Abstract
Implementation of primary human papillomavirus (HPV) testing has been slow in the United States perhaps because of concerns of decreased sensitivity compared with concurrent HPV and cytology testing (“cotesting”). We used the National Breast and Cervical Cancer Early Detection Program and the Kaiser Permanente of Northern California cohort to quantify potential trade-offs with primary HPV compared with cotesting in 4 US populations with differing precancer or cancer prevalence. In all settings, cotesting required more lab tests and more colposcopies compared with primary HPV testing. Additional cervical intraepithelial neoplasia grade 3 or cancer immediately detected from cotesting vs primary HPV decreased with decreasing population-average cervical intraepithelial neoplasia grade 3 or cancer prevalence from 71 per 100 000 screened among never or rarely screened individuals in the National Breast and Cervical Cancer Early Detection Program (prevalence = 1212 per 100 000) to 4 per 100 000 screened among individuals with prior HPV-negative results in Kaiser Permanente of Northern California (prevalence = 86 per 100 000). These data suggest that cotesting confer an unfavorable benefit-to-harm ratio over primary HPV testing.
Although many countries have introduced primary human papillomavirus (HPV)–based screening programs, the transition from concurrent HPV and cytology testing to primary HPV screening has been slow in the United States (1), perhaps because in part to concerns about the slight decrease in sensitivity for detecting precancer inherent with primary HPV testing compared with cotesting (2). To inform different settings about implications of the transition, we quantified clinical parameters of primary HPV testing vs cotesting in 4 US populations with differing precancer prevalence from the Center for Disease Control and Prevention’s National Breast and Cervical Cancer Early Detection Program (NBCCEDP) and the Kaiser Permanente of Northern California (KPNC) cohorts.
The NBCCEDP was established to provide low-income underinsured or uninsured individuals access to screening and diagnostic exams; the cohort consists of individuals, ages 30-64 years, who either had cotesting or were triaged using HPV testing for atypical squamous cells (ASC) of undetermined significance (ASC-US) cytology result from 2009 to 2017 (3). The NBCCEDP did not collect data to differentiate between the 2 screening approaches. The KPNC cohort consists of individuals in an integrated health-care setting, ages 25-65 years, who underwent triennial cotesting at KPNC between 2003 and 2017 (4). The NBCCEDP study was approved by the Center for Disease Control and Prevention’s institutional review board, and the KPNC study has been approved by KPNC and National Cancer Institute institutional review board committees.
Under primary HPV testing, individuals are first tested for oncogenic HPV, and if HPV positive, were triaged with cytology. In contrast, individuals receiving cotesting are concurrently tested for HPV and cytology. For these strategies, US management guidelines recommend immediate colposcopy for 1) HPV-positive individuals with a cytologic result of ASC-US or worse; 2) HPV-negative individuals with severely abnormal cytologic results, defined as ASCs that cannot exclude high-grade squamous intraepithelial lesion (ASC-H), atypical glandular cells (AGC), high-grade squamous intraepithelial lesion (HSIL) and squamous cell carcinoma (SCC) (4). Persons with other combinations of test results were recommended to have 1-year surveillance or routine screening (Supplementary Table 1, available online). HPV-negative individuals with severely abnormal cytologic results are not identified under primary HPV testing; however, this combination is uncommon (5).
We assessed clinical parameters of primary HPV testing vs cotesting in 4 settings, listed in order of decreasing population-average precancer prevalence: 1) 113 126 underinsured or uninsured individuals who never received cervical screening or were screened more than 5 years prior, referred to in this analysis as never or rarely screened; 2) 208 658 underinsured or uninsured individuals screened within 5 years; 3) 1 546 462 individuals in an integrated health-care setting with no prior history of receiving an HPV-based test; and 4) 819 533 individuals in an integrated health-care setting whose prior HPV test was negative and are returning for routine screening. For each setting and screening modality, we evaluated the following clinical parameters per 100 000 individuals: the number of screening tests performed, colposcopies conducted and the cases of cervical intraepithelial neoplasia grade 2 (CIN2) or worse (CIN2+), CIN grade 3 (CIN3) or worse (CIN3+), and cancers immediately detected vs those that are not immediately detected despite being present at the time of screening.
