Gene expression signature associated with in vitro dexamethasone resistance and post-induction minimal residual disease in pediatric T-cell acute lymphoblastic leukemia.

Abstract

10033 Background: T-cell acute lymphoblastic leukemia (T-ALL) is a genetically heterogeneous disease, which has largely precluded the use of genetic mutations for risk stratification. We hypothesized that despite this heterogeneity, diverse T-ALLs may have functional similarities that underlie patterns of chemotherapy sensitivity. Methods: We used flow cytometry to evaluate in vitro dexamethasone (DEX) sensitivity and baseline expression of signal transduction effectors and BCL2-family proteins in 68 fresh diagnostic T-ALL samples from patients enrolled on the Children’s Oncology Group (COG) trial AALL1231. We also performed RNA-sequencing (RNA-seq) on 40 AALL1231 samples and used hierarchical clustering and linear regression to analyze these and published T-ALL RNA-seq data from COG AALL0434. Comparisons between groups were made using t-tests and Fisher’s exact tests. Results: Of the proteins analyzed, only high BCL2 expression was significantly associated with increased in vitro DEX resistance (p = 0.002). Hierarchical clustering of the AALL1231 RNA-seq data identified two distinct clusters. Cluster 1 was associated with significantly higher BCL2 transcript expression (p = 0.0002) and in vitro DEX resistance (p = 0.04) relative to cluster 2. We defined a gene set consisting of the top 210 differentially expressed genes between these clusters and applied this gene set to the COG AALL0434 cohort. In this analysis, the early T-cell precursor (ETP) and near-ETP samples clustered together (p < 0.0001) in cluster 1 along with 39 of 146 non-ETP samples. Not only did these cluster 1 non-ETP samples have significantly higher BCL2 transcript expression relative to the non-ETP samples in cluster 2 (p < 0.0001), but 54% of these non-ETP samples were minimal residual disease (MRD) positive (≥0.01%) at the end of induction, as opposed to only 16% of the non-ETP samples in cluster 2 (p < 0.0001). Conclusions: Gene expression profiling identifies non-ETP T-ALLs that cluster with ETP/near-ETP T-ALLs and have significantly higher BCL2 expression and increased rates of post-induction MRD.