To evaluate the feasibility and utility of intraoperative post-conisation human papillomavirus (IOP-HPV) testing and cytology to detect treatment failure in patients with cervical intraepithelial neoplasia grades 2–3 (CIN2–3).
To evaluate the feasibility and utility of intraoperative post-conisation human papillomavirus (IOP-HPV) testing and cytology to detect treatment failure in patients with cervical intraepithelial neoplasia grades 2–3 (CIN2–3).
Prospective observational pilot study.
A cohort of 132 women treated for CIN2–3 by loop electrosurgical conisation.
An endocervical sample was obtained intraoperatively with a cytobrush from the cervix remaining after the conisation. The material was kept in PreservCyt medium and processed for Hybrid Capture 2 and cytology. Patients were followed-up for 24 months. The performance of IOP-HPV testing and IOP cytology was compared with conventional indicators of recurrence (cone margin, endocervical curettage, and HPV testing and cytology at 6 months).
Treatment failure (i.e. recurrent CIN2–3 during follow-up).
Treatment failure was identified in 12 women (9.1%). IOP-HPV testing for sensitivity, specificity, and positive and negative predictive values for treatment failure were 91.7, 78.3, 62.2, and 96.0%, respectively, which are similar to the figures for conventional HPV testing at 6 months (91.7, 76.0, 64.0, and 95.1%, respectively), and are better than the values of other conventional predictive factors (cone margin, endocervical curettage, and cytology intraoperative at 6 months). IOP-HPV was strongly associated with treatment failure in the multivariate analysis (OR 15.40, 95% CI 1.58–150.42).
IOP-HPV testing is feasible, and accurately predicts treatment failure in patients with CIN2–3. This new approach may allow an early identification of patients with treatment failure, thereby facilitating the scheduling of an attenuated follow-up for negative patients who are at very low risk of persistent disease.
Conisation using the loop electrosurgical procedure (LEEP) is the standard treatment of cervical intraepithelial neoplasia grades 2 or 3 (CIN2–3), a premalignant lesion caused by persistent infection with high-risk human papillomavirus (HPV).[1, 2] Residual or recurrent disease after conisation occurs in 5–25% of cases.[3, 4] Therefore, close follow-up of patients after treatment is recommended to identify failures requiring re-excision.[5-8] Current protocols of post-conisation follow-up are based on serial cytology and HPV testing at 6 or 12 months after treatment.[2-5, 7, 9-19] Nevertheless, an earlier identification of patients at higher risk of harbouring treatment failure might contribute to avoiding delays in the re-treatment of these patients, and would allow reducing unnecessary visits for women at very low risk of post-treatment disease.
Most precancerous lesions arise at the transformation zone, and it has been shown that after excision of the lesion and the transformation zone a high percentage of women clear not only the lesion but also the infection, and become negative for HPV.[2-5, 7, 9-18, 20, 21] This suggests that the HPV infection is mainly limited to the lesion and the transformation zone. This evidence led us to hypothesize that HPV testing performed intraoperatively immediately after the conisation should be negative when the lesion has completely been removed, or positive in the case of persistence of abnormal areas and, consequently, become an early marker of clearance or persistence of the disease.
The objective of this prospective study was to evaluate the feasibility and utility of intraoperative HPV (IOP-HPV) testing performed in the remaining cervix immediately after the conisation procedure for CIN2–3, and to determine whether it could be an early marker of treatment failure and become an alternative to the currently recommended HPV testing performed 6–12 months after treatment.
This prospective study included 132 patients consecutively diagnosed with CIN2–3 by colposcopically directed biopsy (CDB) or endocervical curettage (ECC) within 90 days before treatment, who underwent conisation by LEEP in the Department of Obstetrics and Gynaecology of the Hospital Clinic of Barcelona, from November 2008 to May 2010. All the women were tested for HPV before treatment.
Colposcopy was performed using a colposcope Olympus Evis Exera II CV-180 (Olympus, Barcelona, Spain) after preparing the cervix with 5% acetic acid. Colposcopy findings were described following the criteria of the International Federation for Cervical Pathology and Colposcopy.
