Feasibility of self-sampling and human papillomavirus testing for cervical cancer screening in First Nation women from Northwest Ontario, Canada: a pilot study.

Background The incidence of cervical cancer is up to sixfold higher among First Nation women in Canada than in the general population. This is probably due to lower participation rates in cervical cancer prevention programmes. Objective To raise screening participation in this underserved population by launching an alternative approach to (Pap)anicolaou testing in a clinic-namely, vaginal self-sampling followed by human papillomavirus (HPV) diagnostics. Methods Good relationships were established with a First Nation community of the Northern Superior region in Northwest Ontario, and then 49 community women, aged 25-59, were recruited, who provided a vaginal self-sample and answered a questionnaire. Frequency distributions and cross-tabulations were used to summarise the data. Associations between categorical variables were assessed using the χ(2) test of association, or the Goodman-Kruskal γ if both variables had ordered categories. Self-collected samples were tested for integrity and HPV using optimised molecular biological methods. Results The majority of participants (87.2%) were amenable to future HPV screening by self-sampling. This finding was independent of age, educational level and a previous history of abnormal Pap tests. Interestingly, the preferred way to learn about sexual health remained through interaction with healthcare professionals. As defined by the presence of a housekeeping gene, self-sample integrity was high (96%). Using polymerase chain reaction-based Luminex typing, the overall HPV positivity was 28.6% (ie, with either a low- or high-risk type) and 16.3% were infected with a high-risk type such as HPV16. Conclusion In this pilot study of First Nation women, self-sampling and HPV testing was well received and self-sample quality was excellent. A larger survey to be conducted in other Northern Superior communities in Northwest Ontario will determine whether this approach could become a viable screening strategy for First Nation women.

Introduction

Cervical cancer is among the top three cancers affecting women world wide1 and the third most common cancer in Canada among women aged 20–49.2 Aboriginal populations appear to be particularly affected by this disease. In Canada, cervical cancer is up to sixfold higher in First Nation women than in the general population in the Northwest Territories,3 Manitoba4 and Ontario.5 Similarly, aboriginal women from Australia and the USA have a higher cervical cancer prevalence than the general population in those countries.6 7

Most women who develop cervical cancer have been infrequently or never screened,8 yet such screening is crucial for early detection of precancerous lesions. Accessing health information and preventive medical services can be challenging for First Nation women9 10—their communities are generally rural and remote, transportation is a limiting factor and culturally appropriate, on-site health and educational services may be inadequate. These challenges (as well as the lack of an electronic database to identify seldom or never screened Ontario women) probably contribute to irregular participation of First Nation women in cervical screening in comparison with other Canadian women.4 11 12

Lack of accessible or appropriate screening facilities could be overcome by offering a screening test based on self-collection. Indeed, the Canadian National Aboriginal Health Organization suggested that HPV testing based on self-sampling may be a good alternative to (Pap)anicolaou testing to increase participation among First Nation women.13 In addition, a comprehensive review by the International Agency for Research on Cancer Working Group concluded that human papillomavirus (HPV) testing is a justifiable strategy for cervical cancer prevention.14 Self-collection of vaginal samples for HPV testing has been investigated as a potential cervical screening method in several populations, with good uptake (reviewed by Stewart et al15 and by Huynh et al16and references therein17–20). Two Swedish studies found that, among women who had not been screened for more than 6 years, 32–58% participated in self-sampling.18 19 Similar findings have been obtained in a recent Canadian study, suggesting that Caucasian women who do not participate in cervical cancer screening programmes may be willing to provide a self-collected specimen instead.17 Furthermore, self-collection has been observed to be as reliable as sampling carried out by a doctor for the detection of high-risk HPV associated with an increased risk of cervical cancer.21

A study on cervical cancer screening uptake based on self-sampling and HPV testing among First Nation women has never been conducted in Ontario. Before beginning a large investigation in 10 Northern Superior communities in Northwest Ontario (box 1), we conducted a pilot study with 49 First Nation women in the largest of these communities. Our approach was based on convenient self-sampling and sensitive HPV testing, rather than (Pap)anicolaou screening. To assess the feasibility of this alternative method we used a questionnaire in which demographics and cervical cancer knowledge, self-sampling and sexual health were investigated. Sample adequacy and HPV testing methods were also evaluated.

