Abstract
Background
We demonstrated that cervical cancer risk following any screening result is lower if there is a known prior negative screening history vs an unknown screening history. We extended these findings to look at how screening performs following repeatedly negative screening results.
Methods
Approximately 1.7 million women aged 30-64 years underwent triennial human papillomavirus (HPV) and cytology co-testing from 2003 to 2021. We modeled 5-year risks of cervical intraepithelial neoplasia grade 3 or more severe diagnoses (CIN3+) and invasive cervical cancer for the initial co-test, and then successive rounds following negative co-testing. A logistic-Weibull prevalence-incidence model was used to model risks.
Results
HPV test positivity was greater than cytology positivity for only the first co-test, and both positivity rates decreased with each screening round. Diagnostic yields of CIN3+ and cancer declined with each round of screening so the first screen yielded 8-fold more CIN3+ and invasive cancer than the fifth screen following 4 consecutive negative co-tests. Five-year risks of CIN3+ for positive and negative HPV and cytology results, individually or combined, decreased considerably after the first screen, with smaller decreases in each subsequent round. For cancer, we noticed a considerable decrease with the first screen only. Five-year CIN3+ risks were greater for positive HPV or cytology results, with a longer antecedent screening interval and younger age at screening (Ptrend < .001).
Conclusions
Triennial screening that includes HPV testing becomes inefficient after a single and more so after multiple negative screens. These data support the use of longer screening intervals, especially following negative screen(s).
Introduction
Interim guidance for triennial human papillomavirus (HPV) DNA and cytology co-testing was first published in 2004 following an expert meeting in 2003.1 Kaiser Permanente Northern California (KPNC) introduced triennial co-testing in 2003, a practice that continued until switching to primary HPV screening in 2021. Since then, US cervical screening guidelines have recommended HPV testing or HPV-cytology co-testing every 5 years, an interval that was chosen mainly based on modeling results.2-4
Using electronic health record data, we previously examined the impact of a negative screening history by co-testing at KPNC over 3 rounds of screening (ie, the first co-test and 2 subsequent tests after negative co-test) on the clinical meaning (cervical cancer risk) of a negative5 and a positive co-test.6 In the former analysis,5 the main observation was that the risk of precancer, cervical intraepithelial neoplasia grade 3 (CIN3) or adenocarcinoma in situ (AIS), and invasive cervical cancer (together, CIN3+) was lower for negative screening results (negative co-test, negative HPV, and negative cytology) following a history vs an unknown screening history.
Here we update those analyses with larger numbers of women screened and more rounds of screening to look at the long-term patterns of directly observed CIN3+ risk for positive and negative screens. Importantly, because of those larger numbers, we were able to examine the risk of invasive cervical cancer following multiple rounds of screening. These data are particularly relevant for informing recommendations for screening intervals because a negative screen should reassure patients for many years with respect to the frank invasive cervical cancers we are trying to prevent.
Methods
Population
The KPNC cohort consists of approximately 2.5 million women, approximately 1.7 million of whom are aged 30-64 years, enrolled in co-test screening from January 1, 2003, through December 31, 2021. This cohort has been extensively described.7 KPNC screened women aged 30-64 years by HPV and cytology co-testing (and women aged 21-29 years with cytology alone) until 2013, when it lowered the co-testing screening age to 25 years. We limited the current analysis to women 30-64 years of age. The baseline visit was defined as the first co-test within this period with no prior indication of cervical intraepithelial neoplasia grade 2 (CIN2) or more severe histopathology, no known prior abnormal cytology, and no prior indication of hysterectomy.
Screening and clinical management
At KPNC, women aged 30 years or older (and aged ≥25 years from 2013 or later) underwent triennial HPV-cytology co-testing, wherein 2 cervical specimens were collected: 1 for Paponicolaou cytology (initially conventional smears, then transitioned in 2009 to liquid-based BD SurePath tests [Becton Dickinson]) and a second for high-risk HPV testing using Hybrid Capture 2 (QIAGEN). Cytology was interpreted using the Bethesda System for reporting cervical cytology.8,9 Screen-positive women were managed under internal protocols closely aligned with contemporaneous US guidelines, in which patients with definite cytological abnormalities (HPV-positive atypical squamous cells of undetermined significance [AS-CUS], low-grade squamous intraepithelial lesion (LSIL), or more severe cytologic interpretations) proceeded directly to colposcopy, whereas women with HPV-positive, cytology-negative results were reevaluated annually by co-test and referred if subsequent cytology were abnormal or if a second consecutive HPV-positive cytology-negative co-test occurred (policy effective 2006). Human papillomavirus–negative and cytology-negative women were invited to rescreen at 3-year intervals, whereas women with HPV-negative AS-CUS cytology were retested at 1-year to 3-year intervals as protocols evolved.
Women aged 65 years or older exited screening after a single negative co-test or 2 consecutive negative Papanicolaou tests obtained before co-testing was introduced. Case ascertainment relied on linked electronic health records and the integrated KPNC cancer registry. Throughout the study period, abnormal results and subsequent diagnoses were otherwise managed in accordance with national standards and KPNC guidelines.
All data use was approved by the KPNC Institutional Review Board. The National Institutes of Health Office of Human Subjects Research deemed the study exempt from additional review.
