Real-World Outcomes of Tisagenlecleucel Versus Standard-of-Care Salvage Chemotherapy in Relapsed or Refractory Diffuse Large B-Cell Lymphoma: A Propensity Score-Matched Analysis From Korea.

INTRODUCTION

Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of non-Hodgkin lymphoma, accounting for approximately 30–40% of all cases globally and Korea. 1 2 Although first-line rituximab-based chemoimmunotherapy induces durable remission in many patients, 30–40% experience refractory disease or relapse. 3 4 5 The traditional standard-of-care (SOC) for patients with relapsed or refractory (R/R) DLBCL has been salvage chemotherapy followed by high-dose therapy and autologous stem cell transplantation (ASCT), but a substantial number of patients are either ineligible for ASCT or fail to achieve a durable response after transplantation. 6 7 8 The prognosis for patients who relapse after ASCT or are refractory to chemotherapy is particularly poor, with a reported median overall survival (OS) of less than 6 to 12 months, underscoring the urgent need for more effective treatments in this setting. 9 10

To address this unmet medical need, CD19-targeted chimeric antigen receptor T-cell (CAR-T) therapy has emerged as a transformative treatment option for R/R DLBCL. Tisagenlecleucel, an anti-CD19 CAR-T therapy, demonstrated durable responses in the single-arm phase 2 JULIET trial, which enrolled patients with R/R DLBCL after at least 2 prior lines of therapy, leading to its regulatory approval in the third-line (3L) or later setting. 11 12 In contrast, the phase 3 BELINDA trial evaluated tisagenlecleucel in the second-line (2L) setting for patients with primary refractory disease or relapse within 12 months after completion of first-line therapy, but did not show an improvement in event-free survival (EFS) compared with SOC chemoimmunotherapy followed by ASCT. 13 These findings suggest that outcomes with tisagenlecleucel may vary depending on treatment line, patient population, and logistical factors such as time to infusion and use of bridging therapy.

Several large real-world studies, including those from the Center for International Blood and Marrow Transplant Research and European centers, have evaluated tisagenlecleucel in the ≥ 3L setting, confirming its efficacy and safety in routine practice. 14 15 However, most were single-arm analyses without a contemporaneous SOC control group, limiting the ability to quantify the relative benefit of CAR-T therapy. Moreover, data from Asian populations remain scarce, and differences in treatment patterns, patient characteristics, and healthcare delivery systems may influence outcomes compared with Western cohorts. This is particularly relevant in Korea, where tisagenlecleucel is currently the only commercially available CAR-T therapy for DLBCL and thus represents the sole cellular therapy option in routine practice. Differences in product availability, healthcare infrastructure, and supportive care resources between Korea and western countries underscore the importance of region-specific real-world evidence.

To address these gaps, we conducted a retrospective, registry-based, propensity score-matched cohort study at two tertiary university hospitals in Korea to directly compare the clinical outcomes of tisagenlecleucel vs. SOC therapies in patients with R/R DLBCL treated in the ≥ 3L setting.

METHODS

Study design and data source

This was a retrospective, registry-based cohort study conducted at two tertiary university hospitals in Korea: Seoul St. Mary’s Hospital and Yeouido St. Mary’s Hospital, both affiliated with the Catholic University of Korea. All consecutive patients newly diagnosed with DLBCL between December 1, 2002 and December 31, 2024 were identified from institutional lymphoma registries. These registries are prospectively maintained databases that capture demographic, clinical, pathological, treatment, and survival data for all patients diagnosed with lymphoma at the participating institutions. Clinical and laboratory data were extracted from the registry and cross-checked against electronic medical records to ensure completeness and accuracy.

