THERE IS STRONG epidemiologic and laboratory-based evidence to support an etiologic role for human papillomavirus (HPV) in the development of benign and invasive cervical disease.1 Although HPV is essential for cellular transformation to occur, which depends on the expression of the viral E6 and E7 oncogenes, additional cofactors are required in the development of disease. Many potential risk/protective factors have been implicated, including marital and sexual history, gynecologic and obstetric events, infectious agents, smoking, oral contraceptives, other contraceptive methods, occupational factors, immunosuppression, and dietary habits.2 Many of these factors are highly intercorrelated, and their precise causal relationships and the biologic basis of oncogenesis remain to be clarified. More recent, multicenter case-control and cohort studies indicate that the number of sexual partners, age at first intercourse, number of pregnancies, and the number of sexual partners of the husbands are independent risk factors.3
In Canada and the United States, the Native American (referred to as “Aboriginal” in Canada) population is at an increased risk for cervical cancer, based on several regional and national studies.4–7 Little systematic data are available on the prevalence of HPV infection and other risk factors for cervical cancer in this population. Some limited data have been collected from Greenlandic Inuit,8,9 Indians in New Mexico,10 and Alaska Natives.11 The prevalence of HPV infection in these populations was lower than among non-Natives, despite their higher risk for cervical cancer. These studies, however, used filter-in-situ and dot-blot hybridization, methods now regarded as insensitive, in detecting HPV infection, although the Greenland study has since been repeated using the polymerase chain reaction (PCR) method.12
This article reports on a study that combines recent advances in molecular diagnosis of HPV infection with a comprehensive epidemiologic survey of potential cofactors in an ethnically mixed urban population in Canada.
A cross-sectional survey was conducted among women attending a community health center located in a predominantly low-income, inner-city area of Winnipeg that serves a large Aboriginal population in its environs. Between November, 1992 and March, 1995, women presenting for routine care who either requested to have a Papanicolaou (Pap) smear or were recommended to do so by one of the three staff physicians were recruited into the study.
Data related to exposure to potential risk factors were collected by an interviewer-administered questionnaire. A gynecologic examination provided cervical cells for cytology and viral DNA studies, and specimens for gonorrhea culture and Chlamydia trachomatis detection (Chlamydiazyme; Abbott Laboratories, Abbott Park, IL).
The questionnaire covered basic sociodemographic data such as age, ethnic group, marital status, education, employment, occupation, and residential history. Although a direct inquiry as to personal income was not made, participants' postal codes were obtained and individuals were then categorized into income levels according to aggregate, census-tract information. Behavioral and clinical information obtained included smoking status, obstetric/menstrual history, contraceptive practices, frequency and results of past Pap smears, several measures of sexual activity (age at first intercourse, frequency of intercourse, and number of lifetime and past-year sexual partners), and history of genital infections.
Two trained nurses with experience in sexually transmitted disease contact tracing, interviewing, and counseling conducted the interviews in a confidential and sensitive manner. Of 1,696 women selected for the study, 1,477 (87%) completed the questionnaires.
Exfoliated cervical cells were collected by using an Ayres spatula in conjunction with a cytobrush or endocervical swab. After preparing a Pap smear slide, the residual cells were placed in phosphate-buffered solution and held at −20 °C until DNA analysis.
The DNA was isolated by proteinase K digestion, phenol-chloroform extraction, and ethanol precipitation.13 The DNA was detected by a PCR procedure using consensus primers (MY 09/11) that amplify a 450-base pair (bp) conserved target sequence in the viral L1 gene.14 The quality of the DNA was assessed by coamplifying the human β-globin gene with primers (GH20 and PC04), generating a 268-bp product.14
Each 100-μl PCR contained 50 mmol/l KCl, 10 mmol/l Tris-Cl, pH 8.5, 4.0 mmol/l MgCl2, 200 μmol/l of each nucleotide, 2.5 U Taq polymerase, and 30 ng of DNA. The HPV and β-globin primers were added to a final concentration of 1 and 0.1 μmol/l, respectively. The 30 cycles of amplification comprised denaturation at 94 °C for 30 seconds, annealing at 55 °C for 30 seconds, and extension at 72 °C for 1 minute. Amplification products were resolved by agarose gel electrophoresis, then 20 μl of reactions yielding a 450-bp product underwent dot-blot hybridization with individual oligonucleotide probes for HPV 6, 11, 16, 18, 31, 33, and 35.14 Aliquots of reaction with a 450-bp product that did not hybridize to one of the probes were considered to contain untyped HPV DNA.
