Endocervical sampling using brush versus curette: a single centre experience and literature review

Abstract Endocervical sampling is performed traditionally with an endocervical curette (ECC). The current study objective is to compare the histopathological performance of endocervical brush (ECB) and endocervical curette (ECC). A retrospective review was performed including patients included that underwent colposcopy with endocervical sampling using either method. A total of 127 samples were obtained with ECC and 98 with ECB. Histopathological diagnosis was obtained in 124 (97.6%) ECC samples and in 94 (95.9%) ECB samples (p = 0.46). The incidence of benign results was similar between ECC and ECB (117 (92.1%) versus 88 (89.8%) respectively (p = 0.28)). When combining information from endocervical sampling with cervical biopsies, the detection rate of high-grade pathologies was similar between the groups with 14 cases (17.7%) for ECC and 8 cases (17.0%) for ECB (p = 0.43). A scope review of the topic was performed, illustrating that studies favour either method. In conclusion, ECB and ECC perform similarly for providing a histopathological diagnosis on endocervical samples. IMPACT STATEMENT What is already known on this subject? Endocervical samples in colposcopy were traditionally obtained using an endocervical curette. Similarly, a brush can be used for histological sampling of the endocervical canal. However, it is unclear how the ability to obtain a histopathological diagnosis compares between the two techniques. What do the results of this study add? This single-institution experience with using endocervical brush and curette for endocervical sampling finds that both methods are acceptable and have a high ability to provide a histopathological diagnosis. Precisely, 4.1% of brush and 2.4% of curette samples had insufficient tissue. What are the implications of these findings for clinical practice and further research? The endocervical brush is an adequate sampling method for colposcopy, and can be safely used instead of the curette, based on clinician preference. Further studies could investigate how these methods compare from a patient perspective.


Introduction
Endocervical curettage represents a sampling technique used in colposcopy during investigations for abnormal Papanicolaou (Pap) tests.Indications for endocervical assessment include high-grade intra-epithelial lesion (HSIL) or atypical glandular cell (AGC) pathologies (Feltmate andFeldman 2020, Perkins et al. 2020).As well, endocervical curettage was previously shown to provide additional histopathological information in certain clinical contexts such as when the squamocolumnar junction is not visible (Solomon et al. 2007).
While endocervical curettage has traditionally been obtained using an endocervical curette (ECC), using an endocervical brush for this technique (ECB) has been described as less painful for the patients (Weitzman et al. 1988, Goksedef et al. 2013, Doo et al. 2016) and easier to perform by physicians (Undurraga et al. 2017).The samples obtained with the brush represent an ongoing debate in terms of diagnostic yield, with some studies showing similar results to the curette (Dunn et al. 2000, Klam et al. 2000, Goksedef et al. 2013), and others showing discrepancies (Gibson et al. 2001, Boardman et al. 2003, Doo et al. 2016).While the ASCCP supports performing the sampling with either ECB or ECC (Perkins et al. 2020), other sources recommend using both techniques concomitantly (Feltmate and Feldman 2020).
The purpose of the current study was to investigate whether the feasibility and ability of ECB to provide a histopathological diagnosis is similar to that of ECC at a single tertiary care centre, as well as to evaluate whether patients with discordant results (i.e.abnormal Pap tests and normal endocervical samples and cervical biopsies) had significant pathologies during a 12-month follow-up period.A secondary aim was to conduct a scope review of the literature comparing the two endocervical histopathological sampling techniques.

