A comparative study of self-collected versus clinician-collected specimens in detecting high-risk HPV infection: a prospective cross-sectional study

Article information

Obstet Gynecol Sci. 2024;67(6):557-564

Publication date (electronic) : 2024 October 8

doi : https://doi.org/10.5468/ogs.24117

Natnipa Parapob, MD¹, Suree Lekawanvijit, MD², Theera Tongsong, MD¹, Kittipat Charoenkwan, MD¹, Charuwan Tantipalakorn, MD¹

¹Department of Obstetrics and Gynecology, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand

²Department of Pathology, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand

Corresponding author: Charuwan Tantipalakorn, MD, Department of Obstetrics and Gynecology, Faculty of Medicine, Chiang Mai University, Chiang Mai 50200, Thailand, E-mail: charuwan.t@cmu.ac.th

Received 2024 March 18; Revised 2024 August 9; Accepted 2024 September 29.

Abstract

Introduction

In 2020, based on the global cancer statistics of the World Health Organization (WHO), approximately 604,000 new cases of cervical cancer were diagnosed, and 342,000 deaths were estimated worldwide. Eighty-four percent of the women with cervical cancer were from low-resource countries [1]. In Thailand, cervical cancer is the third leading carcinoma, ranking after breast and colorectal cancers. In 2022, there were 9,157 new cases of cervical cancer among Thai women. Of them, 4,705 Thai women died of cervical cancer, with an estimated 12 deaths per day [2]. Human papillomavirus (HPV) infection is a major cause of cervical neoplasia and has been detected in 99.7 percent of cervical carcinomas [3]. Cervical carcinoma is now recognized as the most preventable cancer, as there are various prevention methods, such as primary prevention with HPV vaccination and secondary prevention with effective screening tests [4–6]. The risk factors associated with HPV-related cervical cancer include early onset of sexual activity, multiple sex partners, history of sexually transmitted diseases, history of vulvar or vaginal squamous intraepithelial neoplasia or cancer, smoking status, and human immunodeficiency virus (HIV) status [7].

Currently, HPV detection plays an important role in gynecological practice. It is applied in combination with cytological examination (co-testing) for primary screening of cervical cancer, triage of abnormal cytology or Ascus cytology, and for follow-up after treatment of CIN2 and CIN3 lesions, the so-called “test of cure”.

Women with abnormal cytology results or HPV detection by cervical cancer screening are typically triaged to undergo colposcopy to exclude invasive cervical cancer, detect and treat high-grade lesions before cancer progression, and guide further proper care. Women with high-risk HPV infections, including HPV16, HPV18, and other 12 high-risk HPV genotypes, are at higher risk of developing cervical cancer. For women aged 25 years or older who have positive HPV testing and abnormal cytology, the American Society for Colposcopy and Cervical Pathology recommends direct referral for colposcopy and a biopsy should be done if the high-grade lesion is suspicious [8]. Cervical cancer screening technology has been further developed to allow women to collect specimens themselves (self-collected vaginal sampling). This is useful for HPV detection in women who do not undergo pelvic examinations and have limited access to medical facilities. However, according to current guidelines, self-HPV collection is not considered a standard screening tool. Recent studies have shown that cervical cancer screening by self-collection is acceptable, and its effectiveness may be comparable to that of cell collection by doctors. This option may be attractive to women who dislike pelvic examinations, leading to an increase in the number of women undergoing cervical cancer screening [9–13]. Accordingly, this option may increase patients’ compliance to continuous follow-up because it does not require hospital visits or pelvic examinations and can reduce the incidence of cervical cancer in the future.

Surprisingly, several original studies and systematic reviews on the acceptability of self-sampling techniques have been published; however, studies on the effectiveness of self-sampling in comparison with clinician sampling are very limited. Whether the self-sampling technique can replace the clinician-sampled technique remains unclear. Accordingly, studies on the efficacy of self-sampling in comparison with conventional standard techniques are required.

Additionally, although the high acceptability of self-sampling has been reported in most studies, this technique may theoretically be associated with educational level, cultural lifestyle, and socioeconomic status, possibly leading to different levels of acceptability among various population groups. Therefore, each population should assess its acceptance within its own population. Accordingly, this study was conducted primarily to compare the detection rate of high-risk HPV infection between the self-sampling and clinician-sampling techniques, as well as the correlation between the two techniques. The secondary objective was to assess satisfaction with self-sampling for HPV testing among Thai women.

Materials and methods

This prospective cross-sectional study was conducted at the Colposcopy Clinic of Maharaj Nakorn Chiang Mai Hospital. The study was approved by the Institutional Review Board (Research Ethics Committee 4, Faculty of Medicine, Study code: OBG-2564-08107, Date of approval: August 11, 2021). Women with positive test results for HPV16, HPV18, or other 12 high-risk HPV genotypes, and abnormal cytology results of Ascus or higher between October 2021 and September 2022 were prospectively enrolled. Written informed consent was obtained from all the participants.

