Cold-knife conization versus loop electrosurgical excision for cervical adenocarcinoma in situ: a retrospective cohort study Authors: Salamaiti Abudurehemu, Xuanhui Wang, Jue Wang, Tianyu Liu, Ming Chen Published: 2026 ABSTRACT AbstractObjectiveThe purpose of this study was to compare the efficacy of cold-knife conization (CKC) and loop electrosurgical excision procedure (LEEP) in treating cervical adenocarcinoma in situ (AIS) and identify relevant risk factors within a recent cohort from a tertiary hospital in China.MethodsWe conducted a retrospective chart review of patients who underwent a conization procedure with a preoperative or postoperative diagnosis of AIS of the cervix from January 2021 to July 2024. Clinicopathological data and follow-up results were collected, with recurrence-free survival (RFS) as the primary endpoint.ResultsAmong the 251 patients included, 161 (64.14%) underwent LEEP and 90 (35.86%) underwent CKC as primary treatment. Overall, 29.9% (75/251) underwent eventual hysterectomy. Median follow-up was 36 months. The positive margin rate was lower in the CKC group than in the LEEP group (16.7% vs. 31.7%, P < 0.001). Multivariate analysis showed that LEEP (aOR 2.28, 95% CI 1.13–4.58, P = 0.021), TZ type 2 or 3 (aOR 2.00, 95% CI 1.12–3.58, P = 0.019), and AIS with concurrent HSIL (aOR 2.07, 95% CI 1.13–3.78, P = 0.018) were independently associated with positive margins. Greater cone depth was associated with lower risk (per 1 mm increase: aOR 0.73, 95% CI 0.55–0.97, P = 0.031) when conization type was excluded due to collinearity. Among patients who underwent hysterectomy, positive conization margin was the only predictor of residual disease (aOR 9.12, 95% CI 1.07–77.8, P = 0.043). Kaplan–Meier analysis showed a significant difference in RFS among the three groups (P = 0.018), but pairwise comparison did not reveal a significant difference between the LEEP and CKC groups (P = 0.109).ConclusionsIn this retrospective cohort, CKC achieved lower positive margin rates than LEEP for AIS. However, short-term RFS did not differ significantly between the two techniques, suggesting that margin status alone may not fully predict oncologic outcomes. Positive margin was a predictor of residual disease at hysterectomy, but this finding requires validation. For fertility-seeking patients, CKC is preferred to optimize margins, though LEEP may be acceptable when CKC is not feasible.Keywords: Cervical Adenocarcinoma in Situ (AIS), Cold-Knife Conization (CKC), Loop Electrosurgical Excision Procedure (LEEP), Positive Margin, Recurrence-Free Survival (RFS) Introduction Adenocarcinoma in situ (AIS) of the uterine cervix, the precursor lesion to invasive cervical adenocarcinoma, is being diagnosed with increasing frequency [1]. Its incidence has been increasing in recent years, reaching 11.2 per 100,000 persons at the peak age of 30–39 years [2]. In cases of AIS, cervical conization with clear margins is advised, especially given that the average age at diagnosis is 37 and many women are postponing pregnancy. The two main approaches for conization are cold knife conization (CKC) and loop electrosurgical excision procedure (LEEP). LEEP is typically performed under local anesthesia in an outpatient setting and is more cost-effective, whereas CKC generally requires regional or general anesthesia in an operating room, resulting in significantly higher costs. Earlier research has examined conservative management of AIS using both CKC and LEEP, yet findings on their comparative therapeutic effectiveness remain mixed. While certain studies and meta-analyses have indicated a preference for CKC over LEEP in treating AIS [3–5], other studies considered LEEP as equally effective as CKC [6, 7], achieving the same rates of negative margins. The 2019 guidelines from the American Society for Colposcopy and Cervical Pathology (ASCCP) do not specify a preference between CKC and LEEP as the primary treatment method. However, the language has evolved since 2001, when CKC was favored over loop excision, to 2006, when diagnostic excision using any technique, including larger loops, was permitted [8, 9]. Whether LEEP has comparable effects to CKC remains debated. Most existing evidence comes from meta-analyses pooling older studies. This study aims to provide contemporary, real-world data from a high-volume Chinese tertiary center, with a specific focus on risk factors for positive margins and