【Nat Med】A real-world analysis of axi-cel versus tisa-cel treatment of 809 cases of R/R DLBCL

CD19 CAR T cell therapy has shown impressive efficacy and controlled toxicity
in the treatment of a variety of histological subtypes of lymphoma, such as mantle-cell lymphoma, follicular lymphoma, and diffuse large B-cell lymphoma 。 Tislecagenleucel (tisa-cel) and axicabtagene ciloleucel (axi-cel) are two CAR T products, both originally approved for the treatment ≥ 3-wire DLBCL; Tisa-cel is a second-generation CAR T based on the 4-1BB co-stimulating domain, while axi-cel is CD28-based
.

Over the past 2 years, many real-world reports have confirmed that CAR T achieves high remission rates, prolonged duration of remission and survival in DLBCL, and it is remarkable that despite rigorous patient selection in clinical trials, efficacy in non-trial settings appears to be similar to results in key studies, and real-world toxicity appears to be significantly reduced
due to early remission strategies using anti-interleukin-6 and steroids.

Many parameters can affect the efficacy and safety of CAR T, such as the use of bridging therapy to control disease progression during product production, tumor bulk, or delay
between leukocyte monotherapy and infusion.

Crude response rates and safety reports from clinical trials suggest that axi-cel is more effective and toxic than tisa-cel, but this conclusion may be misleading due to large differences between study designs: (1) ZUMA-1 enrolled patients with primary mediastinal B-cell lymphoma (PMBCL), but JULIET was not; (2) Fludarabine and cyclophosphamide in ZUMA1 as a pretreatment regimen are higher doses; (3) During CAR T production, the JULIET study allows the use of bridging chemotherapy to control disease progression, while the ZUMA-1 study does not
.

Matching-adjusted indirect comparisons (MAICs) have been reported to compare different CAR T products, using inpidual patient data (IPD) from one study and trial-level data from another study to form a population-corrected indirect comparison
of treatments.

Since 2019, French health authorities have required extensive data collection for every patient who is theoretically indicated for the treatment of CAR T, so the Association for the Study of Lymphoma (LYSA) and the Lymphoma Academic Research Organization (LYSARC) have established the DESCAR-T registry to meet this regulatory requirement and conduct a comprehensive RWE study
.

Given that the efficacy and safety comparison between tisa-cel and axi-cel is insufficient, Professor Emmanuel Bachy of Hospices Civils de Lyon in France led a matching comparison based on individual case data (IPD) based on data from all French DLBCL patients treated with commercially available CAR T in the DESCAR-T registry
.

Research results

Patients and outcomes

Between December 2019 and October 2021, 809 R/R DLBCL patients from 23 French centres who had previously undergone ≥2-line treatment had commercial CAR T orders for axi-cel or tisa-cel and were registered in DESCAR-T (Figure 1).

Patient characteristics are shown in Table 1
.

The median age of patients was 63 years, and 61% of patients were male
.

The median follow-up was 13 months, compared with the predicted median OS of 17 months from the subscription of CAR T (Figure 2a blue line
).

Of the 809 patients who ordered CAR T products, 60 progressed or died between leukocyte solitary collection and driffing, and another 20 did not perform leaching due to physician decision or other reasons (Figure 1), and these 80 patients (tisa-cel n = 38, axi-cel n = 42) had very poor and similar OS expectations based on CAR T products (axi-cel or tisa-cel) (P = 0.

Preference score matching

Propensity scoring is the conditional probability that a patient with observed covariates will receive one treatment or another, and the purpose of a propensity score match (PSM) is to balance the covariates between the axi-cel and tisa-cel groups to account for all possible measured confounding variables (i.

Even so, patients in the tisa-cel group were slightly more severity than patients in the axi-cel group, with a higher proportion of aaIPI scores of 2-3 (57.

Inverse probability handles weighting

Inverse Probability of Treatment Weighting (IPTW) is another method in which patients are assigned weights
based on the inverse probability of receiving one treatment or estimating another treatment by propensity score.

IPTW is used to support the results of PSM analysis and can also be appropriately compared
between two patient populations receiving axi-cel or tisa-cel.

Safety in people with a propensity score match

In the matching population (n=418), 180/209 patients in the axi-cel group (86.
1%) developed cytokine release syndrome (CRS) at any level, compared with 158/209 patients (75.
6%) in the tisa-cel group (Table 3
).
。 Most CRS were Level 1 or Level 2, regardless of CAR T products, and the incidence of Level 1-2 CRS in the axi-cel group was significantly higher than in the tisa-cel group (80.
9% vs 66.
5%; P < 0.
001), while there was no significant difference ≥ level 3 CRS (9.
1% vs 5.
3% in the tisa-cel and axi-cel groups, respectively; P= 0.
130）
。

In terms of ICANS, the frequency of ICANS in the axi-cel group was significantly higher than that in the axi-cel group (i.
e.
, ≤ level 2) and severe (i.
e.
, ≥ level 3
) than in the tisa-cel group (Table 3).
Grade 1-2 ICANS appeared after infusion of axi-cel in 35% of patients, compared with 19.
1% in the tisa-cel group (P< 0.
001); ICANS at ≥ level 3 were 29 (13.
9%) and 6 (2.
9%) (P< 0.
001),
respectively.

