Objective
Endocervical curettage (ECC) specimens obtained during colposcopy can detect cervical cancer and precursors otherwise missed by biopsy alone, but the procedure can be painful and reduce compliance with needed follow-up. ECC is routinely performed in the Calgary Health Region colposcopy clinics, permitting a look at its real-world utility.
Study Design
We analyzed pathology and colposcopy reports from 2003 to 2007. We calculated the added diagnostic utility of ECC compared with cervical biopsy alone.
Results
ECC increased the diagnostic yield of cervical intraepithelial neoplasia grade 2 or worse (cervical intraepithelial neoplasia [CIN]2+) in 1.01% of 13,115 colposcopically guided biopsy examinations. Therefore, 99 ECC specimens were taken to detect 1 additional CIN2+. ECC detected 5.4% of 2443 CIN2+ subjects otherwise missed by biopsy alone. Utility was greatest among women aged 46 years or older referred after a high-grade cytology.
Conclusion
ECC is rarely informative when used routinely in colposcopic practice. Older women referred after high-risk cytology benefit most from ECC.
Each year millions of women in the United States and Canada are referred to colposcopy after abnormal cytology or positive human papillomavirus testing. The colposcopist will take biopsies of visualized lesions and possibly sample the endocervical canal using a curette to rule out the presence of hidden cervical intraepithelial neoplasia (CIN). This latter procedure, the endocervical curettage (ECC), involves circumferentially scraping the endocervical canal. Although several reviews have summarized the findings of numerous heterogeneous studies, the use of ECC remains controversial.
The most recent data indicate that ECC might increase the sensitivity of the colposcopic visit by detecting 2-6% of high-grade CIN or cancer that otherwise would have been missed by cervical biopsies alone. Analyzed from a related but different angle, these studies have also shown that ECC increases the diagnostic yield of colposcopic visits by 1-6% (increasing the total number of cases of high-grade CIN or cancer per colposcopic exam). Yet ECC is often the most painful part of the colposcopy procedure, and pathologic interpretation can be difficult because of small, fragmented, and poorly oriented specimens with deficient stroma. In addition, during the collection process, the presence of an ectocervical lesion might contaminate the ECC specimen, resulting in more extensive treatment of a suspected endocervical lesion in which none exists.
With the exception of its contraindication in pregnant women, the use of ECC varies widely among providers. Some colposcopists limit its use to indications for endocervical sampling recommended as acceptable or preferred by professional guidelines such as when managing a woman with high-grade squamous intraepithelial lesion (HSIL) cytology, unsatisfactory colposcopy, atypical squamous cells of undetermined significance (ASC-US), or low-grade squamous intraepithelial lesion (LSIL), with or without visualized lesions, or suspicion of glandular extent.
Another option is to sample the endocervix using an endocervical brush instead of curettage. Still others uniformly collect ECC in every colposcopy examination of nonpregnant women, regardless of age, adequacy of colposcopy, or cytology result. Further identification of subgroups of women most likely to benefit from ECC is required, but studies have lacked appropriate data or statistical power for such comparisons. Evidence suggests that ECC is less advantageous in younger, nulliparous women referred for low-grade abnormalities with satisfactory examinations.
The Calgary Health Region and the Alberta Cervical Cancer Screening Program in Alberta, Canada, have an extensive data collection system that records histopathology, cytopathology, and colposcopic and patient characteristics for all colposcopy examinations conducted. Because ECC is routinely taken at essentially all colposcopy examinations in the outpatient colposcopy clinics, we had a rare opportunity to assess the benefit of ECC in thousands of colposcopic examinations. This large analysis provides many more outcomes than previous studies and permits careful stratification to identify subgroups of women for whom ECC is most valuable, without selection bias.
Materials and Methods
The Calgary Health Region provides services to a population of approximately 1.2 million. Colposcopy, cytopathology, and histopathology are regionalized services with uniform practice guidelines and standards. Deidentified pathology reports were obtained from Calgary Laboratory Services for histological specimens collected at colposcopy exams at the Women’s Health Centre and the Tom Baker Cancer Centre colposcopy clinics and read between Jan. 1, 2003 and Dec. 31, 2007.
Using the patient number, records from the pathology database were linked with the colposcopic examination database to abstract the colposcopic impression and whether the examination was satisfactory, when available. In addition, personal characteristics of women were obtained including gravidity, parity, use of contraception, and date of last menstrual period. The record review received human subjects research approval from the Conjoint Health Research Ethics Board Review, University of Calgary, and Calgary Health Region and was considered exempt from review by the National Cancer Institute, National Institutes of Health.
The 60,537 histopathology specimens corresponded to 39,476 examinations where up to 4 types of specimens were collected. Types of specimens included cervical biopsy, ECC, loop electrosurgical excision procedure (LEEP), endometrial biopsy, vulvar biopsy, vaginal biopsy, and cervical brush. If more than 1 of a specimen type was taken at the examination and analyzed (for example, multiple cervical biopsies), we considered the worst histopathology reading as a final diagnosis for that specimen type. If a specimen was read more than 1 time, we considered the final reading as the final diagnosis for that specimen.
