Cervical cancer prevention among long-term screening non-attendees by vaginal self-collected samples for hr-HPV mRNA detection

The implementation of nationwide screening programs has caused a reduction in cervical cancer cases in many Western countries [1, 2]. In Sweden, a more than 50% decrease in cervical cancer incidence has been seen since the introduction of an organized screening program in the late 1960s [3]. However, despite huge success in the preventive work against cervical cancer, there are still approximately 500 women being diagnosed with cervical cancer every year in Sweden [4]. The greatest risk factor is non-attendance at cervical cancer screening [5]. In 2017, the coverage level of cervical cancer screening was 82.9% in Sweden [6]. This is below the national recommended coverage level of 85% [7]. With the implementation of screening methods for detection of high-risk human papillomavirus (hr-HPV), the major causal factor of cervical dysplasia and invasive cancer [8], a vaginal hr-HPV self-sampling method performed by the woman herself at home has become an option. These self-collected samples could be one way to reach screening non-attendees and are recommended in the Swedish national guidelines [9]. However, on August 29, 2019, only seven out of 21 regions in Sweden had implemented self-collected sampling of screening non-attendees, making it important to evaluate the results of the few regions that use self-collected HPV samples [10].

Several hr-HPV tests for cervical- and self-sampling are available on the market. The hr-HPV mRNA tests have shown similar sensitivity but improved specificity compared to hr-HPV DNA tests taken from the cervix [11, 12]. For vaginal self-collected samples a meta-analysis by Arbyn et al. showed a lower sensitivity for the hr-HPV mRNA test compared to clinician-taken samples [13]. However, two other studies showed a substantial agreement between a self-collected vaginal sample and a clinician taken sample analyzed by Aptima mRNA assay [14, 15]. Asciutto et al. found a similar clinical sensitivity of self-collected samples analyzed by Aptima mRNA assay as compared to routine cytology. Recently an improvement of this self-collection approach showed a sensitivity of 95.3% for severe dysplasia [16, 17].

Previous studies have shown response rates for self-sampling among non-attendees to be between 15 and 58% [18‐24]. Reasons for declining a vaginal hr-HPV self-sample have been explored by a few studies by sending out questionnaires. Common reasons for not returning self-samples have been; opportunistic screening outside the screening program, preference for the regular screening procedure, insecurity about how to collect the specimen themselves, the belief that screening is unnecessary, pregnancy and previous hysterectomy [20, 25‐27]. To our knowledge, no previous study has used a telephone interview to explore reasons for not returning a self-sample.

The first aim of this study was to investigate the response rate of a free of charge self-collected vaginal hr-HPV sample sent to women who had not attended organized cervical cancer screening for ≥7 years. The second aim was to study the compliance with follow-up among women positive for hr-HPV in the self-collected vaginal sample. The third aim was to analyze the prevalence of severe cervical dysplasia (high grade squamous intraepithelial lesion (HSIL), atypical squamous cells that cannot exclude HSIL (ASC-H) or adenocarcinoma in situ (AIS)) or cancer among the responders. The fourth aim was to explore, by telephone interviews, the reasons for not returning a self-collected vaginal hr-HPV sample.

The response rate of the self-sample was 13.2% [(797/6023), 95% CI 12.4–14.1%] (Fig. 1). One returned self-sample could not be analyzed due to insufficient sample material, leaving 796 self-samples for hr-HPV mRNA analysis. The mean age of the women who submitted their self-samples was 61.2 years (range 33–71 years – the women 71 years of age turned 71 years during the follow-up time). Response rate and prevalence of hr-HPV mRNA stratified by age groups are shown in Table 1. Hr-HPV mRNA was detected among 9.9% [(79/796), 95% CI 7.9–12.2%] of the self-samples (mean age 60.8 years, range 37–70). Out of the women with detection of hr-HPV mRNA in the self-sample 83.5% [(66/79), 95% CI 73.5–90.9%] attended the midwife follow-up examination where 33.3% [(22/66), 95% CI 22.2–46.0%] presented with dysplasia at cytology, and 12.1% [(8/66), 95% CI 5.4–22.5%] with severe dysplasia (Fig. 2). The rate of histologically confirmed severe dysplasia or cancer in the whole group of responders was 1.3% [(10/796), 95% CI 0.6–2.3%], 0.4% [(3/796), 95% CI 0.1–1.1%] were diagnosed with cervical cancer (Fig. 2). The ten women with severe dysplasia or cancer had no registered cervical smear in the registers of the Region of Skåne for the last 16 years or more. Of the women with a positive hr-HPV self-sample, 16.5% [(13/79), 95% CI 9.1–26.5%] did not attend the follow-up examination despite receiving an invitation letter and thereafter a reminder letter. Among these, one woman was reached for a telephone reminder, the other 12 received a second reminder letter. On January 31, 2019, one of the 13 women had given a cervical smear with benign cytology and negative hr-HPV.