To ensure the validity of the PCR results, several steps to guard against potential contamination were taken. Three physically separate rooms were used, respectively, for the preparation of reagents, isolation of DNA from clinical specimens, and PCR amplification followed by dot-blot analysis. Gilson Microman positive displacement pipettors with disposable tips and pistons (Mandel Scientific Co., Guelph, Ontario, Canada) were used throughout the procedures. Several controls were included in each PCR run and hybridization analysis. PCR reactions were set up with the cloned genome of each HPV type of interest and the amplification products were spotted on membranes and included in each hybridization reaction to verify the sensitivity and specificity of the respective probes. Reagent controls were included in each analysis that consisted of PCR reactions containing all PCR components in the absence of DNA, to verify the absence of contamination in the system. In addition, PCR reactions with human DNA known to be negative for HPV sequences were included in each analysis, to control for potential contamination and the specificity of the consensus primers. The hybridization results of each run were read independently by two individuals.
Pap smears were processed according to the clinic's routine practice and interpreted by the cytology service of the university teaching hospital. Copies of the verbatim reports were reviewed separately by a certified pathologist who assigned grades according to the cervical intraepithelial neoplasm (CIN) system.
Characteristics of Study Participants
Of the 1,477 participants, 42% were Aboriginal. The clinic population was predominantly young, with 73% younger than 30 years of age. Aboriginal women in the study differed significantly from non-Aboriginal women in a variety of sociodemographic, behavioral, and clinical characteristics, summarized in Table 1.
The results of tests for cervical cytology, HPV, gonorrhea, and chlamydial infection are shown in Table 2.
About 14% of specimens provided inadequate samples for HPV detection. Among adequate samples, the overall prevalence of HPV was 33%, with little difference between Aboriginal and non-Aboriginal women. Compared to individuals with inadequate samples, those with adequate samples for HPV detection did not differ in terms of any of the sociodemographic and behavioral variables, or in the rates of current gonorrhea, chlamydial infection, and abnormal Pap smears. All subsequent analyses on HPV involve only the 1,263 subjects with adequate samples.
A single type only was found in 38% of typed specimens, two types in 27%, three types in 19%, and four or more types in the remainder.
The commonest HPV type among Aboriginal women was type 18 (15%), whereas type 16 was the most prevalent among non-Aboriginal women (13%). Infection with type 16 or type 18 can be found in 20% of Aboriginal women and 15% of non-Aboriginal women (not significantly different).
Determinants of Human Papillomavirus Infection
Table 3 shows the prevalence of HPV infection by categories of potential risk factors and the Mantel-Haenszel summary odds ratios, adjusting for Aboriginal status. It can be seen that HPV infection is associated with marital status, condom use, number of lifetime sexual partners, number of sexual partners last year, age at first sexual intercourse, a history of sexual abuse, past chlamydial infection, and current abnormal Pap smear.
Those who use condom “most of the time” or “always” reported the highest prevalence of HPV. Condom use is associated with other aspects of sexual behavior: 85% of “never” users of condoms were monogamous or celibate during the past year, compared with 56% of “always/most of the time” users; only 1.5% of never users had five or more partners, compared with 9.5% of those who used condoms always or most of the time.
A history of sexual abuse is also associated with the age of onset of intercourse: women reporting a history of sexual abuse were nine times more likely than those without such a history to have age of sexual debut younger than 12 years of age (with age 19 years and older serving as the baseline).
