MedChinaGo
If CAR-T Comes Back: What Still Works After Relapse in China
November 11, 2026·Updated November 18, 2026|12 min read

If CAR-T Comes Back: What Still Works After Relapse in China

Timing, re-biopsy and the options that remain

OncologyTreatment GuideCancer TreatmentImmunotherapyClinical TrialsMedical TourismWhy Choose China?

When CAR-T fails, the next step depends on when it failed and what the new biopsy shows. A plain guide to second CAR-T, bispecifics, ADCs and trials in China.

The question nobody wants to ask

Every CAR-T conversation has an unspoken second half. The first half is whether the therapy works. The second is what happens if it does not.

It is worth answering plainly, because the answer is not the same for everyone. Some patients who progress after CAR-T still have several effective options, and some have very few. The difference between those two groups is not luck — it is timing, target biology, and how quickly a next plan is made.

This is what the evidence says about relapse after CAR-T in China, what can still be done, and how to tell which situation you are in.

Cartoon illustration of a road that splits into several paths at a fork, with a signpost showing different directions and a small figure standing at the junction looking at a map. Bright flat vector cartoon style, warm clinic colors, no text

"Relapse" is three different situations

Doctors do not treat these the same way, and it helps to know which one you are in.

Primary refractory disease. The disease did not respond at all, or progressed within the first month. The infused cells never gained control. This is the hardest group, and it is also the group where a second attempt is least likely to behave differently.

Early relapse — within three months. There was a response, then it slipped away quickly. The cells either did not expand enough or did not persist. This group has the worst outcomes after salvage, consistently across studies.

Late relapse — beyond six months. There was a durable response, and then the disease returned. This is a fundamentally different biological situation, and it is the group where subsequent therapy works best.

One number separates these groups more sharply than any other: the time from infusion to progression. In a multicentre analysis of 92 patients treated with bispecific antibodies after CAR-T failure, the overall response rate was 29% in early relapse, 54% in the intermediate 4–6 month group, and 60% in late relapse — with median progression-free survival of 2.2, 3.7 and 10.5 months respectively.

The numbers, honestly

In large B-cell lymphoma, roughly 60% of patients progress within two years of a CD19 CAR-T infusion. Once that happens, published outcomes are sobering but not uniform:

Situation after CD19 CAR-T failure Typical outcome with salvage
Median overall survival, whole group 5.7–7.5 months
Salvage chemotherapy alone ORR 8–54%, CR 3.8–35%
Median PFS with conventional salvage 1.4–3.8 months
Bispecific antibody after failure ORR 43%, median PFS 2.8 months
Bispecific after late relapse (>6 months) ORR 60%, median PFS 10.5 months
Bispecific as first salvage after late relapse Median PFS not reached
Second infusion of the same CD19 product ORR 19–54%, median duration of response 81 days

The last two rows matter most. Late relapse treated early with a different mechanism can produce durable control. The same product given again rarely does.

Why it comes back

The mechanism determines which option makes sense, which is why a repeat biopsy is not optional.

Antigen loss. The tumour stops expressing the target. In CD19 CAR-T this is well documented, and it can be complete or partial. A clone that never had CD19 simply survives the therapy.

Antigen density and heterogeneity. The target is still there, but at lower density, or on only some cells. CAR-T killing is density-dependent — below a threshold, the engineered cell engages but does not kill.

T-cell exhaustion and poor persistence. The infused cells arrive but do not expand, or expand and then disappear. This is the dominant failure mode in early relapse, and it is why simply repeating the same product tends to fail again.

Microenvironment and tumour-intrinsic resistance. The tumour is protected, or carries lesions (TP53, double-hit, triple-hit) that make any immunotherapy less effective.

Lineage switch. Mostly seen in leukaemia: the tumour changes identity and stops being a B-cell malignancy at all. Rare, but it is a reason why relapses are re-biopsied rather than assumed.

Before choosing anything: re-biopsy

The single most useful thing to do at progression is a fresh biopsy with immunohistochemistry for the original target and for the targets you might use next.

For a B-cell lymphoma that failed CD19 CAR-T, that means CD19, CD20 and CD22, plus TP53 and MYC/BCL2 status. For myeloma after BCMA CAR-T, it means BCMA and GPRC5D expression — GPRC5D is expressed through a pathway independent of BCMA, which is exactly why it is the usual next choice when BCMA is lost. For gastric cancer after satri-cel, it means repeat CLDN18.2 staining with a percentage and intensity score.

This sounds bureaucratic. It is the difference between choosing a therapy that has a target and one that does not.

Repeating the same target rarely works

It is the most intuitive option — the first infusion helped, so give another — and the data is consistently disappointing.

In the TRANSCEND trial, re-treatment with the same CD19 product after failure produced an overall response rate of 19%. In ZUMA-1, re-treatment with axi-cel produced responses in 54%, but the median duration of response was 81 days — fewer than three months of benefit for a second full course of lymphodepletion, inpatient monitoring and CRS risk.