This analysis was not conducted at the patient level, but we instead analyzed each examination in which a cervical biopsy and ECC specimen was collected (n = 13,476), under the assumption that the diagnostic utility of ECC did not depend on the women per se. Of note, we identified 19,372 examinations in which only an ECC specimen was taken and 2030 examinations in which only a cervical biopsy specimen was taken (0.6% and 4.7% were diagnosed CIN grade 3 [CIN3] or worse including 2 and 8 carcinomas, respectively).
Each woman contributed on average 2.2 examinations to the dataset, and women who contributed 1 examination did not differ from those contributing more than 1 with regard to age, parity, or oral contraceptive use. Among these examinations we excluded those in which another type of histopathology specimen (eg, endometrial biopsy) was taken (n = 361, 2.7%) because the focus of this analysis was on the value added by ECC to colposcopically guided biopsy for diagnosing cervical precancer and cancer. Our final analytic sample was 13,115 examinations.
Cytopathology records for specimens processed at Calgary Laboratory Services up to 2 years prior to the date of the reading of the histopathology specimen were retrieved. Cytology readings were standardized and categorized according to the Bethesda Classification System: ASC-US, atypical squamous cells, cannot rule out HSIL (ASC-H), atypical glandular cells of undetermined significance (AGUS), LSIL, and HSIL.
For this analysis, cytology interpretations of ASC-H were categorized with HSIL because of their rarity and the high risk of cervical precancer and cancer associated with them. The colposcopy visit was considered a referral visit if a cytopathology result was within 270 days of the examination and the result was unsatisfactory or abnormal. The visit was considered follow-up to a previous diagnostic or treatment colposcopy visits and not a referral visit if no cytopathology result was identified for the patient, the cytopathology visit was 270 days or more before the histopathology reading, or the cytopathology result was normal.
We compared the final histopathology diagnosis for ECC specimen with the final histopathology diagnosis for the cervical biopsy specimens. Results were categorized as unsatisfactory, less than CIN grade 2 (CIN2), CIN2, CIN3, or cancer.
For examinations in which both ECC and cervical biopsy specimens were taken, our 2 primary outcomes were: (1) the proportion of examinations in which ECC detected CIN2 or worse that would have been missed by cervical biopsy alone (diagnostic yield) and the number of needed to test (NNT) with ECC to detect 1 additional case of CIN2 or worse (calculated as the inverse of the diagnostic yield) and (2) the proportion of CIN2 or worse diagnoses that would have been diagnosed as less than CIN2 based on cervical biopsy had ECC not been performed (additional case detection). We also considered CIN3 or worse as a more scientifically rigorous precancerous endpoint and better surrogate of cervical cancer risk.
Findings were stratified by the woman’s age at the examination, parity, method of contraception, referral cytology, and colposcopy impression when available. Because satisfactory visualization of the transformation zone is closely associated with menopausal status, we imputed this variable based on their age and date of last menstrual period. Women under age 46 years were classified as premenopausal, women aged 46-59 years with less than 1 year since last menstrual period were classified as perimenopausal, and women either over age 60 years or aged 46-59 years with 1 year or more since last menstrual period were classified as postmenopausal.
Logistic regression and contingency analyses with χ 2 statistics were used to evaluate the influence of potential confounders. All analyses were conducted using Stata 11.0 analytic software (StataCorp LP, College Station, TX).
Results
We identified 13,115 colposcopy examinations in which cervical biopsy and ECC specimens were collected and 79.4% had concordant diagnoses ( Table 1 ). Compared with cervical biopsies, ECC specimens were more likely to be unsatisfactory (4.2% vs 1.2%, McNemar’s χ 2 , P < .001, respectively) and less likely to be diagnosed as CIN2 or worse (4.3% vs 17.6%, McNemar’s χ 2 , P < .001, respectively). Of the 16 cancers detected, 15 were diagnosed as CIN3 or worse by both cervical biopsy and ECC and 1 was diagnosed as less than CIN2 by ECC.
| Histopathology result of cervical biopsy | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Unsatisfactory | Less than CIN2 | CIN2 | CIN3 | Cancer | Total | |||||||
| Histopathology result of ECC | n | Column % | n | Column % | n | Column % | n | Column % | n | Column % | n | Column % |
| Unsatisfactory | 6 | 3.8 | 464 | 4.4 | 44 | 3.2 | 40 | 4.4 | 0 | 0.0 | 554 | 4.2 |
| Less than CIN2 | 150 | 94.3 | 10,052 | 94.4 | 1184 b | 85.9 b | 612 b | 66.7 b | 1 b | 6.7 b | 11,999 | 91.5 |
| CIN2 | 3 | 1.9 | 86 a | 0.7 a | 127 | 8.3 | 44 b | 4.8 b | 0 b | 0.0 b | 260 | 2.0 |
| CIN3 | 0 | 0.0 | 43 a | 0.4 a | 23 a | 1.5 a | 221 | 24.1 | 7 b | 46.7 b | 294 | 2.2 |
| Cancer | 0 | 0.0 | 0 a | 0.0 a | 0 a | 0.0 a | 1 a | 0.1 a | 7 | 46.7 | 8 | 0.1 |
| Total (row %) | 159 | 1.2 | 10,645 | 81.2 | 1378 | 10.5 | 918 | 7.0 | 15 | 0.1 | 13,115 | 100.0 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
