Summary
Efficacy of chimeric antigen receptor (CAR) T‐cell therapy hinges on CAR‐T potency, as well as on tumour traits and disease status. Potency assessment currently relies on post‐infusion in vivo growth; therefore, manufacturing metrics could provide early potency predictions, enabling potency‐guided strategies. To assess the impact of ex vivo cell growth during manufacturing on clinical outcomes, we analysed diffuse large B‐cell lymphoma patients treated with tisagenlecleucel, merging clinical records with manufacturing parameters. Of 75 cases, 21 (28%) showed poor growth, indicated by any decrease in cell number during manufacturing. Good growth correlated with significantly better overall response rate (85.2% vs. 52.4%; p = 0.006), with enhanced lymphocyte increase in peripheral blood after infusion. Multivariate analysis revealed that this group had significantly higher progression‐free survival (PFS) (adjusted hazard ratio [aHR]: 0.377; 95% confidence interval [CI]: 0.180–0.789; p = 0.010) and overall survival (OS) (aHR: 0.191; 95% CI: 0.071–0.510; p = 0.001), with lower cumulative incidence of disease progression (aHR: 0.412; 95% CI: 0.198–0.858; p = 0.018), compared to the poor growth group, even after adjusting for clinical factors, bridging therapies performed between apheresis and infusion and disease status at infusion. Our findings suggest that good cell growth during manufacturing predicts favourable post‐CAR‐T outcomes. Identifying clinical factors that influence manufacturing parameters could improve post‐CAR‐T outcomes.
Keywords: chimeric antigen receptor T‐cell therapy, diffuse large B‐cell lymphoma, ex vivo cell proliferation, tisagenlecleucel
Efficacy of chimeric antigen receptor (CAR) T‐cell therapy depends not only on tumour traits but also on the potency of CAR‐T cells. In clinical practice, the degrees of cell proliferation during CAR‐T‐cell manufacturing vary significantly among individual cases. Notably, consistent cell proliferation during manufacturing is associated with favourable therapeutic outcomes. This study highlights the potential utility of proliferation data obtained during CAR‐T manufacturing, which is available well before infusion, as an early predictive marker of clinical response in patients with relapsed or refractory B‐cell lymphomas receiving CAR‐T therapy.
INTRODUCTION
Chimeric antigen receptor (CAR) T‐cell therapies directed against CD19 have revolutionized the treatment landscape for relapsed or refractory (r/r) B‐cell lymphomas, offering sustained remission in cases in which conventional therapies have failed. 1 , 2 , 3 Despite these advances, the efficacy of CAR‐T‐cell therapy remains highly variable among patients, suggesting a complex interplay of factors influencing treatment outcomes. This variability underscores the need to identify biomarkers that accurately predict therapeutic success.
While the efficacy of CAR‐T‐cell therapies has been attributed to tumour‐intrinsic traits and disease status, emerging evidence suggests that the potency of CAR‐T cells themselves is critical. 4 , 5 Currently, assessments of CAR‐T potency have relied on monitoring the expansion of these cells post‐infusion, yet the information obtained after infusion is too delayed to be useful for clinical decision‐making. Thus, there is growing interest in whether metrics observed during the CAR‐T‐cell manufacturing process, particularly the behaviour of CAR‐transduced cells, could serve as early indicators of CAR‐T‐cell potency and ultimately treatment efficacy.
In this study, we focused on ex vivo cell proliferation after CAR transduction during manufacturing. By evaluating cell proliferation in combination with clinical information, we analysed the impact of cell proliferation during manufacturing on therapeutic outcomes. Our results will enable personalized treatment strategies based on the potency of CAR‐T cells, potentially contributing to the improvement of post‐CAR‐T‐cell therapy outcomes in r/r B‐cell lymphoma patients.
PATIENTS AND METHODS
Patients and date collection
This retrospective study enrolled all consecutive patients with r/r diffuse large B‐cell lymphoma (DLBCL) who received tisagenlecleucel (tisa‐cel) at Kyoto University Hospital between December 2019 and December 2023. Final follow‐up was performed in April 2024. The diagnosis of DLBCL was based on the WHO classification of tumours of haematopoietic and lymphoid tissues (revised 4th edition). 6 This study was approved by the Institutional Review Board of Kyoto University and was conducted according to the principles of the Declaration of Helsinki. Written informed consent was obtained from all patients.
Cell collection and tisagenlecleucel manufacturing
Autologous peripheral blood mononuclear cell (PBMNC) concentrates were collected using Spectra Optia (Terumo BCT, Tokyo, Japan) with the CMNC program. PBMNC concentrates were cryopreserved and shipped to manufacturing facilities for CAR‐T cell production. CD3+ cell counts in peripheral blood and in the collection bag were measured by flow cytometry after staining with fluorochrome‐conjugated anti‐CD3 antibody (clone SK7; BD Biosciences, San Jose, CA). All processes from cell collection to shipment were performed according to procedural guidelines provided by Novartis.
