Prostate Cancer Risk and Serologic Evidence of Human Papilloma Virus Infection

Introduction

There has been substantial progress in epidemiological research on prostate cancer over the last decade (1) . An area of increasing interest is the possible role of number of sexual partners and history of sexually transmitted infections as a risk factor for prostate cancer (2, 3, 4, 5, 6, 7) . However, accurate information on sexual behavior is difficult to obtain, and the association between sexual history and exposure to specific infectious agents may be weak and will vary across populations. Studies of prostate cancer risk in relation to features of sexual life have obtained some evidence for an effect of indicators of high sexual activity, such as early intercourse, and many sexual partners (2 , 4, 5, 6, 7) . We have previously reported tentative evidence from the current study population that marital status, early first intercourse, and a larger number of sexual partners are associated with increased prostate cancer risk (3) . Specific sexually transmitted infections have been associated inconsistently with prostate cancer, with positive associations being reported with syphilis, gonorrhea and HPV4 infection in various studies (2 , 4 , 7, 8, 9) .

There has been particular attention given to HPV infections as possible prostate cancer risk factors, as these are definite causes of several other forms of cancer, particularly cervical cancer (10) . HPV, especially HPV type 18, have tropism for the grandular epithelium and can immortalize prostate epithelial cells (11) . Several studies have reported that HPV DNA was detectable in prostate cancer cells (12, 13, 14, 15, 16, 17) , although other studies have not reproduced this finding (12 , 18, 19, 20) . These inconsistencies in results may be because of technical difficulties in HPV DNA detection. There are several possible reasons for false-positive or false-negative results, most notably PCR contamination, degradation, or cross-linking of DNA and tumor samples not being representative. In addition, control samples from the target tissue cannot be readily obtained from nondiseased persons.

Therefore, serological studies offer an attractive alternative, as they provide an integrated, readily standardized measure of long-term HPV exposure on a biological sample that can be easily obtained from both cases and healthy controls. Indeed, serological studies have contributed substantially to establishing the causal relationship between HPVs and several cancers, such as cervical, vulvar, and oropharyngeal cancers in women and men (21, 22, 23) . They have also generated hypotheses in relation to other tumor sites (24) . With regard to prostate cancer, a prospective study with >20 years of follow-up reported a highly significant increased risk associated with HPV type 18 seropositivity, although the proportion of exposed cases was small (8) . Tentative evidence suggesting an association with HPV type 16 was reported in one serological case-control study (4) but not in others (8 , 19) .

We report here results from a population-based case-control study of prostate cancer, in which serological evidence of infection to three major oncogenic HPV types (16 , 18 , 33) has been evaluated. Serum samples from the same study have been used previously in analyses of steroid hormones (25) and insulin-like growth factor I (26) .

Materials and Methods

Results

The mean value for serological measurement of HPV 16 and 18 was lower for cases than for controls, whereas cases had higher mean values for HPV 33. The distributions for all three of the HPV types were strongly skewed. In simple unadjusted comparisons between groups, the difference between cases and controls with regard to HPV 16, and HPV 18 was not statistically significant. For HPV 33, the difference was also not statistically significant, although the absorbance units were ∼50% higher in cases than in controls (Table 1)⇓ .

The proportion of subjects who were seropositive for HPV 16 and HPV 18 (absorbances exceeding the cutoff value of 0.1) was small, in the range of 11–15%, without any appreciable differences between cases and controls. For HPV 33, the seropositive proportion was 29% among cases and 23% among controls (Table 2)⇓ . Because the age distribution among cases and controls was similar (mean values 69.9 and 69.8 years), age-adjusted logistic regression analyses produced essentially the same results as univariate analyses. With the antibody measurements in their original untransformed form, we found no association between HPV 16 or HPV 18 and risk of prostate cancer in the age-adjusted analyses. However, for HPV 33 there was some evidence of a positive association, which was additionally strengthened after mutual adjustment between the three different HPV types (Table 2)⇓ .

Additional exploratory analyses of HPV 33 indicated an excess risk confined to high values. Indeed, in the antibody measurement interval of absorbances 0.101–0.2 we found an OR, 0.9 (95% CI, 0.5–1.6) for age-adjusted analyses and OR, 1.0 (95% CI, 0.6–1.9) after mutual adjustment. Substantially higher relative risks were obtained for serum titers exceeding 0.2, namely OR, 2.3 (95% CI, 1.2–4.1) in age-adjusted analyses and OR, 3.2 (95% CI, 1.6–6.4) after mutual adjustment.

