Triage using HPV-testing in persistent borderline and mildly dyskaryotic smears: Proposal for new guidelines
Version of Record online: 4 MAR 2005
Copyright © 2005 Wiley-Liss, Inc.
International Journal of Cancer
Volume 116, Issue 1, pages 122–129, 10 August 2005
How to Cite
Bais, A. G., Rebolj, M., Snijders, P. J.F., de Schipper, F. A., van der Meulen, D. A.J., Verheijen, R. H.M., Voorhorst, F., van Ballegooijen, M., Meijer, C. J.L.M. and Helmerhorst, T. J.M. (2005), Triage using HPV-testing in persistent borderline and mildly dyskaryotic smears: Proposal for new guidelines. Int. J. Cancer, 116: 122–129. doi: 10.1002/ijc.20958
- Issue online: 12 MAY 2005
- Version of Record online: 4 MAR 2005
- Manuscript Accepted: 15 DEC 2004
- Manuscript Received: 22 SEP 2004
- Dutch Prevention Fund/ZorgOnderzoekNederland (ZON). Grant Number: 2200.0054
- Revolving Fund Erasmus MC. Grant Number: 98-9
- borderline or mild dyskaryosis;
In the Netherlands 2% of cervical smears in the cervical cancer screening program are read as borderline or mildly dyskaryotic cytology (BMD smear). Only in about 10% of these women a high-grade CIN lesion (CIN II–III) is present; therefore referral is for the majority unnecessary. In our study triage with high-risk HPV (hrHPV) testing was used to identify women at risk for development of high-grade CIN lesions after a repeat BMD smear. A “wait-and-see” period was incorporated allowing clearance of HPV and regression of the lesion. Women with a low-grade lesion, irrespective of their HPV status, were monitored at 12 months; women with a high-grade lesion were monitored at 6 and 12 months. Fifty-one of the 105 women (49%) were hrHPV negative at baseline; none of them showed progression of the lesion within the first year of follow-up (NPV 100%). High-grade CIN was present in 1 patient who was HPV negative at baseline (2%); she demonstrated regression after 12 months. Nineteen of the hrHPV positive women (35%) demonstrated a high-grade CIN lesion at baseline and 3 cleared hrHPV after 6 months, with a subsequent regression of CIN. Ten women remained hrHPV positive with persistence of high-grade CIN and were eventually treated. At baseline, 35 hrHPV positive women demonstrated a low-grade lesion, 19 remained hrHPV positive after 12 months and 5 developed high-grade CIN. Sixteen out of the 35 cleared the hrHPV infection without progression of the lesion. In conclusion, triage, using hrHPV testing for women with persistent BMD cytology, can select women who are not at risk for development of high-grade CIN. We recommend return to the screening program without referral for colposcopic examination if hrHPV is absent. For hrHPV positive women, a repeat hrHPV test after another 6 months is suggested. Referral is only required if persistence of hrHPV is established. © 2005 Wiley-Liss, Inc.
In the Dutch population-based cervical cancer-screening program, women between 30 and 60 years of age undergo a cytological smear every 5th year. Recent surveys showed that about 2% of these smears contained borderline or mildly dyskaryotic (BMD) changes.1, 2 According to current screening guidelines, in these cases cytology is repeated after 6 months and (if negative) after 12 months. When cytology is persistently abnormal, women are referred to the gynaecologist where colposcopic examination is performed and a biopsy is taken for histological examination. When histology shows no or low-grade dysplasia (≤ CIN I) women will be kept under cytological surveillance without immediate treatment. High-grade dysplasia (CIN II–III) or a worse lesion will be treated. Only 10% of women with a single BMD smear show a high-grade CIN lesion.3, 4, 5, 6 Consequently, there is a need for an improved risk assessment of women with BMD.