The inclusion criteria were women aged 25–65 years with the ability to collect vaginal specimens by themselves, and the ability to communicate in Thai. The exclusion criteria were women who were pregnant and those who had previous gynecologic cancer, previous hysterectomy, genital tract infection, bleeding disorders, or severe medical conditions.

Demographic and clinical data were collected, including age, menopausal status, contraception, age at first intercourse, smoking status, HIV status, history of sexually transmitted disease, level of education, duration since the last HPV test, cervical cytology results, HPV testing results, and final histology. After receiving instructions about the methods of vaginal self-sampling (following the video demonstration at: https://www.youtube.com/watch?v=lv8GgOsRCwE) from a well-trained research assistant, participating women collected the vaginal self-sampling specimens by themselves before colposcopy setup on the same day. The vaginal collection equipment used in this study was the Aptima® Multitest Swab Specimen Collection Kit (Hologic, Inc., San Diego, CA, USA), which was approved by the Thai Food and Drug Administration. This kit had sterile swabs and transport tubes containing 2.9 mL of specimen transport medium. All colposcopic procedures were performed by gynecologic oncologists or gynecologic oncology fellows using a standard conventional technique. Cervical specimens were collected by attending colposcopists before performing the colposcopic procedures. Cervical biopsies were obtained from the abnormal acetowhite or suspected areas. All pathological specimens were examined by a gynecologic pathologist. HPVDNA testing was performed in this study for the HPV 16/18 genotypes and other 12 high-risk HPV genotypes: 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68. Satisfaction with vaginal self-collection for high-risk HPV testing was queried and rated using a standard questionnaire with 5-point Likert scales for satisfaction after colposcopic procedures.

The primary endpoints of the study were the detection rate of high-risk HPV genotypes based on the two techniques and the level of agreement or correlation in detecting HPV infection between the two techniques. The secondary endpoint was satisfaction with the women using the vaginal self-collection method. Satisfaction was systematically assessed using the standard Likert 5-point scales, including very dissatisfied, dissatisfied, neither dissatisfied, satisfied, satisfied, and very satisfied.

Results

During the study period, 104 women were recruited as shown in Fig. 1. Six participants were excluded because of a previous hysterectomy or gynecologic cancer. The remaining 98 women were included in the final analysis. All women were successfully sampled using both techniques, providing 98 paired specimens for comparison. Only one self-sampled specimen was invalid for evaluation because the HPV betaglobin was not detected. Demographic variables are presented in Table 1.

Fig. 1

Flow diagram of participating women.

Table 1

Demographic variables of the participants (n=98)

The median age of participants was 45 years. Sixty-one patients (62.2%) had their first sexual intercourse at less than 20 years of age. Most participants were nonsmokers (94.9%), had negative HIV status (94.9%), and had undergone vaginal delivery (62.2%). Eight patients (8.2%) had a history of sexually transmitted diseases. Oral contraceptive pills were the most common contraception method, accounting for 33.7%, followed by tubal sterilization (18.4%), and depot-medroxyprogesterone acetate (13.3%). The median time from the last HPV testing of participants in this study was 3 months (Table 1).

The detection rate of high-risk HPV infection was comparable between the two techniques, 79/98 (80.6%) vs. 81/98 (82.7%) for the self-sampling and clinician-sampling techniques, respectively (McNemar’s test; P=0.774), as presented in Table 2. The correlations of HPV testing between self-collected vaginal specimens and clinician-collected cervical specimens are presented in Table 3. The level of agreement of both techniques in detecting HPV infection was “substantial” with a kappa coefficient of 0.75.

Table 2

Comparison of the detection rate of high-risk HPV infection between self-sampling and clinician-sampling group (McNemar test; P-value=0.774)

Table 3

Kappa’s coefficient for agreement between self-sampling and clinician-sampling

Analysis of satisfaction based on the 5-point scale is presented in Table 4. Most participants rated the method of vaginal self-collection as satisfied to very satisfied because of its convenience (98.0%) and safety (93.9%). The instructions for self-collection provided by the research assistant and the diagrams were considered very good to excellent (96.9%). Overall, satisfaction with the vaginal self-sampling method was excellent in 84 participants (85.7%). Fifty participants (51.0%) preferred follow-up investigation after preinvasive cervical cancer treatment with vaginal self-sampling, and the remaining preferred clinician-sampling.

Table 4

Participants’ satisfaction based on 5-point scales

From the histological results shown in Table 5, the detection of high-grade intraepithelial lesions (HSIL) and cervical cancer from positive self-sampling and clinician collection were similar, and no patients with cervical cancer were missed in negative self-sampling. The detection rate of HSIL or cancer by a positive high-risk HPV test based on the self-sampling technique and that based on the clinician-sampling technique was not significantly different (15/98 and 16/98, respectively; Chi-square test; P-value=0.845).

