Risk of high-grade cervical intra-epithelial neoplasia based on cytology and high-risk HPV testing at baseline and at 6-months

Authors

  • Saskia Bulk,

    1. Department of Pathology, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
    • Saskia Bulk and Nicole W.J. Bulkmans have contributed equally to this manuscript. Bulk performed the analyses and prepared the manuscript. N.W.J. Bulkmans was responsible for the data collection. J. Berkhof was responsible for the conception of the study and statistical methods. P.J.F. Snijders was responsible for the virology determinations. C.J.L.M. Meijer was the principal investigator of the project, and had access to all data and final responsibility for the decision to submit for publication. All authors participated in the preparation of the design of the study, interpretation of data, writing of the manuscript and approved the final version of the manuscript.

  • Nicole W.J. Bulkmans,

    1. Department of Pathology, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
    • Saskia Bulk and Nicole W.J. Bulkmans have contributed equally to this manuscript. Bulk performed the analyses and prepared the manuscript. N.W.J. Bulkmans was responsible for the data collection. J. Berkhof was responsible for the conception of the study and statistical methods. P.J.F. Snijders was responsible for the virology determinations. C.J.L.M. Meijer was the principal investigator of the project, and had access to all data and final responsibility for the decision to submit for publication. All authors participated in the preparation of the design of the study, interpretation of data, writing of the manuscript and approved the final version of the manuscript.

  • Johannes Berkhof,

    1. Department of Clinical Epidemiology & Biostatistics, VU University Medical Center, The Netherlands
    Search for more papers by this author
  • Lawrence Rozendaal,

    1. Department of Pathology, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • A. Joan P. Boeke,

    1. Department of General Practice, Institute for Research in Extramural Medicine, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • René H.M. Verheijen,

    1. Department of Obstetrics and Gynaecology, Division of Gynecologic Oncology, VU University Medical Center, The Netherlands
    Search for more papers by this author
  • Peter J.F. Snijders,

    1. Department of Pathology, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • Chris J.L.M. Meijer

    Corresponding author
    1. Department of Pathology, VU University Medical Center, Amsterdam, The Netherlands
    • Department of Pathology, VU University Medical Center, P.O. Box 7057, 1007 MB Amsterdam, The Netherlands
    Search for more papers by this author
    • Fax: +31-20-444-2964


  • POBASCAM study collaborators other than authors: Dr. K. V Groningen (Spaarne Ziekenhuis, Hoofddorp), Dr. W. Ruitinga (Kennemer Gasthuis, Haarlem), Dr. M.E. Boon (Leiden Cytology and Pathology Laboratory, Leiden), Dr. M. van Ballegooijen (Department of Public Health and Social Medicine, Erasmus University Rotterdam), Dr. F.J. Voorhorst and Dr. F.J. van Kemenade (Unit cytopathology, VU University Medical Center, Amsterdam).

Abstract

Adding a test for high-risk human papillomavirus (hrHPV) to cytological screening enhances the detection of high-grade cervical intraepithelial neoplasia (≥CIN2), but data are required that enable long-term evaluation of screening. We investigated the ≥CIN2 risk for women participating in population-based screening as a function of hrHPV and cytology testing results at baseline and at 6 months. We included 2,193 women aged 30–60 years participating in a population-based screening trial who received colposcopy or a repeat testing advice at baseline. The main endpoint was histologically confirmed ≥CIN2 diagnosed within 36 months. hrHPV testing was more sensitive than cytology for ≥CIN2 (relative sensitivity 1.4, 95%CI: 1.3–1.5; absolute sensitivity 94.1 and 68.0%, respectively). The 18-month ≥CIN2 risks in women with a hrHPV-positive smear and in women with abnormal cytology were similar (relative risk 0.9, 95%CI: 0.8–1.1). Women with HPV16 and/or HPV18 had a higher ≥CIN2 risk than other hrHPV-positive women irrespective of the cytological grade. Repeat testing showed that both cytological regression and viral clearance were strongly associated with a decrease in ≥CIN2 risk. Notably, women who had a double negative repeat test at 6 months had a ≥CIN2 risk of only 0.2% (95%CI: 0.0–1.1) and hrHPV-negative women with baseline borderline or mild dyskaryosis and normal cytology at 6 months had a ≥CIN2 risk of 0% (95%CI: 0.0–0.8). Using hrHPV and/or cytology testing, risk of ≥CIN2 can be assessed more accurately by repeat testing than single visit testing. Hence, when hrHPV testing is implemented, patient management with repeat testing is a promising strategy to control the number of referrals for colposcopy. © 2007 Wiley-Liss, Inc.