Tisa‐cel was manufactured by Novartis, as described elsewhere. 7 In brief, on the manufacturing start day (day 0), leukapheresis material was thawed, followed by cell wash to remove the cryomedium. T cells were selected and activated using anti‐CD3/CD28 antibody‐coated paramagnetic beads, followed by transduction with a self‐inactivating lentiviral vector containing the anti‐CD19 CAR transgenes. Following transduction, excess vector and other residuals were washed from the culture, and cells were expanded in static culture and then in bioreactors. Starting from day 0, cell expansion continued ex vivo until there were sufficient cells to meet the final product dose requirements.
Definition of poor proliferation during manufacturing
The number of viable cells was measured on day 0 and every day from day 3 until the end of cell expansion. Poor cell growth during manufacturing was defined as a decrease in cell number below that of the previous day from day 4 to the final day of cell expansion. Conversely, good cell growth was defined as continuous increases in cell count throughout this period.
End‐points and variables
The primary end‐point of this study was progression‐free survival (PFS). Secondary end‐points included overall survival (OS), disease progression, non‐relapse mortality (NRM), best response rates and incidence of cytokine‐releasing syndrome (CRS), neurological adverse events, prolonged cytopenia and hypogammaglobulinaemia. PFS was defined as the time from CAR‐T‐cell infusion to the date of documented disease progression, relapse, death or the last date of follow‐up. OS was calculated from the date of CAR‐T‐cell infusion to death or the last date of follow‐up. NRM was defined as patients who died of causes unrelated to lymphoma relapse/progression. Prolonged neutropenia was defined as neutropenia <500/mm3 or requiring G‐CSF administration persisting for >3 weeks after CAR‐T infusion, prolonged anaemia was defined as anaemia with haemoglobin <7.0 g/dL or necessitating RBC transfusion persisting for >3 weeks after infusion, and prolonged thrombocytopenia was defined as thrombocytopenia <30 000/μL or requiring platelet transfusion persisting for >3 weeks after infusion. 8 Hypogammaglobulinaemia was defined as a decrease in serum IgG levels to below 400 mg/dL or administration of immunoglobulin preparations. Disease status at apheresis, at infusion and after infusion was assessed by fluorodeoxyglucose positron emission tomography/computed tomography scan using the Revised Response Criteria for Malignant Lymphoma. 9 Progression of relapse was defined based on morphological and clinical evidence of disease activity. Cell‐of‐origin classification was determined by immunostaining of CD10, BCL6 and MUM1 based on the Hans classifier. 10 Protein expression of CD5, BCL2 or cMYC was evaluated by immunostaining. Cytogenetic abnormalities were assessed by G‐banding and fluorescence in situ hybridization. All variables shown in tables and figures were retrospectively obtained from patient records.
Statistical analyses
Continuous variables were summarized using medians and ranges, and categorical variables were summarized as counts and percentages. For baseline patient characteristics, nominal categorical variables were assessed using Fisher's exact test. Ordinal categorical variables, including disease stage, were compared using the Cochran–Armitage trend test. Continuous variables were compared using the Mann–Whitney U‐test. Increase in cell number during manufacturing and increase in lymphocyte counts in peripheral blood after CAR‐T‐cell infusion were compared between groups, using two‐way analysis of variance (ANOVA) test. Best response was compared between groups using the Cochran–Armitage trend test. Probabilities of OS and PFS were estimated according to the Kaplan–Meier method and compared among groups with the Cox proportional‐hazards model. Probabilities of disease progression and NRM were estimated on the basis of cumulative incidence methods and compared among groups with the Fine‐Gray proportional‐hazards model, 11 considering death without relapse/progression as a competing event for disease progression, disease progression/relapse as a competing event for NRM. For PFS, OS and disease progression, multivariate models were constructed. Variables of known clinical relevance (disease stage at infusion) and variables of interest (cell growth during manufacturing) were mandatory included in all models regardless of univariate significance. Other exploratory variables with p < 0.1 in univariate analysis were evaluated through stepwise inclusion and exclusion, and the final models were selected based on the lowest Bayesian information criterion (BIC) value. p‐values <0.05 were considered statistically significant. All statistical analyses were performed using Stata software version 18 (Stata Corp., College Station, TX, USA) and GraphPad Prism software version 10 (GraphPad Software, San Diego, CA, USA).
RESULTS
Patient characteristics
Of 86 patients with r/r DLBCL who underwent apheresis for tisa‐cel, one was excluded because manufacturing was terminated due to disease progression with declining performance status, and 10 were excluded because tisa‐cel was not infused (manufacturing failure, 4; aggregates in product, 1; disease progression with declining performance status, 5) (Figure S1). As a result, the analysis cohort consisted of 75 patients, including those transformed from indolent B‐cell lymphomas (n = 19) (Table 1). The median age at apheresis was 60 years (range, 20–75). Cases included 37 male (49.3%) and 38 female patients (50.7%). Immunohistochemistry revealed that 24 cases (32.0%) of DLBCL expressed both BCL2 and cMYC (double‐expressor). There were 19 cases (25.3%) of primary refractory disease. The median number of chemotherapy regimens before apheresis was 3 (1–12), and 22 patients (29.3%) had received high‐dose chemotherapy with autologous stem cell transplantation (ASCT). At the time of apheresis, disease status was as follows: complete remission (CR) in two cases (2.7%), partial remission (PR) in 31 (41.3%), stable disease (SD) in 24 (32.0%) and progressive disease (PD) or relapsed disease in 18 (24.0%).
