Mixed Lineage Leukemia Rearrangements (MLL-R) Are Determinants of High Risk Disease in Homeobox A (HOXA)-deregulated T-Lineage Acute Lymphoblastic Leukemia: A Children's Oncology Group Study

Acute leukemias are the most commonly occurring cancers among children, adolescents and young adults. Approximately 15% of newly-diagnosed patients present with T-lineage acute lymphoblastic leukemia (T-ALL). In contrast to B-precursor ALL, molecular lesions that deregulate homeobox (HOX) genes appear to be widely prevalent in T-ALL, but their impact on outcome is unclear. We hypothesized that the molecular lesions associated with HOXA-deregulated T-ALL might identify patients with high-risk disease. We performed gene expression profiling (GEP) on a cohort of 213 T-ALL diagnostic samples obtained from children and young adults enrolled on Children’s Oncology Group (COG) treatment studies 9404 (n=50) and AALL0434 (n=163). We identified a cluster of 54 cases (25%) characterized by increased expression of HOXA3, 5, 7, 9, and 10 (FDR ≤ 0.05; fold change cut off 3). We screened our database for the presence of MLL-R, AF10-R (MLLT10) and other HOXA deregulating lesions using an analysis of cytogenetics, FISH, LDI-PCR and RNA sequencing.  We identified 15 cases with MLL-R, including MLL-AF6 (n = 5), del3’MLL (n = 3), MLL-ENL (n = 5), MLL-AF17 (n = 1), and PICALM-MLL (n = 1). Five cases were confirmed to have PICALM-AF10 fusions and two cases showed DDX3X-AF10 lesions. Two cases harbored NUP98 fusions, two cases had inv(7)(p15q34), and one case each was identified for HOXA10-(3’UTR)TRBC, STAG2-LMO2, LOC338817-CCDC91. We could not identify fusion transcripts in 3 cases, but MLL-R and AF10-R were confidently excluded with RNA sequencing. Almost 90% of cases identified to have MLL-R or AF10-R had ≥ 8-fold over-expression of HOXA9/10. Because HOXA-deregulated T-ALL is characterized by a high degree of molecular heterogeneity, we hypothesized that subset analyses might identify lesions that were more likely to be associated with an inferior outcome. We found that chromosomal abnormalities involving the MLL gene, but not AF10, were associated with induction failure (IF) in T-ALL (P = 0.02, OR = 5.34). Since MLL- and AF10-R leukemias also demonstrate features of undifferentiated leukemias, we discriminated early T-cell precursor (ETP) from non-ETP cases using a GEP developed by Coustan-Smith et al. (Lancet Oncol, 2009). Among the 26 cases that showed ETP features, eleven (42%) also co-expressed HOXA9/10 (≥ 8-fold increase over the median).  We found an association between ETP-ALL and early treatment failure (P = 0.01, OR = 4.37), and next assessed whether ETP cases are enriched with translocations harboring MLL or AF10 genes.  We found overlap between MLL-R and the ETP cases (P = 0.03, OR = 4.14).  We confirmed that ETP-ALL and MLL-R are risk features for IF (P = 0.026, OR = 4.37), and that cases with MLL-R (n = 11) had an inferior EFS compared to those that did not (N = 89) (P = 0.0158).  We extended these observations to assess the impact of ETP/MLL-R (n = 6 vs. non-MLL/non-ETP, n = 69) on EFS, and found a significant association with treatment failure (P = 0.0007). For T-ALL, MRD has emerged as a prognostic indicator of high-risk disease. While MLL-ENL (n = 5) cases did not fail therapy regardless of Day 29 MRD levels, all patients with Day 29 MRD > 0.1 and MLL-AF6 (n = 5) or FISH-identified del3’MLL (n = 3) either failed induction or relapsed. Patients with AF10-R (N = 7) have been reported to have inferior EFS, but we observed that only patients with Day 29 MRD ≥10% failed treatment. Our findings show a heterogeneity of outcomes related to MLL-R, but those with MLL-AF6 and del3’MLL should be considered high-risk. We propose that cytogenetic testing including specific FISH should be performed on all T-ALL patients at diagnosis and relapse. The current development of epigenetic modifying therapies targeted against HOXA-deregulating lesions warrants further study in T-ALL.