Early Detection and Diagnosis
Post-treatment CIN: Randomised clinical trial using hrHPV testing for prediction of residual/recurrent disease†
All authors state no conflicts of interest. MvB and MR of the Department of Public Health received a grant from GSK, a manufacturer of a HPV vaccine, for research on the cost-effectiveness of HPV vaccination in 2007 and 2008. This research and manuscript were neither funded nor supported by GSK.
We investigated in a randomised clinical trial whether addition of hrHPV testing (high-risk human papillomavirus) to cytological follow-up after treatment for high-grade CIN (cervical intraepithelial neoplasia 2/3) can lead to a better selection of women at risk for residual/recurrent CIN. We included 210 women with high-grade CIN undergoing treatment in outpatient clinics in The Netherlands. Follow-up was based on cytology alone and cytology combined with hrHPV detection. Our primary outcome measurement was improving specificity for residual/recurrent CIN after treatment. Secondary, we compared health-care costs and impact of individual hrHPV type on the risk of residual/recurrent CIN. Follow-up by abnormal cytology alone (6, 12 and 24 months after treatment according to the Dutch protocol) showed a lower specificity for detection of residual/recurrent CIN than follow-up by abnormal cytology and presence of hrHPV (80 vs. 91%, relative risk 0.87 (95% CI 0.77–0.99)). Both methods showed no significant difference in sensitivity ((86 vs. 100%) RR 0.86 (95% CI 0.63–1.16)). Comparing different post hoc modifications in the strategy of combined testing showed similar test characteristics when low-risk women (normal cytology and hrHPV negative at 6 months) omitted the 12 months visit (specificity 91%, p = 1.00 z = 0.00). Prediction of residual/recurrent CIN by typing of hrHPV could not be confirmed. Total health-care costs using cytology and hrHPV testing during follow-up decreased when low-risk women omit the 12 months visit. Follow-up after treatment for high-grade CIN can be improved by combining cytology with hrHPV testing. We advise combined cytology and hrHPV testing at 6, 12 and 24 months after treatment. Low-risk women may omit the 12 months visit, resulting in cost reduction. © 2008 Wiley-Liss, Inc.
According to current Dutch standards, follow-up after treatment for high-grade CIN (cervical intraepithelial neoplasia 2/3) consists of cervical cytological monitoring at 6, 12 and 24 months, to identify residual/recurrent CIN lesions (CIN 2/3).1 Colposcopic examination is performed where there is abnormal cervical cytology. In case of normal cytology after 24 months, follow-up will be performed within the national screening program. One of the drawbacks of cervical cytological follow-up after treatment is the high number of false-positive findings. Approximately 20% of the women have abnormal cervical cytology within the first 2 years of follow-up. However, in more than half of these women no underlying residual/recurrent CIN will be found, resulting in unnecessary diagnostic and therapeutic procedures.2–6 The reported residual/recurrent high-grade CIN rates vary between 5 and 35%.2, 3, 7–9 A persistent infection with high-risk human papillomavirus (hrHPV) is necessary for the development, maintenance and progression of primary CIN lesions.10–12 It is assumed that effective treatment for CIN lesions results in clearance of the hrHPV DNA, while in residual/recurrent CIN disease hrHPV infection is still present.4, 8, 9, 13
Currently, only cytology is used in the follow-up after treatment. Previous studies have established the value of hrHPV testing for detection of high-grade CIN lesions.4, 8, 14–16 The use of combined testing by cytology and hrHPV can lead to a better selection of women at risk for residual/recurrent CIN after initial treatment for high-grade CIN lesions. This has been confirmed in several retrospective and observational studies.2, 8, 13, 17–22
This selection results in diagnostic procedures being performed only in those patients who have the necessary risk factor present for the development of recurrent/residual CIN lesions and avoids unnecessary diagnostic procedures in patients without this risk factor. Consequently, this policy can lead to an important reduction in health costs and anxiety. We investigated in a randomised clinical trial the role of hrHPV testing in addition to cytology during follow-up of post-treatment CIN.
