|Home | About | Journals | Submit | Contact Us | Français|
Human papillomavirus (HPV) type 16 infection is an etiologic factor in a subset of head and neck squamous cell carcinomas (HNSCC). It is unknown if host genetic susceptibility modifies the HPV16–HNSCC association. DNA samples collected as part of a Boston area case–control study of HNSCC were genotyped for single-nucleotide polymorphisms (SNPs) from the National Cancer Institute's SNP500Cancer database. Analysis of demographic, phenotypic and genotypic data for 319 HNSCC cases and 495 frequency-matched controls was performed using unconditional logistic regression. All reported P-values are two sided. We identified a polymorphism in the sodium-dependent vitamin C transporter SLC23A2 that modifies the risk of HNSCC associated with HPV16 infection. Among those with a wild-type allele at SLC23A2, the risk of HNSCC associated with HPV16-positive serology was 5.0 (95% confidence interval (CI)=3.2–7.8). However, among those with a homozygous variant genotype, the risk of HNSCC associated with HPV16 was attenuated [odds ratio (OR)=2.8; 95% CI = 1.2–6.2]. Further, when we tested whether genotype modified the interaction between citrus exposure, HPV16, and HNSCC, we found a dramatically increased risk of HNSCC for those with a wild-type SLC23A2 allele, HPV16-positive serology and high citrus intake (OR=7.4; 95% CI=3.6–15.1). These results suggest that SLC23A2 genetic variation alters HPV16-associated HNSCC while also highlighting the important role of citrus exposure in this disease.
Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer in men and 13th most common cancer in women in the USA, with ~47500 incident cases in USA in 2007 (1). These cancers arise from the epithelium of the oral cavity, oropharynx and larynx and are predicted to cause >11000 deaths in the USA this year (1). Established risk factors for HNSCC include cigarette smoking, use of smokeless tobacco, alcohol consumption and human papillomavirus (HPV) (primarily HPV16) infection (2–5). HPV infection now accounts for ~25% of head and neck cancer in many centers in North America and Europe and is found primarily in tumors of the oropharynx (3,6). Overall, 5 year survival of HNSCC approaches 60%, but is markedly improved in those whose disease is HPV16 positive (3). HPV16-associated oropharyngeal HNSCC is also known to have distinct risk factors and, taken together, this suggests that the differing etiologies of this disease give rise to clinically, and potentially molecularly, distinct diseases (3,4,7,8). We recently reported a significant interaction between HPV16 serology and citrus fruit intake in HNSCC (9). Among serologic HPV-negative persons, citrus fruit consumption reduces risk of HNSCC, whereas among HPV-positive individuals citrus fruit consumption is associated with an increased risk for HNSCC.
Given these risk profiles, we hypothesized that HPV infection, marked by serology status, modifies the risk of HNSCC associated with normal variation in cancer-associated genes or dietary factors, particularly citrus fruit intake. By identifying subgroups of genetically susceptible individuals, as well as delineating the significant differences seen in HPV-positive compared with HPV-negative disease, we may be better able to characterize disease risk factors. We initially utilized a standardized cancer genotyping platform developed by Illumina, based on the USA National Cancer Institute's Cancer Genome Anatomy Project (the SNP500Cancer database) to identify single-nucleotide polymorphisms (SNPs) differentially associated with HPV-positive and negative HNSCC. Subsequently, additional directed genotyping of the entire case–control study population for the candidate locus most associated with HNSCC revealed a significant interaction between a polymorphism in the vitamin C transporter SLC23A2 and HPV16 antibody serology in HNSCC. Variation in SLC23A2 has been recently reported to be associated with non-Hodgkin lymphoma, colorectal adenocarcinoma and gastric cancer (10–12), but to our knowledge, this is the first report of variation in SLC23A2 associated with HNSCC.
