The study aimed to compare the performance of human papillomavirus (HPV)-based strategies to cytology for detecting cervical intraepithelial neoplasia grade 3 or worse (CIN3+) in routine program in China. The study included 50,434 women ages 30–64 years from January 2015 to December 2019, to assess four strategies: cytology with HPV triage (strategy 1), primary HPV testing with reflex cytology (strategy 2), primary HPV testing with HPV-16/18 genotyping and reflex cytology for non-16–18 high-risk HPV genotypes (strategy 3), and co-testing (strategy 4). The main outcomes were sensitivity for CIN3+ and colposcopy referral rate. Overall, the rates of HPV positivity and cytologic abnormality were 7.0% [95% confidential interval (CI), 6.8–7.2] and 3.5% (95% CI, 3.3–3.6], respectively. The sensitivity for CIN3+ were 88.5% for strategy 4, 83.2% for strategy 3, 64.6% for strategy 1, and 60.1% for strategy 2. The relative sensitivity of strategy 4 and strategy 3 compared with strategy 1 for detecting CIN3+ were 1.38 (95% CI, 1.24–1.54) and 1.29 (95% CI, 1.14–1.46), respectively. The colposcopy referral rates of strategies 4 and 3 were significantly higher than that of strategy 1 (2.4% and 2.2% vs. 1.4%, P < 0.001). In conclusion, co-testing and primary HPV testing with HPV-16/18 genotyping and reflex cytology improved the sensitivity for CIN3+ compared with cytology but increased the colposcopy referral rate. Long-term negative predicted value for HPV-negative women should be studied to determine the screening interval. Our study provides further evidence to the introduction of HPV-based strategies in China.

Prevention Relevance:

Both co-testing and primary HPV testing with HPV-16/18 genotyping and reflex cytology triage provided higher sensitivity for detecting CIN3+; however, the number of colposcopy referrals also increased compared with cytology in a routine program. It has great public health implications for the introduction of HPV-based screening strategies in China.

Cervical cancer is the fourth most commonly diagnosed cancer in women, with 604,000 new cases and 342,000 deaths worldwide (1). The World Health Organization has launched a global strategy to eliminate cervical cancer through the expansion of three-grade prevention strategies (2). Given that HPV vaccination coverage rates are low in developing countries, and it will take decades to realize the full benefits (3), cervical screening is the main approach for the coming decades.

Human papillomavirus (HPV) testing has a higher sensitivity and long-term negative predictive value for precancerous lesions compared with cytology (4, 5); therefore, many countries have switched from cytology to HPV-based strategies (6). However, a concern for HPV testing is that most infections clear spontaneously and a small fraction of women progress to persistent infection, which is likely to cause invasive cancer (7). The introduction of HPV-based strategies inevitably increases the number of colposcopy referrals leading to overdiagnosis and overtreatment (8, 9) and challenges low-resource settings (10, 11). Therefore, a key question is how to choose the appropriate triage modality for HPV-positive women to maximize the benefit-harm ratio in the screening.

Although a wide range of studies have evaluated the performance of HPV-based strategies (12–14), available health resources and characteristics of HPV prevalence are likely to affect routine practice (15–17). Furthermore, less is known about the performance of HPV-based strategies when adherence to follow-up after HPV-positive results cannot be assured. In 2008, an organized cervical cancer screening program was initiated in Beijing, which recommended that eligible women to undergo cytology screening every 2 years (18). From 2015 to 2018, co-testing with a combination of HPV testing and cytology was assessed at some pilot sites.

On the basis of the routine program, this study compared the performance of different HPV-based strategies with that of cytology to provide evidence of different triage strategies for the introduction of HPV testing in China.

Study design and participants

From January 2015 to December 2019, 50,880 women ages 30–64 years participated in the cervical cancer screening program in Beijing, P.R. China and underwent both cytology and high-risk HPV testing. We conducted a population-based, cross-sectional study to compare the performance of different HPV-based screening strategies to cytology for the detection of high-grade cervical intraepithelial neoplasia (CIN). The inclusions were as follows: women ages 30–64 years with an intact uterus and no history of CIN or invasive cancer, cervical surgery, or physical therapy. As the analyses only used anonymous data, individual informed consent was waived. This study was conducted in accordance with Declaration of Helsinki and approved by the Peking University Institutional Review Board (IRB00001052-19077).

Procedures

Women were interviewed in a private room to obtain demographic information and then underwent gynecologic examinations. All procedures were performed by local physicians. Cervical exfoliate cells were obtained in cytologic preservation medium for cytology, and brush specimens were obtained for HPV testing. A liquid-based method was used to produce the cytology slides. Cytologic diagnoses were performed by cytotechnicians based on the Bethesda 2001 classification (19). Two Chinese FDA-approved, PCR-based high-risk HPV genotyping kits (ZJ Bio-Tech; Cobas, Roche) were applied (20), which detected HPV-16/18 and other high-risk genotypes.

Women who were positive for HPV-16 or HPV-18, those with cytologic low‐grade squamous intraepithelial lesions or worse (LSIL+), and those who were positive for non-16/18 HPV genotype with atypical cells of undetermined significance (ASC‐US), were referred to colposcopy. In addition, women with clinically relevant abnormalities (abnormalities visible to the naked eyes and contact bleeding) were referred to colposcopy regardless of the screening results. Women who were positive for non-16–18 HPV genotypes with normal cytology or who were HPV negative with ASC-US were recommended to undergo repeat testing after 12 months. Given the low rate of attendance for repeat testing, approximately 15% of these women were referred to opportunistic colposcopy to adjust for verification bias.

All colposcopies and histologic evaluations were performed at the Tongzhou Maternal and Children's Hospital (Beijing, P.R. China). A colposcopy with directed four-quadrant biopsies at the squamo-columnar junction was performed (directed cervical biopsies of suspicious lesions under colposcopy), and when the colposcopy was unsatisfactory, endocervical curettage was performed. Histologic evaluation was performed by a panel of pathologists, and the results were categorized as negative or inflammation, CIN grade 1, 2, or 3, adenocarcinoma in situ (AIS) or cervical glandular intraepithelial neoplasia (AIS/CGIN), and invasive cancer. Women with CIN grade 2 or worse (CIN2+) were referred for treatment.

