Early CAR-T and T-Cell Engager Trials in Autoimmune Disease: Where They Opened and Pre-Dosing Requirements

CAR-T and T-cell engagers in autoimmune disease: where early trials opened and what regulators expect before first dosing

Engineered T-cell therapies and CD3-based T-cell engagers (TCEs) have moved from oncology into lupus, systemic sclerosis, myositis, rheumatoid arthritis and other B-cell-driven autoimmune diseases. Across Germany, Australia, China and the United States, the registries show one pattern. The earliest registered start in this set is an academic study in China (NCT05030779, estimated start 10 Sep 2021). In Germany the earliest registered starts are industry-sponsored: Kyverna's KYSA-3 (1 Dec 2022) and Miltenyi's CASTLE study at Erlangen (July 2023). Other US and multinational industry programmes followed from 2023. Australia is taking part through industry-sponsored multiregional studies that also need gene technology licensing. TCEs started later, from late 2024, as academic studies in China and industry and academic studies in the EU and US.

The pre-dosing requirements differ in structure more than in scientific substance. All four regulators expect a risk-based quality and nonclinical package, a justified starting dose and staggered early dosing. The differences lie in who authorises the trial, how GMO aspects are handled, and whether the guidance was written for oncology.

For the general first-in-human authorisation routes in the US, China and Australia, see Starting a first-in-human trial: how FDA (IND), China's NMPA, and Australia's CTN/CTX schemes compare.

Registry snapshot: early autoimmune CAR-T and TCE trials by country

The dates below are registry start or decision dates. They are not verified first-patient-dosed dates. Few records describe themselves as first-in-human.

Other points from the registries:

• A second Kite programme in Australia covers multiple sclerosis, myasthenia gravis and CIDP. It was registered on 2 Dec 2025 ³⁶⁵. No autoimmune TCE trial was found on ANZCTR, but Amgen's platform study (NCT06570798) lists a recruiting site in Perth.
• In China, three CAR-T products for refractory SLE hold CDE clinical-trial implied-permission records, as distinct from hospital-registered investigator-initiated studies. The three are a CD19-targeted non-viral PD1-specific CAR-T from Shanghai Bangyao ⁷⁴⁴, ICG318 ⁷⁴⁸, and a CD19/CD22 CAR-T from Shanghai Pharmaceuticals Group Biotherapy ⁷⁵⁶. The records do not carry dates.
• In CTIS, the Leipzig and Fraunhofer TCE trials list Germany as their only location, and the BMS TCE trial lists Germany, Spain, Poland and Italy ^834835836.
• The industry studies are mostly not first-in-human in the strict sense. For example, the Breakfree-SLE protocol for CC-97540 states that the product had already been tested in a Phase 1 autoimmune study ³³⁸. The COMPARE protocol cites a separate first-in-human lymphoma study of the same CAR construct as KYV-101 ³⁰¹.

United States: FDA/CBER

Gate to first dosing. A cell or gene therapy IND generally takes effect 30 calendar days after FDA receives it. Dosing can start then unless FDA imposes a clinical hold. FDA can also tell the sponsor earlier that the study may proceed ⁶⁸⁶. CBER recommends early contact before an IND is filed:

• An INTERACT meeting suits novel products once proof-of-concept work exists but before definitive nonclinical studies begin ⁶⁸⁷⁶⁹¹.
• A pre-IND meeting covers nonclinical design, the initial clinical study, and manufacturing ⁶⁸⁷.

CAR-T products. FDA's January 2024 CAR T guidance is written for oncology. It tells sponsors developing non-oncology indications, which include autoimmune disease, to discuss product-specific issues with OTP/CBER before submitting the IND ⁵⁶⁷. Within that framework the IND must show the following.

• CMC. Stage-appropriate CMC data are required, as follows:
  • vector characterisation and lot release covering safety, identity, purity and biological activity ⁵⁷⁹
  • adventitious-agent and microbiological testing ⁵⁷⁹
  • control of CQAs, with assays for identity, purity and potency ⁵⁶⁶⁵⁵⁹
  • a potency readout for each target of a multi-antigen CAR ⁵⁷¹
• Retroviral and lentiviral vectors. These fall under FDA's RCR guidance. It recommends RCR testing at several points during vector production ⁵¹⁷⁵²³.
• Nonclinical package. It must cover on-target/off-tissue and off-target binding, and it must show that the cells do not signal or proliferate without antigen ⁵⁶¹⁵⁵⁶. For allogeneic products it must also address GVHD and host rejection ⁵⁵⁶.
• Clinical design.
  • The starting dose should not rest only on animal data, because CAR T cells expand in vivo ⁵⁵⁵.
  • Staggered enrolment is recommended for a first-in-human product or one with limited human experience ⁵⁶².
  • DLTs should be defined without regard to attribution to the product ⁵⁷⁵.
  • Stopping rules might include more than two Grade 4 CRS events for a first-in-human CAR T product, or any death within 30 days of dosing ⁵⁷⁵.
• Long-term follow-up. Subjects who receive CAR T cells with an integrated transgene should be followed for 15 years ⁵⁵⁸.

