HLA class II allele control of HPV load in carcinoma in situ of the cervix uteri

Authors


Abstract

Human papillomavirus (HPV) infection is the most important risk factor for development of cervical carcinoma. Carriers of certain HLA class II alleles, e.g., DRB1*1501 and DQB1*0602, are more prone to HPV 16 infection and cervical carcinoma, whereas other alleles, e.g., DRB1*1301 and DQB1*0603, render carriers less susceptible to the disease. In our study comprising 484 cases and 601 controls, we examine the effect of HLA class II alleles on viral load of the oncogenic types HPV 18/45 and HPV 31 and risk of developing cervical carcinoma in situ. We find that carriers of the commonly reported protective DRB1*1301 and DQB1*0603 alleles have lower HPV 18/45 load compared to noncarriers and a lower risk of developing HPV 18/45-positive cervical carcinoma. This provides further evidence that the HLA class II-mediated immune response to HPV is important for controlling viral load and outcome of an infection. © 2005 Wiley-Liss, Inc.

HPV is the major etiologic risk factor for cervical carcinoma and is found in the majority of cervical tumors. Although infection with oncogenic forms of HPV are common among young women, less than 1% of those infected develop cervical carcinoma.1, 2, 3 Both the length of the infection and a high viral load several years before diagnosis have been shown to increase the risk of developing cervical dysplasia and carcinoma in situ (CIS).4, 5, 6 Certain HLA class II alleles, such as DRB1*1501 and DQB1*0602, have been shown to increase the risk for cervical carcinoma, most strongly in HPV 16-positive carcinomas.7, 8, 9, 10, 11, 12, 13 We have previously shown that the increased carcinoma risk primarily depends on that the alleles render carriers more susceptible to HPV16 infections.12 In addition, carriers of these HLA susceptibility alleles had higher HPV 16 load in their cervical smears compared to HPV 16-positive noncarriers. A plausible hypothesis is that HLA is involved in the cellular immunity against HPV. Also, HLA alleles making carriers more susceptible to HPV infections are likely to be less efficient in triggering immunity and inducing viral clearance. As a consequence, carriers of susceptibility alleles harbor higher HPV load than carriers of more efficient alleles. The relation between viral load and carcinoma development has been most extensively studied for HPV 16, presumably because it is the most prevalent HPV type in cervical tumors throughout most of the world.1 We have previously examined the effect of HPV 18/45 and HPV 31 viral load on the risk for cervical CIS.14 Our results showed that higher viral load of HPV 18/45 or HPV 31 increases the risk of CIS, although the risk magnitude is less than that observed for HPV 16. Therefore, it is possible that HLA alleles could influence carcinoma development by affecting viral load also for other HPV types besides HPV 16.

A meta analysis demonstrated a protective effect of the DRB1*1301 and DQB1*0603 alleles against development of cervical carcinoma in 18 of 19 studies.15 The presence of such a consistent effect across studies indicates that it would be important to test the effect of protective HLA class II haplotypes on HPV viral load. Our study focuses on the association between HLA class II alleles and viral load for some of the most frequent oncogenic HPV types besides HPV 16 and present evidence for an inverse relationship between viral load and carrier frequency of protective HLA alleles.

Material and methods

Subjects

The participants in our study were selected from a cohort of women resident in Uppsala County, Sweden, between 1969–1995.16 The participants had to fulfill the following entry criteria: (i) born in Sweden; (ii) less than 50 years of age at entry; (iii) their first smear classified as cytologically normal (PAP = 1). Histologic specimens (either a small biopsy or a complete “cone”) from all eligible cases were reviewed by an experienced pathologist (Prof. Jan Pontén). All cases with invasive carcinoma or adenocarcinoma were excluded from the study and only cases with squamous cell carcinoma in situ, CIS, were included. For each case, 5 separate controls individually matched by date of entry to the cohort (±90 days) and by year of birth were randomly selected from the study cohort. The women included as controls had to be alive without having developed carcinoma in situ or invasive carcinoma before the date of diagnosis for their corresponding matched case. In addition, the first smear of the controls had to be classified as cytologically normal (PAP = 1).5, 16 First smears were reviewed by a cytotechnician blinded to case-control status. Since we did not want to exclude individuals with transient infections, controls having abnormal smears during the study period were not excluded. The study was performed with approval from local ethics committees.

