Introduction

Cervical cancer is the third most commonly diagnosed cancer and the fourth leading cause of cancer death in women worldwide [1]. It is also a grave health problem in Taiwan, with 2700 women developing this disease each year and second in incidence only to breast cancer [2]. Human papillomavirus (HPV) has been widely accepted as an important etiological agent of cervical carcinogenesis [3]. However, only a small portion of HPV-infected women progress to cervical cancer during their lifetime, suggesting that other factors may contribute to the progression of the disease. Along with environmental and lifestyle factors, familial aggregation studies indicate that host genetic variation is likely to involve in cervical cancer pathogenesis [4].

Human leukocyte antigen (HLA) genes are highly polymorphic and encode HLA molecules that are essential for the presentation of viral peptides to the immune system, including HPV. The large number of polymorphisms of the HLA alleles leads to variations in the antigen-recognition site, which may confer susceptibility or resistance to HPV infection and the consequentially neoplastic progression or regression. Due to the pivotal role of HLA molecules in the immune system, several studies have been performed to investigate the association between specific HLA alleles and cervical cancer. In a review article, Hildesheim and Wang [5] reported that HLA-DRB1*15:01, DQB1*03 alleles, and DRB1*15:01-DQB1*06:02 haplotype were positively associated with cervical cancer among certain ethnic groups, whereas DRB1*13 alleles and DRB1*13:01-DQB1*06:03 haplotype were negatively associated with it. Other studies have shown that HLA-A*03:01, *11:04, B*44:02, and Cw*05:01 alleles conferred susceptibility to cervical cancer, but a protective effect against the disease was found in HLA-A*24:02, B*15:01, and Cw*02:02 alleles [6–8]. However, only a few studies have explored the association between HLA-DPB1 and cervical cancer [9–11] and no such data on Taiwanese population have been reported.

In this study, we investigated whether HLA-DPB1 alleles are associated with cervical cancer risk in a hospital-based case–control study of 473 cervical squamous cell carcinoma (CSCC) patients and 676 sex- and age-matched healthy controls. A recently discovered SNP, +550 G/A (dbSNP ID: rs 9277535), located within the 3′ UTR of HLA-DPB1 gene, also has been analyzed because of its putative role in the regulation of HLA-DPB1 mRNA expression [12].

Patients and methods

Study population

A hospital-based case–control study was conducted at the Mackay Memorial Hospital. Four hundred and seventy-three Taiwanese women with pathologically proven CSCC (mean ± SD age at diagnosis: 53.7 ± 12.1 years) were included in the present study. Controls were 676 women (mean ± SD age at sampling: 54.5 ± 8.4 years) selected randomly from a cervical cancer screening program. Exclusion criteria included abnormal Pap result, history of cervical neoplasia, skin or genital warts, immunocompromised conditions, other cancers, and previous operations on the uterine cervix. All the control subjects were frequency matched to the cases by age. After obtaining informed consent, surgical resections or cervical scrapes were collected from all genetically unrelated study participants. The study was approved by the ethics committee and adhered to the tenets of the 1964 Declaration of Helsinki.

DNA isolation

DNA was isolated from formalin-fixed and paraffin-embedded tissue specimens of CSCC patients by use of the Qiagen DNeasy Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol. Qiagen DNA extraction kit was used to isolate DNA from cervical scrapes of controls.

HPV typing

HPV detection and genotyping were performed by polymerase chain reaction (PCR)-based amplification of a fragment of approximately 192 bp in the L1 region of the HPV genome with a pair of degenerate primers, GP6+/MY11 [13, 14]. The HPV genotype was then determined by sequencing the PCR product on an automated sequencer (ABI 377, Applied Biosystems, Foster City, CA). Since stratifications based on HPV types in controls were not performed in this study, no HPV DNA testing was done for the control subjects.

HLA-DPB1 genotyping

HLA-DPB1 alleles were determined using high-resolution sequence-based typing protocol with some modifications from a previous study [15]. This involved a high-resolution HLA typing method using PCR amplification of genomic DNA, followed by direct DNA sequencing. Briefly, PCR using locus-specific primer sets was performed to amplify exon 2 of HLA-DPB1 gene. The presence of PCR product was checked by 2 % agarose gel electrophoresis. The amplicons were then purified and sequenced using the ABI BigDye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Foster City, CA) in an ABI 3730 DNA analyzer. Sequencing was performed in both forward and reverse directions and analyzed using Assign-SBT™ 3.5+ software (Conexio Genomics, Fremantle, Australia). Ambiguous alleles were assigned according to common alleles (allele frequency >0.01) found in the Taiwanese and southern Chinese populations [16, 17] or to the allele with lowest definition number. The genotyping of the +550 G/A SNP was performed with the TaqMan assay (Applied Biosystems) according to the manufacturer’s instructions.

