Introduction:

The InflaMix (INFLAmmation MIXture) model classifies the disease of patients with R/R LBCL as inflamed or noninflamed using routine pre-infusion laboratory tests. In the published cohorts (Raj S, et al. Nat Med 2025) comprised primarily of patients with third-line or later (3L+) LBCL, this classification predicts response to CAR T cell therapy. As liso-cel is a robust treatment option in second-line (2L) LBCL, we sought to understand the value of InflaMix stratification for this population. Here, we retrospectively compared efficacy outcomes by treatment arm in inflamed versus noninflamed groups in patients with 2L LBCL who received liso-cel or historical standard of care (SOC) in TRANSFORM.

Methods

For patients with available data, albumin, hemoglobin, AST, alkaline phosphatase, C-reactive protein, and LDH values were collected before leukapheresis and at the time of infusion from patients with 3L+ LBCL (n = 256; TRANSCEND NHL 001, NCT02631044) and before leukapheresis from patients with 2L LBCL (n = 127; TRANSFORM, NCT03575351). InflaMix, which was trained on the published derivation cohort with laboratory values taken at infusion (n = 149; 98% had 3L+ LBCL; Raj S, et al. Nat Med 2025), classified all patients as either inflamed or noninflamed. Endpoints were CR rate, ORR, PFS, and rates of cytokine release syndrome (CRS) or neurological events (NEs). PFS was analyzed using KM estimates with log-rank tests. Categorical outcomes were compared with chi-square tests of association.

Results

Efficacy outcomes in patients with 3L+ LBCL were compared stratifying based on inflamed (72%) versus noninflamed (28%) status at the time of infusion or before leukapheresis and were consistent with published results on the InflaMix model. The 2L LBCL analysis was based on laboratory features before leukapheresis to enable comparison between treatment arms. In patients with 2L LBCL, InflaMix classified 109/127 patients (86%) as noninflamed and 18/127 patients (14%) as inflamed. A significantly greater proportion of patients were classified as noninflamed in the 2L setting compared with the 3L+ setting (P = 0.004).

Among patients in the liso-cel arm, median PFS was similar between inflamed and noninflamed groups (33.2 versus 30.9 months; P = 0.29). In contrast, InflaMix was predictive in the SOC arm, with a median PFS of 2.1 months for the inflamed group versus 5.7 months for the noninflamed group (P = 0.022).

Across both InflaMix groups, liso-cel outperformed SOC, resulting in longer median PFS, consistent with clinical results from TRANSFORM. In the noninflamed group, median PFS was 30.9 months with liso-cel treatment versus 5.7 months with SOC (P = 0.003). Inflamed patients treated with liso-cel had a median PFS of 33.2 months versus 2.1 months for those treated with SOC (P = 0.146); the lack of statistical significance was likely due to limited patient numbers in the inflamed group for both arms. Differences in CR rate and ORR mirrored PFS results favoring liso-cel in both groups. Notably, liso-cel treatment in the inflamed group resulted in longer median PFS than SOC in the noninflamed group (33.2 versus 5.7 months). InflaMix groups were not associated with incidence of any grade or grade ≥ 3 CRS or NE.

Conclusions

InflaMix is consistently prognostic in patients with LBCL treated with CAR T cell therapy in later-line settings. However, in 2L LBCL, InflaMix stratified outcomes in patients treated with SOC, but not in those treated with liso-cel. The results from this study support prior data showing InflaMix stratification of patients treated with bispecific antibodies (Magno G, et al. Blood 2024), further suggesting that the stratification may be broadly prognostic for patients with LBCL. Importantly, liso-cel outperformed SOC in 2L regardless of inflammatory status, reinforcing liso-cel as a robust treatment option delivering deep and durable efficacy for a broad population of patients with 2L LBCL.

This content is only available as a PDF.
2025
Sign in via your Institution