Three numbers get quoted at every myeloma conference and in almost every patient email we receive. 96%. 92%. 34.9 months. The first is the response rate of equecabtagene autoleucel (eque-cel) in the FUMANBA-1 trial. The second is the response rate of zevorcabtagene autoleucel (zevor-cel) in LUMMICAR-1. The third is the median progression-free survival of ciltacabtagene autoleucel (cilta-cel) in CARTITUDE-1.
Put side by side, they look like a ranking. They are not. They come from three different trials, run in different countries, in patients who had received different amounts of prior treatment, followed for different lengths of time, and measured with different schedules. Reading them as a league table is the single most common mistake made by patients comparing BCMA CAR-T options in China.
This article takes the three datasets apart, one at a time, and puts the pieces back together in a way that is actually useful for a decision. For the separate question of when to use CAR-T versus a bispecific at first relapse, see our sequencing guide. Here the subject is the trial data itself.

The three trials, side by side
| FUMANBA-1 | LUMMICAR-1 | CARTITUDE-1 | |
|---|---|---|---|
| Product | eque-cel (福可苏) | zevor-cel (赛恺泽) | cilta-cel (西达基奥仑赛) |
| Where it ran | 14 centres in China | 23 centres in China | United States |
| Patients infused | 103 | 102 | 97 |
| Median prior lines | 4 (3–23) | 4 (3–15) | 6 (3–18) |
| High-risk cytogenetics | 38.6% | 45.1% | — |
| Extramedullary disease | 13.6% | 10.8% | — |
| Prior anti-CD38 antibody | 20.4% | 29.4% | 84.5% triple-class refractory |
| Prior BCMA therapy | Allowed (11.7% had prior CAR-T) | Excluded | Excluded |
| Bridging therapy used | 46.6% | 25.5% | — |
| Primary publication | JAMA Oncology, 2024 | Experimental Hematology & Oncology, 2025 | JCO / ASCO, final analysis 2023 |
The first thing that jumps out is the anti-CD38 column. In the two Chinese trials, roughly one patient in five to one in three had ever received daratumumab or another CD38 antibody. In CARTITUDE-1, more than eight in ten were triple-class refractory — refractory to a proteasome inhibitor, an immunomodulator and an anti-CD38 antibody. Those are not the same patients, and they are not the same disease.
That single difference explains much of the gap between the Chinese and American response rates. A patient who has never seen a CD38 antibody has more treatment-sensitive disease left. It is not a criticism of the Chinese products; it is a warning against reading 96% and 73% as a head-to-head result.
What each trial actually reported
| eque-cel (FUMANBA-1) | zevor-cel (LUMMICAR-1) | cilta-cel (CARTITUDE-1) | ide-cel (KarMMa, US benchmark) | |
|---|---|---|---|---|
| Overall response rate | 96.0% (97/101) | 92.2% (94/102) | 97.9% | 73% |
| CR or better | 74.3% at 13.8 months; 83.2% at 3 years | 71.6% (sCR 68.6%, CR 2.9%) | 82.5% sCR | 33% |
| MRD negative at 10⁻⁵ | 95.0% | 91.9% | — | — |
| Median PFS | Not reached at 13.8 months; 30.5 months at 3 years | Not mature; 12-month 76.3%, 18-month 61.9% | 34.9 months | 8.8 months |
| Overall survival | 92.2% at 12 months | 79.4% at 30 months | Median not reached | — |
Read the eque-cel column carefully, because it contains two different snapshots. The peer-reviewed JAMA Oncology paper, with a median follow-up of 13.8 months, reported an overall response rate of 96.0% and a CR-or-better rate of 74.3%, with median PFS not reached. The three-year update presented at the 2025 International Myeloma Society meeting reported an ORR of 96.3%, CR-or-better of 83.2%, and a median PFS of 30.5 months. Both are legitimate. They are simply not the same measurement, and quoting "30.5 months" next to a 13-month trial is an apples-to-oranges comparison.