Infection with high-risk HPV (hrHPV) has been established as an important etiological factor in the carcinogenesis of cervical cancer.7, 8 Persistence of hrHPV infection is required for development of cervical cancer.9, 10, 11, 12 Several studies have shown that HPV testing has a high sensitivity (approximately 95%) for identifying high-grade lesions and cervical carcinoma. Moreover, the negative predictive value of HPV testing for detection of a high-grade CIN lesion or even (micro)invasive carcinoma is nearly 100%.13, 14 Consequently, HPV-testing may play an important role in clinical practice, i.e., in triage of women with BMD cytology, as an adjunct to cytology in primary screening and in posttreatment protocols.15, 16, 17
The prevalence of hrHPV in women with BMD smears is about 35%18, 19 and it is likely that hrHPV negative BMD women are not at risk for development of high-grade CIN lesions or carcinoma. Hence, triage with hrHPV testing may improve the selection of women at risk for development of cervical cancer and consequently the management of women with BMD smears.15, 19, 20, 21, 22, 23 This would not only have the advantage of improving the efficiency of the screening program through fewer referrals but also would reduce unnecessary anxiety among the majority of women with BMD who are not at risk for high-grade CIN.3, 5, 24
In this prospective study, triage with hrHPV testing was performed in the follow-up of women referred with a repeated BMD established in the screening program. Unlike most published studies, a wait-and-see period was incorporated in the present protocol to allow potential clearance of hrHPV infection and consequently regression of the cervical lesion. We evaluated the use of triage with hrHPV testing to prevent unnecessary diagnostic procedures and treatment.
Material and methods
All women with a repeated BMD smear referred through the national screening program to gynaecological outpatient clinics were asked to participate in our study. Participating hospitals were the Erasmus University Medical Centre Rotterdam (December 1999–May 2003), the Hospital Walcheren in Vlissingen (January 2000–March 2002) and the VU University Medical Center Amsterdam (June 2000–January 2003). Exclusion criteria were pregnancy at time of enrolment or during follow-up (n = 3), non-Hodgkin lymphoma (n = 1), age younger than 30 years or older than 60 years at time of abnormal smear (n = 5) or insufficient Dutch or English language skills (n = 1).
Figure 1 shows the trial design. At baseline women were asked to complete a questionnaire on education, ethnic background, smoking, number of sexual partners, sexarche, contraceptive use and history of sexually transmitted diseases. A cervical scrape was taken for HPV detection followed by colposcopy. Standard colposcopical assessment with acetic acid and iodine solution was performed by expert gynaecologists. Biopsies were taken from all colposcopic abnormalities. At this first visit no treatment was carried out.
Women who were hrHPV negative were reviewed after 12 months. HrHPV positive women were reviewed after 12 months if the colposcopically-directed biopsy was ≤ CIN I, and after 6 and 12 months if histology showed moderate to severe dysplasia (CIN II–III). During follow-up visits hrHPV-testing and colposcopy were performed. Women with a persistent hrHPV-positive high-grade CIN lesion during follow-up were treated by loop excision of the transformation zone (LETZ). The study endpoint was reached after 12 months, or after treatment, whichever came first. The study protocol was approved by a multicenter research ethics committee and by local committees at all 3 hospitals. All women voluntarily gave signed informed consent before enrolment.
Lesions were histologically defined as mild dysplasia (CIN I), moderate dysplasia (CIN II), severe dysplasia (CIN III) and (micro) invasive cancer. Lesions ≤ CIN I are hereafter referred to as a low-grade lesion and CIN II–III as a high-grade lesion. Regression or progression was defined as histological change from a high-grade lesion to low-grade lesion or vice versa, detected in the biopsy material at 2 consecutive time points as scheduled by the trial design. All histological samples were read by expert pathologists.
Human papillomavirus testing
All HPV samples were taken by a cervical biosampler (Accellon Combi® Medscand Medical, Sweden). Testing for HPV was conducted by using a general/consensus primer based GP5+/GP6+ polymerase chain reaction (PCR) enzyme immunoassay (EIA) for 14 high risk types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68).25 This test has been clinically validated.11, 26, 27 Additionally, reverse line blot (RLB) analysis was used to identify individual HPV types in case the PCR-EIA was positive.25 We used β-globin PCR to identify sampling errors and to monitor for PCR inhibitors.
