|Home | About | Journals | Submit | Contact Us | Français|
The mixed lineage leukemia (MLL) gene is commonly rearranged in infant leukemia (IL). Genetic determinants of susceptibility to IL are unknown. Recent genome wide association studies for childhood acute lymphoblastic leukemia (ALL) have identified susceptibility loci at IKZF1, ARID5B, and CEBPE.
We genotyped these loci in 171 infants with leukemia and 384 controls and evaluated associations overall, by subtype (ALL, acute myeloid leukemia (AML)), and by presence (+) or absence (−) of MLL rearrangements.
Homozygosity for a variant IKZF1 allele (rs11978267) increased risk of infant AML (Odds Ratio (OR)=3.9, 95% Confidence Interval (CI)=1.8–8.4); the increased risk was similar for AML/MLL+ and MLL− cases. In contrast, risk of ALL/MLL− was increased in infants homozygous for the IKZF1 variant (OR=5.1, 95%CI=1.8–14.5) but the variant did not modify risk of ALL/MLL+. For ARID5B (rs10821936), homozygosity for the variant allele increased risk for the ALL/MLL− subgroup only (OR=7.2, 95%CI=2.5–20.6). There was little evidence of an association with the CEBP variant (rs2239633).
IKZF1 is expressed in early hematopoiesis, including precursor myeloid cells. Our data provide the first evidence that IKZF1 modifies susceptibility to infant AML, irrespective of MLL rearrangements, and could provide important new etiologic insights into this rare and heterogeneous hematopoietic malignancy.
Leukemias in infancy (<12 months) are extremely rare; ~40 cases per million infants are diagnosed each year in the US. Infant leukemias (IL) are biologically distinct from leukemias in later childhood and adolescence. In older children, acute lymphoblastic leukemia (ALL) is almost five times more common than acute myeloid leukemia (AML); in infants the ratio is closer to 1.3 . Most infant cases (~60–80% ALL, 30–50% AML) present with mixed lineage leukemia (MLL) gene rearrangements in their leukemia cells [2–7], whereas MLL rearrangements occur in only 6% and 14% of childhood ALL and AML, respectively . Evidence from molecular studies of twins suggests that ILs arise in utero, possibly due to transplacental carcinogen exposure . However, given the rarity of IL, it is extremely difficult to identify contributing factors. Like childhood leukemia, it is feasible that IL arises from a combination of events, perhaps on the background of underlying genetic susceptibility.
Very few genetic susceptibility studies have specifically focused on IL. Candidate gene studies suggest a potential role of single nucleotide polymorphisms (SNPs) in methylenetetrahydrofolate (MTHFR), NAD(P)H:quinone oxidoreductase (NQ01), and N-acetyltransferase 2 (NAT2); some report significant associations only with cases harboring MLL gene translocations [10–12]. However, these studies each had fewer than 50 IL cases, emphasizing challenges with interpretation and replication.
Genome wide association (GWA) studies of childhood ALL have identified candidate susceptibility loci at IKZF1, ARID5B, and CEBPE [13,14]. No studies have explored these loci in IL. We genotyped these SNPs in 171 non-Hispanic white IL cases and 384 non-Hispanic white controls to determine associations by leukemia subtype and presence of a MLL gene rearrangement.
IL cases were enrolled in Children’s Oncology Group (COG) Protocol AE24 [15, 16]. Infants (<12 months) with a confirmed diagnosis of ALL or AML (ICD-9 codes: 9801, 9803, 9821, 9823, 9861, 9863, 9910, 9913) during 1996–2006 and treated at U.S. and Canadian COG institutions were eligible. Down syndrome cases were excluded. Case mothers submitted buccal cell samples or released existing biological specimens from their child for genetic analysis. DNA was extracted from cytobrushes via a Purgene kit protocol (Gentra Systems, Minneapolis, MN). DNA was extracted from archived biological samples (e.g., blood smears, bone marrow samples, cerebrospinal fluid) using a phenol-chloroform extraction protocol . All isolated DNA was stored at −80°C prior to genotyping. Mothers also released their child’s diagnostic information, including results of Southern blot, RT-PCR, fluorescent in situ hybridization, or other cytogenetics testing to permit central review. Three independent reviewers (S.M.D., N.A.H., and J.M.H.) evaluated submitted materials to confirm diagnosis and determine if there was evidence of MLL gene rearrangement (MLL+), no rearrangement (MLL−), or insufficient evidence to classify. The current analysis was restricted to infants who provided DNA specimens and who were categorized as non-Hispanic white based on maternal report (n=189) to avoid potential spurious associations due to confounding by population admixture . Anonymized plasma samples from 384 healthy, non-Hispanic white blood donors were used as controls. DNA was isolated from the plasma via a DNA blood midi column protocol (Qiagen, Inc., Valencia, CA) and was stored at −20°C prior to genotyping. Institutional Review Boards at the University of Minnesota and participating COG institutions approved the study. Mothers of IL cases provided informed consent for the use of the biospecimens in genetic analyses.
