2018 — repair of CRISPR-Cas9-induced breaks leads to large deletions and complex rearrangements. Nat Biotechnol 36(8):765-771.
15. Haapaniemi E et al., 2018 — CRISPR-Cas9 editing induces a p53-mediated DNA damage response. Nat Med 24(7):927-930.
16. Ihry RJ et al., 2018 — p53 inhibits CRISPR-Cas9 engineering in human pluripotent stem cells. Nat Med 24(7):939-946.
17. Leibowitz ML et al., 2021 — chromothripsis as an on-target consequence of CRISPR-Cas9 editing. Nat Genet 53(6):895-905.
18. Cohn DM et al., 2026 — lonvoguran ziclumeran: in vivo CRISPR editing in hereditary angioedema. Placebo-controlled phase 3, 80 patients. N Engl J Med.
19. Cohn DM et al., 2025 — CRISPR-based therapy for hereditary angioedema. Randomized phase 2, 27 patients. N Engl J Med 392(5):458-467.
20. Gillmore JD et al., 2021 — in vivo CRISPR-Cas9 gene editing for transthyretin amyloidosis. First in vivo trial published. N Engl J Med 385(6):493-502.
21. Fontana M et al., 2024 — nexiguran ziclumeran in ATTR amyloidosis cardiomyopathy. Phase 1, 36 patients. N Engl J Med 391(23):2231-2241.
22. Musunuru K et al., 2025 — patient-specific in vivo gene editing to treat a rare genetic disease. Case report, n=1. N Engl J Med 392(22):2235-2243.
23. Hmaidan S et al., 2025 — safety, feasibility and fertility preservation outcomes in patients with sickle cell disease and beta-thalassemia undergoing gene therapy. 40 patients. Transplant Cell Ther.
24. Chetlapalli K et al., 2026 — haploidentical transplant, gene therapy and standard of care in sickle cell disease: cost-effectiveness analysis. Blood.
25. GBD 2021 Sickle Cell Disease Collaborators, 2023 — global, regional and national prevalence and mortality burden of sickle cell disease, 2000-2021. Lancet Haematol 10(8):e585-e599.
26. MHRA — world-first authorization of a gene therapy for sickle cell disease and transfusion-dependent beta-thalassemia, November 15, 2023.
27. FDA — CASGEVY product page, with the approval dates and those of each supplement.
28. FDA — approval of the first gene therapy for young children with sickle cell disease, July 1, 2026.
29. EMA — public assessment report for Casgevy, with the date and type of the European authorization.
30. NICE — guidance TA1044, exagamglogene autotemcel for severe sickle cell disease. Includes the list price, the confidentiality of the discount and the eligibility funnel.
31. NICE — guidance TA1003, exagamglogene autotemcel for transfusion-dependent beta-thalassemia.
32. WHO — recommendations on human genome editing, published July 12, 2021.
33. Greely HT, 2019 — edited babies: human germline genome editing in the He Jiankui case. J Law Biosci 6(1):111-183.
34. Decree No. 7,646, of December 21, 2011 — article 15: the request for incorporation into SUS requires the number and validity of the registration.
35. CONITEC — Recommendation Report No. 924, Clinical Protocol and Therapeutic Guidelines for Sickle Cell Disease, 2024. Source of the Brazilian epidemiology cited.

← Back to all compounds