Collection of smears

The archival smears included in our study were collected during the period of 1969–1995. A total of 4,732 smears were included, with a mean of 3.4 smears per control and 5.2 smears per case.

DNA extraction

DNA was purified from Papanicolaou stained archival smears, as previously described.17

Quantitative PCR

The amount of HPV 18/45, HPV 31 and HPV 16 DNA was estimated using real-time PCR.18, 19, 20 The assay employed has been described in detail previously.21 In brief, the assay generates individual estimates for HPV 16 and HPV 31 and a combined estimate for HPV 18 and HPV 45. In the case of 18/45, the viral load estimate reflects either a single HPV type or a combination of the 2 viral types. Three microliters of DNA extract is added to each real-time PCR reaction to a final reaction volume of 25 μL. Each sample is analyzed in 2 different PCR reactions: Reaction 1 detects and quantifies HPV types 16, 31, 18 and 45 (HPV 18 and 45 together). Reaction 2 detects and quantifies the amount of the human nuclear gene hydroxymethylbilane synthase, used to estimate the number of nuclear genome equivalents in the sample. No template controls were included in each real-time PCR run to monitor contamination of reagents. The generated Ct-value represents the PCR cycle at which the fluorescence of the real-time PCR reaction passes a threshold value set by the machine. The Ct-value is inversely proportional to the amount of target DNA present in the reaction, i.e., a high HPV concentration in a smear yields a low Ct-value. Samples with Ct-values corresponding to <10 copies of the human control gene were considered as negative and were excluded from the analysis.

HLA typing

From each woman, the smear with the highest amount of genomic DNA based on the real-time PCR assay was selected for HLA typing. Two microliters of the DNA from each smear was used for PCR amplification of the HLA class II DQB1 and DRB1 loci, using biotinylated PCR primers. The PCR products were hybridized to reverse-dot blots containing sequence-specific oligonucleotides, as provided by the manufacturer (Dynal RELI™ SSO by DYNAL, Oslo, Norway). These procedures have been described previously.22 For subtyping, group-specific amplifications of the DR*03, *04, *11, *13 and *14 alleles were performed as previously described.23, 24

Statistical analysis

χ2 Statistics were used to test the association of single HLA alleles or haplotypes with specific HPV types in positive carcinoma as compared to the general frequency among controls (FREQ procedure, SAS software package version 6.12, SAS Institute, Cary, NC). T-tests were performed to detect differences in mean HPV load among carriers and noncarriers of the particular HLA alleles studied, independent of case-control status (TTEST procedure, SAS v6.12), using the Ct-values that follow the normal distribution. Mantel-Haenszel statistics were employed to study the relationship between HLA haplotype carrier frequency and HPV titer. Mantel-Haenszel statistics were computed using case-control status as confounding variable, carrier status of the haplotype or allele as row-variable and Ct-category as column-variable (FREQ procedure, SAS v6.12).

Results

The study initially comprised 494 women diagnosed with cervical carcinoma in situ and 645 age-matched controls. HLA-DRB1 typing results were obtained for 433 cases and 469 controls, whereas HLA-DQB1 typing results were obtained for 440 cases and 476 controls.12 Due to poor DNA quality, incomplete typing results (only 1 of the 2 alleles could be detected) were obtained for 6 cases and 12 controls for DRB1 and 9 cases and 17 controls for DQB1. Since the statistical analysis was based on HLA carrier frequency, these individuals were included in the analysis. Smears with <10 copies of the human control gene were excluded from the analysis. As a consequence, 418 smears from cases and 406 smears from controls were excluded, leaving 484 cases with a total of 1,754 smears and 601 controls with 2,154 smears. A woman was considered HPV positive if one or more of her cervical smear samples tested positive with a Ct-value corresponding to 10 or more HPV DNA copies. In total, 28% (135/484) of the cases and 12% (70/601) of the controls tested positive for HPV 18/45 with the real-time PCR assay. Also, 16% (80/484) of the cases and 8% (46/601) of the controls tested positive for HPV 31 and 61% (294/484) of the cases and 12% (73/601) of the controls tested HPV 16 positive.

