Successful implementation of Papanicolaou (Pap) smear screening programs has drastically reduced invasive cervical cancer incidence and mortality in developed countries.1,2 However, Pap screening programs have been difficult to implement in low-resource settings because of limited infrastructure and access to trained cytopathologists and clinicians.3 Consequently, a region with low screening coverage such as Eastern Africa still has among the highest estimated annual incidence of invasive cervical cancer in the world (34/100,000).4
A primary screening approach based on testing for the central etiological risk factor for cervical cancer, human papillomavirus (HPV) infection, in self-collected specimens could help increase access to screening in low-resource settings. Human papillomavirus testing of self-collected specimens can be integrated into a 2-stage cervical cancer screening process.5 Women with positive HPV testing result could be rescreened using a second, more specific test (e.g., Pap test; colposcopy) or referred directly to treatment.6 Because primary HPV testing of self-collected specimens does not require an initial gynecologic examination, HPV testing of self-collected specimens as an initial screen could be an advantage in low-resource settings if the appropriate follow-up of women with positive HPV results can be assured. Furthermore, women’s acceptability of self-collection has generally been positive in various geographical settings worldwide.7
Very few studies have evaluated HPV testing of self-collected specimens for cervical cancer screening in low-resource settings, and to date, all have used HPV DNA testing.6,8,9 Recently developed diagnostic testing allows for the detection of high-risk (hr) HPV messenger RNA (mRNA), which may be a more specific marker than HPV DNA for clinically significant high-grade cervical disease,10 and has not yet been implemented in a high-risk, low-resource setting.
We present here results comparing the performance of APTIMA hrHPV mRNA (AHPV; Hologic/Gen-Probe Incorporated, San Diego, CA) testing of physician-collected (AHPV physician testing) and self-collected (AHPV self-testing) specimens for the detection of cytological high-grade cervical lesions in high-risk female sex workers (FSWs) in Kenya. We also examined risk factors for hrHPV mRNA positivity in our population of FSWs.
MATERIALS AND METHODS
From August 2009 to March 2011, FSWs attending the Korogocho clinic in Nairobi, Kenya, were invited to participate in this study to compare the performance of physician- and self-collected specimens for cervical cancer screening with hrHPV mRNA testing. The clinic provides counseling and medical care including screening and treatment of cervical cancer as well as sexually transmitted infections (STIs) for FSW in the Korogocho slum area.
Women were informed of the study by community peer leaders during “baraza” public meetings. Women were not eligible if they had undergone hysterectomy or were in the second trimester of pregnancy or later. A total of 350 FSWs aged 18 to 49 years provided written informed consent and were subsequently enrolled, from an approximately 425 FSWs who were invited to participate in the study.
At screening, participating women were administered a questionnaire to collect sociodemographic, reproductive, and sexual behavior data. Of the 350 FSWs recruited, 6 women were missing hrHPV mRNA testing results and were excluded from subsequent analyses, resulting in a final sample size of 344.
Sample Collection and Laboratory Analyses
Each woman self-collected a cervicovaginal specimen using the APTIMA Cervical Specimen Collection and Transport cytobrush (Hologic/Gen-Probe Incorporated) according to pictorial instructions. The cytobrush was then swirled in the APTIMA specimen transport medium and then discarded. During a pelvic examination, the physician collected 1 cervical sample from each woman using a Cervex-Brush (Rovers Medical Devices, Oss, the Netherlands), which was then swirled in the PreservCyt medium (Hologic Corporation, Marlborough, MA) and then discarded. The physician then collected a second cervical sample for conventional Pap test.