When HPV antibody levels were analyzed as continuous variables, there was no convincing positive association between HPV 16 or HPV 18 and risk of prostate cancer; indeed, after mutual adjustment, there was some evidence of a negative association in relation to HPV 18. In contrast, significant positive associations were found for HPV 33 both in age-adjusted and, even more convincingly, in mutually adjusted analyses.

Discussion

The results of this population-based case-control study show an association between serological evidence for HPV 33 infection and risk for prostate cancer. The risk was significantly elevated for subjects with high antibody levels against HPV 33. In contrast, no association between the presence of antibodies for HPV 16 or HPV 18 and risk of prostate cancer was apparent.

Most of the earlier studies on the effect of HPV infection on prostate cancer risk have been case series using PCR methods to detect the presence of HPV DNA. Because of the difficulty in obtaining suitable control samples in most of these studies the prevalence of HPV DNA in prostate cancer specimens was compared with that in benign prostatic hyperplasia (BPH) specimens. The results have been inconsistent, with some reports showing no relationship between HPV 16, 18, or 33 with prostate cancer (15 , 18 , 33 , 34) and others reporting a potential effect of HPV 16 (17 , 35) , HPV 18, or HPV 33 (35) . Several studies also failed to detect HPV DNA sequences in any of the specimens, whether cases or controls (19 , 20 , 36) .

Discrepancies between these studies may be attributable to the use of different techniques for HPV DNA detection, including the use of different primer sets (16) . The use of paraffin-embedded archival samples, compared with fresh frozen samples, may also cause inconsistency and nonreproducibility in results (37) . In addition, disagreement between studies may be the result of sampling errors. A lack of association between the presence of HPV DNA and prostate cancer does not definitely exclude a role for HPV infection in prostate cancer, because the presence of HPV DNA does not reflect past exposure (only present exposure), and for carcinogenesis past exposure is likely to be more important.

The detection of HPV immune response in the serum, as was used in the present study, indicates both past and present infection and overcomes many of the limitations of the studies using HPV DNA detection. Our data reported here are consistent with a case-control study comparing prostate cancer cases to BPH controls, which found no evidence for an association between HPV 16 (or HPV 11), whether measured through antibody responses or HPV DNA from prostate specimens, and risk of prostate cancer (19) . Our results are also consistent with a survey of HPV 16 antibodies in patients with epithelial cancer, which found a low prevalence of HPV 16 seropositive prostate cancer cases, both in comparison with other cancer types and to the noncancer study group (38) . However, our data are not consistent with two nested case-control studies, which found a weak positive association between prostate cancer and presence of antibodies to HPV 16 (4 , 8) , a significant positive association with HPV 18 antibodies (8) and a nonsignificant inverse association between HPV 33 antibodies and prostate cancer (8) .

An important strength of this study is that this is the largest study that has been conducted, up until now, investigating the association between HPV infection and prostate cancer. We measured HPV infection using serological, rather than PCR data, and we investigated the independent effects of HPV 16, 18, and 33 on prostate cancer risk. We used a population-based case-control study design, limiting many of the problems associated with hospital-based case-control studies, and we used healthy male controls, rather than controls with BPH.

There are also limitations to our study. It would have been preferable to use a nested case-control design, rather than a case-control design, because we cannot exclude the possibility that our results are produced by virus reactivation by the cancer, or an increased susceptibility for infection caused by prostate cancer. However, this is unlikely to have explained our results, because the association with prostate cancer was specific for HPV 33, whereas no effect was found for either HPV 16 or 18. Because not all of the eligible cases and controls participated in our study, there is room for selection bias. However, the proportion of refusals was low, and earlier reports from this study are in line with current knowledge about prostate cancer (25 , 26) . We cannot exclude the possibility that the association between HPV 33 and prostate cancer risk was produced by chance, and the association appeared to be driven by a few high values. This highlights the need to replicate these findings in other settings.

Inconsistent associations of prostate cancer with seropositivity for different HPV types in different studies could be the result of chance or publication bias, but also because of confounding by some unknown risk factor that may associate with different HPV infections in different populations. Specifically, the spread of the HPV infection is dependent on the contact rates and mixing patterns in the population. These factors vary between different populations and may also vary over calendar time in the same population. Additional studies of the relationship between prostate cancer and other HPV types, notably HPV 33, and their relationship with sexual behavior, could help clarify the possible role of sexual risk factors. Despite the inconsistent epidemiological data, a direct role for HPV 33 in prostate carcinogenesis is biologically plausible because it can be isolated from prostate tissue (12) , and HPV33 is a known oncogenic infection, causally involved in the development of cervical cancer (10) .