TABLE 1.
Patient characteristics.
| Total, N = 75 | Poor growth, N = 21 | Good growth, N = 54 | p‐value | |
|---|---|---|---|---|
| Age at apheresis (year) | 60 (20–75) | 65 (20–75) | 60 (27–75) | 0.263 |
| Sex | 0.801 | |||
| Male | 37 (49.3%) | 11 (52.4%) | 26 (48.1%) | |
| Female | 38 (50.7%) | 10 (47.6%) | 28 (51.9%) | |
| Diagnosis | 0.770 | |||
| DLBCL | 56 (74.7%) | 15 (71.4%) | 41 (75.9%) | |
| Transformed from iNHL | 19 (25.3%) | 6 (28.6%) | 13 (24.1%) | |
| Hans class | 0.375 | |||
| GCB | 32 (42.7%) | 7 (33.3%) | 25 (46.3%) | |
| Non‐GCB | 36 (48.0%) | 13 (61.9%) | 23 (42.6%) | |
| CD5 expression | 17 (22.7%) | 7 (33.3%) | 10 (18.5%) | 0.176 |
| Double expressor (BCL2/cMYC) | 24 (32.0%) | 5 (23.8%) | 19 (35.2%) | 0.241 |
| Double hit (BCL2/cMYC) | 4 (5.3%) | 0 (0.0%) | 4 (7.4%) | 0.137 |
| Primary refractory | 19 (25.3%) | 9 (42.9%) | 10 (18.5%) | 0.040* |
| No. of chemotherapy regimens before apheresis | 3 (1–12) | 4 (2–12) | 3 (1–8) | 0.081 |
| ASCT before apheresis | 22 (29.3%) | 5 (23.8%) | 17 (31.5%) | 0.583 |
| Bendamustine before apheresis | 14 (18.7%) | 4 (19.0%) | 10 (18.5%) | 1.000 |
| No. of bendamustine cycles before apheresis | 0 (0–12) | 0 (0–12) | 0 (0–6) | 0.864 |
| Stage at apheresis | 0.399 | |||
| CR | 2 (2.7%) | 1 (4.8%) | 1 (1.9%) | |
| PR | 31 (41.3%) | 5 (23.8%) | 26 (48.1%) | |
| SD | 24 (32.0%) | 10 (47.6%) | 14 (25.9%) | |
| PD/Rel | 18 (24.0%) | 5 (23.8%) | 13 (24.1%) | |
| Month between diagnosis and apheresis | 16.9 (2.0–254.9) | 13.6 (4.2–109.4) | 18.5 (2.0–254.9) | 0.480 |
| Stage at infusion | 0.242 | |||
| CR | 10 (13.3%) | 2 (9.5%) | 8 (14.8%) | |
| PR | 26 (34.7%) | 6 (28.6%) | 20 (37.0%) | |
| SD | 20 (26.7%) | 6 (28.6%) | 14 (25.9%) | |
| PD | 19 (25.3%) | 7 (33.3%) | 12 (22.2%) | |
| Extranodal lesion at infusion | 36 (48.0%) | 11 (52.4%) | 25 (46.3%) | 0.797 |
| No. of bridging chemotherapy regimens | 1 (0–4) | 1 (0–2) | 1 (0–4) | 0.254 |
| Bridging radiotherapy | 10 (13.3%) | 3 (14.3%) | 7 (13.0%) | 1.000 |
| Months between apheresis to infusion | 2.3 (1.5–31.7) | 2.3 (1.6–4.5) | 2.3 (1.5–31.7) | 0.998 |
Abbreviations: ASCT, autologous stem cell transplantation; CR, complete response; DLBCL, diffuse large B‐cell lymphoma; GCB, germinal centre B‐cell like; iNHL, indolent non‐Hodgkin lymphoma; PD, progressive disease; PR, partial response; Rel, relapsed disease; SD, stable disease.
p < 0.05.
Cell counts in peripheral blood at apheresis were as follows: haemoglobin, 10.1 g/dL (range, 6.8–15.2); platelets, 18.9 × 104/μL (range, 3.0–41.8); lymphocytes, 681/μL (range, 99–4117); CD3+ cells, 576/μL (range, 60–3005) (Table 2). A median of 3.6 × 109 (range, 0.7–9.6) and 3.1 × 109 (range, 0.7–7.7) CD3+ cells were harvested and shipped for CAR‐T‐cell production (Table S1).
TABLE 2.
