Predictive Markers and Cancer Prevention
Human papillomavirus type 16 E6 variants and HLA class II alleles among Japanese women with cervical cancer
The enhanced oncogenicity of particular human papillomavirus type 16 (HPV16) E6 variants is population-dependent, implying the involvement of additional genetic cofactors. This study was designed to investigate the association between E6 variants and human leukocyte antigen (HLA) polymorphism within a Japanese population. Fifty-seven women with HPV16-positive cervical cancer were analyzed for E6 sequence variation and its relationship to HLA class II alleles. Compared with local controls (n = 138) and published controls (n = 916), DRB1*1501 and DQB1*0602 frequencies were significantly increased among patients with HPV16 E6 prototype (n = 11). Additionally, DRB1*1502 was positively associated with a particular E6 variant designated D25E (n = 25), although we could not find a significant association between HLA class II alleles and L83V variants (n = 16). Our observations suggest that a specific match between E6 variant proteins and HLA types may contribute to HPV16-related cervical carcinogenesis. © 2003 Wiley-Liss, Inc.
Infections by oncogenic human papillomaviruses (HPVs) are established as a major risk factor for cervical cancer.1 HPV16 is the most frequently detected papillomavirus in invasive cervical cancer worldwide. HPV16 E6 and E7 proteins are known to inactivate the function of oncosuppressor proteins p53 and pRb, respectively, and immortalize human keratinocytes in vitro.2
Recently, certain HPV16 E6 variants having < 2% sequence variation compared with the HPV16 prototype strain have been reported to confer an increased risk for cervical cancer in a given population.3, 4, 5, 6 Interestingly, the oncogenicity of specific E6 variants differs between various ethnic populations.7 The E6 L83V variant, which has valine (V) substituted for leucine (L) at residue 83, represents a higher risk for cervical cancer in a Swedish population. However, the L83V variant does not carry a significant risk in either Italian or Czech populations. Population-related oncogenicity of HPV16 carrying specific E6 variations suggests the involvement of additional genetic cofactors, for instance human leukocyte antigen (HLA) types. Additionally, there are data suggesting that HLA association with cervical cancer shows HPV type specificity.8 However, little is known about a possible link between HPV intratypic variations and HLA types.
To investigate the role of HLA polymorphism among HPV16 E6 variants in cervical carcinogenesis, we analyzed HPV16 E6 sequence variation and its relationship to HLA class II alleles in 57 Japanese women with HPV16-positive cervical cancer. Since there is little evidence suggesting a possible association between HLA class I alleles and cervical cancer, the present study focused on HLA class II alleles. Although the sample size of the present study was small, we found some significant associations between E6 variants and HLA class II alleles.
MATERIAL AND METHODS
Among 161 women who underwent treatment for primary cervical cancer at University of Tokyo Hospital and Saitama Cancer Center between 1997 and 2001, 57 Japanese women with HPV16-positive cervical cancer [6 carcinoma in situ (CIS) and 51 invasive cervical cancer (ICC)] were tested for E6 sequence variation and HLA class II alleles. Biopsied tissues were collected before treatment and informed consent was obtained from all the patients. The diagnosis of CIS and ICC was confirmed histologically. All CIS and ICC samples enrolled in our previous study of HPV16 E6 variants were also included in the present study.5 Since precursor lesions have the potential to regress spontaneously, samples with cervical dysplasia were excluded from the present study to obtain more conclusive results.
To reduce the risk of chance findings, HLA class II frequencies of women with HPV16-positive cervical cancer were compared with those of 2 control groups. One of the control groups consisted of 138 healthy Japanese subjects from the same geographical area as the cervical cancer patients (Tokyo and Saitama), while the other control group was quoted from a published study of 916 healthy Japanese subjects.9
HPV16 DNA detection
We detected HPV DNA in cervical samples by polymerase chain reaction (PCR)-based methodology described previously.10 In brief, total cellular DNA was extracted from cervical samples by a standard sodium dodecyl sulfate (SDS)-proteinase K procedure. HPV DNA was amplified by PCR using consensus primers for the HPV L1 region. HPV types were identified by restriction fragment length polymorphism (RFLP) that has been shown to identify at least 26 types of genital HPVs.11
HPV16 E6 sequencing
The entire coding region of the HPV16 E6 open reading frame (nt 83–559) from cervical DNA samples was amplified by PCR using HPV16 E6-specific synthetic primers as previously described.5 All samples were successfully amplified with E6-specific primers. Purified PCR products were sequenced by the dideoxy termination method, and DNA sequence data were analyzed with sequence analysis software, DNASIS v.3.3 (Hitachi Software Engineering, Tokyo, Japan). Nucleotide changes were accepted only when confirmed by repeated PCR and sequencing in both directions.