Statistical analysis
We used 5-year cumulative detection of CIN3+ as a proxy for cancer risk as well as 5-year cumulative risk of invasive cancer for all but the fifth round of co-testing (due to a limited number of cancers). Diagnoses of CIN2/3 were not included in our definition of CIN3+. A 5-year horizon accommodates heterogeneity in downstream management (immediate colposcopy vs surveillance), captures late attenders, and mitigates underascertainment at any single screening encounter. For each screening round, the first through the fifth co-test after 0, 1, 2, 3, or 4 consecutive negative co-tests, we calculated 5-year CIN3+ and invasive cervical cancer risk following HPV-positive and HPV-negative results, cytology-positive and cytology-negative results, and combination of HPV-cytology results.
Risks were estimated using a logistic-Weibull prevalence-incidence mixture model, which pairs a logistic regression model for disease present at the index screen with a Weibull survival model for incident lesions detected during follow-up.10 The Weibull survival model produces smoother, more biologically plausible risk curves than nonparametric alternatives and handles interval censoring (ie, the time events that are first detectable lie between the last negative screen/colposcopy and event detection).11 Joint estimation of the logistic and Weibull parameters allows lesions discovered later—but potentially present at the index screen if a screen-positive individual did not attend colposcopy—to be treated probabilistically as missed prevalent disease. This hybrid framework delivers stable, interpretable estimates while naturally handling ascertainment bias related to colposcopy, rescreen referral, and compliance.12 Variance was computed on the complementary log-log scale, then back-transformed for conventional 95% confidence intervals.
To describe how each round of screening affected precancer and cancer risk, we fit change point models to the estimated risks using weighted (by the inverse of the variance in the risk estimate) linear regression. Explanatory effects considered included abrupt changes in the average risk following a specific screening round and more gradual changes (ie, trends) in the average risk following each subsequent screening round. Finally, we explored the effects of age at the time of the screen (30-39, 40-49, and 50-64 years) for screening rounds 1-3 and the antecedent screening interval (<2.5, 2.5 to <3.5, and ≥3.5 years between the prior negative co-test and the index screening co-test, the latter of which is time zero for estimating the subsequent risk) for screening rounds 2 and 3 on the 5-year CIN3+ risks; there were too few CIN3+ cases for stratification for rounds 4 and 5 and too few cervical cancers for stratification for any screening rounds. Tests for all explanatory variables were conducted using the Wald test. All analyses were done using R, version 4.5.0, statistical software (R Foundation for Statistical Computing) and the PIMixture package or SAS, version 9.4, software (SAS Institute Inc). P values were 2 sided, and P of less than .05 was considered significant.
Results
The CONSORT diagram for this screening cohort is shown in Figure 1. Starting with a cohort of 1 733 188 women aged 30-64 years at their inaugural co-test, 1 592 805 women contributed data to the first co-test, 887 994 to the second, 537 912 to the third, 316 094 to the fourth, and 151 694 to the fifth. Basic statistics for the cohort by round of screening are shown in Table 1. Not surprisingly, the cohort aged with each round of co-testing, although the mean age increased less than 3 years between rounds due to the right truncation because screening stops at age 65 years with a negative screening history. Screening intervals between rounds were similar, from a mean of 3.36 years and a median of 3.06 years between rounds 1 and 2 and a mean and median of 3.06 years between rounds 4 and 5.
Figure 1.
CONSORT diagram for the screening cohort. AGC = typical glandular cell; ASC-H = atypical squamous cell, cannot exclude high-grade squamous intraepithelial lesion; AS-CUS = atypical squamous cells of undetermined significance; CIN = cervical intraepithelial neoplasia; HPV = human papillomavirus; HSIL = high-grade squamous intraepithelial lesion; LSIL = low-grade squamous intraepithelial lesion.
Table 1.
Description of the screening cohort, by round of screening.
| Category | First co-test | Second co-testa | Third co-testa | Fourth co-testa | Fifth co-testa |
|---|---|---|---|---|---|
| n = 1 592 805 | n = 887 994 | n = 537 912 | n = 316 094 | n = 151 694 | |
| Age at co-test, y | |||||
| Mean (SD) | 42.33 (10.07) | 45.32 (9.50) | 48.02 (8.76) | 50.41 (7.93) | 52.3 (7.12) |
| Median (IQR) | 40 (33-50) | 44 (36-53) | 48 (40-55) | 50 (43-57) | 52 (46-58) |
| Age group at co-test, No. (%), y | |||||
| 30-39 | 752 475 (47.24) | 314 529 (35.42) | 124 786 (23.20) | 29 887 (9.46) | 1797 (1.18) |
| 40-49 | 411 216 (25.82) | 261 652 (29.47) | 177 242 (32.95) | 118 250 (37.41) | 55 845 (36.81) |
| 50-64 | 429 114 (26.94) | 311 813 (35.11) | 235 884 (43.85) | 167 957 (53.14) | 94 052 (62.00) |
| Screening interval, y | |||||
| Mean (SD) | NA | 3.34 (1.54) | 3.21 (1.13) | 3.13 (0.85) | 3.07 (0.69) |
| Median (IQR) | NA | 3.06 (2.743-3.46) | 3.06 (2.81-3.33) | 3.06 (2.88-3.27) | 3.06 (2.91-3.24) |
| Race and ethnicity, No. (%) | |||||
| Asian or Pacific Islander | 285 706 (17.94) | 179 740 (20.24) | 113 481 (21.10) | 67 768 (21.44) | 31 717 (20.91) |
| Black | 97 089 (6.10) | 59 724 (6.73) | 38 058 (7.08) | 22 973 (7.27) | 10 916 (7.20) |
| Hispanic | 227 583 (14.29) | 136 521 (15.37) | 83 647 (15.55) | 49 277 (15.59) | 23 888 (15.75) |
| Native American | 6114 (0.38) | 3476 (0.39) | 2041 (0.38) | 1167 (0.37) | 525 (0.35) |
| White | 609 713 (38.28) | 38 8834 (43.79) | 247 796 (46.07) | 149 014 (47.14) | 73 502 (48.45) |
| Unknown | 20 928 (1.31) | 9588 (1.08) | 51,04 (0.95) | 2849 (0.90) | 1329 (0.88) |
| Other | 22 309 (1.40) | 13 801 (1.55) | 8659 (1.61) | 5032 (1.59) | 2344 (1.55) |
| Missing | 323 363 (20.30) | 96 310 (10.85) | 39 126 (7.27) | 18 014 (5.70) | 7473 (4.93) |
Abbreviation: NA = not applicable.
