Comparative efficacy and safety of systemic therapies in transplant-eligible and ineligible relapsed/refractory large B-cell lymphoma: a Bayesian network meta-analysis

Introduction

Large B-cell lymphoma (LBCL) constitutes a heterogeneous spectrum of non-Hodgkin lymphomas (1). While frontline rituximab-based immunochemotherapy offers curative potential, 30–40% of patients eventually develop relapsed or refractory (R/R) disease (2,3).

For patients with transplant-eligible R/R LBCL, standard of care (SoC) consists of salvage immunochemotherapy followed by high-dose chemotherapy (HDCT) and autologous stem cell transplantation (ASCT), whereas CD19-directed chimeric antigen receptor (CAR) T-cell therapy has demonstrated durable efficacy for those with primary refractory or early relapsing disease (4,5). However, real-world data indicate that many patients remain ineligible for intensive cellular therapies due to advanced age, severe comorbidities, rapid disease progression, or socioeconomic constraints, relegating them to alternative palliative approaches (6,7).

For transplant-ineligible populations, conventional salvage regimens, such as rituximab plus gemcitabine (RG) and oxaliplatin (R-GemOx) or bendamustine plus rituximab (BR), yield suboptimal outcomes, conferring a median event-free survival (EFS) or progression-free survival (PFS) of merely 2.3–6.7 months (8,9). Subsequently, attempts have been made to incorporate alternative chemotherapeutic agents and novel immunotherapies, prominently featuring the antibody-drug conjugate (ADC) polatuzumab vedotin and CD20/CD3 bispecific antibodies (BsAbs) such as glofitamab and epcoritamab. These agents are being actively integrated with rituximab-based chemotherapy or other combinatorial regimens to manage second-line and subsequent-line disease in clinical practice (10,11).

Despite the rapid expansion of this therapeutic armamentarium, the absence of direct, head-to-head randomized controlled trials (RCTs) comparing these novel agents remains the critical knowledge gap and unmet clinical need. This lack of comparative evidence complicates optimal treatment selection and sequencing. Therefore, we conducted a Bayesian network meta-analysis (NMA) accordingly. The primary endpoint of this study was PFS. The secondary endpoints included EFS, overall survival (OS), objective response rate (ORR) and adverse events (AEs). We present this article in accordance with the PRISMA reporting checklist (12) (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0678/rc).

Methods

Registration

This NMA was prospectively registered in the Prospective Register of Systematic Reviews (PROSPERO) registry (CRD420261293124).

Search strategy and selection criteria

A comprehensive literature search was conducted across three databases—PubMed, Embase, and Cochrane—alongside other relevant resources to identify relevant publications and conference abstracts up to January 15^th, 2026. The search strategy utilized a combination of key terms, including “large B-cell lymphoma”, “relapsed”, “refractory”, “randomized”, etc. The detailed search strategy is presented in Table S1. Furthermore, the reference lists of pertinent articles were manually screened to identify any additional eligible studies.

Study population

The study population was restricted to adult patients with histologically confirmed R/R LBCL following at least one prior line of systemic therapy. Studies were deemed eligible if they met the following criteria: (I) enrolled patients with histologically confirmed LBCL, including diffuse LBCL, not otherwise specified (DLBCL NOS), high-grade B-cell lymphoma (double-hit or triple-hit), primary mediastinal LBCL (PMBCL), T cell/histiocyte-rich LBCL, follicular lymphoma (FL) grade 3b, or transformed FL, etc.; (II) belonged to R/R disease following at least one prior line of treatment; (III) RCT design; (IV) were available as English full-text articles or conference abstracts; (V) studies that reported at least one outcome including PFS, EFS, OS, ORR, AEs [treatment-emergent adverse events (TEAEs), treatment-related adverse events (TRAEs), grade ≥3 TEAEs, and grade ≥3 TRAEs].

PFS was defined as the time from randomization to the date of disease progression or death of any cause. OS was defined as the time from randomization to the date of death of any cause. ORR was defined as the proportion of patients achieving a complete or partial response per the Lugano classification (13). AEs were evaluated according to the National Cancer Institute Common Terminology Criteria for AEs.

While studies were excluded if they met one of the following criteria: (I) non-randomized study design (e.g., single-arm studies, retrospective cohorts); (II) secondary or non-primary clinical literature (e.g., reviews, meta-analyses, prognostic or genetic analyses, study protocols, and previously published indirect comparisons); (III) irrelevant populations or interventions (e.g., not in the relapsed/refractory stage, other unmatched histologies, pediatric patients, primary CNS lymphoma, and receiving radiotherapy); (IV) updated analyses or subgroup analyses of the same study; (V) without relevant outcomes; and (VI) studies with a very limited sample size (n<20 per arm).