For women who did not attend the colposcopy referral, we reviewed the histopathologic database of the hospital and linked them with the women who were lost to follow-up using their unique Chinese identity number. We included the matched results of the histology report within 6 months of the date of initial screening.

Screening strategies

Four screening algorithms were used for evaluation, which included: (i) strategy 1, cytology with reflex HPV for ASC-US: women who were LSIL+ and ASC-US with HPV positive were referred to colposcopy and those who were ASC-US with HPV negative were to undergo repeat testing after 12 months; (ii) strategy 2, HPV testing with reflex cytology: women who were HPV positive with ASC-US+ were referred to colposcopy, whereas women who were HPV positive with normal cytology were to undergo repeat testing; (iii) strategy 3, HPV testing with HPV-16/18 genotyping and reflex cytology: women who were positive for HPV-16/18 and non-16–18 high-risk HPV genotypes with ASC-US+ were referred to colposcopy, whereas women who were positive for non-16–18 HPV genotypes with normal cytology required repeat testing; (iv) strategy 4, co-testing: women who were positive for HPV-16/18, LSIL+, or positive for non-16–18 HPV genotypes with ASC-US were referred to colposcopy, whereas women who were HPV negative with ASC-US or who were positive for non-16–18 HPV genotypes with normal cytology required repeat testing (Fig. 1).

Figure 1.

Study flowchart based on the programme. ASCUS, atypical squamous cells of undetermined significance; CIN, cervical intraepithelial neoplasia; CIN1, 2, or 3, cervical intraepithelial neoplasia grade 1, 2, or 3; HPV, human papillomavirus; LSIL, low-grade squamous intraepithelial lesions; NILM, normal intraepithelial lesion malignancy or inflammation.

Figure 1.

Study flowchart based on the programme. ASCUS, atypical squamous cells of undetermined significance; CIN, cervical intraepithelial neoplasia; CIN1, 2, or 3, cervical intraepithelial neoplasia grade 1, 2, or 3; HPV, human papillomavirus; LSIL, low-grade squamous intraepithelial lesions; NILM, normal intraepithelial lesion malignancy or inflammation.

Close modal

Statistical analysis

The main outcomes were histologically confirmed CIN2+ and CIN3+. AIS/CGIN was categorized as CIN3+. There were 16 cases with ungraded CIN, and we adopted a conservative estimation that categorized them as CIN1 to avoid an overestimation of CIN2+ and CIN3+. Furthermore, we excluded them from the dataset and performed a sensitivity analysis. To consider the women who did not attend the colposcopy, we adopted an inverse probability weighting method (21). Briefly, we accounted for this loss to follow-up in the immediate colposcopy referral by replacing raw counts with expected counts under the assumption that women who did not attend the referral had the same risk as the women who had similar abnormalities and attended the referral.

We used the following indicators to show the performance of strategies: (i) sensitivity, specificity, positive predictive value (PPV), and Youden index, representing the benefit; (ii) the number of tests performed (including primary testing, reflex testing, and repeat testing as expected), colposcopy referral rate, and number of colposcopies required to detect one case, representing the potential harm.

We used McNemar χ2 test to determine the differences in paired nominal data and the Cochran–Armitage method to assess trends. The 95% confidence intervals (CI) for sensitivity, specificity, and PPV were estimated using the Wilson score method. The normal approximation method was used to estimate the 95% CIs for other indicators, and the Clopper–Pearson method was used when the expected value in the cross-table was <5. For comparison, we estimated the relative sensitivity of HPV-based strategies versus cytology using the methods described by Kitchener and colleagues (22). We used logistic regression adjusted for age to show the association between the risk of histologic lesions and screening abnormalities.

All statistical tests were two sided with a significance level of P < 0.05. Statistical analyses were performed using the R software (version 4.0.2). Package CompareTests was used for diagnostic performance, and ggplot was used for visualization.

Data availability

All data that support the findings of this study are included in this published article and its Supplementary Data files. The datasets are available on request from the corresponding author.

Participant characteristics

A total of 50,880 women underwent co-testing and 446 were excluded (Fig. 2). Finally, the study included 50,434 women with a median age of 52 years (interquartile range: 45–57). Most women (96.6%) were ages 35 years and older (Table 1). A total of 1,199 of the 1,555 women who were referred to colposcopy attended the examination, and 35 of the 2,400 women who underwent repeat testing had their colposcopy deferred. In addition, 251 women had negative cytology and HPV testing results, and 521 women who were recommended to have repeat testing attended the immediate colposcopy. Table 1 presents the distribution of histologic outcomes in the women who underwent colposcopy. Finally, 744 women were diagnosed with histologic abnormalities, including 495 with CIN1, 132 with CIN2, 108 with CIN3, and 9 with cervical cancer.

Figure 2.

Four screening strategies framed in the study. Note: Strategy 1 represented cytology with HPV triage. Strategy 2 represented HPV testing with cytology triage. Strategy 3 represented HPV testing with HPV-16/18 genotyping and reflex cytology for other high-risk HPV types. Strategy 4 represented co-testing with HPV genotyping testing and cytology. ASC-US, atypical squamous cells of undetermined significance; CIN, cervical intraepithelial neoplasia; HPV, human papillomavirus; LSIL, low-grade squamous intraepithelial lesions; NILM, normal intraepithelial lesion malignancy or inflammation.

Figure 2.

Four screening strategies framed in the study. Note: Strategy 1 represented cytology with HPV triage. Strategy 2 represented HPV testing with cytology triage. Strategy 3 represented HPV testing with HPV-16/18 genotyping and reflex cytology for other high-risk HPV types. Strategy 4 represented co-testing with HPV genotyping testing and cytology. ASC-US, atypical squamous cells of undetermined significance; CIN, cervical intraepithelial neoplasia; HPV, human papillomavirus; LSIL, low-grade squamous intraepithelial lesions; NILM, normal intraepithelial lesion malignancy or inflammation.

Close modal
Table 1.

Characteristics of participants and distribution of histologic outcomes.

