A common genetic variant in XPD associates with risk of 5q- and 7q-deleted acute myeloid leukemia

Acute myeloid leukemia (AML) can be characterized by the somatic acquisition of gross genomic abnormalities, including translocations, inversions, deletions, and chromosomal trisomy and monosomy. The structural diversity of these lesions suggests that the mechanisms underlying their generation are likely to be varied. Given this, we hypothesized that whereas some risk factors for AML will be common to all subtypes and independent of karyotype, other risk factors, including some genetic susceptibility alleles, will be unique to subgroups with defined molecular lesions. We have previously shown that carriers of the xeroderma pigmentosum group D (XPD) codon 751 glutamine-encoding polymorphism were more likely to have a karyotype associated with a less favorable prognosis (n = 308 cases, test for trend P = .008).¹ XPD encodes a protein with DNA helicase activity that mediates the DNA unwinding required for basal transcription and nucleotide excision DNA repair (NER). Given its role in DNA repair we hypothesized that polymorphic XPD predisposes to the formation of 1 or more specific chromosomal lesions, and that this might explain the observed association with a less-than-favorable karyotype. In order to test this, we determined XPD codon 751 polymorphism status in a series of 927 AML cases, enriched for those with a complex karyotype and chromosome 5 and/or 7 deletions.

We examined XPD codon 751 status in 927 individuals diagnosed with AML. These included 343 patients recruited as part of a large population-based case-control study of acute leukemia conducted in the United Kingdom,²  558 patients enrolled in the United Kingdom Medical Research Council AML trials (nos. 10, 11, and 12) between May 1988 and May 2002,^3-6  and 66 patients recruited as part of a therapy-related leukemia study conducted in the United Kingdom^1,7,8  (Table 1). Ethical approval was obtained for these studies. Informed consent was provided according to the Declaration of Helsinki. This study has been registered with the University of York, Department of Biology Ethics Committee.

DNA was extracted from whole frozen blood or white cell pellets using proteinase K digestion and phenol/chloroform extraction, or from archived bone marrow smears using the QIAamp DNA mini extraction kit according to the manufacturer's instructions (Qiagen, Hilden, Germany).

Leukemia cases were genotyped using polymerase chain reaction/restriction fragment-length polymorphism assay, as previously described.¹  Direct sequencing was used to genotype 4% of randomly selected samples; genotyping concordance for different techniques was 100%.

Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using unconditional logistic regression⁹  routinely adjusting for age and sex. XPD genotype was analyzed either as a trichotomous (lys/lys, lys/gln, gln/gln) or dichotomous (lys/lys, lys/gln + gln/gln) variable. All analyses were performed using Stata 9.1 1999 (Stata, College Station, TX). Cases were stratified according to the presence of defined cytogenetic abnormalities, irrespective of the presence of other abnormalities (some groups are therefore not mutually exclusive), and ORs and 95% CIs were calculated using AML cases with a normal karyotype as the reference population.

When grouped according to the presence of specific cytogenetic lesions, those cases without any apparent abnormality (normal karyotype) formed the largest group. The XPD codon 751 polymorphism distribution in this group (n = 363; lys/lys 38.3%, lys/gln 47.1%, gln/gln 14.6%) was not significantly different from United Kingdom cancer-free individuals reported previously (n = 696; lys/lys 42.1%, lys/gln 43.0%, gln/gln 15.0%).¹  The normal karyotype case subgroup was used as the reference population in all subsequent analysis.

We can postulate that structurally similar lesions, irrespective of their location in the genome, might have arisen via the same mechanism. As such, for the purposes of an initial analysis we grouped cases according to the presence of structurally similar, commonly occurring cytogenetic aberrations. XPD codon 751 status was not significantly associated with risk of developing AML defined by the presence of a trisomy (+8, +21, +22; n = 122), a monosomy (−5, −7; n = 100), a translocation (t(15;17), t(8;21); n = 137), or a complex karyotype (defined as 5 or more unrelated abnormalities; n = 112; Table 2), but was significantly associated with risk of developing AML with a deletion (del(5q), del(7q), del(9q); n = 128) when compared with the normal karyotype group (OR for lys/gln + gln/gln vs lys/lys 1.65; 95% CI 1.06-2.58; Table 2; P = .03). When stratified by the site of deletion, carrier status for the XPD codon 751 glutamine-encoding variant (lys/gln + gln/gln) was independently associated with risk of del(5q)-positive AML (OR for lys/gln + gln/gln vs lys/lys 2.09; 95% CI 1.14-3.81; P = .02) and del(7q)-positive AML (OR for lys/gln + gln/gln vs lys/lys 2.27; 95% CI 1.09-4.71; P = .03), but was not associated with risk of del(9q)-positive AML (Table 2). One hundred seven patients had a karyotype that included a 5q deletion and/or a 7q deletion, risk for which was strongly associated with carrier status for the XPD codon 751 glutamine-encoding variant (OR for lys/gln + gln/gln vs lys/lys 1.93; 95% CI 1.18-3.16; P = .009; data not shown). Of these, 9 patients had a karyotype that included a 5q deletion and a 7q deletion, and all of these were carriers of the glutamine variant (5 heterozygotes and 4 homozygotes for the glutamine-encoding allele).