Because adherence to recommended management differed between the settings and across test results, we assessed these clinical parameters using published estimates (3,4) of CIN2, CIN3, and cancer prevalence that account for differential ascertainment bias, as opposed to using the raw counts of CIN2, CIN3, and cancers detected. The CIN2, CIN3, and cancer prevalences were estimated using prevalence-incidence mixture models, which jointly fit a logistic regression model for latent prevalent precancer or cancer (allowing that some precancer or cancer detected in follow-up could have been detectable if the individual had received colposcopy at the initial visit) and a proportional hazards model for left-, interval-, and right-censored incident precancer or cancer (6,7). In cervical screening data, the time precancer or cancer first becomes detectable by colposcopy is only partially known with left, interval, and right censoring referring to where that time occurs on the timeline relative to a known timepoint: left censoring is when the time occurs before detection at the first administered colposcopy; interval censoring is when the time occurs between the last negative screening or colposcopy result and detection; right censoring (for noncases) is when the time occurs after an individual’s last negative screening/colposcopy result (Supplementary Figure 1, available online).
All clinical parameters were estimated using R statistical software, version 4.1.2 (8). The variances for each clinical parameter estimate were derived using the Delta method and accounted for the standard errors around prevalence estimates (Supplementary Methods, available online).
The average prevalence of CIN3 or cancer ranged from 1212 (95% confidence interval [CI] = 1134 to 1290) per 100 000 among underinsured or uninsured never or rarely screened individuals to 86 (95% CI = 78 to 93) per 100 000 among individuals with a prior HPV-negative result. The number of additional CIN3 or cancer detected immediately from cotesting vs primary HPV decreased with decreasing population-average CIN3 or cancer prevalence from 71 (95% CI = 57 to 85) per 100 000 screened among never or rarely screened individuals to 4 (95% CI = 3 to 6) per 100 000 screened among individuals with prior HPV-negative results (Figure 1, A). Cotesting required more lab tests (86 000-96 000 more per 100 000 screened) and colposcopies (200-500 more per 100 000 screened) compared with primary HPV screening (Table 1). Further, cotesting was less efficient in lower risk settings, with the number of additional colposcopies needed to find 1 additional CIN3 or cancer increasing from 8 (95% CI = 7 to 9) among never or rarely screened individuals to 59 (95% CI = 27 to 92) among individuals with prior HPV-negative results (Figure 1, B).
Figure 1.
Trade-offs of cotesting vs primary human papillomavirus (HPV) testing in 4 populations with differing cervical intraepithelial neoplasia grade 3 or cancer (CIN3+) prevalence. Left panel) Additional CIN3+ detected immediately per 100 000 individuals screened when using cotesting as opposed to primary HPV. Right panel) The number of additional colposcopies needed to find 1 additional CIN3+ when using cotesting as opposed to primary HPV. Settings: 1) Center for Disease Control and Prevention’s National Breast and Cervical Cancer Early Detection Program (NBCEEDP) never or rarely screened—underinsured or uninsured individuals in the NBCEEDP cohort who never received cervical screening or were screened more than 5 years ago; 2) NBCEEDP screened no more than 5 years—underinsured or uninsured individuals in the NBCEEDP cohort who have been screened within the past 5 years; 3) Kaiser Permanente of Northern California (KPNC) no prior HPV—individuals in the KPNC cohort with no prior history of receiving an HPV-based test; 4) KPNC prior HPV negative—individuals in the KPNC cohort returning for routine screening whose prior HPV result was negative.
Table 1.