Methods
The current study is a retrospective cohort study including a medical record review of all patients that underwent endocervical sampling during a two-month period in 2020 at the Jewish General Hospital in Montreal, Quebec.The Research Ethics Committee of the Jewish General Hospital granted ethics approval for this study (project 2021-2557).
The inclusion criteria for the study are patients who underwent colposcopy and had endocervical sampling performed using either a Kevorkian curette (ECC) or a cytobrush (ECB, Innovatek Medical) (Figure 1).Patients were included in the study if they underwent endocervical sampling with a technique that is clearly documented in the chart.The specimen label was endocervical curettage with no specification of the technique used.A gradual change in practice occurred at our institution in 2019-2020, where some physicians transitioned from using ECC to using histological ECB.The specific technique for ECB sampling involved a back-and-forth movement in the endocervical canal while concomitantly turning the brush counter-clockwise several times.Of note, the ECB is 196 mm long and the diameter of its cone-shaped brush is 4 mm at the narrowest part and 6 mm at the widest part while the length of its brush part is 21 mm.For ECC sampling, the Kevorkian curette was inserted in the endocervical canal, and a back-and-forth movement was performed in each quadrant of the canal.The samples obtained with either technique were placed in 10% formalin and sent for histopathological diagnosis.Patients were excluded from the study if the sampling method was not indicated in the chart or if both sampling methods were used during the same visit.If a patient underwent two different samplings during the study period during two different visits using the two different techniques, there were still included in the study.If this last scenario occurred, the two endocervical sampling results were entered as two separate data points and analysed according to the technique used.
Data were extracted from the computerised database containing the electronic medical records by one researcher and verified by another researcher.The following demographic and clinical characteristics were retrieved: age, gravidity, parity, menopausal status, smoking status, method of contraception, and human papilloma virus (HPV) vaccination status.The retrieved colposcopy results included Pap tests, cervical biopsies, endocervical samples, as well as endometrial biopsies, if applicable.Depending on the clinical scenario or whether discrepancies in results were encountered, colposcopy results up to 18 months prior and 12 months after the endocervical sampling were reviewed to assess whether high grade lesions were missed and whether access to care was delayed in either technique.Analysis was performed using the Statistical Package for Social Sciences (SPSS version 26, IBM New York) to obtain averages, standard deviations, means, as well as t-test and p-value assessments.
For the scope review, a librarian designed the search strategy (Appendix 1) and searched MEDLINE, Embase, CINAHL from database inception until August 2022.The Covidence platform was used for screening and inclusion (Vh 2022).Two reviewers (C.M. and R.L.) independently screened the articles and solved disagreement by discussion.Included studies had to be published in English and to have a cohort, cross-sectional, case series, or randomised controlled trial design.Studies were excluded if they were only available in abstract form, and if they were designed as case reports, systematic reviews and editorial comments.Studies had to compare histopathological endocervical samples for both techniques for any outcome.Studies were excluded if only cytology samples were obtained with ECB.Information extraction was performed independently by the two reviewers and disagreements were solved by discussion (Figure 2).

Results
A total of 212 patients met the inclusion criteria, of which 117 patients were in the ECC group and 95 in the ECB group.The demographics of the groups are included in Table 1.The total number of samples obtained was 127 ECCs and 98 ECBs.The occurrence of a non-diagnostic sample was rare for both ECB (4.1%) and ECC (2.4%), with no statistical significance between the two groups (p ¼ 0.46).Of the four patients with insufficient tissue in the ECB group, two were post-menopausal, one had prior cryotherapy, and one was nulliparous with a history of LEEP (loop electrosurgical excision procedure).Of the three patients with insufficient tissue in the ECC group, two were post-menopausal.
The incidence of pathologies such as strips of neoplastic endocervix (SONE), low-grade intra-epithelial (LSIL) lesion, high-grade intra-epithelial lesion (HSIL), and adenocarcinoma in situ (AIS) was not statistically different between the groups (p ¼ 0.28).When looking at the detection rate of significant pathologies (HSIL, AIS, or malignancy) based on the complete colposcopy assessment including endocervical samples and cervical biopsies, the detection rate was similar between both groups (p ¼ 0.43).The results are illustrated in Table 2.When discrepancies occurred between an abnormal Pap test result and the combination of endocervical sampling and cervical biopsy, abnormal Pap tests were reviewed up to 18 months prior to and 12 months after the target endocervisample date.Overall, no significant pathology was missed or resulted in a delayed treatment.
For the secondary objective, a total of 163 studies were screened after removing four duplicates.The title and abstract screening showed a 0.88 proportionate agreement and a 0.60 Cohen's kappa between the two reviewers.Full screening of 25 articles showed a 0.84 proportionate agreement and a 0.63 Cohen's kappa coefficient between the two reviewers.A total of 7 articles were included in the scope review.Figure 1 reports the literature review using the PRISMA protocol (Page et al. 2021).A summary of their comparative findings for histopathological endocervical sampling is illustrated in Table 3. Overall, 3 out of 8 (37.5%) studies recommend ECB, 3 (37.5%)studies either, one (12.5%)study both, and one (12.5%)study ECC.