Denotes screening rounds following a negative co-test.
Figure 2 shows the test positivity and diagnostic yield of important clinical endpoints by each round of screening. HPV test and cytology positivity, defined as AS-CUS or more severe interpretations, was highest in the first round of screening and was statistically significantly higher for HPV than for cytology, 6.76% and 5.12%, respectively (P < .001, McNemar’s chi-square test) (Figure 2, A). HPV test and cytology positivity then decreased and converged in subsequent rounds of screening (round 2, 3.73% vs 3.76%, P = .004; round 3, 3.02% vs 2.93%, P = .001; round 4, 2.72% vs 2.67%, P = .10; round 5, 2.53% vs 2.58%, P = .22) (Figure 2, A and B). Adjusting for age, the HPV positivity was slightly higher than the cytology positivity due to older populations being less likely to be HPV positive and more likely to have HPV-negative cytology (eg, HPV-negative ASC-US) than younger populations.
Figure 2.
Screening positivity (A) and age-standardized positivity (B) for human papillomavirus test (HPV+) and cytology (atypical squamous cells of undetermined significance or more severe interpretation [Cyto+]) are shown following multiple rounds of co-testing. (C) Diagnostic yield of cervical intraepithelial neoplasia grade 3 or more severe diagnoses (CIN3+) and cervical cancer. (D) Diagnostic yield of squamous (CIN3 or squamous cell carcinoma) and glandular (adenocarcinoma and adenocarcinoma in situ) diagnoses, by round of screening. Cytology was considered positive if interpreted as atypical squamous cells of undetermined significance or more severe interpretation and negative if interpreted as negative for intraepithelial neoplasia or malignancy. Diagnostic yields are presented on a log10 scale.
aDenotes screening rounds following a negative co-test result.
Diagnostic yield of CIN3+ and cervical cancer decreased with each round of screening, decreasing approximately 3-fold from the first to the second round of screening, then approximately by 50% each subsequent round (Figure 2, C). Similar patterns were seen for glandular and squamous histology (Figure 2, D). Overall, the diagnostic yields of CIN3+ and cervical cancer were approximately 8-fold lower on the fifth screen than on the first screen.
Figure 3 and Table S1 presents 5-year risks of CIN3+ and cervical cancer by screening result and round of screening. The HPV testing showed greater risk stratification between positive and negative results than did cytology for CIN3+ (Figure 3, A) and cervical cancer (Figure 3, B) for all rounds of screening. Importantly, a negative HPV test provided greater reassurance against cervical cancer than did negative cytology (negative for intraepithelial neoplasia or malignancy) (Figure 3, D). A negative co-test result—negative HPV testing and cytology—was only incrementally more reassuring against cervical cancer than a negative HPV test (regardless of the cytology result) (Figure 3, D vs 3, B) for all screening rounds; the small difference in risk between a negative co-test and a negative HPV test was due to the small group of women with HPV-negative/cytology-positive disease, a group that was at higher risk than all women who were HPV negative.
Figure 3.
Five-year risk of cervical intraepithelial neoplasia grade 3 or more severe diagnoses (CIN3+) (A, C) and cervical cancer (B, D) following multiple rounds of co-testing by human papillomavirus (HPV) and cytology (Cyto). The results are presented for single test results for CIN3+ and cancer (A and B, respectively) and paired test results for CIN3+ and cancer (C and D, respectively). Cytology was considered positive (Cyto+) if interpreted as atypical squamous cells of undetermined significance or more severe disease and negative (Cyto-) if interpreted as negative for intraepithelial neoplasia or malignancy.
aDenotes screening rounds following a negative co-test.
Five-year risks of CIN3+ (Figure 3, A and C) and cervical cancer (Figure 3, B and D) were the greatest in the first round of screening compared with subsequent rounds. Notably, in explaining the effect of screening rounds on risk, models using a change point at the second co-test generally fit patterns of 5-year risks over sequential screens better than strictly linear models (Table S2). Generally, there was a larger decrease (2- to 3-fold for CIN3+ disease and 50% for cervical cancer) in risk between the first and second co-test (for CIN3+, P < .001 for all screening results; for cervical cancer, P < .001 for cytology positive and HPV positive and P = .01 for cytology negative but P = .73 for HPV negative and P = .39 for a negative co-test because of the small number of cancers) and a linear, smaller decrease in risk with each subsequent round for CIN3+ disease (negative coefficient and P < .01 for all results except HPV negative/cytology positive) and no decrease in risk with each subsequent round for cervical cancer (either coefficient close to zero or P > .05 for all results). Other models considering a large decrease in risk occurring after the second, third, fourth, or fifth co-test were worse fit to these data (smaller log likelihood) (data not shown).
We stratified CIN3+ risks by antecedent screening intervals and age, as shown in Table 2. Five-year CIN3+ risks following a positive HPV or cytology result for the second and third co-tests were greater if the antecedent screening interval was longer (Ptrend < .001 for all), but this finding was not the case following a negative HPV or cytology result. Five-year CIN3+ risks following any HPV or cytology result for the first, second, and third co-tests were lower with older ages at the time of screening (Ptrend < .001 for all).