Estimated trade-offs of cotesting vs primary HPV testing in 4 populations
| Clinical outcome | CDC NBCCEDP cohorta |
KPNC cohortb |
||
|---|---|---|---|---|
| Never or rarely screened | Screened within past 5 years | No prior HPV result | Prior HPV-negative result | |
| Estimate (95% CI) | Estimate (95% CI) | Estimate (95% CI) | Estimate (95% CI) | |
| Total CIN3 or cancer per 100 000 screened | 1212 (1134 to 1290) | 772 (727 to 816) | 455 (443 to 467) | 86 (78 to 93) |
| Cotesting | ||||
| No. of tests per 100 000 screened | 200 000 | 200 000 | 200 000 | 200 000 |
| No. of colposcopy referrals per 100 000 screened | 7742 | 8669 | 4277 | 2092 |
| CIN3 or cancer detected immediately per 100 000 screened | 1001 (950 to 1052) | 637 (605 to 670) | 364 (355 to 374) | 69 (63 to 75) |
| CIN3 or cancer not detected immediately per 100 000 screened | 211 (152 to 270) | 134 (104 to 164) | 90 (83 to 98) | 17 (12 to 22) |
| No. tests needed to find 1 CIN3 or cancer | 200 (190 to 210) | 314 (298 to 330) | 549 (535 to 563) | 2912 (2659 to 3165) |
| No. of colposcopies needed to find 1 CIN3 or cancer | 7.7 (7.3 to 8.1) | 13.6 (12.9 to 14.3) | 11.7 (11.4 to 12.0) | 30 (28 to 33) |
| Primary HPV | ||||
| No. of tests per 100 000 screened | 114 230 | 113 589 | 108 175 | 103 851 |
| No. of colposcopy referrals per 100 000 screened | 7184 | 8167 | 4067 | 1831 |
| CIN3 or cancer detected immediately per 100 000 screened | 930 (882 to 979) | 608 (576 to 640) | 358 (349 to 367) | 64 (59 to 70) |
| CIN3 or cancer not detected immediately per 100 000 screened | 282 (221 to 343)c | 164 (133 to 195)d | 96 (89 to 104)e | 21 (16 to 26)f |
| No. tests needed to find 1 CIN3 or cancer immediately | 121 (115 to 128) | 187 (177 to 197) | 302 (294 to 310) | 1616 (1471 to 1761) |
| No. of colposcopies needed to find 1 CIN3 or cancer immediately | 7.7 (7.3 to 8.1) | 13.4 (12.7 to 14.1) | 11.4 (11.1 to 11.6) | 28 (26 to 31) |
| Cotesting vs primary HPV | ||||
| No. of additional cotests needed under cotesting vs primary HPV per 100 000 screened | 85 770 | 86 411 | 91 825 | 96 149 |
| No. of additional colposcopy referrals under cotesting per 100 000 screened | 558 | 502 | 210 | 261 |
| Additional CIN3 or cancer not detected immediately under primary HPV per 100 000 screened | 71 (57 to 85) | 30 (22 to 37) | 6.3 (5.0 to 7.6) | 4.4 (2.9 to 5.9) |
| No. of additional tests needed to find 1 additional CIN3 or cancer detected immediately under cotesting | 1229 (1101 to 1358) | 2922 (2399 to 3445) | 14 560 (9430 to 19 690) | 21 842 (9820 to 33 865) |
| No. of additional colposcopies needed to find 1 additional CIN3 or cancer detected immediately under cotesting | 7.9 (7.1 to 8.7) | 17 (14 to 20) | 33 (22 to 45) | 59 (27 to 92) |
The Center for Disease Control’s National Breast and Cervical Cancer Early Detection Program cohort includes individuals, aged 30-64 years, receiving cotesting or HPV triage for atypical squamous cells of undetermined significance cytology results.
The Kaiser Permanente Northern California cohort includes individuals, aged 25-65 years, receiving HPV and cytology cotesting.
CIN3 or cancer not detected immediately per 100 000 screened attributed to HPV negative with ASC-H cytology result: 7.2 (2.5%); HPV negative with AGC cytology result: 26 (9.1%); HPV negative with HSIL or squamous cell carcinoma cytology result: 38 (13.2%).
CIN3 or cancer not detected immediately per 100 000 screened attributed to HPV negative with ASC-H cytology result: 6.9 (4.1%); HPV negative with AGC cytology result: 9.6 (5.8%); HPV negative with HSIL or squamous cell carcinoma cytology result: 13 (7.9%).
CIN3 or cancer not detected immediately per 100 000 screened attributed to HPV negative with ASC-H cytology result: 1.7 (1.8%); HPV negative with AGC cytology result: 1.6 (1.7%); HPV negative with HSIL or squamous cell carcinoma cytology result: 3.0 (3.1%).
CIN3 or cancer not detected immediately per 100 000 screened attributed to HPV negative with ASC-H cytology result: 1.9 (9.3%); HPV negative with AGC cytology result: 1.4 (7.1%); HPV negative with HSIL or squamous cell carcinoma cytology result: 1.1 (5.4%).
