Treating a lymphoma with the patient’s own cells appears to be the future of oncology. A study published in Nature Medicine on 580 patients shows an association between the number of prior treatments and the cellular composition of the product. The immune composition collected by apheresis is correlated with that of the manufactured CAR-T product. But this result, precise as it is, fits into a larger movement: that of living medicines, whose scope, ambitions and constraints are expanding as biology advances.
The essentials
- The cellular state at apheresis is correlated with product composition, and the proportion of so-called “naïve” cells is associated with clinical outcomes.
- Over 580 patients followed for a median of 41.9 months, a high proportion of immune cells called “naïve” (CCR7+CD45RA+) in the product is associated with a 54% reduced risk of progression or death, according to Nature Medicine on the ZUMA trials.
- This result was validated on 203 independent patients: durable response doubles between groups with low and high proportions of these cells.
- The challenge shifts from the laboratory to health systems: a greater number of prior therapeutic lines is associated with a lower proportion of naïve cells in the product.
- Beyond lymphomas, living medicines are extending to autoimmune diseases: CAR-T directed against CD19 deplete autoreactive B lymphocytes, potentially by resetting the immune system to achieve durable remissions.
CAR-T cells are manufactured from what the patient still has
To understand why timing matters, you need to understand what a CAR-T medicine is. Doctors extract T lymphocytes from the patient’s blood. A laboratory genetically modifies them so they recognize and attack cancer cells. Then this product, unique to each individual, is reinjected.
The mechanism is elegant. It is also fragile. T lymphocytes are not all equivalent. Cells identified by the markers CCR7+CD45RA+ are described as naïve-like or stem cell-like memory, less differentiated than conventional naïve cells, and associated with a better capacity for durable response. A greater number of prior treatment lines is associated with a lower proportion of naïve cells in the CAR-T product.
Prior treatments are associated with a decrease in the proportion of naïve-like T cells, an indicator of differentiation and fitness of the cells used to manufacture the CAR-T. A low proportion of naïve-like cells at apheresis is correlated with a low proportion in the product; a low proportion in the product is associated with less durable responses.
Personalized medicine has a constraint that standard medicine does not: the patient’s characteristics and prior treatments are associated with the composition of the CAR-T product.
The ZUMA trials and the biology of therapeutic success
The ZUMA trials involve axi-cabtagene ciloleucel, an anti-CD19 CAR-T therapy used in aggressive B-cell lymphomas. The main cohort of 580 patients was followed for a median of nearly four years. Researchers analyzed the cellular composition of manufactured products and correlated it with clinical outcomes.
The central result is clear: patients whose product contains a high proportion of CCR7+CD45RA+ cells have a 54% lower risk of progression or death than those whose product contains few. This gap is considerable. Durable response—that is, maintenance of remission over time—doubles between the two groups.
To ensure this result holds, the researchers submitted it to an independent validation cohort of 203 patients. The signal was confirmed. This type of double validation on an emerging technology is rare and carries weight to the conclusion.
The exact causal direction remains undetermined. A patient in better immunological condition is also generally a patient less advanced in his disease. However, the researchers controlled for the usual clinical factors, and the cellular signature of the product remains independently predictive. The composition of the medicine matters in itself, and not merely as a reflection of the patient’s overall condition.
The treatment line as a biological parameter
This tension reveals a structural mismatch between the temporality of reimbursement decisions and the biological temporality of the patient. Access criteria were built around disease progression, without integrating the parallel progression of immune exhaustion. The two evolve simultaneously, but only the first is measured and considered when deciding on eligibility.
This is where biological data becomes a question of health policy. Depending on the indication and jurisdiction, some CAR-T therapies are accessible as early as the second line while others remain reserved for later lines. The number of treatments required depends on the cancer, the product and the jurisdiction; some indications begin after a single line. This logic is understandable: treatments cost hundreds of thousands of dollars per patient, and reimbursement systems seek to reserve them for cases where alternatives have been exhausted.
A lower proportion of naïve cells in the product is associated with less favorable progression-free survival. Patients with more prior therapeutic lines tend to have fewer naïve cells in the CAR-T product.
The study suggests that prior lines and the cellular composition of the product are linked. This observation is uncomfortable, because it suggests that short-term cost optimization could compromise treatment efficacy, and thus, ultimately, its true benefit-cost ratio.
