Non-Classical Monocyte Abundance Is an Independent Adverse Risk Factor for Relapse in Pediatric B-ALL

Background Acute Lymphoblastic Leukemia (ALL) is the most common pediatric cancer and while curable in the majority of cases, 15%-20% of children will relapse with only 50% surviving long-term. Treatment failures arise from the outgrowth of pre-existing or acquired sub-clones that are genetically or epigenetically primed to resist treatment. In addition, the bone marrow microenvironment is known to provide a protective niche. We performed the first mapping of the human B-cell ALL (B-ALL) immune bone marrow (BM) microenvironment at single cell resolution at diagnosis, remission and relapse (Witkowski, 2020). We uncovered a striking rewiring of the myeloid compartment during B-ALL progression with significant over-representation of a leukemia-associated monocyte subpopulation expressing high levels of the Macrophage Colony Stimulating Factor Receptor (M-CSFR/CSF1R). Using both peripheral blood (PB) complete blood count analysis and RNA-seq data, we demonstrated that high monocyte abundance at B-ALL diagnosis is predictive of inferior pediatric and adult overall survival in two large independent clinical cohorts. To determine the association of non-classical monocyte abundance in BM and PB with risk of relapse, we examined a cohort of clinical samples from children enrolled on Children's Oncology Group (COG) protocols. Methods Using an unmatched case-control design, we established a preliminary cohort of PB and BM samples collected at diagnosis from 24 B-ALL patients with eventual relapse and 24 patients in long-term remission. Four remission samples from an NYU Langone cohort were used to validate the expansion of this population in the presence of B-ALL. We applied a customized flow cytometry based assay to identify CD115-expressing human monocyte subsets: classical (CD45 +CD56 -CD14 +CD16 -), non-classical (CD45 +CD56 -CD14 -CD16 +), as well as B-cells (CD19, CD22, CD10) and T/NK cells (CD3, CD56). We then performed univariate and multivariable analysis of outcome (relapse versus long-term remission) compared to monocyte subset abundance, adjusting for potential confounding factors (age, gender, CNS status, NCI risk, genetic subtype, WBC at diagnosis, and end of induction minimal residual disease). Results We observed a significantly higher percentage of non-classical monocytes in the diagnostic BM from the COG cohort when compared to remission samples (COG diagnostic B-ALL BM non-classical percentage mean 52.19% vs NYU B-ALL remission BM non-classical per