Similar findings of cotesting being increasingly inefficient with lower risk settings occurred when using either CIN2+ or cancer as the primary endpoint (Supplementary Tables 2 and 3, available online). Whereas additional cancers detected immediately under cotesting as opposed to primary HPV screening decreased with decreasing population-average cancer prevalence from 29 (95% CI = 15 to 42) per 100 000 screened to 0.8 (95% CI = 0.2 to 1.5) per 100 000 screened, the number of additional tests needed to detect 1 additional cancer increased substantially from 3000 to 114 000 more per 100 000 screened.
Although prior articles have focused on the relative sensitivity and specificity of cotesting vs primary HPV testing (9), our study focused on absolute measures by assessing the clinical implications of the transition from cotesting to primary HPV testing in 4 settings with different population-average precancer prevalence. The relationships between relative measures and our absolute measures (precancers detected and colposcopies conducted) are further described in the Supplementary Methods (available online).
In our study, we found that cotesting results in many additional tests and colposcopies, particularly in low prevalence settings and among HPV-negative individuals returning for retesting. The main goal of cervical screening is to detect and treat precancers before cancer develops (5). A short delay in detection of a precancer is not harmful, but depending on the setting, there may be a risk of loss to follow-up. We found that cotesting can initially have value in detecting CIN3 or cancer earlier in underinsured or uninsured individuals who have never previously been screened or are not up to date with their screening; however, with continued regular screening of this population, precancer prevalence can reduce to the point that cotesting may have more harms than benefits compared with primary HPV screening.
In our analysis, the clinical implications of transitioning from cotesting to primary HPV are described as per 100 000 individuals screened. This is equivalent to 20 years of practice for an individual practitioner seeing 20 patients per day with 250 practice days a year. For a practitioner who only serves underinsured or uninsured never or rarely screened individuals, it would require an average of 3.4 months to encounter CIN3 or cancer detectable by cotesting but not primary HPV. For a practitioner who only serves individuals returning for routine screening after a prior HPV-negative result, it would require an average of 4.5 years. Most practitioners would fall somewhere between these 2 extremes.
Economic analyses using various outcome metrics have had mixed conclusions about primary HPV screening compared with cotesting (10-12). Given our study results showing that the benefit and harm trade-offs depend on disease prevalence, additional economic analyses may be useful in understanding whether disease prevalence affects the cost effectiveness of various screening approaches.
A limitation of the analysis is that the NBCCEDP cohort is a convenience sample of individuals with HPV test results rather than a random sample of low-income underinsured or uninsured individuals. Because we are unable to differentiate between individuals receiving cotesting vs receiving HPV triage for an ASC-US cytology result, there is a much greater proportion of ASC-US results than would normally be found in a screening population. Additionally, symptomatic individuals may have more readily used the program than asymptomatic individuals, which would also result in greater proportions of high-grade cytology. As a result, the number of additional cotests and colposcopies needed to detect precancer or cancer under cotesting may be underestimated for underinsured or uninsured settings. Furthermore, the number of precancers not immediately detected from primary HPV screening may also be overestimated in these settings. Finally, our analysis assumes 100% adherence with referral to colposcopy; lower adherence would lead to fewer numbers of precancer or cancers immediately detected and greater numbers not immediately detected regardless of whether primary HPV or cotesting is employed.
In sum, these data suggest that cotesting confer an unfavorable benefit-to-harm ratio over primary HPV testing, especially in populations with a low prevalence of cervical precancer and cancer.
Supplementary Material
Acknowledgements
We extend a special thanks to the Center of Disease Control personnel, and particularly Mona Saraiya, for their helpful suggestions and answers to questions regarding the NBCCEDP data.
The funder did not play a role in the design of the study; the collection, analysis, and interpretation of the data; the writing of the manuscript; and the decision to submit the manuscript for publication.
Contributor Information
Shrutikona Das, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Rockville, MD, USA.
Nicolas Wentzensen, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Rockville, MD, USA.
George F Sawaya, Department of Obstetrics, Gynecology, & Reproductive Sciences, University of California San Francisco, San Francisco, CA, USA.
Didem Egemen, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Rockville, MD, USA.
Alexander Locke, Retired, Department of Obstetrics and Gynecology, The Permanente Medical Group, Sacramento, CA, USA.
Walter Kinney, Retired, Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, The Permanente Medical Group, Sacramento, CA, USA.
Thomas Lorey, Kaiser Permanente, The Permanente Medical Group Northern California Regional Laboratory, Berkeley, CA, USA.
Li C Cheung, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Rockville, MD, USA.