The same caution applies in the other direction. Zevor-cel's median PFS is still not mature at a median follow-up of 20.3 months; the honest summary is "12-month PFS 76.3%, 18-month 61.9%", not a median that does not yet exist.
Response rate is the wrong headline
An overall response rate counts a partial response the same as a stringent complete response. That is why 96% sounds better than it is. What predicts durability is not whether the disease shrank, but how deep and how long the remission went.

The ladder runs: partial response → very good partial response → complete response → stringent complete response → MRD negativity. Most patients who benefit from BCMA CAR-T reach the top two rungs. In FUMANBA-1, 96 of 103 patients (95%) became MRD-negative at 10⁻⁵, and every patient who achieved CR or better was MRD-negative. In LUMMICAR-1, 91.9% of evaluable patients were MRD-negative at 10⁻⁵, and 100% of those in CR or sCR were MRD-negative.
That is genuinely strong. But 10⁻⁵ is the sensitivity most Chinese centres can report routinely, and it is not the same as 10⁻⁶. "MRD-negative" is a threshold, not an absolute. Ask which assay, which sensitivity, and whether the negativity was sustained at two consecutive tests — not just once.
30.5 months is a median, not a cure
A median PFS of 30.5 months means that, at that point, half of the patients in the analysis had progressed or died and half had not. It does not mean 30.5 months of benefit for everyone, and it does not mean the other half stopped responding at month 31.
The cilta-cel number is the most mature of the three: 34.9 months median PFS, with an estimated 47.5% of patients progression-free at 36 months. Even in that trial, roughly half the patients had progressed by three years. In myeloma, "half progression-free at three years" is a major advance over what came before — KarMMa's 8.8 months — but it is not a cure, and no single-arm trial can prove one.
There is also a structural trap in comparing medians across trials. A trial that follows patients longer will eventually report a longer median. A trial that enrols patients with less prior treatment will report a longer median. A trial that allows bridging chemotherapy and then measures PFS from infusion will report a different median from one that measures from apheresis. None of these differences is visible in the headline number.
Safety is where the choice is actually made
The efficacy numbers are close enough that, for most patients, safety and access decide the product. The three datasets differ more here than in response rates.
| eque-cel | zevor-cel | cilta-cel | |
|---|---|---|---|
| Cytokine release syndrome, any grade | 93.2% | 90.2% | About 95% |
| CRS grade 3 or higher | 1.0% (one grade 4) | 6.9% | About 4% |
| ICANS (neurotoxicity) | 1.9%, grade 1–2 only | 2.0%, grade 1 only | About one in five, mostly low grade |
| Serious infection | 31.1% | 27.5% grade ≥3 | A leading cause of non-relapse death |
| Delayed movement disorders | None reported in the pivotal trial | None reported | Rare, but reported across the programme |
| Related deaths in the pivotal trial | 2 (cerebral infarction, intracranial haemorrhage) | 1 (pneumonia, day 149) | — |
Two patterns are worth naming. Eque-cel has an unusually clean neurological profile — 1.9% ICANS, both grade 1–2 — but a high burden of grade 3–4 cytopenias and a 31% rate of serious infection; in FUMANBA-1, infection was the leading cause of death. Zevor-cel has a higher rate of grade 3–4 CRS (6.9%) but almost no ICANS and no movement disorders, and its cytopenias mostly recovered within 30 days.
Cilta-cel's efficacy is the most durable, and its neurotoxicity profile is the heaviest: ICANS in roughly one in five patients, plus the delayed movement and neurocognitive disorders that have been reported across the CARTITUDE programme. That trade-off belongs in the conversation, especially for older patients or those with a neurological history.
The lymphodepletion regimens also differ, and this is easy to miss. FUMANBA-1 used fludarabine 30 mg/m² with cyclophosphamide 500 mg/m²; LUMMICAR-1 used fludarabine 25 mg/m² with cyclophosphamide 300 mg/m². A gentler conditioning regimen may reduce toxicity, but it also changes the comparison with US trials, which mostly used fludarabine 30 mg/m² plus cyclophosphamide 300 mg/m².