The haematological toxicity of axi-cel is also significantly more frequent and severe than that of tisa-cel (Table 3
).
At 1 month after CAR T infusion, 64.
6% of patients in the axi-cel and tisa-cel groups developed any level of cytopenia, and 34.
0% and 12.
4% of patients developed ≥grade 3 cytopenia (Table 3).

Hematological toxicity after axi-cel transfusion was significantly elevated compared with tisa-cel and was consistent across all hematologic lineages (i.
e.
, neutropenia, anemia, and thrombocytopenia; Table 3).

The same is true for long-term cytopenia 3 months after CAR T transfusion (Table 3
).
It is important to note that significant differences in cytopenia between patients treated with axi-cel or tisa-cel were not observed prior to debridement, meaning that excessive hematologic toxicity of axi-cel cannot be attributed to significant baseline differences
.

Grade 5 CRS associated with axi-cel was not observed in the match population, while 2 cases of tisa-cel associated CRS were observed
.
One case of Level 5 ICANS was reported in the axi-cel group, but not in the tisa-cel group
.
No other grade 5 adverse events
directly related to CAR T infusion were reported in the matching population.

Subgroup analysis

First, outcomes
were assessed in the PSM population based on age (≤ 70 years and > 70 years).
PFS after axi-cel infusion was significantly longer than tisa-cel in patients aged ≤ 70 years and > 70 years: ≤the median PFS, axi-cel and tisa-cel were 5.
9 months and 3.
1 months (P = 0.
0128) in patients aged > 70 years, respectively, and not achieved and 3 months (P = 0.
0026)
in patients aged 70 years, respectively 。 For OS, survival of axi-cel was also longer in both age groups, but statistical significance was not achieved ≤ patients aged 70 years (P = 0.
0779 ≤ 70 years of age, P = 0.
0167 in the > 70 age group).

Second, because the efficacy of CAR T in the case of large tumor volume may depend on the type of co-stimulating domain, efficacy
was assessed based on the maximum diameter of the largest lymph node or extranodal mass (i.
e.
, ≤5 cm or > 5 cm) that is a factor associated with tumor bulk 。 Results Compared with tisa-cel, PFS after axi-cel infusion was significantly prolonged and independent of tumor volume: median PFS was 7.
9 months in the axi-cel group and 3.
5 months in the tisa-cel group (P= 0.
0164) in the absence of large masses; In the presence of large lumps, it was 8.
2 months and 2.
1 months (P = 0.
0023),
respectively.
Therefore, the outcome of axi-cel was better than tisa-cel regardless of tumor volume, which also applies to OS
.

conclusion

This study analyzed 809 patients who ordered CAR T and who had real-world DLBCL
in a second or subsequent recurrence.
Across the patient cohort, the median OS of patients with CAR T subscription and CAR T infusion was 17 months and 19 months,
respectively.
Notably, in the 1:1 match population of 418 patients considered after rigorous PSM statistical methods, IRA-cel and axi-cel had ORR/CRR of 66%/42% and 80%/60%, respectively, and this result reflected the response rates of two key studies: JULIET and ZUMA-1 (52%/40% and 82%/58%, respectively).

。 Similarly, the median OS in the tisa-cel group was 11.
2 months, while the median OS in the axi-cel group was not reached, echoing
the recently updated median OS of the JULIET and ZUMA-1 studies (11.
1 months and 25.
8 months, respectively).

In summary, this study found that in R/R DLBCL ≥3-line therapy, axi-cel was more effective and more toxic than tisa-cel
.
While this result still needs to be confirmed by other large RWE studies with similar statistical methods to explain the imbalance between patient characteristics, the findings of the Larger Study can help refine the choice
of CAR T products for specific patients based on the trade-off between safety and efficacy.

References

Emmanuel Bachy ,et al.
A real-world comparison of tisagenlecleucel and axicabtagene ciloleucel CAR T cells in relapsed or refractory diffuse large B cell lymphoma.
Nat Med .
2022 Sep 22.
doi: 10.
1038/s41591-022-01969-y

： ，
。 Video Mini Program like, tap twice to cancel like in viewing, tap twice to cancel in viewing