Table 2.
Five-year risks of CIN3+ disease following positive or negative HPV test result and a positive or negative cytology result, stratified by age group at the time of the screen or on the antecedent screening interval (the time between this screen and the last negative screen).a
| First co-testb |
Second co-testb,c |
Third co-testb,c |
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Screening test and result | Age group, y | Antecedent screening interval, y | No. | No. with CIN3+ | CIN3+ 5-y cumulative risk, % | 95% CI | No. | No. with CIN3+ | CIN3+ 5 -y cumulative risk, % | 95% CI | No. | No. with CIN3+ | CIN3+ 5-y cumulative risk, % | 95% CI |
| HPV positive | 30-39 | All | 69 268 | 4920 | 8.42 | 8.19 to 8.65 | 16 237 | 596 | 4.33 | 4.00 to 4.69 | 5212 | 145 | 3.48 | 2.95 to 4.09 |
| <2.5 | 3996 | 121 | 3.22 | 2.77 to 3.74 | 1152 | 21 | 2.43 | 1.80 to 3.26 | ||||||
| 2.5 to <3.5 | 9026 | 337 | 4.43 | 4.03 to 4.87 | 3314 | 108 | 3.70 | 3.10 to 4.40 | ||||||
| 3.5 | 3215 | 138 | 5.71 | 5.02 to 6.49 | 746 | 16 | 4.19 | 3.23 to 5.44 | ||||||
| 40-49 | All | 21 992 | 1587 | 8.18 | 7.80 to 8.57 | 9311 | 288 | 3.45 | 3.08 to 3.86 | 5898 | 151 | 2.99 | 2.56 to 3.51 | |
| <2.5 | 1698 | 42 | 2.45 | 2.04 to 2.93 | 949 | 16 | 1.93 | 1.42 to 2.62 | ||||||
| 2.5 to <3.5 | 4803 | 148 | 3.34 | 2.94 to 3.79 | 3504 | 87 | 3.08 | 2.58 to 3.66 | ||||||
| 3.5 | 2810 | 98 | 4.38 | 3.80 to 5.05 | 1445 | 48 | 3.57 | 2.82 to 4.52 | ||||||
| 50-64 | All | 16 433 | 942 | 6.78 | 6.37 to 7.21 | 7578 | 184 | 2.86 | 2.48 to 3.30 | 5110 | 120 | 2.87 | 2.40 to 3.43 | |
| <2.5 | 1059 | 17 | 2.00 | 1.63 to 2.45 | 591 | 15 | 1.95 | 1.40 to 2.70 | ||||||
| 2.5 to <3.5 | 4112 | 103 | 2.76 | 2.37 to 3.22 | 3273 | 80 | 2.91 | 2.40 to 3.52 | ||||||
| 3.5 | 2407 | 64 | 3.56 | 3.00 to 4.21 | 1246 | 25 | 3.26 | 2.50 to 4.25 | ||||||
| HPV negative | 30-39 | All | 683 207 | 1534 | 0.22 | 0.21 to 0.24 | 298 292 | 474 | 0.16 | 0.14 to 0.17 | 119.574 | 144 | 0.15 | 0.12 to 0.18 |
| <2.5 | 65 543 | 131 | 0.15 | 0.13 to 0.17 | 23 710 | 41 | 0.16 | 0.12 to 0.20 | ||||||
| 2.5 to <3.5 | 175 789 | 270 | 0.16 | 0.14 to 0.18 | 79 377 | 86 | 0.14 | 0.12 to 0.17 | ||||||
| 3.5 | 56 960 | 73 | 0.16 | 0.13 to 0.19 | 16 487 | 17 | 0.17 | 0.12 to 0.23 | ||||||
| 40-49 | All | 389 223 | 626 | 0.12 | 0.11 to 0.13 | 252 341 | 308 | 0.10 | 0.08 to 0.11 | 171 344 | 152 | 0.09 | 0.08 to 0.11 | |
| <2.5 | 44 411 | 74 | 0.09 | 0.08 to 0.11 | 26 700 | 34 | 0.10 | 0.07 to 0.12 | ||||||
| 2.5 to <3.5 | 139 480 | 178 | 0.10 | 0.08 to 0.11 | 106 935 | 104 | 0.09 | 0.07 to 0.10 | ||||||
| 3.5 | 68 450 | 56 | 0.10 | 0.08 to 0.12 | 37 709 | 14 | 0.10 | 0.08 to 0.14 | ||||||
| 50-64 | All | 412 680 | 410 | 0.11 | 0.10 to 0.12 | 304 234 | 244 | 0.09 | 0.08 to 0.10 | 230 773 | 127 | 0.07 | 0.06 to 0.09 | |
| <2.5 | 44 006 | 43 | 0.08 | 0.07 to 0.10 | 28 887 | 28 | 0.08 | 0.06 to 0.11 | ||||||
| 2.5 to <3.5 | 181 609 | 155 | 0.09 | 0.08 to 0.10 | 154 856 | 78 | 0.07 | 0.06 to 0.08 | ||||||
| 3.5 | 78 619 | 46 | 0.09 | 0.07 to 0.11 | 47 030 | 21 | 0.08 | 0.06 to 0.11 | ||||||
| Cytology positive | 30-39 | All | 46 628 | 3632 | 8.82 | 8.56 to 9.09 | 13 106 | 441 | 3.61 | 3.32 to 3.93 | 4343 | 112 | 3.05 | 2.61 to 3.56 |