Several clinical trials are already exploring the use of CAR-T in the second line. Early results are encouraging, but regulators and payers have not yet adjusted their criteria accordingly.
Indications expanding far beyond blood cancers
The question of timing is not unique to hematologic oncology. It runs through the entire field of living medicines, whose scope has expanded considerably over the past two years.
The most striking advances concern autoimmune diseases. Published in Nature Medicine in January 2026, the CASTLE phase 1/2a study evaluated anti-CD19 CAR-T therapy in patients with refractory systemic lupus erythematosus, systemic sclerosis and idiopathic inflammatory myopathies. Of 24 patients treated, 22 met efficacy criteria: 9 of 10 lupus patients achieved DORIS remission, 9 of 9 patients with systemic sclerosis had no progression, and 4 of 5 patients with inflammatory myopathy responded by ACR criteria—with no cytokine release syndrome grade higher than 2, nor neurotoxicity.
This result points to a new logic: the living medicine acts as a treatment that persists in the body, continuing its action long after administration. For some clinicians, this marks the beginning of an era defined not by disease control, but by the possibility of eliminating it.
In solid tumors, progress remains more modest, but a threshold has been crossed. An important result concerns certain gastric and gastro-esophageal junction cancers expressing CLDN18.2: the study on satricabtagene autoleucel drew attention as the first positive randomized trial of a CAR-T in a solid tumor. Three structural obstacles remain: tumor heterogeneity, immunosuppressive microenvironment and on-target off-tumor toxicity.
The year 2025 saw the continuation of scientific advances in cell therapy and commercial expansion of CAR-T, accompanied by notable strategic contractions mainly related to manufacturing cost constraints. This paradox—increasingly convincing clinical results, fragmented industrialization—is at the heart of current debates.
Early identification of eligible patients as a new clinical challenge
If the immune state at the time of collection conditions the response, the logical next step is to identify eligible patients before their immune capital erodes. This is technically feasible: CCR7+CD45RA+ markers are measured by flow cytometry, a standard technique in hematology laboratories.
The real difficulty is organizational. Centers that perform CAR-T are few in number. In France, about twenty institutions are authorized to administer them. In the United States, access remains unevenly distributed across the territory. The delay between the medical decision and administration of the finished product often reaches four to eight weeks, during which the disease can progress and the immune system continue to deplete.
Work is underway to reduce this manufacturing delay. Ultra-rapid manufacturing platforms reduce the delay below 24 hours: clinical studies confirm that shorter ex vivo cultures paradoxically improve the phenotype of T cells—less exhaustion, more stem potential—questioning the idea that longer expansions guarantee a better quality product.
More radical still: in vivo programming of CAR-T involves delivering genetic cargo directly to circulating T cells via targeted lipid nanoparticles or pseudo-viral particles, eliminating ex vivo manufacturing entirely. This approach reduces total manufacturing time to an estimated one to three weeks and lowers projected costs by at least 50% per treatment. The cost of ex vivo autologous manufacturing currently runs between 100,000 and 150,000 dollars per dose; in vivo approaches aim for 2,000 to 10,000 dollars at commercial scale, with vein-to-vein time reduced from several weeks to a few hours. No in vivo product has yet received regulatory approval, but several are in phase 1.
So-called “off-the-shelf” CAR-T cells, manufactured from healthy donors rather than from the patient, could partially solve this timing problem, at the cost of sometimes shorter immune response. No allogeneic CAR-T therapy has yet received FDA approval, despite rapid clinical progress. This biological trade-off is not yet resolved.
Another lever is early cryopreservation: collecting and storing cells at the time of diagnosis, before the first treatment lines, to have higher quality raw material if the disease relapses later. This approach raises logistical, economic and equity questions—not all potentially eligible patients can be collected preventively.
Debates on hospital system efficiency, which work on spending and governance in Europe demonstrates, remind us that access depends on more than drug price: the capacities for referral, coordination and early identification carry equal weight.
Reimbursement as a biological decision
By 2030-2035, reimbursement systems will need to decide whether to treat CAR-T as an expensive last resort or as an intervention to schedule early, when the patient’s immune capital makes a durable response more likely.
The two trajectories have very different consequences.
In the first, CAR-T remains reserved for patients in third line or beyond. Payers maintain the therapeutic escalation logic: you don’t engage an expensive treatment as long as a cheaper alternative is available. This logic is budgetarily defensible in the short term. But it means administering a product of lower biological quality to patients whose disease is more advanced and prognosis more bleak. Average clinical outcomes suffer, and hidden costs—relapses, hospitalizations, salvage treatments—could erase much of the anticipated savings.