Laboratory data at apheresis.
| Total, N = 75 | Poor growth, N = 21 | Good growth, N = 54 | p‐value | |
|---|---|---|---|---|
| Haemoglobin (g/dL) | 10.1 (6.8–15.2) | 9.0 (6.8–12.5) | 10.5 (8.3–15.2) | <0.001* |
| Platelet counts (104/μL) | 18.9 (3.0–41.8) | 15.2 (3.0–41.8) | 19.8 (5.2–39.3) | 0.092 |
| WBC counts (/μL) | 3710 (1130–11 120) | 3380 (1560–11 120) | 4340 (1130–9920) | 0.047* |
| Monocyte counts (/μL) | 460 (140–1503) | 419 (140–1334) | 496 (187–1503) | 0.223 |
| Lymphocyte counts (/μL) | 681 (99–4117) | 608 (187–1390) | 887 (99–4117) | 0.300 |
| CD3+ cell counts (/μL) | 576 (60–3005) | 469 (177–1290) | 719 (60–3005) | 0.168 |
| CD4+ cell counts (/μL) | 217 (8–690) | 164 (8–605) | 229 (19–690) | 0.334 |
| CD8+ cell counts (/μL) | 303 (26–2334) | 226 (73–960) | 382 (26–2334) | 0.147 |
| CD4/CD8 ratio | 0.545 (0.049–3.474) | 0.619 (0.049–3.474) | 0.500 (0.127–2.277) | 0.704 |
| LDH (U/L) | 225 (138–613) | 216 (168–613) | 228 (138–575) | 0.736 |
Abbreviations: LDH, lactate dehydrogenase; WBC, white blood cell.
p < 0.05.
The median time from apheresis to CAR‐T‐cell infusion was 2.3 months (range, 1.5–31.7). Disease status at the time of CAR‐T‐cell infusion was CR in 10 cases (13.3%), PR in 26 (34.7%), SD in 20 (26.7%) and PD in 19 (25.3%). Bridging treatments between apheresis and infusion included a median of one line of chemotherapy (range, 0–4) and radiation therapy in 10 cases (13.3%).
Cell growth during manufacturing
The extent of cell proliferation during manufacturing varied significantly between cases (Figure S2). The median number of seeded cells at the start of culture on day 0 was 2.05 × 108 (range, 0.61–4.78). Although there was a decreasing trend in cell number by day 3, cell counts began to increase after day 4. By the final day of culture, median cell counts reached 66.4 × 108 (range, 15.7–158), which represented a median 34.64‐fold increase (range, 4.06–164.16) compared to the number of cells seeded at day 0 (Table S2).
Of 75 cases, 21 (28.0%) experienced poor cell proliferation during manufacturing (poor growth group), whereas 54 (72.0%) exhibited good cell proliferation (good growth group). Among the poor growth group, in 12 cases (57.1%), a decrease in cell number was observed at day 4 of culture, and by day 7 of culture, 18 of 21 cases in the poor growth group (85.7%) showed at least one decrease in cell number, suggesting that poor growth is determined early in the culture process. The poor growth group showed a reduced increase in cell number during manufacturing (Figure 1; Table S2).
FIGURE 1.
Increase in cell numbers during manufacturing process. Cell counts are plotted starting from the day of incubation initiation (Day 0). Dots represent the median cell counts at each time point for each group, and the light shading depicts the 95% confidence intervals (CIs) for the medians at each time point. * indicates p < 0.05, as determined by a two‐way analysis of variance (ANOVA) test.
Comparison of characteristics in the good and poor growth groups
Then, we compared clinical factors between the good (n = 54) and poor growth group (n = 21) (Tables 1 and 2; Table S1). Median age at tisa‐cel infusion was similar in the two groups (median, 60 vs. 65 years; p = 0.263). The proportion of patients with primary refractory disease was lower in the good growth group than in the poor growth group (18.5% vs. 42.9%; p = 0.040). There were no significant differences in other patient backgrounds between the two groups. The proportion of patients with a history of ASCT was 31.5% vs. 23.8% (p = 0.583). While haemoglobin concentration (10.5 vs. 9.0 g/dL, p < 0.001) and white blood cell counts (4340 vs. 3380/μL, p = 0.047) at apheresis were higher in the good growth group, there were no significant differences in other blood cell counts at apheresis or in variables of apheresis procedures between the groups. At the time of infusion, disease status was CR in 14.8% vs. 9.5%, PR in 37.0% vs. 28.6%, SD in 25.9% vs. 28.6% and PD in 22.2% vs. 33.3% (p = 0.242). Duration from apheresis to infusion did not differ between these groups (2.3 vs. 2.3 months; p = 0.998). Regarding bridging therapies administered between apheresis and infusion, no significant differences were observed between the two groups in terms of numbers of chemotherapy regimens or proportions of patients who received radiation therapy.