Statistical analysis

Statistical significance of differences in frequency between each allele, genotype, and haplotype of CSCC, HPV-16 positive CSCC, and control groups was compared using Chi-square test with Yates’ correction or Fisher’s exact test (when the number of subjects in a cell was <5). Odds ratio (OR) and 95 % confidence interval (CI) were calculated to determine the magnitude of associations. Hardy–Weinberg equilibrium, haplotype frequencies, and linkage disequilibriums (LDs) were evaluated with PyPop 0.7.0 statistical software [18] from the International Histocompatibility Working Group based on the method described by Guo and Thompson [19]. Haplotypes with frequency <1.5 % in the cases or controls were excluded. The Bonferroni correction was used to correct P values when several statistical tests were being performed simultaneously on a single data set. To do a Bonferroni correction, the formula P c = 1 − (1 − P)n was applied, where P c is the corrected P value, P the uncorrected one, and n the number of comparisons [20]. The comparison number was 8 for HLA-DPB1 alleles and haplotypes. As for +550 G/A SNP, n is 2 for each genotype and allele for simultaneously testing genotype and allele frequencies [21]. A corrected P value (P c) of <0.05 (2-tailed) was considered statistically significant. Using the Quanto Version 1.1 software (Department of Preventive Medicine, University of Southern California, CA, USA), we designed the study to have >80 % power at a 5 % significance level to determine a genotype relative risk of 1.5 conferred by HLA-DPB1 variants with an estimated prevalence of 360/100,000 [22].

Results

HPV DNA was detected in 72.7 % of the 473 tumors tested, with 65.4 % positive for HPV type 16, 10.2 % for HPV type 18, and 24.4 % for other HPV types.

The distribution of the HLA-DPB1 alleles and +550 G/A genotypes and alleles in the controls and cases is shown in Tables 1 and 2. The observed genotype frequencies for these DPB1 variants did not deviate significantly from the Hardy–Weinberg equilibrium in the controls (P > 0.05). A significantly increased frequency in DPB1*13:01 allele (OR = 1.46, 95 % CI 1.10–1.96) and +550 A allele (OR = 1.25, 95 % CI 1.05–1.49) and a significantly decreased frequency in DPB1*05:01 allele (OR = 0.72, 95 % CI 0.61–0.86), +550 G/G genotype (OR = 0.75, 95 % CI 0.59–0.96), and G allele (OR = 0.80, 95 % CI 0.67–0.96) were found in CSCC patients as compared with controls (Tables 1, 2). The significance was still retained in DPB1*05:01 allele (P c = 0.001), +550 G/G genotype (P c = 0.04), and A and G alleles (P c = 0.03) after Bonferroni correction.

Table 1 Distribution of HLA-DPB1 alleles in controls, women with CSCC and those with HPV-16 positive CSCC
Full size table
Table 2 Genotype and allele frequencies of the HLA-DPB1 +550 G/A polymorphism in controls, women with CSCC and those with HPV-16 positive CSCC
Full size table

In order to clarify whether interactions between oncogenic HPV-16 infection and HLA-DPB1 variants would affect susceptibility or progression to CSCC, we further stratified CSCC patients by positivity of HPV type 16. The results showed that the frequencies of DPB1*13:01 allele increased significantly (OR = 1.83, 95 % CI 1.31–2.57) and the DPB1*05:01 allele decreased significantly (OR = 0.65, 95 % CI 0.52–0.80) in women with HPV-16 positive CSCC in comparison with controls, which remained significant after Bonferroni correction (P c = 0.004 and 0.0007 for DPB1*13:01 and *05:01 alleles, respectively) (Table 1). However, the significance of increased DPB1*02:01 allele frequency (OR = 1.36, 95 % CI 1.02–1.81) in women with HPV-16 positive CSCC disappeared after Bonferroni correction (P c = 0.27). For the +550 G/A polymorphism, no significant difference in genotype and allele distribution was observed between women with HPV-16 positive CSCC and controls (Table 2). LD analysis of the HLA-DPB1 variants between exon 2 alleles and +550 G/A SNP revealed high LDs in both controls and patients (D′ = 0.99 and 0.95, respectively).

We also analyzed the possible haplotypes of HLA-DPB1 alleles and +550 G/A SNP in controls, women with CSCC, and those with HPV-16 positive CSCC (Table 3). Among the eight haplotypes, we found that haplotype frequencies of DPB1*02:01-A and DPB1*13:01-G increased significantly and DPB1*05:01-G decreased significantly in both CSCC patients (OR = 1.31, 95 % CI 1.03–1.67; OR = 1.55, 95 % CI 1.15–2.08; OR = 0.72, 95 % CI 0.61–0.85) and HPV-16 positive CSCC patients (OR = 1.51, 95 % CI 1.12–2.02; OR = 1.84, 95 % CI 1.30–2.62; OR = 0.63, 95 % CI 0.51–0.79), whereas DPB1*03:01-G frequency increased significantly (OR = 1.64, 95 % CI 1.07–2.53) only in HPV-16 positive CSCC patients as compared with controls. The significance remained for the DPB1*02:01-A haplotype (P c = 0.05 for HPV-16 positive CSCC patients), DPB1*05:01-G haplotype (P c = 0.0009 and 0.0004 for CSCC and HPV-16 positive CSCC patients, respectively) and DPB1*13:01-G haplotype (P c = 0.03 and 0.004 for CSCC and HPV-16 positive CSCC patients, respectively) after Bonferroni correction.