The selected SNPs (IKZF1 rs11978267, ARID5B rs10821936 and rs10994982, CEBPE rs2239633) were genotyped by fluorogenic PCR-based allelic discrimination (Taqman) assays using the ABI Prism 7900HT RT-PCR System (Applied Biosystems, Inc., Carlsbad, CA). Primers and probes were designed by ABI (Taqman SNP assays C_199413_10, C_26140184_10, C_30824850_10, and C_335486_1). PCR conditions are available on request. Genotype calls were made manually upon visual inspection of allelic discrimination plots by identifying clusters around reference controls for each allele. Among the case samples, there were 18 samples that could not be genotyped across all SNPs. Of the remaining 171 cases (including 102 ALL, 67 AML, and 2 biphenotypic), genotypes for the IKZF1 and ARID5B SNPs could not be called for 3–7% and 19% for the CEBPE SNP. Of these 171 cases, 94 were MLL+, 54 MLL− and 23 indeterminate. For the control samples, the call rate was ≥95% for each SNP. As a quality control measure, 5–8% of samples were selected for duplication for each SNP, and there was 100% concordance between the replicates. Control genotypes for all SNPs were in Hardy Weinberg Equilibrium (p>0.05) and closely followed the genotype distributions reported in similar populations [13, 16].
Case genotypes were compared to those of controls at each locus of interest via unconditional logistic regression (SAS version 9.2, SAS Institute Inc., Cary, NC). Genomic odds ratios (ORs), as well as corresponding 95% confidence intervals (CIs), were calculated to compare individual genotypes (AA vs. BB, AB vs. BB), and allelic ORs and p-values for trend were calculated from ordinal variables indicating the number of high risk alleles (0, 1, 2) to assess potential dose response. An ARID5B risk score was summed for each subject indicating the number of risk alleles across the two ARID5B SNPs and an overall genetic risk score was likewise compiled representing the total number of risk alleles across the four SNPs examined. Individuals with missing genotypes were excluded from the risk scores. Case-control comparisons were performed for all cases combined and for subgroups by MLL gene status (MLL+, MLL−), leukemic subtype (ALL, AML), and combinations of these categories versus the entire control group.
There was a significantly increased risk of IL overall associated with two copies of the IKZF1 variant (OR=2.3, 95%CI=1.3–4.2) (Table 1). This association was seen in MLL− and AML cases but not in ALL cases overall, and was only apparent for cases homozygous (OR=4.0, 95%CI=1.8–8.6; OR=3.9, 95%CI=1.8–8.4, respectively) for the variant allele. Upon further stratification by subtype and MLL status (Table 2), strong associations were observed among AML cases with two copies of the variant allele regardless of MLL involvement. For ALL, there was a strong association for MLL− cases with two copies of the variant allele (OR=5.1, 95%CI=1.8–14.5), but no association in ALL/MLL+ cases (OR=0.7, 95%CI=0.2–2.2).
There were no significant associations observed overall for either of the variant ARID5B alleles (Table 1). However, when stratified by leukemia subtype or MLL status, there was an increased risk associated with two copies of the variant rs10821936 allele among MLL− cases only (OR=2.8, 95%CI=1.3–6.2). Moreover, there was an inverse association among MLL+ cases heterozygous for rs10994982 (OR=0.5, 95%CI=0.3–0.8). Further stratification revealed a significant risk with rs10821936 among ALL/MLL− cases only; homozygosity for the variant allele was associated with a 7·2-fold increased risk (95%CI=2.5–20.6) (Table 2). An inverse effect of rs10994982 was only apparent among AML/MLL+ cases, although this observation is based on a small number of cases. Finally, for CEBPE (rs2239633) there was a significant inverse association among heterozygous AML/MLL− cases (OR=0.3, 95%CI=0.1–0.9), but no reduction in risk in infants with the homozygous variant genotype, raising concern that this may be a spurious observation. There were no associations of CEBPE (rs2239633) with any of the other case groups. Across all SNPs, we found little evidence of an increasing risk associated with increasing number of variant alleles, suggesting each locus imparts an independent role.