Invasive cervical cancer is one of the leading causes of cancer-related death in women of childbearing age worldwide.1, 2 As a preventive measure, screening by cervical cytology (i.e., the Pap test) has been shown to dramatically decrease the cervical cancer incidence and mortality.3 Another possibility is to screen for infections with high-risk human papillomavirus (hrHPV), the causative agent for cervical cancer, and combined cytology and hrHPV testing seems to be a promising strategy to improve cervical screening. Previous studies have shown that cytology combined with hrHPV testing improves the sensitivity to detect high-grade cervical lesions.4,5 The positive predictive value of a single positive hrHPV test, however, remains low for women with a normal smear or mild cytological abnormalities, and referral rates for colposcopy may increase substantially with combined testing.6 Hence, implementation of hrHPV testing needs to be preceded by an evaluation of various screening strategies using hrHPV and cytological testing.

Abbreviations:

BMD, borderline or mild dyskaryosis; >BMD, moderate dyskaryosis or worse; ≥CIN2, high-grade cervical intraepithelial neoplasia or worse, i.e., CIN2, CIN3 or invasive cancer; ≥CIN3, high-grade cervical intraepithelial neoplasia or worse, i.e., CIN3 or invasive cancer; HPV, human papillomavirus; hrHPV, high-risk human papillomavirus; POBASCAM, population based screening Amsterdam.

We investigated the risk of high-grade cervical intraepithelial neoplasia in women with hrHPV test results and cytology at baseline and at 6 months. We used data obtained from a population-based cervical screening trial. Since women in whom HPV16 or HPV18 is detected seem to have a substantially elevated risk of high-grade lesions compared with other hrHPV-positive women, and since 70% of all cases of cervical cancer are caused by HPV16 and HPV18, we evaluated risks for hrHPV-positive women with HPV16 and/or HPV18 and women positive for other high-risk types separately.7, 8, 9

Material and methods

Study population and procedures

In this study, we included all women participating in the POBASCAM (Population-Based Screening Amsterdam) trial who had received an advice to have repeat cytology at 6 and 18 months, or who had been referred for immediate colposcopy. The POBASCAM trial is a population-based double blind randomized controlled trial to evaluate the efficacy of screening using hrHPV testing in conjunction with conventional cytology (intervention group) compared with cervical screening with classical cytology (control group). All participants gave written informed consent. The design, methods and baseline results of the POBASCAM trial have been described previously.10 A flowchart of the randomization, selection and screening procedure of the POBASCAM trial is presented in Figure 1.

Figure 1.

N indicates normal cytology, BMD indicates borderline or mild dyskaryosis and >BMD indicates moderate dyskaryosis or worse. Participants in the intervention group (n = 21,996) were referred for colposcopy directed biopsy when the cytology result was moderate dyskaryosis or worse (≫BMD) irrespective of hrHPV status. Participants returned to the next screening round after 5 years if the cytological result was normal and the hrHPV test was negative. Otherwise, participants were followed both cytologically and with hrHPV testing at 6 and 18 months. Participants with BMD at baseline were referred for colposcopy at 6 months if the 6-month result was hrHPV-positive BMD, or ≫BMD irrespective of hrHPV status, and participants with hrHPV positive normal cytology at intake were referred at 6 months only for ≫BMD. All participants in the intervention group were referred at 18 months for colposcopy if the combined test suggested the presence of a cervical lesion. The participants in the control group (n = 22,106) followed the current Dutch screening guidelines. Participants were referred for colposcopy if the result was ≫BMD at intake. Participants returned to the next screening round at 5 years if the result was normal. Smears were repeated at 6 or 18 months in case the result was BMD. All participants in the control group were referred at 18 months for colposcopy if cytology was ≥BMD.

Conventional cytological smears were taken using a Cervex-Brush (Rovers Medical Devices, Oss, The Netherlands). The brush was placed in a vial containing a collection medium (i.e., 5 ml PBS and 0.5% thiomersal) for hrHPV testing.10 Cervical smears were classified according to the Dutch CISOE-A classification blinded to hrHPV status of participants.11 In short, cytological results were grouped as normal, borderline or mild dyskaryosis (BMD; translating into ASC-US/ASC-H/LSIL) and moderate dyskaryosis or worse (≫BMD; translating into HSIL).12 Detection of hrHPV was performed by GP5+/6+ PCR enzyme immunoassay, using a cocktail of 14 high-risk types, i.e., HPV16, HPV18, HPV31, HPV33, HPV35, HPV39, HPV45, HPV51, HPV52, HPV56, HPV58, HPV59, HPV66 and HPV68.13 HrHPV tests were performed in duplicate, and all hrHPV-positive samples were typed by reverse line blotting.14 Technicians performing hrHPV testing were blinded to the cytology results.

