Age-related differences in leukemia biology and prognosis: the paradigm of MLL-AF4-positive acute lymphoblastic leukemia

Chromosomal rearrangements can fuse the MLL (mixed lineage leukemia) gene to any of more than 50 different partner genes in both acute lymphoblastic leukemia (ALL) and acute myeloid leukemia. The prognosis of leukemias with an MLL rearrangement varies widely according to the partner gene, the leukemia cell lineage, age of the patient and the treatment administered.¹ The most prevalent MLL rearrangement in ALL generates the MLL-AF4 fusion gene owing to the t(4;11)(q21;q23) chromosomal translocation. MLL-AF4-positive ALL has a bimodal age distribution with a major peak incidence in early infancy, and accounts for over 50% of ALL cases in infants less than 6 months of age, 10–20% in older infants, 2% in children and up to 7% in adults.^{1, 2, 3, 4} In an early study of childhood t(4;11)-positive ALL, we observed that treatment outcome differed by age group with infants having the worst outcome,⁵ a finding that has since been confirmed by many other studies including one published in this issue of Leukemia.^{6, 7, 8} Despite recent improvements in the overall treatment outcome for ALL patients, MLL-AF4 fusion is still associated with a dismal prognosis in infants (especially those younger than 6 months) and adults.^{4, 6, 7, 8} Of note, a difference in outcome by age has also been observed in cases with the t(9;22)(q34;q11.2) and BCR-ABL fusion.⁹

The underlying basis of age differences in outcome among leukemias with the same primary genetic abnormality remains unknown, but has been the subject of many investigations that raise several intriguing hypotheses. The very short latency period in infant leukemias suggests that leukemogenic events in these cases occur prenatally, during the early stages of hematopoietic cell development. Two studies reported in this issue of Leukemia demonstrate a high prevalence of immature, nonproductive and/or oligoclonal antigen-receptor gene rearrangements in infant ALL with MLL fusions.^{8, 10} Overall, infant leukemias had gene rearrangement patterns that were different from those of leukemias diagnosed in older children (including those with MLL fusions),^{8, 10} suggesting that infant cases originated mostly from a lymphoid progenitor with germline or incompletely rearranged antigen-receptor genes.^{8, 10} Leukemogenic events during fetal development may also result in distinct MLL breakpoints. Thus, Reichel et al.¹¹ found that 13 of 24 infants less than 1 year of age had breakpoints in the telomeric part of the MLL breakpoint cluster region, whereas 29 of 34 children and adults had breakpoints in the centromeric part of the MLL breakpoint cluster region. They suggested that certain chromosomal areas of the MLL breakpoint cluster region are more susceptible to DNA damage at particular stages of embryonic or hematopoietic development.

Despite these advances in knowledge, further elucidation of the biologic features between MLL-rearranged leukemias diagnosed at different ages is required. The gene expression profile of primary leukemic cells with MLL rearrangements fits with that of cells at a very early stage of differentiation, but whether this pattern differs between cases diagnosed in infants or in older age groups remains unclear.¹² Also unclear is whether the molecular pathways needed for lymphoid cell differentiation are altered in cases with an MLL rearrangement and, if so, whether these alterations differ between the leukemias of infants and older children. These issues can now be addressed by novel methods for large-scale screening of molecular events that might be involved in leukemia pathogenesis (Mullighan et al., Blood 2006; 108: 68A). Although MLL-rearranged leukemias are thought to arise from the oncogenic subversion of undifferentiated hematopoietic cells, it was recently shown that that enforced MLL-AF9 expression in discrete populations of more differentiated hematopoietic progenitors can also initiate leukemia.^{13, 14} Importantly, the leukemic cells in these two studies differed in immunophenotype and in the aberrant expression of ‘stem-cell genes.’ Hence, the characteristics of leukemic cells with the same primary genetic abnormality may vary according to the developmental stage of the target cells. If these findings extend to MLL-AF4-induced transformation, one could hypothesize that such leukemias arise from different target cells in infants compared with older children, leading to different responses to chemotherapy.

Novel therapies are urgently needed for infant leukemias with MLL rearrangements. Such patients might well be among the first candidates for cellular therapies that could enhance the efficacy of hematopoietic stem cell transplantation or even replace it.^{15, 16, 17} Besides FLT3 inhibitors, options for molecularly targeted therapies for MLL leukemias are currently limited.¹⁸ Hence, it is essential to elucidate the molecular pathways subverted by the gene fusion and determine their relative impact on disease aggressiveness and drug resistance, so that promising candidate targets can be identified. Finally, there is increasing evidence that the bone marrow microenvironment can protect leukemic cells from the cytotoxic effects of chemotherapy.¹⁹ If bone marrow mesenchymal cells are found to play a major role in the different treatment responses of MLL-rearranged leukemias by age group, it may be possible to manipulate the interaction between the leukemic cells and their microenvironment to therapeutic advantage.