Cytological smears were evaluated at the University of Nairobi and classified according to the 2001 Bethesda System (TBS 2001) for cervical cytology. All smears were independently read by 2 cytopathologists (Dr M. Mungania, MBChB, MMed in Pathology, and Dr W. Waweru, MBChB, MMed in Pathology) blinded to HPV and STI testing results. For discrepant cases, the final diagnosis was made based on the consensus of the reviewing cytopathologists. Study participants were notified of their Pap test results 2 weeks after their screening visit. Women with low-grade squamous intraepithelial lesions (LSIL) or atypical squamous cells of undetermined significance (ASCUS) were instructed to undergo a repeat cytology 4 months later. Women with high-grade squamous intraepithelial lesions (HSIL) or ASCUS with possibility of high-grade changes were immediately referred to a colposcopy-directed biopsy. In the event of histological cervical intraepithelial neoplasia 2 or worse (≥CIN 2), women received standard care and treatment at Kenyatta National Hospital. Women who had <CIN 2 were considered disease negative for statistical analyses.
HPV and STI Testing
The physician- and self-collected specimens were transported to Hologic/Gen-Probe Incorporated in San Diego for HPV and STI testing. Laboratory testing for HPV in our study was by the AHPV (Hologic/Gen-Probe Incorporated), which qualitatively detects E6/E7 mRNA of 14 hrHPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68).11
The physician- and self-collected specimens were also tested for Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (GC) with the APTIMA Combo 2 assay, for Trichomonas vaginalis (TV) with the APTIMA TV assay, and for Mycoplasma genitalium (MG) with the APTIMA research use only assay, using the same target capture, transcription-mediated amplification, and hybridization steps as AHPV detection. All assays were performed according to the manufacturer’s instructions, without knowledge of the Pap test or other study results.
Serum was tested for HIV antibodies by enzyme-linked immunosorbent assay (ELISA), with positive results confirmed by a second ELISA. Peripheral blood CD4 cells were enumerated for HIV-seropositive women. The HIV ELISA and CD4 assays were conducted in University of Nairobi.
Agreement between AHPV positivity in physician- and self-collected specimens was measured by the κ statistic. Median unbiased estimates12 and their mid-P 95% confidence intervals (CIs) were computed for sensitivity, specificity, positive predictive value, and negative predictive value (NPV) of AHPV physician testing and self-testing for the detection of ≥HSIL. For small sample sizes, the median unbiased estimate potentially provides a more accurate estimate than that obtained by conventional maximum likelihood estimation.12 Data on AHPV and ≥HSIL prevalence as well as performance of AHPV testing to detect ≥HSIL were stratified into age 2 groups (<30 and ≥30 years). Currently, the US Preventive Services Task Force recommends HPV cotesting with Pap for women 30 years or older and does not recommend HPV testing for women younger than 30 years.13 We thus present results overall and stratified by age.
Potential risk factors for AHPV positivity in physician- and self-collected specimens were determined, and directed acyclic graphs14 were analyzed with online software15 to identify minimally sufficient adjustment variables to reduce confounding in binomial regression estimates of the prevalence difference of AHPV positivity between categories of potential risk factors. All statistical analyses were performed using SAS 9.2.
Overall prevalence of AHPV was similar in physician-collected (30%) and self-collected specimens (29%; Table 1). APTIMA high-risk HPV mRNA prevalence in both physician- and self-collected specimens was slightly higher in women younger than 30 years than in older women. Prevalence of any abnormal cytology (≥ASCUS) in the population was 19% and was similar in women 30 years or older (21%) than in younger women (17%).
Performance of AHPV Testing of Physician- and Self-Collected Specimens
The overall sensitivity of AHPV for ≥HSIL seemed similar to that of self-testing (Table 2). APTIMA high-risk HPV mRNA physician testing detected only 1 ≥HSIL case more than that of self-testing (n = 13 vs. n = 12; Table 3). This ≥HSIL case, which was negative by AHPV self-testing, was also disease negative by histology. Overall specificity for ≥HSIL seemed similar in both AHPV physician testing and self-testing (Table 2).
The NPV for AHPV physician testing and self-testing was 97% to 99% overall and in both age groups. The positive predictive value was 12% to 13% overall and varied by age group, being somewhat lower in women younger than 30 years (9%–10%) than in older women (17%–18%). The agreement between AHPV physician testing and self-testing was κ (κ statistic) = 0.59 (95% CI, 49–68) overall, κ = 0.76 (95% CI, 32–100) in women with ≥HSIL and κ = 0.55 (95% CI, 45–65) in women with <HSIL.