Colposcopically directed biopsies were taken for histological examination when suspected areas on the cervix were present according to standard procedures in The Netherlands, and abnormal results were classified histologically as CIN 1, 2 or 3, or invasive cancer according to international criteria.15, 16 We included all lesions diagnosed after the referral smear and within 3 years after baseline. Histology samples were read in a community setting and were not subjected to revision.10

Statistical analysis

All participants received cytological analyses (i.e., Pap tests) and hrHPV testing at baseline. Using the screening results leading to a repeat or referral advice at baseline, we defined groups based on the combinations of cytology (normal, BMD, ≫BMD) and hrHPV (positive (+)/negative (−)) test result. hrHPV positive samples were further stratified on the presence of HPV16 and/or HPV18 in the baseline sample, since these 2 types account for ∼70% of cervical cancer cases. Participants with follow-up were further stratified on cytology or hrHPV test result at 6 months. The outcome of interest was defined as a lesion of at least CIN2 (≥CIN2, i.e., CIN2, CIN3 or invasive cancer). All analyses were repeated using lesions ≥CIN3 as outcome measure. Cumulative 18-month incidences as a measure of risk of lesions ≥CIN2 were assessed using Kaplan–Meier methods and 95% CI's were calculated. The 95% CI's on the original ≥CIN2 risk scale were obtained by exponentiating the upper and lower bounds of the 95% confidence intervals constructed on the log risk scale. Specific groups were compared using log-rank testing. Reported sensitivities and specificities were adjusted for nonverification occurring because women in the control group with normal cytology were sent back to routine screening regardless of the blinded hrHPV test result and because some women were lost to follow-up.17 Because the Medical Ethics Committee did not allow recalling participants with normal cytology and a negative hrHPV test for repeat testing earlier than the regular screening interval of 5 years, we assumed women who were sent back to routine screening not to have an underlying CIN lesion. This assumption does not affect the relative sensitivity.17, 18, 19

Role of the funding source

The funding source had no involvement in study design, data collection, data analysis, data interpretation or writing the report. The corresponding author had full access to all data in the study and had final responsibility for the decision to submit for publication.

Results

Of 44,102 participants of the POBASCAM trial, we included 763 participants with normal cytology and a positive hrHPV test from the intervention group, 1,080 participants with a BMD result of whom 381 (35.3%) tested hrHPV positive and350 participants with a ≫BMD result of whom 313 (89.4%) tested hrHPV positive. Mean age was 38.5 years (range 29–60) for participants with normal cytology and a positive hrHPV test, 40.0 years (range 29–60) for participants with BMD and 37.3 years (range 29–60) for participants with ≫BMD. The nonresponse rate was 23.1% (165/713) at 6 months and 28.0% (146/522) at 18 months for participants with normal cytology. The nonresponse rates at 6 and 18 months for participants with BMD were 9.9% (37/374) and 28.8% (53/184), respectively.

In the control arm, ≥CIN2 risk for women with a baseline abnormal smear (i.e., ≥BMD) was 27% (95%CI: 24–31). In the intervention arm, ≥CIN2 risk was 26% (95%CI: 23–29) for women with a baseline positive hrHPV test, 28% (95%CI: 24–31) for women with abnormal cytology at baseline and 19% (95%CI: 17–21) for women with abnormal cytology and/or a positive hrHPV test. The relative risk of ≥CIN2 for women with a positive hrHPV result in the intervention arm compared with abnormal cytology in the control arm was 25.6/27.2 = 0.94 (95%CI: 0.8–1.1). The relative risk of ≥CIN2 for women with a positive hrHPV result and/or abnormal cytology in the intervention arm compared with abnormal cytology in the control arm was 0.71 (95%CI: 0.60–0.83). Detection rates of ≥CIN2 in women with BMD did not differ between intervention (14%, 95%CI: 11–18) and control group (13%, 95%CI: 11–17).

Because ≥CIN2 risks in women with abnormal cytology did not depend on the allocation to either intervention or control group (Fig. 1), risks were pooled in further analyses. Furthermore, only results for ≥CIN2 are discussed. Results were comparable, albeit with lower absolute risks, using lesions ≥CIN3 as outcome measure.

