Approximately one-third of patients with diffuse large B-cell lymphoma (DLBCL) and high-grade B-cell lymphoma (HGBCL) do not achieve remission or relapse following receipt of first-line therapy.1 Patients with relapsed or refractory (R/R) disease have a poor prognosis, particularly those who are ineligible for, or relapse after, autologous stem cell transplant.2 3 Tisagenlecleucel (Kymriah) is a CD19-directed, autologous chimeric antigen receptor (CAR)-T-cell therapy, approved for adult patients with R/R DLBCL after ≥2 lines of treatment, based on the pivotal JULIET trial (NCT02445248).4–6 An earlier report from the Center for International Blood and Marrow Transplant Research (CIBMTR) cellular therapy registry showed that real-world use of tisagenlecleucel in patients with lymphoid malignancies (155 patients with non-Hodgkin's lymphoma (NHL); median follow-up of 11.9 months (range: 3.8–19.0)) resulted in similar efficacy and improved safety outcomes compared with reports of clinical trials.7 Here, we expand on this work, reporting extended real-world outcomes, including more patients and longer follow-up than the previous analysis,7 specifically for patients with R/R DLBCL/HGBCL. We also further analyse these real-world CIMBTR data to assess the impact of pre-infusion clinicopathologic and treatment characteristics on efficacy and safety outcomes.

In this noninterventional, prospective study, clinical data on tisagenlecleucel were reported to the CIBMTR by treatment centres across the USA, Canada and Israel. Tisagenlecleucel manufacturing data were provided by Novartis. CIBMTR data collection7 and the tisagenlecleucel manufacturing process8 have been described previously. All patients signed informed consent to share data with the CIBMTR, and the study was approved by the institutional review board of the National Marrow Donor Program/Be The Match.7 Eligible patients with R/R DLBCL/HGBCL received commercial tisagenlecleucel after 1 May 2018.9 The infused set included all patients who received tisagenlecleucel. Efficacy and safety sets included patients with completed Day 100 efficacy and safety forms, respectively. Outcomes were assessed in all patients and in subgroups based on pre- and post-infusion characteristics (see online supplemental methods for additional details). Outcomes included overall response rate (ORR), complete response (CR) rate, progression-free survival (PFS), duration of response (DOR), overall survival (OS), cytokine release syndrome (CRS) and immune effector-cell-associated neurotoxicity syndrome (ICANS). Endpoint definitions and statistical analyses are described in online supplemental methods.

As of 4 May 2022, 1159 patients had received tisagenlecleucel (infused set), with a median follow-up of 20.9 months (range: 0.3–46.4 months). The median turnaround time (receipt of apheresis product to shipment) for tisagenlecleucel was 26 days (IQR 24–30). Of all infused patients, 189 patients (16%) received tisagenlecleucel that was outside of release specifications and were included in the analyses.

The median age at baseline was 67.1 years, and most patients had DLBCL (81.4%), an Eastern Cooperative Oncology Group performance status (ECOG PS) of 0–1 (83.4%) and ≥3 prior lines of therapy (61.0%). Patients who achieved CR prior to tisagenlecleucel infusion (n=78) had received a variety of prior therapies, which included systemic therapy (n=71, 91.0%), radiation therapy (n=22, 28.2%), intrathecal therapy (n=21, 26.9%) and surgery (n=5, 6.4%). Among patients who achieved a CR prior to infusion, 47.4% had normal LDH (<upper limit of normal (ULN)) at baseline. Additional baseline characteristics can be found in online supplemental table 1.