The risk of development of high-grade CIN associated with HPV presence was assessed with the Fisher's exact test and chi-square. Confidence intervals of 95 percent and 2-sided p-values were used.
A total of 105 women met the inclusion criteria. The mean age was 39 years (range 30–60 years). The median time-lag between the 2 BMD smears on the basis of which women were referred was 7 months, with a range of 2–20 months. The median time-lag between the 2nd BMD smear and colposcopic examination was 2 months (range 0–8 months).
At baseline 54 women (51%) were hrHPV positive, leaving 51 hrHPV negative subjects (Fig. 1). HPV 16 and 31 were the most frequently detected hrHPV types (40% and 20% of all infections, respectively). Other types were less common: HPV 18 (7%), HPV 33 (10%), HPV 35 (7%), HPV 42 (7%), HPV 45 (2%), HPV 51 (7%), HPV 52 (10%), HPV 56 (10%), HPV 58 (7%), HPV 59 (2%) and HPV 66 (2%). Of the observed HPV positive scrapings at baseline, 68% contained single infections and in the remaining 32% a multiple infection was detected. With 1 exception, all hrHPV negative women demonstrated no or a low-grade lesion at baseline biopsy: 27 women without dysplasia (53%) and 23 women with CIN I (45%). One hrHPV negative woman had a CIN II lesion at baseline.
Among hrHPV positive women at baseline, 35 cases (65%) had ≤ CIN I and 19 cases (35%) CIN II–III. The first group consisted of 10 women (19%) without dysplasia and 25 women (46%) with a CIN I lesion, whereas the high-grade CIN group comprised 14 women (26%) with a CIN II lesion and 5 women (9%) with a CIN III lesion.
Table I shows the characteristics of the study population stratified according to hrHPV presence at baseline. When age is stratified in 3 categories, 30–40, 40–50 and 50–60 years, respectively, we see a significant difference for the youngest group (30–40 years) where HPV is more frequently present (p < 0.01). The presence of CIN II–III was significantly higher in women who were hrHPV positive at baseline, compared to hrHPV negative women (p < 0.01 with an odds ratio of 27 [95% CI 3–211]). Women who were hrHPV negative were less likely to have a history of sexually transmitted disease (p < 0.01). Other risk factors such as education, ethnic background, sexarche, number of sexual partners in the preceding year, smoking and oral contraceptive showed no difference of statistical significance between groups with and without HPV infection.
|Total number (%) N = 105||hrHPV negative at intake number (%) N = 51||hrHPV positive at intake number (%) N = 54||Significance|
|Age at intake (years)1||p < 0.01|
|30–40||58 (55)||17 (33)||41 (76)|
|40–50||26 (25)||16 (31)||10 (19)|
|50–60||21 (20)||18 (35)||3 (6)|
|Histology at intake1||p < 0.01|
|Low-grade||85 (81)||50 (98)||35 (65)|
|High-grade||20 (19)||1 (2)||19 (35)|
|Caucasian||82 (78)||43 (84)||39 (72)|
|Asian||4 (4)||2 (4)||2 (4)|
|Negroid||10 (10)||4 (8)||6 (11)|
|Mediterranean||4 (4)||0||4 (7)|
|Other||5 (5)||2 (4)||3 (6)|
|Primary or less||9 (9)||4 (8)||5 (9)|
|Secondary, incomplete||55 (52)||32 (63)||23 (43)|
|Secondary or more||41 (39)||15 (29)||26 (48)|
|Age at first intercourse (years)|
|≤ 15||23 (22)||8 (16)||15 (28)|
|16–18||51 (49)||26 (51)||25 (46)|
|≥ 19||31 (30)||17 (33)||14 (26)|
|No. of sexual partners last year|
|0–1||89 (85)||45 (88)||44 (81)|
|2–4||16 (15)||6 (12)||10 (19)|
|No||56 (53)||32 (63)||24 (44)|
|Yes||49 (47)||19 (37)||30 (56)|
|Oral contraceptive use|
|No||62 (59)||34 (67)||28 (52)|
|Yes||43 (41)||17 (33)||26 (48)|
|History of Sexual Transmitted disease1||p < 0.01|
|No||84 (80)||48 (94)||36 (67)|
|Chlamydia trachomatis||9 (9)||3 (6)||6 (11)|
|Condylomata accuminata||5 (5)||5 (9)|
|Other||7 (7)||7 (13)|
Follow-up of hrHPV negative women
One woman acquired hrHPV infection after 12 months, without progression to high-grade CIN. All other women (50/51) remained hrHPV negative and did not develop a high-grade CIN lesion. The woman with a high-grade CIN lesion at baseline showed histological regression to no dysplasia after 12 months (Table II).