Of the 15 women with ≥HSIL, 14 underwent colposcopy-directed biopsy. One woman could not be traced and was lost to follow-up. Twelve had histological ≥CIN 2, and 2 were considered disease negative (both had normal histology). Of the 12 ≥CIN 2 cases, 10 were AHPV positive in both physician- and self-collected specimens. The remaining 2 ≥CIN 2 cases were AHPV negative in both physician- and self-collected specimens.
Risk Factors for AHPV Positivity in Physician- and Self-Collected Specimens With Normal Cervical Cytology
Adjusted prevalence differences (APDs) of AHPV positivity in both physician- and self-collected specimens were generally 5% or less (“Appendix 1”). Adjusted prevalence difference was greater than 5% in women who were younger (<30 years), had TV or MG, or had more education (>8 years). In physician-collected specimens, APD was also greater than 5% in women who reported less frequent condom use with sexual clients and regular partners, as well as women who charged more per transaction. In self-collected specimens, APD was also greater than 5% in HIV-seropositive women and in women with CT/GC infection.
Owing to small sample size, the 95% CIs for the APDs were generally wide. The 95% confidence limit differences (difference between upper and lower 95% confidence limits, CLD)16 for the APDs were generally between 0.19 and 0.25 (“Appendix 1”). The least precise APD estimates (difference between upper and lower 95% confidence limit >0.25) in both physician- and self-collected specimens were those relating HIV seropositivity and CT/GC, TV, and MG positivity to the number of regular sexual partners and to frequency of condom use with regular sexual partners.
APTIMA high-risk HPV mRNA physician and self-testing demonstrated a high sensitivity for the detection of ≥HSIL. APTIMA high-risk HPV mRNA self-testing in our population of FSWs with high AHPV prevalence seemed to have similar sensitivity and specificity for ≥HSIL as that of physician testing. APTIMA high-risk HPV mRNA positivity in both physician- and self-collected specimens was also somewhat higher in women who were younger than 30 years, had TV or MG infection, or had higher educational attainment.
In our study, overall sensitivity of AHPV physician testing for ≥HSIL seemed similar to that of self-testing. Our finding of similar sensitivity for ≥HSIL between AHPV physician testing and self-testing is in contrast to previous population-based studies,6,8,9 where overall sensitivity of Hybrid Capture 2 hrHPV DNA physician testing for ≥CIN 2 was higher than that of self-testing. However, the small number of women with ≥HSIL (n = 15) in our study made comparing sensitivity estimates of AHPV physician testing and self-testing somewhat difficult. Our finding of similar overall specificity of AHPV physician testing and self-testing is consistent with studies based on Hybrid Capture 2 hrHPV DNA testing.6,8,9
Prevalence of ≥HSIL in our study (4%) was similar to that in other African studies.6,17 Prevalence of ≥HSIL in women younger than 30 years was similar to that in women 30 years or older (4% vs. 5%) in this group of high-risk FSW in Kenya (Table 3), suggesting that AHPV self-testing could be a viable option for cervical cancer screening in high-risk FSW populations 18 to 29 years of age in Kenya. Although AHPV self-testing in our study missed 1 ≥HSIL case that was detected by physician testing, that particular ≥HSIL case was later confirmed to be negative by histology. Thus, in our study, AHPV self-testing detected as many ≥CIN 2 cases as did that of physician testing, suggesting that in low-resource areas where Pap screening is not routinely available, AHPV self-testing can be used to identify high-risk women at higher risk of ≥CIN 2.
The somewhat higher APDs (>0.05) in women younger than 30 years and in women with MG and TV infection were consistent with earlier findings, where younger age18–20 and infection with MG or TV 21–23 were risk factors for hrHPV DNA positivity. The reason for the difference in APDs of specific STIs in physician collection and self-collection (HIV and CT/GC seemed higher in self-collection, TV seemed higher in physician collection) is unclear, although it could also be an effect of the small number of women who were positive with either infection (n ≤ 15).
