Synthetic biology & cell therapy/China
Two targets instead of one, and 87 percent complete responses with nothing to compare them to
Dual CD19/CD20 CAR T cells with decitabine priming produced an 87 percent complete response rate in 23 lymphoma patients, uncontrolled.
A Chinese PLA General Hospital team ran an open-label non-randomised phase I/II trial of decitabine-primed CD19/CD20 dual-targeted CAR T cells in 23 patients with relapsed or refractory non-Hodgkin lymphoma. They report an 87 percent complete response rate and 77 percent two-year progression-free survival at 24.3 months median follow-up.
What happened
A team from the Chinese PLA General Hospital, Southern Medical University, People’s Liberation Army units, Nankai University and Changping Laboratory published, in Nature Communications, a phase I/II trial of decitabine-primed CD19/CD20 dual-targeted CAR T cells in 23 patients with relapsed or refractory non-Hodgkin lymphoma.
Reported: an 87 percent complete response rate and 77 percent two-year progression-free survival at a median follow-up of 24.3 months.
Open-label. Non-randomised. No control arm.
The test
CAR T cell therapy against CD19 transformed the treatment of several blood cancers and then ran into a predictable problem. If you kill every cell displaying a particular marker, you select for cells that have stopped displaying it. Antigen escape is a leading cause of relapse after an initially successful response.
The straightforward answer is to aim at two markers at once. A tumour cell then has to lose both CD19 and CD20 to escape, which is a less probable event.
The second idea is subtler. Decitabine is a hypomethylating agent – it removes methyl groups from DNA, which can reactivate silenced genes. The rationale for priming with it is that it raises expression of the target antigens, making the tumour more visible to the CAR T cells.
So the trial tests a combination:
– Dual targeting: CD19 and CD20 in a tandem construct – Priming: decitabine before the CAR T infusion – Design: open-label, non-randomised phase I/II; registration NCT04697940 – Patients: 23 with relapsed or refractory non-Hodgkin lymphoma – Follow-up: 24.3 months median
The numbers
| Measure | Value |
|---|---|
| Patients | 23, relapsed or refractory non-Hodgkin lymphoma |
| Construct | CD19/CD20 dual-targeted, decitabine-primed |
| Complete response rate | 87 percent |
| Two-year progression-free survival | 77 percent |
| Median follow-up | 24.3 months |
| Tolerability | described as well tolerated |
| Control arm | none |
| Randomisation | none |
The result
Eighty-seven percent complete responses in relapsed or refractory lymphoma is a high figure. Twenty of 23 patients, if the percentage is read against the enrolled number. And 77 percent two-year progression-free survival means most of those responses held.
The design is what limits how much that can mean. Three specific issues, in order of importance.
No control arm. Single-arm response rates in cancer depend enormously on who was enrolled. “Relapsed or refractory non-Hodgkin lymphoma” spans histologies with wildly different prognoses, patients who have had one prior line and patients who have had five, bulky disease and minimal disease. An 87 percent complete response rate in a favourable selection and in an unfavourable one are different results carrying the same number.
Two changes at once. The trial combines dual targeting with decitabine priming. If the result is good – and it looks good – the design cannot say which component delivered it, or whether both were needed. Decitabine is itself an active agent in some haematological cancers, so it is not a neutral pre-treatment.
Twenty-three patients. A phase I/II trial is sized to detect signals and establish safety, not to measure efficacy precisely. The confidence interval around 87 percent in 23 patients is wide, and the record does not give it.
On the follow-up figure: a median of 24.3 months means half the patients have been observed for longer and half for less. The two-year progression-free survival estimate therefore relies partly on censoring-adjusted projection rather than on every patient having reached two years. This is standard survival analysis and worth understanding when reading the number.
None of this means the result is not real. Dual-targeted CAR T is a well-motivated approach and 87 percent with durable responses is the kind of signal that justifies a randomised trial. It is a reason to run that trial, not a substitute for it.
How this compares to what came before
This batch contains three cell therapy trials of almost identical size and very different evidential strength:
| Study | Design | n | Control | Follow-up |
|---|---|---|---|---|
| mRNA BCMA CAR T, myasthenia gravis | phase 2b randomised double-blind | 26 | placebo | 12 months |
| CASTLE, CD19 CAR-T autoimmune basket | phase 1/2a, prespecified endpoints | 24 | none | 24 weeks |
| This paper, tandem CD19/CD20 | open-label non-randomised phase I/II | 23 | none | 24.3 months median |
Each has one clear advantage. The myasthenia gravis trial has a placebo arm, which is the strongest design feature available in this field. CASTLE has a protocol where every patient stopped their other immunosuppressants, which substitutes for a control in a limited way. This paper has by far the longest follow-up – 24.3 months median against 12 months and 24 weeks.
In oncology, durability is the outcome that matters, so long follow-up is a genuine strength and not a consolation. A complete response that holds for two years is clinically different from one measured at three months.
