A prospective double-blind cross-sectional study of the accuracy of the use of dry vaginal tampons for self-sampling of human papillomaviruses

Authors


Abstract

Objective

To evaluate if a dry vaginal tampon can be used to accurately detect high-risk human papillomaviruses (HPV) and so be used as a cervical screening tool for the early detection of high-grade cervical intraepithelial neoplasia (CIN2+).

Design

Prospective double-blinded cross-sectional study.

Setting

Colposcopy unit in North London.

Population

Women referred for colposcopy with both abnormal and normal cervical cytology were invited to participate in this study.

Methods

Women inserted a dry tampon in the vagina before colposcopic examination. Polymerase chain reaction (PCR) HPV test was carried out on the dry tampon. Cervical samples were collected by the colposcopist for HPV testing using Hybrid Capture (an HC2 kit).

Main outcome measure

Detection of CIN2+.

Results

In all, 501 women participated in the study. The majority of participants (69%) were in the 25–35-year age group. Overall sensitivity and specificity for detection of CIN2+ by vaginal tampon were 76% (95% confidence interval [95% CI] 65–85) and 61% (95% CI 56–66), respectively, and HC2 had a sensitivity of 92% (95% CI 83–97) with a specificity of only 46% (95% CI 41–51). Sensitivity ratio was 0.83 (95% CI 0.73–0.94) (P = 0.004) and Specificity ratio for CIN2+ was 1.33 (95% CI 1.22–1.45). Indicating that the tampon test was significantly (P < 0.0001) more specific. Sensitivity decreased and specificity improved with increasing age.

Conclusions

Dry tampon test showed reasonably high sensitivity and specificity when detecting CIN2+. The majority of participants (98%) approved of the use of dry tampons as a method of sample collection, which could be an effective alternative method for detecting HPV infection.

Introduction

There is little doubt that well-organised cytology-based screening programmes for cervical cancer have been effective in reducing its incidence and preventing premature deaths. However, morbidity and mortality secondary to cervical cancer continue to be global problems, with estimates of nearly 471 000 new cases and 233 000 deaths annually.

The majority of cervical cancer cases (60%) are associated with inadequate screening[1] and are, therefore, preventable. Though the introduction of cervical screening programmes to most countries has led to a decline in mortality, there are still women who, due to inconvenience, the time taken, or discomfort, do not participate in them. In the developing world, it may not be women's choice but the lack of adequate screening resources that impedes progress in improving the outcomes of this disease.

The main aetiological agent implicated in the development of cervical intraepithelial neoplasia (CIN) is human papillomavirus (HPV), which is present in 99.7% of cervical cancers.[2] Types 6 and 11 are the most common low-risk types; however, these are associated with benign disease. Types 16 and 18 have been reported to be the most common high-risk HPVs, being the most frequently found in cervical cancers.

There are around 20 types that are considered to be ‘high-risk’.[3] The main bulk of evidence derived from randomised controlled trials shows reduction of the incidence of CIN3+ and of invasive cervical cancer by offering HPV-based instead of cytology-based screening.[4, 5]

Direct testing for HPV relies on molecular biological techniques using nucleic acid probes, as HPVs cannot be cultured in vitro. There are several molecular methodologies for the detection of HPV including: the polymerase chain reaction (PCR), DNA sequencing, branch-chain DNA analysis and reverse blotting.

Two of the most commonly used techniques for HPV detection are PCR and Hybrid Capture-2® (HC2; Digene Co., Gaithersburg, MD, USA). These techniques are employed to detect the most clinically relevant high-risk HPV types. HC2 assays use cocktails of specific RNA probes that are directed towards target HPV DNA sequences.

