Validated PET-CT metabolic tumor volume threshold identifies high-risk relapsed/refractory LBCL and predicts 12-month toxicity-free progression-free survival Post–CAR-T therapy

Abstract Introduction: Tumor burden (TB) is a well-established prognostic factor in relapsed/refractory large B cell lymphoma (R/R LBCL) treated with CAR-T therapy, and metabolic tumor volume (MTV) derived from 18F-FDG PET-CT offers a comprehensive measure of total metabolically active disease. In this study of a large cohort, we aimed to establish a clinically meaningful risk stratification cutpoint for baseline MTV and applied findings from a discovery cohort to an independent validation cohort. We also propose 12-month toxicity-free progression-free survival (TFPFS12) as a composite endpoint to integrate treatment efficacy and high-grade toxicity into a unified clinical outcome measure. Methods: We retrospectively identified patients (pts) in our institution from 2017-2024, measuring baseline MTV prior to apheresis using PET Edge, a gradient-based method included in MIM software. All pts had serum glucose 180 mg/dL and received 12-18 mCi 18F-FDG 45-60 minutes prior to imaging acquisition. Pts were randomly assigned to discovery and validation cohorts in a 70: 30 ratio, stratified by year of CAR-T infusion to account for evolving clinical practices. The threshold for high/low MTV was derived using a maximally selected log-rank statistic for PFS in a discovery cohort and confirmed through two-fold cross validation and performance testing in a validation cohort. Kaplan-Meier and Cox proportional hazards models assessed associations between the MTV cutpoint and survival. Logistic regression was used to evaluate treatment response and toxicity with multivariable (MV) adjustment for age, ECOG performance status, disease type, cell of origin, prior autologous stem cell therapy, and number of prior therapy lines. Serial blood samples are collected over 28 days post CAR-T to assess CAR T-cell kinetics (Hamilton, Blood Adv 2024). TFPFS-12, defined as the absence of Grade 3 cytokine release syndrome (CRS) or immune effector cell associated neurotoxicity syndrome (ICANS), progressive or stable disease, or non-relapse mortality at month 12 after CAR T infusion, was analyzed as an exploratory endpoint. Results: We identified 239 pts (92% (n=220) axicabtagene ciloleucel, 7% (n=17) lisocabtagene maraleucel, and 1% (n=2) tisagenlecleucel) with R/R LBCL. A data-driven cutpoint of 125. 5 mL was identified in the discovery cohort (n=167) and was independently validated in the validation cohort (n=72) McNemar's test: p=0. 5. For the entire cohort, we analyzed high (n=61, 26%) vs. low (n=178, 74%) MTV; baseline characteristics were similar, including age, ECOG performance status, histologic subtype, and line of therapy. Median follow-up was 39. 5 months. Pts with high MTV had significantly worse outcomes; median PFS was 4. 2 vs. 36. 3 months (HR 2. 65, p0. 01), and OS was 20. 1 vs. 65. 5 months (HR 2. 89, p0. 01) in high vs. low MTV groups, respectively. Complete response rate was 46% in the high MTV cohort vs 79% in the low MTV cohort (p0. 01). There were no significant differences in CRS or ICANS incidence between MTV groups (p=0. 2/0. 2). Pre-apheresis LDH had a positive correlation with baseline MTV (r=0. 41; p0. 01). Similar cutpoints were derived for TLG (1593. 8 SUV*mL) and SUVmax (9. 9), but only TLG's cutpoint was prognostic for PFS (p0. 01) ; SUVmax was not (p=0. 09). TLG and MTV agreed in 92% of cases. TFPFS12 rate across all pts was 54% and was achieved in 30% of high MTV pts (n=18) vs. 52% in the low MTV group (n=91). High MTV was independently associated with worse TFPFS12 on MV analysis (OR 0. 36, 95% CI 0. 17–0. 76, p=0. 01), adjusted for relevant clinical covariates. CAR-T expansion kinetics did not differ by MTV category at days 7 (p=0. 2) or 14 (p=0. 3). MV linear regression fit for log-10 transformed median area-under-the-curve over 28 days (AUC-28), MTV group, and relevant clinical covariates did not show a significant association between baseline MTV and peak CAR (p=0. 3) or AUC-28 (p=0. 2). Conclusions: This is the largest study to identify through robust validation that a baseline MTV cutpoint of 125 mL independently predicts PFS and OS in R/R LBCL treated with CAR-T, defining a high-risk subgroup with unmet therapeutic need. These findings support using MTV for risk-stratification in clinical trials and for identifying pts who may benefit from intensified or novel regimens. Integrating MTV-guided risk groups with composite endpoints like TFPFS-12 could refine efficacy and safety assessments.