Material and methods
Women who were to be treated for high-grade CIN lesions were asked to participate in the study. From July 2002 till September 2004 patients were enrolled at the outpatient clinic of the Department of Gynaecology and Obstetrics of the Erasmus University Medical Center in Rotterdam, the VU University Medical Centre in Amsterdam and the Albert Schweitzer Hospital in Dordrecht. Exclusion criteria were previous treatment for high-grade CIN, immune compromising conditions or previous or current cancer.
At baseline, women were asked to complete a questionnaire on education, smoking, number of sexual partners and history of sexually transmitted diseases. Before treatment, a cervical scrape was taken for hrHPV detection in all study participants. Treatment was performed according to standard methods such as loop excision of the transformation zone (LETZ), cold-knife-exconisation or laser-exconisation. All treatment was performed by experienced gynaecologists. After treatment, patients were randomised by computer in 2 groups: Group A, follow-up by cytology and Group B, follow-up by cytology and hrHPV testing at 6, 12 and 24 months after treatment.
Colposcopic examination was performed in Group A if cervical cytology was abnormal (according to the standard Dutch protocol) and in Group B if abnormal cytology was accompanied by a positive hrHPV test.
At the end of the study all women, irrespective of the test results, underwent colposcopic examination for end-histology to exclude residual/recurrent CIN. The study endpoint was reached 24 months after treatment or in the case of re-treatment, whichever came first. The study protocol was approved by a Multicenter Research Ethics Committee and by local review boards at all hospitals. All women gave signed informed consent before enrolment.
The primary outcome measure is an increase in specificity achieved by combined testing. This will secondarily result in a decrease in unnecessary examinations and treatment. Health-care costs and the possible impact of hrHPV type on the risk of residual/recurrent CIN were assessed.
Sample size calculation
Under the H0 hypothesis, it is assumed that the specificity of cytology and combined testing (cervical cytology and hrHPV testing) is equal. To prove a higher specificity of combined testing in comparison to cytology alone, we used power of 80 and 5% level of significance. Assuming the specificity of cytology alone to be 80% and of combined testing being 15% points higher, 148 women treated for CIN lesions without residual/recurrent CIN were needed for inclusion. Given the incidence of residual/recurrent CIN of 10–20%,2, 3, 7–9 185 women treated for CIN lesions had to be included. Taking into account, some patients lost to follow-up, 210 women were finally included.
We used block randomisation with random blocks of 4 and 6. The patient's physician reported each study candidate to the independent study coordinator through a central telephone number. This coordinator assigned participants to Group A or B. The results of randomisation were not blinded since the different groups used distinct follow-up policies based on their results. The results of hrHPV testing in Group A were blinded until the end of the study.
Cervical scrapes for HPV detection and genotyping were taken using a cervical bio-sampler (Accellon Combi® Medscand Medical, Sweden). HPV testing was performed with the consensus GP5+/GP6+ PCR enzyme immunoassay (EIA) using a cocktail probe covering all (probably) hrHPV types (i.e., HPV 16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 68, 73 and 82) and a yet unclassified HPV type (i.e., HPV 67) of which an oncogenic potential cannot be excluded, as previously described.23 This test is clinically validated.24 We used β-globin PCR to identify sampling errors and to monitor for PCR inhibitors. Additionally, reverse line blot (RBL) analysis was performed on PCR-EIA positive cases to identify individual HPV types.