There are exceptions: patients who lost the cells to a manufacturing or persistence problem rather than to tumour resistance can respond again. But as a general strategy, giving the same product to the same resistant tumour is spending a large amount of money and risk on a low probability.

Switching the target is the strongest option in blood cancers

If the tumour has lost or reduced CD19, the logical move is to attack a different surface protein.

CD20-directed CAR-T. In a Chinese series of CD19 CAR-T failures, CD20-directed T-cell therapy — ten patients with anti-CD20 CAR-T and two with a CD20 bispecific — produced an overall response rate of 100% with 91.7% complete remissions, against 18.2% in patients who received non-CD20 salvage. The C-CAR066 CD20 CAR-T study reported ORR 92.9% and CR 57.1% in the same setting. CD20 expression tends to survive CD19 loss, which is the whole rationale.

CD22-directed CAR-T. CD22 CAR-T after CD19 failure has reported ORR 68% and CR 53%.

Dual-target products. CD19/CD22 bispecific CAR-T (CAR2219) in relapsed large B-cell lymphoma showed cytokine release syndrome in 74% of patients with no grade 3 or higher, and ICANS in 6% — a notably mild toxicity profile for a dual-target construct. A separate Chinese series using CD19/CD22 CAR-T after bispecific antibody failure reported ORR 100% with 50% complete responses.

CAR-T plus transplant. In 124 patients not in remission after bridging, CD19/CD22 cocktail CAR-T followed by autologous transplant produced ORR 93.2% versus 78.5% for CAR-T alone, with median progression-free survival not reached versus 5.68 months. For younger, fit patients, consolidation with transplant remains a real option.

Bispecifics: the option that does not wait for manufacturing

A bispecific antibody redirects your existing T cells to the tumour. It is off-the-shelf, subcutaneous or intravenous, and available in days rather than the two to three weeks a CAR-T needs.

That speed matters enormously in early relapse, where the disease can double in the time a CAR-T would spend in the manufacturing facility.

In lymphoma, CD20×CD3 bispecifics produce complete remissions in roughly a third of patients who failed CAR-T. Glofitamab is commercially available in China and was used in the Tongji series above, including in a patient with central nervous system relapse.

In myeloma, the picture changed materially in October 2026. Teclistamab — a BCMA×CD3 bispecific — received Chinese approval for patients after at least one prior line, moving it into first relapse rather than last resort. In the MajesTEC-9 trial it reduced the risk of progression or death by 71% (HR 0.29) and the risk of death by 40% (HR 0.60) against standard combinations, with 65.9% achieving complete response or better versus 16.8%. Cytokine release syndrome occurred in 66% and was mostly grade 1–2; ICANS in 4.1%.

The trade-off is different from CAR-T: bispecifics are given continuously, infections are more frequent (grade 3 or higher in 41.6%), and immunoglobulin replacement becomes part of life. But there is no manufacturing wait and no apheresis.

Which of the two to use first is now a sequencing question with its own rules, including how earlier BCMA exposure affects a later BCMA CAR-T — CAR-T or bispecific first at myeloma relapse.

Solid tumours: what comes after satri-cel

This is the least settled area, and the one where patients most need a straight answer.

satri-cel, approved in China in June 2026 for CLDN18.2-positive gastric and gastroesophageal junction cancer, is the first CAR-T approved anywhere for a solid tumour. There is no second solid-tumour CAR-T approved to fall back on. What exists is a set of CLDN18.2-directed alternatives that work by a different mechanism:

Therapy after CLDN18.2 CAR-T Mechanism Reported result in later-line gastric cancer
Sonesitatug vedotin (CMG901) CLDN18.2 antibody–drug conjugate 3L+ OS 10 vs 5 months, HR 0.47
Arcotatug tavatecan (IBI343) CLDN18.2 ADC PFS 7.1 vs 3.1 months, OS 10.2 vs 6.9 months
CLDN18.2×CD3 bispecific (IBI389) T-cell engager ORR 30.8%, median PFS 3.5 months
Dual-target CAR-T (CLDN18.2/NKG2DL) Investigational ORR 75% in early phase

Two caveats keep this honest. These trials were run in later-line gastric cancer generally, not exclusively in patients who had failed CAR-T — the subset with prior CLDN18.2-directed therapy in the IBI343 trial was small (34 patients), though the benefit held (OS HR 0.39). And CLDN18.2 expression can fall after targeted treatment: in one paired-biopsy study after zolbetuximab, 53.3% of patients no longer met the high-expression threshold, but 73.3% still met a lower one. That is why ADCs with a 25% expression cutoff may reach patients whom a stricter CAR-T criterion would exclude.

The practical point: after satri-cel, the realistic next step is a CLDN18.2 ADC, and it depends on what the post-progression biopsy shows.