T-cell engagers. FDA's 2021 bispecific antibody guidance sets the following expectations:

• Sponsors should assess homodimers of anti-CD3 constructs, because they can trigger cytokine release ⁵⁸⁵.
• For agonistic bispecifics, sponsors should consider a MABEL-based first dose and discuss dose selection with the review division ⁵⁸³.
• The guidance cites an analysis of 17 CD3 bispecifics in which a first-in-human dose giving 10 to 30% pharmacological activity was acceptable ⁵⁸³.
• Healthy volunteers may be unsuitable for these studies ⁵⁸³.

A June 2026 draft guidance on QSP-based MABEL takes a conservative line. Where estimates diverge, sponsors should use the lowest one for novel targets. It also recommends safeguards such as staggered enrolment and intensive monitoring ⁵⁸⁹.

Germany: PEI or BfArM under EU CTR and the AMG

Gate to first dosing. A trial may start only once the competent federal authority has authorised it under Article 8 of Regulation (EU) 536/2014. The application goes through the EU portal. The ethics committee's assessment binds the authority on the matters assigned to it ⁶³³⁶³⁵. Other national conditions in the AMG:

• For purely national trials, or trials run in Germany and third countries, the sponsor or a sponsor's representative must be established in the EU/EEA ⁶³⁴.
• Subject insurance is required, with at least EUR 500,000 per case of death or permanent incapacity ⁶³⁴.
• Consent must meet AMG § 40b ⁶³⁸.

Which authority. Under AMG § 77, the Paul-Ehrlich-Institut is competent for ATMPs, which include gene therapy medicinal products such as CAR-T ⁴⁶⁷. PEI describes CAR-T cells as gene therapy ⁵³⁶. Monoclonal antibodies are not among PEI's listed categories, so BfArM is competent by default unless a product is reassigned ⁴⁶⁷. This means antibody-format TCEs would normally go to BfArM.

GMO overlay for CAR-T. Where the investigational product consists of or contains GMOs, the sponsor must add the following to the application:

• a GMO risk assessment
• an environmental assessment
• a monitoring plan
• information on residues
• emergency plans

The authority consults the Federal Office of Consumer Protection and Food Safety (BVL). The trial authorisation then includes authorisation to release the GMO within the trial ⁶³³⁶³⁴.

Scientific expectations (EMA investigational ATMP guideline).

• GMP specific to ATMPs is a prerequisite for the trial ⁶¹⁵.
• Most nonclinical data should be available before first administration ⁶¹⁸⁶¹⁹. The minimum package is:
  • proof of concept
  • support for a safe starting dose
  • biodistribution and persistence data
  • an assessment of insertional mutagenesis for integrating vectors
  • an assessment of tumourigenicity
• Administration should be staggered between subjects. The waiting period should reflect the timing of toxicity seen in animals and in human experience with related products ⁶²³.

For TCEs, the EMA first-in-human guideline (Rev. 1, 2018) sets these expectations:

• The starting dose should be anchored to MABEL, PAD or NOAEL, with justified safety factors ⁶⁰⁰⁵⁹⁶.
• A single subject normally receives the first dose, and the rest of the cohort is dosed in sequence ⁶⁰³.
• The next cohort should not be dosed until the previous cohort's data have been reviewed ⁶⁰³.

The 2007 version of the guideline flagged immune targets that can drive cytokine cascades, such as CD3 or CD28 super-agonists, for special attention ⁵⁹⁷.

Hospital exemption. AMG § 4b allows some non-routinely manufactured ATMPs to be used without a marketing authorisation, under PEI approval and GMP. PEI still stresses that clinical trial data remain important ⁵³⁶.

Australia: TGA, OGTR and HREC in parallel

TGA gate. Under the CTN scheme (Schedule 5A item 3 of the Therapeutic Goods Regulations):

• The sponsor, or the organisation conducting the trial, approves it after considering the ethics committee's advice ⁷⁸⁸.
• The sponsor must notify the Secretary before the goods are first used in the trial, and again for each additional site ⁷⁸⁸⁷⁹⁹.

Under the CTA pathway, the approval holder and each principal investigator must give written GCP assurances and undertakings to the Secretary before trials start ⁷⁸⁷. Under either scheme, use must follow an ethics-approved protocol, the GCP guideline and the National Statement. Use must stop if the ethics committee advises it is inconsistent with the approval ⁵⁰⁹.