For this study, the HPV16 E6 sequence reported by Seedorf et al.12 was regarded as the prototypical HPV16 E6 protein sequence. Sequence variations in the HPV16 E6 DNA sequence leading to amino acid changes were regarded as HPV16 E6 variants.
Cervical tissues from patients and blood leucocytes from healthy controls were used for HLA genotyping. Total cellular DNA was extracted from these specimens and amplified by PCR using locus-specific primers. All samples were initially typed at the HLA-DRB1 and -DQB1 loci using a commercially available reverse sequence-specific oligonucleotide (SSO) probe typing kit (Dynal RELI SSO; Dynal Biotech, Oslo, Norway). For subtyping, group-specific amplifications were performed as previously described.13 DRB1 and DQB1 alleles were identified by single-strand conformation polymorphism (SSCP) and restriction fragment length polymorphism (RFLP) using the PCR products. RFLP was performed for the samples that were difficult to genotype with SSCP only.
HLA class II allele frequencies among patients and controls were compared by Fisher's exact test or chi-square test. When an expected cell value in the 2 × 2 tables was less than 5, Fisher's exact test was used. To correct for multiple testing, logistic regression analysis was performed on the data from patients and local controls, as proposed by Farewell and Dahlberg,14, and p-values were adjusted for multiple alleles (17 DRB1 and 10 DQB1 alleles) excluding rare alleles in both patients and controls. We also analyzed the data using Bonferroni adjustment, a conventional method for correcting for multiple comparisons, although several investigators have indicated that this method is likely to miss significant differences.15, 16 All analyses were carried out using JMP 4.0J statistics package (SAS Institute, Cary, NC). p-values were considered to be significant when below 0.05. An association of certain E6 variants with HLA class II alleles was considered significant when we obtained consistent findings in comparison with the 2 control groups.
HPV16 E6 variations
DNA sequence analysis of the HPV16 E6 region in DNA derived from CIS and ICC biopsies revealed 10 variations with predicted E6 amino acid changes at residues 10, 16, 24, 25, 27, 56, 58, 78, 83 and 113 (Table I). In all cases tested, frame shift mutations derived by the insertion or deletion of a nucleotide was not observed and no nucleotide alteration conferred a premature stop codon. HPV16 E6 variants were found in 50% (3/6) of CIS and 84% (43/51) of ICC DNA samples, respectively. The rank order of incidence of each amino acid alteration was as follows: D25E (44%) > L83V (28%) > E113D (19%) > R10T, and H78T (4%) > C16H, H24N, D56N, I27R, N58S (2%). Interestingly, E113D was always associated with either D25E or L83V substitutions, whereas no cases harbored both D25E and L83V concurrently.
Table I. Distribution of HPV16 E6 Variations
HLA class II alleles and HPV16 E6 variants
The distribution of HLA-DRB1 and –DQB1 alleles in 57 patients with HPV16-positive cervical cancer and the 2 control groups is shown in Table II. We could not find a significant association between HPV16-positive cervical cancer and HLA class II alleles.
Table II. Correlation Between HLA Class II Carrier Frequency and HPV16 E6 Variants
|DRB1|| || || || || || || || || || || || |
|DQB1|| || || || || || || || || || || || |
To examine a putative link between specific HPV16 E6 variants and HLA class II alleles, HPV16 E6 variants were subdivided into 3 different groups: E6 prototype (n = 11), D25E variants (n = 25) with or without another variation, and L83V variants (n = 16) with or without another variation. A significant association between certain E6 variants and HLA class II alleles was observed for 3 alleles (DRB1*1501, DRB1*1502 and DQB1*0602; Table II).
DRB1*1501 frequency was significantly increased among patients with the E6 prototype sequence compared with local controls [45% vs. 13%, odds ratio (OR) = 5.6, 95% confidence interval (CI) = 1.7–18.0, p = 0.01, p corrected by logistic regression analysis (pc) = 0.03] and published controls (45% vs. 12%, p = 0.006), while DRB1*1501 frequency among patients with L83V or D25E variants was similar to that among controls. In addition, DQB1*0602 frequency was also significantly increased among patients with the E6 prototype sequence [45% vs. 12% (local controls), OR = 6.4, 95% CI = 2.0–20.4, p = 0.01, pc = 0.02; 45% vs. 12% (published controls), p = 0.006]. Also, DRB1*1501 and DQB1*0602 frequencies were significantly higher among patients with the E6 prototype compared with patients with nonprototype variants (45% vs. 12%, p = 0.02, for each allele). By contrast, DRB1*0405 and DQB1*0401 frequencies were decreased among patients with the E6 prototype sequence [0% vs. 28% (local controls), p = 0.03, for each allele], but the significance disappeared after adjustment for other alleles.