Age at first intercourse shows a U-shaped relationship with HPV prevalence. The highest prevalence is among those who first had sex before age 12 years. The prevalence declines among those who first had sex during the teen years, but rises again after age 19 years.
In a multiple logistic regression model consisting of Aboriginal status, smoking status, marital status, age, oral contraceptive use, condom use, age at first sex, lifetime number of sexual partners, number of sexual partners last year, result of Pap smear, and number of past pregnancies, only Pap test was found to be a significant factor associated with HPV infection. When Pap test result was dropped from the model—assuming an abnormal Pap test is a consequence rather than risk factor of HPV infection—both marital status and the number of lifetime sexual partners emerged as significant independent predictors of HPV infection.
A series of multiple logistic regressions limited to two independent variables (marital status and number of lifetime sexual partners) were done separately in Aboriginal and non-Aboriginal women, and separately with groups of HPV genotypes (6/11 vs. 16/18/31/33/35). Table 4 shows that individuals with 20+ partners are more likely than those with one or fewer partner to have HPV detected. The odds ratios associated with HPV types with low oncogenic risk (6/11) tend to be lower than those associated with the high-risk types (16/18/31/33/35). The association is present in both Aboriginal and non-Aboriginal women when all types are considered or when only the high-risk types are considered.
Although Aboriginal women are at increased risk for cervical cancer, the prevalence of HPV infection, detected by the sensitive PCR method, does not differ between Aboriginal and non-Aboriginal women. This is also observed in the study comparing Native Greenlanders and Danes.12 The prevalence of type 18, the most oncogenic genotype, however, is higher among Aboriginal women, among whom there is also a higher proportion with known or potential clinical and behavioral risk factors for cervical cancer, such as early onset and frequency of sexual activity, multiple partners, high parity, smoking, and coinfection with other sexually transmitted microorganisms, especially C. trachomatis. Participation in preventive screening is also lower. The strong association between abnormal cervical cytology and HPV infection is confirmed also by this study: the odds ratio for HPV infection is 9.0, comparing CIN grade III with normal (Table 3).
The “paradoxical” lack of association between HPV prevalence and incidence of cervical cancer has been reported from high-cancer-risk areas such as Greenland12 and northeastern Brazil.15 In Greenland, women with multiple partners are at lower risk for HPV infection, in contrast to herpes simplex virus. This led Kjaer and colleagues to propose that women submitted to high exposure to multiple partners and venereal pathogens may suppress HPV viral replication immunologically, resulting in a lower detection rate.12 In Brazil, Franco and coworkers demonstrated that high-oncogenic-risk HPV types are associated with measures of sexual activity such as multiple partners and age at first intercourse, whereas infection with low-risk types is not.15 In our study, the association between multiple partners and HPV infection can be demonstrated in both Aboriginal and non-Aboriginal women. Detection of both low- (types 6/11) and high-oncogenic-risk (types 16/18/31/33/35) types of HPV is associated with the number of lifetime sexual partners in our study, although the association is stronger with the high-oncogenic-risk types. However, the size of the odds ratio is modest, compared with those observed, for example, in Brazil.15
The proportion of multiple (more than one) genotypes among typed specimens was over 60%, higher than some studies reported in the literature.14,16 The sample sizes of these studies are smaller than ours, with a lower precision of the prevalence estimates. In a study of university students in California, as much as 40% of individuals with identifiable HPV types had two or more types.16 We have exercised great care to ensure that type specificity is not compromised by laboratory contamination.
Of interest is the observation that frequent condom use is not associated with a lower prevalence of HPV. That this could be the result of confounding with other aspects of sexual behavior, because condom users were more likely to have multiple partners, is supported by the lack of significant association with HPV in the presence of other factors in multivariate analyses. However, laboratory and epidemiologic studies have cast doubt on the protective effectiveness of condoms against HPV17; see, for example, a study among prostitutes in Nairobi, Kenya reported in this journal.18 Ascertainment of condom use depends on respondent recall. We also did not explore whether the condom was applied correctly by the individual or her partner or when the condom was used during the sexual act.