Patient selection

Between December 2002 and December 2024, a total of 1,575 patients were newly diagnosed with DLBCL at the participating institutions. Patients with an unclear diagnosis of DLBCL or with transformed DLBCL from indolent lymphoma were excluded at the outset (n = 49 for transformed DLBCL; n = 2 for unclear diagnosis), leaving 1,524 patients eligible for screening. Of these, 77 patients received tisagenlecleucel as 3L or later therapy (CAR-T cohort candidates), while the remaining 1,447 patients did not receive CAR-T therapy (SOC cohort candidates). For the purpose of this study, SOC therapy was defined as salvage chemoimmunotherapy regimens commonly used for relapsed or refractory DLBCL in Korea during the study period (e.g., R-dexamethasone, cytarabine, and cisplatin [DHAP], R-ifosfamide, carboplatin, and etoposide [ICE], gemcitabine, dexamethasone, and cisplatin [GDP], or gemcitabine-based combinations), with or without subsequent high-dose therapy and ASCT. In the CAR-T cohort, we excluded any patients who received 3L or later therapy involving investigational or non-standard agents, including bispecific antibodies (e.g., glofitamab, epcoritamab), antibody–drug conjugates (e.g., polatuzumab vedotin), other novel targeted agents, unapproved cellular therapies, or enrollment in clinical trials (n = 10). In the SOC cohort, we excluded patients who received only ≤ 2L therapy (n = 1,302), and those who received investigational or non-standard agents as defined above (n = 34). After these exclusions, 67 patients remained in the CAR-T cohort and 111 patients in the SOC cohort. A schematic of the patient selection process is provided in Fig. 1 .

Next, we performed 1:1 propensity score matching based on age, sex, year of diagnosis, line of therapy (third, fourth, or ≥ fifth), International Prognostic Index (IPI) at diagnosis, and Eastern Cooperative Oncology Group (ECOG) performance status at diagnosis, resulting in 50 patients in each group for the matched analysis.

Efficacy and safety assessment

OS was defined as the duration from the start of index therapy (index date) to death from any cause or to the date of the last follow-up. Progression-free survival (PFS) was assessed as the time from index date to confirmation of disease progression or death from any cause, whichever occurred first. The index date was defined as the infusion date for CAR-T and the initiation date of the index salvage regimen for SOC. Adverse events were graded according to the Common Terminology Criteria for Adverse Events (version 5.0). Specifically, cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) were defined and graded according to the American Society for Transplantation and Cellular Therapy criteria.

Statistical analysis

Hypothesis testing was conducted using the Mann-Whitney U test and the χ 2 test for non-parametric data. We used 1:1 propensity score matching to minimize bias due to possible confounders. Propensity scores for SOC cohort selection were determined using a logistic regression model including sex, age, line of therapy, IPI, ECOG, and diagnosis year as covariates. A caliper width equal to 0.2 of the standard deviation of the logit of the propensity score was used to improve covariate balance and reduce residual confounding. The balances of matched variables between the cohorts were evaluated using standardized mean difference (SMD). Kaplan-Meier survival analyses were conducted to estimate PFS and OS. Differences in PFS and OS rates were assessed using the log-rank test. The Cox proportional hazard model estimated hazard ratios (HRs) for PFS and OS. To ensure the quality of real-world data (RWD)-based comparative analyses and overcome biases inherent in RWD, we performed a quantitative bias analysis using E-values to assess the strength of unmeasured confounders and quantify the sensitivity of adjusted HRs to residual confounding. 16 Subgroup analyses were performed for each covariate, included in the matching process except diagnosis year. Variables demonstrating a SMD greater than 0.2 were subjected to double adjustment in the HR estimation process. Statistical significance was defined as a two-sided P value < 0.05. All statistical analyses and figure generation were performed using R statistical software version 4.4.0 (R Foundation for Statistical Computing, Vienna, Austria).

Ethics statement

The study was approved by the Institutional Review Board (IRB) of Catholic Hematology Hospital (IRB No. XC24OIDI0042), and the requirement for informed consent was waived owing to the retrospective nature of the study. All procedures were conducted in accordance with the Declaration of Helsinki.

