Expression Profiling for MEIS1 and HOXA9/10 Identifies an Increased Incidence of MLL Rearrangements in T-ALL: A Children's Oncology Group Study.

Abstract 2505

The 11q23 MLL gene is commonly rearranged by a diverse range of chromosomal translocations in B-progenitor acute lymphoblastic leukemia (ALL), acute myeloid leukemia and, in particular, in infants less than one year of age. These genomic lesions, commonly identified as MLL (MLL-R) rearrangements via fluorescence in situ hybridization (FISH), are highly variable but consistently encode for MLL fusion proteins that function as transcriptional deregulators of HOX genes, leading to sustained high expression of MEIS1 and many HOXA gene family members. The incidence of MLL-R in T-ALL has not been studied extensively, but has been reported to occur in approximately 1–3% of cases. Using the Affymetrix U133 Plus 2.0 microarray to assess gene expression, we hypothesized that over-expression of MEIS1, HOXA9 and HOXA10 could be use to screen for MLL-R in a cohort of 214 T-ALL cases from recent Children's Oncology Group (COG) studies. Fifty cases and 164 cases, respectively, were obtained from COG studies P9000/9404 and AALL03B1/AALL08B1/AALL0434; all cases underwent nucleic acid extraction, hybridization and profiling as previously described. Following RMA normalization and ROSE analysis for expression outliers, 27 cases (12.6%) showed ≥4-fold increase over the median expression of MEIS1 and 58 cases (27.0%) had ≥4-fold increase over the median expression of HOXA9/10. Interestingly, of the 27 cases with high MEIS1, 25 (92.6%) also had high expression of HOXA9/10 (Fisher's exact test P<0.001). This signature of elevated expression of both MEIS1 and HOXA9/10 is a hallmark of MLL-R in pediatric ALL. For the 50 unselected P9000/9404 cases, 6 cases were identified to be possibly MLL-R based on elevated expression of both MEIS1 and HOXA9/10; however, no specific testing for MLL-R was required on those studies. Twenty-one of the 164 AALL0434 cases profiled demonstrated MEIS1 and HOXA9/10 over-expression; 7 cases had MLL-R confirmed by cytogenetics and/or FISH: MLL-AF6 (x2), del3'MLL, t(11;19)(q23;p13.3) (MLL-ENL; x3) and der(11)t(11;17)(q23;q21). Cases with HOXA9/10 over-expression alone showed one case each for t(11;19)(q23;p13.3), MLL-AF6 and a del3'MLL lesion, and one case without MEIS1 or HOXA9/10 over-expression showed a del3'MLL by FISH. To screen for MLL-R in this study, we also utilized the NG2 mAb, which has been previously shown to identify MLL-R in leukemic samples. Six cases with MLL translocations confirmed by cytogenetics showed detectable NG2 expression by flow cytometry. In our study of 214 T-ALL samples, 30 (14.0%) cases with MEIS1 and/or HOXA9/10 overexpression co-localized with 11 (5.1%) cases confirmed by cytogenetics to have MLL-R. Because of the higher-than-expected number of MLL-R in our series, we reviewed the cytogenetic information provided for 1088 patients currently enrolled onto AALL0434; MLL FISH was performed in 574 (53%) cases, in which 38 (6.62%) MLL-R cases were identified. Our results suggest that MLL-R is more common in T-ALL than previously appreciated. We propose that further confirmatory cytogenetic and molecular testing will identify additional cryptic/unusual translocations, and that MLL-Rs differentially activate MEIS1 and HOXA signaling pathways in T-ALL.