In summary, our study supports the association of HPV infection with sexual activity and abnormal cytology. Although there is no current treatment for HPV infection per se, education efforts directed at curtailing and protecting sexual behavior and improving participation in Pap screening will ultimately reduce the incidence of cervical cancer.
1. Schiffman MH. Recent progress in defining the epidemiology of human papillomavirus infection and cervical neoplasia. J Natl Cancer Inst 1992; 84:394–398.
2. Brinton LA. Epidemiology of cervical cancer: Overview. In: Muñoz N, Bosch FX, Shah KV, Meheus A, eds. The Epidemiology of Cervical Cancer and Human Papillomavirus. Lyon, France: International Agency for Research on Cancer, 1992:3–23.
3. Franco EL. Viral etiology of cervical cancer: A critique of the evidence. Rev Infect Dis 1991; 13:1195–1206.
4. Nutting PA, Freeman WL, Risser DR, et al. Cancer incidence among American Indians and Alaska Natives, 1980 through 1987. Am J Public Health 1993; 83:1589–1598.
5. Mahony MC, Michalek AM. A meta-analysis of cancer incidence in United States and Canadian Native population. Int J Epidemiol 1991; 20:323–327.
6. Young TK, Choi NW. Cancer risks among residents of Manitoba Indian reserves, 1970–1979. Can Med Assoc J 1985; 132:1269–1272.
7. Young TK, Frank JW. Cancer surveillance in a remote Indian population in northwestern Ontario. Am J Public Health 1983; 73:515–520.
8. Kjaer SK, Teisen C, Haugaard BJ, et al. Risk factors for cervical cancer in Greenland and Denmark: A population-based cross-sectional study. Int J Cancer 1989;44:40–47.
9. Kjaer SK, Engholm G, Teisen C, et al. Risk factors for cervical human papillomavirus and herpes simplex virus infections in Greenland and Denmark: A population-based study. Am J Epidemiol 1990; 131:669–682.
10. Becker TM, Wheeler CM, McGough NS, et al. Cervical papillomavirus infection and cervical dysplasia in Hispanic, Native American, and non-Hispanic white women in New Mexico. Am J Public Health 1991; 81:582–586.
11. Davidson M, Schnitzer PG, Bulkow LR, Parkinson AJ, et al. The prevalence of cervical infection with human papillomaviruses and cervical dysplasia in Alaska Native women. J Infect Dis 1994; 169:792–800.
12. Kjaer SK, de Villiers EM, Çaahglayan H, et al. Human papillomavirus, herpes simplex virus and other potential risk factors for cervical cancer in a high-risk area (Greenland) and a low-risk area (Denmark): a second look. Br J Cancer 1993; 67:830–837.
13. McNicol PJ, Dodd JG. Detection of human papillomavirus DNA in prostate gland tissue by using the polymerase chain reaction amplification assay. J Clin Microbiol 1990; 28:409–412.
14. Bauer HM, Ting Y, Greer CE, et al. Genital human papillomavirus infection in female university students as determined by a PCR-based method. JAMA 1991; 265:472–477.
15. Franco EL, Villa LL, Ruiz A, Costa MC. Transmission of cervical human papillomavirus infection by sexual activity: Differences between low and high oncogenic risk types. J Infect Dis 1995; 172:756–763.
16. Moscicki AB, Palefsky J, Gonzales J, Schoolnik GK. Human papillomavirus infection in sexually active adolescent females: Prevalence and risk factors. Pediatr Res 1990; 28:507–513.
17. D'Oro LC, Parazzini F, Naldi L, La Vecchia C. Barrier methods of contraception, spermicides, and sexually transmitted diseases: A review. Genitourin Med 1994; 70:410–417.
18. Kreiss JK, Kiviat NB, Plummer FA, et al. Human immunodeficiency virus, human papillomavirus, and cervical intraepithelial neoplasia in Nairobi prostitutes. Sex Transm Dis 1992; 19:54–59.