Age Is the Strongest Determinant of Leukemia Blast Cell Gene Expression in MLL-Rearranged Infant ALL and MLL-AF4 Directs a Distinct Gene Expression Profile Related to CNS Disease.

In infant ALL a constellation of features (young age, hyperleukocytosis, extramedullary/CNS disease, CD10⁻ phenotype, MLL translocation, slow early response to treatment) are associated with poor outcome. Despite the frequent MLL translocations, the biologic basis for these features is unknown. In this study, gene expression profiles of leukemia cells from diagnosis were determined in MLL-rearranged infant ALL and relationships were sought between gene expression and clinical covariates.

Patients and Methods: Seventeen infants treated on COG protocol CCG 1953 were studied. Features at diagnosis were: age, 0–333 d, median 121 d; WBC count, 10.2–1,286 ×10³/μL, median 229 ×10³/μL; CD10⁻, 14, CD10⁺, 2, CD10 unknown, 1; CNS1 status, 11, CNS2 status, 2, CNS3 status, 4. Eight were alive and 9 were dead at survival times from 41–3384 d. MLL translocations were identified using karyotype, FISH, Southern blot and PCR. Gene expression profiling was performed with Affymetrix HG_U133 Plus2.0 arrays. Pearson’s correlation coefficients and significance levels were computed between age and log-2 transformed expression levels of each probeset. To find associations between gene expression and MLL partner genes with the age effect controlled, linear regression was run using gene expression as the dependent variable and age and partner gene (AF4 v. other) as the independent variable. Similar linear regressions were run for WBC count, CNS status (excluding CNS2) and survival status.

Results: PCR identified the MLL partner genes AF4, ENL, AF9, AF10 and EPS15 in 6, 4, 2, 1 and 1 cases, respectively. In 1 case each with t(4;11)(q21;q23) or t(11;19)(q23;p13.3) partner genes were assigned based on karyotype. Another case harbored a t(6;9;11) (q21;p22;q23). Gene expression was affected more by age at diagnosis than any other factor. Remarkably, gene expression analysis by age at diagnosis as a continuous variable showed correlations of 2072 probesets, all but one of which were positive (p<0.0001). The most differentially expressed genes were related to glycosylation and signaling, adhesion, membrane and development. Gene expression analysis by partner gene with the age effect controlled yielded 118 probesets (p<0.001) that segregated cases with MLL-AF4 from cases with MLL-ENL. Many more genes were up- than down-regulated in cases with MLL-AF4. Expression of these 118 probesets in other non MLL-AF4 cases more closely resembled those in cases with MLL-ENL. CNS status significantly correlated with 208 probesets (p<0.001). Even though the partner gene was AF4 in all CNS3 status cases and there was substantial overlap, the genes associated with CNS3 status or MLL-AF4 were not entirely overlapping. The most differentially expressed genes that were correlated with CNS status and MLL-AF4 were related to macromolecule/protein localization and protein transport. With age effect controlled, few genes correlated with WBC count or outcome.

Conclusions: Within infant ALL there are profound intrinsic biologic differences in leukemia cells due to age at diagnosis. MLL-AF4 translocations in infant ALL have variably been associated with more inferior outcomes than other MLL translocations. This analysis indicates that the different MLL translocations are biologically dissimilar and that cases with MLL-AF4 are distinct from cases with MLL-ENL or other partner genes. The study also shows that a specific gene expression program directs the clinical/biologic property of CNS disease, which is overlapping with the effect of AF4 involvement. The complex correlations between clinical covariates, MLL translocations and gene expression in infant ALL merit further study. It will also be important to determine if a similar gene expression profile is associated with CNS disease in other leukemia subtypes.