RESULTS

Patient characteristics

Table 1 shows the baseline characteristics of our study cohort. Among the 178 eligible patients, 67 were assigned to the CAR-T cohort, with a median follow-up of 15.2 months (95% confidence interval [CI], 11.6–20.0), and 111 to the SOC cohort, with a median follow-up of 45.7 months (95% CI, 17.3–65.8). Baseline demographic characteristics were well balanced between the cohorts: the proportion of male patients was 65.7% in the CAR-T cohort and 63.1% in the SOC cohort ( P = 0.849), and the distribution of age groups showed no statistically significant difference ( P = 0.740). There were no significant between-group differences in line of therapy, ECOG, Ann-Arbor stage, and BM involvement (all P > 0.05). Several laboratory and disease-related factors significantly differed between cohorts. Patients in the CAR-T cohort had higher proportion of hemoglobin ≥ 10 g/dL (74.6% vs. 55.9%, P = 0.032) and albumin ≥ 3.5 g/dL (88.1% vs. 47.7%, P < 0.001), but lower proportion of lactate dehydrogenase above the upper limit of normal (58.2% vs. 83.8%. P < 0.001), WBC (5.81 vs. 9.79 × 10 3 /µL, P = 0.025) compared to the control cohort. After 1:1 propensity score matching, a sufficient balance was achieved for all matching variables ( Table 2 , Supplementary Fig. 1 ).

Treatment details

Among the matched CAR-T cohort, 40 patients (80%) received bridging therapy, most commonly with polatuzumab, bendamustine, and rituximab (n = 14), followed by etoposide, prednisolone, vincristine, cyclophosphamide, and doxorubicin (n = 9), bendamustine and rituximab (BR; n = 7), DHAP (n = 2), ICE (n = 1), bendamustine monotherapy (n = 1), doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD; n = 1), corticosteroids (n = 1), and radiation therapy (n = 4). Among infused patients, the infused cell count was a median 3.6 × 10 8 cells (interquartile range 2.0–5.8 × 10 8 ), with product viability a median 94.2% (range 76.9–99.6%). The vein-to-vein time was a median 40 days (range 29–68 days). Given limited power, we did not detect obvious monotonic patterns between cell count/viability and overall response rate (ORR) or survival on descriptive inspection; formal modeling was not pursued.

In the matched SOC cohort, salvage treatment regimens consisted primarily of platinum-based chemotherapy (n = 31), followed by anthracycline-based regimens (n = 11), L-asparaginase–based regimens (n = 2), BR (n = 2), ABVD (n = 2), and IVAM (ifosfamide, etoposide, cytarabine, and methotrexate; n = 2).

Response outcomes

The ORR was significantly higher in the CAR-T cohort compared with the SOC cohort (68% vs. 28%; odds ratio [OR], 5.46; 95% CI, 2.32–12.87; P < 0.001). Similarly, the complete response (CR) rate was significantly greater in the CAR-T cohort (60% vs. 22%; OR, 5.32; 95% CI, 2.21–12.77; P < 0.001).

Survival outcomes

The OS of the matched CAR-T cohort was significantly better than that of the matched SOC cohort ( Fig. 2A ): for median OS, 11.9 months (95% CI, 9.3, NA) vs. 3.3 months (95% CI, 2.3–11.2); for 1-year OS rate, 47.6% (95% CI, 34.6–65.5%) vs. 30.4% (95% CI, 18.9–48.8%), P = 0.003; for 2-year OS rate, 35.1% (95% CI, 20.0–61.8%) vs. 27.4% (95% CI, 16.3–45.9%), P = 0.006; with a HR of 0.512 (95% CI, 0.303–0.867), P = 0.013. The PFS of the matched CAR-T cohort was significantly better than that of the matched SOC cohort ( Fig. 2B ): for median PFS, 4.8 months (95% CI, 4.1–18.0) vs. 1.6 months (95% CI, 1.2–3.8); for 1-year PFS rate, 34.2% (95% CI, 23.0–50.8%) vs. 18.5% (95% CI, 10.0–34.3%), P < 0.001; for 2-year PFS rate, 29.9% (95% CI, 18.6–48.1%) vs. 18.5% (95% CI, 10.0–34.3%), P = 0.001; and a HR of 0.494 (95% CI, 0.310–0.788), P = 0.003, respectively. Sensitivity analyses restricting follow-up to 1.5 and 2.5 years showed consistent findings, with the CAR-T cohort maintaining significantly better OS and PFS than the SOC cohort (all P < 0.05) ( Supplementary Fig. 2 ).