The association between del(5q)-positive, del(7q)-positive AML and XPD codon 751 status remained significant when the United Kingdom cancer-free control series¹  (n = 696) was used as the reference population (del(5q) OR for lys/gln + gln/gln vs lys/lys 2.57; 95% CI 1.42-4.65; P = .002; del(7q) OR for lys/gln + gln/gln vs lys/lys 2.77; 95% CI 1.35-5.69; P = .006). It will of course be important to replicate this study in a second independent series of AML cases with 5q and/or 7q deletions such that we can be confident the association reported is not a false positive.

These data suggest that the XPD locus, located on chromosome 19q13, carries a low penetrance susceptibility allele for AML with somatically acquired long-arm deletions of chromosomes 5 and/or 7. Although rare, a putative susceptibility locus for familial AML with chromosome 5 and/or 7 deletions has been localized to 16q22.^10,11  Efforts to identify the responsible locus have proven difficult, although one hypothesis suggests that polymorphic NAD(P)H:quinone oxido-reductase 1, encoded by a locus at 16q22, may affect risk of 5q- and/or 7q-deleted AML by modifying host response to leukemogens, including benzene¹²  and chemotherapeutic alkylating agents.¹³  AML that develops after occupational exposure to benzene frequently carries somatic aberrations involving loss of all or part of chromosomes 5 and/or 7, and benzene metabolites have been shown to induce these deletions in cultured cells.^14-16  These observations are particularly relevant to our findings because human cells deficient in NER, of which XPD is one component, are sensitive to the mutagenic effects of the benzene metabolite p-benzoquinone in comparison with NER-proficient cells,¹⁷  implicating NER in the repair of benzene-induced DNA damage. These observations suggest one plausible biologic mechanism by which polymorphic XPD might predispose to the development of AML with long-arm chromosome 5 and/or 7 deletions.

A second model predicts a role for polymorphic XPD in modulating p53-mediated cell death signaling. XPD is a component of the p53-mediated apoptosis pathway, and the 2 proteins interact directly via the carboxy terminus of XPD, which includes the polymorphic codon 751 residue.¹⁸  Consistent with a role for XPD in mediating cell death, fibroblasts with inactivating XPD mutations have attenuated p53-mediated apoptosis.^19,20  Given a functional interaction between XPD and p53 it is noteworthy that somatic alterations involving p53 (chromosome 17 monosomy, 17p deletions, and point mutations) are relatively rare in AML per se, but they occur with high frequency in AML with chromosomal losses of 5 and/or 7, and particularly so when found in the context of therapy-related AML after alkylating chemotherapy.^21-23  Indeed, there is now considerable evidence to suggest that chromosome 5 and 7 alterations cooperate with p53 alterations to promote myeloid neoplasia. Given this, we can speculate that in a background of chromosome 5 and/or 7 long-arm deletions polymorphisms in XPD may directly impact the ability to signal cell death via a p53-dependent pathway, or that XPD polymorphic variants may predispose to the acquisition of p53 mutations, as has been suggested for lung and bladder cancer.^24,25  An inability to signal cell death is predicted to result in the persistence of preleukemic clones and increase the risk of malignant transformation. Consistent with this model, the XPD codon 751 glutamine-encoding allele is significantly associated with an increased risk of developing therapy-related AML, as we have previously reported.¹ 

In summary, we have identified a polymorphic variant in the XPD DNA repair gene that associates specifically with risk of developing 5q- and/or 7q-deleted AML, supporting the hypothesis that AML subgroups defined by karyotype have independent etiologies and unique risk factors. The identification of common risk factors for independent cytogenetic aberrations might explain why some lesions are seen together more frequently that would otherwise be expected.

Contribution: A.G.S. performed research and analyzed the data; L.J.W. performed research and managed the case series; J.M.A. conceived the hypothesis, performed research, and wrote the paper; all authors commented on and approved the final manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: James M. Allan, Epidemiology and Genetics Unit, Department of Biology, University of York, Heslington, York YO10 5YW, United Kingdom; e-mail: jim.allan@egu.york.ac.uk.

We are grateful to staff at the Leukaemia Research Fund Epidemiology and Genetics Unit (University of York), and the Medical Research Council Adult Acute Leukaemia Working Party for supporting this work. We acknowledge the support of the Medical Research Council DNA and RNA bank, University College London Hospital, a facility funded in part by the Leukaemia Research Fund, the Medical Research Council, and the Kay Kendall Leukaemia Fund. This work was supported by the Leukaemia Research Fund of the United Kingdom. The Epidemiology and Genetics Unit is a member of the Haematology Research Network for Yorkshire and Humber.