Histologic outcomes
N (%)Number of colposcopiesNegativeCIN1CIN2CIN3Invasive cancer
Total 50,434 2,006 1,262 (62.9) 495 (24.7)a 132 (6.6) 108 (5.4) 9 (0.4) 
Age group 
 30–34 1,704 (3.4) 65 34 (52.3) 19 (29.2) 6 (9.2) 6 (9.2) 0 (0) 
 35–39 5,003 (9.9) 215 108 (50.2) 64 (29.8) 24 (11.2) 19 (8.8) 0 (0) 
 40–44 5,044 (10.0) 232 136 (58.6) 69 (29.7) 18 (7.8) 8 (3.4) 1 (0.4) 
 45–49 8,365 (16.6) 391 255 (65.2) 99 (25.3) 19 (4.9) 16 (4.1) 2 (0.5) 
 50–54 10,893 (21.6) 463 286 (61.8) 121 (26.1) 30 (6.5) 22 (4.8) 4 (0.9) 
 55–59 10,862 (21.5) 382 261 (68.3) 77 (20.2) 23 (6.0) 20 (5.2) 1 (0.3) 
 60–64 8,563 (17.0) 258 182 (70.5) 46 (17.8) 12 (4.7) 17 (6.6) 1 (0.4) 
Cytology results 
 NILM 48,685 (96.5) 1,011 704 (69.6) 218 (21.6) 53 (5.2) 34 (3.4) 2 (0.2) 
 ASC-US or worse 1,749 (3.5) 995 558 (56.1) 277 (27.8) 79 (7.9) 74 (7.4) 7 (0.7) 
  ASC-US with HPV+ 423 (0.8) 317 165 (52.1) 106 (33.4) 29 (9.1) 15 (4.7) 2 (0.6) 
  ASC-US with HPV− 792 (1.6) 250 217 (86.8) 24 (9.6) 5 (2.0) 3 (1.2) 1 (0.4) 
  LSIL or worse 534 (1.1) 428 176 (41.1) 147 (34.3) 45 (10.5) 56 (13.1) 4 (0.9) 
HPV testing 
 HPV negative 46,907 (93.0) 618 524 (84.8) 70 (11.3) 9 (1.5) 12 (1.9) 3 (0.5) 
 HPV positive 3,527 (7.0) 1,388 738 (53.2) 425 (30.6) 123 (8.9) 96 (6.9) 6 (0.4) 
  16/18+ & ASC-US+ 227 (0.5) 187 66 (35.3) 50 (26.7) 23 (12.3) 45 (24.1) 3 (1.6) 
  16/18+ & cytology− 598 (1.2) 454 277 (61.0) 116 (25.6) 35 (7.7) 25 (5.5) 1 (0.2) 
  Other hrHPV+ & ASC-US+ 499 (1.0) 381 165 (43.3) 159 (41.7) 38 (10.0) 17 (4.5) 2 (0.5) 
  Other hrHPV+ & cytology− 1,966 (3.9) 282 179 (63.5) 81 (28.7) 17 (6.0) 5 (1.8) 0 (0) 
Histologic outcomes
N (%)Number of colposcopiesNegativeCIN1CIN2CIN3Invasive cancer
Total 50,434 2,006 1,262 (62.9) 495 (24.7)a 132 (6.6) 108 (5.4) 9 (0.4) 
Age group 
 30–34 1,704 (3.4) 65 34 (52.3) 19 (29.2) 6 (9.2) 6 (9.2) 0 (0) 
 35–39 5,003 (9.9) 215 108 (50.2) 64 (29.8) 24 (11.2) 19 (8.8) 0 (0) 
 40–44 5,044 (10.0) 232 136 (58.6) 69 (29.7) 18 (7.8) 8 (3.4) 1 (0.4) 
 45–49 8,365 (16.6) 391 255 (65.2) 99 (25.3) 19 (4.9) 16 (4.1) 2 (0.5) 
 50–54 10,893 (21.6) 463 286 (61.8) 121 (26.1) 30 (6.5) 22 (4.8) 4 (0.9) 
 55–59 10,862 (21.5) 382 261 (68.3) 77 (20.2) 23 (6.0) 20 (5.2) 1 (0.3) 
 60–64 8,563 (17.0) 258 182 (70.5) 46 (17.8) 12 (4.7) 17 (6.6) 1 (0.4) 
Cytology results 
 NILM 48,685 (96.5) 1,011 704 (69.6) 218 (21.6) 53 (5.2) 34 (3.4) 2 (0.2) 
 ASC-US or worse 1,749 (3.5) 995 558 (56.1) 277 (27.8) 79 (7.9) 74 (7.4) 7 (0.7) 
  ASC-US with HPV+ 423 (0.8) 317 165 (52.1) 106 (33.4) 29 (9.1) 15 (4.7) 2 (0.6) 
  ASC-US with HPV− 792 (1.6) 250 217 (86.8) 24 (9.6) 5 (2.0) 3 (1.2) 1 (0.4) 
  LSIL or worse 534 (1.1) 428 176 (41.1) 147 (34.3) 45 (10.5) 56 (13.1) 4 (0.9) 
HPV testing 
 HPV negative 46,907 (93.0) 618 524 (84.8) 70 (11.3) 9 (1.5) 12 (1.9) 3 (0.5) 
 HPV positive 3,527 (7.0) 1,388 738 (53.2) 425 (30.6) 123 (8.9) 96 (6.9) 6 (0.4) 
  16/18+ & ASC-US+ 227 (0.5) 187 66 (35.3) 50 (26.7) 23 (12.3) 45 (24.1) 3 (1.6) 
  16/18+ & cytology− 598 (1.2) 454 277 (61.0) 116 (25.6) 35 (7.7) 25 (5.5) 1 (0.2) 
  Other hrHPV+ & ASC-US+ 499 (1.0) 381 165 (43.3) 159 (41.7) 38 (10.0) 17 (4.5) 2 (0.5) 
  Other hrHPV+ & cytology− 1,966 (3.9) 282 179 (63.5) 81 (28.7) 17 (6.0) 5 (1.8) 0 (0) 

Note: 16 women detected ungraded cervical intraepithelial neoplasia which were classified into CIN1.