short-term recurrence. Methods Study design and clinical information This study was designed as a single-center, retrospective study of women who received conization for AIS at the Department of Gynecology of the First Affiliated Hospital of Sun Yat-sen University. We extracted patients’ clinicopathological data including Pap test results, high-risk human papillomavirus (HR-HPV) DNA test results, and surgical pathology findings including endocervical curettage (ECC), cervical biopsy, and LEEP/CKC/hysterectomy pathology from our database. The LEEP procedure was performed using a loop (15 × 15 mm or 20 × 10 mm), whereas CKC was performed with a cold knife, excising 5–10 mm outside the iodine-free zone of the cervix to a depth of 15–20 mm. The decision to undergo LEEP or CKC was mainly based on surgeon preference. In general, LEEP was performed if the doctor believed that the cervical lesion seen on colposcopy could be entirely removed in one pass. Due to the retrospective nature of this study, data on lesion size and endocervical involvement were not consistently available and could not be included in the analysis. Follow-up data were based on clinic medical records and were supplemented through telephone contact and subsequent visits. Conization was defined as LEEP or CKC. Pathology reports were analyzed for descriptions of histology, margin status of the specimen, location of positive margins, and final pathologic diagnosis. Positive margins were defined as AIS identified within 1 mm of the surgical margin or a positive endocervical curettage (ECC) performed after conization. Residual lesions after hysterectomy were defined as HSIL or AIS identified in the hysterectomy specimen. The Ethics Committee of the First Affiliated Hospital of Sun Yat-sen University approved this study. All samples and data were collected after written informed consent was provided by the participants. Inclusion and exclusion criteria Inclusion criteria: age 20–50 years, received conization at our center, histological diagnosis of AIS based on cervical biopsy before conization or on the conization specimen. Exclusion criteria: concurrent invasive carcinoma on the conization specimen, concomitant immunocompromised disease or other cancer, prior treatment for CIN lesions, or incomplete pretreatment or follow-up data. Follow-up Primary treatment was defined as the most invasive surgical procedure performed within the first three months of diagnosis of AIS. The follow-up period was defined as time from primary treatment to the date of last patient contact. Recurrence was defined as a histological diagnosis of high-grade cervical or vaginal dysplasia (CIN 2/3, vaginal intraepithelial neoplasia [VAIN] 2/3, or AIS) after primary treatment. Recurrence-free survival (RFS) was defined as time from primary treatment to histologic confirmation of recurrence. All 251 patients were followed for 12 to 36 months. When a positive margin was found after conization, patients were advised to undergo repeated conization or hysterectomy within three months. After conization with negative margins or after hysterectomy, follow-up was performed at 6 and 12 months and annually thereafter, including co-testing and colposcopy. Statistical analysis Descriptive statistics were summarized by medians and ranges for numerical variables and by frequencies and percentages for categorical variables. Comparisons between treatment groups were performed using the ANOVA test or independent t-test for normally distributed numerical variables, the Kruskal–Wallis or Mann–Whitney U test for non-normally distributed numerical variables, and the chi-squared test for categorical variables. Logistic regression analysis was used to identify risk factors for positive conization margins and for residual disease at hysterectomy. Univariate analysis was first performed for each variable. Variables with a P-value < 0.10 in univariate analysis, along with clinically relevant factors, were considered for entry into multivariate models. Due to collinearity between cone depth and conization type, two separate multivariate models were constructed for positive margin analysis: Model 1 included cone depth, transformation zone (TZ) type, and histology (excluding conization type); Model 2 included conization type, TZ type, and histology (excluding cone depth). For residual disease analysis, a single multivariate model including margin status and histology was