Fort William First Nation

Gull Bay First Nation

Lake Nipigon First Nation

Long Lake No 58 First Nation

Pays Plat First Nation

Pic River First Nation

Pic Mobert First Nation

Red Rock First Nation (Lake Helen)

Rocky Bay First Nation

Whitesand First Nation

Methods

Participating First Nation community

Fort William First Nation (FWFN), the community that participated in our pilot study, is situated near Thunder Bay on the northern shore of Lake Superior in Northwest Ontario, Canada. FWFN was created in 1853, as a result of the 1850 Robinson-Superior Treaty. With a total of 1798 individuals registered, of whom 832 live on-reserve, FWFN is the largest of the Northern Superior communities (box 1). The mean population of all communities is 313 (range 70–832). We chose these communities for our investigation since they are part of one strategic region of the Anishinabek Nation inhabiting the northern shore of Lake Superior from Pigeon River to Batchawana Bay. They are under the healthcare portfolio held by the Regional Grand Chief in FWFN.

Approvals

After a meeting with the band council a research agreement, identifying potential benefits for the participants' community, data ownership and plans for dissemination and publication of the results, was signed in September 2009 by FWFN Grand Chief Peter Collins. The agreement adhered to guidelines formulated by the First Nations Information Governance Committee through ownership, control, access and possession.22 The study was also approved by the local research ethics board of Thunder Bay Regional Health Sciences Centre (TBRHSC REB#2009125). The on-site Dilico family health team (DFHT) agreed to recruit participants and provide feedback about the study approach.

Participant recruitment

A DFHT nurse practitioner (project nurse) served as primary contact for the participating women. To enrol approximately 50 volunteers, recruitment and information dissemination was carried out between 16 November and 18 December, 2009 through a community meeting with an information poster; flyers posted in public places within the community; flyers sent to all community households through the weekly Band newsletter and flyers distributed at parenting workshop nights through the Band office and DFHT staff. Women were not approached individually. All participants automatically took part in incentive draws for five, CAD$100 grocery certificates.

Participant eligibility

Eligibility for our pilot study required that participants: were female; self-reported First Nation ethnicity; registered in FWFN; were aged 25–59 years and had a command of the English language. For safety reasons, women who knew they were pregnant or menstruating were excluded. Women were enrolled from age 25 rather than age 30 because Canadian First Nation women exhibit earlier onset of cervical cancer than the general population4 and HPV testing is recommended at an earlier age.23

Protocol

Eligible women who wanted to participate in the pilot study contacted DFHT staff. The research nurse provided information orally and through the informed consent form, which was explained to, and signed by, the volunteer before taking a self-sample. After taking the self sample, the questionnaire was answered, numbered and sent to the research team in a sealed envelope. Participants could choose whether or not they wanted their test results sent to their healthcare provider (HCP) or to their homes, or to both.

Self-sampling

The project nurse provided each participant with a sterile, plain polyester Dacron swab15 and transportation tube (Copan Diagnostics Sterile Plain Swab; licensed in Canada by Inverness Medical: #552C). Swabs were sent at ambient temperature in the transportation tube to the National Microbiology Laboratory in Winnipeg. For confidentiality, HPV test results were blinded to the project nurse (sent in individually sealed envelopes, each identifiable by the participant's number). This ensured that members of the research team did not have access to the names of the participants, and the project nurse, to individual test results.

Self-sampled DNA testing

DNA from self-samples was purified using Roche MagnaPure (automated, magnetic bead-based DNA extraction; Roche, Mississauga, Canada), validated specifically for HPV genotyping. Sample integrity was assayed by amplifying the housekeeping gene β-globin by PCR, as described previously.24

HPV testing was done by Hybrid Capture II (HCII), based on an antibody capturing RNA:DNA hybrids.24 HCII generically detects the 13 most common high-risk HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68) at a viral load that best correlates with cervical dysplasia. For specific typing, our in-house Luminex technology,25 based on nested PCR amplification,26 was concomitantly used on all 49 participant samples The Luminex method is less expensive than comparable commercial tests and has been successfully used previously.27

Reporting of results

Only samples that were positive for high-risk HPV were reported to the participants as “Your sample is positive with a high-risk HPV” followed by the corresponding HPV type with which they were infected. Cases testing negative for high-risk, but positive for low-risk HPV, were reported to the participants as “Your sample is negative for high-risk HPV”. When communicating HPV test results to the participating women, it was emphasised that this study was not a substitute for their biennial Pap test. Women who tested positive for a high-risk HPV type were requested to contact their family doctor or the local colposcopy clinic at the Thunder Bay Regional Health Sciences Centre. The research team provided names and contact information of the principal investigator (IZ) and the colposcopist (NE).