After having established the HPV and allele frequencies, we examined whether HLA class II alleles alter the risk of HPV-associated CIS. Due to discordant coverage of the HLA typing and HPV load assay, the analysis includes 442 controls and 420 cases typed for both HPV and DRB1. The distribution of DRB1 allele frequencies is shown in Table I. For HPV 18 and/or 45 the alleles *0101 (16/123 (13%) vs. 99/442 (22%); p = 0.02), *1101 (18/123 (15%) vs. 38/442 (9%); p = 0.05) and *1301 (7/123 (6%) vs. 66/442 (15%); p = 0.01) show significant differences between HPV-positive cases and the controls. For HPV 31, the allele *1301 (5/78 (6%) vs. 66/442 (15%); p = 0.04) show significant difference between HPV-positive cases and the controls. For HPV 16, the results were in accordance with our previous study.25 The analysis includes 429 cases and 454 controls typed for both HPV and DQB1. The distribution of DQB1 alleles is shown in Table II. For HPV 18/45 the allele *0603 (8/124 (6%) vs. 65/454 (14%); p = 0.02) show significant frequency differences between positive women and the controls. For HPV 31, only *0501 display a difference between cases and controls (11/77 (14%) vs. 114/454 (25%), respectively; p = 0.04). In these analyses, no correction of the significance level for the number of tests performed has been done.

Table I. HLA DRB1 Carrier Frequencies Among Cases and Controls in Human Papillomavirus (HPV)-Infected Individuals and Stratified According to HPV Type
AlleleControlsCases
AllHPV 18/45 positiveHPV 31 positiveHPV 16 positiveAllHPV 18/45 positiveHPV 31 positiveHPV 16 positive
(n = 442)(n = 52)(n = 32)(n = 56)(n = 420)(n = 123)(n = 78)(n = 255)
n (%)n (%)n (%)n (%)n (%)n (%)n (%)n (%)
10199 (22)12 (23)8 (25)15 (27)69 (16)16 (13)10 (11)47 (18)
1022 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
1033 (1)0 (0)0 (0)0 (0)4 (1)2 (2)1 (1)3 (1)
30183 (19)10 (19)5 (16)10 (18)84 (20)29 (24)19 (22)43 (17)
40199 (22)14 (27)8 (25)14 (25)105 (25)33 (27)25 (28)61 (24)
4024 (1)0 (0)0 (0)0 (0)5 (1)1 (1)1 (1)4 (2)
4034 (1)0 (0)0 (0)0 (0)6 (1)3 (2)2 (2)4 (2)
40451 (12)9 (17)4 (13)7 (13)46 (11)16 (13)9 (10)23 (9)
4050 (0)0 (0)0 (0)0 (0)3 (1)1 (1)0 (0)1 (0)
4071 (0)0 (0)0 (0)0 (0)5 (1)2 (2)2 (2)4 (2)
4081 (0)0 (0)0 (0)0 (0)1 (0)0 (0)1 (1)0 (0)
70155 (12)5 (10)3 (9)8 (14)68 (16)21 (17)14 (16)50 (20)
80134 (8)3 (6)5 (16)4 (7)49 (12)12 (10)11 (13)29 (11)
8020 (0)0 (0)0 (0)0 (0)1 (0)1 (1)0 (0)0 (0)
8042 (0)0 (0)0 (0)0 (0)1 (0)0 (0)0 (0)1 (0)
8051 (0)1 (2)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
9010 (0)2 (4)1 (3)5 (9)0 (0)2 (2)1 (1)4 (2)
10017 (2)1 (2)0 (0)0 (0)5 (1)2 (2)1 (1)2 (1)
110138 (9)5 (10)4 (13)4 (7)37 (9)18 (15)9 (10)23 (9)
11021 (0)0 (0)0 (0)0 (0)1 (0)0 (0)0 (0)1 (0)
11040 (0)0 (0)0 (0)0 (0)1 (0)0 (0)0 (0)1 (0)
12018 (2)1 (2)1 (3)2 (4)5 (1)2 (2)0 (0)3 (1)
12039 (2)0 (0)0 (0)1 (2)7 (2)1 (1)0 (0)5 (2)
130166 (15)6 (12)6 (19)8 (14)41 (10)7 (6)5 (6)23 (9)
130240 (9)4 (8)4 (13)5 (9)33 (8)5 (4)5 (6)12 (5)
130310 (2)1 (2)0 (0)0 (0)10 (2)1 (1)0 (0)8 (3)
13150 (0)0 (0)0 (0)0 (0)1 (0)1 (1)0 (0)1 (0)
13191 (0)1 (2)1 (3)1 (2)0 (0)0 (0)0 (0)0 (0)
140116 (4)1 (2)0 (0)3 (5)20 (5)7 (6)2 (2)12 (5)
14020 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
14047 (2)0 (0)0 (0)0 (0)3 (1)0 (0)1 (1)2 (1)
1501124 (28)15 (29)6 (19)12 (21)149 (35)40 (33)24 (27)104 (41)
15021 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
15031 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
16016 (1)1 (2)1 (3)2 (4)6 (1)2 (2)0 (0)3 (1)
Table II. HLA DQB1 Carrier Frequencies Among Cases and Controls in Human Papillomavirus (HPV)-Infected Individuals and Stratified According to HPV Type
AlleleControlsCases
AllHPV 18/45 positiveHPV 31 positiveHPV 16 positiveAllHPV 18/45 positiveHPV 31 positiveHPV 16 positive
(n = 454)(n = 55)(n = 33)(n = 56)(n = 429)(n = 124)(n = 77)(n = 261)
n (%)n (%)n (%)n (%)n (%)n (%)n (%)n (%)
201119 (26)14 (25)8 (24)14 (25)121 (28)38 (31)27 (35)71 (27)
301110 (24)19 (35)11 (33)12 (21)114 (27)39 (31)21 (27)78 (30)
302127 (28)15 (27)9 (27)17 (30)130 (30)44 (35)29 (38)68 (26)
30337 (8)3 (5)1 (3)7 (13)35 (8)13 (10)7 (9)22 (8)
40234 (7)3 (5)4 (12)4 (7)49 (11)11 (9)10 (13)28 (11)
501114 (25)13 (24)8 (24)15 (27)81 (19)21 (17)11 (14)53 (20)
5026 (1)2 (4)1 (3)2 (4)6 (1)2 (2)0 (0)3 (1)
50320 (4)1 (2)0 (0)3 (5)23 (5)7 (6)3 (4)14 (5)
602121 (27)15 (27)6 (18)10 (18)151 (35)39 (31)24 (31)105 (40)
60365 (14)6 (11)6 (18)9 (16)42 (10)8 (6)5 (6)25 (10)
60440 (9)3 (5)4 (12)5 (9)34 (8)5 (4)5 (6)12 (5)