Data extraction and risk of bias assessment

Data extraction for each eligible study was conducted independently by two reviewers (L.H. and K.C.). Any discrepancies were resolved through negotiation, with a third author (Z.L.) providing adjudication when necessary. The extracted baseline characteristics including study demographics (first author, publication year, phase), clinical parameters (treatment lines, experimental and control regimens, sample size per arm), the percentage of international prognostic index (IPI) ≥2 and primary refractory disease, as well as median follow-up time. For efficacy endpoints, hazard ratios (HRs) and their corresponding 95% credible intervals (CrIs) were extracted for PFS, EFS, and OS from reports of the original RCT studies. Additionally, incidence data for ORR, any-grade AEs, and grade ≥3 AEs were collected to estimate odds ratios (ORs) or risk ratios (RRs).

The quality and risk of bias of the included studies were evaluated using the modified Cochrane Risk of Bias Tool 2.0 (14). This assessment categorized studies as “low”, “high”, or “some concerns” across six standardized domains: bias arising from the randomization process, deviations from intended interventions, missing outcome data, outcome measurement, selective reporting, and other potential sources of bias.

Statistical analysis

An NMA was conducted to compare the efficacy and safety profiles of the diverse systemic treatment regimens using Markov chain Monte Carlo simulations. Summary effect estimates were appropriately reported as HRs, ORs, or RRs, together with their respective 95% CrIs.

Given that all direct evidence was derived from single trials per comparison, a fixed-effects consistency model was employed. Model fit was evaluated via the deviance information criterion (DIC), while the I² statistic quantified the heterogeneity of treatment effects across studies. To establish a hierarchy of efficacy, a league table was generated for all pairwise comparisons, and surface under the cumulative ranking curve (SUCRA) probabilities ranging from 0–1 were calculated for each treatment. To evaluate the robustness of our findings, a sensitivity analysis was performed by excluding trials with a median follow-up of <12 months. All statistical analyses were performed using the “gemtc” package in R software (version 4.4.3), with statistical significance defined by the exclusion of unity within the 95% CrI.

Results

Patient characteristics

A systematic search of three databases including PubMed, Cochrane, and Embase, supplemented by other relevant sources, yielded 5,474 records. Following the removal of duplicates, 3,862 publications underwent title and abstract screening, resulting in 70 studies for full-text review. Ultimately, 14 RCTs involving 3,329 patients were included in the NMA, comprising 6 studies for transplant-eligible populations and 8 for transplant-ineligible populations (Figure 1).

Figure 1 Study flow diagram. CNS, central nervous system; RCT, randomized controlled trial.

The transplant-eligible studies included three trials investigating CAR-T therapies [axicabtagene ciloleucel (axi-cel), lisocabtagene maraleucel (liso-cel), and tisagenlecleucel (tisa-cel)] (4,15,16), while the remaining three trials assessed ofatumumab plus cisplatin, cytarabine, and dexamethasone (O-DHAP) (17), dacetuzumab plus rituximab, ifosfamide, carboplatin and etoposide (Dace-R-ICE) (18), and rituximab plus dose-intensive cyclophosphamide, etoposide, cisplatin (R-DICEP) (19), respectively. All six trials targeted second-line treatment for R/R LBCL, utilizing SoC—defined as salvage chemotherapy [such as rituximab plus ifosfamide, carboplatin and etoposide (R-ICE), R-GemOx, RG, dexamethasone and cisplatin (R-GDP), rituximab plus cisplatin, cytarabine, and dexamethasone (R-DHAP), or rituximab plus etoposide, cytarabine, cisplatinum and methylprednisolone (R-ESHAP)] followed by HDCT and ASCT for responders—as the control arm. In the transplant-ineligible cohort, individual studies evaluated glofitamab plus gemcitabine and oxaliplatin (Glofit-GemOx) (20), polatuzumab vedotin plus bendamustine and rituximab (Pola-BR) (21), polatuzumab vedotin plus rituximab, gemcitabine and oxaliplatin (pola-R-GemOx) (22), mosunetuzumab plus polatuzumab vedotin (mosun-pola) (6), decitabine-RDHAP (23), nivolumab plus R-GemOx (nivo-R-GemOx) (24), rituximab plus pixantrone (R-PIX) (25), and rituximab plus inotuzumab ozogamicin (R-InO) (26) compared to R-Chemotherapy (R-GemOx, BR, or RG) as second-line or subsequent treatment options. Table 1 and Table S2 summarize the main characteristics of these studies. Figure 2 presented the network plots of all included treatments and endpoints comparisons.