Impact of cell proliferation during manufacturing on efficacy of CAR‐T‐cell therapy
Next, we evaluated the impact of cell proliferation during manufacturing on the efficacy of CAR‐T‐cell therapy. Good proliferation correlated strongly with better lymphocyte increase in peripheral blood after tisa‐cel infusion (median, 1200 vs. 570/μL at day 21; p = 0.024, two‐way analysis of variance test) (Figure 2A). Regarding the best response after tisa‐cel infusion, ORR was significantly higher in the good growth group than the poor growth group (85.2% vs. 52.4%; p = 0.006), as rates of CR were 61.1% vs. 38.1%, and PR rates were 24.1% vs. 14.3% respectively (Figure 2B; Table 3). For post‐infusion therapy, consolidation or maintenance therapy was administered in six patients (8.0%) for reasons other than relapse or progression, and treatments for relapse or progression were administered in 31 (41.3%); however, no significant differences were observed between the two groups in these proportions (Table S3).
FIGURE 2.
Comparison of treatment response between the groups. (A) Increases in circulating lymphocyte counts after CAR‐T‐cell infusion are compared between the groups. Lymphocyte counts in peripheral blood are plotted starting from the day of CAR‐T‐cell infusion (Day 0). Dots represent the median lymphocyte counts at each time point for each group, and the bars depict the 95% confidence intervals (CIs) for the medians at each time point. (B) Best response rates are compared between the groups. Overall response rates (ORR), complete response (CR) rates and partial response (PR) rates in each group are shown. * indicates p < 0.05, as determined by two‐way analysis of variance (ANOVA) test (A) or Fisher's exact test (B).
TABLE 3.
Treatment outcomes following tisagenlecleucel.
| Total, N = 75 | Poor growth, N = 21 | Good growth, N = 54 | p‐value | |
|---|---|---|---|---|
| Best response | ||||
| Overall response | 57 (76.0%) | 11 (52.4%) | 46 (85.2%) | 0.006* |
| CR | 41 (54.7%) | 8 (38.1%) | 33 (61.1%) | 0.010* |
| PR | 16 (21.3%) | 3 (14.3%) | 13 (24.1%) | |
| SD | 4 (5.3%) | 3 (14.3%) | 1 (1.9%) | |
| PD | 14 (18.7%) | 7 (33.3%) | 7 (13.0%) | |
| CRS | ||||
| Any grade | 66 (88.0%) | 17 (81.0%) | 49 (90.7%) | 0.258 |
| Grade 3–4 | 2 (2.6%) | 0 (0.0%) | 2 (3.8%) | 1.000 |
| Neurological adverse events | ||||
| Any grade | 3 (4.0%) | 1 (4.8%) | 2 (3.7%) | 0.418 |
| Grade 3–4 | 1 (1.3%) | 1 (4.8%) | 0 (0.0%) | 0.280 |
| Prolonged cytopenia | ||||
| Any | 40 (53.3%) | 11 (52.4%) | 29 (53.7%) | 1.000 |
| Neutropenia | 38 (50.7%) | 11 (52.4%) | 27 (50.0%) | 1.000 |
| Anaemia | 19 (25.3%) | 7 (33.3%) | 12 (22.2%) | 0.380 |
| Thrombocytopenia | 31 (41.3%) | 9 (42.9%) | 22 (40.7%) | 1.000 |
| Hypogammaglobulinaemia | 46 (61.3%) | 13 (61.9%) | 33 (61.1%) | 1.000 |
Abbreviations: CR, complete response; CRS, cytokine‐releasing syndrome; PD, progressive disease; PR, partial response; SD, stable disease.
p < 0.05.
We compared PFS, OS and disease progression between groups in univariate and multivariate analyses (Table 4; Table S4). Multivariate analysis showed that the good growth group had significantly better 1‐year PFS (46.7% vs. 26.2%; adjusted hazard ratio [aHR], 0.377; 95% confidence interval [CI], 0.180–0.789; p = 0.010) and OS (79.3% vs. 52.7%; aHR, 0.191; 95% CI, 0.071–0.510; p = 0.001) with lower cumulative incidence of disease progression (51.0% vs. 73.8%; aHR, 0.412; 95% CI, 0.198–0.858; p = 0.018) than the poor growth group (Figure 3A−C; Table 4).
TABLE 4.
Multivariate analyses for outcomes.
| PFS | OS | Disease progression | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | p‐value | HR | 95% CI | p‐value | HR | 95% CI | p‐value | ||
| Cell growth during manufacturing | Good vs. poor | 0.377 | (0.180–0.789) | 0.010* | 0.191 | (0.071–0.510) | 0.001* | 0.412 | (0.198–0.858) | 0.018* |
| Double expressor (BCL2/cMYC) | Yes vs. no | 1.829 | (0.891–3.754) | 0.100 | 4.260 | (1.635–11.103) | 0.003* | 1.827 | (0.887–3.764) | 0.102 |
| Stage at infusion | SD/PD vs. CR/PR | 1.889 | (0.938–3.804) | 0.075 | 7.957 | (2.273–27.855) | 0.001* | 2.035 | (1.035–4.003) | 0.040* |
| No. of bridging chemotherapy regimens | ≥2 vs. 0–1 | 2.455 | (1.207–4.993) | 0.013* | 2.521 | (1.055–6.024) | 0.038* | 2.432 | (1.140–5.185) | 0.021* |
Abbreviations: CI, confidence interval; CR, complete response; HR, hazard ratio; OS, overall survival; PD, progressive disease; PFS, progression‐free survival; PR, partial response; SD, stable disease.
p < 0.05.