Abbreviations: AGC, atypical glandular cells; AIS, adenocarcinoma in situ; ASC-H, atypical squamous cell, cannot exclude high-grade squamous intraepithelial lesion; ASCUS, atypical squamous cells of undetermined significance; CIN 1, 2, or 3, cervical intraepithelial neoplasia grade 1, 2, or 3; HPV, human papillomavirus; HSIL, high-grade squamous intraepithelial lesion; LSIL, low-grade squamous intraepithelial lesions; NILM, normal intraepithelial lesion malignancy or inflammation.

Prevalence of screening abnormalities and lesion detection rates

The rate of HPV positivity was 7.0% (95% CI, 6.8%–7.2%), and the rate of HPV-16/18 positivity was 1.6% (1.5%–1.8%). The abnormal cytology rate was 3.5% (3.3%–3.6%). As shown in Fig. 3 and Supplementary Table S1, the HPV-positive rate remained stable among women ages 30–59 years and decreased in women ages 60–64 (Ptrend = 0.02). The highest abnormal cytology rate was observed in women ages 30–34 years (5.7%, 95% CI, 4.7%–6.9%). The proportion of women who were HPV negative with LSIL+ was 0.3% (0.2%–0.4%), and the proportion was relatively higher in women ages < 45 years. The proportion of women who were HPV positive with normal cytology was 5.4% (5.2%–5.6%), accounting for 77% of HPV-positive women. In combination with HPV genotyping, the proportion of women who were positive for HPV-16/18 with normal cytology was 1.2% (1.1%–1.3%), whereas the proportion of women who were positive for non-16/-18 HPV genotypes with normal cytology was 3.9% (3.7%–4.1%).

Figure 3.

Prevalence of HPV positivity and cytologic abnormality by age groups. A, Prevalence of HPV positivity. B, Prevalence of cytologic abnormality. C, Proportions of different screening abnormalities. D, Proportions of different screening abnormalities incorporating HPV−16/18. Note: + represented HPV positive; − represented HPV negative. ASC-US, atypical squamous cells of undetermined significance; hrHPV, high-risk human papillomavirus; LSIL, low-grade squamous intraepithelial lesions; NILM, normal intraepithelial lesion malignancy or inflammation.

Figure 3.

Prevalence of HPV positivity and cytologic abnormality by age groups. A, Prevalence of HPV positivity. B, Prevalence of cytologic abnormality. C, Proportions of different screening abnormalities. D, Proportions of different screening abnormalities incorporating HPV−16/18. Note: + represented HPV positive; − represented HPV negative. ASC-US, atypical squamous cells of undetermined significance; hrHPV, high-risk human papillomavirus; LSIL, low-grade squamous intraepithelial lesions; NILM, normal intraepithelial lesion malignancy or inflammation.

Close modal

Figure 4 shows the detection rates of CIN2+ and CIN3+ in women with different screening abnormalities. The detection rate of CIN3+ was significantly higher in women who were positive for HPV-16/18 (OR = 7.98; 95% CI, 3.27–26.39; P < 0.001) compared with that in women who were HPV negative with ASC-US (reference group). For HPV-16/18–positive women, the detection rate of CIN3+ was significantly higher in those who were LSIL+ (38.0%, 28.6%–48.3%), ASC-US (11.5%, 5.9%–20.6%), and those with normal cytology (5.7%, 3.8%–8.4%) than in reference group. The detection rate of CIN3+ was 5.1% (95% CI, 2.1–11.3) in women who were HPV negative and LSIL+, which was higher than that in reference group, though this was not statistically significant (OR = 3.32; 0.93–13.23; P = 0.067). For those who were not positive for non-16–18 HPV genotypes, the detection rate of CIN3+ was 5.0% (3.1%–7.8%) for those women who were ASC-US+, whereas the detection rate in women who were not positive for non-16–18 HPV genotypes and normal cytology was 1.8% (0.7%–4.3%). Similar results were obtained for the detection rate of CIN2+ and for the analysis stratified by age group (Supplementary Table S2).

Figure 4.

Detection of CIN2+ and CIN3+ among women with abnormal screening results. Note: OR was estimated by the logistic regressions using CIN2+ or CIN3+ as dependent variable and screened abnormality classifications as independent variables. The reference group was women screened with HPV- and ASC-US. N represented the number of screening abnormalities, and n represented the number of CIN2+ or CIN3+. The point and error bar represented the estimation of OR and 95% CI. ASC-US, atypical squamous cells of undetermined significance; CI, confidence interval; CIN, cervical intraepithelial neoplasia grade; hrHPV, high-risk human papillomavirus; LSIL, low-grade squamous intraepithelial lesions; NILM, normal intraepithelial lesion malignancy or inflammation; OR, odd ratio.

Figure 4.

Detection of CIN2+ and CIN3+ among women with abnormal screening results. Note: OR was estimated by the logistic regressions using CIN2+ or CIN3+ as dependent variable and screened abnormality classifications as independent variables. The reference group was women screened with HPV- and ASC-US. N represented the number of screening abnormalities, and n represented the number of CIN2+ or CIN3+. The point and error bar represented the estimation of OR and 95% CI. ASC-US, atypical squamous cells of undetermined significance; CI, confidence interval; CIN, cervical intraepithelial neoplasia grade; hrHPV, high-risk human papillomavirus; LSIL, low-grade squamous intraepithelial lesions; NILM, normal intraepithelial lesion malignancy or inflammation; OR, odd ratio.

Close modal

Performance of HPV-based strategies versus cytology

Table 2 shows the performance of different HPV-based strategies compared with cytology for detecting CIN3+. Strategy 1 had a sensitivity of 64.6% (95% CI, 55.9%–72.4%) for CIN3+, which was significantly lower than that of strategy 4 (88.5%; 95% CI, 81.3%–93.2%; P < 0.001) and strategy 3 (83.2%; 75.3%–88.9%; P < 0.001). Strategy 4 (1.38, 1.24–1.54) and strategy 3 (1.29, 1.14–1.46) had a greater relative sensitivity compared with strategy 1 for the detection of CIN3+. Conversely, the specificities for CIN3+ were lower in strategies 4 (97.7%) and 3 (98.0%) than in strategy 1 (98.8%, P < 0.001 for both). Strategy 2 had a lower sensitivity (60.1%, P = 0.013) but a higher specificity (98.8%, P < 0.001) for CIN3+ compared with strategy 1. Youden index for the detection of CIN3+ was highest for strategy 4 (0.862) and lowest for strategy 2 (0.589). The PPV for CIN3+ in strategies 3 and 4 was 8.6% (7.1%–10.3%) and 8.2% (6.9%–9.8%), respectively, both of which were lower than that in strategy 1 (10.5%, 8.6%–12.9%) though the difference was not statistically significant. Similar results were obtained for the detection of CIN2+. The results of the sensitivity analysis were consistent with the main results (Supplementary Table S3).