We then studied the association between carrier status for susceptible and protective HLA alleles and viral load of HPV 18/45 and HPV 31. A t-test shows that carriers of the alleles DRB1*1301 harbor significantly lower viral loads of HPV 18/45 (p = 0.04) and HPV 31 (p = 0.02). Similarly, carriers of DQB1*0603 have lower loads of HPV 31 (p = 0.02) and a trend for lower loads of HPV 18/45 (p = 0.08). For the haplotype DRB1*1301-DQB1*0603, the viral loads of HPV 18/45 and HPV 31-positive women are lower in carriers compared to noncarriers (p = 0.03 and p = 0.02, respectively). No differences were found for the other susceptibility or protective alleles.

To visualize the difference in mean HPV titer between carriers and noncarriers, we divided the carriers into 3 viral-load percentiles for HPV 18/45. A negative relationship between carrier status and viral load is seen for both for the DQB1*0603 and DRB1*1301 alleles analyzed separately and for the haplotype with these 2 alleles (Mantel-Haenszel statistics, p = 0.02, 0.04 and 0.05, respectively) (Fig. 1). In this analysis, adjustment was made for case-control status because in previous analyses the HPV 18/45 load was found to affect the risk for cervical CIS.14 The corresponding analysis performed on HPV 31 load did not show a significant relationship between load and carrier status, possibly due to the small number of subjects in each group.

Figure 1.

(a) Carrier frequency of the HLA class II DQB1*0603 allele in HPV 18/45-negative women and HPV 18/45 load percentiles. (b) Carrier frequency of the HLA class II DRB1*1301 allele in HPV 18/45-negative women and HPV 18/45 load percentiles.