ThinPrep Pap tests were prepared according to the manufacturer's protocol from PreservCyt samples using an automated processor (ThinPrep 2000). All specimens were stained using the Papanicolaou method and were evaluated by a skilled cytopathologist following the criteria of the 2001 Bethesda System.
Detection of HPV was performed using the commercially available Hybrid Capture 2 (HC2) system (Qiagen, Gaithersburg, MD, USA) on the material collected in liquid-based media (ThinPrep). All the samples were analysed for the presence of HPV types (16, 18, 32, 34, 36, 39, 45, 51, 52, 56, 58, 59, and 68). The test provides relative quantification of the viral load present in each individual sample. A relative light unit (RLU) of 1 (1.0 pg/ml) was used as the cut-off to classify a specimen as positive or negative for HPV. The RLU value of each sample was recorded.
The cervix was exposed using a speculum adapted for smoke evacuation. After delineating the area of abnormality with acetic acid and lugol iodine, 1 ml of 1% mupivacain was injected into each quadrant of the cervix. The loop size was selected according to the size of the area to be excised. In all cases the excision was performed under colposcopic guidance and included both the abnormal colposcopy area as well as the complete transformation zone. On suspicion of endocervical involvement, a second selective endocervical sweep was performed using a smaller loop (top hat). Exceptionally, when an exocervical lesion was too large to be excised in a single sweep, two or more systematic sweeps were performed. After the excision, careful inspection of the cervical bed was performed with selective coagulation of the bleeding areas by ball diathermy, if necessary. Subsequently an endocervical sample of the remaining cervix was collected with a cytobrush and kept in PreservCyt (Hologic Corp, Marlborough, MA, USA) for liquid-based cytology (LBC) and HPV testing. Finally, ECC was performed with a Kervokian curette, systematically scraping the whole endocervical surface.
The LEEP conisation specimens were anatomically oriented, pinned to a cork support, and fixed in 10% neutral buffered formalin. The excisional samples were thoroughly examined after processing the whole specimen in between three and 14 paraffin blocks (with a median of six). If there was more than one sweep, each was independently included. Surgical margins were identified with ink and carefully examined.
Post-treatment follow-up visits were scheduled at 6, 12, 18 and 24 months. Colposcopy, with cervical sampling with a cytobrush, was performed at each visit with PreservCyt (Hologic) for LBC and HPV testing. At each visit (immediate, 6, 12, 18, and 24 months) an LBC was performed. HPV was analysed in all patients intraoperatively and at 6 months. Thereafter, HPV testing was performed in all cases found to be positive intraoperatively or at 6 months. HPV detection was interrupted after two consecutive negative results.
At follow-up, patients diagnosed with high-grade or low-grade squamous intraepithelial lesion (HSIL or LSIL), or atypical squamous cells that cannot exclude an HSIL cytological result (ASC-H), were offered colposcopy and CDB when a lesion was identified. When the transformation zone was not completely visible or no colposcopy abnormalities were identified, an ECC using a Kervokian curette was also performed. Patients with atypical squamous cells of unknown significance (ASC-US) found in the Pap test, and/or an isolated positive HPV test, underwent a colposcopy, but CDB or ECC was only performed in cases of abnormal colposcopy or when the transformation zone was not completely visible.
Diagnosis of post-treatment CIN2–3 was established only after histological confirmation. Diagnosis of post-treatment CIN1 was established based on an LSIL result in LBC, and/or a CIN1 result in a CDB or ECC. All patients with histologically confirmed CIN2–3 diagnosed during follow-up underwent re-conisation. Other criteria of persistent disease, including involvement of the cone margin by the lesion or positive ECC performed after conisation, were not considered as criteria for systematic re-treatment. These patients were followed using the same protocol. Patients developing CIN1 after conisation were followed-up following the study protocol without treatment, except for patients with endocervical involvement, who underwent re-conisation.