Long-term non-attendees of the organized cervical cancer screening program are at greater risk of cervical cancer [5, 29]. Our study found that 13% of 6023 women with no cervical smear for ≥7 years returned a self-collected vaginal sample for hr-HPV testing. The prevalence of severe dysplasia and cancer was 1.3% and the prevalence of solely cervical cancer was 0.4%, which was several times higher than expected.

The histologically confirmed prevalence of severe dysplasia and cancer of 1.3% in this study corresponded to the results of 0–3.1% found in other studies using self-sampling among women with no cervical sample for 3–9 years [18, 20‐23, 30, 31]. This was also in line with the prevalence of severe dysplasia or cancer of 1.4% found in the organized cervical cancer screening program in Sweden 2016 [32]. However, the prevalence of cervical cancer (0.4%) was found to be almost seven times higher in our study compared to the cervical cancer prevalence in the organized screening in Sweden 2016 (cervical cancer prevalence 2016 = 0.06%) [32]. Considering the number of self-samples received in our study, no cases of cervical cancer should have been found. Our results indicate that non-attendees of cervical cancer screening have an increased risk of cervical cancer development, a conclusion which corresponds to the findings of a systematic-review and meta-analysis by Spence et al. [29]. It is also in agreement with a Swedish study by Andrae et al. where 64% of all cervical cancer cases and 83% of advanced cases of cancer were found in women with no cervical smear taken during the recommended screening interval [5]. However, it is important to emphasize that the women diagnosed with cervical cancer in this study had no registered cervical smear in the registers in Region of Skåne in the last 20 years. This period is more than four times the length of the recommended screening interval. A similar trend with a long interval (≥16 years) since the last cervical smear could also be seen for the women with severe dysplasia. This was the first time these women had been offered a self-sampling test. Our study shows that self-samples can reach women who have not attended cervical cancer screening for a very long time. However, we need to reach the non-attending women earlier to prevent early dysplasia from developing into severe dysplasia or cancer. Thus, offering a self-sampling test after 7–9 years of non-attendance as recommended in the Swedish National guidelines could have potential for decreasing the incidence of cervical cancer among non-attendees.

The prevalence of hr-HPV mRNA in the genital tract was 10% in this study. This result was similar to that of Des Marais et al. who found an hr-HPV prevalence of 12.4% using self-collected samples analyzed by Aptima mRNA assay among women with no cervical sample in the past four years in North Carolina, USA [14]. Compared to women ≥30 years in the county of Skåne who had attended cervical cancer screening regularly, our hr-HPV prevalence was slightly higher (7% in regularly screened women vs. 10% in this study) [33]. It is also of interest that 29.5% (13/44) of women with benign cytology at the follow-up were positive for cervical hr-HPV mRNA, compared to 4% hr-HPV mRNA positivity among women aged 40–42 years with normal cytology in the regular screening program [33]. Apart from indicating that long-term non-attendees are a risk population, this could also indicate that the women participating in self-sampling may categorize themselves as individuals at high risk of HPV infection and therefore they chose to participate. However, our data cannot provide any further information on this matter.

Arbyn et al. recently published a meta-analysis investigating strategies to reach under-screened or not screened women by offering self-samples. The meta-analysis showed a response rate of between 6.4 to 34.0% (average 19.2%) when a self-sample was mailed to a woman’s home [13]. In our study, the response rate for self-collected vaginal hr-HPV samples was 13%. All self-samples returned until May 31, 2018 (approximately six months after the self-sampling offer) were included in our study. After this date, an additional of 217 women returned their self-collected sample, these samples were not included in the study. Szwarewski et al. found a response rate for self-samples returned within six months among persistent non-responders of 6.4% (10.2% including women who chose clinician cervical sampling instead) [34] and Stenvall et al. found a response rate for self-samples returned within five months among women with no cervical sample for ≥6 years of 32% [35]. Furthermore, Sancho-Garnier et al. found a response rate for self-samples of 18.3% of which 13.3% of the women had never taken a cervical smear according to the records [36]. These data show a large variation in response rate for self-collected samples among non-attendees. Previous studies have shown that reminders could increase the self-sampling participation rates among non-attendees [18, 22], however, no reminder was sent in this study. In 2017, the coverage of the cervical cancer screening program in Skåne was 81.5% [6], an addition of 13% among all the former non-attendees in the entire county of Skåne generates a total coverage level of nearly 84%. This is a small but important step in a desirable direction. The mean age of the women returning the self-collected sample was 61.6 years in this study. Comparing to a study by Darlin et al. conducted in the same geographical area as our study, Darlin et al. found a mean age of women returning a self-collected HPV sample of 51 years (range 32–64 years) [18]. Although in another more recent study conducted in the same geographical area we studied the response rate of a self-collected HPV sample sent to screening non-attending women 69–70 years old and found a high response rate of 43.3% [37]. This could indicate that older women in this area have a positive attitude towards self-sampling.