Statistical analyses

All analyses were carried out using PASW (formerly SPSS), version 18.0. Data obtained from the questionnaires were examined by frequency distributions and cross-tabulations. Associations between primary questions (general, self-sampling and sexual health) and secondary questions (age, education and abnormal Pap history) were assessed with the Goodman–Kruskal γ (when both variables were ordinal) or Pearson's χ2 test (p≤0.05 was considered significant). For 2×2 tables with any expected counts <5, we used the ‘N−1’ χ2 test.28

Results

The questionnaire (Appendix) was divided into three sections: general questions (including participants' demographics—table 1), questions about self-sampling and questions about sexual health—that is, HPV and cervical cancer (table 2).

Table 1

Demographics of pilot study participants (n=49, unless otherwise indicated)

Metric	Number	Percentage
Ethnicity
First Nation	47	96.0
Metis	2	4.0
Age range*
25–39	27	56.2
40–49	12	25.0
50–59	9	18.8
Education†
University undergraduate	23	48.9
College	12	25.5
High school or lower	12	25.5
Smoker
At some point	45	91.8
Tampon user	45	91.8

Only 48 participants responded to the age question.

Only 47 participants responded to the education question.

Table 2

Self-sampling and sexual health questions of pilot study participants

Self-sampling	Sexual health
87.2% willing to participate in self-sampling in the future	Educational methods: 35% prefer learning through HCP 31.6% want to use audiovisual material 9.8% want to learn on their own
67% prefer self-sampling to HCP sampling	7.5% want to learn with their partner
77.1% found self-sampling easy and 61.7% found it comfortable	Knowledge, attitude and behaviour: 87.8% have some to very good knowledge about cervical cancer 69.4% have some to very good knowledge about HPV 83.7% did not know that both men and women can be HPV-infected Relevance and comprehension of questions: 74% found questions important 69% found questions easy

HCP, healthcare provider.

General questions

All participants reported having had previous Pap tests, with 67.3% (33/49) screened at least biennially. Fifty-one per cent (25/49) of participants reported a previous abnormal Pap test; this statistic is almost 10-fold higher than that generally seen in Ontario women.2 Participants with previous abnormal Pap tests reported that they had undergone (52%, 13/25) or were undergoing (36%, 9/25) treatment. Participants without regular Pap tests (32.7%, 16/49) reported a hiatus of up to 20 years. Regular participation in Pap screening was associated with a higher level of education (p=0.035) but not with age or previous history of abnormal Pap tests. Willingness to self-sample was not associated with previous level of participation in cervical screening. Four participants were unsure about whether they had received HPV vaccine (Gardasil); everyone else reported that they had not.

Self-sampling questions

As shown in the Appendix, questions 11 and 12 each had nine ordinal response options, ranging from 1=easy/comfortable to 9=difficult/uncomfortable. For analysis, both questions were recorded such that 1–3 became ‘easy/comfortable’, 4–6 ‘mid-range’ and 7–9 ‘difficult/uncomfortable’. For acceptability of self-sampling, 77.1% (37/48) found self-sampling easy, 6.3% (3/48) were mid-range, 16.7% (8/48) found it difficult and one did not answer (question 11). Likewise, 61.7% (29/47) participants experienced comfort with self-sampling, 23.4% (11/47) were mid-range, 14.9% (7/47) were uncomfortable and two did not answer (question 12). Thus the majority of participants found self-sampling easy and were comfortable with using it. Consequently, 87.2% (41/47) indicated willingness to participate in self-sampling screening in the future; 8.5% (4/47) did not know if they would participate more regularly; 4.2% (2/47) answered no; and two did not answer (question 13). Sixty-seven per cent (32/48) preferred self-sampling rather than an HCP taking the sample; 18.8% (9/48) had no preference, while 14.6% (7/48) preferred an HCP to take their sample (question 14). For cross-tabulations of questions 11–14 with age, educational level or a previous history of abnormal Pap tests, no statistical significance was achieved in any case.