In Table I, separate ≥CIN2 risks are presented for strata defined by baseline cytology and hrHPV status. The lowest risk of 2.5% (95%CI: 1.5–4.2) was observed in women with BMD and a negative hrHPV test. A low ≥CIN2 risk of 5.3% (95%CI: 3.3–8.6) was also observed in women with normal cytology who were infected with hrHPV but not HPV16 and/or HPV18. In the subsets of hrHPV-positive women with normal cytology and HPV16 and/or HPV18, hrHPV-positive women with BMD, and women with ≫BMD, ≥CIN2 risks were moderate to high. Risks for HPV16-positive women only were comparable with the risk of women with HPV16 and/or 18 (data not shown).

Table I. Cumulative 18-Month Risk of Histologically Diagnosed CIN2 or Worse Stratified by Cytology and HPV Status at Baseline
Intake ≥CIN2≥CIN3
  1. + Indicates a positive hrHPV test, and − indicates a negative hrHPV test. The 16+ and/or 18+ indicates an intake hrHPV test positive for either HPV16 or HPV18, and 16− and 18− indicates an intake hrHPV test positive for types other than HPV16 or HPV18. N indicates normal cytology, BMD indicates borderline or mild dyskaryosis, and >BMD indicates moderate dyskaryosis or worse. CIN indicates cervical intra-epithelial neoplasia, and CxCa indicates cervical cancer (i.e., squamous cell carcinoma, adenocarcinoma and adenocarcinoma in situ).

 Total≥CIN2≥CIN3CxCaRisk (95%CI)Risk (95%CI)
N and HPV+7635829112 (9.7–16)6.1 (4.2–8.7)
 16+ and/or 18+262422325 (19–33)13 (8.9–19)
 16− and 18−50116615.3 (3.3–8.6)2.2 (1.0–4.8)
BMD and HPV−69915822.5 (1.5–4.2)1.8 (0.9–3.3)
BMD and HPV+38110863533 (28–38)20 (16–25)
 16+ and/or 18+1615941543 (35–52)31 (23–39)
 16− and 18−220492226 (20–33)12 (8.3–18)
>BMD and HPV−37168149 (34–66)27 (16–44)
>BMD and HPV+313246192979 (74–83)62 (56–67)
 16+ and/or 18+205169137982 (77–87)67 (61–74)
 16− and 18−108775571 (63–60)51 (42–61)

A detailed overview of the role of repeat cytology on ≥CIN2 risks is displayed in Table II, showing ≥CIN2 risks for strata defined by hrHPV and cytology at baseline, and cytology at 6 months. In women with hrHPV-positive normal cytology at baseline, cumulative risk of ≥CIN2 increased from 5.5% (95%CI: 3.3–9.2) to 79% (95%CI: 61–92) when comparing women with a second normal smear with women who had ≫BMD at 6 months. Women who had HPV16 and/or HPV18 at baseline with 2 normal smears had a risk of 13% (95%CI: 7.8–22), whereas women who tested hrHPV-positive for other high-risk types had a much lower risk of 1.2% (95%CI: 0.3–4.9). Risks in HPV16 and/or HPV18-positive women with normal cytology at baseline and BMD at 6 months were 52% (95%CI: 34–73), and 11% (95%CI: 4.2–26) for hrHPV-positive women positive for other types. In women with a hrHPV-negative BMD smear at baseline and cytological regression at 6 months (n = 485), ≥CIN2 risk was 0% (95%CI: 0.0–0.8). In contrast, women with hrHPV-positive BMD and cytological regression still had a substantial risk of ≥CIN2 of 14% (95%CI: 8.5–23). Women with ≫BMD at 6 months irrespective of baseline cytology had high ≥CIN2 risks.

Table II. Cumulative 18-Month Risk of Histologically Diagnosed CIN2 or Worse Stratified by Cytology and hrHPV at Baseline and Cytology at 6 Months
Intake6 months ≥CIN2≥CIN3
  1. + Indicates a positive hrHPV test, and − indicates a negative hrHPV test. 16+ and/or 18+ indicates an intake hrHPV test positive for either HPV16 or HPV18, and 16− and 18− indicates an intake hrHPV test positive for types other than HPV16 or HPV18.

  2. N indicates normal cytology, BMD indicates borderline or mild dyskaryosis, and >BMD indicates moderate dyskaryosis or worse. CIN indicates cervical intraepithelial neoplasia, and CxCa indicates cervical cancer (i.e., squamous cell carcinoma, adenocarcinoma and adenocarcinoma in situ). Participants without a hrHPV test result at 6 months were excluded from the analyses.