DRB1*1502 was overrepresented among patients with D25E variants compared with local controls (40% vs. 23%, OR = 2.2, 95% CI = 0.9–5.3, p = 0.07, pc = 0.04) and published controls (40% vs. 17%, p = 0.002), whereas DRB1*1502 frequency among patients with E6 prototype or L83V variants was similar to that among controls.
DQB1*03032 was more frequent among patients with the L83V variant than among controls, but the difference did not reach statistical significance [44% vs. 28% (local controls), p = 0.16, pc = 0.12]. Neither DRB1*04 nor DRB1*07 correlated with the L83V variant in our study, although other studies suggested the possible associations of DRB1*07 and the DR04-DQ03 haplotype with L83V variants.7, 17
The HLA data were also analyzed by the conventional Bonferroni method to correct for multiple comparisons. After Bonferroni adjustment, the association between DRB1*1502 and D25E variants remained significant in comparison with published controls (corrected p = 0.04), while the other associations lost significance due to limitations imposed by the sample size.
The distribution of HPV16 E6 variant proteins among Japanese women with cervical cancer was considerably different from those reported in other ethnic populations.18 For instance, the incidence of the E6 prototype (European variant) and L83V variant (another European variant) was 19% and 28% in the present study, 53% and 40% in North America, 40% and 44% in Europe, 25% and 52% in Central and South America, 54% and 6% in Southeast Asia and 8% and 2% in Africa, respectively. The D25E variant (Asian variant) was far more frequently detected in Japan (44%) compared with Southeast Asia (26%), North America (3%), Europe (2%), Central and South America (0%) and Africa (0%). In our previous study, we reported that nonprototype E6 variants, especially D25E variants, carry a higher risk for cervical cancer in the Japanese population.5
The present study indicated some significant links between particular E6 variants and HLA class II alleles. DRB1*1501 and DQB1*0602 frequencies were significantly increased among patients with the E6 prototype. Additionally, DRB1*1502 was positively associated with the D25E variant. One study suggested an increase in DRB1*1501 and DQB1*0602 among patients with the HPV16 E6 prototype, although the association did not reach statistical significance.19 The association of HLA DRB1*1502 with a particular HPV16 E6 variant has not been previously reported.
In the present study, we could not find any significant association between L83V variants and HLA class II alleles. Failure to find a significant association may be due to the small sample size and frequent combination with other variations (mainly the E113D variation). One study suggested an increased frequency of HLA DRB1*07 frequency among cancer patients with L83V variants in a Dutch population (p = 0.08).17 Another recent study reported that the DR04-DQ03 haplotype is positively associated with L83V variants in Swedish, Italian and Czech populations.7 To address a possible link between L83V variation and HLA class II alleles, larger studies may be required.
To date, several groups have suggested a significant association of DRB1*1501 and DQB1*0602, or the corresponding haplotype 1501/0602 with HPV16-positive cervical cancer,8, 19, 20 while other groups have reported no association.17, 21 However, these conflicting results may be explained by different distributions of the E6 prototype among HPV16-positive isolates. Although DRB1*1501 and DQB1*0602 were significantly associated with the E6 prototype in the present study, their association with HPV16-positive cervical cancer was not statistically significant because of the low prevalence of E6 prototype among Japanese women with HPV16-positive cervical cancer. To fully evaluate HLA associations with HPV-induced cervical cancer, the analysis of both HPV intratypic variations and HLA polymorphism, as demonstrated in this study, is necessary.
The mechanism by which certain E6 variants confer an increased risk for cervical cancer is not fully understood. Several E6 variations are located in the functional or antigenic domains of E6 protein, potentially introducing different biological properties. An in vitro study has demonstrated that a certain E6 variant (African variant) has a different biological activity in the degradation of p53.22 Another possible hypothesis is that amino acid substitutions in the E6 protein may result in reduced immunogenicity with consequent evasion of host immune surveillance. In fact, the E6 R10T variant that affects a putative HLA-B7-restricted CTL epitope is more frequently found among patients with HLA-B7-positive cervical cancer.23 Since HLA molecules play a central role in determining which peptide can be presented to specific T cells in the recognition of viral antigens, the observed links between specific E6 variants and HLA types appear to support this hypothesis. Additionally, HLA types are known to have population-related distributions. Thus, the specific association between E6 variants and HLA types may contribute to the population-related oncogenicity of HPV16 E6 variants.
In the present study, we have identified some significant links between certain E6 variants and HLA class II alleles, suggesting that HLA polymorphism may influence the oncogenicity of HPV16 E6 variants, most likely through an immunologic mechanism. However, the sample size of the current study was small. To further evaluate the risk of HPV16 E6 variants in relation to HLA polymorphism, larger studies will be required. The establishment of HLA associations with HPV intratypic variations may allow insight into the mechanism of immunologic control of HPV infections.
The authors thank Graham R. Leggatt and Rachel L. de Kluyver for helpful discussion and Mami Kimura and Sumiko Mitsumata for assistance.