Childhood B-cell precursor (BCP) ALL is thought to be caused by a delayed immune response to an unidentified postnatal infection. An association between BCP ALL and HLA class II (DR, DQ, DP) alleles could provide further clues to the identity of the infection, since HLA molecules exhibit allotype-restricted binding of infection-derived antigenic peptides. We clustered 430 HLA-DPB1 alleles into six predicted peptide-binding supertypes (DP1, 2, 3, 4, 6, and 8), based on amino acid di-morphisms at positions 11 (G/L), 69 (E/K), and 84 (G/D) of the DPb 1 domain. We found that the DPb11-69-84 supertype GEG (DP2), was 70% more frequent in BCP ALL (n ¼ 687; Po10 À4 ), and 98% more frequent in cases diagnosed between 3 and 6 years (Po10 À4 ), but not o3 or 46 years, than in controls. Only one of 21 possible DPB1 supergenotypes, GEG/GKG (DP2/DP4) was significantly more frequent in BCP ALL (P ¼ 0.00004) than controls. These results suggest that susceptibility to BCP ALL is associated with the DP2 supertype, which is predicted to bind peptides with positively charged, nonpolar aromatic residues at the P4 position, and hydrophobic residues at the P1 and P6 positions. Studies of peptide binding by DP2 alleles could help to identify infection(s) carrying these peptides.

The extended human major histocompatibility complex (MHC) is a gene-rich region of about 7.6 Mb on chromosome 6, and includes a high proportion of genes involved in the immune response. Among these are the two Human Leukocyte Antigen (HLA) gene clusters, class I and class II, which encode highly polymorphic classical HLA-A, B, C and HLA-DR, DQ and DP genes, respectively. The protein products of the classical HLA genes are heterodimeric cell surface molecules that bind short peptides derived from non-self and self proteins, including infections and auto-antigens. The presentation of these HLA-anchored peptides to T lymphocytes triggers a cascade of responses in immune-associated genes that leads to adaptive immunity. Associations between HLA class II alleles and childhood leukemia have been reported in a number of studies. This could be due to the role of HLA allele-restricted peptide binding and T cell activation, or linkage disequilibrium to an MHC-linked "leukemia gene" in the pathogenesis of childhood leukemia. Efforts are currently in progress to resolve these questions, using large leukemia case-control sample series such as the UK Childhood Cancer Study (UKCCS) and the Northern California Childhood Leukemia Study (NCCLS). Here we review the background to these studies, and present a novel hypothesis based on the paradigm of HLA-associated auto-immune disease that might explain an infection-based etiology of childhood leukemia.

Most childhood B cell precursor (BCP) acute lymphoblastic leukaemia (ALL) cases carry the reciprocal translocation t(12;21)(p13;q22) ( approximately 25%), or a high hyperdiploid (HeH) karyotype (30%). The t(12;21) translocation leads to the expression of a novel fusion gene, TEL-AML1 (ETV6-RUNX1), and HeH often involves tri- and tetrasomy for chromosome 21. The presence of TEL-AML1+ and HeH cells in utero prior to the development of leukaemia suggests that these lesions play a critical role in ALL initiation. Based on our previous analysis of HLA-DP in childhood ALL, and evidence from in vitro studies that TEL-AML1 can activate HLA-DP-restricted T cell responses, we hypothesised that the development of TEL-AML1+ ALL might be influenced by the child's DPB1 genotype. To test this, we analysed the frequency of six HLA-DPB1 supertypes in a population-based series of childhood leukaemias (n = 776) classified by their karyotype (TEL-AML1+, HeH and others), in comparison with newborn controls (n = 864). One DPB1 supertype (GKD) conferred significant protection against TEL-AML1+ ALL (odds ratio (OR), 95% confidence interval (95% CI): 0.42, 0.22-0.81; p < 0.005) and HeH ALL (OR; 95% CI: 0.44, 0.30-0.65; p < 0.0001). These negative associations were almost entirely due to a single allele, DPB1*0101. Our results suggest that DPB1*0101 may afford protection from the development of TEL-AML1+ and HeH BCP ALL, possibly as the result of a DP-restricted immune response to BCP ALL-associated antigen(s), the identification of which could have important implications for the design of prophylactic vaccines.

We previously reported that susceptibility to childhood B cell precursor ALL (BCP ALL) is associated with HLA-DPB1 alleles having glutamic acid (E) rather than lysine (K) in the P4 antigenic peptide-binding pocket. Clustering approximately 90% of DPB1 alleles into DPB69E (DP2, 6, 8) and DPB69K (DP1, 3, 4) supertypes revealed that DP2 and DP8 are associated with BCP ALL, but DP6 is also associated with non-BCP leukaemia. Here, we report that only one of seven alleles with the DP6 supertype (DPB1(*)0601) is associated with childhood leukaemia (leukaemia vs controls: odds ratio, 95% confidence interval [OR, CI]: 4.6, 2.0-10.4; corrected P=0.019), but not with childhood solid tumours or lymphomas. DPB1(*)0601 is also significantly associated with leukaemia subtypes, including BCP ALL, Pro-B ALL, T-ALL and AML. DPB1(*)0601 is significantly over-transmitted (76.9%) from parents to children with BCP ALL (OR; CI: 4.7; 1.01-22.2). Sequencing the coding region of DPB1(*)0601 revealed an exon 1-4 haplotype [T-DEAV-KIL-RVI] shared with DPB1(*)0301 and 0901, but no evidence of germline mutations in childhood leukaemia. These results suggest that the DPbeta0601 molecule may be functionally involved in childhood leukaemia. Analysis of peptide binding and T-cell activation by DPbeta0601-peptide complexes should help determine its role in childhood leukaemia causation.