| <2.5 | 3088 | 87 | 2.80 | 2.35 to 3.34 | 953 | 16 | 1.99 | 1.42 to 2.78 | ||||||
| 2.5 to <3.5 | 7374 | 241 | 3.64 | 3.26 to 4.07 | 2747 | 82 | 3.16 | 2.59 to 3.86 | ||||||
| 3.5 | 2644 | 113 | 5.27 | 4.57 to 6.08 | 643 | 14 | 4.05 | 3.03 to 5.42 | ||||||
| 40-49 | All | 20 023 | 1225 | 6.51 | 6.20 to 6.84 | 10 385 | 227 | 2.25 | 2.01 to 2.52 | 5999 | 124 | 2.22 | 1.91 to 2.59 | |
| <2.5 | 1855 | 31 | 1.61 | 1.31 to 1.98 | 1110 | 12 | 1.39 | 0.99 to 1.95 | ||||||
| 2.5 to <3.5 | 5570 | 116 | 2.09 | 1.80 to 2.42 | 3495 | 70 | 2.30 | 1.89 to 2.79 | ||||||
| 3.5 | 2960 | 80 | 3.05 | 2.60 to 3.58 | 1394 | 42 | 2.95 | 2.27 to 3.82 | ||||||
| 50-64 | All | 14 964 | 723 | 4.93 | 4.63 to 5.26 | 9094 | 149 | 1.79 | 1.57 to 2.04 | 5440 | 92 | 1.85 | 1.56 to 2.20 | |
| <2.5 | 1330 | 13 | 1.27 | 1.00 to 1.60 | 830 | 14 | 1.24 | 0.86 to 1.77 | ||||||
| 2.5 to <3.5 | 5139 | 79 | 1.67 | 1.41 to 1.99 | 3370 | 57 | 1.96 | 1.57 to 2.44 | ||||||
| 3.5 | 2625 | 57 | 2.42 | 2.00 to 2.92 | 1240 | 21 | 2.51 | 1.87 to 3.37 | ||||||
| Cytology negative | 30-39 | All | 705 847 | 2822 | 0.49 | 0.47 to 0.50 | 301 423 | 629 | 0.25 | 0.23 to 0.27 | 120 443 | 177 | 0.21 | 0.18 to 0.24 |
| <2.5 | 66 451 | 165 | 0.23 | 0.20 to 0.26 | 23 909 | 46 | 0.21 | 0.16 to 0.26 | ||||||
| 2.5 to <3.5 | 177 441 | 366 | 0.25 | 0.22 to 0.27 | 79 944 | 112 | 0.20 | 0.17 to 0.23 | ||||||
| 3.5 | 57 531 | 98 | 0.25 | 0.21 to 0.29 | 16 590 | 19 | 0.24 | 0.18 to 0.31 | ||||||
| 40-49 | All | 391 193 | 988 | 0.25 | 0.23 to 0.26 | 251 267 | 369 | 0.14 | 0.13 to 0.15 | 171 243 | 179 | 0.12 | 0.11 to 0.14 | |
| <2.5 | 44 254 | 85 | 0.13 | 0.11 to 0.15 | 26 539 | 38 | 0.12 | 0.10 to 0.15 | ||||||
| 2.5 to <3.5 | 138 713 | 210 | 0.14 | 0.12 to 0.15 | 106 944 | 121 | 0.12 | 0.10 to 0.14 | ||||||
| 3.5 | 68 300 | 74 | 0.14 | 0.12 to 0.16 | 37 760 | 20 | 0.14 | 0.11 to 0.18 | ||||||
| 50-64 | All | 414 150 | 629 | 0.20 | 0.19 to 0.22 | 302 719 | 279 | 0.11 | 0.10 to 0.13 | 230 444 | 155 | 0.10 | 0.09 to 0.12 | |
| <2.5 | 43 735 | 47 | 0.10 | 0.09 to 0.12 | 28 648 | 29 | 0.10 | 0.08 to 0.13 | ||||||
| 2.5 to <3.5 | 180 582 | 179 | 0.11 | 0.10 to 0.13 | 154 759 | 101 | 0.09 | 0.08 to 0.11 | ||||||
| ≥3.5 | 78 402 | 53 | 0.12 | 0.10 to 0.14 | 47 037 | 25 | 0.11 | 0.08 to 0.14 | ||||||
Abbreviations: CIN3+ = cervical intraepithelial neoplasia grade ≥3; HVP = human papillomavirus.
Cytology was considered positive if interpreted as atypical squamous cells of undetermined significance or more severe and negative if interpreted as negative for intraepithelial neoplasia or malignancy.
P values for trends across age groups were <.001 for age group for all screening rounds for HPV positive, HPV negative, cytology positive, and cytology negative. P values for trends across antecedent screening interval category were <.001 for screening rounds 2 and 3 (not applicable to screening round 1) for HPV positive and cytology positive; .49 and .40 for HPV negative and cytology negative, respectively, for screening round 2; and .94 and .58 for HPV negative and cytology negative, respectively, for screening round 3.
Denotes screening rounds following a negative co-test.