Details of this case–control study of HNSCC are described elsewhere (3,7). Briefly, cases were collected from Boston area hospitals from 1999 to 2003. Controls were randomly selected from Massachusetts town books to match cases on town of residence, gender and age (Table I). All case and control subjects were provided written informed consent and the study was approved by institutional review boards of all participating institutions. Questionnaires were administered to collect data on smoking, drinking, diet, family history, medication history and demographics. Diet information was collected using a semi-quantitative food frequency questionnaire assessing food consumption 5 years prior to diagnosis or date of enrollment. The food frequency questionnaire has been previously validated (13,14). Fruit intake was calculated as the amount consumed per day (or week or month if less frequent than once per day) based on subjects’ response for average year-round consumption. Citrus fruit consisted of grapefruit, grapefruit juice, orange and orange juice. A serving was considered to be equivalent to one glass of juice, an orange, or half of a grapefruit. Low citrus intake was defined as <0.67 servings per day and high citrus intake was defined as ≥0.67 servings per day, based on the median value in the control group. Subjects who completed less than half of the food frequency questionnaire or with missing citrus fruit intake data were eliminated from the analysis as were subjects who reported consuming <800 calories (500 calories if female) or >4200 calories (3500 calories if female). HPV16 serology was determined using the HPV competitive Luminex immunoassay as described previously (15).
Genomic DNA was extracted from whole blood using standard protocols as described previously (16). To identify SNPs for further gene–environment interaction testing, a subset of 138 cases (69 HPV16 serology positive and 69 HPV16 serology negative) and 138 controls were genotyped on the GoldenGate SNP platform for 1536 loci on the SNP500Cancer array (17). Analysis of the initial genotyping screen was performed in the R software environment for statistical computing (18) and employed an interaction model including variables for SNP genotype, smoking and drinking. The SLC23A2 polymorphism rs4987219 was most significantly associated with case status. Subsequently, we used Taqman allelic discrimination to genotype rs4987219 in the entire study population (96 HPV16 serology-positive cases, 223 HPV16 serology-negative cases and 495 controls) to confirm the initial results and to provide a larger sample size for examination of effect modification. For the 272 subjects that were genotyped on both platforms, we had a 99.3% concordance rate.
Unconditional logistic regression in SAS 9.1 was utilized to examine the association between rs4987219 and HNSCC, while controlling for age, gender, race, education, smoking, drinking and HPV16 serology as previously reported (7). In addition, we tested for statistical interaction with HPV16 serology using a logistic model controlled for age, gender, race, education, smoking and drinking. The significance of the interaction was tested using the Wald test. All reported P-values are two sided. Because HPV16 positive serology confers a 10-fold risk of HNSCC of the pharynx and only a minimal, if any, increased risk of HNSCC of the larynx and oral cavity (7), we tested the association between HPV16 serology and genotype, stratified by site of tumor. Further, as our previous report identified an interaction between HPV16 seropositivity and citrus consumption (9), we re-examined the statistical interaction between HPV16 serology and dietary citrus intake for each of two rs4987219 genotype groups: any wild-type allele (includes GG and GC genotypes) and homozygous CC genotype.
Using the SNP500Cancer genotyping panel, we identified SNP rs4987219, a non-coding SNP found in intron 8 of SLC23A2, one of two required sodium-dependent vitamin C transporters (reviewed in ref. 19), as a potential modifier of HPV-associated HNSCC. Upon genotyping the entire case–control population for rs4987219, we found that, although there was no association between this SNP and case status (Table II), among HPV16 seropositives, homozygous variant genotype decreased the risk of HNSCC (Table III). This effect was not seen in subjects who were serologically HPV16 negative, indicating there is a potential interaction between rs4987219 and HPV16 serology in HNSCC (Wald test, P=0.10, Table III). We have previously reported that positive antibody titer for HPV16 results in a >4-fold risk of HNSCC (7) and found this to be true in this subset of the study population [odds ratio (OR) 4.2, 95% confidence interval (CI) 2.8, 6.3]. When considering the SNP in conjunction with HPV16 serology, we observed the risk of HNSCC associated with HPV16 positivity was highest among those with a wild-type allele (OR 5.0, 95% CI 3.2, 7.8), and HNSCC risk was significantly attenuated among those who were homozygous variant (OR 2.8, 95% CI 1.2, 6.2). The OR comparing HPV positive with HPV negative within the homozygous variant group is 2.1 (95% CI 0.8, 5.6), which is markedly <5.0, the corresponding OR for the wild-type/heterozygous.
When we tested the association between HPV16 serology and genotype by site of tumor, the interaction between genotype and HPV was strongest for HNSCC of the pharynx (data not shown); however, there was not sufficient power to test for a three-way interaction.