To rule out systematic differences in the amount of cellular DNA between HPV load groups, potentially caused by variation in sampling over time, we compared the smears' mean content of human DNA. The mean HMBS Ct-values for the smears are similar between the percentiles, and there is no indication of a systematic difference in DNA amount between the viral load groups of HPV 18/45 and HPV 31 (Table III).

Table III. Mean Ct-Values for Human DNA in Smears and Viral Load Groups
 Viral load groupNo. of smearsMean human DNA Ct-valueSD
HPV 18/45Negative85929.41.6
35.9 ≤ Ct < 33.36729.41.4
33.3 ≤ Ct < 29.47229.51.3
Ct ≤ 29.46629.41.5
HPV 31Negative93829.41.5
37.1 ≤ Ct < 34.04229.21.5
34.0 ≤ Ct < 29.73929.41.8
Ct ≤ 29.74529.11.4

Discussion

The results of this and our previous studies reveal a striking association of HLA haplotype with HPV load and cervical carcinoma in situ. The DRB1*1301 and DQB1*0603 alleles are protective against HPV 18/45-associated cervical carcinoma in situ. Here, we have shown that carriers of these HLA alleles have lower HPV 18/45 load relative to carriers of other alleles. In contrast, our previous findings show that carriers of the susceptibility alleles DRB1*1501 and DQB1*0602 have higher viral load of HPV 16 compared to noncarriers and are prone to more long-term infections. Due to the low number of observations and uneven distribution of smears per woman, it is difficult to evaluate the persistence of HPV 18/45 infections. Nevertheless, our results indicate that certain HLA class II alleles affect the risk for progression toward cervical carcinoma and that this effect is coupled to increase or decrease in viral load.

Although several HLA class II alleles are associated with cervical carcinoma development, only weak or no association with HLA class I alleles have been reported.7, 11, 26, 27, 28, 29 Also, in an affected sib-pair study, we have found further evidence that the class II region rather than the region containing the HLA class I loci is linked to cervical carcinoma in situ.30 A high viral load of HPV 16 several years before diagnosis markedly increases the risk of CIS development.6 Similarly, the risk of CIS has been shown to increase with higher viral loads of other high-risk types, although not to the magnitude seen for HPV 16. For HPV 16, the most common high-risk type, an effective cytotoxic T-lymphocyte (CTL) response is important for clearance of an infection.31 It has been shown that HPV 16 E5 has the ability to selectively retain HLA class I A and B complexes in the golgi apparatus, preventing peptide presentation to CTLs.32 This is consistent with our observation that the variation in class I genes may not determine the efficiency of the immune response. Rather, the observed associations imply that HLA class II antigens act indirectly on the CTL response by activation of T-helper cells. If this is correct, it will have consequences for the development of effective vaccines against HPV infection. A prophylactic vaccine, employing HPV 16 VLPs to induce immunity, has shown promising results.33 VLPs have been shown to be endocytosed, processed and presented by APCs to naïve T cells,34 thereby enabling class II presentation of the VLPs to activate the immune system. It is therefore likely that vaccinated women will differ in their capability to mount an effective immune response against an HPV infection, depending on their HLA class II genotype. For therapeutic vaccines, it has been suggested that the optimal vaccines for carcinoma should incorporate tumor-specific cytotoxic as well as helper epitopes.35, 36 In a series of ex vivo experiments, it has been demonstrated that anti-wt p53110-124 CD4+ T cells have the ability to enhance the generation and antitumor functions of CD8+ effector cells.37 Our results indicate that further information is needed regarding the means by which different genetic factors, such as the HLA class II alleles, affect whether an HPV infection will evade or activate an immune response. Studies of the relationship between low-prevalent HPV infections, carrier status of HLA alleles and cervical carcinoma development often suffer from lack of power. However, infection by HPV 18 and 45 is less prevalent in Europe (8.1% and 2.3%, respectively) compared to North America (15.8% and 14%, respectively),1 indicating that powerful studies of the association of HLA haplotype and load of these viral types may be conducted in North American populations.

Our study shows that genetic factors such as the HLA class II alleles influence HPV load of HPV 18/45 and thereby predispose to carcinoma in situ. Protective HLA class II alleles could affect risk of carcinoma in situ risk indirectly via HPV 18/45 viral load, in a manner similar to the situation for the more prevalent HPV 16.

Ancillary