Discussion
Although the ASCCP recommends the use of either ECB or ECC for obtaining an endocervical canal sample (Perkins et al. 2020), other sources recommend using both methods to obtain an adequate sample (Feltmate and Feldman 2020), resulting in a controversy.The present study illustrates that the cytobrush has a similar ability to provide sufficient tissue for histopathological diagnosis when compared to the traditional endocervical sampling method with a Kevorkian curette.Our rate of insufficient tissue was 4.1% in ECB group, which compared to the rates obtained in two previous randomised controlled trials (Klam et al. 2000, Undurraga et al. 2017), but was lower than other reported rates of 9.1-22% (Gibson et al. 2001, Goksedef et al. 2013).Such high rates of inadequate samples have also been reported for ECC, with values reaching 9.6-14.3%(Tate and Strickland 1997, Goksedef et al. 2013, Undurraga et al. 2017).The difference between observed rates in the literature might be related to the sampling technique itself, as well as to the lack of standardised pathological criteria in what constitutes an adequate sample or a scant sample.
A recent systematic review compared diagnostic accuracy for ECB and ECC and found higher sensitivity (81% vs. 70%) and lower specificity (73% vs 81%) for ECB compared to ECC (Damkjaer et al. 2022).Our study differs from the recently published review in that it focuses specifically on endocervical brush technique to obtain a histopathologic rather than a cytological sample.Despite the limited data with heterogeneity in terms of outcomes analysed, the literature provides reassuring data that ECB appears to be at least equivalent to the standard ECC.
Although the study did not assess measures of patient perspective such as pain experience or physician perspective such as ease of the procedure, the observation driving the study at our institution was that ECB constitutes a less painful and easier to perform technique.The decreased pain benefit was previously documented in the literature with improved patient-reported VAS (visual assessment scale) pain scores (Goksedef et al. 2013), as well as physician perception of patients' pain experience (Undurraga et al. 2017), when ECB is performed.As well, gynaecologists previously described the endocervical sampling with ECB as easier to perform, with ECC being more operator dependant and more difficult to perform especially in patients with stenotic os or post-menopausal patients (Martin et al. 1995, Undurraga et al. 2017).Overall, the pain associated with endocervical canal sampling can be difficult to assess as the procedure is short-lasting and coincides with other uncomfortable procedures such as cervical biopsy and speculum placement.
The current study includes several limitations, for instance, it did not directly compare the two methods in the same group of colposcopy patients.The design was observational rather than a prospective randomised trial, and the number of patients with subsequent treatment procedure was small and did not allow for comparison with final pathology.As such, calculations of sensitivity and specificity of ECB versus ECC were not possible in the current study.An ideal study would involve a randomised control trial where all patients undergo both ECB and ECC, with a randomised order for sampling.These results would be compared to results of LEEP or cone biopsy, and the pathological criteria for what constitutes an adequate sample would be standardised.To have adequate power, such a design would need to involve multiple centres and would be rather challenging to conduct from a time and a financial perspective.One of the advantages of our study was the ability to review clinical follow-up details about the patients who had discrepancies in their results or insufficient samples and provide the reassurance that no patient experienced an adverse colposcopy outcome.To our knowledge, this study is the only one to combine a retrospective cohort study and a literature scope review.This allows to compare our centre's findings to those of different colposcopy centres.

Conclusion
In conclusion, ECB is an adequate endocervical sampling method to obtain a histopathological diagnosis as it performs similarly to the endocervical curette.