Building on our prior finding of an interaction between HPV16 serology and citrus intake (9), we tested whether the effect of this interaction on risk of HNSCC varied by SLC23A2 genotype (Figure 1). Dietary citrus intake modified the association of a positive serologic response to HPV16 only in those who were wild-type or heterozygous for the SLC23A2 genotype. HNSCC risk was >7-fold higher in the stratum containing those with a SLC23A2 wild-type GG or heterozygous GC genotype, positive HPV16 serology and high citrus consumption. HNSCC risk was markedly reduced among those with low citrus fruit exposure, homozygous SLC23A2 variant CC genotype or negative HPV16 serology. This interaction was found to be statistically significant (P=0.05).
We report that a SNP in intron 8 of SLC23A2 (rs4987219) alters susceptibility to HNSCC associated with HPV16. This result strongly suggests that vitamin C metabolism, manifested by an altered transporter, modifies the likelihood or consequences of HPV16 infection and subsequent associated HNSCC. Alternatively, it is possible that the variant gene transporters have a differential exposure to HPV16, although it is unlikely that genotype of a SNP in a vitamin C transporter would influence or be associated with HPV16 exposure.
The concentration of vitamin C (ascorbate) in the intracellular environment is accomplished through passive and facilitated diffusion as well as via active transport of ascorbate through the sodium-dependent vitamin C transporters (reviewed in ref. 19). The proteins that facilitate this active transport, termed SVCT1 and SVCT2, are encoded by the genes SLC23A1 and SLC23A2, respectively. The SVCT1 protein is responsible for active transport of L-ascorbate from the luminal surface of the gastrointestinal tract and kidney (20–22). In contrast, SVCT2, which is expressed in most human tissues examined with the exception of lung and skeletal muscle, is thought to regulate intracellular levels of ascorbic acid for protection of the cell from oxidative stress (23–25) and promote the maturation of type I collagen (26). Intracellular levels of vitamin C are tightly controlled through regulation of the transporters that, in turn, are regulated by substrate concentrations of ascorbate, by local and systemic inflammatory cytokines, by the presence of a dominant negative isoform of SVCT2, by hormones and by protein kinase C signaling (27–32). The role of the SLC23A2 SNP, rs4987219, in this complex environment has not been established.
HPV16 infection initially occurs as the result of epithelial damage and subsequent exposure and viral invasion of the basal epidermis. This results in the expression of viral transcription regulating genes, alterations in cellular transcription factor regulation and viral oncogenes that alter cell cycle progression (reviewed in ref. 33). In our study, the association with HPV16 was established by a serological marker of HPV16: antibodies specifically directed to the L1 viral capsid protein. While we have established the correlation between serological presence of HPV16 antibodies and presence of HPV16 DNA in HNSCC tumors (3), it is not known whether the SLC23A2 loci can be directly or indirectly regulated via viral or altered cellular transcription factors or, if so, if the polymorphic variants could affect any such regulation. One could hypothesize that the sequence differences resulting from SLC23A2 polymorphisms could influence transcription factor binding and regulation of vitamin C transporter expression. In fact, rs4987219 is predicted to provide a CCAAT enhancer binding protein beta site with the C allele that is not present with the G allele (http://www.cbrc.jp/research/db/TFSEARCH.html).
Although no phenotype for any of the polymorphisms in SLC23A2 has been established, polymorphisms in intron 2 of SLC23A2 have previously been associated with an increased risk of preterm delivery (34). These authors assert that the variant may be associated with diminished tensile strength of membranes and hence contribute to premature rupture of membranes and preterm infant delivery, although the data were not adjusted for dietary intake of ascorbate. These data suggest a possible novel mechanism of action of this variant in HNSCC; namely, the polymorphism may be associated with a difference in the potential strength of the epithelial surface barrier, giving rise to a different potential for viral entry.
In another recent study, intronic SNPs in SLC23A2 were associated with an increased risk of non-Hodgkin lymphoma (10). Again, no specific phenotype was established for the SNPs, but the authors suggest one of the intronic SNPs studied may alter transcriptional regulation of the gene, with the variant genotype causing decreased binding of transcriptional enhancers. The authors also suggest that vitamin C, as a free radical scavenger, may be essential for preventing DNA damage from oxidative stress. Dietary intake of vitamin C was not reported in this study nor was viral presence, although a viral etiology for this disease has been suggested, and therefore the results cannot be adjusted for this factor.