Table 2.

Comparison of HPV-based strategies and cytology-based strategy for the detection of CIN2+ and CIN3+.

Screening strategiesSensitivityRelative sensitivity (95% CI)SpecificityRelative specificity (95% CI)Youden indexPPV (95% CI)Tests performedColposcopy referral, %NNR
CIN2+ 
Strategy 1 58.3 (52.5–63.9) Reference 98.9 (98.8–98.9) Reference 0.572 19.8 (17.2–22.7) 52,161 1.4 (1.3–1.5) 5.0 
Strategy 2 55.7 (50.0–61.4) 0.93 (0.89–0.97) 99.0 (98.9–99.0) 1.003 (1.002–1.003) 0.547 20.4 (17.7–23.5) 56,051 1.3 (1.2–1.4) 4.6 
Strategy 3 81.7 (76.4–86.0) 1.40 (1.27–1.54) 98.2 (98.1–98.2) 0.992 (0.991–0.993) 0.799 17.5 (15.5–19.8) 54,628 2.2 (2.1–2.3) 5.5 
Strategy 4 85.5 (80.5–89.4) 1.47 (1.35–1.60) 97.9 (97.9–98.0) 0.989 (0.989–0.990) 0.844 16.5 (14.6–18.6) 103,152 2.4 (2.3–2.6) 5.8 
CIN3+ 
Strategy 1 64.6 (55.9–72.4) Reference 98.8 (98.7–98.8) Reference 0.634 10.5 (8.6–12.9) 52,161 1.4 (1.3–1.5) 9.4 
Strategy 2 60.1 (51.5–68.3) 0.91 (0.85–0.97) 98.8 (98.8–98.9) 1.003 (1.003–1.004) 0.589 10.6 (8.6–13.0) 56,051 1.3 (1.2–1.4) 8.8 
Strategy 3 83.2 (75.3–88.9) 1.29 (1.14–1.46) 98.0 (98.0–98.0) 0.992 (0.991–0.993) 0.812 8.6 (7.1–10.3) 54,628 2.2 (2.1–2.3) 11.2 
Strategy 4 88.5 (81.3–93.2) 1.38 (1.24–1.54) 97.7 (97.7–97.8) 0.989 (0.988–0.990) 0.862 8.2 (6.9–9.8) 103,152 2.4 (2.3–2.6) 11.7 
Screening strategiesSensitivityRelative sensitivity (95% CI)SpecificityRelative specificity (95% CI)Youden indexPPV (95% CI)Tests performedColposcopy referral, %NNR
CIN2+ 
Strategy 1 58.3 (52.5–63.9) Reference 98.9 (98.8–98.9) Reference 0.572 19.8 (17.2–22.7) 52,161 1.4 (1.3–1.5) 5.0 
Strategy 2 55.7 (50.0–61.4) 0.93 (0.89–0.97) 99.0 (98.9–99.0) 1.003 (1.002–1.003) 0.547 20.4 (17.7–23.5) 56,051 1.3 (1.2–1.4) 4.6 
Strategy 3 81.7 (76.4–86.0) 1.40 (1.27–1.54) 98.2 (98.1–98.2) 0.992 (0.991–0.993) 0.799 17.5 (15.5–19.8) 54,628 2.2 (2.1–2.3) 5.5 
Strategy 4 85.5 (80.5–89.4) 1.47 (1.35–1.60) 97.9 (97.9–98.0) 0.989 (0.989–0.990) 0.844 16.5 (14.6–18.6) 103,152 2.4 (2.3–2.6) 5.8 
CIN3+ 
Strategy 1 64.6 (55.9–72.4) Reference 98.8 (98.7–98.8) Reference 0.634 10.5 (8.6–12.9) 52,161 1.4 (1.3–1.5) 9.4 
Strategy 2 60.1 (51.5–68.3) 0.91 (0.85–0.97) 98.8 (98.8–98.9) 1.003 (1.003–1.004) 0.589 10.6 (8.6–13.0) 56,051 1.3 (1.2–1.4) 8.8 
Strategy 3 83.2 (75.3–88.9) 1.29 (1.14–1.46) 98.0 (98.0–98.0) 0.992 (0.991–0.993) 0.812 8.6 (7.1–10.3) 54,628 2.2 (2.1–2.3) 11.2 
Strategy 4 88.5 (81.3–93.2) 1.38 (1.24–1.54) 97.7 (97.7–97.8) 0.989 (0.988–0.990) 0.862 8.2 (6.9–9.8) 103,152 2.4 (2.3–2.6) 11.7 

Note: There were 50,197 women in the analysis and 237 women who did not have HPV genotyping results were excluded. Strategy 1 represented primary cytology test with HPV triage for ASC-US. Strategy 2 represented primary HPV testing with cytology triage. Strategy 3 represented HPV testing with HPV-16/18 genotyping and reflex cytology for non-16–18 high-risk HPV types. Strategy 4 represented co-testing with HPV genotyping testing and cytology.

Abbreviations: ASC-US, atypical squamous cells of undetermined significance; CI, confidence interval; CIN, cervical intraepithelial neoplasia grade; HPV, human papillomavirus; NNR, the number of colposcopies required to detect 1 CIN2+ or CIN3+; PPV, positive predictive value.

Strategy 4 was associated with the highest colposcopy referral rate (2.4%; 95% CI, 2.3%–2.6%) and number of colposcopies performed to detect 1 CIN3+ case (11.7). Furthermore, the number of tests performed using strategy 4 was approximately 2-fold greater than that of strategy 1. Strategy 2 had the lowest colposcopy referral rate (1.3%, 1.2%–1.4%); however, approximately one-third of CIN3+ cases were missed at the initial screening. Strategy 3 had a higher colposcopy referral rate (2.2%, 2.1%–2.3%) than strategy 1 and required more colposcopies to detect 1 CIN3+ case; however, the number of tests performed was comparable.