Patient-obtained vaginal samples (self-collected) for analysis of HPV DNA can have the same sensitivity to detect high-grade cervical lesions and invasive cancer as a Papanicolaou smear.[6] This is emerging as a tool with the potential to make screening more efficient and convenient. The concept of self-sampling is gaining popularity. A recent meta-analysis also suggested that HPV testing on a self-sample can be suggested as an additional strategy to reach women not participating in the regular screening programme.[7]

In the present study, patient-obtained vaginal samples have been evaluated to determine the feasibility and accuracy of self-sampling for HPV DNA as a tool for detecting cervical pre-cancers. We also evaluated the acceptability to women of the self-sampling test using tampons. The CIN detection rate from the self-sampling method was compared with that from clinician-collected HPV samples.

Methods

Women were referred to a busy colposcopy unit (in London, UK), where this trial was carried out, with abnormal cervical cytology, postcoital bleeding, intramenstrual bleeding, or an abnormal looking cervix. A small proportion of the women attended with benign conditions (cervicitis, polyps and ectropion) and normal cervical smear results. Women who were pregnant, immune-compromised, had undergone hysterectomy or could not speak adequate English, were excluded from the study.

A designated research clinic was set up to conduct this prospective double-blind cross-sectional study. British Society for Colposcopy and Cervical Pathology accredited colposcopists performed all the colposcopic assessments.

Before their attendance at the unit information leaflets explaining the study were sent by post to the women with their appointment letters. A further verbal explanation was given on arrival. Written informed consent was taken from all participants. Epidemiological data were stored on an electronic database.

Women were invited to insert a standard-sized commercially manufactured vaginal tampon themselves. The changing area for women was situated in the corner of the colposcopy room. Once prepared, women had a short walk to the examination couch. On average, tampons remained in the vagina for between 90 and 120 seconds. Tampons were removed by the participants immediately before colposcopic examination, and placed in sterile collection pots (plastic urine collection pots). These were placed in sealed, transparent bags and stored at room temperature. Stickers with only the study ID number were used to identify samples (Figure 1). Six batches made up of 500 samples were sent to Australia using a courier service. The analytical PCR-based HPV assay was carried out by SDS Pathology (Maquarie Park, NSW, Australia). Because dry tampons, with no preservatives, were used to transport our samples to the laboratory; there was less possibility of contamination or degradation of samples.

Figure 1.

Study design for evaluating the accuracy of self-collected dry tampon samples.

SDS Pathology has developed a real-time PCR method for HPV detection and typing based on Specific Pathogen Free (SPF) primer sets and run on the Corbett Rotorgene 3000 platform. The most commonly used primers for the detection of HPV are the consensus primers GP5+/6+ and the degenerate primers MY09/11.[8, 9] One major problem with the GP5+/6+ and the MY09/11 primers is that they substantially underestimate the prevalence of some HPV types (type 55 and to a lesser extent type 52). The SPF primers are a multiple set of consensus primers that amplify a small fragment, around 65 bp of the L1 gene.[10] This is a relatively small product compared with the GP primers, which amplify a 150-bp fragment, and the MY09/11 primers, which amplify a 450-bp fragment.

The assay is a two-step process, involving an initial screening step, followed by typing of HPV. In the first step samples were screened for HPV, using a variation of the SPF primers to form a sequence able to pick up the majority of high-risk HPV types. The screening result would therefore be either positive or negative. DNA extracted from Factor V Leiden samples was used as controls in screening PCR. The second step is confirmatory and types the HPV present. Typing was carried out using 17 sets of primers specific for types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68, 70, 73 and 82.

Further samples were collected by the colposcopist using a Hybrid Capture kit brush (Qiagen Gaithersburg Inc., Gaithersburg, MD, USA [previously Digene Corp.]). These samples were sent to an independent laboratory in London (TDL) for analysis using the hybrid capture technique. Digene's Hybrid Capture 2® platform uses specific RNA probes, hybridisation, antibody capture and signal amplification to test the genetic material of infectious agents. The Hybrid Capture 2® System amplifies the signal by combining antibody capture with chemiluminescent signal detection.

Colposcopy-directed punch biopsies were taken when indicated. Further cone biopsies of the transformation zone of the cervix were taken. Where punch biopsies were reported CIN2+ or colposocopic appearances were suggestive of high-grade premalignant changes. Histology was reviewed and reported in the hospital's pathology laboratory.

