Comprehensive Characterization of Inflammatory Cytokines in Chronic Lymphocytic Leukemia Highlight Immune Dysregulation and Associates with MRD

Background: Chronic lymphocytic leukemia (CLL) is characterized by immune dysfunction and dysregulation resulting in increased risk for infection and second cancers. Prior studies indicated that chronic inflammation may contribute to worse clinical course in CLL. Comprehensive assessment of the inflammatory mediators and in context of response to treatment, particularly measurable residual disease (MRD), remains unexplored. We aimed to leverage a novel proteomic profiling of >250 inflammatory proteins with high sensitivity assay to assess serum inflammatory profiles at baseline and post-treatment and associations with MRD status.

Methods: The study included 64 patients with CLL treated first-line therapy on the randomized trial of acalabrutinib (ACA) + venetoclax (VEN) +/- early obinutuzumab (OBIN) (NCT04169737) who had baseline plasma samples available for testing. 18 patients had a paired end of treatment (EOC26) sample available from whom all had bone marrow (BM) undetectable MRD status (uMRD4; 10^-4 sensitivity). Healthy donor plasma samples (n=26) were used as control. Inflammatory proteome was evaluated through blood-based proteomic profiling of 251 soluble inflammatory proteins using NUcleic acid Linked Immuno-Sandwich Assay (NULISA), a proximity-ligation assay based on NGS or PCR allowing attomolar (10^-18) detection. Comparative analysis was performed to identify differentially expressed cytokines at baseline or end of therapy compared to healthy to define dysregulated inflammatory protein networks. Logistic regression was applied to assess cytokines prognostic for early uMRD4 at EOC9.

Results: The median age at baseline was 61 years (range 54-68); 52 (83%) had unmutated IGHV, 15 (23%) had del(17p)/TP53 mutation, 16 (25%) had del(11q), and 16 (25%) had complex cytogenetics (>=3 abnormalities). The most frequent mutations were in ATM (30%), NOTCH1 (24%), and TP53 (17%). Stringent internal and inter-plate normalization was performed to control for potential variation. The average detectability was 97.1% across 251 cytokines. Interestingly, supervised hierarchical clustering demonstrated no clear pattern of cytokine expression in key CLL molecular subgroups. However, distinct inflammatory profile was observed between healthy (n=26), and patients at time of therapy initiation and at EOC26 with BM uMRD4 status. Differential analysis with FDR-correction identified 131 significantly upregulated (e.g. TNFSF9, IRAK4, MIF) and 49 significantly downregulated proteins (E.g. IL4, BDNF, CXCL8) at baseline compared to healthy. This suggests a group of inflammatory mediators specifically altered in CLL. By hierarchical clustering, EOC26 samples clustered more closely with healthy than with baseline indicating partial correction of inflammatory proteome with uMRD4 status. However, 50 proteins remained persistently upregulated compared with healthy at EOC26 including key pro-inflammatory cytokines such as IFNG, IL12, and IL18.

To test the prognostic ability of the inflammatory proteomic profiles for early BM uMRD4 status at EOC9, we built logistic regression models on each cytokine for uMRD4, which revealed 14 significant targets (P<0.05). After fitting L1 regularized multivariable logistic model incorporating 14 targets and clinical variables (age, IGHV mutation, complex karyotype, del(17p)/TP53-mutated, therapy regimen), 2 cytokines (MUC16 and HGF) and the receipt of early OBIN remained significant and were retained in the model. The final model achieved high performance in predicting early BM uMRD4 status using leave-one-out cross validation (AUC=0.79; p = 0.003).

Conclusions: Through highly multiplexed proteomic profiling, we identified an integrated profile of coordinately expressed inflammatory proteins in patients with CLL versus normal healthy donors and which may be associated with immune dysfunction and dysregulation in the disease. MUC16 and HGF were associated with achieving early BM uMRD4 status in patients treated with ACA+VEN+/-OBIN. These findings demonstrated the value of high throughput cytokine profiling and may indicate therapeutic targets in CLL.