Cytology, colposcopy and histology
Cervical smears were read according to the Dutch CISOE-A (KOPAC-B) classification, the standard classification used in The Netherlands.1, 25 The relationship between CISOE-A and histological CIN and SIL classification (Bethesda 2001) has been previously described.1
Smears were classified as normal (Pap 1) or abnormal, i.e., borderline dyskaryosis (Pap 2), mild dyskaryosis (Pap 3a1), moderate dyskaryosis (Pap 3a2), severe dyskaryosis (Pap 3b), suspected for carcinoma in situ (Pap 4) or suspect for invasive cancer (Pap 5). Experienced gynaecologists performed standard colposcopic assessment using acetic acid and iodine solution for the identification and delineation of the lesions. Biopsies were taken from all colposcopic abnormalities. Histology results were defined as mild dysplasia (CIN 1), moderate dysplasia (CIN 2), severe dysplasia (CIN 3) or (micro) invasive cancer. High-grade CIN and residual/recurrent CIN were defined as ≥CIN 2. A gynaecological pathologist revised all cytological and histological samples.
Post hoc modifications in the use of hrHPV testing during follow-up
In earlier retrospective and observational studies, various modifications in the use of hrHPV testing in the follow-up after treatment for high-grade CIN have been suggested.13, 18, 20, 26 We assessed the data from Group B also according to these different modifications.
The modifications were defined as B1, follow-up by combined testing whereby low-risk women (normal cytology and hrHPV negative at 6 months after treatment) omit the 12th months visit; B2, follow-up by hrHPV testing alone at 6, 12 and 24 months after treatment and B3, follow-up by hrHPV testing at 6 months and combined testing at 24 months after treatment.
For calculations of the health-care costs, we looked at 2 major sources of cost difference. These were reduction in unnecessary examinations (colposcopy and treatment) and increase in hrHPV tests. Additionally in the case of modification B1, the omission of cytology at the 12th months visit was taken into account. The reduction in unnecessary examinations was calculated through the observed difference in specificity and adjusted for the proportion of residual/recurrent CIN. We assumed that all women in Group B underwent hrHPV testing 3 times during 24 months of follow-up. We compared Group A, Group B and modification B1 within Group B. The average total cost per referral per CIN grade (including all related follow-up) was established in an earlier study27 and reflects the current Dutch situation. As costs were calculated from a socio-economic point of view, they included both direct medical (material, lab, consultations) and non-medical costs (time and travel costs of the women), which were valued according to Dutch guidelines.27 In this study, average costs of (unnecessary) examinations were estimated at €582 or €652 if hrHPV was present. The costs of cytology and hrHPV testing were estimated at €89 and €34, respectively.27, 28 All costs are calculated at price levels for 2005.
The specificity and sensitivity of both follow-up policies for residual/recurrent CIN were determined on basis of the outcome of the colposcopic examinations. Women who were lost to follow-up were excluded.
In the analyses 2 groups were compared:
We used two-by-two tables to assess the diagnostic value of the different follow-up policies for post-treatment CIN 2/3. In these analyses, women without a suspect cervical lesion on colposcopic examination or with CIN 0 (no CIN) or CIN 1 (mild dysplasia) in the biopsy were considered as ‘negative’. Relative risk and 95% confidence interval (CI) were calculated. We used McNemar's test for paired proportions to establish possible differences between scenarios for variations in the use of hrHPV testing during follow-up with respect to sensitivity and specificity.
A total of 210 women who attended the outpatient clinics for treatment of high-grade CIN were eligible for participation. Six of them were excluded: 4 women with previous treatment of CIN, 2 women diagnosed with cancer in the past.
The remaining 204 women were randomised (Fig. 1). No important differences in baseline characteristics were found between the 2 groups or between women who were successfully treated and women who developed residual/recurrent CIN (Table I). The residual/recurrent CIN rates were comparable in both groups (6.9 vs. 8.8%, relative risk 0.77 (95% CI 0.43–1.37)).