Discussion
We observed that successive rounds of cervical cancer screening by triennial co-testing led to fewer positive screens, fewer diagnoses of CIN3+ disease and invasive cervical cancers per screen, and a lower risk of cancer and CIN3+ disease for both positive and negative test results, with the largest reduction after the first negative co-test. The decreasing risk of cancer is likely the result of 2 factors. First, there are successively fewer false-negative screens preceding the cancer diagnoses with a sensitive HPV testing–based screening (ie, risks of invasive cervical cancer following negative tests decrease in subsequent rounds). Second, those CIN3 cases that do arise following a negative screen are most likely incident (except for a small number that tested falsely negative), some of which probably arose from recently acquired HPV infection, with a relatively low invasive potential compared with prevalent CIN3 that has been present for an unknown but likely longer time. Indeed, approaching a nadir in CIN3+ risk would suggest a sort of equipoise/steady state between screening identifying women with CIN3, which is then excised, and newly developed CIN3. Modeling of the natural history of cervical cancer finds that the transition from CIN3 to cancer takes much longer than does HPV infection to CIN3,13,14 and more sensitive, HPV testing–based screening will find CIN3 earlier, extending that difference in sojourn time.
These data highlight that fewer and fewer women benefitted with subsequent rounds of triennial cervical screening by co-testing because of the decrease in HPV positivity and lower positive predictive value of screening. Consequently, cervical screening by triennial co-testing became statistically significantly less efficient and more cost-ineffective with subsequent screens. These data support using screening intervals longer than 3 years, if not initially as currently recommended nationally2,3,15 and internationally,16 then following 1 or 2 rounds of negative screens at shorter intervals. Decision support tools and implementation of capitation/valued-based care may be necessary to avoid overscreening. Although modeling results4 previously suggested that 3-year screening intervals should be extended to 5 years when using HPV-based screening, this is the first direct observation of risk of CIN3+ disease and cancer following up to 5 rounds of organized screening in a large population.
We also observed that older age at the time of the screening was associated with lower risks. We interpret this finding as older age being a proxy for past effective screening either at KPNC or elsewhere before joining KPNC. These data may support even longer screening intervals in older women than in younger women.
We note that the Hybrid Capture 2 HPV test was used over this period at KPNC, which does not offer individual HPV genotyping for HPV16 and HPV18 as the next generation of HPV tests do, which would have resulted in earlier and perhaps increased detection of CIN3+. Clinical algorithms did not refer HPV-positive/cytology-negative disease or HPV-negative disease/AS-CUS cases immediately to colposcopy,17-20 and by 2019, HPV-negative/LSIL cases were referred based on individual risk assessment.17 As a result, biases in ascertainment may have influenced and perhaps underestimated differences in CIN3+ detection between positive HPV and cytology results.
We highlight several limitations of this analysis. First, we used interval-censored observational data, with the obvious caveats and limitations of such data. We can make counterfactual inferences only related to primary HPV screening vs co-testing for the safety of a negative screen. Nevertheless, these data are similar to what have previously been reported in this cohort5,21 and others22,23 in that co-testing adds little benefit in terms of increased detection or improved safety following a negative screen and increases screening (false) positivity compared with HPV testing alone. These data are consistent with the long-term safety afforded by negative screening that includes HPV testing, and the excellent safety against invasive cervical cancer derived from a negative co-test is primarily the consequence of the negative HPV test.24,25 Cost-effectiveness analyses have shown that the small incremental added benefit of co-testing over primary HPV screening is cost inefficient.26 Together, these data support worldwide adoption of primary HPV screening as the new standard of care for routine cervical cancer screening.2,3,15,16
A second limitation is that KPNC screened by co-testing every 3 years rather than every 5 years, as has been the recommendation for screening, including HPV testing, since 2012.2,3,15 Inferences of what these results might be for quintennial vs triennial cervical screening are more uncertain, but we hypothesize that more rounds of quintennial screening might be needed to achieve this nadir (irreducible) risk than triennial screening or that nadir might be slightly greater (ie, longer interval to allow more new abnormalities to develop before being detected and treated). The ideal program would achieve zero cancer risk and reach a steady state for CIN3/AIS disease. Even triennial co-testing, however, could not achieve zero cancer risk at the cost of the noted inefficiencies. It is also worth pointing out that quintennial HPV testing–based cervical screening is rarely practiced in the United States, and the “average” screening interval is closer to triennial screening.27-31
Notably, we observed that the antecedent screening interval had little effect on the subsequent 5-year CIN3+ risks following a negative HPV or cytology test result, which seems logical because a new infection would need to be acquired after the index screen. Perhaps not surprisingly, a longer antecedent screening interval was associated with greater 5-year CIN3+ risks following a positive HPV or cytology test result because the longer interval may be selecting for interval HPV infections that have persisted for a longer vs shorter interval. Although we do not have sufficient numbers of cancers to look at the cancer-only outcome, we suspect that most of these CIN3+ are incident CIN3 with relatively low invasive cancer potential.
Even in this large cohort of women, the number of invasive cervical cancers was limited, and therefore some analyses were not possible with a cancer endpoint. This limitation is important in the context of cervical cancer screening because its primary goals are to find and treat CIN3 before it becomes cancer and to rule out cancer if the screen is negative (ie, cervical cancer incidence should be considered a screening failure).
Importantly, a negative screening history suggests adjustments for subsequent screening and management based on risk-based care.17 Already women who are found to have low-grade cytologic abnormalities (AS-CUS and LSIL) following a negative screening that includes HPV testing are not recommended for immediate colposcopy, as was done historically,18-20 but undergo annual surveillance.17 Future analyses will examine specific combinations of positive screening and negative screening history to determine which might warrant less aggressive management based on cervical cancer risk and the principle of equal care for equal risk.
Supplementary Material
Acknowledgments
The funder had no role in the design of the study; the collection, analysis, or interpretation of the data; or the writing of the manuscript and decision to submit it for publication.
Additional Information: Co-author Thomas Lorey, MD, was retired.