To assess the robustness of our adjusted worst-case HR of 0.512 (95% CI, 0.320–0.867) against residual confounding, we calculated E-values. An unmeasured confounder would need to be associated with both mortality and CAR-T treatment by a risk ratio of 2.551 to nullify the observed protective effect (E-value = 2.551). Based on the magnitude of associations typically observed for measured confounders in this population, substantial residual confounding seems less probable. Also, to render our results nonsignificant, a confounder would need associations of 1.443 with both outcomes (E-value for CI upper limit = 1.443).

To evaluate the robustness of our findings to residual imbalance and unmeasured treatment-related factors, we performed additional weighted and covariate-adjusted analyses within the matched cohort. Using stabilized inverse probability of treatment weights truncated at the [1st–99th] percentiles, the inverse probability of treatment weighting (IPTW)-adjusted Cox proportional hazards model yielded OS HR 0.496 (95% CI, 0.273–0.903; P = 0.022) and PFS HR 0.600 (95% CI, 0.362–0.997; P = 0.048) ( Supplementary Fig. 3 ). Similarly, when prior ASCT before the index date was further incorporated into a doubly robust Cox regression framework, the results remained consistent (OS HR, 0.475; P = 0.007; PFS HR, 0.471; P = 0.002) ( Supplementary Fig. 3 ). The persistence of the survival advantage across these complementary adjustment approaches supports the robustness of our primary findings.

We performed a subgroup analysis for OS and PFS across various demographics, and risk categories ( Fig. 3 ). For OS, patients aged 60–69 years (HR, 0.268; 95% CI, 0.098–0.732), female patients (HR, 0.281; 95% CI, 0.101–0.783) and high IPI patients (HR, 0.333; 95% CI, 0.158–0.701) derived statistically significant OS advantages. For PFS, aged 50–59 years (HR, 0.292; 95% CI, 0.126–0.677) and 60–69 years (HR, 0.326; 95% CI, 0.143–0.743), female patients (HR, 0.169; 95% CI, 0.069–0.415)–high IPI patients (HR–0.330; 95% CI, 0.175–0.624)–patients with 3 rd line of therapy (HR, 0.450; 95% CI, 0.229–0.882) and ≥ 5 th line of therapy (HR, 0.292; 95% CI, 0.090–0.945), and patients with 2–4 ECOG (HR, 0.171; 95% CI, 0.046–0.641) derived statistically significant PFS advantages. Taken together, these subgroup findings reinforce the robustness of the progression-free-survival benefit.

Safety

For safety evaluation, neutropenia was the most common in matched CAR-T cohort (98.0%), followed by lymphopenia (94.0%), thrombocytopenia (38.0%), hypogammaglobulinemia (24.0%), CRS (16.0%), and neurotoxicity (6.0%). Grade ≥ 3 neutropenia was more frequent in the matched CAR-T cohort, whereas thrombocytopenia was more common in the matched SOC cohort; lymphopenia occurred at similar rates in both groups. Other toxicities did not differ significantly between the groups ( Table 3 ).

DISCUSSION

In this propensity score–matched, registry-based cohort from two tertiary centers in Korea, tisagenlecleucel was associated with significantly higher response and superior survival compared with contemporary SOC salvage chemoimmunotherapy in patients with R/R DLBCL treated in the 3L or later setting. The tisagenlecleucel cohort achieved an ORR of 68% and a CR rate of 60%, which translated into longer PFS (median 4.8 vs. 1.6 months) and OS (median 11.9 vs. 3.3 months) relative to matched SOC. Notably, these outcomes were observed despite frequent bridging (80%) and a median apheresis-to-infusion (vein-to-vein) interval of 40 days, reflecting real-world delivery constraints. Real-world practice often features heterogeneous timing and criteria for response assessment during bridging, which can introduce measurement noise and limit comparability across cohorts. Future studies should prespecify uniform evaluation windows aligned with Lugano/LYRIC principles (e.g., fixed imaging landmarks between bridging initiation and CAR-T infusion), record concomitant immunomodulators such as corticosteroids and tocilizumab, and, where feasible, implement central blinded review to minimize information bias and misclassification.