Figure 2 Comparative network plots illustrating the direct and indirect comparisons of endpoints including PFS, OS, ORR, and grade ≥3 TEAEs for patients with relapsed/refractory LBCL: transplant-eligible patients (A,B) and transplant-ineligible patients (C,D). Nodes represent treatment strategies, with node size proportional to the total number of participants. Edges (lines) indicate direct comparisons between treatments. axi-cel, axicabtagene ciloleucel; Dace-R-ICE, dacetuzumab plus rituximab, ifosfamide, carboplatin and etoposide; Decitabine-RDHAP, decitabine plus rituximab, cisplatin, cytarabine, and dexamethasone; Glofit-GemOx, glofitamab plus gemcitabine and oxaliplatin; LBCL, large B-cell lymphoma; liso-cel, lisocabtagene maraleucel; Mosun-Pola, mosunetuzumab plus polatuzumab vedotin; Nivo-R-GemOx, nivolumab plus rituximab, gemcitabine and oxaliplatin; O-DHAP, ofatumumab plus cisplatin, cytarabine, and dexamethasone; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; pola-BR, polatuzumab vedotin plus bendamustine and rituximab; pola-R-GemOx, polatuzumab vedotin plus rituximab, gemcitabine and oxaliplatin; R-Chemo, rituximab plus chemotherapy; R-DICEP, rituximab plus dose-intensive cyclophosphamide, etoposide, cisplatin; R-InO, rituximab plus inotuzumab ozogamicin; R-PIX, rituximab plus pixantrone; SoC, standard of care; TEAEs, treatment-emergent adverse events; tisa-cel, tisagenlecleucel.

NMA

Transplant-eligible population

In the primary outcome analysis, in terms of PFS (Figure 3A), five regimens were included except tisa-cel and R-DICEP. Among them liso-cel and axi-cel achieved significantly improved PFS compared to SoC and conventional immunochemotherapy including O-DHAP and Dace-R-ICE. While no significant difference was observed in the comparison between liso-cel and axi-cel.

Figure 3 Pooled estimates of the NMA in the transplant-eligible patients. (A) Pooled HRs (95% CrIs) for OS (upper triangle) and PFS (lower triangle). (B) Pooled ORs (95% CrIs) for ORR (upper triangle) and RRs (95% CrI) for grade ≥3 TEAE (lower triangle). Estimates in the upper triangles represent row-defined treatments vs. column-defined treatments, whereas those in the lower triangles represent column-defined vs. row-defined comparisons. axi-cel, axicabtagene ciloleucel; CrI, credible interval; Dace-R-ICE, dacetuzumab plus rituximab, ifosfamide, carboplatin and etoposide; HR, hazard ratio; liso-cel, lisocabtagene maraleucel; NA, not available; NMA, network meta-analysis; O-DHAP, ofatumumab plus cisplatin, cytarabine, and dexamethasone; OR, odds ratio; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; R-DICEP, rituximab plus dose-intensive cyclophosphamide, etoposide, cisplatin; RR, risk ratio; SoC, standard of care; TEAEs, treatment-emergent adverse events; tisa-cel, tisagenlecleucel.

In the secondary outcome analysis, in terms of EFS, data were only available from three CAR-T studies, liso-cel and axi-cel demonstrated superior EFS benefits compared to SoC or tisa-cel (Figure S1). Six regimens were included except R-DICEP in terms of OS analysis. However, no treatment strategy demonstrated a statistically significant OS advantage (Figure 3A). Regarding ORR, all seven regimens were included, among them axi-cel and liso-cel showed superior ORR than the rest four treatments regimens except R-DICEP (Figure 3B). Data regarding grade ≥3 TEAEs were unavailable for O-DHAP and R-DICEP. Among the remaining five regimens, axi-cel was associated with the highest incidence of grade ≥3 TEAEs, reaching statistical significance when compared with tisa-cel (HR =3.64, 95% CrI: 1.42–9.6) and SoC (HR =2.09, 95% CrI: 1.08–4.19) (Figure 3B). Data on TRAEs were only available for tisa-cel, precluding any comparison.

Transplant-ineligible population

In the primary outcome analysis, in terms of PFS, all nine strategies were included, novel BsAbs- or ADC-containing treatment combinations (Glofit-GemOx, pola-BR, pola-R-GemOx, and Mosun-Pola) significantly outperformed R-PIX, R-InO, nivo-R-GemOx, and R-chemo (Figure 4A). While no statistically significant differences were observed among the four BsAbs- or ADC-containing treatment combinations as mentioned above.

Figure 4 Pooled estimates of the NMA in the transplant-ineligible patients. (A) Pooled HRs (95% CrIs) for OS (upper triangle) and PFS (lower triangle). (B) Pooled ORs (95% CrIs) for ORR (upper triangle) and RRs (95% CrI) for grade ≥3 TEAE (lower triangle). Estimates in the upper triangles represent row-defined treatments vs. column-defined treatments, whereas those in the lower triangles represent column-defined vs. row-defined comparisons. CrI, credible interval; Decitabine-RDHAP, decitabine plus rituximab, cisplatin, cytarabine, and dexamethasone; Glofit-GemOx, glofitamab plus gemcitabine and oxaliplatin; HR, hazard ratio; Mosun-Pola, mosunetuzumab plus polatuzumab vedotin; NA, not available; Nivo-R-GemOx, nivolumab plus rituximab, gemcitabine and oxaliplatin; NMA, network meta-analysis; OR, odds ratio; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; pola-BR, polatuzumab vedotin plus bendamustine and rituximab; pola-R-GemOx, polatuzumab vedotin plus rituximab, gemcitabine and oxaliplatin; R-Chemo, rituximab plus chemotherapy; R-InO, rituximab plus inotuzumab ozogamicin; R-PIX, rituximab plus pixantrone; RR, risk ratio; TEAEs, treatment-emergent adverse events.