In the second, regulators and payers integrate collection biology into access criteria. A patient in the second line, with a favorable immune signature, would be directed to a CAR-T before his naïve T cells have disappeared. This approach requires an early identification capacity that current systems don’t yet have: immune stratification tools deployed routinely, a sufficient number of centers, controlled manufacturing times.
It also requires an evolution in the standard of evidence required by payers. Today, reimbursement is granted on the basis of randomized trials comparing CAR-T to the best available therapy in the treatment line concerned. To validate second-line use, trials of the same caliber would be needed, with the costs and timelines that entails. Manufacturers have an incentive to fund them: earlier use means a larger market. But trials take time, and disease waits for no one.
A signal to watch: several regulatory agencies, including the FDA in the United States and the EMA in Europe, are examining second-line applications based on observational data and biomarkers. In June 2025, the FDA eliminated the REMS program for CAR-T, removing site certification and tocilizumab storage requirements, and shortening proximity monitoring obligations—a signal that the regulator is seeking to expand access to these therapies. The decision to validate this type of indirect evidence will largely determine how quickly the paradigm can shift. Work on health spending governance in central and eastern Europe, where institutional capacities condition investment effectiveness, shows that the quality of regulatory and reimbursement decisions is a structuring factor, not an administrative detail.
A second signal: trials comparing second-line CAR-T to standard chemotherapy (some preliminary results positive) should produce long-term survival data by 2027-2028. If this data confirms a net benefit in the second line, pressure on payers to revise their criteria will become hard to ignore.
A third signal concerns the economic models themselves. Manufacturers and some health systems are experimenting with outcome-based contracts: the payer reimburses fully if the patient responds durably, partially otherwise. This type of contract changes manufacturers’ incentive to improve the biological quality of the product, and thus to support earlier access. If these contracts become generalized, they could align interests where regulatory criteria lag in evolving.
Manufacturing as care
This logic means that collection can no longer be treated as a purely logistical step sandwiched between two clinical decisions. Cellular composition at apheresis is correlated with that of the product, and the fraction of naïve cells is associated with clinical outcomes. Integrating this constraint into routine practice means requalifying the collection procedure as an intervention in its own right, subject to its own criteria of appropriateness, and not merely as a simple supply of raw material.
The ZUMA study raises a question standard medicine has never had to address: the timing of drug manufacturing. For a chemical molecule, this timing is inconsequential. For the axi-cel studied, product composition is associated with outcomes, and this composition is linked to the cellular profile collected.
This reality changes the definition of care. The decision to collect early or late is a medical decision just as much as the choice of molecule or dose. Oncologists who refer patients to CAR-T now make a biological prescription as much as a clinical one, even if reimbursement systems haven’t yet integrated this dimension.
The cell and gene therapy sector has spent the last decade proving the potential of living medicines to alter the course of disease. Well advanced in 2026, the question is no longer whether these therapies work, but whether they can be delivered to patients at scale, consistently and in a financially viable manner. This tension between scientific validation and industrial deployment cuts across all boundaries of the class: from lymphoma to lupus, from solid tumors to neurological disease.
Immune biomarkers like CCR7+CD45RA+ could become screening tools as common as the tumor markers now used to qualify access to certain immunotherapies. Precision medicine thus extends to the medicine itself, not merely to the disease it targets.
Decision-makers must now answer two distinct questions: who can benefit from a CAR-T, and when should it be manufactured for it to be effective. Answering the first without answering the second amounts to optimizing a process while ignoring its most decisive constraint. And with the broadening of indications—cancers, autoimmune diseases, tomorrow perhaps other chronic pathologies—this constraint will only grow in importance.
Sources
- Nature Medicine, Multicenter study on axi-cabtagene ciloleucel (ZUMA trials), 2026
- Autoimmune Institute, CAR T-Cell Therapy for Autoimmune Disease: 2026 Update
- AJMC, CAR T-Cell Therapy in 2026 and Beyond: Evolving Evidence, Expanding Indications, and the Path to Broader Access
- Nature, The in vivo revolution in CAR-T therapy medicinal products: challenges and regulatory prospects, 2026
- Hartmann Institute, CAR-T therapies in solid tumors, 2026