Our study has several advantages. First, all cervical smears were independently read by 2 cytopathologists, followed by consensus of reviewing cytopathologists for discrepant cases, ensuring accurate cytological diagnoses. Second, a woman’s physician collection and self-collection of specimens were performed on the same day, enabling direct comparison of AHPV testing results of these 2 sample types.
One limitation of our study was that histological results were obtained only for women who had cytological ≥HSIL. Because women with normal Pap reading were not systematically referred to colposcopy, histological ≥CIN 2 could not be used as reference standard for evaluating test sensitivity and specificity because of the potential for verification bias.24 Second, our FSW study was originally established to examine the role of CT and GC infections in the pathogenesis of pelvic inflammatory disease. We thus did not account for an a priori sample size to determine the number of women needed to detect a difference between AHPV physician testing and self-testing for ≥HSIL detection. Our small sample size also limited our ability to compare the agreement (as measured by the κ statistic) between AHPV physician testing and self-testing in women with <HSIL compared with those with ≥HSIL. Third, given that the APTIMA assay does not differentiate between hrHPV types, genotype-specific results were not presented. Furthermore, the APTIMA was the only HPV detection assay used, and thus, we could not directly compare our results of mRNA testing to those of a DNA-based assay. Lastly, self-collection was always conducted before physician collection, and thus, we were not able to determine if the order of self-collection (before or after physician collection) may have potentially affected our results. We do not, however, expect that specimen collection order would have made a meaningful difference given that self-sampling involves collection of cervicovaginal cells, whereas physician sampling involves collection of cervical cells near the cervical os.
In terms of public health implications, our results of AHPV self-testing versus physician testing performance were from a high-risk population in a low-resource setting and thus not necessarily generalizable to low-risk populations. Also, one concern of self-collection in primary screening is that a woman who tested HPV positive may not return for follow-up screening or treatment.6 Other commonly reported issues include difficulty of use and potential contamination of the self-collection brush.7 Despite potential limitations, data from a meta-analysis25 supported increased use of self-collection in epidemiological studies. Future research should address if self-collection can improve screening coverage in low-resource settings.
High-risk HPV self-testing does not seem sufficiently specific to be a stand-alone test for primary cervical cancer screening.8 The small sample size and absence of complete histological diagnosis in our study limited our ability to make definitive conclusions about the comparative performance of hrHPV mRNA physician testing versus self-testing for ≥CIN 2 detection. Our results, however, are generally consistent with earlier studies,6,8,9 which showed that hrHPV self-testing have high NPV and have the potential to effectively identify women at higher risk for high-grade lesions without an initial gynecologic examination. Limited resources may then be channeled into clinical follow-up (e.g., rescreening using a different test) of a woman who was positive by AHPV self-testing, based on specific local capacity.
1. Gustafsson L, Pontén J, Zack M, et al. International incidence rates of invasive cervical cancer after introduction of cytological screening. Cancer Causes Control 1997; 8: 755–763.
2. Nygard JF, Skare GB, Thoresen SO. The cervical cancer screening programme in Norway, 1992–2000: Changes in Pap smear coverage and incidence of cervical cancer. J Med Screen 2002; 9: 86–91.
3. Sankaranarayanan R, Budukh AM, Rajkumar R. Effective screening programmes for cervical cancer in low-and middle-income developing countries. Bull World Health Organ 2001; 79: 954–962.
4. Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. GLOBOCAN 2008 v1.2, Cancer incidence and mortality worldwide: IARC Cancer Base no. 10 [Internet]. Lyon, France: International Agency for Research on Cancer; 2010. Available at: http://globocan.iarc.fr
. Accessed November 17, 2011.
5. Denny L, Kuhn L, Risi L, et al. Two-stage cervical cancer screening: An alternative for resource-poor settings. Obstet Gynecol 2000; 183: 383–388.
6. Wright TC, Denny L, Kuhn L, et al. HPV DNA testing of self-collected vaginal samples compared with cytologic screening to detect cervical cancer. JAMA 2000; 283: 81–86.