Statistical analysis was performed using spss 18.0 (SPSS Inc., Chicago, IL, USA). The results are presented as absolute numbers and percentages or means and SD. An association between histological factors (cone margins and ECC) or tests (HPV testing and cytology performed intraoperatively and at 6 months) and residual or recurrent disease was analysed using either chi-square or Fisher's exact tests. The diagnostic efficacy of these tests to identify treatment failure was evaluated as sensitivity, specificity, and positive and negative predictive values (PPV and NPV). Univariate logistic regression analysis was performed to determine predictors of treatment failure with independent variables such as cone margin, ECC, and HPV testing and cytology, performed intraoperatively and at 6 months after treatment. Multivariate Cox regression models were applied to evaluate interaction effects with regard to recurrent or residual CIN2–3. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were also estimated. All statistical tests were two-sided, and P ≤ 0.05 was considered to be statistically significant.
The mean age of the 132 patients with histological diagnosis of CIN2–3 previous to conisation included in the study was 36.2 ± 9.6 years. Pre-treatment HPV detection was positive in 128/132 women (97.0%), and the mean viral load was 550.6 ± 637.2 RLU.
Histological analysis of the 132 LEEP specimens confirmed CIN2–3 in 96 women (72.7%), and showed CIN1 in 25 women (19.0%). In 11 cases (8.3%), no lesions were detected in the conisation specimen after thorough examination. The margin of the surgical specimens was involved with the lesion in 30/132 cases (22.7%; 28 out of 96 with a LEEP diagnosis of CIN2–3 [30.2%] and two out of 25 with CIN1 [8.0%], P = 0.036). The ECC performed after the procedure was positive in 11/132 patients (8.3%) and was non-evaluable in 26 women (19.7%).
Six months after treatment residual or recurrent CIN2–3 was histologically confirmed in seven women and CIN1/LSIL was found in eight women. At 12 months, CIN2–3 was identified in four women and CIN1/LSIL was found in two additional women. At 24 months, CIN2–3 was identified in one woman and CIN1/LSIL was identified in another additional patient. Overall, 24 months after treatment, residual or recurrent CIN2–3 had been identified in 12 patients (9.1%) and CIN1/LSIL had been identified in 11 (8.3%) patients. Residual or recurrent disease was histologically confirmed as recurring CIN2–3 in 12 women, and as recurring CIN1/LSIL in 5/11 women. One hundred and nine women were considered to be negative for residual or recurrent disease (91 with negative cytology and colposcopy; 16 with ASC-US cytology and/or colposcopy, but with a negative CDB and/or ECC).
All patients with residual or recurrent CIN2–3 were treated, and the recurrent disease was confirmed in the re-conisation specimen in all of them. Five out of 11 (45.4%) patients diagnosed with CIN1/LSIL at 6–12 months follow-up (two of them with a histological diagnosis of CIN1) spontaneously regressed, showing normal colposcopy and negative HPV testing and cytology at the 24-months visit. A re-conisation was performed in two patients with recurrent CIN1, and the lesion was confirmed in the cone specimen.
Table 1 shows the correlation between residual or recurrent CIN and the histological factors (cone margin status, post-conisation ECC) and IOP and 6-months tests (IOP-HPV testing and IOP cytology, and 6-month HPV testing and 6-month cytology). No correlation was observed between diagnosis in the conisation specimen (negative, CIN1, or CIN2–3) and the development of residual or recurrent CIN (P = 0.182, data not shown). In 13 women IOP cytology was not evaluable because of the scarce number or poor preservation of the cells. In contrast, IOP-HPV testing could be performed in all cases and no differences in the percentages of positive IOP-HPV testing between women having not evaluable or adequate results in the IOP Pap test were observed (3/13 or 23.1% versus 34/119 or 28.6%; P = 1). The samples obtained at 6 and 12 months were adequate for both cytology and HPV testing in all cases.