FIGURE 3.
Comparison of outcomes according to cell growth during manufacturing. (A) Progression‐free survival (PFS). (B) Overall survival (OS). (C) Disease progression. aHR, adjusted hazard ratio; CI, confidence interval. * indicates p < 0.05.
For adverse events, there was no significant difference in the frequency of CRS (90.7% vs. 81.0%; p = 0.258), neurological events (3.7% vs. 4.8%; p = 0.418), prolonged cytopenia (53.7% vs. 52.4%; p = 1.000) or hypoimmunoglobulinaemia (61.1% vs. 61.9%; p = 0.258) between the two groups (Table 3). Death without disease progression was rare, and no significant difference was observed between the two groups (Figure S3; Table S5).
DISCUSSION
We evaluated the influence of ex vivo cell proliferation during CAR‐T‐cell manufacturing on clinical outcomes following CD19‐targeted CAR‐T‐cell therapy for r/r B‐cell lymphomas and found that good cell proliferation after CAR transduction is a significant predictor of favourable prognosis after CAR‐T infusion.
First, we found that, even in cases in which in‐specification tisa‐cel products were manufactured, consistent cell proliferation was not achieved during manufacturing in 28% of the cases, resulting in low final cell numbers. By day 7 of culture, the majority of poor growth cases exhibited at least one decrease in cell number, suggesting that poor proliferation is determined early in the cell culture process. These results suggest that while CAR‐T‐cell manufacturing has been optimized since clinical trials, 7 the degree of cell proliferation varies significantly among individual cases in real‐world practice.
We explored the clinical background associated with good proliferation. There were fewer patients with primary induction failure and higher haemoglobin concentrations and white blood cell counts at the time of apheresis in the good growth group, compared to the poor growth group. These factors may be associated with maintained fitness of autologous T cells due to mechanisms such as less bone marrow suppression caused by chemotherapy. However, because no significant differences were observed in other variables potentially associated with T‐cell fitness, including age, number of chemotherapy regimens and disease status before apheresis, these observations need to be confirmed in a larger investigation.
In our previous multicentre study, 12 we reported that repeated use of bendamustine with a short washout period before apheresis, thrombocytopenia and a low CD4/CD8 ratios were risk factors for manufacturing failure, and it was hypothesized that these factors could affect cell proliferation during manufacturing. However, in this current analysis, no significant differences were found between the poor growth and good growth groups regarding these factors. It may be attributed to the modest differences in clinical factors among patients, because this single‐centre cohort only included cases that reached CAR‐T‐cell infusion, excluding cases of manufacturing failure or those who were unable to receive infusion due to disease progression. Actually, among the 10 cases in which manufacturing was attempted but did not reach infusion (excluded from the current analysis cohort), four showed poor growth, all of which resulted in manufacturing failure (Figure S1). When including cases that did not proceed to infusion, poor growth group cases tended to have higher numbers of chemotherapy regimens before apheresis and lower platelet counts (data not shown). To predict cell proliferation more accurately before attempted manufacturing, it is crucial to develop methods that evaluate T‐cell function based on biological mechanisms, as well as clinical factors.
Next, we evaluated the impact of cell proliferation during manufacturing on clinical outcomes after CAR‐T‐cell infusion. Our results indicate a direct correlation between cell proliferation and circulating lymphocyte increases in peripheral blood after infusion, which is reportedly a marker for robust anti‐tumour effects. 5 , 13 Moreover, we observed a clear distinction in clinical outcomes between patients exhibiting good versus poor cell growth during CAR‐T‐cell manufacturing. Notably, patients in the good growth group demonstrated significantly better PFS and OS, as well as response rates, compared to those in the poor growth group. In addition to the minimal differences in patient characteristics between the groups, good growth remained a significant prognostic factor even after conducting multivariate analysis that included known prognostic factors. These results suggest that the biological behaviour of CAR‐T cells during manufacturing may independently influence patient prognosis.