constructed. Results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). For RFS, the Kaplan–Meier method was used, and comparisons among groups were performed using the log-rank test. Univariable and multivariable Cox regression analyses were performed to assess the association between treatment and RFS. The proportional hazards assumption was tested and met. All statistical analyses were performed using SPSS (version 22.0) and R (version 4.2.0), specifically using the packages “survival” and “stats”. Results Clinicopathological characteristics and clinical outcomes Of 362 patients screened, 251 (69.3%) were included (Fig. 1). The mean age was 35.8 years (SD 7.9; range 23–49). As shown in Table 1, 161 patients (64.14%) underwent LEEP and 90 (35.86%) underwent CKC as primary treatment; 29.9% (75/251) eventually underwent hysterectomy. There were no significant differences in age, parity, cytology, or HPV distribution between the LEEP and CKC groups. A definitive cytologic diagnosis of atypical glandular cells (AGC, AGC‑NOS, or AGC‑FN) was made in only 49 patients (19.5%). Normal cytology was found in 77 patients (30.7%). The most common HR‑HPV type was HPV16 (42.6%). Nearly half of AIS patients had concurrent HSIL (LEEP 47.7%, CKC 42.2%). Flowchart of patient selection for the study Table 1Baseline patient characteristics of patients treated with LEEP versus CKCLEEPCKC P 161900.377Age23–4925–46 < 40y10262 >=40y5928Parity0.080 04938 17630 >=23622Cytology before conization0.404 AGC /AGC-NOS2511 AGC-FN85 ASC-US/LSIL4627 ASC-H/HSIL3814 NILM4433HPV0.674 169645 183832 Other hr-HPV208 Negative75Histopathological diagnosis at conization0.702 AIS8952 AIS+HSIL7238Margin status0.010 Positive5115 Negative11075Hysterectomy after conization0.160 Yes5322 No10868Abbreviations: LEEP loop electrosurgery excisional procedure, CKC Cold knife conization, NILM intraepithelial lesions or malignancy, ASC-US atypical Squamous Cells of Undetermined Significance, LSIL low-grade squamous intraepithelial lesion, ASC-H atypical squamous cells, HSIL high-grade squamous intraepithelial lesion, AGC atypical glandular lesion, AGC-NOS atypical glandular cells-not otherwise specified, AGC-FN atypical glandular cells-favor neoplasia Baseline patient characteristics of patients treated with LEEP versus CKC Abbreviations: LEEP loop electrosurgery excisional procedure, CKC Cold knife conization, NILM intraepithelial lesions or malignancy, ASC-US atypical Squamous Cells of Undetermined Significance, LSIL low-grade squamous intraepithelial lesion, ASC-H atypical squamous cells, HSIL high-grade squamous intraepithelial lesion, AGC atypical glandular lesion, AGC-NOS atypical glandular cells-not otherwise specified, AGC-FN atypical glandular cells-favor neoplasia Treatment with LEEP resulted in a higher positive margin rate (51/161, 31.7%) compared with CKC (15/90, 16.7%) (P < 0.001). Among patients with positive margins in the LEEP group, 7 underwent repeated conization, 32 underwent hysterectomy, and 12 refused further surgery. In the CKC group with positive margins, 2 received repeated conization, 8 received hysterectomy, and 5 chose observation. Risk factors for positive margin and residual disease after hysterectomy Overall, 66 of 251 patients had positive margins after the first conization. Univariate logistic regression analysis (Table 2) showed that abnormal cytology (low-grade: OR 6.53, 95% CI 2.44–17.5, P < 0.001; high-grade: OR 10.3, 95% CI 3.69–28.6, P < 0.001), TZ type 2 or 3 (OR 2.06, 95% CI 1.17–3.63, P = 0.012), LEEP (OR 2.32, 95% CI 1.22–4.41, P = 0.010), and AIS with concurrent HSIL (OR 2.14, 95% CI 1.22–3.76, P = 0.008) were associated with higher risk of positive margins. Greater cone depth was associated with lower risk (per 1 mm increase: OR 0.71, 95% CI 0.54–0.94, P = 0.015). Table 2Univariate and multivariate logistic regression analysis of risk factors for positive conization margins in patients with cervical AISVariablesPositive marginsNegative marginsUnivariate analysisMultivariate analysis (model 1)*1Multivariate analysis (model 2)*2OR (95%CI) P aOR *3 (95%CI) P aOR (95%CI) P Age <40y41ReferenceNot includedNot included >=40y251.21 (0.68–2.14)0.517Parity 018ReferenceNot includedNot included ≥1481.59 (0.86–2.94)0.142Cytology before conization Normal5ReferenceNot includedNot included Low-grade*4346.53 (2.44–17.5)<0.001 High-grade*52710.3 (3.69–28.6)<0.001HPV genotype 1639102ReferenceNot includedNot