In the secondary outcome analysis, eight regimens were included except nivo-R-GemOx in the analysis of OS and ORR. In the OS analysis, the four BsAbs- or ADC-containing treatment combinations mentioned above demonstrated significantly improved OS compared to R-PIX, R-InO, and R-chemo (Figure 4A). Regarding ORR, R-Chemo and R-InO demonstrated significantly lower response rates than the other evaluated strategies (Figure 4B). Data regarding grade ≥3 TEAEs were available for five regimens. Among them, Glofit-GemOx was associated with a significantly higher incidence of grade ≥3 TEAEs (Figure 4B).

Bayesian ranking profiles

The Bayesian ranking profiles for comparable treatments across each outcome were presented in Figures 5,6 and Table S3. In transplant-eligible patients, liso-cel achieved the highest ranking across PFS (SUCRA =0.72), ORR (SUCRA =0.79), all grade TEAEs (SUCRA =0.80), axi-cel ranks first in grade ≥3 TEAEs (SUCRA =0.60), whereas dacetumumab plus R-ICE ranked highest for OS (SUCRA =0.49). In transplant-ineligible patients, Glofit-GemOx attained the highest ranking for PFS (SUCRA =0.51), overall TEAEs (SUCRA =0.54), and grade ≥3 TEAEs (SUCRA =1). Whereas, Pola-BR achieved the highest ranking for OS (SUCRA =0.72) and ORR (SUCRA =0.62).

Figure 5 The SUCRA ranking plots showing all different treatment strategies for transplant-eligible patients. (A) Axi-cel; (B) Liso-cel; (C) Tisa-cel; (D) Dace-R-ICE; (E) O-DHAP; (F) R-DICEP; (G) Soc. axi-cel, axicabtagene ciloleucel; Dace-R-ICE, dacetuzumab plus rituximab, ifosfamide, carboplatin and etoposide; EFS, event-free survival; liso-cel, lisocabtagene maraleucel; O-DHAP, ofatumumab plus cisplatin, cytarabine, and dexamethasone; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; R-DICEP, rituximab plus dose-intensive cyclophosphamide, etoposide, cisplatin; SoC, standard of care; SUCRA, surface under the cumulative ranking curve; TEAEs, treatment-emergent adverse events; tisa-cel, tisagenlecleucel.

Figure 6 The SUCRA ranking plots showing all different treatment strategies for transplant-ineligible patients. (A) Glofit-GemOx; (B) Pola-R-GemOx; (C) Pola-BR; (D) Mosun-Pola; (E) Nivo-R-GemOx; (F) R-InO; (G) R-PIX; (H) Decitabine-RDHAP; (I) R-Chemo. AEs, adverse events; Decitabine-RDHAP, decitabine plus rituximab, cisplatin, cytarabine, and dexamethasone; Glofit-GemOx, glofitamab plus gemcitabine and oxaliplatin; Mosun-Pola, mosunetuzumab plus polatuzumab vedotin; Nivo-R-GemOx, nivolumab plus rituximab, gemcitabine and oxaliplatin; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; pola-BR, polatuzumab vedotin plus bendamustine and rituximab; pola-R-GemOx, polatuzumab vedotin plus rituximab, gemcitabine and oxaliplatin; R-Chemo, rituximab plus chemotherapy; R-InO, rituximab plus inotuzumab ozogamicin; R-PIX, rituximab plus pixantrone; SUCRA, surface under the cumulative ranking curve; TEAEs, treatment-emergent adverse events.

Quality of evidence

In the transplant-eligible cohort, the trial by Steward lacked sufficient disclosure of its randomization and stratification methods to rule out bias arising from the randomization process, and was consequently rated as having “some concern”, the remaining five studies were assessed as “low risk”. In the transplant-ineligible cohort, the trials from Matthew et al. and Held et al. were derived exclusively from conference abstracts, precluding an adequate assessment of bias in the selection of reported results, while the study by Kong featured a limited inclusion size that might diminish statistical power, so they evaluated as having “some concern”, the remaining six studies were assessed as “low risk”. Figures S2-S14 summarized the risk of bias for all included studies.

Heterogeneity analysis was conducted for each endpoint across all treatment strategies. The results revealed minimal heterogeneity in all comparisons (I² 0–25%), with the specific DIC and I² values summarized in Table S4. A sensitivity analysis was performed by excluding trials with a median follow-up of <12 months (specifically, tisa-cel and O-DHAP), the comparative outcomes for the remaining treatment network remained stable (Figure S15 and Table S5).

Discussion

The advent of novel therapeutic strategies, including CAR-T, BsAbs, and ADCs, has markedly expanded treatment options and improved outcomes for R/R LBCL (2,27). In this NMA, we comprehensively evaluated the efficacy and safety of available systemic treatments, and provides valuable comparative evidence and guidance for clinical decision-making in R/R LBCL.