The study showed that both co-testing and primary HPV testing using HPV-16/18 genotyping triage significantly increased not only the sensitivity for CIN3+ but also the number of colposcopy referrals compared with cytology. However, the number of tests performed with co-testing was 2-fold greater than that in cytology, which reduces its applicability in low-resource settings. Primary HPV testing with reflex cytology had a similar sensitivity and number of colposcopy referrals compared with cytology; however, approximately one-third of the CIN3+ cases were missed. To our knowledge, this is the first study to evaluate the performance of different HPV-based strategies based on a real-world screening program in China.

The rate of high-risk HPV in this study was 7.0%, which was lower than anticipated, as a previous population-based study found a high-risk HPV-positive rate of 11% (23). In addition, the detection rates for CIN2+ and CIN3+ in our study were lower than those reported in a nationwide study (24). This difference may be explained by the low number of women ages <45 years who participated in the program and the high coverage of previous screenings in the population. Moreover, the absence of random biopsies on the four quadrants of the cervix may explain the low rate of detection of lesions. Nonetheless, the HPV-positivity rate was 2-fold higher than the cytologic abnormality rate, indicating a potentially higher rate of colposcopy referrals. In HPV-positive women, more than 70% had normal cytology and might have had a spontaneous infection that would not progress to CIN3 or cancer. Furthermore, the detection rate of CIN 3+ in women who were HPV positive with normal cytology was 4.2%, which was higher than the risk threshold (4.0%) for immediate colposcopy recommended by the ASCCP guidelines (25). This suggests that appropriately triaging HPV-positive women is essential after the introduction of HPV testing in this program.

Our study showed that co-testing not only significantly increased the sensitivity for CIN3+ compared with cytology, but also provided more stratification for the risk of precancerous lesions or cancer, which allowed for the appropriate management of women with different screening results (26, 27). However, the method has some disadvantages. First, the number of tests performed with co-testing increased to nearly two times more than that with cytology and primary HPV-based strategies, and resource-poor settings may not be able to afford this cost (28, 29). Second, co-testing requires more cytology tests for HPV-negative women than other HPV-based strategies, which is still a controversial issue in population-based screening (30, 31). Our study showed that the detection rate of CIN3+ in women who were HPV negative and LSIL+ was 5.1%, suggesting that co-testing could be the clinically preferred choice given the greater protection of HPV-negative women. Nonetheless, the proportion of women who were HPV negative and LSIL+ accounted for only 0.3% of the population. Therefore, the necessity of cytology for HPV-negative women should be considered, and the cost-benefit ratio associated with the application of co-testing for population-based programs need further studies.

Primary HPV testing with reflex cytology has been the most common HPV-based strategy in Western countries, as it provides higher sensitivity and acceptable colposcopy referral rates compared with cytology (5, 32); however, the benefits were not observed in our study. The lower sensitivity for CIN3+ in HPV testing with reflex cytology mainly resulted from the loss of CIN3+ who were HPV-positive with normal cytology. These cases would not be referred to immediate colposcopy but would require repeat testing at 12-month intervals. However, longer screening intervals may lead to lower adherence to screening timeframes, particularly in low-resource settings (33). This suggests that adequate follow-up for HPV-positive women with normal cytology is essential according to the timeframe. In addition, the main aim of screening was to detect more high-grade lesions, although the number of colposcopy referrals was lower in primary HPV testing with reflex cytology.

HPV testing with HPV-16/18 genotyping triage had a much higher sensitivity for CIN3+ compared with cytology with HPV triage, which is consistent with the ATHENA (13) and HERMES studies (14). In Mexico, HPV testing with HPV-16/18 genotyping had higher sensitivity than cytology or HPV testing with cytology triage (12). Furthermore, HPV-16/18 genotyping decreased the proportion of women who required repeat testing and could potentially be lost to follow-up, and provided the benefit of immediate CIN3+ treatment, thus preventing misdiagnoses of CIN3+. Moreover, the detection rate of CIN3+ in women who were positive for HPV-16/18 was 11.5%. Even in women who were HPV-16/18 positive with normal cytology, the detection rate of CIN3+ was higher than the risk threshold for immediate colposcopy (5.7% vs. 4.0%; ref. 25). These results demonstrate that HPV-16/18 genotyping triage should be integrated into the HPV-based strategies to improve the sensitivity and efficiency.

The number of colposcopy referrals is usually used as a surrogate to measure potential harm caused by the introduction of HPV-based screening. Unnecessary referrals to colposcopy of women with low risk may lead to overdiagnosis (10, 11), and thus overtreatment (e.g., Loop Electrosurgical Excision Procedure [LEEP]), which may increase the risk of premature pregnancy (34). Our study showed that both co-testing and HPV testing with HPV-16/18 genotyping triage increased the number of colposcopy referrals by approximately 50% and increased the number of colposcopies to detect 1 CIN3+ by 30% compared with cytology. Nonetheless, potential harm should not be given more concern than the concurrent benefit. This is because cytology usually results in immediate colposcopy referrals to women with abnormal cytology (ASC-US+) rather than HPV triage or repeated tests for ASC-US in China, which adopted the “screen-to-treat” approach to increase sensitivity (18). This means that 3.5% of women would be referred to colposcopy when using cytology alone. Hence, colposcopy referrals would not increase when switching to HPV-based strategies.

Our results provide further evidence of the benefits and costs of HPV-based strategies in a real-world screening program. The sample size of our study was large enough to assess the capacity for different HPV-based strategies to detect CIN3+ and provide robust estimations. However, some limitations should be discussed. First, most women had previously been screened; therefore, the risk of CIN3+ was low in the population. Hence, our results may reflect the effect of multiple rounds of screening and should be interpreted with caution. Second, not all women were confirmed by colposcopy, except for a small fraction of opportunistic colposcopies. Hence, we could not evaluate the false negative value because it was difficult to refer women with negative test results to colposcopy and biopsy. Furthermore, the study was based on a single round of screening. We could therefore not estimate the long-term negative predicted values of HPV-based strategies for the risk of CIN3+.