Despite matching procedures, residual confounding cannot be entirely excluded. Notably, the control cohort had higher LDH levels and lower albumin levels at baseline, which can both establish adverse prognostic factors. This imbalance would theoretically bias results against the control group, potentially favoring CAR-T therapy. However, the survival benefit associated with CAR-T remained consistent after IPTW adjustment for these variables, suggesting that the observed treatment effect is not driven solely by this residual imbalance.

Our response and survival results are broadly consistent with the pivotal phase 2 experience and large real-world cohorts of tisagenlecleucel, while providing direct comparative evidence against SOC—an element missing from most real-world reports. 11 14 15 The magnitude of benefit supports the view that, in heavily pretreated patients who remain candidates for cellular therapy, CAR-T yields not only higher remission rates but also more durable disease control than cytotoxic salvage regimens delivered at similar lines of therapy.

The modest activity of SOC in our matched cohort (ORR, 28%; CR, 22%) likely reflects real-world features of refractory disease. First, SOC regimens were heterogeneous and many are known to have limited efficacy in chemoresistant DLBCL. Second, a portion of the SOC cohort was treated before commercial CAR-T availability in Korea, a temporal factor that may have influenced referral thresholds and supportive-care practices. Third, novel agents such as bispecific antibodies or antibody–drug conjugates were largely unavailable during most of the study window (through 2023) and therefore excluded. Importantly, the observed SOC performance aligns with historical refractory DLBCL benchmarks (SCHOLAR-1: ORR, 26%; CR, 7%; median OS, 6.3 months), underscoring persistent limitations of chemotherapy alone in this population. 9

The survival advantage with tisagenlecleucel remained robust to follow-up differences between cohorts. Given the longer calendar-time follow-up in the SOC arm, we repeated the survival analyses after administratively truncating follow-up at prespecified horizons (e.g., 18 and 24 months). As shown in Supplementary Fig. 2 , the separation of Kaplan-Meier curves and the direction/magnitude of HRs were consistent with the primary analysis (all comparisons remained statistically significant), indicating that the observed benefits of tisagenlecleucel are not artifacts of longer tail follow-up in the SOC cohort but reflect a true treatment effect over clinically relevant time windows. These findings strengthen the inference that timely identification and referral of eligible patients can translate into meaningful survival gains even when observation windows are restricted.

Our results should be interpreted in the context of the mixed 2L randomized data. In BELINDA (2L, primary refractory/early relapse), tisagenlecleucel did not improve EFS vs. SOC plus ASCT, a result frequently attributed to differences in line of therapy, disease biology/risk, vein-to-vein time (with attendant reliance on bridging), and supportive-care evolution. 13 In contrast, our cohort reflects the approved ≥ 3L indication, where CAR-T has shown the most consistent benefit, and demonstrates superiority over a matched, contemporaneous SOC comparator in that setting. 11 14 15

Safety profiles diverged in clinically intuitive ways. As expected with CAR-T, CRS (16%) and ICANS (6%) occurred but were predominantly low grade and manageable; no treatment-related mortality occurred in either cohort. Hematologic toxicities showed distinct patterns: neutropenia was more frequent with tisagenlecleucel (consistent with lymphodepleting chemotherapy and early post-infusion cytopenias), whereas thrombocytopenia (and anemia) were more common with SOC, in line with cumulative marrow suppression from multi-agent cytotoxic therapy. Hypogammaglobulinemia occurred only after tisagenlecleucel (24%), consistent with B-cell aplasia biology and supporting infection prophylaxis and intravenous immunoglobulin as indicated. 17 18 19 Overall, these data suggest that, while CAR-T introduces unique immune-mediated risks, the toxicity burden is predictable and manageable within standardized pathways and may be less chronically marrow-suppressive than intensive chemotherapy.

Exploratory analyses provide additional nuance. When restricting to patients who achieved CR, survival did not favor tisagenlecleucel over SOC in our dataset. This contrasts with several single-arm reports in which CR after CAR-T portends excellent durability. 14 15 The discrepancy likely reflects small sample size, selection effects (e.g., favorable biology among chemotherapy-sensitive CRs), and post-response management differences. Practically, these findings suggest that achieving CR by any modality can confer favorable prognosis, while also highlighting the need for larger, prospectively harmonized landmark analyses to clarify outcome determinants among CR achievers.