For transplant-eligible patients, the results reaffirm the superiority of CD19-directed CAR-T therapies (specifically liso-cel and axi-cel) over SoC in terms of EFS and PFS, solidifying the role of CAR-T therapies as preferred treatment strategies for primary refractory/early relapsed (≤12 months) disease. Although the significant PFS advantage of CAR-T therapies failed to translate into a statistically significant OS benefit across the network, this paradox is likely attributed to crossover effects. In the pivotal phase III studies, 56–60% of patients in the SoC arms who experienced disease progression subsequently crossed over to receive CAR-T therapy, which may dilute the OS difference between the study arms (4,15). Nevertheless, long-term follow-up data from the ZUMA-1 trial demonstrated durable responses of axi-cel with a median OS of 25.8 months, and an estimated 5-year OS rate of 42.6% (28). In the comparison of the three CAR-T agents in our study, no significant differences in efficacy or safety were observed between axi-cel and liso-cel in the indirect analysis, but liso-cel ranked the first in EFS, PFS, and ORR according to the SUCRA value. However, both demonstrated superior efficacy compared with tisa-cel, consistent with findings from real-world studies (29,30). Additionally, axi-cel was found to be associated with the highest incidence of grade ≥3 TEAEs. In a large real-world meta-analysis, axi-cel was associated with a higher incidence of grade ≥3 immune effector cell–associated neurotoxicity syndrome [OR =3.95; 95% confidence interval (CI): 3.05–5.11] compared to liso-cel (30). Now axi-cel and liso-cel are currently approved for use in the second-line setting of early R/R LBCL, whereas tisa-cel was approved in the third-line setting (2,31). Recently, tailored clinical guidance has been developed to optimize the care of elderly patients with R/R DLBCL who are candidates for CAR-T cell therapy (32).

Dacetuzumab plus R-ICE ranked highest for OS (SUCRA =0.49) in this study. However, dacetuzumab, a CD40-directed monoclonal antibody, has not received approval in the management of LBCL. In its phase II trial, the addition of dacetuzumab to R-ICE failed to meet its primary endpoint of complete response rate (CRR) (18), and no subsequent studies have provided further support for its clinical value in this setting. Therefore, the finding is likely a statistical artifact, potentially driven by limited follow-up duration, small sample size, etc.

For transplant-ineligible patients, our analysis categorically demonstrates the inadequacy of conventional immunochemotherapy. The integration of novel BsAbs and ADCs represents a critical breakthrough. Glofit-GemOx exhibited the most robust PFS benefit compared to R-Chemo (HR =0.32, 95% CrI: 0.23–0.45) and the highest SUCRA ranking for PFS (SUCRA =0.51), while also accompanied by significantly higher incidence of grade ≥3 TEAEs. Cytokine release syndrome was the most common AE related to glofitamab; however, the majority of cases were grade 1–2. Another reason for the higher incidence of AEs is that patients in the Glofit-GemOx group received more treatment cycles prior to disease progression than those in the R-GemOx group (median, 11 vs. 4 cycles) (20). Our findings indicated that Pola-BR may provide optimal OS and ORR benefits. However, the trial contributing these data was limited by a small sample size of 40 patients per arm. Furthermore, the Pola-BR group had numerically fewer patients with an IPI >2 (55% vs. 73%) and primary refractory status (53% vs. 70%) than the control group, indicating a potential selection bias that could inflate the observed efficacy (21). More evidence is awaited. The study of pola-BR was not included in the AE analysis because the grade ≥3 TEAEs of BR was not revealed. While the grade ≥3 TEAEs rate of pola-BR was revealed as 80% which even higher than 78% of Glofit-GemOx. In the safety profile of pola-BR, the most common grade ≥3 TEAEs were hematologic toxicities including neutropenia, thrombocytopenia, and anemia (21). The reason for using grade ≥3 TEAEs rather than grade ≥3 TRAEs as an endpoint is that data on grade ≥3 TRAEs were available in only 3 of the 14 included trials, whereas data on grade ≥3 TEAEs were reported in 8 of the 14 trials. Overall, the safety profiles of Glofit-GemOx and pola-BR were manageable. Beyond considerations related to safety profile, real-world data highlighted that the efficacy of glofitamab or epcoritamab is closely associated with tumor CD20 expression levels, moreover, 89.5% of patients who experienced disease progression on CD3/CD20 BsAbs were found to have CD20 loss (33). Loss of the target antigen represents a recognized mechanism of resistance to anti-CD19 CAR T-cell therapies (34), and this similar phenomenon is increasingly being found in patients receiving BsAbs (33,35).