Results

Of 511 women recruited, ten did not provide a tampon sample (of whom six also did not provide a sample for HC2) and a further six had an inadequate PCR result from the tampon samples. Results are presented for the remaining 495 women. Most women (65%) were in the age group 25–34 years, and the main indication for referral was either a borderline (31%) or mild (38%) abnormality on the cervical smear test. Colposcopic impressions and HPV detection rates in participating women according to referral cytology are shown in Table 1. Forty-nine per cent of the women had some form of histology taken, and this was mainly in the form of punch biopsy. Of these, most women either had CIN1 (44%) or CIN2 (20%) identified. Six samples (of 501) collected by tampon were inadequate for HPV testing: two of them were from women with CIN2 on biopsy (and none had CIN3 or cervical glandular intraepithelial neoplasia [CGIN]).

Table 1. Colposcopic impressions and HPV detection rates in participating women, according to referral cytology
Referral cytologyColposcopy impressionTampon testHC2 test

Normal

(n = 47)

Normal (23)Positive (36)Positive (33)
Low grade (21)Negative (10)Negative (14)
High grade (2)Inadequate (1) 
Unsatisfactory (1)  

Low-grade cytology

(n = 346)

Normal (97)Positive (149)Positive (135)
Low grade (208)Negative (192)Negative (211)
High grade (37)Inadequate (5) 
Unsatisfactory (4)  

High-grade cytology

(n = 108)

Normal (21)Positive (62)Positive (75)
Low grade (24)Negative (46)Negative (33)
High grade (63)  
Unsatisfactory (0)  

Punch or excisional biopsy detected 75 cases of CIN2+ (among 495 women with adequate tampon and HC2 samples) including one case of CGIN, 76% of which were identified by the vaginal tampon, compared with 96% detected by the cervical brush. Overall sensitivity and specificity for detection of CIN2+ by vaginal tampon were 76% (95% confidence interval [95% CI] 65–85) and 61% (95% CI 56–66), respectively, and HC2 had a sensitivity of 92% (95% CI 83–97) with a specificity of only 46% (95% CI 41–51). Sensitivity ratio was 0.83 (95% CI 0.73–0.94) (P = 0.004) and Specificity ratio for CIN2+ was 1.33 (95% CI 1.22–1.45). Indicating that the tampon test, Tampap, was significantly (P < 0.0001) more specific.

Twenty-six cases of CIN3 or CGIN were seen, 77% of which were identified using the vaginal tampon and 89% of which were detected with the cervical brush and HC2. The sensitivity of the tampon test for CIN3+ was 77% (95% CI 56–91). Sensitivity ratio, to that of HC2 was 0.87 (95% CI 0.71–1.07). However, this was not significant enough (P = 0.38). The vaginal tampon missed 18 cases of high-grade disease (including CIN2), whereas the cervical brush method failed to identify six cases. Out of 198 tests reported as HPV negative using the cervical brush, the vaginal tampon identified 16 as being positive, and one of these was CIN3 and another CIN2 (Table 2). Out of 297 testing positive for HPV by the cervical brush, 92 of these were reported as negative with the vaginal tampon. Further analysis of the performance of these tests in different age groups revealed that both methods had decreasing sensitivity and improved specificity with increasing age (Table 3).

Table 2. Accuracy of the tampon test at detecting CIN (compared to HC2)
Histology resultHC2 testTampap test
Negative (%)Positive (%)TotalNegative (%)Positive (%)Total
No biopsy142 (54.8)117 (45.2)259172 (67.7)82 (32.3)254
<CIN253 (31.6)115 (68.5)16884 (50.6)82 (49.4)166
CIN23 (5.8)49 (94.2)5212 (24.5)37 (75.5)49
CIN33 (12)22 (88)255 (20)20 (80)25
CGIN01 (100)11 (100)01
Total201 (39.8)304 (60.2)505274 (55.4)221 (44.7)495
<CIN2192228427256164420
CIN2+66978185775
Table 3. Accuracy of the test to detect CIN2+ in different age groups
AgeHC2 testTampap test
Sample sizeSpecificity (%)Sensitivity (%)Sample sizeSpecificity (%)Sensitivity (%)
  1. Linear trend with different age groups. Specificity improves with age by Tampap test (P = 0.044) and by HC2 test (P = 0.001). Sensitivity decreases with age by Tampap test (P = 0.030) and by HC2 test (P = 0.11).