OGTR gate for CAR-T. Whether a CAR-T trial needs a licence depends on the product and on whether manufacture takes place in Australia.

• Administering GM somatic cells counts as an exempt dealing, with no OGTR licence needed, if all of the following apply (Schedule 3, Part 3.1(n) of the Gene Technology Regulations 2001) ⁴²⁶⁴³¹:
  • the cells cannot produce infectious agents
  • no viruses able to recombine with the GM nucleic acid were detected in testing
  • the viral vector is no longer present in the cells
• If any of these conditions is not met, the trial needs a GMO licence ⁴²⁶⁴²⁷.
• Manufacturing with a replication-defective lentiviral or retroviral vector cannot proceed as an NLRD. It needs its own authorisation, even if the final infusion qualifies as exempt ⁴²⁶⁴³¹.

Autoimmune programmes have in practice obtained DNIR licences. DNIR-717 covered CB-010 in SLE (issued 25 Mar 2025, now surrendered) ⁸³³. DNIR-735 covered BMS-986515 in severe refractory autoimmune diseases (issued 6 Nov 2025) ⁴²³.

HREC gate. The National Statement requires the following for trials like these:

• Risks must be weighed against the risks of the condition and of usual care ⁴⁵³.
• Safety monitoring must be proportionate to risk. A DSMB is recommended and in some cases required ⁴⁵⁶⁴⁵⁷.
• The trial must be registered publicly before the first participant is recruited ⁴⁵³.

China: CDE implied approval, plus a large investigator-initiated sector

Gate to first dosing (registration pathway). NMPA Announcement No. 50 of 2018 set up a 60-day implied (tacit) approval system for clinical trial applications ⁸²¹⁸²⁶. The 2019 Drug Administration Law (Article 19) now sets the period at 60 working days. CDE guidance describes pre-IND communication as required before a Phase I application ⁸¹⁹. The CDE implied-permission records for SLE CAR-T products noted above went through this route ^744748756.

Many of the earliest Chinese autoimmune CAR-T and TCE studies are hospital-sponsored and registered on ClinicalTrials.gov, for example at Zhejiang University, Ruijin Hospital and Shandong First Medical University ^271421420. The national rules for investigator-initiated clinical research were not in the sources reviewed here. The regulatory basis for those studies should be checked separately.

CAR-T expectations. CDE's cell therapy guidance is mostly written for oncology:

• a 2020 draft pharmaceutical guideline for immune cell products ⁶⁴⁹
• a 2024 clinical pharmacology guideline for cell therapy ⁶⁴⁶
• 2024 CAR-T clinical guidelines for haematological malignancies ⁶⁴⁷⁶⁵¹

The 2020 draft clinical guideline for immunotherapy products is the most directly relevant. It says cohort sizes in trials for non-malignant indications should reflect that population's risk tolerance, and that larger samples may be needed for safety assessment ⁵⁴⁶. The 2024 cell therapy clinical pharmacology guideline adds two points ⁵⁴⁵:

• Interactions with immune-modulating co-medications should be assessed.
• The effect of lymphodepletion on cell kinetics should be considered.

CDE also expects the following for CAR-T:

• ongoing monitoring for clonal expansion and secondary T-cell tumours ⁶⁶⁴
• a multidisciplinary safety panel for trial safety issues ⁶⁶⁴

The CDE CAR-T guideline does not set a numerical long-term follow-up period ⁶⁶⁴.

TCE expectations. CDE's 2025 guideline on antibody clinical pharmacology says MABEL should be the priority method for high-risk antibodies. That covers antibodies with:

• immune-agonist activity
• novel mechanisms
• cascade-amplification effects

Where methods give different starting doses, the sponsor should choose the lower one ⁴⁸⁸⁴⁹⁷. The 2022 guideline on bispecific antitumour antibodies has a section on risk control in first-in-human trials ⁴⁸⁶⁵⁰⁵. CDE's immunogenicity guidance supports staggered dosing between individuals and cohorts, and PK/PD review before dose escalation ⁵⁰³. CDE's guideline on immune-related adverse events sets no CRS-specific operational rules for TCEs ⁵⁰⁶.

Comparative takeaways for regulatory strategy

Two practical points follow for sponsors.

• Guidance gap. In the sources reviewed, none of the four regulators has final guidance written specifically for autoimmune CAR-T or TCE programmes. Both the FDA and CDE CAR-T frameworks are oncology-based. FDA explicitly sends non-oncology sponsors to pre-IND discussion ⁵⁶⁷⁶⁴⁷.
• GMO pathway. This is the main cross-border difference for CAR-T. Germany folds GMO release into the trial authorisation ⁶³⁴. Australia runs a separate OGTR track, where licensing depends on whether the final product meets the somatic-cell exemption conditions and where manufacture takes place ⁴²⁶.