Table 3 Analysis of HLA-DPB1 haplotypes in controls, women with CSCC and those with HPV-16 positive CSCC
Full size table

Discussion

In the present study, we evaluated whether HLA-DPB1 variants and their haplotypes were associated with the susceptibility to cervical cancer in Taiwanese women. We found a decreased risk of CSCC in association with the HLA-DPB1*05:01 allele and +550 G/G genotype and G allele and an increased risk in association with the +550 A allele. When the analysis was restricted to the subgroup of women with HPV-16 positive CSCC, the association of HLA-DPB1*05:01 allele still existed and the DPB1*13:01 allele was shown to associate with CSCC risk. However, some nonsignificant associations such as DPB1*03:01 allele (OR = 1.52, 95 % CI 0.99–2.33), +550 A/A genotype (OR = 1.51, 95 % CI 0.97–2.36), and G (OR = 0.82, 95 % CI 0.65–1.02) and A (OR = 1.23, 95 % CI 0.98–1.54) alleles found in the HPV-16 subgroup could be due to the small sample size. In addition, DPB1*05:01-G and DPB1*13:01-G haplotypes were revealed to confer protection and risk in both CSCC patients and HPV-16 positive CSCC patients, respectively. Another risk haplotype, DPB1*02:01-A, was also found in the HPV-16 positive CSCC women. Our results suggest that the HLA-DPB1 gene plays a significant role in the pathogenesis of CSCC in Taiwanese women.

Certain HLA class II alleles have been reported to influence cervical cancer development by affecting the interactions between host immune response and HPV infection [23–26]. The DRB1 and DQB1 loci therefore have been thoroughly investigated in several cervical cancer association studies [5]. However, the HLA-DPB1 locus was seldom studied because of its few polymorphisms and low cell surface expression levels [27]. A significant increase in DPB1*13:01 allele in cervical cancer patients and a significant decrease in DPB1*04:01:01 allele in HPV-16 positive cervical cancer patients were previously reported in a central Chinese population [10]. Another study in southern Chinese women demonstrated an increased risk of cervical cancer with both HLA-DPB1*02:02 and *13:01 alleles [11]. In contrast, associations between HLA-DPB1 allele and cervical intraepithelial neoplasia were not found among southwestern American Indian women [9]. It is likely that differences in the study design, ethnicity, sample sizes, cancer types, and pertinent risk factors may result in the disconcordance between these studies and ours.

A number of genome-wide association studies have identified a SNP in the 3′ UTR of HLA-DPB1 gene (+550 G/A, dbSNP ID: rs 9277535), which was associated with chronic hepatitis B virus infection in various Asian populations [28–30]. These findings were validated subsequently in other studies [31–35]. The +550 G/A polymorphism has been reported to regulate HLA-DPB1 mRNA expression according to the mRNA levels obtained from both liver cells and macrophages [12]. Recently, Jiang et al. [36] also have investigated this SNP in cervical cancer and found that +550 GA, AA, and GA/AA genotypes were associated with cervical cancer susceptibility in Chinese women. Our findings of increased CSCC risk in association with +550 A allele in the current study are also consistent with the results of Jiang et al. [36].

The molecular mechanisms underlying the association between CSCC risk and HLA-DPB1 gene are still not well understood. However, on the basis of our results, it is reasonable to infer that HLA-DPB1*13:01 and +550 A alleles are able to affect the antigen presentation and cellular expression of HLA-DPB1 molecules. These specific HLA-DPB1 alleles therefore are likely to associate with persistent HPV infections and thus increase the risk of CSCC. If these associations can be replicated in different populations, it could help to provide better prevention and treatment strategies for women to carry these risk alleles. However, this study still has its limitations need to be considered. The correlations between HLA-DPB1 gene variants and viral load, length, and severity of the HPV infection cannot be addressed due to lack of these data. Furthermore, comparison between HPV-infected individuals with and without CSCC would better clarify the link between HPV infection, host immune response to the HPV and cervical cancer.

Our findings indicate that the HLA-DPB1*13:01 and +550 A alleles are significantly associated with CSCC risk in Taiwanese women. In addition, significant increase in CSCC risk was observed for the HLA-DPB1*13:01-G and DPB1*02:01-A haplotypes. Large-scale epidemiological studies are required to consolidate the HLA-DPB1 role in the development of cervical cancer.