20–2920835.595.220255.685.7
30–4927751.290.927463.464.5
50–692064.766.71976.550.0

Ninety-eight per cent of the women tested liked the concept of using the self-sampling test to screen for cervical disease. Those who used tampons were pleased (100%) with it. However, 18% of postmenopausal women felt mild discomfort during insertion of the tampon.

Discussion

Main findings

Our study showed that over all ages combined, sensitivity of the self-sample was good; that of the clinician samples was excellent. Conversely, the specificity of HPV testing was better on the self-sample (Table 4). Although the specificity of both tests was low, this is a referral population and the use of either test as a triage would have substantially reduced the numbers referred to colposcopy. Both positive and negative predictive values are similar in both methods (Table 5). The κ-value between these two groups was 0.5728.

Table 4. Sensitivity, specificity odds ratio of the HPV test by the tampon test and by Hybrid Capture (95% CI) for detection of CIN2+
PrevalencePr(A)Vaginal tamponCervical brush
SensitivityPr(+|A)76% (95% CI 65–85)92% (95% CI 83–97)
SpecificityPr(−|N)61% (95% CI 56–66)45.7% (95% CI 41–51)
Odds ratioLR(+)/LR(−)4.949.68
Table 5. Positive predictive value (PPV) and negative predictive value (NPV) of both tests at different prevalence rates
 15%12%18.6%
Tampon test
PPVPr(A|+)25.820.232.1
NPVPr(N|−)93.489.896.1
Hybrid capture test
PPVPr(A|+)23.218.528.5
NPVPr(N|−)9793.598.9

Our study also attempted to discover the preferred collection method of women. Just over 2% (2.15%) of women declined use of the tampon, mainly due to inexperience in using one. The majority of women found the test simple to use and preferred it to smear tests at primary-care clinics (general/family physicians’ surgery). All the women liked the concept of self-testing at home as an efficient time-saving and convenient method.

Strengths and limitations

In this study, we tried to assess the accuracy of a unique method of HPV test by using dry vaginal tampons. These were cross-checked by a well-established and widely used Hybrid capture method. To our knowledge, there is no published study that reports on this type of HPV test. This is a double-blinded study where the participants, clinicians and laboratory personnel were not aware of the HPV status of the participants. However, we also accept the fact that most of the women who attended our colposcopy clinic were referred with abnormal cytology results. Hence, they were already in a high-risk group and very likely to be HPV positive. To address this issue, we also recruited women with normal cytology. Though the number in the normal cohort was not very high.

We acknowledge that the sample size of 501 is not big enough to reach any robust conclusion. However, this study can be a starting point for larger population-based studies. Most of the women in our study already participated in normal screening programmes, hence they were fully motivated. However, their views on HPV sampling methods can be applied to the general population.

Interpretation

In high-income countries, a substantial proportion of women diagnosed with cervical cancer have never had a Papanicolaou test or were infrequently screened. In the UK, recent studies have suggested a decline in participation in the routine screening programme, particularly among women aged 25–29 years, to below the target rate of 80%.[11, 12] However, almost 50% of all cervical cancers occur in the age group of 25–49 years.[12-14] Many factors may be responsible for the lack of uptake, including inconvenience, lack of time, discomfort associated with a gynaecological examination, as well as cultural objections to the latter. In low-resource regions, additional problems include a lack of resources and access to adequate screening programmes. Offering a simple, inexpensive and convenient self-test that respects individual privacy may improve the participation of women who might otherwise be reluctant to undergo screening, and those who live in areas with poor access to health care.