Our results provide evidence for a distinct genetic etiology for infant ALL cases with and without MLL gene rearrangements, but similarity for infants with AML with or without MLL rearrangement. Clinical observations support these observations. Five-year survival rates for infant ALL vary markedly by MLL status. Eighty percent of infants with ALL and no MLL rearrangement survive at five years, while survival rates for infant ALL/MLL+ cases are less than 50% [19, 20]. In contrast, MLL status has little influence on survival for infant AML . Further, infant ALL/MLL+ cases are more often diagnosed prior to six months of age, whereas infant ALL/MLL− cases are more frequently diagnosed in older infants; age differences by MLL status are less distinct among AML cases . Gene expression studies demonstrate clustering by selected cell phenotypes (including pre-B cell versus infant ALL, pediatric ALL versus AML) as well as by MLL status (rearranged versus germline) [23–26]. However, we are not aware of a specific gene expression study that directly compares infant ALL and AML cases by MLL status. Taken together, the biological distinctions between these subgroups of leukemia may reflect origins in precursors arising at different stages of hematopoietic development, and modifiers of susceptibility may be expected to vary also.
Ikaros is a zinc finger transcription factor that plays a major role in B cell development ; it is highly expressed in liver, lymph node, thymus, bone marrow, and placenta . Ikaros drives differentiation of hematopoietic stem cells and multi-potent progenitors into the B-cell lineage, and suppresses differentiation into the myeloid lineage. The functional significance of the IKZF1 polymorphism is not fully elucidated. mRNA expression was evaluated in transformed Epstein Barr Virus lymphocytes and significantly lower dose-dependent expression was found with each copy of the variant IKZF1 allele . Our results, as well as those in B-lineage leukemia in older children suggest that reduced expression of Ikaros increases risk of leukemogenesis, and it is possible that low levels of Ikaros delay progression through B-cell maturation and increase the number of precursors at risk of leukemogenesis. Data suggest that MLL rearrangements associated with fusion to AF4 (most common in infant ALL) occur very early in embryological development, perhaps in a population of mesodermal prehematopoietic precursors, still capable of differentiation into multiple cell lineages, including mesenchymal stem cells . It is possible that such an early precursor would be less influenced by Ikaros, as our data show, in contrast to leukemia arising later in the pathway of lymphoid differentiation.
Involvement of Ikaros in AML has not been reported previously, and is perhaps more surprising. While Ikaros appears to play a less significant role in myelopoiesis, the protein is highly expressed in early myeloid precursor cells, and it is clear that suppression of myeloid differentiation is as important a function as the promotion of lymphoid development . Thus, it is conceivable that reduced expression of Ikaros during these early stages contributes to susceptibility to infant AML overall. The IKZF1 genotype has not been studied in older children with AML. In an ad hoc analysis, we genotyped 450 AML cases (ages 0–19 years) enrolled on COG studies 2941 and 2961 and found no overall associations with the variant allele, suggesting that this susceptibility is specific to infants and is not a feature of AML generally.
ARID5B encodes the AT-rich interactive domain 5B, a member of the ARID family of proteins, which are highly conserved and play a key role in development. ARIDB5 homozygous null mice show transient immune abnormalities including a reduction in early B-cell progenitors . Expression of ARID5B also correlated with RAG1 expression in bone marrow , which supports a role for this gene in early B-cell development. Our data point to a role of ARID5B only in ALL/MLL− cases, in agreement with observations in childhood pre-B cell ALL [13, 28]. Fewer studies suggest a role for CEBPE, a regulator of myelopoiesis in childhood ALL ; our data do not support an important role in IL.
In conclusion, this study suggests a role for IKZF1 in early myeloid leukemia. While there were a relatively small number of IL cases in our study, the ORs are quite strong, and higher than those typically reported in genetic susceptibility studies. While the data would be strengthened by replication in independent datasets, this is challenging due to the low frequency of IL, and the need for cases to be carefully phenotyped for lineage (ALL vs AML) and MLL involvement. Additional functional studies are also needed to understand the role of these genes in early hematopoiesis.
Supported by: National Institute of Health grants R01 CA079940, T32 CA099936, K05 CA157439, U10CA13539, U10CA98543, and the Children’s Cancer Research Fund, Minneapolis, MN.