In the efficacy set (n=968; median follow-up: 23.2 months, range: 3.1–46.4 months), ORR was 59.5% (95% CI 56.3 to 62.6) and CR rate was 44.5% (95% CI 41.4 to 47.7). Among patients with active disease prior to tisagenlecleucel infusion (n=892), ORR was 57.4% (CR 41.3%). Among patients who were in CR prior to tisagenlecleucel infusion (n=71), 87.3% remained in CR after infusion. Median PFS, DOR and OS for the overall population were 4.1 months (95% CI 3.5 to 4.9), 27.6 months (95% CI 17.4 to not estimable (NE)) and 16.4 months (95% CI 14.6 to 21.0), respectively (figure 1), with 24 month rates of 28.4% (95% CI 24.7 to 32.1), 52.6% (95% CI 46.9 to 58.0) and 43.6% (95% CI 39.1 to 48.0), respectively (table 1). While the overall patient population achieved positive outcomes, patients who achieved a CR after tisagenlecleucel infusion showed better PFS, DOR and OS (figure 1, table 1).

Real-world efficacy outcomes after tisagenlecleucel infusion. Kaplan–Meier estimates for (A) progression-free survival, (B) duration of response and (C) overall survival in the efficacy set. For PFS and DOR, eight patients in ongoing CR/PR were censored at the time of post-infusion consolidative haematopoietic stem cell transplantation; for OS, patients were not censored at the time of post-infusion HSCT. For DOR, time is relative to the onset of the first remission, 1 month=30.4 days. For PFS and OS, time is relative to the first infusion, 1 month=30.4 days. Nominal P values are reported and calculated by using the log-rank test to compare the best overall response (BOR) of CR vs BOR of PR. P values are not adjusted for multiplicity. CR, complete response; NE, not estimable; PR, partial response.

Median and month 6, 12 and 24 PFS, DOR and OS rates in all patients and in patients who achieved BOR of CR

Univariable and multivariable analyses of efficacy outcomes are shown in figure 2A and online supplemental table 2. Of note, achieving CR prior to infusion was associated with higher odds of remaining in CR compared with patients who had active disease at infusion and a lower hazard ratio (HR) for progression; prior autologous or allogeneic haematopoietic stem cell transplantation (HSCT) was associated with higher odds of a response and lower HR for progression and death; elevated lactate dehydrogenase (LDH) prior to infusion was associated with a higher HR for relapse and death; low International Prognostic Index score was associated with a lower HR for progression and death; and lymphodepleting chemotherapy (LDC) regimen of bendamustine was associated with lower odds of a response. Kaplan–Meier survival curves are shown for the subgroups of patients with factors that were found to be significant in the multivariable analyses for DOR (online supplemental figure 1), PFS (online supplemental figure 2) and OS (online supplemental figure 3). The cumulative incidence of deaths among patients who were in remission at the time of death is shown in online supplemental figure 4. Overall, the infusion of tisagenlecleucel that was outside of product release specifications did not have an impact on efficacy or safety.

Effects of patient demographics and disease characteristics on efficacy and safety outcomes after tisagenlecleucel infusion. Multivariable analyses of patient demographics and disease characteristics on (A) efficacy outcomes in the efficacy set and (B) safety outcomes in the safety set with tisagenlecleucel. Multivariable logistic regression model estimates are presented for ORR, any cytokine release syndrome (CRS), CRS grade of ≥3, any immune effector-cell-associated neurotoxicity syndrome (ICANS) and ICANS grade ≥3; multivariable Cox regression model estimates are presented for DOR, OS and PFS. CRS and ICANS were graded according to the American Society for Transplantation and Cellular Therapy criteria and were based on the maximum grade that occurred within 100 days of tisagenlecleucel infusion. For each comorbidity group, ‘No’ indicates the absence of the specific comorbidity of interest. a71/968 (7.3%) patients were in CR at the time of infusion in the efficacy evaluable set. b73/990 (7.4%) patients were in CR at the time of infusion in the safety evaluable set. CR, complete response; DLBCL, diffuse large B-cell lymphoma; DOR, duration of response; ECOG, Eastern Cooperative Oncology Group; HGBCL, high grade B-cell lymphoma; HSCT, haematopoietic stem cell transplantation; IPI, International Prognostic Index; LD, lymphodepleting; LDH, lactate dehydrogenase; ORR, overall response rate; OS, overall survival; PFS, progression-free survival.