Recently, it has become clear that self-collected vaginal specimens from women who have received appropriate instruction produce HPV test results similar to cervical specimens collected by healthcare professionals.[15, 16] A recent meta-analysis suggested that testing for high-risk HPV on self and clinician-sampled specimens is similarly accurate with respect to CIN2+ detection both in high risk women and in the screening population.[7] Another meta-analysis also reported a high level of concordance of 0.87 (95% CI 0.82–0.91) between self-collected and physician-collected samples.[17] The authors indicated that variations in clinical performance are likely to reflect the use of different combinations of collection devices and methods of HPV testing. In the present study we found that PCR-based HPV assays on tampons have a sensitivity of 85.1% (estimated prevalence of 18.6%) for detection of CIN2+ compared with 97% by HC2. However, the specificity was 65.6% for HPV testing using the former technique, with HC2 having a specificity of 50.6%. This study also found high concordance (a κ-value of 0.572) between the two techniques.

With respect to resource-poor regions where the cytology screening programme may be restricted, the high sensitivity of vaginal tampon-testing is crucial, as women may only be screened at a few points in their lives. Other studies have shown the superior sensitivity of HPV testing for detecting high-grade cervical disease in comparison to conventional cytology. Potentially, self-sampled HPV testing could serve as a primary screening tool and use of more specific tests, cytology or HPV genotyping, could be reserved for those who test positive for HPV.

In our study, the majority of women found the test simple to use and preferred it to smear tests at primary-care clinics (general/family physicians’ surgery). All the women liked the concept of self-testing at home as an efficient time-saving and convenient method. So, it is plausible that self-sampling may increase compliance with cervical cancer screening programmes. In another recently published study[18] on vaginal self-sampling by standard versus dry swabs for HPV testing found no differences in sensitivity for CIN1 or worse between the two self-HPV tests. Women in this study also found the self-sampling method highly acceptable.

Studies in western countries with organised screening programmes[19, 20] have shown that the provision of self-sampling kits to women who did not participate in conventional screening programmes, may improve their screening adherence. Further investigation is needed to evaluate the responses to self-sampling and its cost-effectiveness in different demographic and geographic settings.[21]

The wide variation in HPV types in the vaginal samples was not unexpected. In view of the paucity of robust evidence establishing a relationship between the presence of less common high-risk HPVs in the vagina and high-grade CIN, vaginal HPV subtyping should not influence clinical management.

Conclusion

The current study reinforces the increasing body of evidence suggesting that offering self-sampling for HPV DNA testing may be an appropriate alternative for high-risk HPV detection. This may lead to a decrease in the incidence of high-grade CIN and ultimately reducing mortality from cervical cancer. However, the feasibility and impact of introducing this method beyond the research context requires further evaluation. Challenges include the cost of testing and ensuring adequate follow up and treatment is available to those who test positive. We have found that a self-sampling tampon is an effective way of detecting high grade CIN and is widely acceptable to women as the preferred method of sample collection.

Disclosure of interests

All the authors declare that they do not have any conflict of interest.

Contribution to authorship

AMK was the study coordinator and chief investigator, helped to conceive the project, designed the study, formulated the methodology and wrote the manuscript. PS carried out the data analysis, interpretation and data presentation, and manuscript editing. AS gave the idea and assisted in securing a research grant and helped in editing the manuscript.

Details of ethics approval

This study was approved by National Research Ethics service (East Central London REC1, ref: 08/H0721/88).

Funding

The research project was mainly funded by the Research fund of the Colposcopy Unit of Whittington Health NHS. Unconditional grants were also received from Home Test Direct Pty. Ltd. (owner of Tampap) to cover the administrative costs of the project.

Acknowledgements

The authors are grateful to research Nurse Veronica Larbi and trainee doctors Dr Larisa Corda, Dr Yuiliya Ashton and Dr Deepali Bhatte for helping AMK in his research clinic to take consent from participants and to prepare the samples. We also thank Mrs Freeman- Wang, Mr Narendra Pisal and Mr Oliparambil Ashok Kumar for their support and guidance.

Ancillary