  Total≥CIN2≥CIN3CxCaRisk (95%CI)Risk (95%CI)
N and HPV+N3991445.5 (3.3–9.2)1.6 (0.6–4.2)
 BMD70168127 (17–40)17 (7.0–25)
 >BMD24191279 (62–92)50 (32–71)
 16+ and/or 18+N13512313 (7.8–22)3.3 (1.1–10)
 BMD2812652 (34–73)27 (13–51)
 >BMD15131187 (65–98)73 (50–92)
 16− and 18−N264211.2 (0.3–4.9)0.6 (0.1–4.3)
 BMD4242111 (4.2–26)5.4 (1.4–20)
 >BMD96167 (38–92)11 (1.6–57)
BMD and HPV−N4850.0 (0.0–0.8)0.0 (0.0–0.8)
 BMD814216.4 (2.7–15)4.0 (1.3–12)
 >BMD764186 (54–99)71 (39–96)
BMD and HPV+N14013414 (8.5–23)4.6 (1.7–12)
 BMD1094120138 (29–48)18 (12–27)
 >BMD483826479 (67–89)54 (41–69)
 16+ and/or 18+N5272 21 (11–30)6.1 (1.6–22)
 BMD43201546 (33–62)35 (23–51)
 >BMD282316182 (66–94)57 (40–75)
 16− and 18−N8862411 (4.8–22)3.8 (1.0–14)
 BMD6621532 (22–45)7.6 (3.2–17)
 >BMD20151075 (55–91)50 (31–73)

Table III gives a detailed presentation of the role of hrHPV clearance and displays the cumulative risk of a lesion ≥CIN2 stratified for baseline results, and hrHPV test result at 6 months. In women with hrHPV-positive normal cytology at baseline, cumulative risk of ≥CIN2 was 2.4% (95%CI: 0.6–9.1) in women testing negative for hrHPV at 6 months and 20% (95%CI: 15–25) in women who tested hrHPV-positive twice. In the group of hrHPV-positive women with normal cytology at baseline and a negative hrHPV test at 6 months, all ≥CIN2 cases diagnosed tested positive for HPV16 and/or HPV18 at baseline. For the other strata with a negative hrHPV test at 6 months, ≥CIN2 risks were low but not negligible.

Table III. Cumulative 18-Month Risk of Histologically Diagnosed CIN2 or Worse Stratified by Cytology and hrHPV at Baseline and hrHPV at 6 Months
Intake6 months ≥CIN2≥CIN3
  1. + Indicates a positive hrHPV test, and − indicates a negative hrHPV test. 16+ and/or 18+ indicates an intake hrHPV test positive for either HPV16 or HPV18, and 16− and 18− indicates an intake hrHPV test positive for types other than HPV16 or HPV18.

  2. N indicates normal cytology, BMD indicates borderline or mild dyskaryosis, and >BMD indicates moderate dyskaryosis or worse. CIN indicates cervical intra-epithelial neoplasia, and CxCa indicates cervical cancer (i.e., squamous cell carcinoma, adenocarcinoma and adenocarcinoma in situ). Participants without a cytology result at 6 months were excluded from the analyses.

  Total≥CIN2≥CIN3CxCaRisk (95%CI)Risk (95%CI)
N and HPV+HPV−16522.4 (0.6–9.1)0.0 (0.0–2.2)
 HPV+2664623120 (15–25)9.8 (6.6–14)
 16+ and/or 18+HPV−4827.4 (1.9–26)0.0 (0.0–7.4)
 HPV+1123520 36 (27–46)20 (13–30)
 16− and 18−HPV−1170.0 (0.0–3.0)0.0 (0.0–3.0)
 HPV+15411317.9 (4.5–14)2.4 (0.8–7.3)
BMD and HPV−HPV−378541.5 (0.6–3.6)1.2 (0.4–3.3)
 HPV+10110 (1.5–53)0.0 (0.0–31)
BMD and HPV+HPV−555211 (4.7–26)4.9 (1.2–19)
 HPV+1425432339 (32–48)24 (17–32)
 16+ and/or 18+HPV−112124 (6.3–70)17 (2.5–73)
 HPV+733123344 (33–57)33 (23–46)
 16− and 18−HPV−44318.2 (2.6–24)2.3 (0.3–1.5)
 HPV+6923934 (24–46)14 (7.5–25)

Finally, we evaluated the cumulative risk of a lesion ≥CIN2 based on hrHPV status and cytology at baseline and at 6 months (Table IV). Overall, the ≥CIN2 risk of women with a hrHPV-negative normal smear at 6 months (n = 522) was 0.2% (95%CI: 0.0–0.8). In women with hrHPV-positive normal cytology at baseline, cumulative risk of ≥CIN2 was 1.3% (95%CI: 0.0–2.4) in women double negative at 6 months and 41% (95%CI: 31–53) in women who tested positive for cytology and/or hrHPV at 6 months. In the group of hrHPV-positive women without HPV16 and/or HPV18 at baseline with a hrHPV-negative normal test at 6 months, no cases of ≥CIN2 were diagnosed. In women with hrHPV-negative BMD at baseline, risk was 0.0% (95%CI: 0.0–1.1) in women who had a hrHPV-negative normal smear at 6 months and 11% (95%CI: 5.1–23) in women positive for either test at 6 months. Women with BMD cytology and a positive hrHPV test at baseline that tested hrHPV-negative with normal cytology at follow-up had a risk of 0.0% (95%CI: 0.0–8.4) and women who tested positive for cytology and/or hrHPV had a risk of 48% (95%CI: 38–58).

