HPV-induced cervical neoplasia progresses through a predictable natural course of disease, starting with mild cervical intraepithelial neoplasia (CIN1) and developing into severe degrees (CIN3) and eventually to invasive cervical cancer. A positive finding on initial screening for cervical neoplasia results in further diagnostics involving sampling with an endocervical brush (EB) or endocervical curettage (ECC). There exists a considerable variation in the choice of test method for obtaining these endocervical or cervical canal samples.
This meta-analysis aimed to retrospectively determine the diagnostic test accuracy (DTA) and patient discomfort of the EB and ECC for the detection of cervical neoplasia in women with any indication for colposcopy, biopsies, and endocervical sampling. All diagnostic studies and randomized clinical trials that included women with abnormal cytology results in cervical cancer screening or women with red flag symptoms undergoing colposcopy, biopsy, EB, or ECC were included. The target condition was cervical neoplasia, and the reference standard was the final histological result. A bivariate random-effects model was used to determine summary estimates of sensitivity and specificity with 95% confidence intervals (CIs). A meta-analysis assessing summary relative risk for inadequate samples was also conducted.
A total of 7 studies were included in the meta-analysis with a total sample size of 1097 women. Three studies were randomized clinical trials, and 4 were cohort studies, and they took place between 1988 and 2013. A DTA meta-analysis for summary sensitivity and specificity found an overall pooled sensitivity of 81% (95% CI, 48–95) for EB and 70% (95% CI, 42–89) for ECC. The overall pooled specificity was 73% (95% CI, 36–93) for EB and 81% (95% CI, 56–94). The overall relative risk was found to be 2.53 (95% CI, 0.58–11.0; P = 0.215) for ECC compared with EB for inadequate samples. Two studies reported patient discomfort as an outcome, and only a single study reported a significant difference between ECC and EB in favor of ECC (visual analog scale score 2.55 vs 1.99).
The results of this meta-analysis comparing the DTA of EB and ECC found no significant difference between EB and ECC in DTA, inadequate sampling rate, or adverse effects. The overall sensitivity was found to be slightly higher for EB than for ECC and the specificity to be slightly higher for ECC than EB, although these results were not statistically significant.