Our findings are broadly concordant with a 2023 Korean external-comparator analysis that contrasted tisagenlecleucel outcomes from the JULIET full analysis set with conventional chemotherapy using matching-adjusted indirect comparison. That study reconstructed OS from published JULIET Kaplan-Meier curves and weighted a Korean real-world cohort on major prognostic variables (age, ECOG status, Ann Arbor stage, number of prior lines, relapse/refractory status, prior autologous hematopoietic stem-cell transplantation), reporting longer median OS with tisagenlecleucel (11.7 vs. 5.4 months) and a reduced risk of death (crude HR, 0.55; 95% CI, 0.37–0.83; adjusted HR, 0.59; 95% CI, 0.40–0.85), with a time-aligned sensitivity analysis showing aHR 0.52 (95% CI, 0.23–1.16); conventional regimens were predominantly ICE or GDP. 20 In contrast, our study leveraged a contemporaneous SOC cohort from two tertiary centers within a single health system and used 1:1 propensity-score matching on prespecified covariates (age, sex, line of therapy, IPI, ECOG, year of diagnosis), enabling patient-level comparison across multiple endpoints. Consistent with the external comparator, we observed a clear advantage for tisagenlecleucel—higher ORR (68% vs. 28%) and CR (60% vs. 22%) and longer PFS (median, 4.8 vs. 1.6 months; HR, 0.49) and OS (median, 11.9 vs. 3.3 months; HR, 0.51)—together with an on-treatment safety profile characterized by manageable CRS/ICANS and no treatment-related mortality in our cohort. Complementing both, recent Korean single-arm data (n = 96) described response and survival broadly in line with our CAR-T cohort but, by design, could not estimate the relative benefit vs. SOC. 21 Taken together, the matching-adjusted indirect comparison-based external comparator and the single-arm series provide supportive context, while our contemporaneous, propensity-matched analysis quantifies the incremental advantage of tisagenlecleucel over chemotherapy and extends the comparison beyond OS to response and toxicity in an Asian ≥ 3L setting. These complementary datasets also motivate practical implications for referral timing, bridging strategy, and standardized toxicity pathways in Korean practice.

Clinically, the implications are immediate for Korea, where tisagenlecleucel has been the only commercially available CAR-T product for DLBCL since 2021 and thus the predominant cellular option in routine practice. 22 23 Our data support early referral, patient optimization before apheresis, and standardized CRS/ICANS pathways to preserve safety while maximizing benefit. Table 4 summarizes tisagenlecleucel–specific evidence—integrating pivotal trials, real-world cohorts, and our propensity-matched comparison—to provide a historical benchmark for the Korean setting. As Korea transitions to a multi-product CAR-T environment (ex. axicabtagene ciloleucel or anbalcabtagene autoleucel), this tisagenlecleucel benchmark will serve as the reference point for future comparative-effectiveness, sequencing, and health-economic analyses.

This study has several limitations. First, its retrospective design is subject to inherent selection bias and unmeasured confounding, even though propensity score matching was performed; moreover, post-match imbalances in disease-severity markers (e.g., higher LDH, lower albumin, and higher WBC in the SOC cohort) may bias effect estimates toward CAR-T. Nonetheless, the direction and magnitude of benefit remained robust in administrative truncation analyses (18–24 months) and in doubly robust, covariate-adjusted Cox models where data permitted, although residual confounding cannot be fully excluded. Second, the SOC cohort comprised a heterogeneous mix of salvage regimens, reflecting real-world practice but complicating interpretation of comparative efficacy. Third, novel agents such as bispecific antibodies or antibody–drug conjugates were not included, as they were largely unavailable in Korea during most of the study period. Fourth, although key CAR-T product parameters were captured—including infused cell count, product viability, and vein-to-vein time—the cohort was underpowered to model dose/viability–outcome relationships, and manufacturing-process metrics (e.g., CD4:CD8 composition, vector copy number) were not systematically collected. Fifth, the median follow-up time was relatively short, limiting the assessment of long-term durability of responses. Finally, the sample size was modest, which may restrict the precision and generalizability of subgroup analyses.