In the safety set (n=990; median follow-up: 23.1 months, range: 3.1–46.4), the safety profile presents no new or unexpected safety findings. Clinically relevant adverse events (AEs; CRS, ICANS, prolonged cytopenias, infections) are shown in table 2.

Clinically relevant adverse events of interest

Univariable and multivariable analyses of safety outcomes are described in figure 2B and online supplemental table 3. Of note, ECOG PS of ≥2 was associated with higher odds of high-grade CRS and ICANS; elevated LDH (>ULN) prior to infusion and ≥3 prior lines of therapy were associated with higher odds of any-grade CRS and ICANS. Interestingly, bendamustine LDC was associated with lower odds of any-grade CRS and ICANS. Among patients who received fludarabine-based chemotherapy and those who received bendamustine, the proportions of deaths were comparable (48/825 (5.8%) and 5/127 (3.9%), respectively); the most common cause of death was infection (24 (2.9%) and 2 (1.6%), respectively).

This noninterventional, prospective study of the CIBMTR registry represents the largest and most recent dataset of over 1100 patients with R/R DLBCL/HGBCL treated with tisagenlecleucel in the real-world setting, building on the previous report.7 Overall, this study confirms that treatment with tisagenlecleucel results in durable efficacy for a subset of patients, particularly those who achieved a BOR of CR (median DOR of 35.2 months and median PFS of 27.6 months). In these patients, durable responses were noted, with a 61% probability of remaining relapse-free 24 months after infusion. Importantly, efficacy outcomes were comparable between CIBMTR and the JULIET trial despite less restricted use of tisagenlecleucel in the real-world setting; patients in the CIBMTR study were older than in JULIET with a median age of 67 years vs 56 years and 58% vs 23% were ≥65 years old at baseline.5

Several pre-infusion clinicopathologic and treatment characteristics were associated with efficacy and safety outcomes. Patients achieving CR prior to infusion demonstrated better ORR and PFS following tisagenlecleucel than those with active disease. This is consistent with a recent report showing that achievement of CR prior to CAR-T-cell infusion results in superior survival compared with patients who had persistent disease.10 The finding that elevated LDH (>ULN) prior to infusion is also associated with lower post-CAR-T efficacy supports the use of bridging therapy to normalise LDH levels prior to infusion in patients receiving CAR-T-cell therapy.10–13 More favourable outcomes were observed in patients with prior autologous or allogeneic HSCT and may be due to selection bias for more fit patients with disease that is more likely to be relapsed rather than refractory prior to tisagenlecleucel infusion.14 Additionally, patients with ECOG PS of ≥2 and elevated LDH prior to infusion experienced higher rates of CRS and ICANS, which supports either closer outpatient or inpatient monitoring following tisagenlecleucel infusion for patients with these characteristics to facilitate timely identification and treatment of these toxicities. Of note, the use of bendamustine LDC was associated with lower toxicity and response rates but similar survival outcomes, which is in contrast with a recent report wherein bendamustine did not affect ORR.15 Future studies are needed to understand what might be driving these observations. Finally, there were no significant differences in efficacy or safety outcomes based on older age or organ comorbidities, suggesting that tisagenlecleucel might be a viable therapeutic option for these patients. Overall, patient characteristics and prior therapies had minimal effects on safety and efficacy outcomes; the positive response seen in the overall population demonstrates that these characteristics should not prevent patients from considering CAR-T-cell therapy.