included 1824461.41 (0.76–2.61)0.279 Other/ Negative3370.29 (0.10–0.82)0.020Preoperative TZ types under colposcopy TZ 129107ReferenceReference TZ 2 or 337782.06 (1.17–3.63)0.0122.00 (1.12–3.58)0.0192.00 (1.12–3.58)0.019Conization type CKC1575ReferenceReferenceNot included LEEP511102.32 (1.22–4.41)0.0102.28 (1.13–4.58)0.021Cone depth (per 1 mm increase) CKC2.00.4 mm2.2 mmNot included LEEP1.70.3 mm1.8 mm0.71(0.54-0.94)0.0150.73(0.55-0.97)0.031Histology AIS28113ReferenceReference AIS with HSIL38722.14 (1.22–3.76)0.0082.07 (1.13–3.78)0.0182.07 (1.13–3.78)0.018*1. Model 1 included conization type, TZ type, and histology (cone depth excluded due to collinearity)2. Model 2 included cone depth, TZ type, and histology (conization type excluded due to collinearity)3. aOR: Adjusted odds ratio4. Low-grade cytology includes ASC-US, LSIL, AGC and AGC-NOS5. High-grade cytology includes ASC-H, HSIL and AGC-FN Univariate and multivariate logistic regression analysis of risk factors for positive conization margins in patients with cervical AIS *1. Model 1 included conization type, TZ type, and histology (cone depth excluded due to collinearity) 2. Model 2 included cone depth, TZ type, and histology (conization type excluded due to collinearity) 3. aOR: Adjusted odds ratio 4. Low-grade cytology includes ASC-US, LSIL, AGC and AGC-NOS 5. High-grade cytology includes ASC-H, HSIL and AGC-FN Due to collinearity between cone depth and conization type, two separate multivariate models were constructed. In Model 1 (excluding conization type), cone depth (aOR 0.73, 95% CI 0.55–0.97, P = 0.031), TZ type 2 or 3 (aOR 2.00, 95% CI 1.12–3.58, P = 0.019), and AIS with HSIL (aOR 2.07, 95% CI 1.13–3.78, P = 0.018) remained independently associated with positive margins. In Model 2 (excluding cone depth), LEEP (aOR 2.28, 95% CI 1.13–4.58, P = 0.021), TZ type 2 or 3 (aOR 2.00, 95% CI 1.12–3.58, P = 0.019), and AIS with HSIL (aOR 2.07, 95% CI 1.13–3.78, P = 0.018) remained significant. Cytology was not included in multivariate models due to collinearity with histology, and HPV was excluded due to small subgroup sample sizes and zero events in the negative group. Among 75 patients who underwent hysterectomy, residual disease was identified in 10 (13.3%) specimens. On univariate logistic regression analysis (Table 3), positive conization margin was the only factor significantly associated with residual disease (OR 9.87, 95% CI 1.19–81.9, P = 0.034). Age, HPV genotype, conization type, and histology were not significantly associated (all P > 0.05). In multivariate analysis adjusting for histology, positive margin remained independently associated with residual disease (aOR 9.12, 95% CI 1.07–77.8, P = 0.043). Histology showed a trend toward higher risk for AIS with concurrent HSIL but did not reach statistical significance (aOR 2.85, 95% CI 0.54–15.0, P = 0.218), likely due to the limited number of events (n = 10). Table 3Univariate and multivariate logistic regression analysis of risk factors for residual disease at hysterectomy in patients with cervical AISResidualNo residualUnivariate analysisMultivariate analysisOR (95%CI) P aOR (95%CI) P Age < 40y425ReferenceNot included >=40y6401.50 (0.39–5.78)0.556Margin status Negative134ReferenceReference Positive9319.87 (1.19–81.9)0.0349.12 (1.07–77.8)0.043HPVNot included 16529Reference 184240.97 (0.23–4.00)0.962 Other170.83 (0.09–7.87)0.871 Negative05— (0 events)—Conization typeැNot included LEEP647Reference CKC4180.57 (0.15–2.25)0.428Histology AIS230ReferenceReference AIS with HSIL8353.43 (0.68–17.3)0.1362.85 (0.54–15.0)0.218 Univariate and multivariate logistic regression analysis of risk factors for residual disease at hysterectomy in patients with cervical AIS Recurrence Median follow-up was 36 months (range 12–36). After primary treatment with LEEP or CKC, hysterectomy during follow-up was performed in 53 of 151 (35.1%) and 22 of 90 (24.4%) patients, respectively. Overall, 14 patients (5.6%) recurred: 12 with high-grade dysplasia (4.8%) and 2 with adenocarcinoma (0.8%) (Table 4). Table 4Recurrence-free survival (RFS) after primary treatment for cervical AISTreatmentTotal (n)Recurrence (n)Recurrent AIS or CIN 2–3 or VAIN 2–3 (n)Cervical cancer (n)Median RFS (months, 95% CI)P*LEEP108119234.5(33.4–35.6)Positive margin12532Negative margin96660CKC6811035.2 (34.3–36.1)Positive margin5110Negative margin63000Hysterectomy7522035.8 (35.3–36.3)Total251141220.018*P value from