BRENNEN

Industries / Frontier science / Cell and gene therapy

Testing and approval

This work tests cell and gene therapies in people and decides, through regulators, whether they may be sold and on what evidence.

6

Milestones achieved, from 1990 to 2024.

3

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1990

    Marked cells tracked in people

    A team led by Steven Rosenberg at the US National Cancer Institute (NCI) added a bacterial gene, used only as a tag, to tumour-fighting white blood cells (tumour-infiltrating lymphocytes) taken from five people with advanced melanoma, then infused the cells back. The tag showed the cells in the blood of all five for three weeks and in tumours up to 64 days later, with no side effects put down to the gene transfer; the results were published in August 1990. NCI says Rosenberg "was the first to successfully insert foreign genes into humans".

    europepmc.orgccr.cancer.gov

    Credit

    Steven A. Rosenberg and colleagues, NCI

  2. 2007

    Europe's rules for advanced therapies

    On 13 November 2007 the EU adopted Regulation 1394/2007 on advanced therapy medicinal products (ATMPs), the legal class that covers gene therapies, cell therapies and engineered tissue. It requires one EU-wide approval, but leaves out products made "on a non-routine basis" in a hospital for one patient, under a doctor's responsibility and within one country: the hospital exemption.

    Lessoneach member state runs the hospital exemption its own way, and a product made under it can be used only in the country that authorised it.

    eur-lex.europa.eueuropepmc.org

    Credit

    European Parliament and Council of the EU

  3. 2016

    A designation for regenerative therapies

    The 21st Century Cures Act, enacted on 13 December 2016, created the regenerative medicine advanced therapy (RMAT) designation. The FDA grants it to cell therapies and, by its own reading of the law, to some gene therapies, when the product is for a serious disease and early clinical evidence suggests it could meet an unmet need. By the FDA's count, by 30 September 2025 it had received 388 requests, granted 193 and denied 165, and 14 products holding the designation had been approved, from Breyanzi in 2021 to Zevaskyn in 2025.

    fda.govfda.govfda.gov

    Credit

    US Congress; FDA Center for Biologics Evaluation and Research (CBER)

  4. 2019

    Trial plans reviewed once

    From the first trials, the NIH's Recombinant DNA Advisory Committee (RAC) reviewed US gene therapy trial plans. On 25 April 2019 NIH said it had changed its guidelines to drop duplicate review and reporting of these plans, which the FDA, ethics boards and local biosafety committees must still approve before any patient is treated, and to turn the RAC toward emerging technologies in general.

    grants.nih.govgenome.gov

    Credit

    NIH Office of Science Policy, with the FDA

  5. 2019

    Approved against natural history

    On 24 May 2019 the FDA approved Zolgensma, a one-dose gene therapy for spinal muscular atrophy (SMA) in children under 2, on two open-label trials with no control group. Results were compared with the known course of the untreated disease, in which, by the maker's account, only about 25% of comparable babies would be expected to stay alive without permanent breathing support. On 6 August 2019 the FDA said the maker, AveXis, had known before approval that some animal-testing data in its application had been manipulated, but told the FDA only afterwards; the FDA kept the product on sale because the human trial data were not affected.

    Lessonthe FDA said complete and accurate data across the whole application is the maker's legal duty, and that it could seek civil or criminal penalties.

    novartis.comfda.gov

    Credit

    AveXis, a Novartis company; FDA

  6. 2024

    Early approval taken back

    In 2014 Japan began a conditional, time-limited approval route for regenerative medicine products: a product can be sold on limited evidence, then must confirm its benefit within a set period or lose its approval. Terumo's HeartSheet, sheets grown from a patient's own skeletal muscle cells to treat heart failure, was approved this way on 18 September 2015; a 2025 review says it was the only product so approved "without any improvement reported" when its maker applied. Follow-up studies did not show a benefit, and the PMDA, Japan's medicines agency, lists the approval as withdrawn on 25 July 2024. The same review says a conditionally approved gene therapy, Collategene from AnGes and Mitsubishi Tanabe Pharma, was also withdrawn for lack of evidence.

    Lessona conditional approval on weak evidence can keep a product on sale, and paid for by public insurance, for years before its benefit is tested.

    europepmc.orgpmda.go.jp

    Credit

    Japan's Ministry of Health, Labour and Welfare and PMDA; Terumo

Watch for

  1. Approval on one study

    On 3 September 2025 the FDA introduced the Rare Disease Evidence Principles (RDEP) process, for treatments that correct a known inborn genetic defect in diseases with generally fewer than 1,000 US patients, that worsen fast and that have no adequate alternative. For these, the FDA expects that one adequate and well-controlled study, which may have a single arm, plus strong confirmatory evidence, such as comparison with natural-history studies, can be enough. Sponsors must apply before the main trial starts.

    Achieved whenthe FDA approves a gene therapy it reviewed through the RDEP process and says so.

    fda.govfda.gov

    Claimed by

    FDA (CDER and CBER), September 2025

  2. Randomised trials for CAR-T

    In December 2025 FDA officials, among them the agency's biologics chief, Vinay Prasad, wrote in JAMA that evidence for a new CAR-T therapy in cancer "should generally demonstrate superiority" over a control group in a randomised trial. By Applied Clinical Trials' count, all seven approved CAR-T therapies reached the market on single-arm trials, with no control group. The FDA's decision on Gilead and Arcellx's anito-cel for multiple myeloma, whose pivotal trial was single-arm, is due on 23 December 2026.