The prior-BCMA question is the one that affects you
The most useful number in these datasets for a real decision is not the ORR. It is the subgroup result in patients who had already received a BCMA-directed therapy.
In FUMANBA-1, 12 patients had prior CAR-T. Their ORR was 75%, versus 99% in CAR-T-naive patients; CR-or-better was 42% versus 79%. Zevor-cel's trial excluded prior BCMA therapy outright, so no equivalent estimate exists. CARTITUDE-1 also excluded it.
That is a consistent signal: BCMA CAR-T works less well after prior BCMA therapy. It does not mean a second BCMA CAR-T is pointless — a 75% response rate in a heavily pretreated salvage population is still meaningful — but it does mean the order of treatment matters. If a bispecific antibody is used first and the disease then progresses, the later CAR-T starts from a weaker position. The full argument is in our first-relapse sequencing article.
Why the Chinese numbers are not a marketing claim
It would be easy to read this as "Chinese CAR-T is better". That is not what the data show, and it is not what this site will tell you.
Three structural reasons explain most of the apparent gap:
- Less CD38 exposure. Chinese registration trials enrolled patients with 20–30% prior anti-CD38 exposure; CARTITUDE-1 enrolled a triple-class-refractory population.
- Different follow-up and reporting. A three-year update reports deeper responses and a reached median; a 13-month analysis reports neither. Quoting one against the other is a presentational trick, not a scientific one.
- No head-to-head trials. No randomised study has compared eque-cel, zevor-cel and cilta-cel in the same population. Any ranking you see is inferred, not demonstrated.
What the Chinese data do support is narrower and more defensible: in heavily pretreated Chinese patients, both products produce deep, MRD-negative remissions at rates comparable to the best published BCMA CAR-T data, with safety profiles that differ in ways that matter clinically.
The ten questions that turn a percentage into a plan
- How many prior lines did the patients in this trial have, and how many had prior anti-CD38 treatment?
- What proportion were triple-class refractory, and how does that compare with me?
- Was the response rate measured at one month or at best response?
- What was the CR-or-better rate, not just the ORR?
- Was MRD tested at 10⁻⁵ or 10⁻⁶, and was it sustained?
- Is the median PFS mature, or is it a landmark rate (12-month, 18-month) instead?
- What were the grade 3–4 CRS and ICANS rates, not the "any grade" rates?
- What was the infection rate and the non-relapse mortality?
- Does the trial exclude prior BCMA therapy, and if so, why?
- Is the data peer-reviewed, or a conference abstract or company press release?
The honest limits
Every number in this article comes from a single-arm trial. There is no control group, no randomisation, and no blinding. Single-arm results are vulnerable to selection bias: investigators enrol patients well enough to wait for manufacturing, with adequate organ function and without rapidly progressive disease.
The Chinese trials enrolled only Chinese patients, so the data do not directly tell us how these products perform in other populations. Follow-up for overall survival is still short; the 30-month and three-year figures are estimates, not final. The three-year eque-cel data were presented at a conference and have not yet gone through peer review in the form quoted here. CARTITUDE-1 was run in the United States, so it is a benchmark for the construct, not evidence about care delivered in China.
None of that makes the data weak. It makes them specific — and it is exactly why the questions above matter more than the headline percentages.
Getting started
If you are considering BCMA CAR-T in China, the practical sequence is: confirm your prior lines and CD38 exposure, confirm whether you have had any prior BCMA-directed therapy, then ask each centre which product it can offer and under what label or trial. The Chinese labels for eque-cel and zevor-cel sit at three or more prior lines; earlier use usually means a clinical trial or off-label treatment with written consent.
Our CAR-T therapy in China page sets out what the programme includes, and the approved-products guide lists current list prices: eque-cel at ¥1,166,000 and zevor-cel at ¥1,150,000, with cilta-cel approved but not publicly priced. For total cost of care rather than the product alone, see CAR-T cost in China vs the US.
The numbers are the beginning of the conversation, not the end of it.