Another key and increasingly debated issue is the optimal sequencing of CAR-T therapy and BsAb. Importantly, prior exposure to CD3/CD20 BsAbs seems not compromise the efficacy of subsequent CAR-T cell therapy. In one retrospective study with 47 patients previously treated with BsAbs, the subsequent CAR-T therapy yielded an ORR of 85% and a CRR of 43%, which were comparable to those observed in matched BsAb-naïve patients (36). A large meta-analysis including 2,122 patients demonstrated that the CRR was significantly higher when CAR-T therapy was administered after BsAb treatment compared to BsAb use following CAR-T therapy (53.7% vs. 29.4%) (37). The use of BsAbs as bridging therapy may potentiate the efficacy of subsequent CAR-T infusion, more data are awaited. Beyond prior BsAb exposure, previous treatment history with other targeted agents may also influence therapeutic sequencing. For instance, for patients who have already received Pola in the frontline setting, the suitability of Pola-BR as a second-line option warrants careful reconsideration (38).

In addition, there are several relevant therapeutic strategies that were not included in this study. Tafasitamab in combination with lenalidomide has been approved for patients with R/R LBCL, however, this approval was based on a single-arm phase II study, precluding its inclusion in our NMA (39). Furthermore, epcoritamab monotherapy has recently been evaluated in the phase III EPCORE DLBCL-1 trial, in which it was compared with R-GemOx or BR in patients with R/R LBCL. According to preliminary results reported in a press release, PFS favored epcoritamab (HR =0.74, 95% CI: 0.60–0.92), whereas OS failed to demonstrate a statistically significant difference (HR =0.96, 95% CI: 0.77–1.20) (40). Detailed data are anticipated to be presented at an upcoming medical meeting.

This study existed several limitations. Firstly, node-splitting analysis for inconsistency was not feasible in this study as all direct comparisons were conducted against the same control arm. This network structure reflects the current clinical trial landscape in R/R LBCL. Secondly, while limited sample size and the single-center nature of the study exploring decitabine-RDHAP may weaken the reliability of its specific estimates, this regimen is rarely used in clinical practice and does not affect the primary findings of our manuscript. Thirdly, although in the transplant-ineligible cohort the inclusion of patients receiving second-line and subsequent therapies (≥2 lines) introduces clinical heterogeneity in prior treatments, our strict restriction to RCTs minimized selection bias as much as possible. Besides, our statistical models revealed overall low heterogeneity, supporting the robustness of the network estimates and the credibility of our findings. Fourthly, although two studies in the transplant-eligible cohort reported relatively short follow-up durations of approximately 10 months, a sensitivity analysis excluding these trials confirmed the robustness of our overall findings.

Conclusions

This NMA provides a comprehensive indirect comparison of systemic therapies for R/R LBCL. For transplant-eligible patients, our findings suggest that CAR-T therapies, particularly liso-cel, may offer the most favorable probability of PFS and EFS benefit. For transplant-ineligible patients, Glofit-GemOx appears to maximize PFS, whereas Pola-BR suggests an optimal OS and ORR advantage. However, given the indirect nature of these comparisons and the associated toxicities, the rankings should be interpreted cautiously. Head-to-head RCTs with larger sample sizes are needed to provide more definitive evidence. Moreover, future research may prioritize biomarker-guided prospective trials and the integration of molecular risk profiling to rigorously validate these findings, individualize treatment selection, and delineate the optimal sequencing of these potent therapies.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the PRISMA reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0678/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0678/prf