Cartoon illustration of a laboratory slide being examined under a microscope with colourful cell markers highlighted, and a friendly scientist pointing at two different protein markers on the same cell. Bright flat vector cartoon style, clean warm background, no text

Cartoon illustration of a hospital building with several research wings, each marked by a different coloured molecule icon, and a friendly clinician leading a patient toward one of them with a clipboard. Bright flat vector cartoon style, warm palette, clean composition, no text

Trials: China's pipeline is unusually deep here

For patients who have exhausted approved options, China currently runs an unusually large number of trials in exactly this post-CAR-T space.

In multiple myeloma alone, as of late 2026 there were 27 recruiting BCMA-directed trials and 14 GPRC5D-directed trials across more than 300 centres — including BCMA/GPRC5D dual-target CAR-T, allogeneic off-the-shelf CAR-T, and trispecific antibodies targeting BCMA, GPRC5D and CD3.

In lymphoma, CD19/CD22 dual-target CAR-T trials are recruiting across several centres, and in-vivo CAR-T — where the vector is injected and the engineering happens inside the body, with no apheresis and no manufacturing wait — is in early-phase testing.

Trial entry is not a consolation prize. In a disease where standard salvage produces a median progression-free survival under two months, a trial with a plausible mechanism is often the better clinical choice, and it is frequently the only route to a novel target.

What makes a second attempt work — and the timing trap

Five factors predict whether a second cell therapy is worth attempting:

Time since the first infusion. Beyond six months is favourable; under three months is not.

T-cell fitness. Prior therapies, prior CAR-T and prolonged infection all degrade the T cells you would collect next. This is the main argument against waiting.

Disease burden at the time of the next treatment. Low burden responds better. Every month of uncontrolled growth makes the next attempt less likely to work.

Organ function and performance status. Lymphodepletion and CRS are poorly tolerated in a patient whose liver, kidneys or heart have already been stressed.

Washout from the last therapy. In the Chinese series of patients receiving CAR-T after bispecific failure, the median interval from last bispecific dose to apheresis was 35–42 days — long enough for T-cell recovery, short enough not to lose ground.

The trap is the same in every country: patients who relapse early are precisely the ones who respond worst to salvage, so they are often managed with chemotherapy while everyone waits for a better option to become available. Published Chinese experience argues the opposite — go to a targeted T-cell therapy early rather than exhausting fitness on ineffective chemotherapy first.

If you are already home

Most international patients return home after Day 30 or Day 90. Relapse is then detected by a local oncologist, often without access to the treating centre's protocol.

Three things make that handover work:

Get the discharge summary in a usable form. The exact product, target, dose, infusion date, CRS and ICANS history, and the imaging that established response. Without the infusion date, nobody can interpret the timing of a relapse.

Agree in advance who manages progression. Ask the Chinese centre, in writing, whether they will accept a returning patient for salvage, and under what conditions. Some will; some will not repeat cell therapy in a patient who already failed it.

Know the red flags that are not relapse. Persistent cytopenias, hypogammaglobulinaemia and late infections are common for months after CAR-T and are not the same thing as disease progression. Distinguishing them changes the treatment entirely.

If the answer is that you go back, the order — records first, then pathology, then visa, then travel — is in the patient journey guide.

The honest limits

Not every patient has a second option. For primary refractory disease with rapid progression and poor performance status, the honest conversation is about control and comfort, not about a third line of intensive therapy. That conversation is better had early than in an emergency.

Second attempts are not first attempts. Response rates are lower, durations are shorter, and toxicity in an already-treated patient is less forgiving. Anyone presenting a second CAR-T as equivalent to the first is overselling it.

The evidence is mostly retrospective. Many of the numbers above come from single-centre series and registry analyses, not randomised trials. They describe what happened to similar patients; they do not guarantee what will happen to you.

Availability differs from approval. A product approved in China may be at a limited number of centres, and a trial may have eligibility criteria that exclude you. Verify at the specific hospital rather than from a press release.

If the patient is a child, read our guide to CAR-T for children with relapsed B-ALL in China first — the paediatric indication, the trial data and the bridge to transplant work differently from the adult pathway.

Getting started

The useful question is not "can I have CAR-T again" but "what does my biopsy show, and what does the timeline say".

MedChinaGo's CAR-T programme reviews the pathology report, the infusion date, the response history and any post-progression biopsy, then returns a written assessment covering which targets remain, whether a second cell therapy, a bispecific, an ADC or a trial is the better fit, and what the realistic timeline looks like. Learn more about CAR-T therapy in China — or start with the background reading: which CAR-T products are approved in China, what satri-cel achieved in gastric cancer, how CAR-T compares with transplant and chemotherapy, and what the whole course costs.

Frequently Asked Questions

Author

M

MedChinaGo Medical Team

Medical Tourism Specialists

https://www.chinamedicaltrips.com/about

This information is for educational purposes only and does not constitute medical advice. Always consult with a qualified healthcare provider for personal medical decisions.

Ready to Start Your Medical Journey?

Contact us today for a free consultation and personalized treatment plan

Free Consultation