Previous analyses of this population have found arsenic was a potential risk factor for NMSC in the United States (Karagas et al. 2001
), and the current analysis suggests that the relationship between arsenic and NMSC may be modified by NER polymorphisms. Having at least one variant at both XPD
polymorphisms was observed less frequently in both BCC and SCC cases than in controls. Among these carriers of variants at multiloci, subjects with toenail arsenic > 0.286 μg/g had twice the risk of SCC compared with those with lowest arsenic exposure. The results from the XPA
analysis suggested that an elevated risk of BCC from exposure to high arsenic may occur in subjects homozygous wild-type for the A23G polymorphism.
Previous studies have examined the XPD
codon 751 polymorphism in relation to benign keratoses. In a study of residents of Bangladesh, Ahsan et al. (2003)
examined the relationship between urinary arsenic, XPD
Lys751Gln, and risk of hyperkeratosis, a potential precursor lesion for NMSC. The investigators found a statistically significant trend for increased arsenic-associated hyper-keratosis among those with the Lys/Lys genotype and a weaker dose response for subjects with one or two copies of the XPD
codon 751 variant allele. In West Bengal India, investigators reported suboptimal DNA repair (as measured by frequency of chromosomal aberrations and aberrant cells) among arsenic-exposed subjects with the Lys/Lys genotype compared with those with one or more variant alleles (Banerjee et al. 2007
). Those with the Lys/Lys genotype also were at greater risk of hyperkeratosis. However, there are several key differences between the studies in Asia and ours from the United States. First, the minor allele frequency was higher in Bangladesh (50%) and India (40%) than in U.S. Caucasians (30%) (Ahsan et al. 2003
; Banerjee et al. 2007
; Shen et al. 1998
; Spitz et al. 2001
). Pigmentary differences influence the amount of UV radiation exposure that reaches keratinocytes, which may alter the nature of the arsenic–gene interaction. In addition, the differences in arsenic dose may underlie the population differences in gene–environment interaction, as studies have shown that the reaction of cells to arsenic exposure differs by dose (Andrew et al. 2006
; Barchowsky et al. 1999
; Del Razo et al. 2001
). These differences in dose are dramatic; arsenic in drinking water in Bangladesh ranges from 10 to 2,040 μg/L (Tondel et al. 1999
), whereas in New Hampshire the range is considerably lower (0.01–180 μg/L among controls) (Karagas et al. 1998
). Finally, although hyperkeratosis may be a precursor lesion for some types of keratinocyte malignancies, in particular SCC, it is not a malignant end point. Our study included incident, histologically confirmed invasive squamous cell carcinomas.
There are a number of ways that arsenic may interfere with the NER pathway and removal of DNA lesions, such as pyrimidine dimers and 6,4-photoproducts from UV radiation (Danaee et al. 2004
; Hartwig 1998
; Lee-Chen et al. 1992
; Rossman 2003
). First, NER proteins have zinc fingers, where zinc is surrounded by four cysteine and/or histidine residues containing sulfhydryl groups (Mackay and Crossley 1998
). Arsenic has a high affinity for sulfhydryl groups, which would allow arsenic to bind to the repair proteins. Consequently, this would inhibit the ability of NER proteins to repair DNA damage, and as a result, increase the risk of cancer. The 312 and 751 polymorphisms in XPD
alter the acidity of the amino acids on the protein. If these non-synonymous polymorphisms change protein structure, this could influence arsenic–protein binding, which may explain the mechanism of XPD
polymorphism effect modification.
Another hypothesis involves the expression of NER genes. Arsenic has been demonstrated to reduce the expression of NER genes, including the incision proteins ERCC1 and XPF (Andrew et al. 2003
). This reduced expression could decrease DRC, a phenotype which has been linked to cancer susceptibility, including skin cancer (Brockmoller et al. 2000
; Wei et al. 1994
). The polymorphisms we studied in XPA
have already been linked to influencing DRC (Hemminki et al. 2001
; Qiao et al. 2002a
; Spitz et al. 2001
; Wu et al. 2003
). If arsenic influenced gene expression levels and therefore DRC, the polymorphisms may further amplify the association between arsenic and NMSC.
We observed that, among those not exposed to high arsenic, subjects with genotypes containing the variant allele were at reduced risk for NMSC. This is consistent with other studies of NMSC and polymorphisms in DNA repair genes (Han et al. 2005
; Marin et al. 2004
; Nelson et al. 2005
; Popanda et al. 2004
; Sanyal et al. 2004
; Shen et al. 2001
). Keratinocytes are thought to have a greater apoptotic response to DNA damage than other cell types (Bowen et al. 2003
). As a result, keratinocytes containing the variant allele, which may have suboptimal DNA repair compared with the wild-type (described earlier), are more likely to undergo apoptosis due to insufficient repair of DNA damage, which can reduce the risk of NMSC (Nelson et al. 2002
). However, when subjects are exposed to higher concentrations of arsenic, there may be interference with cell machinery that influences DNA damage burden and the apoptotic threshold. As a result, cells with the variant allele may have more DNA damage that is not repaired and be unable to undergo apoptosis, resulting in an increased risk of NMSC.
We chose to examine the relationship between arsenic, NER, and NMSC with these particular polymorphisms because previous studies found that they influenced susceptibility to various cancers. However, there are many other polymorphisms in XPA
, which raises the question of whether other polymorphisms could influence these associations and should be analyzed simultaneously. For XPA
, we previously conducted a haplotype analysis, accounting for coding throughout the gene (Miller et al. 2006
). These results suggested that the association with NMSC susceptibility was captured by the A23G polymorphism and that haplotypes accounting for variation across XPA
did not contribute more information. Therefore, we focused on the A23G polymorphism for this gene. For XPD
, we chose to focus on two nonsynonous polymorphisms. We observed that the 312 and 751 polymorphisms in XPD
were in linkage disequilibrium, which has been reported by other investigators (Butkiewicz et al. 2001
; Caggana et al. 2001
; Han et al. 2005
; Hou et al. 2002
; Qiao et al. 2002b
; Spitz et al. 2001
; Vogel et al. 2001
). In addition, the combined genotype data for the two polymorphisms suggested that having coding changes at both loci together influenced risk of NMSC more than just one coding change. Therefore, our results suggest that these polymorphisms should be considered together and may identify individuals who are more susceptible to the carcinogenicity of arsenic. At this point we do not know how additional coding variation in XPD
would influence this finding.
By using toenail measurements of arsenic, we have a measure of arsenic intake through all routes of exposure. A limitation of this measure is that it reflects exposure at one point in time. As previously reported, this New Hampshire population was relatively stable, with over half of subjects using the same water system for at least 15 years (Karagas et al. 2001
). In our study population, arsenic in toenails measured 3–5 years apart were correlated (Karagas et al. 2001
), and in the Nurses’ Health Study measurements 6 years apart were correlated (Garland et al. 1993
). In addition, because our arsenic measurements were blinded to case status, exposure misclassification would result in attenuated estimates.
Our findings provide additional support for the co-carcinogenic action of arsenic via the NER pathway. Additional work is needed to further define the biologic mechanism underlying the interaction, as well as confirm these results in a second population. We chose to focus on a particular mechanism of arsenic co-carcinogenicity with UV radiation in order to identify individuals who may be most susceptible to the effects of arsenic. More research is needed to determine how chronic exposure to low concentrations of arsenic in groundwater as experienced in the United States contributes to the risk of NMSC.