In conclusion, both co-testing and primary HPV testing with HPV-16/18 genotyping and reflex cytology triage provided higher sensitivity for detecting CIN3+; however, the number of colposcopy referrals also increased compared with cytology in a routine program. The results provide real-world evidence to support the introduction of HPV-based strategies in China. Further studies are needed to investigate the long-term negative predicted values of HPV-based strategies to determine the appropriate screening interval.

H. Bao reports grants from National Natural Science Foundation of China during the conduct of the study. No disclosures were reported by the other authors.

S. Wang: Data curation, formal analysis, visualization, methodology, writing–original draft, writing–review and editing. L. Li: Resources, data curation, investigation, project administration, writing–review and editing. J. Yang: Resources, data curation, investigation, project administration, writing–review and editing. N. Han: Resources, data curation, investigation, project administration, writing–review and editing. H. Bao: Conceptualization, funding acquisition, methodology, project administration, writing–review and editing. H.-J. Wang: Resources, supervision, methodology, project administration, writing–review and editing.

We thank all research staff from institutes for maternal and child health care in Longzhu district for their hard work in the implementation of the programme and data collection. We wish to thank all experts from the technical group of the programme for the quality control. This study was supported by grants from National Natural Science Foundation of China (81903328). The funders had no role in study design, data collection and interpretation, or the decision to submit the article for publication.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Sung
H
,
Ferlay
J
,
Siegel
RL
,
Laversanne
M
,
Soerjomataram
I
,
Jemal
A
, et al
Global Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
.
CA Cancer J Clin
2021
;
71
:
209
49
.
2.
Das
M
. 
WHO launches strategy to accelerate elimination of cervical cancer
.
Lancet Oncol
2021
;
22
:
20
1
.
3.
Bruni
L
,
Diaz
M
,
Barrionuevo-Rosas
L
,
Herrero
R
,
Bray
F
,
Bosch
FX
, et al
Global estimates of human papillomavirus vaccination coverage by region and income level: a pooled analysis
.
Lancet Glob Health
2016
;
4
:
e453
63
.
4.
Ronco
G
,
Dillner
J
,
Elfström
KM
,
Tunesi
S
,
Snijders
PJ
,
Arbyn
M
, et al
Efficacy of HPV-based screening for prevention of invasive cervical cancer: follow-up of four European randomised controlled trials
.
Lancet
2014
;
383
:
524
32
.
5.
Ogilvie
GS
,
van Niekerk
D
,
Krajden
M
,
Smith
LW
,
Cook
D
,
Gondara
L
, et al
Effect of screening with primary cervical HPV testing vs cytology testing on high-grade cervical intraepithelial neoplasia at 48 months: the HPV FOCAL Randomized Clinical Trial
.
JAMA
2018
;
320
:
43
52
.
6.
Maver
PJ
,
Poljak
M
. 
Primary HPV-based cervical cancer screening in Europe: implementation status, challenges, and future plans
.
Clin Microbiol Infect
2020
;
26
:
579
83
.
7.
de Sanjosé
S
,
Brotons
M
,
Pavón
MA
. 
The natural history of human papillomavirus infection
.
Best Pract Res Clin Obstet Gynaecol
2018
;
47
:
2
13
.
8.
Ronco
G
,
Giorgi-Rossi
P
,
Carozzi
F
,
Confortini
M
,
Dalla Palma
P
,
Del Mistro
A
, et al
Efficacy of human papillomavirus testing for the detection of invasive cervical cancers and cervical intraepithelial neoplasia: a randomised controlled trial
.
Lancet Oncol
2010
;
11
:
249
57
.
9.
Ogilvie
GS
,
Krajden
M
,
van Niekerk
DJ
,
Martin
RE
,
Ehlen
TG
,
Ceballos
K
, et al
Primary cervical cancer screening with HPV testing compared with liquid-based cytology: results of round 1 of a randomised controlled trial– the HPV FOCAL study
.
Br J Cancer
2012
;
107
:
1917
24
.
10.
Velentzis
LS
,
Caruana
M
,
Simms
KT
,
Lew
JB
,
Shi
JF
,
Saville
M
, et al
How will transitioning from cytology to HPV testing change the balance between the benefits and harms of cervical cancer screening? Estimates of the impact on cervical cancer, treatment rates and adverse obstetric outcomes in Australia, a high vaccination coverage country
.
Int J Cancer
2017
;
141
:
2410
22
.
11.
Tebeu
PM
,
Fokom-Domgue
J
,
Crofts
V
,
Flahaut
E
,
Catarino
R
,
Untiet
S
, et al
Effectiveness of a two-stage strategy with HPV testing followed by visual inspection with acetic acid for cervical cancer screening in a low-income setting
.
Int J Cancer
2015
;
136
:
E743
50
.
12.
Torres-Ibarra
L
,
Cuzick
J
,
Lorincz
AT
,
Spiegelman
D
,
Lazcano-Ponce
E
,
Franco
EL
, et al
Comparison of HPV-16 and HPV-18 genotyping and cytological testing as triage testing within human papillomavirus-based screening in Mexico
.
JAMA Netw Open
2019
;
2
:
e1915781
.
13.
Cox
JT
,
Castle
PE
,
Behrens
CM
,
Sharma
A
,
Wright
TC
 Jr
,
Cuzick
J
. 
Comparison of cervical cancer screening strategies incorporating different combinations of cytology, HPV testing, and genotyping for HPV 16/18: results from the ATHENA HPV study
.
