CAR-hematotox score for predicting outcomes among myeloma patients receiving BCMA targeting CART constructs (ide-cel and cilta-cel)

Abstract

Abstract Background CAR-HEMATOTOX score, a cumulative score of key independent laboratory variables, is a clinical surrogate of bone marrow reserve and inflammation that predicts delayed hematotoxicity after chimeric antigen receptor (CAR)-T administration. This tool likely may have predictive value for all CAR T-cell–related toxicities across hematological malignancies. We evaluated the utility of CAR-HEMATOTOX score among myeloma patients treated with two CART constructs targeting the B-cell maturation antigen (BCMA) - Ide-cel and Cilta-cel. Methods 253 CART constructs were used among 247 patients from 2/2018 until 3/2025 at the Winship Cancer Institute of Emory University. Three groups of patients: cohort 1 (research, N=62) cohort 2 (commercial Ide-cel, N=54) and cohort 3 (commercial Cilta-cel, N=137) comprised the study population. 143 patients (ide-cel, N=49, cilta-cel, N=94) received CART for the indication of ≥4 LOT including an immunomodulatory agent, a proteasome inhibitor, and an anti-CD38 monoclonal antibody. We hypothesized that CAR-HEMATOTOX score at the time of apheresis will predict higher grade CRS (≥grade 2), higher grade neurotoxicity (≥grade 2), day 100 mortality, day 365 mortality, progression free survival (PFS) and overall survival (OS). Demographic and clinical characteristics and outcomes data were obtained from our institutional review board-approved myeloma database and with manual abstraction. The 5 baseline hematotoxicity biomarkers were collected for each patient to calculate the HTscore - HThigh ( ≥2) or HTlow (≤1) risk. Results Overall, 68 patients (26.9%) were scored as HThigh [(cohort 1: N=21 (33.9%), cohort 2: N=28 (20.4%), cohort 3: N=19 (35.2%)]. For the overall cohort HThigh did not predict ≥grade 2 CRS or neurotoxicity – Hthigh vs HTlow: [(cohort 1: NS, cohort 2: NS, cohort 3: NS]. However, day 100-mortality was significantly higher for Hthigh vs HTlow (11.8% vs 2.2%, p=0.004). This also was replicated for day 365-mortality - Hthigh vs HTlow (27.9% vs 8.6%, p<0.001). These differences were highly pronounced in cohort 2 both for day 100-mortality - Hthigh vs HTlow (10.7% vs 0.9%, p<0.027) and for day 365-mortality - Hthigh vs HTlow (17.9% vs 4.6%, p<0.030). Next, we looked at the median PFS by the entire cohort – 37.78 months (Hthigh vs HTlow – 40.5 vs 22.1 months, p<0.013). Similar results were shows for median OS – 51.78 months (Hthigh vs HTlow – 53.65 vs 35.38 months, p<0.005). On univariate analysis, hazards ratio (HR) for progression for Hthigh was 1.8 (95% CI 1.12-2.88, p=0.014). Similarly, HR for mortality for Hthigh was 2.03 (95% CI 1.23-3.38, p=0.006). We have further evaluated the weight of each variable for the outcome of mortality (including the ones not in the CAR -HEMATOTOX score). CRP – HR 2.0 (95% CI 0.96-4.5, p=0.06), Hb – HR 2.23 (95% CI 1.33-3.72, p=0.002), ANC – HR 0.951 (95% CI 0.43-2.09, p=0.90) Plts <75K HR 3.26 (95% CI 1.48-7.12, p=0.003) plts 75 - 175K HR 0.951 (95% CI 0.43-2.09, p=0.08). Ferritin, >2000 HR 7.6 (95% CI 2.28 – 25.27, p<0.001) ferritin 650-2000 HR 2.12 (95% CI 1.07-4.02, p=0.03). Lastly, we checked B2M (cut-off 5.5) – HR 2.33 (95% CI 1.23-4.22, p=0.01). Conclusions CAR-HEMATOTOX score, with 5 baseline hematotoxicity biomarkers is a good predictive tool for assessing post CART hematotoxicity. In our analysis, it is not predictive of higher grade CRS or neurotoxicity, but clearly predictive of day-100 and day-365 mortality. It may have additional prognostic value for myeloma patients receiving BCMA CART if it can be refined to include B2-microglobulin. While ferritin appears to carry the highest discriminant value and ANC the lowest, CAR-HEMATOTOX certainly has an opportunity for refinement.