Factors influencing prognosis after CAR‐T‐cell therapy are being explored in real‐world settings. While previous reports have suggested various factors related to tumour intrinsic factors such as genetic alterations, factors related to host inflammatory state such as serum ferritin levels and factors related to disease controls such as LDH levels and metabolic tumour volume, 2 , 14 , 15 , 16 , 17 , 18 , 19 , 20 there are only a few reports on factors related to quality and potency of CAR‐T cells. Recently, the importance of maintaining CAR‐T‐cell potency, as well as autologous T‐cell function, has been pointed out, as a previous report suggested that the use of bendamustine prior to apheresis may impair T‐cell function consequently compromising therapeutic efficacy. 4 In this context, the findings of this study, suggesting that metrics in the manufacturing process can predict therapeutic outcomes, provide the following implications. First, metrics during the manufacturing can be used as early predictors of therapeutic effects of CAR‐T‐cell therapies, potentially allowing more tailored treatment strategies for each patient. For instance, these metrics could be used to decide whether to initiate consolidation or maintenance therapies, such as bispecific antibodies or other targeted therapies after CAR‐T‐cell infusion. Second, the approach of maintaining good proliferation during manufacturing by harvesting autologous T cells at the optimal time for their function may enhance therapeutic efficacy.
While this study includes detailed analyses of real‐world data, some limitations exist. First, its retrospective design and the single‐centre nature of the study may restrict the generalizability of findings. Second, in addition to the possibility of errors in cell counting during the manufacturing process, the definition of good versus poor growth was based on a specific time window and this threshold that may not apply universally to all CAR‐T manufacturing protocols or patient populations. Future studies involving multicentre trials and standardized proliferation metrics are needed to validate and refine our findings. Third, while we hypothesize that higher proliferation rates during manufacturing may correlate with robust proliferation and persistence of CAR‐T cells in vivo after infusion, biological mechanistic insights into how different proliferation rates affect the functional capability of CAR‐T cells were not analysed in this study. Direct evidence from cellular and molecular analyses would be crucial to confirm this hypothesis. Additionally, since novel CAR‐T‐cell therapies are under development that incorporate modifications of the manufacturing process to preserve T‐cell stemness and enhance efficacy, 21 it is necessary to investigate whether manufacturing metrics of these novel therapies are associated with post‐infusion outcomes.
In conclusion, this study highlights the potential of using information on cell proliferation during manufacturing that is available well before infusion, as predictive markers for clinical outcomes in patients treated with CAR‐T therapy for r/r B‐cell lymphomas. Identifying clinical factors that influence manufacturing parameters and using a potency‐guided, personalized approach would improve post‐CAR‐T outcomes.
AUTHOR CONTRIBUTIONS
TJ and YA designed the study, reviewed and analysed data and wrote the paper; T Shimizu, TK, T Sakamoto, CM, JK, M Nishikori, KY, M Nagao and ATK interpreted data. All authors critically reviewed the draft and approved the final version of the manuscript.
FUNDING INFORMATION
The authors declare no competing interests of any sort. This work was supported in part by MEXT to YA.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
ETHICS STATEMENT
This study was approved by the Institutional Review Board and Ethic Committee of Kyoto University (G0697).
PATIENT CONSENT STATEMENT
Written informed consent was obtained from all participating patients.
Supporting information
Data S1.
ACKNOWLEDGEMENTS
We thank Dr. Shin Kawamata (Cyto‐Facto Inc.) for insightful discussions. We also express our gratitude to Novartis for providing CAR‐T manufacturing metrics. We are grateful to Team CAR‐T members at Kyoto University Hospital for their dedicated care of patients. This work was supported, in part, by the Program for Development of Next‐generation Leading Scientists with Global Insight (L‐INSIGHT), sponsored by the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan to YA.
Jo T, Arai Y, Shimizu T, Kitawaki T, Sakamoto T, Mizumoto C, et al. Consistent ex vivo cell proliferation during manufacturing predicts favourable outcomes post‐CAR‐T‐cell therapy. Br J Haematol. 2025;207(3):1002–1010. 10.1111/bjh.20266
DATA AVAILABILITY STATEMENT
Data employed in this study are not publicly available due to ethical restrictions. Releasing the data would exceed the scope of patient's consent for research use.