|Baseline||t = 6 months2||t = 12 months|
|≤ CIN I||CIN II–III||Total n (%)||≤ CIN I||CIN II–III||Total n (%)||≤ CIN I||CIN II–III||Total n (%)|
|HPV pos (n)1||35||19||54 (51%)||9||10||19 (100%)||37||6||43 (41%)|
|HPV neg (n)1||50||1||51 (49%)||—||—||—||51||0||51 (49%)|
Follow-up of hrHPV positive women
Nineteen of the 35 hrHPV positive women (54%) with CIN ≤ I at baseline demonstrated persistence of hrHPV infection after 12 months. Five out of these 19 (26%) showed progression to high-grade CIN. No (micro)invasion was detected. Fourteen women without HPV clearance continued to have CIN ≤ I. Clearance of hrHPV infection occurred in 16/35 women (46%), none of whom showed progression detected in histological biopsies, i.e., 14 women without dysplasia and 2 women who still had CIN I after 12 months.
Among the 19 women with a high-grade lesion, 3 (16%) cleared the hrHPV infection after 6 months, with histological regression to a low-grade lesion, and remained hrHPV negative after 12 months. Amongst the 16 women who were still hrHPV positive after 6 months (84%) 10 (63%) revealed a (persistent) high-grade lesion. No (micro)invasion was detected. These 10 women were treated. The remaining 6 women (37%) demonstrated regression to a low-grade lesion. One of them was treated at her own request and therefore left the study group prematurely. After 12 months, 2 out of 5 women remained persistently hrHPV positive, 1 of whom had a high-grade lesion (CIN II). The other 3 women showed clearance of hrHPV and a low-grade lesion on histology.
In our study, progression to a high-grade CIN lesion was not seen in hrHPV negative women with a persistent BMD smear detected in the national screening program. All women with a high-grade lesion were hrHPV positive at baseline, except for 1 woman who had a CIN II lesion at baseline and was hrHPV negative. Histology from this patient showed regression to a low-grade lesion after 12 months, suggesting that for this woman clearance of hrHPV was already evident at baseline, with subsequent regression of the lesion becoming apparent after 12 months. This corresponds with the findings of other studies.17, 20, 28
In our study, the negative predictive value (NPV) of hrHPV testing for having high-grade CIN was 98% at baseline for women with a repeated BMD smear, and 100% after 6 and 12 months follow-up. This suggests that hrHPV negative women, despite a persistent BMD cytology, are not at risk for high-grade CIN.