Despite these limitations, our study provides robust evidence that tisagenlecleucel significantly improves outcomes compared with conventional salvage therapy in Korean patients with R/R DLBCL treated in the ≥ 3L setting.

In conclusion, tisagenlecleucel provided superior survival and response outcomes compared with SOC salvage chemotherapy in Korean patients with R/R DLBCL treated in the ≥ 3L setting, with a manageable safety profile. These findings confirm CAR-T therapy as the standard of care in this population and underscore the importance of timely access and region-specific real-world evidence to inform global practice.

ACKNOWLEDGMENTS

The authors thank Subong Jeong for the co-ordination of data.

Funding: This research was supported by a grant (RS-2023-00216446) from Ministry of Food and Drug Safety in 2023.

Disclosure: The authors have no potential conflicts of interest to disclose.

Author Contributions: Conceptualization: Byeon S, Choi S. Data curation: Kim KY, Kwag DH. Formal analysis: Jung S, Lee JY. Methodology: Choi S, Park SS. Software: Park S, Yoon JH, Lee SE. Validation: Cho BS, Kim YJ, Min CK, Kim HJ. Writing - original draft: Byeon S, Choi S. Writing - review & editing: Jeon JW, Eom KS.

1 Wang SS Epidemiology and etiology of diffuse large B-cell lymphoma Semin Hematol 2023 60 5 255 266 38242772 10.1053/j.seminhematol.2023.11.004 PMC10962251

2 Lee H Park HJ Park EH Ju HY Oh CM Kong HJ nationwide statistical analysis of lymphoid malignancies in Korea Cancer Res Treat 2018 50 1 222 238 28361523 10.4143/crt.2017.093 PMC5784621

3 Coiffier B Thieblemont C Van Den Neste E Lepeu G Plantier I Castaigne S Long-term outcome of patients in the LNH-98.5 trial, the first randomized study comparing rituximab-CHOP to standard CHOP chemotherapy in DLBCL patients: a study by the Groupe d’Etudes des Lymphomes de l’Adulte Blood 2010 116 12 2040 2045 20548096 10.1182/blood-2010-03-276246 PMC2951853

4 Sehn LH Salles G Diffuse large B-cell lymphoma N Engl J Med 2021 384 9 842 858 33657296 10.1056/NEJMra2027612 PMC8377611

5 Sehn LH Berry B Chhanabhai M Fitzgerald C Gill K Hoskins P The revised International Prognostic Index (R-IPI) is a better predictor of outcome than the standard IPI for patients with diffuse large B-cell lymphoma treated with R-CHOP Blood 2007 109 5 1857 1861 17105812 10.1182/blood-2006-08-038257

6 Gisselbrecht C Glass B Mounier N Singh Gill D Linch DC Trneny M Salvage regimens with autologous transplantation for relapsed large B-cell lymphoma in the rituximab era J Clin Oncol 2010 28 27 4184 4190 20660832 10.1200/JCO.2010.28.1618 PMC3664033

7 Van Den Neste E Schmitz N Mounier N Gill D Linch D Trneny M Outcome of patients with relapsed diffuse large B-cell lymphoma who fail second-line salvage regimens in the International CORAL study Bone Marrow Transplant 2016 51 1 51 57 26367239 10.1038/bmt.2015.213

8 Gisselbrecht C Schmitz N Mounier N Singh Gill D Linch DC Trneny M Rituximab maintenance therapy after autologous stem-cell transplantation in patients with relapsed CD20(+) diffuse large B-cell lymphoma: final analysis of the collaborative trial in relapsed aggressive lymphoma J Clin Oncol 2012 30 36 4462 4469 23091101 10.1200/JCO.2012.41.9416 PMC3646314

9 Crump M Neelapu SS Farooq U Van Den Neste E Kuruvilla J Westin J Outcomes in refractory diffuse large B-cell lymphoma: results from the international SCHOLAR-1 study Blood 2017 130 16 1800 1808 28774879 10.1182/blood-2017-03-769620 PMC5649550