Table IV. Cumulative 18-Month Risk of Histologically Diagnosed CIN2 or Worse Stratified by Cytology and hrHPV at Baseline and Cytology and hrHPV at 6 Months
Intake6 months ≥CIN2≥CIN3
  1. + Indicates a positive hrHPV test, and − indicates a negative hrHPV test. 16+ and/or 18+ indicates an intake hrHPV test positive for either HPV16 or HPV18, and 16− and 18− indicates an intake hrHPV test positive for types other than HPV16 or HPV18.

  2. N indicates normal cytology, BMD indicates borderline or mild dyskaryosis, and >BMD indicates moderate dyskaryosis or worse. CIN indicates cervical intra-epithelial neoplasia, and CxCa indicates cervical cancer (i.e., squamous cell carcinoma, adenocarcinoma and adenocarcinoma in situ). Participants without a cytology or hrHPV test result at 6 months were excluded from the analyses.

  Total≥CIN2≥CIN3CxCaRisk (95%CI)Risk (95%CI)
N and HPV+N and HPV−15511.3 (0.0–2.4)0.0 (0.0–2.4)
 ≥BMD and/or HPV+893419141 (31–53)24 (16–35)
 16+ and/or 18+N and HPV−4713.9 (0.6–24.3)0.0 (0.0–7.7)
 ≥BMD and/or HPV+422517 65 (50–80)45 (31–63)
 16− and 18−N and HPV−1080.0 (0.0–3.4)0.0 (0.0–3.4)
 ≥BMD and/or HPV +4792120 (11–35)5.4 (1.4–20)
BMD and HPV−N and HPV−3250.0 (0.0–1.1)0.0 (0.0–1.1)
 ≥BMD and/or HPV+566411 (5.1–23)7.5 (2.9–19)
BMD and HPV+N and HPV−420.0 (0.0–8.4)0.0 (0.0–8.4)
 ≥BMD and/or HPV+994728348 (38–58)28 (20–38)
 16+ and/or 18+N and HPV−90.0 (0.0–34)0.0 (0.0–34)
 ≥BMD and/or HPV+432520358 (44–73)47 (33–62)
 16− and 18−N and HPV−330.0 (0.0–11)0.0 (0.0–11)
 ≥BMD and/or HPV+5622839 (28–53)14 (7.4–27)

To evaluate the test characteristics of cytology and hrHPV screening, we calculated sensitivity and specificity for different thresholds of test positivity (Table V). Since lesions could only have been detected in case of either ≥BMD or hrHPV positivity, the sensitivity of combined testing was assumed to be 100%. Using a threshold of ≥BMD, sensitivity for the detection of lesions ≥CIN2 was 68.0% (95%CI: 64.2–71.6) and specificity was 97.7% (95%CI: 97.4–98.1). Using hrHPV positivity as threshold, the sensitivity was 94.1% (95%CI: 91.7–95.9) and specificity was 96.1% (95%CI: 96.0–96.1). With hrHPV positivity for HPV16 and/or HPV18 as threshold, sensitivity was 62.5% (95%CI: 58.1–66.7) and specificity was 98.8% (98.6–99.0). The relative sensitivity of hrHPV testing compared with cytological testing was 0.941/0.680 = 1.38 (95%CI: 1.25–1.56) and the relative specificity was 0.961/0.977 = 0.98 (95%CI: 0.98–0.99). Using both ≥BMD and hrHPV positivity as threshold, estimates for sensitivity decreased slightly to 61.6% (95%CI: 57.7–65.4) for hrHPV-positive ≥BMD compared with a ≥ BMD cytology threshold, and specificity increased substantially to 99.3% (95%CI: 99.0–99.5). Using a positive test result of either test as threshold, in which case sensitivity was assumed to be 100%, the specificity was 94.5%.