The assessment of efficacy and safety of CAR-T-cell therapies can often influence treatment decisions. Indirect comparison of pivotal trial data suggests a higher initial response rate but a higher risk of toxicities, particularly ICANS, for axicabtagene ciloleucel vs tisagenlecleucel.5 16 17 A real-world analysis of axicabtagene ciloleucel (n=275) with a 58 month median follow-up reported acceptable long-term survival outcomes (median OS 34.9 months; 5 year OS 40.3%; median PFS 8.7 months; 5 year PFS rate 28.5%), but did not report short-term post-infusion AEs.18 These real-world efficacy data for axicabtagene ciloleucel were consistent with long-term efficacy reported in the pivotal ZUMA-1 clinical trial.19 Another real-world analysis of CIBMTR data (n=1297) for axicabtagene ciloleucel in R/R large B-cell lymphoma (LBCL) reported 8% grade ≥3 CRS and 24% grade ≥3 ICANS,20 suggesting that a significant proportion of patients treated with commercial axicabtagene ciloleucel experienced high-grade ICANS. In the primary analysis of the pivotal TRANSCEND NHL 001 clinical trial in patients with R/R LBCL in the third-line setting, lisocabtagene maraleucel has shown high response rates (ORR 73%, CR 53%) with a low rate of grade ≥3 AEs, especially ICANS (10%) and CRS (2%).21 22 A real-world analysis of CIBMTR data in patients with R/R LBCL who received lisocabtagene maraleucel (n=323) reported similar response and CRS/ICANS rates.23 A study that used a matching-adjusted indirect comparison analysis, however, showed that although patients who received lisocabtagene maraleucel had a higher ORR compared with those given tisagenlecleucel, there were no significant differences in CR rates observed.24 These real-world data are informative to clinicians making recommendations regarding CAR-T-cell therapy product choice to patients, and consideration of response rates, estimates of survival outcomes based on follow-up time and toxicities is warranted.

It is worth noting that the real-world safety profile of tisagenlecleucel appears to be more favourable than that reported in JULIET. Here, we find lower rates of grade ≥3 CRS (6.0% vs 13.5% in JULIET)25 and ICANS (7.4% vs 12.6% in JULIET),26 all by the American Society for Transplantation and Cellular Therapy grading criteria. It is worth noting that, in JULIET, tocilizumab was used mainly for grade 3/4 CRS,25 whereas here tocilizumab was mostly used to treat grade 1/2 CRS (table 2). The lower rates of severe CRS and ICANS observed in the real-world setting likely reflect general advancements in toxicity management and the trend for earlier intervention.20 27–29

Key limitations of the current analysis include the voluntary nature of the data collection in the real-world setting that resulted in differences in the completeness of data and timing of assessments.

In conclusion, patients with R/R DLBCL/HGBCL receiving commercial tisagenlecleucel and captured in the CIBMTR registry experienced durable efficacy and well-characterised safety outcomes including older patients (≥65 years) and those with organ comorbidities. Detailed analysis of the impact of baseline characteristics on efficacy and safety outcomes may inform management for future patients treated with tisagenlecleucel. Further analysis from this post-authorisation study will yield additional insights into real-world outcomes for patients infused with tisagenlecleucel.

Data are available upon reasonable request. CIBMTR supports accessibility of research in accordance with the National Institutes of Health data-sharing policy and the National Cancer Institute Cancer Moonshot public access and data-sharing policy. The CIBMTR only releases de-identified datasets that comply with all relevant global regulations regarding privacy and confidentiality.

Not applicable.

This study involves human participants and was approved by the CIBMTR, a collaborative working group of more than 380 international transplant and cellular immunotherapy centres managed by the Medical College of Wisconsin (MCW) and the National Marrow Donor Program (NMDP). All centres must obtain institutional review board (IRB)-approved informed consent from recipients to allow data submitted to the Research Database to be used for research studies, regardless of the level of data the centre submits to the CIBMTR. All US centres must have IRB approval for their centre for the Research Database Research protocol. The CIBMTR tracks the IRB approval for the Research Database at each participating centre. The centre’s IRB approval for this protocol must be current at all times. Participants gave informed consent to participate in the study before taking part.

The authors sincerely thank the patients who participated in the CIBMTR registry and their families, as well as all participating treatment centres, the treating physicians and the support staff.