10 Nagle SJ Woo K Schuster SJ Nasta SD Stadtmauer E Mick R Outcomes of patients with relapsed/refractory diffuse large B-cell lymphoma with progression of lymphoma after autologous stem cell transplantation in the rituximab era Am J Hematol 2013 88 10 890 894 23813874 10.1002/ajh.23524

11 Schuster SJ Bishop MR Tam CS Waller EK Borchmann P McGuirk JP Tisagenlecleucel in adult relapsed or refractory diffuse large B-cell lymphoma N Engl J Med 2019 380 1 45 56 30501490 10.1056/NEJMoa1804980

12 Schuster SJ Tam CS Borchmann P Worel N McGuirk JP Holte H 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 34516954 10.1016/S1470-2045(21)00375-2

13 Bishop MR Dickinson M Purtill D Barba P Santoro A Hamad N Second-line tisagenlecleucel or standard care in aggressive B-cell lymphoma N Engl J Med 2022 386 7 629 639 34904798 10.1056/NEJMoa2116596

14 Landsburg DJ Frigault MJ Heim M Foley SR Hill B Schofield G Real-world outcomes with tisagenlecleucel in aggressive B-cell lymphoma: subgroup analyses from the CIBMTR registry J Immunother Cancer 2025 13 2 e009890 39924174 10.1136/jitc-2024-009890 PMC11808862

15 Iacoboni G Villacampa G Martinez-Cibrian N Bailén R Lopez Corral L Sanchez JM Real-world evidence of tisagenlecleucel for the treatment of relapsed or refractory large B-cell lymphoma Cancer Med 2021 10 10 3214 3223 33932100 10.1002/cam4.3881 PMC8124109

16 VanderWeele TJ Ding P Sensitivity analysis in observational research: introducing the E-value Ann Intern Med 2017 167 4 268 274 28693043 10.7326/M16-2607

17 Wat J Barmettler S Hypogammaglobulinemia after chimeric antigen receptor (CAR) T-cell therapy: characteristics, management, and future directions J Allergy Clin Immunol Pract 2022 10 2 460 466 34757064 10.1016/j.jaip.2021.10.037 PMC8837681

18 Epperly R Giordani VM Mikkilineni L Shah NN Early and late toxicities of chimeric antigen receptor T-cells Hematol Oncol Clin North Am 2023 37 6 1169 1188 37349152 10.1016/j.hoc.2023.05.010 PMC10592597

19 Kampouri E Walti CS Gauthier J Hill JA Managing hypogammaglobulinemia in patients treated with CAR-T-cell therapy: key points for clinicians Expert Rev Hematol 2022 15 4 305 320 35385358 10.1080/17474086.2022.2063833

20 Park S Kim JH Kim S Kang J Moon S Kim SJ Comparison of tisagenlecleucel with conventional treatments for relapsed/refractory diffuse large B-cell lymphomas: a retrospective external comparator study Blood Cancer J 2023 13 1 123 37591839 10.1038/s41408-023-00889-5 PMC10435565

21 Yoon SE Cho J Cho D Kang ES Kim SJ Kim WS Clinical outcomes of tisagenlecleucel in relapsed/refractory diffuse large B-cell lymphoma: insights from a single-center study Int J Hematol 2025 122 4 533 545 40847083 10.1007/s12185-025-04006-z PMC12476396

22 Byun JM Practical issues in CAR T-cell therapy Blood Res 2023 58 S1 S11 S12 36843380 10.5045/br.2023.2023015 PMC10133847

23 Lee S Lee JH Cell and gene therapy regulatory, pricing, and reimbursement framework: with a focus on South Korea and the EU Front Public Health 2023 11 1109873 36908458 10.3389/fpubh.2023.1109873 PMC9998493

SUPPLEMENTARY MATERIALS

Supplementary Fig. 1

Love plot: covariate balance across matched CAR-T and SOC cohort.

Supplementary Fig. 2

Comparison of survival outcomes across different follow-up periods.

Supplementary Fig. 3

Inverse-probability-of-treatment–weighted survival comparing tisagenlecleucel (CAR-T) vs. standard-of-care (SOC) in ≥ 3L R/R DLBCL.