Table V. Specificity and Sensitivity of Cytology Results and hrHPV Test Result at Baseline at Different Thresholds for Histological Detection of Lesions ≥CIN2 (Cumulative 18-Month Risk)
  ≥CIN2 ≥CIN3 
  1. N indicates normal cytology, BMD indicates borderline or mild dyskaryosis, and >BMD indicates moderate dyskaryosis or worse. NA indicates not applicable. Sensitivity is defined relative to the sensitivity of women with abnormal cytology and/or a positive hrHPV test (reference group).

Screening testThresholdSensitivity (95%CI)Specificity (95%CI)Sensitivity (95%CI)Specificity (95%CI)
Cytology≥BMD68.0 (64.2–71.6)97.7 (97.4–98.1)75.5 (71.0–79.5)97.4 (97.2–97.9)
 >BMD44.0 (40.1–48.0)99.8 (99.6–100)52.5 (47.5–57.4)99.7 (99.4–99.9)
hrHPVPositive94.1 (91.7–95.9)96.1 (96.0–96.1)94.2 (91.8–95.9)95.6 (95.5–95.8)
  16+ and/or 18+62.5 (58.1–66.7)98.8 (98.6–99.0)67.0 (61.8–71.8)98.5 (98.3–98.7)
  16− and 18−31.6 (27.7–35.8)97.3 (97.1–97.5)27.2 (22.7–32.2)97.1 (96.9–97.3)
hrHPV and CytologyPositive and ≥BMD61.6 (57.7–65.4)99.3 (99.0–99.5)69.3 (65.3–73.1)99.0 (98.8–99.3)
  16+ and/or 18+ and ≥BMD39.7 (25.2–31.9)99.7 (99.5–100)48.4 (28.4–37.1)99.6 (99.3–99.9)
  16− and 18− and ≥BMD22.4 (19.2–25.9)99.6 (99.3–99.8)21.3 (17.5–25.6)99.4 (99.2–99.7)
 Positive and >BMD40.7 (36.9–44.7)99.8 (99.6–100)49.6 (45.2–54.0)99.7 (99.5–100)
  16+ and/or 18+ and >BMD28.2 (24.7–31.9)99.9 (99.7–100)35.7 (31.1–40.6)99.8 (99.6–100)
  16− and 18− and >BMD12.8 (10.4–15.8)99.9 (99.7–100)14.2 (11.1–18.1)99.9 (99.6–100)
hrHPV and CytologyPositive or ≥BMD100 (reference)94.5100 (reference)94.0
  16+ and/or 18+ or ≥BMD90.8 (88.0–93.0)96.7 (96.5–96.9)94.1 (91.1–96.1)96.3 (96.1–96.5)
  16− and 18− or ≥BMD78.3 (74.6–81.7)98.7 (98.4–98.9)83.8 (79.5–87.3)98.3 (98.1–98.6)
 Positive or >BMD97.1 (95.2–98.3)96.0 (95.9–96.1)96.7 (94.8–98.0)95.6 (95.5–95.7)
  16+ and/or 18+ or >BMD77.2 (73.6–80.5)95.4 (95.3–95.5)81.4 (77.1–85.1)95.1 (94.9–95.2)
  16− and 18− or >BMD62.8 (58.7–66.7)97.2 (97.0–97.3)65.5 (60.5–70.2)96.9 (96.7–97.1)

Discussion

In this implementation study of hrHPV testing in population-based primary screening, we showed that primary hrHPV testing is more sensitive than cytology to detect ≥CIN2 lesions at the cost of slightly lower specificity. Moreover, we showed that women with HPV16 and/or HPV18 have a much higher ≥CIN2 risk than women positive for another hrHPV type. Retesting at 6 months showed that both cytological regression and hrHPV clearance are associated with decreased risks of ≥CIN2. Interestingly, women who had a double negative test at 6 months had a ≥CIN2 risk of 0.2% (95%CI: 0–1.1) and women with a hrHPV-negative BMD smear at baseline and normal cytology at 6 months had a ≥CIN2 risk of 0.0 (95%CI: 0–0.8). These data show that the risk of ≥CIN2 is better identified by 6 months with cytology and/or hrHPV testing than by single visit testing. Finally, our data support colposcopy referral for women with either hrHPV-positive BMD, or ≫BMD regardless of hrHPV status because of the magnitude of the risk of ≥CIN2.

In this study, retesting moments were 6 and 18 months as mandatory in the Dutch cervical screening programme.20 If hrHPV testing is implemented in primary screening, other retesting moments might be more cost-effective, with for example shorter intervals for HPV16 and HPV18 and longer intervals for other hrHPV types.

