INSIGHT: A baseline laboratory profile independently predicts outcomes and demonstrates superiority of lisocabtagene maraleucel (liso-cel) over standard of care (SOC) in second-line (2L) large B-cell lymphoma (LBCL) regardless of baseline profile

Abstract

Abstract Introduction: CD19-directed CAR T cell therapies have transformed the treatment of patients with R/R LBCL, but the field is still defining those patients most likely to experience durable efficacy and have a low-risk safety profile. Here, we present INSIGHT, a framework based on unsupervised clustering routine clinical laboratory tests at screening that identifies 2 patient profiles: PRIME and STANDARD. These profiles provide a data-driven context for understanding projected efficacy and safety outcomes for patients receiving liso-cel, empowering more tailored treatment decisions for patients with LBCL. Methods: INSIGHT clinical profiles, PRIME and STANDARD, were defined by unsupervised consensus k-means clustering of common clinical laboratory test results at screening (complete blood count with differential, LDH, metabolic panel/electrolytes, and liver and kidney function markers) from clinical trial data in adults with LBCL treated with CAR T cell therapy or immunochemotherapy; cohorts included TRANSCEND NHL 001 (3L+, NCT02631044), TRANSFORM (2L HSCT eligible, NCT03575351), PILOT (2L HSCT ineligible, NCT03483103), and ROBUST (1L, NCT02285062) (n = 1024; liso-cel, n = 384; immunochemotherapy, n = 640). Progression-free survival (PFS) was compared using Kaplan–Meier analysis, log-rank testing, and multivariable Cox models that included sum of the product of perpendicular diameters (SPD) to confirm independence from radiographic tumor burden. Incidences of CRS werecompared using a chi-square test. Results: Across all treatment lines and modalities, PRIME patients (n = 715) had superior PFS compared to STANDARD patients (n = 309) (hazard ratio [HR], 0.62; P < 10⁻⁴), independent of SPD (HR, 0.57; P < 0.001). PRIME patients receiving liso-cel also had lower incidence of CRS versus STANDARD patients (grade ≥ 2: 11% versus 22%, P = 0.01; grade ≥ 3: 1% versus 3%, P = 0.39). Furthermore, the proportion of PRIME patients increased from 58% in 3L+ to 77% in 2L LBCL. In TRANSFORM (2L LBCL), INSIGHT profiles stratified PFS within the SOC arm (HR, 0.33; P = 0.002) but not the liso-cel arm (HR, 0.77; P = 0.51), reflecting the broad benefit of liso-cel in 2L LBCL. Indeed, STANDARD patients treated with liso-cel had longer median PFS compared with PRIME patients treated with SOC (14.8 versus 6.5 months). From screening to infusion, 77% of STANDARD patients transitioned to PRIME profiles; this subgroup achieved similar median PFS compared to patients who were PRIME throughout (P = 0.27), suggesting a window for pre-infusion optimization. Conclusions: INSIGHT's stratification of patients with LBCL across treatment lines and modalities suggests these profiles capture prognostic information. By using readily available laboratory tests at screening, it provides an opportunity to fine-tune patient management before CAR T cell therapy infusion. PRIME patients had a high likelihood of achieving durable response and a lower risk of having grade ≥ 2 CRS, making them potentially ideal candidates for outpatient CAR T cell therapy administration and monitoring. Importantly, both PRIME and STANDARD patients derived greater benefit from liso-cel than from SOC in 2L, underscoring the advantage of early CAR T cell therapy intervention. Integration of INSIGHT into clinical practice could improve access to CD19-directed CAR T cell therapy and optimize the overall clinical management of patients with LBCL.