Compliance with follow-up is essential if self-sampling is to be used. In this study, 83.5% of women with a HPV positive self-collected sample attended the midwife follow-up examination. This is similar to a recent meta-analysis by Arbyn et al. where the average participation rate for follow-up was 80.6% [13]. Compared to other studies conducted in Sweden among women who had provided no cervical sample for ≥6–9 years the compliance with follow-up varied between 70 and 100% [18, 20, 21, 35, 38].

Our study is the first, as far as we know, using telephone interviews to explore reasons for not returning a vaginal hr-HPV self-sample among long-term non-attendees of the organized cervical cancer screening program. The participation rate of 12% in the telephone interviews was below our expectations, but the results should be interpreted with some caution because of the small sample size. Previous studies using questionnaires have reached participation rates of 3.4–38% [19, 20, 25]. The most common reasons for not taking part in the organized cervical cancer screening in Sweden are “uncomfortable with vaginal examination”, “feel healthy”, “lack of time” and “experience of unfriendly health workers” [18]. In our study, the six most common reasons for not returning a vaginal hr-HPV self-sample were “did not receive a self-sampling kit”, “forgot”, “phobia/fear of cancer”, “ignorance of cervical cancer screening”, “laziness” and “other diseases prioritized”. A common reason for non-attendance in both settings was “forgot”. It is also noteworthy that 22% of the women answered that they did not receive their self-sampling kit. An individual telephone dialog would provide the opportunity to ensure that all women have received the self-sampling kit, and would also serve as a reminder and give a chance to motivate the women based on their individual reasons. This strategy is also in line with the recommendations in the National Guidelines [39], but it is time-consuming for health personnel.

This study was performed in a setting of non-attendees of a current population-based cervical cancer screening program, which is a strength of the study. Furthermore, only one self-collected sample was invalid for analysis in the final compilation of results. Notably, during the analytical process we experienced problems with invalid samples (in one batch 25% (110/434) of the samples was invalid), but re-analysis of ¼ diluted samples (1 ml sample were transferred to 2.9 ml) rendered such samples as valid. However, for future studies of self-collected samples analyzed for HPV mRNA by the Aptima-system we will instead add a pre-heating step of samples, in order to reduce the proportion of invalid samples [17]. One limitation of the study is that women with a previous total hysterectomy, which is a criterion of exclusion from cervical cancer screening [39], were not excluded from the study. According to figures from the Swedish National Patient registry, 5–6% of Swedish women in the age group 40–60 years have had a hysterectomy with removal of the cervix [40]. This may have affected the number of self-samples submitted and shows that improvements are needed to prevent women who have no cervix from being incorrectly invited to attend cervical cancer screening. In this study, women in the Region of Skåne who had provided no cervical sample for ≥7 years were included. However, we do not have information about whether the women had provided a more recent cervical sample in a different region of Sweden or in another country. The participation rate for returning the self-sample is rather low in this study, and no reminder was sent out. A reminder may increase the participation rate, which is why we recommend a reminder letter or a phone call in future studies. The participation rate in the telephone interviews was also low, partly because of invalid telephone numbers which might have been a consequence of the switch from land-line to mobile phone systems currently taking place in Sweden.

In conclusion, offering self-collected vaginal samples for hr-HPV mRNA testing was a promising strategy to increase participation in the cervical cancer screening among long-term non-attendees. The compliance with follow-up was 83.5%, which is a good level, comparable to other studies. The prevalence of cervical cancer was almost seven times higher among long-term non-attendees than in the organized screening population. No particular reason for not returning self-samples was noted from the conducted telephone interviews.