With lesions ≥CIN2 as the outcome of interest, screening sensitivity of classical cytology at a threshold of BMD was 68.0% and specificity was 97.7%. The sensitivity increased to 94.1% when using hrHPV testing, and specificity became slightly lower (96.1%). These data are in line with previous estimates obtained in screening studies using classical cytology and hrHPV testing by either GP5+/6+ PCR or Hybrid Capture 218, 21 and support the opinion that hrHPV testing should be used either alone or in conjunction with cytology in cervical screening. We also showed that the risk of ≥CIN2 in women with a hrHPV infection and normal cytology is higher than in women with hrHPV-negative BMD (12% vs. 2.5%) Besides, women with normal cytology harboring HPV16 and/or HPV18 had a substantially higher risk of ≥CIN2 than women infected with another high-risk type (25% vs. 5.3%), although the ≥CIN2 risk of the latter group was still higher than in the group of women with a hrHPV-negative BMD smear. Similar data have been described by Khan et al. and Berkhof et al.8, 9 These results underline that hrHPV-positive women with normal cytology, and especially women with HPV16 or HPV18, should have shorter intervals for retesting than women with BMD and a hrHPV-negative test result.

We assume that among women with an abnormal smear or a positive hrHPV test at baseline the vast majority of ≥CIN2 lesions diagnosed during the study were prevalent, and the histological diagnosis was postponed due to the screening algorithm as women with BMD or normal cytology were not referred for colposcopy immediately. Thus, a difference between prevalent cases diagnosed at intake and incident cases only diagnosed during follow-up was not made. Several studies, using either histological or cytological data, have indicated that disease detected during short-term follow-up corresponds to “missed” prevalent disease.22, 23

Until now most studies evaluated the risk of ≥CIN2 for strata defined by hrHPV test results and cytology at baseline.4, 8, 24, 25, 26, 27 Our study shows that for women with BMD or hrHPV-positive normal cytology a second test at 6 months, whether it be cytology or hrHPV testing, is more accurate in detecting ≥CIN2 than a single test at baseline. Tailoring the follow-up to allow for clearance of hrHPV and cytological regression of lesions will lead to a decrease in referrals for colposcopy. This is especially useful for hrHPV-positive women with normal cytology, since their baseline risk of a high-grade lesion is moderate. By retesting at 6 months, women can be distinguished with either a high or low ≥CIN2 risk. For instance, in our study HPV16 and/or HPV18-positive women with normal cytology at baseline followed by abnormal cytology at 6 months had a 4-fold higher risk of ≥CIN2 than women with normal cytology at 6 months. In the group of hrHPV-positive women without HPV16 and/or HPV18, the risk of ≥CIN2 was 9-fold higher when the repeat test was ≥BMD compared with normal. Results were even more pronounced when distinguishing women with a hrHPV-negative normal smear at 6 months as they had virtually no ≥CIN2 risk.

Although the negative predictive value for ≥CIN2 after a negative hrHPV test at baseline is higher than after negative cytology, the risk of a high-grade cervical lesion is not completely absent. Some participants in our study had high-grade cervical lesions but a negative hrHPV test (n = 31), a phenomenon also found by others.17, 23 These failures may be attributable to failure of cervical cell sampling, false-negative hrHPV test results or possibly incident disease. Additional analyses of samples from participants with ≥CIN2 that tested negative for hrHPV using the crude sample indicate that approximately half of the samples were negative due to inadequate material for PCR analysis. Half of the remaining samples were positive by E7 PCR, suggesting that integration of hrHPV DNA caused a negative GP5+/6+ PCR result (data not shown).

In conclusion, hrHPV is a major risk factor of high-grade cervical lesions and cervical cancer. We have now shown that repeat testing for women with BMD or hrHPV-positive normal cytology using either cytology or hrHPV testing detects the risk of ≥CIN2 better than single visit testing, and that HPV16 and/or HPV18 identifies women with the highest risk of ≥CIN2. Moreover, in women with hrHPV-negative normal cytology at the second test the risk of ≥CIN2 is virtually absent. At present, we are conducting cost-effectiveness analyses to determine the optimal algorithm for the use of cytology and hrHPV testing to detect cervical cancer and its precursor lesions in cervical screening programmes.

Acknowledgements

We gratefully acknowledge the work of the 242 GPs and their assistants, the District Health Authority Amstelveen, Medial, and District Health Authority Kennemerland-Haarlemmermeer e.o. We especially thank the research analysts of the Unit Molecular Pathology (VU University Medical Center, Amsterdam) and the Cytotechnologists (Spaarne Ziekenhuis, Hoofddorp; Kennemer Gasthuis, Haarlem; Leiden Cytology and Pathology Laboratory, Leiden; Unit Cytopathology, VU University Medical Center, Amsterdam).

Ancillary