Data availability
The data used in this study can be found in the tables of previously published papers (3,4) and are also available from the excel worksheet provided in the Supplementary Material (available online).
Author contributions
Shrutikona Das, BA (Conceptualization; Data curation; Formal analysis; Methodology; Validation; Visualization; Writing—original draft; Writing—review & editing), Nicolas Wentzensen, MD, PhD (Conceptualization; Methodology; Validation; Writing—review & editing), George F. Sawaya, MD (Conceptualization; Methodology; Validation; Writing—review & editing), Didem Egemen, PhD (Conceptualization; Validation; Writing—review & editing), Alexander Locke, MD (Conceptualization; Validation; Writing—review & editing), Walter Kinney, MD (Conceptualization; Validation; Writing—review & editing), Thomas Lorey, MD (Conceptualization; Validation; Writing—review & editing), and Li Cheung, PhD (Conceptualization; Data curation; Formal analysis; Methodology; Validation; Writing—original draft; Writing—review & editing).
Funding
National Cancer Institute Intramural Research Program.
Conflicts of interest
All authors have nothing to disclose.
References
- 1. Downs LS Jr, Nayar R, Gerndt J, Saslow D; American Cancer Society Primary HPVSISC. Implementation in action: collaborating on the transition to primary HPV screening for cervical cancer in the United States. CA Cancer J Clin. 2023;73(5):458-460. doi: 10.3322/caac.21786 [DOI] [PubMed] [Google Scholar]
- 2. Bouvard V, Wentzensen N, Mackie A, et al. The IARC perspective on cervical cancer screening. N Engl J Med. 2021;385(20):1908-1918. doi: 10.1056/NEJMsr2030640 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Saraiya M, Cheung LC, Soman A, et al. Risk of cervical precancer and cancer among uninsured and underserved women from 2009 to 2017. Am J Obstet Gynecol. 2021;224(4):366.e1-366.e32. doi: 10.1016/j.ajog.2020.10.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Egemen D, Cheung LC, Chen X, et al. Risk estimates supporting the 2019 ASCCP risk-based management consensus guidelines. J Low Genit Tract Dis. 2020;24(2):132-143. doi: 10.1097/LGT.0000000000000529 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Schiffman M, Kinney WK, Cheung LC, et al. Relative performance of HPV and cytology components of cotesting in cervical screening. J Natl Cancer Inst. 2018;110(5):501-508. doi: 10.1093/jnci/djx225 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Cheung LC, Pan Q, Hyun N, et al. Mixture models for undiagnosed prevalent disease and interval-censored incident disease: Applications to a cohort assembled from electronic health records. Stat Med. 2017;36(22):3583-3595. doi: 10.1002/sim.7380 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Cheung LC, Egemen D, Chen X, et al. 2019 ASCCP risk-based management consensus guidelines: methods for risk estimation, recommended management, and validation. J Low Genit Tract Dis. 2020;24(2):90-101. doi: 10.1097/LGT.0000000000000528 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. https://www.R-project.org/. Accessed April 30, 2024.
- 9. Arbyn M, Ronco G, Anttila A, et al. Evidence regarding human papillomavirus testing in secondary prevention of cervical cancer. Vaccine. 2012;30(suppl 5):F88-99. doi: 10.1016/j.vaccine.2012.06.095 [DOI] [PubMed] [Google Scholar]
- 10. Jin XW, Lipold L, Foucher J, et al. Cost-effectiveness of primary HPV testing, cytology and co-testing as cervical cancer screening for women above age 30 years. J Gen Intern Med. 2016;31(11):1338-1344. doi: 10.1007/s11606-016-3772-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Sawaya GF, Sanstead E, Alarid-Escudero F, et al. Estimated quality of life and economic outcomes associated with 12 cervical cancer screening strategies: a cost-effectiveness analysis. JAMA Intern Med. 2019;179(7):867-878. doi: 10.1001/jamainternmed.2019.0299 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Huh WK, Williams E, Huang J, Bramley T, Poulios N.. Cost effectiveness of human papillomavirus-16/18 genotyping in cervical cancer screening. Appl Health Econ Health Policy. 2015;13(1):95-107. doi: 10.1007/s40258-014-0135-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data used in this study can be found in the tables of previously published papers (3,4) and are also available from the excel worksheet provided in the Supplementary Material (available online).