log-rank test comparing the three treatment groups Recurrence-free survival (RFS) after primary treatment for cervical AIS *P value from log-rank test comparing the three treatment groups Based on primary treatment, patients were divided into three groups: LEEP (n = 108, after excluding those who underwent immediate hysterectomy), CKC (n = 68), and hysterectomy (n = 75). Kaplan–Meier analysis showed a significant difference in RFS among the three groups (χ²=8.056, P = 0.018, Fig. 2). In the LEEP group, 11 patients (10.2%) recurred, 2 of whom developed cervical cancer. In the CKC group, 1 patient (1.5%) was diagnosed with recurrent AIS. In the hysterectomy group, 2 patients (2.7%) developed VAIN 2, and no patient developed local cervical cancer. Kaplan–Meier curves for recurrence-free survival (RFS) among patients treated with LEEP, CKC, or hysterectomy Pairwise comparisons revealed no statistically significant difference in RFS between the LEEP and CKC groups (χ²=2.537, P = 0.109), nor between CKC and hysterectomy groups (χ²=0.524, P = 0.469). Only the RFS of the LEEP group was significantly shorter than that of the hysterectomy group (χ²=5.737, P = 0.017). Discussion While prior meta-analyses have established that CKC achieves lower positive margin rates than LEEP for AIS, most evidence is derived from studies conducted before 2020 and lacks detailed analysis of contemporary clinical practice. This study provides incremental value in three aspects: (1) it offers recent (2021–2024) real-world data from a large Chinese cohort, where AIS management patterns may differ from Western populations; (2) it identifies mixed histology (AIS with HSIL) as an independent predictor of positive margins, a finding not consistently reported in previous meta-analyses; and (3) it demonstrates that despite superior margin clearance with CKC, short-term RFS does not significantly differ from LEEP, consistent with recent evidence from a multicenter study [10], suggesting that margin status alone may be an imperfect surrogate for recurrence in conservatively managed patients. In our comparison, the positive margin rate was higher with LEEP (31.7%) than with CKC (16.7%). This finding is consistent with three previous systematic reviews [3, 5, 11]. The reason may be that AIS lesions are often located deep within the cervical canal, and a superficial LEEP may fail to capture them completely [12, 13]. However, pairwise comparisons revealed no significant difference in RFS between LEEP and CKC (P = 0.109). This finding—that superior margin clearance with CKC does not translate into significantly better short-term RFS—challenges the assumption that margin status serves as a direct surrogate for recurrence after conservative management. The clinical significance of a positive margin may therefore lie primarily in its strong association with residual disease at hysterectomy. Additionally, though no universal definition of positive margin exists in the literature, our definition of AIS within 1 mm of the surgical margin is consistent with recent large-scale studies and meta-analyses [14–16], as well as contemporary guidelines that consider involved or close margins as conferring increased risk of residual disease [2, 12]. Nearly half of AIS patients had concurrent HSIL (43.8%), consistent with the literature [17, 18]. The presence of HSIL was an independent predictor of positive margins (P = 0.007), possibly indicating more extensive or multifocal disease. For such patients, we recommend CKC to optimize margin clearance. Historically, hysterectomy has been recommended as definitive treatment for AIS based on early guidelines and concerns regarding skip lesions or occult invasion [19]. In the present cohort, only 10 of 75 (13.3%) hysterectomy specimens had residual disease, and positive conization margin was the sole predictor. Our finding that positive cone margin is associated with residual disease at hysterectomy is consistent with the 2023 Canadian Colposcopy Guideline [12]. This suggests that routine hysterectomy for all AIS patients is not justified, as the majority do not benefit. Instead, conservative management with close follow-up is reasonable, particularly for those with negative margins and fertility desire [20]. We acknowledge that these findings are based on a limited number of hysterectomy specimens (n = 75) and events (n = 10), which may yield unstable estimates; therefore, they should