Endocervical curettage for diagnosing high-grade squamous intraepithelial lesions or worse in women with type 3 transformation zone lesions: a retrospective, observational study

Cervical cancer is the sixth most common malignancy and eighth most common cause of cancer-related mortality in Chinese women, with 119,300 new cases and 59,060 deaths from cervical cancer in 2020. This represents 5% of all female cancer-related deaths in China and both indicators highlight a continuous upward trajectory over the past five years [1]. Early interventions in cervical cancer development rely on adequate, accurate, and efficient screening. The current gold standard for cervical cancer diagnosis in patients with abnormal cytologies or with human papillomavirus (HPV) is colposcopic tissue removal from lesions and histopathological testing. However, for between 2–12% of women with transformation zone type 3 (TZ3) lesions, atrophy analysis and conventional lesion-directed ectocervical biopsies fail to detect occult lesions in the cervical canal [2‐4]. Undiagnosed occult lesions can progress rapidly and can have dire consequences.

At present, endocervical curettage (ECC) is generally only used in clinical practice for adjunctive biopsies when part or all of the TZ is not visible. However, in China, there are no consistent indications for screening high-risk subgroups. In fact, many Chinese clinicians perform ECC simply to avoid missing HSIL + cases, despite knowing that not all women will benefit from the procedure. The value of ECC in HSIL+ detection remains controversial, with a detection rate of only 1–9.3% observed in previous studies [4, 5]. ECC is also invasive, and repeated pulling and scraping in the cervical canal can be painful for patients and reduce the likelihood of attending follow-up appointments [6‐8]. Additionally, specimen cosistency obtained by ECC is only moderate (κ = 0.52) [9] and interobserver agreement is poor [10]. Numerous studies have also found that CIN2+ is more prevalent in women with TZ3 lesions, compared to those without (8–27% vs 1.3–12%) however, this is only when colposcopy is satisfactory [9, 11, 12]. The aim of this study was to determine whether the detection performance and ECC accuracy are sufficiently high for TZ3 lesion cases to outweigh the disadvantages.

ECC is often used as an adjunct to biopsy to diagnose HSIL+ in women with TZ3 lesions who have cervical canal atrophy, and sometimes to detect occult lesions that are difficult to observe with a colposcope. However, the value of ECC as a complementary test in the clinical setting has been controversial. The purpose of this study was to identify women with TZ3 lesions who would benefit most from ECC for HSIL+ and to assess the advantages of this test. Analysis of data from 1,905 women whose ECC and biopsy information was clear revealed that routine ECC following biopsy detected 20.5% of the HSIL+ cases. The HSIL+ diagnostic yield associated with ECC was 0.8% which suggests that to detect one additional HSIL+ case, missed by biopsy, would require screening of 125 women. This means, many women would be subjected to the protracted pain and discomfort associated with unnecessary ECC procedures. Therefore, it would be unwise to perform ECC in all women with abnormal cytology or HPV test results, and a high-risk group should be selected.

Stratified results suggest that age, cytology, HPV status and colposcopic impressions are risk factors for HSIL+ detection with ECC. HSIL cytology, HPV 16/18 infections, and a high-grade colposcopic impression favourably impacted the detection rate in middle-aged and older women. Specifically, we found that ECC detection of HSIL+ was more likely in women over 50 years of age and those aged 40–49 years compared to those younger than 30 years. This finding reflects the age-related decrease in hormone levels and atrophy of the cervical canal to the point where part of all of the TZs become invisible. This is consistent with the findings of Schneider et al. [15] and Shepherd et al. [16], who found that ECC was of most benefit to women older than 50 years, in terms of decreasing the incidence of cervical cancer and mortality. In the latest standarizing colposcopy guidelines, it is suggested to American practitioners that all patients over 40 years old should initially choose cervical curettage [8]. However, there is still no united view of the cut-off age and studies of more varied ethnicities are required to understand this.

In our study, the cytological subgroups of HSIL and ASC-H were the recommended group for doing the ECC procedure compared to normal cytological women. Poomtavorn et al. also concluded that ECC should not be performed in women with ASC-US or LSIL in view of the extremely small risk of HSIL+ [17]. In our study, Multivariate analysis revealed that the probability of diagnosing HSIL+ with ECC was approximately twice as much in women who were HPV 16/18-positive compared to those who were HPV 16/18-negative. The risk for those with non-16/18 but high-risk HPV cases was 0.898. HPV 16/18 infection has been found in 70% of women with invasive precancerous lesions [18, 19]; however, 13 HPV subtypes are carcinogenic when non-16/18 high-risk HPV is included [20, 21]. We also found that seven of 160 patients who were HPV-negative had HSIL according to ECC. 11 of the 234 with HSIL detected by biopsy, were HPV-negative. This evidence supports the notion that HPV detection alone increases the number of missed diagnoses. Bogani et al.'s research illustrates the universality of high-grade cervical lesion in 15% of high-risk-HPV-negative patients after conization [22].