    Achieved whenthe FDA's next new CAR-T approval rests on a randomised trial, or the FDA turns one down for relying on single-arm data.

    appliedclinicaltrialsonline.comajmc.com

    Claimed by

    FDA officials (Mahat, Das, Kumar and Prasad), December 2025

  3. Clearer rules for early approval

    Japan put draft rules for its conditional, time-limited approval route out for public comment from 25 July to 25 August 2025. They cover advice before applying, review, follow-up after approval, re-application within the set period, and use of the route for rare-disease products already approved abroad.

    Achieved whenthe final rules are in force and a product has been approved, or has had its conditional approval confirmed, under them.

    pmda.go.jp

    Claimed by

    Japan's health regulators, as presented by the PMDA, July 2025

Open

  1. 01

    A running clinical-hold log

    A public table of every clinical hold, an FDA order that stops a trial, placed on a cell or gene therapy trial since 2020, as disclosed in company filings and press releases. Each row gives the product type, the dates the hold was placed and lifted, the stated reason (patient harm, manufacturing, animal data) and the quoted sentence it came from. The table refreshes monthly.

    The brief
    Credit · unclaimed
  2. 02

    What each approval rested on

    A public table of the main trials behind every cell and gene therapy approved by the FDA, the European Medicines Agency (EMA) and Japan's PMDA. One row per trial gives the kind of control (randomised, natural-history comparison, the patient's own baseline, none), number of patients, main measure of benefit, follow-up time at approval, and the regulator's stated doubts, quoted with page numbers.

    The brief
    Credit · unclaimed

Open 01 · the brief

A running clinical-hold log

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public table of every clinical hold, an FDA order that stops a trial, placed on a cell or gene therapy trial since 2020, as disclosed in company filings and press releases. Each row gives the product type, the dates the hold was placed and lifted, the stated reason (patient harm, manufacturing, animal data) and the quoted sentence it came from. The table refreshes monthly.

Why it is still open

The only list-based count found is a 2023 paper by Wills, Drago and Pietrusko, which analysed 33 holds announced from January 2020 to December 2022 (15 on AAV gene therapies, 9 on CAR-T) and found about 80% lifted, after 6.2 months on average; it stops at 2022. A 2024 Clinical Leader column estimated from public disclosures that at least 136 cell and gene therapy trials had been put on hold since January 2020, and cited a Pink Sheet count, obtained through a freedom-of-information request, of 147 holds in 2018 and 70 in 2022; neither publishes a list. The column notes that the FDA often cannot share what it learns from holds because the data are commercially confidential.

europepmc.orgclinicalleader.com

Built with what exists

The US Securities and Exchange Commission's EDGAR full-text search to find "clinical hold" in company filings; company press-release pages; an AI model such as Claude to pull dates and reasons with the quoted sentence; a person checking each row; a CSV file in a public GitHub repository. The note "Findings that cite notes" (Cursor's agent kit) covers an agent built to cite the source behind each finding.

  1. First step

    Run an EDGAR full-text search for "clinical hold" with "gene therapy" or "CAR-T" for 2020 to 2022 and compare the hits with the holds the 2023 paper reports.

  2. Done looks like

    The 2020 to 2022 rows include at least 30 of the 33 holds in Wills and colleagues' study, every later row links to a filing or release, a second person's check of 30 random rows finds at least 95% correct, and the table has refreshed on schedule for 3 months.

Open 02 · the brief

What each approval rested on

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public table of the main trials behind every cell and gene therapy approved by the FDA, the European Medicines Agency (EMA) and Japan's PMDA. One row per trial gives the kind of control (randomised, natural-history comparison, the patient's own baseline, none), number of patients, main measure of benefit, follow-up time at approval, and the regulator's stated doubts, quoted with page numbers.

Why it is still open

A 2021 review coded the 23 main trials behind the 17 advanced therapies approved in the EU up to January 2021 and found 13 (56.5%) single-arm and only 7 with a placebo or active control. A 2024 study coded single-arm approvals of 2019 to 2022 at the FDA and EMA, but only for non-cancer drugs of every kind. Neither covers cell and gene therapies approved since 2021 at all three regulators, or is kept current, while the FDA's 2025 RDEP process makes single-arm evidence more common.

europepmc.orgeuropepmc.org

Built with what exists

FDA product pages and clinical review memos, EMA public assessment reports (EPARs), PMDA English review reports, pdftotext, an AI model such as Claude to draft each code with a page citation, and a person to confirm it.

  1. First step

    Write a one-page coding guide and code Zolgensma, Casgevy, Lenmeldy, Elevidys and Aucatzyl from their FDA review memos.

  2. Done looks like

    Every product on the FDA's approved list and every EMA-authorised advanced therapy coded; the recoding of the 23 EU trials in the 2021 review matches its control type for at least 21; and a second coder agrees on at least 90% of fields in a sample of 30 trials.