Prognosis Score for Overall Survival with CAR-T Therapy in Aggressive Lymphoma

Background: Since 2017, six chimeric antigen receptor T cell (CAR-T) therapies have been approved by the FDA for hematologic malignancies. Despite the transformative potential of this therapy, more tools to assist with identifying patients with increased likelihood of benefitting from this therapy will be helpful, particularly given the logistical complexity and socio-economic demands for CAR-T relative to other therapies. Health care resource restriction during the COVID-19 pandemic highlights the need for these tools. We present a simple survival score that uses 3 readily available clinical labs: platelet (plt), absolute lymphocyte count (ALC), and Lactate dehydrogenase (LDH) to predict the risk of dying within 6 months of CAR-T therapy in patients with aggressive lymphoma.

Methods: We conducted a retrospective chart review of patients with aggressive non-Hodgkin lymphoma (NHL) who received FDA-approved CAR-T between Jan 2018 to Jan 2022 at Mayo Clinic Rochester. Demographics and lab variables were selected based on univariate analysis for clinical and statistical significance and tested in a multivariate logistic regression model. Bootstrapping test was used to test the accuracy of the model to predict for death within 6 months of CAR-T infusion. Patients were scored based on the model, and Kaplan-Meier curve for overall survival (OS) based on the score was plotted.

Results: Among a total of 110 pts who received CAR-T, 27 (25%) pts died within the first 6 months post CAR-T infusion (OS ≤ 6 mo). Disease progression was the main cause of death (18/25, 72%), followed by infection (4/25, 16%), CAR-T related (HLH/MAS, 2/25, 8%), second primary malignancy (1/25, 4%) and unknown (2/25, 8%).Baseline demographics were comparable between the OS>6mo and ≤6months groups (Table 1). Patients’ ECOG, Karnofsky performance status and 11 labs at the time of evaluation for CAR-T therapy (initial eligibility assessment, prior to leukapheresis) were compared between those who died from any cause within 6 months of CAR-T infusion and those who did not. Hemoglobin, plt, ALC, absolute monocyte count, C- Reactive Protein, ferritin, and LDH were selected as clinically and/or statistically significant variables for multivariate testing. Multivariate regression with boot-strap testing identified plt, ALC, and LDH as the most predictive variables with 80.9 ± 11.7% accuracy for predicting death within 6 months of CAR-T infusion. Patients were scored 0-3 using these 3 labs, with 1 point assigned for plt ≤ 100 X10⁹/L, ALC ≤ 0.4 X10⁹/L, or LDH > 222 U/L (upper limit of normal). OS by this survival score is shown in Figure 1.

Discussion: Due to the curative potential of CAR-T, patients with broader characteristics than those treated on registration studies have been treated in standard of care practice. While an estimated 5%-10% risk of CAR-T associated deaths in the first 3 months is seen across all patients in clinical trials, predictors for early death after CAR-T in real-world patient populations can provide additional context for pts and providers when selecting treatment. This survival score is important proof of concept that a simple model using readily accessible clinical labs at the time of CAR-T evaluation could provide additional context to help with additional clinical decision-making. Multicenter prospective studies will help define and validate the definitive survival scoring system.