Serum-Based Proteomic Analysis of Cytokine Release Syndrome in Pediatric AML Patients Receiving CART123

Introduction The prognosis for children with relapsed acute myeloid leukemia (AML) is poor, spurring interest in the development of novel therapies. Since a chimeric antigen receptor T-cell (CAR-T) directed against CD19 or CD22 induces high rates of durable remission in children with B-cell acute lymphoblastic leukemia (B-ALL), it has been postulated that a CAR-T directed against myeloid antigens might also be efficacious in children with AML. However, preliminary results from our institution and others show that response rates in AML are low despite high rates of cytokine release syndrome (CRS). Since CRS is often considered a hallmark of CAR-T activity, we hypothesized that broad-based serum proteomics could identify differences in CRS following CAR-T therapy and shed light on mechanisms of resistance in AML compared with B-ALL. Methods Serum samples from pediatric patients treated with CAR-T targeting CD123 (CART123) for AML on trial NCT04678336 were compared with those treated with CAR-T targeting CD19 (CART19) for B-ALL on trials NCT01626495 and NCT02906371 (Diorio et al., Clinical Cancer Research 2022). Baseline samples from day 1 pre-infusion were compared with samples collected at the clinically identified peak CRS day for each patient. Of the 6 AML patients receiving CART123, 5 experienced grade 1-2 CRS. These patients were compared with 12 B-ALL patients that received CART19 with similar CRS severity (grades 1-2). One AML patient that received CD123 experienced high-grade CRS (grade 4). To ensure fair biological comparison between the patient cohorts, only patients with minimal CRS (grades 1-2) were included for this study. The Olink Explore 1536/384 panel (n = 1,463 analytes) was used for the B-ALL patients and the Olink Explore HT platform (n = 5,440 analytes) was used for the AML patients. The Olink platform measures protein expression levels as a normalized protein expression (NPX) value on a log2 scale. To facilitate comparison, we focused on the 1,426 analytes that were measured in both sets of patients. To normalize for batch effect between studies, we compared each NPX value to the patient-matched pre-infusion baseline sample's NPX value to produce an adjusted NPX value. The analysis functions applied determined the NPX difference, i.e., the log2 fold change, between the CRS peak value and the pre-infusion baseline value to assess protein expression levels in CRS in response to CAR-T therapy. Statistically significant assays were identified using a nominal p-value calculated from a Welch two-sample t-test and a p < 0.05 threshold. Results & Discussion AML CRS was characterized by elevation in several proteins not seen to be upregulated in the B-ALL CRS cohort, including PTEN, YTHDF3, IL-2, CXCL16, TNFRSF1B, and IL1RN. Downregulated proteins unique to AML CRS included DSG3, PRTG, CES2, BTC, and GLT8D2. B-ALL CRS was also uniquely characterized by elevation in several proteins, including CD300E, CRTAM, CLEC4C, SIGLEC6, GZMH, and SIGLEC9. Downregulated proteins in B-ALL CRS only included CCL2, PGF, ANXA3, FAS, and FKBP1B. AML and B-ALL patients receiving CAR-T therapy both had elevated CLEC6A, GZMA, GZMB, IFNG, NOS3, and OSCAR. Both cohorts shared downregulation of proteins BOC, CCL16, and CDON. Some differences in the directionality of expression were observed between the two cohorts, for example, GDF15 and MZT1 were upregulated in AML CRS but downregulated in B-ALL CRS. Unique protein signatures identified in this study suggest potential lineage specific differences in immune response to CAR-T therapy. Together, these results warrant further exploration to identify clinically targetable regulators of CRS to decrease toxicity and increase efficacy in AML patients, where efficacy has not been observed to the extent of B-cell malignancies.