Table 5

Correlation between HPV typing results and pathological diagnosis

Discussion

Insights gained from this study are as follows. 1) The HPV detection rate of self-sampling is comparable with that of clinician- sampling, and also the two approaches are substantially correlated, in agreement with a Kappa index of 0.75 (95% CI, 0.64–0.85). Note that in contrast to most previous studies on the self-sampling technique, this study focused primarily on the effectiveness of the detection of high-risk HPV infection, rather than on the acceptability rate. And 2) most participants were satisfied or very satisfied with self-sampling, indirectly indicating a high acceptability rate among our population.

Most importantly, both techniques detected HPV at comparable rates. Although it was not known whether all 98 cases were still positive for HPV at the time of sampling, since some cases with previous positivity might have become negative due to spontaneous regression, all cases with positive results, based on either technique in this study, should be interpreted as persistently positive HPV infection. The detection rate of 79 out of 98 in the self-sampling group was comparable to that of 81 out of 98 in the clinician-sampled group (P>0.05). A kappa index of 0.75 indicates that the two techniques were substantially in agreement in the detection of HPV infection. However, this index does not indicate a superior sampling technique. The agreement was substantial but not perfect. In a small number of cases, the two techniques provided different results for HPV detection. In other words, some cases were detected using one technique and not using the other technique, and vice versa. The observed discrepancies in detection may be attributed to sampling errors that could have occurred in both techniques. However, the overall effectiveness of both techniques for detection was comparable (McNemar’s test; P-value >0.05). Therefore, the findings of this study support the idea that the self-sampling technique can potentially replace or serve as a viable alternative to conventional clinician sampling techniques in terms of effectiveness. Discordant results may stem from false negatives in one approach, as a single positive test should be considered indicative of the true presence of viral DNA. The underlying reasons for this discordance are not readily explained by demographic factors or serotypes as they were evenly distributed between the two approaches. Thus, viral detection failures can occur when using either technique. Missing detections appeared to occur randomly and at equal frequencies in both methods. A minimal viral load may contribute to this issue, as it is susceptible to being overlooked during detection, and the most sensitive sampling site cannot always be precisely identified. Consequently, even using identical methods, consistent detection may not be achieved, and potentially leading to inconsistent results. Therefore, positive cases may yield variable outcomes (either negative or positive), with either test contributing to the observed discordance.

The secondary outcome of this study is also very important, as satisfaction with these methods is expected to be associated with patients’ compliance with screening and follow-up tests. In addition, end-user satisfaction with self-sampling is critical for reaching the WHO global target of 70% HPV screening coverage by 2030 [9–11,13,15,16]. Although many studies have reported a high acceptability rate of self-sampling, either for HPV testing or Pap smears, evaluation of the acceptability rate among various populations is essential, and since cultures or attitudes toward pelvic examination or self-sampling of vaginal specimens are very different among various ethnicities or religions. This study supports the notion that self-sampling is well-accepted among Thai women. Together with its comparable effectiveness to the conventional approach, self-sampling of HPV tests outside a clinical setting has the potential to expand women’s choices of cervical cancer screening and facilitate earlier detection. The HPV self-sampling approach may be the only way to improve access to and uptake of screening, especially in regions of the world with the highest burden of disease, and is now considered a key pillar for reaching the WHO elimination target. Despite their high satisfaction with self-sampling, a significant percentage of women still preferred expert sampling in their next test. This may be due to their belief that care providers are likely to perform more effective sampling than they can themselves as well as their traditional beliefs or personal attitudes that favor conventional methods.

Interestingly, the agreement of self-sampling and clinician-sampling for HPV test in this study was better than the agreement of self-sampling and clinician sampling for Pap smear; Cohen’s κ statistic of 0.75 (95% CI, 0.64–0.85) in this study vs. 0.54 (95% CI, 0.27–0.82) in a previous study by Singla and Komesaroff [12]. These findings indirectly suggest that self-sampling for HPV testing has a higher success rate than self-sampling for Pap smears.

However, there were a few limitations. Our sample size might be too small for comparison of the prediction rate of precancerous or cancerous lesions because the prevalence of HSIL or cervical cancer in this small group was low. Note that the prediction rates of HSIL/cancer between self-sampling and clinician-based techniques were also comparable. Nevertheless, the significant difference in such a prediction rate, if it existed, and might not have been demonstrated because of the small sample size or the low power of the test. Additionally, this study was confined to a group of women with previous HPV-positive results to ensure an adequate number of positive cases, which may limit its applicability to a broader population. Nevertheless, the effectiveness of the sampling appears to be more directly associated with patients’ understanding and the informative instructions provided by caregivers than with ethnicity or high-risk group status.

The strength of this study is that the results of the comparison between the two groups were highly reliable because both techniques were provided to the same participant, resulting in perfect pair matching. In addition, the sample size used in this study was adequate for the primary objective. Notably, the detection rates of both approaches were very high in this study because the study population included only women who were previously positive for high-risk HPV. Therefore, the sample size was large enough to address the primary objective of this study.

Article information Continued