7. Stewart DE, Gagliardi A, Johnston M, et al. Self-collected samples for testing of oncogenic human papillomavirus: A systematic review. J Obstet Gynaecol Can 2007; 29: 817–828.
8. Salmeron J, Lazcano-Ponce E, Lorincz A, et al. Comparison of HPV-based assays with Papanicolaou smears for cervical cancer screening in Morelos State, Mexico. Cancer Causes Control 2003; 14: 505–512.
9. Zhao FH, Lewkowitz AK, Chen F, et al. Pooled analysis of a self-sampling HPV DNA test as a cervical cancer primary screening method. J Natl Cancer Inst 2012; 104: 178–188.
10. Monsonego J, Hudgens MG, Zerat L, et al. Evaluation of oncogenic human papillomavirus RNA and DNA tests with liquid based cytology in primary cervical cancer screening (the FASE study). Int J Cancer 2011; 129: 691–701.
11. Hologic Gen-probe APTIMA® HPV assay [package insert]. San Diego, CA: Hologic Gen-Probe, 2008.
12. Rothman KJ, Greenland S, Lash T. Median unbiased estimates. In: Rothman KJ, Greenland S, Lash TL, eds. Modern Epidemiology, 3rd ed. Philadelphia, PA: Lippincott Williams & Wilkins, 2008: 221, 255–256.
13. Moyer VA. on behalf of the U.S Preventive Services Task Force. Screening for cervical cancer: US preventive services task force recommendation statement. Ann Intern Med 2012; 156: 880–890.
14. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology 1999; 10: 37–48.
15. Knüppel S, Stang A. DAG program: Identifying minimal sufficient adjustment sets. Epidemiology 2010; 21: 159.
16. Poole C. Low P
-values or narrow confidence intervals: Which are more durable? Epidemiology 2001; 12: 291–294.
17. Luchters S, Broeck D, Chersich M, et al. Association of HIV infection with distribution and viral load of HPV types in Kenya: A survey with 820 female sex workers. BMC Infect Dis 2010; 10: 18.
18. Del Amo J, González C, Belda J, et al. Prevalence and risk factors of high-risk human papillomavirus in female sex workers in Spain: Differences by geographical origin. J Women Health 2009; 18: 2057–2064.
19. Juarez-Figueroa LA, Wheeler CM, Uribe-Salas FJ, et al. Human papillomavirus: A highly prevalent sexually transmitted disease agent among female sex workers from Mexico City. Sex Transm Dis 2001; 28: 125–130.
20. Kjaer SK, Svare EI, Worm AM, et al. Human papillomavirus infection in Danish female sex workers: Decreasing prevalence with age despite continuously high sexual activity. Sex Transm Dis 2000; 27: 438–445.
21. Depuydt CE, Leuridan E, Van Damme P, et al. Epidemiology of Trichomonas vaginalis
and human papillomavirus infection detected by real-time PCR in Flanders. Gynecol Obstet Invest 2010; 70: 273–280.
22. Noël JC, Fayt I, Romero Munoz MR, et al. High prevalence of high-risk human papillomavirus infection among women with Trichomonas vaginalis
infection on monolayer cytology. Arch Gynecol Obstet 2010; 282: 503–505.
23. Biernat-Sudolska M, Szostek S, Rojek-Zakrzewska D, et al. Concomitant infections with human papillomavirus and various mycoplasma and ureaplasma species in women with abnormal cervical cytology. Adv Med Sci 2011; 56: 299–303.
24. Begg CB, Greenes RA. Assessment of diagnostic tests when disease verification is subject to selection bias. Biometrics 1983; 39: 207–215.
25. Petignat P, Faltin DL, Bruchim I, et al. Are self-collected samples comparable to physician-collected cervical specimens for human papillomavirus DNA testing? A systematic review and meta-analysis. Gynecol Oncol 2007; 105: 530–535.
APPENDIX 1. Association of Potential Risk Factors With AHPV Positivity Among 279 FSWs With Normal Cytology in Kenya, 2009–2011
No caption available...Image Tools