Sexual health questions

Educational methods

We received 133 suggestions from the 49 participants about the best way to provide education about sexual health (question 15). The preferred way to learn about sexual health was through interaction with an HCP: 35% (47/133). This was followed by the use of audiovisual material—that is, watching a DVD or looking at a poster: 31.6% (42/133). Learning “on my own” and “together with my partner” were the least attractive options: 9.8% (13/133) and 7.5% (10/133), respectively.

Cervical cancer and HPV: knowledge, attitude and behaviour

Participants' knowledge about cervical cancer and HPV was scored as follows, with each of two questions having five possible correct answers (questions 16 and 17; 19 and 20): 1/5 correct=some knowledge; 2/5 correct=fair knowledge; 3/5 correct=good knowledge and 4 or 5 correct=very good knowledge. Importantly, 87.8% (43/49) had knowledge ranging from “some to very good” and about half of the participants had “good or very good” knowledge about cervical cancer. For HPV biology, 69.4% (34/49) of participants had knowledge ranging from “some to very good” and about half of the participants had “good or very good” knowledge. Awareness about cervical cancer and HPV biology was not significantly associated with age, educational level or a history of abnormal Pap tests. Most participants (83.7%; 41/49) were not aware that both men and women can contract HPV infection. Information about cervical cancer and HPV was mainly obtained from an HCP—65.9% (29/44) and five participants did not answer the question or know about cervical cancer and HPV.

Relevance and comprehension of questions

As shown in the Appendix, questions 22 and 23 had nine ordinal response options ranging from 1=important/easy to 9=not important/difficult. For analysis, these questions were recorded such that 1–3 became ‘important/easy’, 4–6 ‘mid-range’ and 7–9 ‘not important/difficult’. Seventy-three per cent of participants (36/49) found the questions important, 6.1% (3/49) were mid-range and 20.4% (10/49) found them unimportant. Sixty-nine per cent (34/49) found the questions easy, 14.3 (7/49) were mid-range and 16.3% (8/49) found them difficult. Thus the majority of participants found the questions that they were asked to answer in the questionnaire both easy and important. The importance of the questions and ease of answering them were positively related (p<0.001). Comfort was also related to ease of answering and importance: those who felt most comfortable with self-sampling perceived the questions to be easier (p=0.000) and more important (p=0.001). However, no statistical significance was reached when both questions were cross-tabulated with age, education or abnormal Pap test history.

HPV testing and typing

Integrity of the self-sampled DNA was high, with 47/49 (96%) testing positive for the β-globin housekeeping gene. Overall, HPV prevalence was 28.6% (14/49); this is within the 10–30% range found in the Canadian adult population (Society of Obstetricians and Gynecologists, Ottawa, Canada).

HPV typing indicated the prevalence of all HPV types (oncogenic and non-oncogenic) in our participants. The detected high-risk HPVs (n=8) included types 16, 35, 52 and 58, which all belong to the phylogenetically related species A9 (HPV 16, 31, 33, 34, 35, 52 and 58).29 Interestingly, only one case was positive for HPV 39, a member of the other high-risk species, A7 (HPV 18, 26, 30, 39, 45, 51, 53, 56 and 59).29 Several women (n=6) were positive for low-risk types 13, 54, 83, 89 and 90, which belong to the species A929 or to other low-risk groups.30 Of eight positive cases with high-risk HPV examined using Luminex, only four cases were positive using the less sensitive HCII method. No association was found between testing positive for high-risk HPV and a previous history of abnormal Pap tests.

Discussion

This initiated investigation examines HPV testing based on self-sampling in First Nation women in Ontario for the first time. Our pilot study relied on HPV-specific DNA assays of vaginal swabs provided by participants to a blinded nurse volunteer, complemented with participant feedback via a detailed questionnaire (Appendix). The participants in this pilot study were positively inclined towards self-sampling, preferring self-sampling over HCP sampling. If it becomes clear that self-sampling is an acceptable screening strategy for First Nation women in other Northwest Ontario communities, offering self sampling might be a significant alternative for the recruitment of First Nation women for cervical cancer screening. Some women (14.7%) reported discomfort with self-sampling in our pilot study, yet surprisingly there was no association between discomfort and preference for HCP sampling.