|Residual or recurrent disease||P|
|No (n = 109)||LSIL/CIN1 (n = 11)||CIN2–3 (n = 12)|
|Negative||90 (82.6)||10 (90.9)||2 (16.7)||<0.001|
|Positive||19 (17.4)||1 (9.1)||10 (83.3)||<0.001|
|Not evaluable||22 (20.2)||4 (36.4)||0 (0.0)||<0.001|
|Negative||83 (76.1)||7 (63.6)||5 (41.7)||<0.001|
|Positive||4 (3.7)||0 (0.0)||7 (58.3)||<0.001|
|Negative||85 (78.0)||9 (81.8)||1 (8.3)||<0.001|
|Positive||24 (22.0)||2 (18.2)||11 (91.7)||<0.001|
|Not evaluable||11 (10.1)||1 (9.1)||1 (8.3)||<0.001|
|Negative||89 (81.7)||9 (81.8)||4 (33.3)||<0.001|
|≥ASC-US||9 (8.2)||1 (9.1)||7 (58.4)||<0.001|
|Negative||88 (80.7)||1 (9.1)||1 (8.3)||<0.001|
|Positive||21 (19.3)||10 (90.9)||11 (91.7)||<0.001|
|Negative||91 (83.5)||3 (27.3)||4 (33.3)||<0.001|
|≥ASC-US||18 (16.5)||8 (72.7)||8 (66.7)||<0.001|
Eleven of the 12 women with residual or persistent CIN2–3 (treatment failure) were positive in the IOP-HPV testing (91.7%), and the same number (11/12, 91.7%) were positive in the 6-month HPV test. In ten out of 12 women with treatment failure (83.3%) at least one margin of the cone showed disease involvement, and 7/12 (58.3%) had positive post-conisation ECC. In contrast, only two out of 11 (18.2%) women with CIN1/LSIL diagnosed during follow-up had a positive result in the IOP-HPV testing, whereas 6-month HPV testing was positive in ten out of 11 (90.9%) women, recurring as CIN1/LSIL. All women who developed CIN1/LSIL after treatment that spontaneously regressed during follow-up were negative in the IOP-HPV testing and positive at 6 months. Only one out of 11 women (9.1%) recurring as CIN1/LSIL had a cone margin involved with disease, and none had a positive post-conisation ECC. The percentage of positive results in the cone margins and post-conisation ECC in women recurring as CIN1/LSIL was significantly lower than that observed in the women recurring as CIN2–3 (P < 0.001).
Figure 1 shows the correlation between the results of IOP and 6-month HPV testing. In 89 women (67.4%) the same result (70 negative–negative and 19 positive–positive) was observed in both tests. In 43 women (32.6%) a different result was observed in the IOP and 6-month HPV tests. The percentage of women with negative IOP-HPV and 6-month HPV testing was similar (71.2 versus 67.4%, respectively).
Table 2 shows the analysis of sensitivity, specificity, PPV and NPV, the 95% CIs for treatment failure for the different histological variables (cone margin and ECC), and the IOP or 6-month tests (HPV testing and cytology). The addition of cytology or cone margin status information to IOP-HPV testing did not increase the results of the test to predict treatment failure (data not shown). Simultaneous positivity for IOP-HPV and cone margin had a positive predictive value of 55.6%.
|Positive cone margin||83.3 (55.2–95.3)||83.3 (75.6–88.9)||33.3 (23.8–44.6)||98.0 (93.3–99.4)|
|Positive endocervical curettage||58.7 (31.9–80.7)||95.7 (89.6–98.3)||57.8 (32.0–80.0)||95.8 (92.1–97.8)|
|IOP-HPV testing positive||91.7 (64.6–98.5)||78.3 (70.1–84.8)||29.8 (22.4–38.3)||98.9 (93.5–99.8)|
|IOP cytology ≥ ASC-US||63.6 (35.4–84.8)||90.7 (83.8–94.9)||40.8 (24.7–59.1)||96.1 (91.9–98.2)|
|Six-month HPV testing positive||91.7 (64.6–98.5)||76.0 (67.9–82.5)||27.6 (21.2–35.2)||98.9 (93.3–99.8)|
|Six-month cytology ≥ ASC-US||66.7 (39.1–86.2)||78.3 (70.1–84.8)||23.5 (15.4–34.2)||95.9 (91.3–98.1)|
Table 3 shows the odds ratios and the 95% confidence intervals of the different histological factors (cone margins and ECC), and IOP or 6-months tests (HPV testing and cytology) associated with treatment failure. All had a significantly positive association with the probability of post-treatment CIN2–3. Tables 4 and 5 show the multivariate logistic regression of factors associated with treatment failure at IOP (Table 4) and after 6 months (Table 5). IOP and 6-month HPV testing showed the strongest association with treatment failure in the multivariate analysis (OR 15.40, 95% CI 1.58–150.42, and OR 12.40, 95% CI 1.13–135.49, respectively).
