In:
Cancer Research, American Association for Cancer Research (AACR), Vol. 76, No. 14_Supplement ( 2016-07-15), p. 2693-2693
Abstract:
B-cell precursor acute lymphoblastic leukemia (BCP-ALL) is the most common type of childhood cancer and is characterized by the malignant expansion of B-lymphocyte progenitors in the bone marrow (BM). Current therapy improves the relapse-free survival in children to over 80%. However, the ∼20% of patients who relapse have a poor prognosis and there are no reliable tests that predict relapse using diagnostic samples. We reasoned that aligning BCP-ALL cells according to a formalized context of normal B-lymphocyte development would reveal hidden cell states associated with relapse, and potentially expose targets to augment therapy for patients at risk. Until recently, our ability to pinpoint the identities of B-cell progenitors had been hindered by the vast cellular diversity within the BM and by the scarcity of the primary BM samples. We applied a single-cell proteomics platform termed mass cytometry by time-of-flight (CyTOF). In CyTOF, elemental mass reporter tagged antibodies probe proteins defining cellular identity and signaling within those cells. CyTOF simultaneously quantifies & gt; 40 proteins per cell in millions of individual cells. We defined a cell-state signature for 15 developmental populations of B lymphocytes within the normal human BM. Using this signature we assigned each leukemia cell from 52 primary diagnostic samples to its closest match in B lymphopoiesis using a classifier based on Mahalanobis distance. When applied to BM samples from 4 healthy donors our classifier correctly assigned cells to the true developmental population (accuracy = 0.92, F-measure = 0.92). Using this classifier it was determined that each BCP-ALL sample contains a mix of developmental populations - with 97% of samples enriched in populations that span the pre-pro-B to pre-B transition. We identified 20 predictors (using a machine learning approach) in diagnostic samples that perfectly separate patients who will relapse from those who will not (lasso; predictive AUC = 0.83). This is superior to the NCI risk that is currently employed at clinical diagnosis. These predictors are informative and suggest that high basal activation of IL-7 signaling nodes (pSTAT5, pAKT) in pre-pro-B to pro-BII cells and poor response following pre-B-cell receptor engagement in pre-BI cells portend relapse. As such, these pathways might eventually be targeted via drug repurposing to improve outcomes and to guide therapy in the high-risk childhood BCP-ALL patients identified with our predictor signature. Such an approach to cancer cell developmental classification could be generally applicable across various investigations on understanding and preventing relapse. Citation Format: Zinaida Good, Jolanda Sarno, Astraea Jager, Nikolay Samusik, Wendy Fantl, Nima Aghaeepour, Robert Tibshirani, Sean C. Bendall, Giuseppe Gaipa, Andrea Biondi, Garry P. Nolan, Kara L. Davis. Relapse in BCP-ALL predicted by activated signaling in pro-B cell subsets. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2693.
Type of Medium:
Online Resource
ISSN:
0008-5472
,
1538-7445
DOI:
10.1158/1538-7445.AM2016-2693
Language:
English
Publisher:
American Association for Cancer Research (AACR)
Publication Date:
2016
detail.hit.zdb_id:
2036785-5
detail.hit.zdb_id:
1432-1
detail.hit.zdb_id:
410466-3
Permalink