Various studies have shown that better results are obtained using HPV detection when compared to conventional cytology (i.e., NPV cytology 93–98%).7, 14, 19 Triage with HPV-testing as an adjunct to cytology has been proposed for women with minimal cytological changes.15, 17, 23 Several studies have used different classifications of mildly abnormal cytology.17, 19, 20, 23, 30, 31, 32 Cuzick et al. included women with borderline cytology and negative cytology but were hrHPV positive (HART study).23 The ALTS study describes triage in women with ASCUS and LSIL cytology (ALTS Group).29 An overview of triage studies with corresponding data is shown in Table III. Selection was based on HPV detection methods that were clinically validated in longitudinal studies (i.e., HCII and GP5+/6+ PCR) since the clinical rather than the analytical performance of HPV detection methods should be considered for triage policies.33 The main benefit that can be concluded from all listed studies is a reduction in referral of hrHPV negative women with minimal abnormal cytology for colposcopy. For more severe cytological abnormalities the use of triage with HPV-testing is not indicated. Our study was based on women referred to the gynaecologist for 2 times BMD smear according to the national guidelines. Based on our results, we suggest that women with a persistent BMD smear and a concomitant negative hrHPV test result are not at risk for development of cervical cancer and should stay in the screening program, without colposcopic evaluation. In the Netherlands the next scheduled screening would then take place after 4.5 years. The risk that a woman will develop high-grade CIN within that period of time seems acceptable since it will be exceptional to take place within 5 years given that a persistent hrHPV infection is required.4, 34 Precursor lesions are detectable within an average of 10–15 years before progression to cancer.35 Thus, even though our study only covers 12 months of follow-up, it seems reasonable to return women with a hrHPV negative persistent BMD smear into the screening program without additional surveillance or treatment. However, an estimate of excess risk among hrHPV negative women with BMD cytology is expected within the next few years when, e.g., the ongoing Dutch POBASCAM screening study and Swedish screening study are expected to yield final results.1, 36
|Study||Population||n||Mean age (years)||Follow-up||HPV+||HPV clearance||CIN 2/3||Sens %||Spec %||PPV %||NPV %||Reduction of referral|
|Cuzick et al.,2003||Borderline cytology negative cytology HPV+||825||42 y||6–12 months||27%||35–45%||3%||97||93||13||HPV- borderline cytology return to screeningHPV+ borderline cytology retest HPV after 12 monthsHPV+ mild dyskaryosis refer|
|Ho et al., 2003||Repeated mild dyskaryosis or less||149||28–32 y||Observational study||59%||44%||24%||100||42% reduction based on HPV+|
|Every 6 monthsMean 3 years||HPV- mild dyskaryosis retest 6–12 months|
|retest HPV: total of 67% less referral|
|Solomon et al., 2001||ASCUS||2,198||29 y||6–12–18 months||54%||12%||96||20||99||44% reduction based on ASCUS HPV+|
|Sherman et al., 2002||ASCUS-LSIL||3,046||29–25 y||89%||96–97||46% reduction based on ASCUS HPV+, 15% reduction if LSIL HPV+|
|Guido et al., 2003||ASCUS||881||25 y||CIN I or less follow-up||18%||92||45% reduction based on HPV retest after 12 months|
|Cox et al., 2003||ASCUS HPV+||1,132||51%||18%||49% reduction based onASCUS HPV+|
|Zielinski et al., 2001||Single and repeated borderline or mild dyskaryosis||278||41 y||Observational study||42%||45%||10%||96||63||22||98||58% reduction based on HPV+|
|Every 6 months median 1.4 years|
|Current study||Repeated borderline or mild dyskaryosis||105||39 y||Intervention study: treatment postponement 6–12 months||51%||41%||19%||95||59||35||98||49% reduction based on HPV+|
In our study 54/105 women (51%) were hrHPV positive at baseline of whom 19 (35%) had a histologically confirmed high-grade CIN lesion. The overall frequency of high-grade CIN was 19%, which is comparable with the findings of Ho et al.20 In contrast, a previous study by our group demonstrated a prevalence of 10%.19 In that study however, about half of the women were referred after 1 BMD smear in contrast to our current study where women were referred after 2 BMD smears in 6 months. Consequently, this group with BMD contains more women with mildly dyskaryosis than the previous group (38% vs. 26%) (data not shown). Therefore, our results can be explained by the fact that the prevalence of high-grade CIN increases with severity of abnormal cytology.12, 19
In this trial women with a high-grade lesion were not treated immediately. A “wait-and-see” period of 6 months showed persistence of high-grade CIN only in women with persistent hrHPV infection. No patient developed (micro)invasive carcinoma. These findings suggest that a wait-and-see period for (at least) 6 months involves no additional health risks. In addition, 16% of these women showed clearance of hrHPV with a subsequent regression of the lesion. They were no longer at risk of development of high-grade CIN and consequently an expectative policy could result in a reduction of the need for treatment.