|Positive cone margin||25.00||5.72–121.87||<0.001|
|Positive endocervical curettage||31.50||6.87–144.49||<0.001|
|IOP-HPV testing positive||39.77||4.96–322.39||0.001|
|IOP-cytology ≥ ASC-USa||17.15||4.27–68.86||<0.001|
|Six-month HPV testing positive||31.58||3.92–254.70||0.001|
|Six-month cytology ≥ ASC-US||7.23||2.02–25.91||0.002|
|Positive cone margin||4.34||0.55–34.12||0.163|
|Positive endocervical curettage||5.73||0.78–41.81||0.085|
|IOP-HPV testing positive||15.40||1.57–150.42||0.019|
|IOP cytology ≥ ASC-USa||1.00||0.97–1.03||0.966|
|Positive cone margin||7.57||1.07–53.53||0.043|
|Positive endocervical curettage||7.83||0.96–64.04||0.085|
|Six-month HPV testing positive||12.40||1.13–135.49||0.039|
|Six-month cytology ≥ ASC-US||4.02||0.59–27.36||0.155|
Our study shows that HPV testing performed intraoperatively in the remaining cervix immediately after the removal of cervical tissue by LEEP conisation (IOP-HPV testing) might be an earlier risk marker of treatment failure. Interestingly, in the present study this technique detected treatment failure with a sensitivity of 91.7%, anticipating the results obtained with the conventional approach of HPV testing at 6 months. Previous studies have shown that HPV testing at 6–12 months is negative in approximately two-thirds of the patients, which is similar to the percentage of negative results in the IOP-HPV testing. Thus, although there is no previous evidence about the timing of HPV clearance after conisation, our study shows that clearance of the infection in the cervix occurs in a high percentage of cases immediately after the removal of the dysplastic epithelium, and suggests that the viral infection is mainly limited to the lesion.
The main strength of our report is that it is the first pilot study showing the feasibility and usefulness of a new approach for the earlier identification of treatment failure after conisation for CIN2–3. The main limitations are the small sample size and the fact that all the cases have been recruited and analysed in a single institution. Nevertheless, the main objective of this pilot study was to demonstrate the feasibility of this technique. Our results show similar values of accuracy for IOP and 6-month HPV testing, indicating that the technique is feasible, and stressing the need for subsequent studies confirming these promising data. A second limitation is the relatively short follow-up period reported in this series. Long-term follow-up is necessary to confirm the accuracy of these data and to determine the significance of persistent long-term infections. Nevertheless, all the patients were followed for at least 24 months, a period of time in which most of the recurrences are expected to develop. Finally, our study does not provide information on the type of HPV causing the lesion. These data are necessary to obtain stronger evidence on persistent disease (recurrence as CIN2–3) or newly acquired infections (recurrence as CIN1/LSIL), as suggested in our study.
A number of studies in the last decade have shown the value of HPV testing at 6–12 months in the surveillance of patients treated for CIN2–3. This test has become widely accepted in clinical practice, and a recent meta-analysis concluded that HPV testing at 6 months should be included in the post-treatment follow-up. Indeed, HPV has consistently shown sensitivity and negative predictive values close to 95–100%.[2-5, 7, 9-13, 15, 16] In agreement with our hypothesis, the accuracy of IOP and 6-month HPV testing to predict post-treatment CIN2–3 was similar and close to 95–100%.