Table I. Baseline Characteristics
|Age at intake (years)|
| <30||10 (6)||10 (100)||0|
| 30–40||108 (59)||98 (91)||10 (9)|
| 40–50||53 (29)||47 (89)||6 (11)|
| >50||12 (7)||12 (100)||0|
|No. of sexual partners last year|
| 0–1||157 (86)||144 (92)||13 (8)|
| 2–4||26 (14)||23 (89)||3 (12)|
| No||95 (52)||90 (95)||5 (5)|
| Yes||88 (48)||77 (88)||11 (13)|
| Primary or less||29 (16)||29 (100)||0|
| Secondary||87 (48)||78 (90)||9 (10)|
| More than secondary||67 (37)||60 (90)||7 (10)|
|History of sexual transmitted disease|
| No||136 (74)||125 (92)||11 (8)|
| Chlamydia trachomatis||25 (14)||22 (88)||3 (12)|
| Condylomata acuminata||16 (9)||13 (81)||3 (19)|
| Other||11 (6)||11 (100)||0|
| LETZ||172 (94)||158 (92)||14 (8)|
| Exconisation||11 (6)||9 (82)||2 (18)|
Primary outcome measures
Abnormal cytology was seen after 6 months in 14/96 women in Group A (Fig. 1, Table II). Three of them had residual/recurrent CIN, leaving 11 women with a false-positive cytology result. After 12 months, 5/90 women had abnormal cytology and none of them had residual/recurrent CIN. After 24 months, 4/89 women had abnormal cytology, of which 3 had residual/recurrent CIN. In Group B, 7/96 women had abnormal cytology and hrHPV after 6 months, 4 of these had residual/recurrent CIN, leaving 3 women with false-positive test results. After 12 months, 5/87 women had abnormal cytology and hrHPV, of whom 2 had residual/recurrent CIN, leaving 3 false-positive combined tests. After 24 months, 4/85 women had abnormal cytology and hrHPV, 3 of these had residual/recurrent CIN, leaving 1 false-positive combined test. In a total follow-up of 24 months, we found 17/23 false-positive cytology results in Group A compared to 7/16 false-positive combined test results in Group B. With the exception of 1 woman, all cases of residual/recurrent CIN that occurred during follow-up were picked up in both strategies. After 24 months, 1 woman in Group A had residual/recurrent CIN with normal cytology.
Table II. Test Results in Relation to Residual/Recurrent CIN
Follow-up by conventional cytology (strategy for A) had a lower specificity for detection of residual/recurrent CIN compared to follow-up by cytology and hrHPV testing (strategy for B) (80 vs. 91%, relative risk 0.87 (95% CI 0.77–0.99)) (Table III). There was no significant difference in sensitivity (86 vs. 100%, relative risk 0.86 (95% CI 0.63–1.16)). PPV for strategy A was 26% compared to 56% for strategy B (relative risk 0.46 (95% CI 0.21–1.05)). NPV was 99% en 100% (relative risk 0.99 (95% CI 0.96–1.01)), respectively.
Table III. Test Characteristics for Detecting Residual/Recurrent CIN
Post hoc modifications in the use of hrHPV testing during follow-up
Since the test characteristics were more favourable in the strategy for B, we compared combined testing with various modifications of hrHPV testing (Table IV). First, modification B1 showed a specificity of 91% and sensitivity of 100%, which was not statistically different from Group B (specificity p = 1.00, z = 0.00; sensitivity p = 1.00 z = 0.71). Evaluating modifications B2 and B3, the test characteristics were even less favourable compared to combined testing (specificity 73%, p = 0.0002, z = 3.75; sensitivity 100%, p = 1.0, z = 0.00 and specificity 79%, p = 0.003, z = 3.02; sensitivity 86%, p = 0.5, z = 0.71, respectively).
Table IV. Test Characteristics of Post Hoc Modifications in the Use of hrHPV Testing for Detecting Residual/Recurrent CIN
|Sensitivity||100% (n = 9/9)||1.0||100% (n = 7/7)2||1.0||86% (n = 6/7)2||0.5|
|Specificity||91% (n = 75/82)||1.0||73% (n = 60/82)||0.0002||79% (n = 65/82)||0.003|
Secondary outcome measures
For Group A, the total average cost of unnecessary referral [(1 − specificity =) 20% × (1 − residual/recurrent CIN) 93% = 18.6% × €582 = ] was €108 per woman treated for high-grade CIN. For Group B, this cost was €50 per woman. The extra cost for hrHPV testing in Group B is €102 (3 times follow-up × €34).