The broader pattern across the three: cell therapy is producing impressive effects in small numbers of patients and has not yet produced many controlled comparisons. Within this batch, one of three trials randomised. That ratio is the field’s evidential state, and readers should calibrate accordingly.
What it means
For CAR T development, dual targeting addresses a mechanistically understood failure mode and this trial supports taking it further.
For trial design, the next experiment is to separate the two interventions. Dual-targeted CAR T with and without decitabine priming, randomised, would tell you whether the priming earns its place.
For clinicians, an 87 percent complete response rate from an uncontrolled 23-patient trial should inform which trials to refer patients to, not current practice.
Business ideas from this paper
Tandem CAR construct engineering. Designing a single receptor that recognises two antigens well is a genuine protein engineering challenge. Construct design and optimisation is a licensable capability.
Antigen expression modulation. The priming idea generalises: any agent that raises target expression makes immunotherapy more effective. Screening for safer, more specific expression modulators than a chemotherapy agent is a drug discovery programme.
Antigen escape monitoring. The problem this therapy addresses is detectable before relapse, by watching antigen expression on residual disease. Diagnostic monitoring for escape would guide treatment sequencing.
Factorial trial design services. Combining two interventions and testing them together is common and uninformative. Expertise in factorial designs that separate contributions without doubling the patient count is a methodological service.
What could be wrong with it
Open-label and non-randomised in 23 patients, so the 87 percent complete response rate has no comparator and depends heavily on who was enrolled.
Two interventions were combined – dual targeting and decitabine priming – so neither contribution can be isolated.
Decitabine has activity against some haematological cancers in its own right, making it an active rather than neutral pre-treatment.
Two-year progression-free survival at 24.3 months median follow-up means a substantial share of patients have not been observed for two years.
No confidence intervals are reported in the record for the response or survival estimates, and no data availability statement was identified.
How sure can you be?
Confident that the trial was conducted and the responses observed, and that 24.3 months median follow-up is a real strength relative to most cell therapy reports.
Confident that dual targeting addresses a well-characterised failure mode, so the approach is mechanistically sound regardless of this trial’s precision.
Less confident in the 87 percent figure as a transferable efficacy estimate, because an uncontrolled response rate in 23 selected patients is the weakest form of efficacy evidence.
Not confident about which component produced the benefit, since the design combines two.
Not confident in the two-year survival estimate’s precision, given the follow-up distribution and the absence of reported intervals.
Our rubric places this at the floor of the range for this batch: peer-reviewed in a strong journal, five institutions, a prospectively registered clinical trial with long follow-up, without randomisation, a control arm or a public data deposit.
Do this today
When you read a single-arm response rate, ask who was enrolled. In relapsed lymphoma, patient selection moves response rates more than most interventions do.
If a trial combines two new things, note that its result belongs to the combination. That is a common and avoidable waste of patients.
If you follow CAR T, watch for the randomised dual-target trials. The approach is sound and the evidence is not yet there.
Source: Demethylation-primed tandem CD19/CD20 CAR T cells in relapsed or refractory B-cell lymphoma, Nature Communications, 2026-04-15. DOI 10.1038/s41467-026-72040-4 · doi.org ().
Just Out Tech explains new research in plain language. This article was drafted with AI assistance and checked by a human against the original source.
- An open-label non-randomised phase I/II trial of decitabine-primed CD19/CD20 dual-targeted CAR T cells in 23 patients with relapsed or refractory non-Hodgkin lymphoma.
- An 87 percent complete response rate is reported, with two-year progression-free survival of 77 percent at a median follow-up of 24.3 months.
- Treatment is described as well tolerated with durable responses.
- The design combines two changes at once - dual targeting and demethylation priming - so neither contribution can be separated.
Open-label non-randomised phase I/II trial in 23 patients with 24.3 months median follow-up; no control arm and two interventions combined.
Questions people ask
Why does single-target CAR T fail?
Because tumours adapt. CAR T cells aimed at CD19 kill every cell displaying CD19, which selects for cells that have stopped displaying it. That is called antigen escape and it is a leading cause of relapse after initially successful treatment. Targeting two antigens at once means a cell must lose both to escape, which is a less likely event.
What does decitabine priming do?
Decitabine is a hypomethylating agent: it strips methyl groups from DNA, which can switch genes back on. The rationale here is that it increases expression of the target antigens, making tumour cells more visible to the CAR T cells. It is also an active chemotherapy for some blood cancers in its own right, which is part of why the design is hard to interpret.
How should an 87 percent complete response rate be read?
With the denominator and the design in view. Twenty-three patients, open-label, no control arm. Relapsed or refractory lymphoma covers a very wide range of prognoses depending on histology, prior lines of therapy and disease bulk, and single-arm response rates in selected patients are the least reliable efficacy evidence there is.
What does 77 percent two-year progression-free survival at 24.3 months median follow-up mean?
That half the patients have been followed for at least 24.3 months, which means the other half have been followed for less - so the two-year estimate rests partly on statistical projection rather than on everyone actually reaching two years. That is standard practice and worth knowing when reading the figure.