In our study both IOP-HPV and 6-month HPV testing showed higher sensitivity than cytology, as has been reported in most previous studies.[2-5, 7, 9-13, 15, 16]As shown in previous studies based on HPV testing at 6–12 months, IOP-HPV testing showed a slightly lower specificity than IOP cytology.[25, 26] However, in our series, the addition of cytology to IOP-HPV testing did not substantially increase either the sensitivity or the specificity. Moreover, a significant percentage of intraoperative Pap tests were not evaluable, probably because of poor preservation and the scarce number of the remaining cells after the excisional procedure. Therefore, HPV testing is preferable to cytology as a test for treatment failure.
In this study 81.8% of women showing CIN1/LSIL during follow-up were negative in the IOP-HPV testing, suggesting that these lesions may represent new HPV infections acquired after treatment. In keeping with this hypothesis, only a few patients recurring as CIN1/LSIL had positive cone margins (9.1%), in contrast with the high percentage of positive margins in the women recurring as CIN2–3 (83.3%). Moreover, 45.4% of patients showing CIN1/LSIL at 6–12 months spontaneously regressed at 24 months, with a negative result for HPV testing and LBC. Thus, HPV testing at 6 months may not only reflect a persistence of the previous HPV infection associated with the treated lesion, but may also identify new post-treatment HPV infections, and most of these will clear spontaneously according to the natural history of HPV infection.
Although all patients had a CDB or ECC confirming CIN2–3 prior to conisation, a high-grade lesion was found in the cone specimen in only 72.7% of the women, whereas 19% showed CIN1 and 8.3% had no lesions in the surgical specimen. These results are in keeping with recent studies on the treatment of CIN2–3 lesions reporting on histological data.[20, 27] The overall rate of post-treatment CIN2–3 (9.1%) found in our study is also consistent with the data previously reported by other authors after a follow-up period of 24 months.[6-8]
In conclusion, IOP-HPV testing is feasible and useful, and may identify treatment failure earlier than conventional strategies. The favourable results of this new approach should encourage new research to confirm these data and assess whether IOP-HPV testing is cost-effective in the follow-up of patients treated for CIN2–3. The potential benefits of this strategy would be a reduction in the anxiety of many patients and a decrease of the costs to the healthcare system, which may be particularly useful in settings with limited healthcare resources.[28, 29] Further larger studies are needed to confirm whether IOP-HPV testing may be used to replace the conventional 6- or 12-month strategies.
The authors declare they have no conflict of interests.
AT and IA participated in the study design, data collection, data interpretation, reference search, and writing of the article. PF participated in the study design, data collection, data analysis, reference search, and writing of the article. LR-C participated in the data collection, data analysis, data interpretation, reference search, and writing of the article. MdP participated in the data collection, data analysis, data interpretation, reference search, and writing of the article. RN, MC and ÁR participated in the data collection, data analysis, and writing of the article. PC participated in the data analysis, data interpretation, reference search, figure design, and writing of the article. JP, JB and JR participated in the data interpretation and writing of the article. JO participated in the study design, data analysis, data interpretation, reference search, figure design, and writing of the article.
Approved by the Ethics Committee of Clinical Research of the Hospital Clinic on 10 April 2008 (registry 2008/4188).
Supported in part by grants PI09/1084 and PI09/1524 from the Fondo de Investigaciones Sanitarias and by a grant from the Agéncia de Gestió d'Ajuts Universitaris i de Recerca - Generalitat de Catalunya (2009SGR 1099).
The authors are grateful to Ms Roser Esteve and Ms Lorena Marimon for their technical assistance with the HC2 test and the cytological evaluation. We thank Ms Rosana Millan for her help with the histological processing of the LEEP samples, and Ms Marta Sanchez for her administrative support. We thank Ms Donna Pringle for her English revision of the article.