The majority of our study participants had at least some knowledge of cervical cancer and HPV. This acute awareness could be attributable to the patient demographics in our small sample size: they reported higher formal educational levels than those of average First Nation women31 and several participating women were local health centre employees. An unexpected finding was the low awareness that HPV can infect both men and women.

Although self-sampling was widely embraced, our participants still preferred receiving information about sexual health from an HCP and/or using audiovisual material; self-study and learning with a partner were much less popular. Indeed, most participants obtained HPV knowledge through an HCP, confirming the important role of this professional group in First Nations' health education.

A reliable cervical cancer screening programme has to use state of the art technology. Our study used the most common, ‘best practice’ self-sampling device described in several Canadian studies—the Dacron swab.15 The use of vaginal tampons32 is not recommended because DNA extraction from tampons is time consuming and inefficient (Dr Alberto Severini, unpublished results). Two available tools can be used for cervical screening: the Pap test,33 with its high specificity but low sensitivity (detecting only 50% of high-grade cervical lesions), and the highly sensitive (close to 100%) HPV tests.11 34 35 As a primary screening tool, HPV testing can lead to increased detection of high-grade cervical lesions and allow larger screening intervals than the Pap test,33–37 probably resulting in lower costs and higher screening participation. Furthermore, HPV testing can be conveniently performed on self-collected samples, which further reduces HCP hours and costs. Consequently, our approach included HPV testing and typing.

Self-sample integrity in our study was found to be excellent and at least similar to, or even better than, that found in other studies.23 The overall HPV prevalence of 28.6% (and 16.3% for high-risk types) was within the upper range of the overall Canadian population.2 Interestingly, typing revealed almost exclusively HPV types phylogenetically like type 16, but not type 18. Similar findings were reported recently in a study involving aboriginal women from a Northern Plain American Indian reservation outpatient clinic.38 Of note, our more sensitive in-house Luminex technique detected twice as many positive cases of high-risk HPV as did HCII because HCII was developed to detect only clinically relevant cases whereas the Luminex technique detects latent HPV infections.

Limitations

Despite the small sample size of our pilot study, similar studies among Caucasian populations in Canada and elsewhere are in agreement with our findings.15–20 Our study population differs from that which might be expected in First Nation communities, which renders generalisation difficult. The high participation in regular Pap screening may be connected to a higher level of education in our participants than among most Canadian First Nation women. Indeed, we found a positive, statistically significant association between higher education and participation in cancer screening. On the other hand, participants who did not have regular Pap screening reported intervals of up to 20 years. Self-reported rates have to be considered cautiously because of over-reporting. Indeed discrepancies between self-reporting and medical charts have been published by several independent studies39 with concordance rates between 65% and 89%. Our participants also had easy access to the collaborating health centre in FWFN, which advocates Pap tests for First Nation women and may explain the rather high reported rate of previous Pap tests. The situation differs for members of the other Northern Superior communities who are required to travel to larger cities like Thunder Bay for their health needs (Lee Sieswerda, epidemiologist, Public Health Unit Thunder Bay, personal communication).

Conclusions and future directions

Our findings indicate that HPV testing based on self-sampling is feasible among First Nation women in Ontario. The majority of women agreed that self-sampling would be the preferred way of taking the sample. We are aware, however, that the women who took part in this pilot study reported having had cytology at least once and that accessing unscreened First Nation women is challenging, a factor that will be dealt with in our larger study. A high sample quality and HPV prevalence, comparable to that of the general Canadian population, was obtained in this pilot study. Based on the success of this pilot study, we will conduct a larger-scale study of cervical cancer screening in 10 Northern Superior communities in Northwest Ontario. In view of the demographics in the communities we will be able to recruit more than 800 women. Owing to over-rated self-reporting when assessing the screening history,39 a chart review of previous Pap screening will be performed. A key question to be answered in the larger study will be how best to reach and provide sexual health education to underscreened women. We will also ask participating women to explicitly rate their discomfort by comparing HCP sampling with self-sampling.