Contributor Information
Philip E Castle, Division of Cancer Prevention, US National Cancer Institute, Rockville, MD, United States; Division of Cancer Epidemiology and Genetics, US National Cancer Institute, Rockville, MD, United States.
Brian Befano, Department of Epidemiology, University of Washington, Seattle, WA, United States; Information Management Services, Calverton, MD, United States.
Marianne Hyer, Information Management Services, Calverton, MD, United States.
Li C Cheung, Division of Cancer Epidemiology and Genetics, US National Cancer Institute, Rockville, MD, United States.
Thomas Lorey, Permanente Medical Group Regional Laboratory, Kaiser Permanente, Berkeley, CA, United States.
Nancy Poitras, Permanente Medical Group Regional Laboratory, Kaiser Permanente, Berkeley, CA, United States.
Nicolas Wentzensen, Division of Cancer Epidemiology and Genetics, US National Cancer Institute, Rockville, MD, United States.
Author contributions
Philip Castle (Conceptualization, Formal analysis, Investigation, Project administration, Writing—original draft, Writing—review & editing), Brian Befano (Data curation, Formal analysis, Methodology, Writing—review & editing), Marianne Hyer (Formal analysis, Writing—review & editing), Li Cheung (Conceptualization, Formal analysis, Writing—review & editing), Thomas S. Lorey (Data curation, Project administration, Resources, Supervision, Writing—review & editing), Nancy Poitras (Data curation, Project administration, Writing—review & editing), and Nicolas Wentzensen (Data curation, Formal analysis, Funding acquisition, Project administration, Writing—review & editing)
Supplementary material
Supplementary material is available at JNCI: The Journal of the National Cancer Institute online.
Funding
This work was supported by the intramural research program (ZIACP101237-01) of the US National Cancer Institute. Brian Befano was supported by the National Cancer Institute under grant No. T32CA09168.
Conflicts of interest
P.E.C. has received HPV tests and assays for research at reduced or no cost from Cepheid and Atila Biosystems. P.E.C., who is a JNCI associate editor and co-author on this article, was not involved in the editorial review or decision to publish the manuscript.
Data availability
The data underlying this article cannot be shared publicly because they are the property of KPNC.
References
- 1. Wright TC, Schiffman M, Solomon D, et al. Interim guidance for the use of human papillomavirus DNA testing as an adjunct to cervical cytology for screening. Obstet Gynecol. 2004;103:304-309. 10.1097/01.AOG.0000109426.82624.f8 [DOI] [PubMed] [Google Scholar]
- 2. Fontham ETH, Wolf AMD, Church TR, et al. Cervical cancer screening for individuals at average risk: 2020 guideline update from the American Cancer Society. CA Cancer J Clin. 2020;70:321-346. 10.3322/caac.21628 [DOI] [PubMed] [Google Scholar]
- 3. Curry SJ, Krist AH, Owens DK, et al. ; US Preventive Services Task Force. Screening for cervical cancer: US Preventive Services Task Force recommendation statement. JAMA. 2018;320:674-686. 10.1001/jama.2018.10897 [DOI] [PubMed] [Google Scholar]
- 4. Kim JJ, Burger EA, Regan C, Sy S. Screening for cervical cancer in primary care: a decision analysis for the US Preventive Services Task Force. JAMA. 2018;320:706-714. 10.1001/jama.2017.19872 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Castle PE, Kinney WK, Xue X, et al. Effect of several negative rounds of human papillomavirus and cytology co-testing on safety against cervical cancer: an observational cohort study. Ann Intern Med. 2018;168:20-29. 10.7326/M17-1609 [DOI] [PubMed] [Google Scholar]
- 6. Castle PE, Kinney WK, Xue X, et al. Role of screening history in clinical meaning and optimal management of positive cervical screening results. J Natl Cancer Inst. 2019;111:820-827. 10.1093/jnci/djy192 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Katki HA, Kinney WK, Fetterman B, et al. Cervical cancer risk for women undergoing concurrent testing for human papillomavirus and cervical cytology: a population-based study in routine clinical practice. Lancet Oncol. 2011;12:663-672. 10.1016/S1470-2045(11)70145-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Solomon D, Davey D, Kurman R, et al. ; Bethesda 2001 Workshop. The 2001 Bethesda System: terminology for reporting results of cervical cytology. JAMA. 2002;287:2114-2119. 10.1001/jama.287.16.2114 [DOI] [PubMed] [Google Scholar]
- 9. Nayar R, Wilbur DC. The Bethesda System for Reporting Cervical Cytology: Definitions, Criteria, and Explanatory Notes. Springer; 2015. [Google Scholar]
- 10. 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;36:3583-3595. 10.1002/sim.7380 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Armitage P, Doll R. The age distribution of cancer and a multi-stage theory of carcinogenesis. Br J Cancer. 1954;8:1-12. 10.1038/bjc.1954.1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Landy R, Cheung LC, Schiffman M, et al. Challenges in risk estimation using routinely collected clinical data: the example of estimating cervical cancer risks from electronic health-records. Prev Med. 2018;111:429-435. 10.1016/j.ypmed.2017.12.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Burger EA, de Kok IMCM, Groene E, et al. Estimating the natural history of cervical carcinogenesis using simulation models: a CISNET comparative analysis. J Natl Cancer Inst. 2020;112:955-963. 10.1093/jnci/djz227 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Vink MA, Bogaards JA, van Kemenade FJ, de Melker HE, Meijer CJ, Berkhof J. Clinical progression of high-grade cervical intraepithelial neoplasia: estimating the time to preclinical cervical cancer from doubly censored national registry data. Am J Epidemiol. 2013;178:1161-1169. 10.1093/aje/kwt077 [DOI] [PubMed] [Google Scholar]