The majority of hrHPV positive women with a repeat BMD smear have low-grade CIN lesions [in our study 35 out of 54 (65%) at baseline]. This is in agreement with the findings of previous studies.6, 17, 19, 20, 23, 30, 31, 32 Approximately half of these women cleared hrHPV in 12 months and none of them developed a high-grade CIN during this time. Fourteen percent developed high-grade CIN after 12 months (all were persistently hrHPV positive), without an observed development of (micro)invasive carcinoma. It should, however, be considered that the biopsies taken to establish the severity of premalignant lesions may have interfered with the natural course of disease. On the other hand, several natural history studies have demonstrated that HPV clearance and histological regression occur after 6–12 months.4, 28, 37 Our follow-up time of 12 months should therefore cover the major part of regression. Consequently, referral can therefore be restricted to women with a repeat BMD smear who remain persistently hrHPV positive for at least another 6 months.
Although our proposal requires more hrHPV testing (100%), according to the present study it will lead to a 49% reduction in referrals for colposcopy. A “wait-and-see” period in the HPV positive group for at least 6 months will result in a supplement of 51% HPV-testing and a further 18% reduction in referrals and 9 percentage points treatment reduction. In summary, a total reduction of 67% referrals for colposcopic evaluation and 38% less treatment can be obtained by addition of 151% HPV-testing. The 6 months wait-and-see period in hrHPV positive women may also have negative influences on the women involved. Some women might prefer a see-and-treat policy instead of a more conservative approach. Proper information and explanation of the natural development of premalignant lesions in relation to hrHPV infection by the physician is therefore required, after which patients preference may influence further choices especially in this group.38 The potential gain of less treatment however remains to be weighted against potentially changes in quality-of-life and effectiveness. Our recommendations are summarised in Figure 2.
In conclusion, triage using hrHPV testing for women with a persistent borderline or mildly dyskaryotic cytology is recommended. Women can stay in the screening program without referral to the gynaecologist if no hrHPV infection can be determined. We suggest a repeat HPV test after another 6 months for hrHPV positive women. Referral for colposcopy is only required if persistence of hrHPV is demonstrated. Women who cleared hrHPV infection are no longer at risk of development of premalignant lesion. Further surveillance can take place within the screening program at the protocollized interval of 5 years.
We are indebted to R. van den Andel, S. van Schaik, N. Fransen Daalmeijer and M. Lettink for excellent technical assistance and P. Ewing, pathologist at the Erasmus University Medical Centre Rotterdam, for reviewing the paper. The authors certify that they have not entered into any agreement that could interfere with their access to the data on the research, nor upon their ability to analyze the data independently, to prepare manuscripts and to publish them.
- 1POBASCAM, a population-based randomized controlled trial for implementation of high-risk HPV testing in cervical screening: design, methods and baseline data of 44,102 women. Int J Cancer 2004; 110: 94–101., , , , , , , , , , , , et al.
- 2De praktijk van het bevolkingsonderzoek naar baarmoederhalskanker in Nederland in 2001: rapport in het kader van de landelijke evaluatie van het bevolkingsonderzoek naar baarmoederhalskanker (LEBA): Deel 3, 2003., , , , , .
- 22A systematic review of the role of human papillomavirus testing within a cervical screening programme. Health Technol Assess 1999; 3: 1–196., , , , , , , .
- 31ALTS-group). Postcolposcopy management strategies for women referred with low-grade squamous intraepithelial lesions or human papillomavirus DNA-positive atypical squamous cells of undetermined significance: a two-year prospective study. Am J Obstet Gynecol 2003; 188: 1401–5., , , (
- 32ALTS-group). Prospective follow-up suggests similar risk of subsequent cervical intraepithelial neoplasia grade 2 or 3 among women with cervical intraepithelial neoplasia grade 1 or negative colposcopy and directed biopsy. Am J Obstet Gynecol 2003; 188: 1406–12., , (