In total, the cost of unnecessary referral for Group A was €108 and for Group B was €152.
Secondary, because post hoc modifications showed comparable test characteristics using modification B1, we calculated health costs as well for modification B1. The total average cost of unnecessary referral was €23 per woman. Using modification B1 included low-risk women to omit the 12th months visit. At 6 months, 62% of the treated women had normal cytology with hrHPV negative test results. Therefore, 2 times hrHPV testing and cytology are required for 62% in modification B1. The extra cost of hrHPV testing is [(62% × €34 × 2 + 38% × €34 × 3) = ] €81, whereas the reduction in cost of cytology at the 12th months visit is 62% × €89 = €55. In total, the cost of unnecessary referral for modification B1 was €49. For Group B, the cost was on average €44 more than for Group A. However, for modification B1 the cost decreases by €59.
A total of 182 women were hrHPV positive at baseline, 10 women were hrHPV negative (Table V). The most common types were HPV 16, 31 and 33, although a wide variety of other types was present.
Table V. Prevalence of hrHPV Types During Follow-Up
|hrHPV type (n)||S/M||S/M||S/M||S/M|
There was no significant difference in the distribution of hrHPV types between the 2 randomization arms (data not shown). These findings persisted during follow-up. For further analyses, all data were combined to establish any possible influence of specific hrHPV genotypes in the development of residual/recurrent CIN within the total follow-up of 24 months. Single infections were more common than multiple infections for all women (72 vs. 28%); this ratio remained stable during follow-up. In 21 of the 49 (43%) hrHPV positive women after treatment for high-grade CIN, we found newly acquired hrHPV types. In 28 out of 49 (57%), the same hrHPV type was detected. For women who developed residual/recurrent CIN, 11 out of 16 (69%) revealed the same hrHPV type as seen before treatment, the remaining 5 (31%) showed newly acquired types.
HrHPV type 16 was present in the majority of women with residual/recurrent CIN before treatment (n = 12/16) and at time of presence of residual/recurrent CIN (n = 11/16).
Follow-up of women after treatment for high-grade CIN is more efficient and effective using both cytology and hrHPV testing. Compared with the conventional protocol using cytology alone at 6, 12 and 24 months after treatment (current national guidelines), a higher specificity can be obtained using cytology in combination with hrHPV (i.e., 80 vs. 91%). As a consequence, fewer false-positives lead to fewer unnecessary renewed examinations. This is in agreement with previous observational studies where hrHPV testing has been proposed as a screening tool for follow-up after treatment for high-grade CIN.13, 18–20, 22, 26 Although improvement of specificity to detect residual/recurrent CIN was obtained using combined testing, no significant change in sensitivity was observed. Because of the small numbers of residual/recurrent CIN, this finding remains uncertain based on the confidence interval (Table III). Larger studies are necessary for evaluation of the sensitivity.
Our results indicate an improvement of the positive predictive value by a third using combined testing, however just not statistically significant given the small numbers of residual/recurrent CIN.