Am J Obstet Gynecol
2013
;
208
:
184.e1
11
.
14.
Chatzistamatiou
K
,
Moysiadis
T
,
Moschaki
V
,
Panteleris
N
,
Agorastos
T
. 
Comparison of cytology, HPV DNA testing and HPV 16/18 genotyping alone or combined targeting to the more balanced methodology for cervical cancer screening
.
Gynecol Oncol
2016
;
142
:
120
7
.
15.
Bruni
L
,
Diaz
M
,
Castellsagué
X
,
Ferrer
E
,
Bosch
FX
,
de Sanjosé
S
. 
Cervical human papillomavirus prevalence in 5 continents: meta-analysis of 1 million women with normal cytological findings
.
J Infect Dis
2010
;
202
:
1789
99
.
16.
Franco
EL
,
Cuzick
J
,
Hildesheim
A
,
de Sanjosé
S
. 
Chapter 20: Issues in planning cervical cancer screening in the era of HPV vaccination
.
Vaccine
2006
;
24
;
171
7
.
17.
Torres-Poveda
K
,
Ruiz-Fraga
I
,
Madrid-Marina
V
,
Chavez
M
,
Richardson
V
. 
High risk HPV infection prevalence and associated cofactors: a population-based study in female ISSSTE beneficiaries attending the HPV screening and early detection of cervical cancer program
.
BMC Cancer
2019
;
19
:
1205
.
18.
Han
L
,
Chang
X
,
Song
P
,
Gao
L
,
Zhang
Y
,
An
L
, et al
An on-going study of three different cervical cancer screening strategies based on primary healthcare facilities in Beijing China
.
J Infect Public Health
2020
;
13
:
577
83
.
19.
Solomon
D
,
Davey
D
,
Kurman
R
,
Moriarty
A
,
O'Connor
D
,
Prey
M
, et al
The 2001 Bethesda System: terminology for reporting results of cervical cytology
.
JAMA
2002
;
287
:
2114
9
.
20.
Xue
P
,
Gao
LL
,
Yin
J
,
Han
LL
,
Zhao
J
,
Li
L
, et al
A direct comparison of four high-risk human papillomavirus tests versus the cobas test: Detecting CIN2+ in low-resource settings
.
J Med Virol
2019
;
91
:
1342
50
.
21.
Katki
HA
,
Li
Y
,
Edelstein
DW
,
Castle
PE
. 
Estimating the agreement and diagnostic accuracy of two diagnostic tests when one test is conducted on only a subsample of specimens
.
Stat Med
2012
;
31
:
436
48
.
22.
Kitchener
HC
,
Blanks
R
,
Cubie
H
,
Desai
M
,
Dunn
G
,
Legood
R
, et al
MAVARIC - a comparison of automation-assisted and manual cervical screening: a randomised controlled trial
.
Health Technol Assess
2011
;
15
:
iii–iv, ix–xi,1–170
.
23.
Bao
HL
,
Jin
C
,
Wang
S
,
Song
Y
,
Xu
ZY
,
Yan
XJ
, et al
Prevalence of cervicovaginal human papillomavirus infection and genotypes in the pre-vaccine era in China: a nationwide population-based study
.
J Infect
2021
;
82
:
75
83
.
24.
Zhang
J
,
Zhao
Y
,
Dai
Y
,
Dang
L
,
Ma
L
,
Yang
C
, et al
Effectiveness of high-risk human papillomavirus testing for cervical cancer screening in China: a multicenter, open-label, randomized clinical trial
.
JAMA Oncol
2021
;
7
:
263
70
.
25.
Egemen
D
,
Cheung
LC
,
Chen
X
,
Demarco
M
,
Perkins
RB
,
Kinney
W
, et al
Risk estimates supporting the 2019 ASCCP risk-based management consensus guidelines
.
J Low Genit Tract Dis
2020
;
24
:
132
43
.
26.
Massad
LS
,
Einstein
MH
,
Huh
WK
,
Katki
HA
,
Kinney
WK
,
Schiffman
M
, et al
2012 updated consensus guidelines for the management of abnormal cervical cancer screening tests and cancer precursors
.
Obstet Gynecol
2013
;
121
:
829
46
.
27.
Perkins
RB
,
Guido
RS
,
Castle
PE
,
Chelmow
D
,
Einstein
MH
,
Garcia
F
, et al
2019 ASCCP risk-based management consensus guidelines for abnormal cervical cancer screening tests and cancer precursors
.
J Low Genit Tract Dis
2020
;
24
:
102
31
.
28.
Sahasrabuddhe
VV
,
Parham
GP
,
Mwanahamuntu
MH
,
Vermund
SH
. 
Cervical cancer prevention in low- and middle-income countries: feasible, affordable, essential
.
Cancer Prev Res
2012
;
5
:
11
7
.
29.
Wright
TC
,
Stoler
MH
,
Behrens
CM
,
Sharma
A
,
Zhang
G
,
Wright
TL
. 
Primary cervical cancer screening with human papillomavirus: end of study results from the ATHENA study using HPV as the first-line screening test
.
Gynecol Oncol
2015
;
136
:
189
97
.
30.
Katki
HA
,
Schiffman
M
,
Castle
PE
,
Fetterman
B
,
Poitras
NE
,
Lorey
T
, et al
Benchmarking CIN 3+ risk as the basis for incorporating HPV and Pap cotesting into cervical screening and management guidelines
.
J Low Genit Tract Dis
2013
;
17
:
S28
35
.
31.
Demarco
M
,
Lorey
TS
,
Fetterman
B
,
Cheung
LC
,
Guido
RS
,
Wentzensen
N
, et al
Risks of CIN 2+, CIN 3+, and cancer by cytology and human papillomavirus status: the foundation of risk-based cervical screening guidelines
.
J Low Genit Tract Dis
2017
;
21
:
261
7
.
32.
Leinonen
MK
,
Nieminen
P
,
Lönnberg
S
,
Malila
N
,
Hakama
M
,
Pokhrel
A
, et al
Detection rates of precancerous and cancerous cervical lesions within one screening round of primary human papillomavirus DNA testing: prospective randomised trial in Finland
.
BMJ
2012
;
345
:
e7789
.
33.
Gago
J
,
Paolino
M
,
Arrossi
S
. 
Factors associated with low adherence to cervical cancer follow-up retest among HPV+/cytology negative women: a study in programmatic context in a low-income population in Argentina
.
BMC Cancer
2019
;
19
:
367
.
34.
Heinonen
A
,
Gissler
M
,
Riska
A
,
Paavonen
J
,
Tapper
AM
,
Jakobsson
M
. 
Loop electrosurgical excision procedure and the risk for preterm delivery
.
Obstet Gynecol
2013
;
121
:
1063
8
.