Funding: The study was supported by Special Research Project of Beijing Association for the Promotion of Public Health Science Popularization (No. 2025-HX-206); National High Level Hospital Clinical Research Funding; Elite Medical Professionals Initiative of China-Japan Friendship Hospital (No. ZRJY2025-QMPY44).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0678/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Li X, Singhal K, Deng Q, et al. Large B cell lymphoma microenvironment archetype profiles. Cancer Cell 2025;43:1347-1364.e13. [Crossref] [PubMed]
2. Bock AM, Epperla N. Therapeutic landscape of primary refractory and relapsed diffuse large B-cell lymphoma: Recent advances and emerging therapies. J Hematol Oncol 2025;18:68. [Crossref] [PubMed]
3. Sesques P, Manson G, Dubois S, et al. Large B-cell lymphoma: The LYSA pragmatic guidelines. Eur J Cancer 2026;232:116070. [Crossref] [PubMed]
4. Locke FL, Miklos DB, Jacobson CA, et al. Axicabtagene Ciloleucel as Second-Line Therapy for Large B-Cell Lymphoma. N Engl J Med 2022;386:640-54. [Crossref] [PubMed]
5. Abramson JS, Solomon SR, Arnason J, et al. Lisocabtagene maraleucel as second-line therapy for large B-cell lymphoma: primary analysis of the phase 3 TRANSFORM study. Blood 2023;141:1675-84. [Crossref] [PubMed]
6. Budde LE, Zhang H, Kim WS, et al. Mosunetuzumab Plus Polatuzumab Vedotin in Transplant-Ineligible Refractory/Relapsed Large B-Cell Lymphoma: Primary Results of the Phase III SUNMO Trial. J Clin Oncol 2025;43:3799-811. [Crossref] [PubMed]
7. Westin JR, Oluwole OO, Kersten MJ, et al. Survival with Axicabtagene Ciloleucel in Large B-Cell Lymphoma. N Engl J Med 2023;389:148-57. [Crossref] [PubMed]
8. Yamshon S, Koff JL, Larson MC, et al. Outcomes of Relapsed or Refractory Diffuse Large B-Cell Lymphoma Treated With R-GemOx: A Multicenter Cohort Study. Am J Hematol 2025;100:606-15. [Crossref] [PubMed]
9. Vacirca JL, Acs PI, Tabbara IA, et al. Bendamustine combined with rituximab for patients with relapsed or refractory diffuse large B cell lymphoma. Ann Hematol 2014;93:403-9. [Crossref] [PubMed]
10. Thieblemont C, Gomes Da Silva M, Leppä S, et al. Large B-cell lymphoma (LBCL): EHA Clinical Practice Guidelines for diagnosis, treatment, and follow-up. Hemasphere 2025;9:e70207. [Crossref] [PubMed]
11. Wallace DS, Loh KP, Casulo C. How I treat older patients with relapsed/refractory diffuse large B-cell lymphoma. Blood 2025;145:277-89. [Crossref] [PubMed]
12. Hutton B, Salanti G, Caldwell DM, et al. The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations. Ann Intern Med 2020;162:777-84. [Crossref] [PubMed]
13. Cheson BD, Fisher RI, Barrington SF, et al. Recommendations for initial evaluation, staging, and response assessment of Hodgkin and non-Hodgkin lymphoma: the Lugano classification. J Clin Oncol 2014;32:3059-68. [Crossref] [PubMed]
14. Sterne JAC, Savović J, Page MJ, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ 2019;366:l4898. [Crossref] [PubMed]
15. Kamdar M, Solomon SR, Arnason J, et al. Lisocabtagene Maraleucel Versus Standard of Care for Second-Line Relapsed/Refractory Large B-Cell Lymphoma: 3-Year Follow-Up From the Randomized, Phase III TRANSFORM Study. J Clin Oncol 2025;43:2671-8. [Crossref] [PubMed]
16. Bishop MR, Dickinson M, Purtill D, et al. Second-Line Tisagenlecleucel or Standard Care in Aggressive B-Cell Lymphoma. N Engl J Med 2022;386:629-39. [Crossref] [PubMed]
17. van Imhoff GW, McMillan A, Matasar MJ, et al. Ofatumumab Versus Rituximab Salvage Chemoimmunotherapy in Relapsed or Refractory Diffuse Large B-Cell Lymphoma: The ORCHARRD Study. J Clin Oncol 2017;35:544-51. [Crossref] [PubMed]
18. Fayad L, Ansell SM, Advani R, et al. Dacetuzumab plus rituximab, ifosfamide, carboplatin and etoposide as salvage therapy for patients with diffuse large B-cell lymphoma relapsing after rituximab, cyclophosphamide, doxorubicin, vincristine and prednisolone: a randomized, double-blind, placebo-controlled phase 2b trial. Leuk Lymphoma 2015;56:2569-78. [Crossref] [PubMed]
19. Stewart DA, Kuruvilla J, Lee D, et al. Canadian cancer trials group LY.17: A randomized phase II study evaluating novel salvage therapy pre-autologous stem cell transplant in relapsed/refractory diffuse large B-cell lymphoma-outcome of rituximab-dose-intensive cyclophosphamide, etoposide, cisplatin (R-DICEP) versus R-GDP. Br J Haematol 2024;205:881-90. [Crossref] [PubMed]
20. Abramson JS, Ku M, Hertzberg M, et al. Glofitamab plus gemcitabine and oxaliplatin (GemOx) versus rituximab-GemOx for relapsed or refractory diffuse large B-cell lymphoma (STARGLO): a global phase 3, randomised, open-label trial. Lancet 2024;404:1940-54. [Crossref] [PubMed]