Since residual/recurrent CIN only develops when a hrHPV infection is present,4, 21 our proposed policy includes colposcopic examination when both abnormal cytology and hrHPV are detected. Thus, cases with only hrHPV infection and no cytological cervical changes or those with abnormal cytology without a hrHPV infection do not need additional investigation.19, 22, 29, 30
In Group A (follow-up by cytology), we found 1 woman with residual/recurrent high-grade CIN (CIN 2) after 24 months but with normal cytology and presence of hrHPV. Because this was our study endpoint including colposcopic examination, the residual/recurrent CIN was detected. In the current national follow-up protocol, this lesion would have been left undetected and the patient would have been returned to the national screening program where she would have been screened 2–3 years later. According to the standard Dutch protocol, this is considered an accepted level of risk, as the probability of developing invasive cervical cancer within these 2–3 years is extremely low.12, 31, 32
Post hoc modifications in the use of hrHPV testing after treatment for high-grade CIN were also evaluated. Some studies suggest omitting the 12th months visit for low-risk women (normal cytology and hrHPV negative 6 months after treatment).13, 18, 30, 33 According to our evaluation, the test characteristics (described as modification B1) were comparable to values obtained for combined testing at 6, 12 and 24 months after treatment. Other modifications such as hrHPV testing alone (B2) and hrHPV testing at 6 months with combined testing at 24 months (B3) showed less favourable test characteristics. As a result of the low specificity rates of hrHPV as a stand-alone test, unnecessary examinations could not be prevented in these groups. Although Coupé et al. previously recommended the latter strategy on basis of a simulation prediction model, we could not confirm the advantage in our study.26 An explanation might be the difference in follow-up time for residual/recurrence, which was 5 years in the prediction model in contrast with 2 years follow-up in this randomised clinical trial. Conversely, our proposed modification (B1) of combined testing at 6 and 24 months for low-risk women and combined testing at 6,12 and 24 months for high-risk women was not evaluated in the prediction model.
Taking the costs into account, follow-up after treatment for high-grade CIN using both cytology and hrHPV testing seems to be slightly more expensive. Although unnecessary examinations could be prevented, the total costs of the additional hrHPV testing were higher, even though this is only a fraction of the total costs of treatment for high-grade CIN (≤ 10% extra costs).27 Yet, in the future these costs may decrease (using liquid based cytology). In addition to costs, effects on quality-of-life must also be taken in account. When low-risk women omit the 12th months visit, reduction in health costs can be achieved.
HrHPV genotyping may potentially be a useful tool to determine the development of residual/recurrent CIN.34–36 Gok et al. have suggested an increased risk for residual/recurrent CIN for HPV 16.34 In this study, we could not confirm these results, although the majority of women who developed residual/recurrent CIN had hrHPV type 16 present before treatment and at time of residual/recurrent CIN. The difference might be explained by the small number of residual/recurrent CIN. Also the number of follow-up visits was distinct; we evaluated patients after 6, 12 and 24 months and Gok et al. evaluated additionally after 3 and 18 months. Van Ham et al. have reported a high level of newly detected hrHPV types after treatment.35 Different hrHPV types were detected in 75% of all hrHPV positive women 6 months after treatment, without description of presence or absence of residual/recurrent CIN. We found newly acquired hrHPV types in 21 out of 49 (43%) hrHPV positive women after treatment during follow-up of 24 months. Women with residual/recurrent CIN showed a newly acquired hrHPV type in only in 5 out of 16 (31%). It thus remains unconfirmed whether hrHPV genotyping has clinical value in follow-up policies after treatment for high-grade CIN.
Summarised, using combined testing at 6, 12 and 24 months after treatment for high-grade CIN prevention of 11% of unnecessary referral can be obtained. This could have major consequences for the women involved. Even the total health-care costs will decrease as a consequence of less unnecessary examinations. Aspects of quality-of-life (e.g., less anxiety) remain to be evaluated.
On the basis of our randomised clinical trial, we recommend a follow-up policy for women treated for high-grade CIN consisting of both cytology and hrHPV testing at 6, 12 and 24 months after treatment. Subsequently, follow-up will be performed within the national screening program. Low-risk women (with normal cytology and a negative hrHPV test at 6 months) may omit the 12th months visit during follow-up, resulting in cost reduction.
We are indebted to Rick van den Andel and Muriel Verkuijten, Laboratory of Molecular Pathology, VU University Medical Centre, Amsterdam for excellent technical assistance, Johan Spruijt, gynaecologist of the VU University Medical Center, Amsterdam, Cees Hogewoning, gynaecologist of the Albert Schweitzer Hospital in Dordrecht for patient care, Patricia Ewing, pathologist and Ilse Beckmann of the Department of Obstetrics and Gynaecology 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 analyse the data independently, to prepare manuscripts, and to publish them. Trial registration: The Dutch Cochrane Center ISRCTN 31244687.