A case–control study of gastric cancer showed a significant association between a SNP in intron 2 of SLC23A2 and a borderline significant synonymous SNP in exon 12 (12). Additionally, a three SNP haplotype-containing SNPs located in intron 3 was significantly associated with gastric cancer. There was no evidence of effect modification by dietary intake of ascorbic acid while consumption of fruits and juices was independently statistically significantly protective of gastric cancer. These authors list several possible mechanisms through which ascorbic acid may play a role in gastric cancer prevention, but there is no known mechanism for the polymorphisms studied.
A nested case–control study of colorectal adenoma looked at the association of 11 SNPs in SLC23A2, including rs4987219, in 656 colorectal adenoma cases and 665 controls (11). None of the SNPs were associated with cancer but a haplotype containing two SNPs, rs4987219 and a synonymous SNP in exon 11, was associated with a decreased risk of colorectal adenoma. There was no association seen between vitamin C intake and disease. The authors suggest no mechanism but recommend further study of the region around the two SNPs in the associated haplotype.
The studies mentioned above, as well as our own investigation, have found that there are numerous low frequency haplotypes of the large and complex SLC23A2 gene (10,34). Fine mapping of this locus in a large number of individuals will be needed to further elucidate any allele that is associated with differential ascorbate absorption.
While the biologic mechanism underlying HNSCC disease in light of the modification of HPV16-associated cancer risk and citrus intake is not clear, our data suggest that variation in systemic ascorbate levels is crucially important for either establishing or maintaining head and neck HPV16 infection or for its progression to a malignant phenotype. The larger cancer risk associated with increased intake of citrus in the wild-type allele carriers might easily be explained by mucosal abrasion and enhancement of the likelihood of local oropharyngeal infection with HPV16, perhaps attributable to the acidic nature of citrus fruit. Further, the fact that susceptibility to HPV16-associated HNSCC risk is modified by genetic variation in a vitamin C transporter argues that the strength of the epidermal barrier is modified by this polymorphism, akin to the mechanism posited by Erichsen et al. (34) to explain the reported association of this variant with premature rupture of membranes in preterm delivery. The role of ascorbate transport in the maturation of type I procollagen and increased intracellular and extracellular levels of type I collagen (26) further supports this connection. Of course, it is possible that, in addition to enhancing viral entry, a high vitamin C diet provides antioxidant and immunostimulatory effects that are modulated by variation in the transporter. If the polymorphism is associated with less efficient regulation of intracellular transport of ascorbate, the result of increased dietary vitamin C could be paradoxical; that is, increased vitamin C intake might induce a decreased intracellular transport of vitamin C in HPV infected cells of wild-type individuals but not in the homozygous variant carriers. In this case, mucosal abrasion and enhanced viral entry act equally but the high citrus intake would be mitigated in the homozygous wild-type individuals where increased dietary citrus would be less effective at lowering intracellular oxidant levels thereby providing less protection against DNA damaging agents, increasing risk of progression to malignancy. As the phenotype of this polymorphism is not known, it is not possible to precisely understand the mechanism responsible for the susceptibility we have observed. In vitro studies examining the functional consequence of this variation are needed.
Limitations of this study include the sample size. Larger sample size would make the associations more stable. At the same time, it seems unlikely that there is any bias associated with participation that would influence our results, as differential participation by genotype or HPV16 exposure seems unlikely. This appears even more unlikely given that subjects were not informed of the study hypotheses.
Regardless of the mechanism, our observation of an interaction of the variant form of the ascorbate transporter SVCT2 with HPV16-associated HNSCC that also modifies the previously observed interaction of citrus intake with HPV16 suggests that the HNSCC risk associated with citrus intake in HPV16-positive individuals is altered by variation in intracellular vitamin C concentrations, perhaps augmenting any increased risk potentially associated with the breakdown of oral mucosal barriers. This data also points to the necessity of examining host susceptibility in light of the major etiologic contributor to this disease (i.e. HPV infection or tobacco and alcohol use), as there are biologically and clinically distinct pathways to carcinogenesis of the head and neck that can be differentially effected by host susceptibility. Such distinctions may point to novel modes of treatment and prevention of this disease.
National Institutes of Health (CA100679, CA78609, T32 ES07155 and T32 CA 09001); Friends of Dana-Farber Cancer Institute; Dan and Gloria Schusterman Foundation.
Conflict of Interest Statement: J.T.B and J.F.S hold stock options employed by Merck and Co., the company that developed and performed the CLIA testing and the makers of Gardasil vaccine. J.T.B is currently conducting research as an employee of the company and J.F.S is currently conducting vaccine basic research at Merck.