Pre-lymphodepletion CRP and ferritin risk stratification as a predictor of early immune toxicities and hematologic complications after CAR-T therapy: A propensity-matched cohort study.

Abstract

7030 Background: Inflammation before CAR-T has been linked to higher rates of immune effector cell-associated toxicities and worse outcomes. A pragmatic pre-lymphodepletion risk stratification using C-reactive protein (CRP) and ferritin has been proposed to identify patients at elevated risk. We used the TriNetX Network to evaluate whether CRP/ferritin risk groups predict clinically relevant early outcomes following CAR-T in a real-world setting. Methods: We performed a retrospective study using TriNetX Network. Adults (≥18 years) treated with CAR-T (day 0) were included if CRP & ferritin were available in the pre-lymphodepletion window (day −20 to −5). Patients were classified as low risk (CRP <4 and ferritin <400), intermediate risk (CRP ≥4 with ferritin <400 or CRP <4 with ferritin ≥400), and high risk (CRP ≥4 and ferritin ≥400). Propensity score matching was performed using demographics, comorbidities, transplant status, malignancy diagnoses & lymphodepletion therapy. Outcomes were assessed from day 0 to +30. We analyzed outcomes using odds ratio (OR) with 95% confidence interval (CI) and time-to-event comparison using log-rank p-value. Results: After propensity matching, high vs low risk cohorts included 246 patients per group. For intermediate vs low risk, cohorts included 298 per group. High-risk patients had higher odds of cytokine release syndrome (CRS) (OR 1.56, 95% CI 1.1–2.2; p=0.04) and immune effector cell-associated neurotoxicity syndrome (ICANS) (OR 1.86, 95% CI 1.1–3.1; p=0.01) versus low risk. In intermediate vs low risk, ICANS remained increased (OR 1.69, 95% CI 1.1–2.7; p=0.02), while CRS showed a nonsignificant trend (OR 1.36, 95% CI 0.98–1.9; p=0.06). High risk cohort was associated with greater supportive care utilization, including higher odds of G-CSF use (OR 2.93, 95% CI 1.5–5.8; p<0.01) and blood transfusion (OR 2.65, 95% CI 1.2–5.9; p=0.01) versus low risk. High vs low-risk was associated with higher odds of platelets <50 (OR 3.59, 95% CI 1.98–6.5; p<0.01) and hemoglobin <8 (OR 2.81, 95% CI 1.3–5.9; p<0.01). Neutropenia (ANC <500) was not significantly different. In intermediate vs low risk, cytopenia-related endpoints were not significantly different. All-cause mortality analysis was not performed due to low event rates. Conclusions: In a TriNetX propensity-matched real-world cohort, a simple pre-lymphodepletion CRP and ferritin stratification identified CAR-T recipients at significantly higher risk for early immune toxicities (CRS, ICANS) and clinically meaningful hematotoxicities (severe thrombocytopenia and anemia) within 30 days, along with increased need for G-CSF & transfusion support. These findings support using simple pre-infusion inflammatory biomarkers to guide counseling, risk-adapted monitoring, and proactive supportive care strategies in routine practice.