Colposcopic impressions revealed HSIL+ in 36 times as many instances in severely ill women compared to healthy individuals. These findings suggest that when screening a large group of women, cytological AS-CUS, negative HPV status, and a low-grade colposcopic impressions for women younger than 30 years of age may help prevent unnecessary flesh contusions. Furthermore, they suggest a need to incorporate these prognostic factors into a convenient risk assessment tool for accurate, precise, and standardized quantitative clinical decision-making. The five-factor nomogram developed by Li et al. was found to have a high degree of discrimination and calibration and performed well in terms of utility in an internal and external validation set of 2,088 patients [23].

In previously reported studies, the rate of detection of HSIL+ by ECC has ranged widely from 1.1% to 18.5% [4, 5, 24]. In our present study, the additional detection rate was 0.8%, which is comparable with the figure of 0.6% in another Chinese study [25]. The variable detection rates could reflect use of different study endpoints and patient populations. The majority of the women in our study visited a clinic rather than a screening facility, and the characteristics of CIN2 lesions, are frequently constrained, small, and have low reproducibility, making a diagnosis uncertain and challenging. Given this low homogeneity restriction, integrating techniques into ECC may be a viable option to bring clarity to diagnostic classification. Shah et al. applied p16/Ki67 dual stain to 58 ECC specimens and diagnosed 18 additional cases of HSIL [26]. Maximiliano et al. attempted to overcome the drawbacks associated with the sparseness of ECC tissue using cell concentration methods, which may be a suitable strategy for qualitative improvement of the sensitivity of ECC and its diagnostic value [27]. Furthermore, Rubesa-Mihaljevic et al. found ECC sensitivity was higher for samples with abundant materials compared to samples with relatively few [28]. Uses of inexpensive and less painful sampling instruments to increase the rate of satisfactory specimen and reduce the reliance on senior physicians are additional ways of increasing patients’ compliance with procedures and improving lesion detection rates [29]. These measures will not only help to increase the accuracy of ECC but also make it possible for clinical examinations to be effectively integrated in low-income and middle-income countries with limited resources, striking a balance between the costs of clinical examinations and the availability of experienced practitioners.

Without high-risk population screening, only 0.8% (15/1905) of the women in the current study had TZ3 lesions, which resulted in an unsatisfactory overall gain. However, this figure increased to 3.4% in the subgroup with cytological HSIL and to 2.3% in the subgroup with low-grade colposcopy impressions. Following a Chinese study, only four additional cases of HSIL+ were detected by ECC, all of whom were in patients aged at least 40 years and with TZ3 lesions [30]. Gage et al. reported an additional detection rate of 5.4% (132/2433) in all cases with CIN2+ pathology when ECC was used and observed that the less visible the TZ, the higher the additional cytology detection rate [4]. This suggests that colposcopists should concentrate on older but still accurate colposcopy impressions when assessing patients with TZ3 lesions. A supplemental ECC method should therefore be used to clarify diagnosis if lesion level in the visible area is not compatible with the risk assessment based on screening findings. This is also in line with the most recent American Society of Colposcopy and Cervical Pathology recommendations, which state that ECC is preferable for non-pregnant women in whom colposcopy is insufficient and there is a slight risk of morbidity but no obvious lesion [31].

In addition to use of ECC for diagnosis, some guidelines suggest large loop excision of the transformation zone (LLETZ) in patients with TZ3 and abnormal cytological results to remove all the transformed areas (not just the diseased portions) and the cervical canal with sufficient length to ensure that there is no diseased tissue within at least 2–3 mm from the cutting edge and remove all the recesses in the transformed areas to reduce the risk of missed diagnosis [31]. A previous study found that diagnostic LLETZ detected six cases of CIN2+ in 40 patients with persistent HPV infection and normal cytology that were not found by biopsy, giving a detection rate of 15% [32]. Another study reported that 25% (n = 6) of 24 women with HPV-positive/normal cytology who underwent diagnostic conization for TZ3 had CIN2+ [33]. Moreover, preoperative diagnosis by conization protects against recurrence after radical hysterectomy [34, 35]. It is noteworthy that an ECC abnormality is an independent risk factor for residual disease after LLETZ [35, 36]. However, the latest guidelines stipulate that if a subsequent resection is planned, the sampling device should not be placed in the cervical canal [8]. Therefore, attention should be paid to the results of ECC, follow-up monitoring should be strengthened, and risk factors should be considered in the selection of individualized diagnostic and treatment methods, which may play an active role in cervical intraepithelial neoplasia and residual tumors after surgery.

This study's biggest strength is that it adds to the evidence regarding the value of ECC for women with TZ3 lesion, patients since they had the highest rate of HSIL+ discovery. Our results in a large research population drawn from seven provinces in China underscore the abuse risk of ECC. However, the study has some limitations. First, this study was retrospective so the possibility of bias in the data cannot be excluded. Second, the majority of the study population in this study was middle-aged and older women with TZ3 lesion, with only a small number of young women in the specific population.

This study was performed in an effort to improve the rate of detection of HSIL+ in women with TZ3 lesions without subjecting these women to unnecessary discomfort. We identified high-risk groups in whom ECC is warranted, namely, middle-aged and elderly women with a high-grade colposcopic impression, high-grade cytology, and HPV 16/18 infection. These findings may reduce the number of missed occult HSIL+ cases and adds to the evidence base related to the use of ECC in clinical practice.