be interpreted cautiously and considered exploratory rather than definitive. Larger prospective studies are needed to validate predictors of residual disease. Our study found that HPV16 (42.6%) and HPV18 (32.4%) were the predominant subtypes in AIS lesions, collectively accounting for nearly 75% of cases. A large Chinese cohort study similarly reported that HPV18 was the most frequent genotype in AIS, followed by HPV16, confirming that HPV16/18 are the dominant drivers of AIS in the Chinese population [21]. The high prevalence of HPV18 is clinically significant, as this subtype demonstrates a higher rate of viral integration into the host genome (59% vs. 13% for HPV16), which may explain its propensity for rapid progression and skip lesions [22]. Furthermore, recent reviews highlight that HPV18 possesses unique biological characteristics distinct from other high-risk HPV types, including a tendency for latent infection in stem cell-like cells and frequent detection in highly malignant adenocarcinoma [23]. Beyond cervical AIS, HPV-related intraepithelial neoplasia can affect the entire lower genital tract. For instance, a recent study by Vural et al. evaluated topical imiquimod for VAIN and reported favorable responses, supporting the feasibility of non-excisional, fertility-sparing strategies in select cases [24]. Although imiquimod is not currently indicated for AIS, such findings reinforce our conclusion that for AIS patients with negative margins and strong fertility desire, conservative follow-up is a reasonable alternative to routine hysterectomy. This study has several limitations. First, its retrospective, single-center design introduces selection bias, as treatment allocation was not randomized. Second, data on lesion size, endocervical involvement, and specimen fragmentation were not consistently available. Third, the median follow-up of 36 months is insufficient to capture late recurrences typical for AIS. Fourth, the small number of events in the hysterectomy subgroup (n = 10) yields unstable estimates. Finally, our definition of positive margins (AIS within 1 mm) may limit comparability with some prior studies. Conclusion In this retrospective cohort, CKC achieved lower positive margin rates than LEEP for AIS. However, short-term RFS did not differ significantly between the two techniques, suggesting that margin status alone may not fully predict oncologic outcomes. Positive margin was a predictor of residual disease at hysterectomy, but this finding requires validation. For fertility-seeking patients, CKC is preferred to optimize margins, though LEEP may be acceptable when CKC is not feasible. Abbreviations Adenocarcinoma in situ Cold knife conization Loop electrosurgery excisional procedure American Society for Colposcopy and Cervical Pathology High-risk human papillomavirus Endocervical curettage High-grade squamous intraepithelial lesion Cervical intraepithelial neoplasia Vaginal intraepithelial neoplasia Confidence interval Odds ratio Authors’ contributions Ming Chen designed the study, carried out the colposcopies, followed up with the patients and revised the manuscript. Salamaiti Abudurehemu drafted the manuscript, completed the follow-up and analyzed the data. Xuanhui Wang collected the data and followed up with the patients. Jue Wang reviewed all the pathological slides. Tianyu Liu helped carry out the colposcopies and follow-up. All the authors provided final approval for the version to be submitted. Funding This study was funded by the Guangdong Natural Science Foundation (2024A1515012946) and the Baiqiuen Foundation (ffbt-B-2022-006). The funders played no role in the study design, data collection, analysis, decision to publish, or preparation of the manuscript. Data availability The data cannot be shared publicly due to institutional policy. Data are available upon reasonable request to the corresponding author. Declarations Ethics approval and consent to participate This retrospective study was approved by the Ethics Committee of the First Affiliated Hospital of Sun Yat-Sen University and adhered to the principles of the Helsinki Declaration as well as national and international standards. The patients provided written informed consent to participate in this study. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The data cannot be shared publicly due to institutional policy. Data are available upon reasonable request to the corresponding author.