REFERENCES
- 1. Iacoboni G, Villacampa G, Martinez‐Cibrian N, Bailén R, Lopez Corral L, Sanchez JM, et al. Real‐world evidence of tisagenlecleucel for the treatment of relapsed or refractory large B‐cell lymphoma. Cancer Med. 2021;10(10):3214–3223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Nastoupil LJ, Jain MD, Feng L, Spiegel JY, Ghobadi A, Lin Y, et al. Standard‐of‐care axicabtagene ciloleucel for relapsed or refractory large B‐cell lymphoma: results from the US lymphoma CAR T consortium. J Clin Oncol. 2020;38(27):3119–3128. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Schuster SJ, Tam CS, Borchmann P, Worel N, McGuirk JP, Holte H, et al. Long‐term clinical outcomes of tisagenlecleucel in patients with relapsed or refractory aggressive B‐cell lymphomas (JULIET): a multicentre, open‐label, single‐arm, phase 2 study. Lancet Oncol. 2021;22(10):1403–1415. [DOI] [PubMed] [Google Scholar]
- 4. Iacoboni G, Navarro V, Martin‐Lopez AA, Rejeski K, Kwon M, Jalowiec KA, et al. Recent bendamustine treatment before apheresis has a negative impact on outcomes in patients with large B‐cell lymphoma receiving chimeric antigen receptor T‐cell therapy. J Clin Oncol. 2024;42(2):205–217. [DOI] [PubMed] [Google Scholar]
- 5. Wada F, Jo T, Arai Y, Kitawaki T, Mizumoto C, Kanda J, et al. T‐cell counts in peripheral blood at leukapheresis predict responses to subsequent CAR‐T cell therapy. Sci Rep. 2022;12(1):18696. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Swerdlow SHCE, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J. WHO classification of tumors of hematopoietic and lymphoid tissues (revised 4th edition). Lyon Cedex, France: International Agency for Research on Cancer (IARC); 2017. [Google Scholar]
- 7. Tyagarajan S, Spencer T, Smith J. Optimizing CAR‐T cell manufacturing processes during pivotal clinical trials. Mol Ther Methods Clin Dev. 2020;16:136–144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Nakamura N, Jo T, Arai Y, Kitawaki T, Nishikori M, Mizumoto C, et al. Clinical impact of cytokine release syndrome on prolonged hematotoxicity after chimeric antigen receptor T cell therapy: KyoTox a‐score, a novel prediction model. Transplant Cell Ther. 2024;30(4):404–414. [DOI] [PubMed] [Google Scholar]
- 9. Cheson BD, Pfistner B, Juweid ME, Gascoyne RD, Specht L, Horning SJ, et al. Revised response criteria for malignant lymphoma. J Clin Oncol. 2007;25(5):579–586. [DOI] [PubMed] [Google Scholar]
- 10. Hans CP, Weisenburger DD, Greiner TC, Gascoyne RD, Delabie J, Ott G, et al. Confirmation of the molecular classification of diffuse large B‐cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103(1):275–282. [DOI] [PubMed] [Google Scholar]
- 11. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446):496–509. [Google Scholar]
- 12. Jo T, Yoshihara S, Okuyama Y, Fujii K, Henzan T, Kahata K, et al. Risk factors for CAR‐T cell manufacturing failure among DLBCL patients: a nationwide survey in Japan. Br J Haematol. 2023;202(2):256–266. [DOI] [PubMed] [Google Scholar]
- 13. Faude S, Wei J, Muralidharan K, Xu X, Wertheim G, Paessler M, et al. Absolute lymphocyte count proliferation kinetics after CAR T‐cell infusion impact response and relapse. Blood Adv. 2021;5(8):2128–2136. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Goto H, Kitawaki T, Fujii N, Kato K, Onishi Y, Fukuhara N, et al. Safety and efficacy of tisagenlecleucel in patients with relapsed or refractory B‐cell lymphoma: the first real‐world evidence in Japan. Int J Clin Oncol. 2023;28(6):816–826. [DOI] [PubMed] [Google Scholar]
- 15. Jacobson CA, Hunter BD, Redd R, Rodig SJ, Chen PH, Wright K, et al. Axicabtagene ciloleucel in the non‐trial setting: outcomes and correlates of response, resistance, and toxicity. J Clin Oncol. 2020;38(27):3095–3106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Jain MD, Ziccheddu B, Coughlin CA, Faramand R, Griswold AJ, Reid KM, et al. Whole‐genome sequencing reveals complex genomic features underlying anti‐CD19 CAR T‐cell treatment failures in lymphoma. Blood. 2022;140(5):491–503. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Kwon M, Iacoboni G, Reguera JL, Corral LL, Morales RH, Ortiz‐Maldonado V, et al. Axicabtagene ciloleucel compared to tisagenlecleucel for the treatment of aggressive B‐cell lymphoma. Haematologica. 2023;108(1):110–121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Scholler N, Perbost R, Locke FL, Jain MD, Turcan S, Danan C, et al. Tumor immune contexture is a determinant of anti‐CD19 CAR T cell efficacy in large B cell lymphoma. Nat Med. 2022;28(9):1872–1882. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Shouval R, Alarcon Tomas A, Fein JA, Flynn JR, Markovits E, Mayer S, et al. Impact of TP53 genomic alterations in large B‐cell lymphoma treated with CD19‐chimeric antigen receptor T‐cell therapy. J Clin Oncol. 2022;40(4):369–381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Vercellino L, Di Blasi R, Kanoun S, Tessoulin B, Rossi C, D'Aveni‐Piney M, et al. Predictive factors of early progression after CAR T‐cell therapy in relapsed/refractory diffuse large B‐cell lymphoma. Blood Adv. 2020;4(22):5607–5615. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Dickinson MJ, Barba P, Jager U, Shah NN, Blaise D, Briones J, et al. A novel autologous CAR‐T therapy, YTB323, with preserved T‐cell stemness shows enhanced CAR T‐cell efficacy in preclinical and early clinical development. Cancer Discov. 2023;13(9):1982–1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1.
Data Availability Statement
Data employed in this study are not publicly available due to ethical restrictions. Releasing the data would exceed the scope of patient's consent for research use.