In a previous study, we obtained preliminary evidence in a small series of patients (n = 63) suggesting that susceptibility to childhood common acute lymphoblastic leukaemia (c-ALL) was associated with an allele at the HLA-DPB1 locus, DPB1*0201. We have now tested this hypothesis by comparing the frequency of children with leukaemia (n = 982) who typed for specific DPB1 alleles and two groups of non-leukaemic children, one consisting of children with solid tumours, excluding lymphomas (n = 409), the other consisting of normal infants (n = 864). We found that significantly more children with c-ALL and T-ALL, but not pro-B ALL or acute non-ALL typed for DPB1*0201 as compared with children with solid tumours [odds ratio (OR), 95% confidence interval (CI) for c-ALL: 1.76, 1.20-2.56; T-ALL: 1.93, 1.01-3.80] and normal infants (OR, 95% CI for c-ALL: 1.83, 1.34-2.48; T-ALL: 2.00, 1.10-3.82). In childhood c-ALL, significantly more children than those with solid tumours or normal infants typed for DPB1 alleles coding specific polymorphic amino acids lining the antigen-binding site of the DPbeta1*0201 allotypic protein, suggesting that susceptibility to childhood c-ALL may be influenced by DPbeta ABS amino acid polymorphisms shared by DPbeta1*0201 and other DPbeta1 allotypes. These results point to a mechanism of c-ALL susceptibility that involves the presentation of specific antigenic peptides, possibly derived from infectious agents, by DPbeta1*0201-related allotypic proteins, leading to the activation of helper T cells mediating proliferative stress on preleukaemic cells.

Gardner and co-workers advanced the hypothesis that the Seascale leukaemia cluster could have been caused by new mutations in germ cells, induced by paternal preconceptional irradiation (PPI) exposure at the Sellafield nuclear installation. Since evidence has shown that PPI can increase the de novo germline mutation rate in hypervariable minisatellite loci, we investigated the hypothesis that sporadic childhood leukaemia might be associated with an increased parental germline minisatellite mutation rate. To test this hypothesis, we compared de novo germline mutation rates in the hypervariable minisatellite locus, CEB1, in family trios (both parents and their child) of children with leukaemia (n ¼ 135) compared with unaffected control families (n ¼ 124). The majority of case and control germline mutations were paternal (94%); the mean paternal germline mutation rates of children with leukaemia (0.083) and control children (0.156) were not significantly different (odds ratio, 95% confidence interval: 0.50, 0.23 -1.08; P ¼ 0.11). There were no significant differences in case and control parental allele sizes, case and control germline mutation progenitor allele sizes (2.74 vs 2.54 kb; P ¼ 0.56), case and control mutant allele sizes (2.71 vs 2.67 kb; P ¼ 0.90), mutant allele size changes (0.13 vs 0.26 kb; P ¼ 0.10), or mutational spectra. Within the limitation of the number of families available for study, we conclude that childhood leukaemia is unlikely to be associated with increased germline minisatellite instability.

SummaryWe recently reported that two of six HLA-DP supertypes 6, 8) were associated with susceptibility (DP2) and resistance (DP1) to childhood acute lymphoblastic leukaemia (ALL). To determine whether DP supertypes are associated with childhood ALL prognosis, we compared treatment outcomes in a cohort (n = 798) of DPB1-typed ALL cases in the UK Medical Research Council UKALL XI trial. No differences in clinical characteristics and outcome between DPB1-typed and untyped (n = 1292) cases suggest no selection bias. Event-free survival (EFS) rates in patients with DP1 and DP3 supertypes were significantly worse than in patients with DP2, DP4, DP6 and DP8 [10-year EFS: 55%; 95% confidence interval (CI) = 49-61%; compared with 64% (61-68%), P = 0AE006]. Ten-year EFS in DP1/DP3 heterozygous patients [30% (2-58%)] was significantly worse than in patients with DP1, DP3 or neither allele [56% (50-62%); P = 0AE02]. Lack of evidence that DP1 or DP3 are associated with known prognostic factors leads us to suggest that these two supertypes exert an independent effect on prognosis. This may involve abrogation of DP1/3-restricted T-cell control of residual disease due to selective effects of chemotherapy. Further studies of HLA supertypes in relation to outcome in recent childhood ALL trials may resolve this question.

No abstract