21. Sehn LH, Hertzberg M, Opat S, et al. Polatuzumab vedotin plus bendamustine and rituximab in relapsed/refractory DLBCL: survival update and new extension cohort data. Blood Adv 2022;6:533-43. [Crossref] [PubMed]
22. Matasar M, Li ZM, Vassilakopoulos TP, et al. Polatuzumab Vedotin, Rituximab, Gemcitabine and Oxaliplatin (Pola-R-Gemox) for Relapsed/Refractory (R/R) Diffuse Large B-cell Lymphoma (DLBCL): Results from the Randomized Phase III Polargo Trial. EHA Library 2025;4159178:S101.
23. Kong X, Zhang X, Ding M, et al. Decitabine combined with RDHAP regimen in relapsed/refractory diffuse large B cell lymphoma. Cancer Med 2023;12:8134-43. [Crossref] [PubMed]
24. Held G, Altmann B, Kerkhoff A, et al. R-GemOx Plus Nivolumab Vs R-GemOx As Second-Line Therapy for Large B-Cell Lymphoma in Transplant-Ineligible Patients: Interim Analysis of the Niveau Trial, an International, Randomized Phase 3 Study of the AGMT, GLA, HOVON, Lysa and PLRG. Blood 2023;142: [Crossref]
25. Pettengell R, Długosz-Danecka M, Andorsky D, et al. Pixantrone plus rituximab versus gemcitabine plus rituximab in patients with relapsed aggressive B-cell non-Hodgkin lymphoma not eligible for stem cell transplantation: a phase 3, randomized, multicentre trial (PIX306). Br J Haematol 2020;188:240-8. [Crossref] [PubMed]
26. Dang NH, Ogura M, Castaigne S, et al. Randomized, phase 3 trial of inotuzumab ozogamicin plus rituximab versus chemotherapy plus rituximab for relapsed/refractory aggressive B-cell non-Hodgkin lymphoma. Br J Haematol 2018;182:583-6. [Crossref] [PubMed]
27. Zelenetz AD, Gordon LI, Abramson JS, et al. NCCN Guidelines® Insights: B-Cell Lymphomas 3.2025. J Natl Compr Canc Netw 2025;23:e250048. [Crossref] [PubMed]
28. Neelapu SS, Jacobson CA, Ghobadi A, et al. Five-year follow-up of ZUMA-1 supports the curative potential of axicabtagene ciloleucel in refractory large B-cell lymphoma. Blood 2023;141:2307-15. [PubMed]
29. Portuguese AJ, Huang JJ, Jeon Y, et al. Real-world comparison of lisocabtagene maraleucel and axicabtagene ciloleucel in large B-cell lymphoma: an inverse probability of treatment weighting analysis with 3-year follow-up. Haematologica 2025;110:2040-54. [Crossref] [PubMed]
30. Jacobson CA, Munoz J, Sun F, et al. Real-World Outcomes with Chimeric Antigen Receptor T Cell Therapies in Large B Cell Lymphoma: A Systematic Review and Meta-Analysis. Transplant Cell Ther 2024;30:77.e1-77.e15. [Crossref] [PubMed]
31. Maziarz RT, Bishop MR, Tam CS, et al. Five-Year Analysis of the JULIET Trial of Tisagenlecleucel in Patients With Relapsed/Refractory Large B-Cell Lymphoma. J Clin Oncol 2026;44:86-91. [Crossref] [PubMed]
32. Shune L, Frigault MJ, Riedell PA. CAR-T cell therapy in older adults with relapsed/refractory LBCL: benefits and challenges. J Immunother Cancer 2025;13:e009793. [Crossref] [PubMed]
33. Brooks TR, Zabor EC, Bedelu YB, et al. Real-world outcomes of patients with aggressive B-cell lymphoma treated with epcoritamab or glofitamab. Blood 2025;146:2177-88. [Crossref] [PubMed]
34. Plaks V, Rossi JM, Chou J, et al. CD19 target evasion as a mechanism of relapse in large B-cell lymphoma treated with axicabtagene ciloleucel. Blood 2021;138:1081-5. [Crossref] [PubMed]
35. Schuster SJ, Huw LY, Bolen CR, et al. Loss of CD20 expression as a mechanism of resistance to mosunetuzumab in relapsed/refractory B-cell lymphomas. Blood 2024;143:822-32. [Crossref] [PubMed]
36. Crochet G, Iacoboni G, Couturier A, et al. Efficacy of CAR T-cell therapy is not impaired by previous bispecific antibody treatment in large B-cell lymphoma. Blood 2024;144:334-8. [Crossref] [PubMed]
37. Sorin M, Okde R, Goulet M, et al. Chimeric antigen receptor T-cell therapy and bispecific antibody sequence for large B-cell lymphoma: a systematic review and meta-analysis. Lancet Haematol 2026;13:e86-97. [Crossref] [PubMed]
38. Iacoboni G, Morschhauser F. Selecting the best treatment approach and optimizing sequencing strategies in large B-cell lymphoma. Hematology Am Soc Hematol Educ Program 2025;2025:482-8. [Crossref] [PubMed]
39. Salles G, Duell J, González Barca E, et al. Tafasitamab plus lenalidomide in relapsed or refractory diffuse large B-cell lymphoma (L-MIND): a multicentre, prospective, single-arm, phase 2 study. Lancet Oncol 2020;21:978-88. [Crossref] [PubMed]
40. AbbVie Announces Topline Results for Epcoritamab (DuoBody® CD3xCD20) from Phase 3 EPCORE® DLBCL-1 Trial in Patients with Relapsed/Refractory Diffuse Large B-cell Lymphoma (DLBCL). 2026. Available online: https://news.abbvie.com/2026-01-16-AbbVie-Announces-Topline-Results-for-Epcoritamab-DuoBody-R-CD3xCD20-from-Phase-3-EPCORE-R-DLBCL-1-Trial-in-Patients-with-Relapsed-Refractory-Diffuse-Large-B-cell-Lymphoma-DLBCL#:~:text=About%20the%20EPCORE%C2%AE%20DLBCL,www.clinicaltrials.gov/