- 15. Saslow D, Solomon D, Lawson HW, et al. ; ACS-ASCCP-ASCP Cervical Cancer Guideline Committee. American Cancer Society, American Society for Colposcopy and Cervical Pathology, and American Society for Clinical Pathology screening guidelines for the prevention and early detection of cervical cancer. CA Cancer J Clin. 2012;62:147-172. 10.3322/caac.21139 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Bouvard V, Wentzensen N, Mackie A, et al. The IARC perspective on cervical cancer screening. N Engl J Med. 2021;385:1908-1918. 10.1056/NEJMsr2030640 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Perkins RB, Guido RS, Castle PE, et al. ; 2019 ASCCP Risk-Based Management Consensus Guidelines Committee. 2019 ASCCP risk-based management consensus guidelines: updates through 2023. J Low Genit Tract Dis. 2024;28:3-6. 10.1097/LGT.0000000000000788 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Wright TC Jr, Cox JT, Massad LS, Carlson J, Twiggs LB, Wilkinson EJ; 2001 ASCCP-Sponsored Consensus Workshop. 2001 Consensus guidelines for the management of women with cervical intraepithelial neoplasia. J Low Genit Tract Dis. 2003;7:154-167. 10.1097/00128360-200307000-00002 [DOI] [PubMed] [Google Scholar]
- 19. Wright TC Jr, Massad LS, Dunton CJ, Spitzer M, Wilkinson EJ, Solomon D; 2006 ASCCP-Sponsored Consensus Conference. 2006 Consensus guidelines for the management of women with abnormal cervical screening tests. J Low Genit Tract Dis 2007;11:201-222. 10.1097/LGT.0b013e3181585870 [DOI] [PubMed] [Google Scholar]
- 20. Massad LS, Einstein MH, Huh WK, et al. ; 2012 ASCCP Consensus Guidelines Conference. 2012 Updated consensus guidelines for the management of abnormal cervical cancer screening tests and cancer precursors. J Low Genit Tract Dis. 2013;17(5 suppl 1):S1-S27. 10.1097/LGT.0b013e318287d329 [DOI] [PubMed] [Google Scholar]
- 21. Gage JC, Schiffman M, Katki HA, et al. Reassurance against future risk of precancer and cancer conferred by a negative human papillomavirus test. J Natl Cancer Inst. 2014;106:dju153. 10.1093/jnci/dju153 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Dillner J, Rebolj M, Birembaut P, et al. ; Joint European Cohort Study. Long term predictive values of cytology and human papillomavirus testing in cervical cancer screening: Joint European Cohort Study. BMJ. 2008;337:a1754. 10.1136/bmj.a1754 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Liang LA, Einzmann T, Franzen A, et al. Cervical cancer screening: comparison of conventional Pap smear test, liquid-based cytology, and human papillomavirus testing as stand-alone or cotesting strategies. Cancer Epidemiol Biomarkers Prev. 2021;30:474-484. 10.1158/1055-9965.EPI-20-1003 [DOI] [PubMed] [Google Scholar]
- 24. Yao Q, Wang J, Elfström KM, Strander B, Dillner J, Sundström K. Evaluation of primary HPV-based cervical screening among older women: long-term follow-up of a randomized healthcare policy trial in Sweden. PLoS Med. 2024;21:e1004505. 10.1371/journal.pmed.1004505 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Inturrisi F, Rozendaal L, Veldhuijzen NJ, Heideman DAM, Meijer C, Berkhof J. Risk of cervical precancer among HPV-negative women in the Netherlands and its association with previous HPV and cytology results: a follow-up analysis of a randomized screening study. PLoS Med. 2022;19:e1004115. 10.1371/journal.pmed.1004115 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Kim JJ, Ortendahl J, Goldie SJ. Cost-effectiveness of human papillomavirus vaccination and cervical cancer screening in women older than 30 years in the United States. Ann Intern Med. 2009;151:538-545. 10.7326/0003-4819-151-8-200910200-00007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Kruse GR, Lykken JM, Kim EJ, et al. Provider beliefs in effectiveness and recommendations for primary HPV testing in 3 health-care systems. JNCI Cancer Spectr. 2023;7:pkac086. 10.1093/jncics/pkac086 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Castle PE, Kinney WK, Chen L, et al. ; New Mexico HPV Pap Registry Steering Committee. Adherence to national guidelines on cervical screening: a population-based evaluation from a statewide registry. J Natl Cancer Inst. 2021;114:626-630. 10.1093/jnci/djab173 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Kulkarni A, Chen L, Gockley A, et al. Patterns of cervical cancer screening follow-up in the era of prolonged screening intervals. Gynecol Oncol. 2023;175:53-59. 10.1016/j.ygyno.2023.06.007 [DOI] [PubMed] [Google Scholar]
- 30. Salingaros S, Shieh Y, Finkel ML, Polaneczky M, Korenstein D, Marti JL. Public cervical cancer screening recommendations from US cancer centers: assessing adherence to national guidelines. J Med Screen. 2024;31:201-204. 10.1177/09691413241238960 [DOI] [PubMed] [Google Scholar]
- 31. Wright JD, Chen L, Tergas AI, et al. Overuse of cervical cancer screening tests among women with average risk in the United States from 2013 to 2014. JAMA Netw Open. 2021;4:e218373. 10.1001/jamanetworkopen.2021.8373 [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 underlying this article cannot be shared publicly because they are the property of KPNC.