The relationship between atypical glandular cells and malignancy risk: a retrospective single-center study.

The study protocol was reviewed and approved by The First Affiliated Hospital of Chongqing Medical University (202416601) and was conducted in accordance with the 1964 Helsinki Declaration and its amendments or comparable ethical standards. We retrospectively retrieved the records of patients who underwent cervical cytology for routine CC screenings for the general population or HPV-positive follow-up in the gynecology department of our hospital between March 2017 and June 2024. Patients with cytological results diagnosed as AGC were included in this study. Data including age, previous medical history, hrHPV testing results, and final pathological diagnosis within 1-year of AGC diagnosis were collected. Histological specimens were obtained through colposcopy-guided endocervical sampling, cold-knife conization, loop electrosurgical excision procedure, endometrial sampling, or hysterectomy. The inclusion criteria were as follows: (1) patients with AGC diagnosed via cervical cytology, (2) patients having clear histopathological biopsy follow-up results. The exclusion criteria were as follows: (1) history of gynecological malignancy, or other systemic malignancy, (2) inconclusive cytological results, and (3) absent or inconclusive histopathological results. All pathological specimens and cervical cytology samples were processed following a standardized procedure and examined independently by two experienced pathologists for diagnosis. According to the 2014 revised Bethesda System (TBS) Interpretation Criteria, AGC was divided into five subcategories: atypical endocervical cells (AEC), atypical endometrial cells (AEM), AGC-not otherwise specified (AGC-NOS), AEC-favor neoplastic (AEC-FN), and AGC-favor neoplastic (AGC-FN) ( Nayar and Wilbur, 2015 ). The aptima HPV E6/E7 mRNA testing (Hologic, San Diego, California, USA) was utilized to detect the following HPV types: 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68. Of these, HPV16 and HPV18/45 were detected in separate lanes, while the remaining types were identified collectively. Normal or benign lesions included inflammatory lesions, low-grade squamous intraepithelial lesions (LSIL), cervical or endometrial polyps, endometrial glandular hyperplasia, mild atypical hyperplasia of the glands, uterine or cervical myoma, and endometriosis. Precancerous lesions and malignancies included cervical high-grade squamous intraepithelial lesions (HSIL), cervical SCC, cervical adenocarcinoma in situ (AIS), AC, endometrial atypical hyperplasia (EAH), endometrial carcinoma (EC), pelvic malignancy (including ovarian cancer, fallopian tube cancer, and peritoneal cancer), and other precancerous lesions and malignancies (including uterus malignant mullerian mixed tumor, small-cell neuroendocrine carcinoma of the cervix, SMARCA4-deficient undifferentiated uterine sarcoma, low-grade endometrial stromal sarcomas, vaginal HSIL, and vaginal SCC). Statistical analyses and data visualization were performed using SPSS 26.0 (IBM, Armonk, New York, USA) and GraphPad Prism 9.0 (GraphPad Software, La Jolla, California, USA). Chi-square and Fisher’s exact tests were used to assess histological correlations, and multinomial logistic regression analysis was used to identify risk factors for precancerous lesions and malignancies. P  < 0.05 was considered statistically significant.

A total of 447 patients diagnosed with AGC through cervical cytology, who visited our hospital between March 2017 and June 2024, were retrospectively analyzed. Based on the inclusion and exclusion criteria, 402 patients were included in this study. The details are presented in Table 1 . The average age of patients with AGC was 49.98 ± 10.233 years, and the samples were divided into two groups based on the average age: <50 and ≥50 years old. The distribution of AGC subcategories was as follows: AEC, 13.4%; AEM, 10.9%; AGC-NOS, 49.8%; AEC-FN, 6.7%; and AGC-FN, 19.2%. The histopathology distribution is shown in Fig. 1 . Among the 402 patients, 12.6% had precancerous lesions and 46.3% had malignancies. EC had the highest incidence at 24.4%, followed by AC with an incidence rate of 10%. Among patients with AGC, 358 underwent hrHPV testing simultaneously. The positive rate of hrHPV was 35.2%, type 16 being the most common, accounting for 10.1%, followed by HPV 18/45 at 8.1%. Characteristics of patients with atypical glandular cells One case had EC combined with cervical HISL, and one case had vaginal SCC combined with ovarian cancer. AC, cervical adenocarcinoma; AEC, atypical endocervical cells; AEC-FN, atypical endocervical cells-favor neoplastic; AEM, atypical endometrial cells; AGC-FN, atypical glandular cells-favor neoplastic; AGC-NOS, atypical glandular cells-not otherwise specified; AIS, cervical adenocarcinoma in situ; EAH, endometrial atypical hyperplasia; EC, endometrial carcinoma; HSIL, high-grade squamous intraepithelial lesions; SCC, squamous cell carcinoma. The histopathological distribution of patients diagnosed with AGC. AC, cervical adenocarcinoma; AEC, atypical endocervical cells; AEC-FN, atypical endocervical cells-favor neoplastic; AEM, atypical endometrial cells; AGC-FN, atypical glandular cells-favor neoplastic; AGC-NOS, atypical glandular cells-not otherwise specified; AIS, cervical adenocarcinoma in situ; EAH, endometrial atypical hyperplasia; EC, endometrial carcinoma; HSIL, high-grade squamous intraepithelial lesions; SCC, squamous cell carcinoma. Other precancerous lesions and malignancies included one case of uterus malignant mullerian mixed tumor, one case of small-cell neuroendocrine carcinoma of the cervix, one case of SMARCA4-deficient undifferentiated uterine sarcoma, one case of low-grade endometrial stromal sarcomas, two cases of vaginal HSIL and two cases of vaginal SCC. One case had EC combined with cervical HISL, and one case had vaginal SCC combined with ovarian cancer. The detailed histopathological results are presented in Table 2 . Among patients with AEC, 50% had normal or benign lesions, followed by cervical HSIL (20.4%) and AC (14.8%). In AEM, 59.1% were diagnosed with EC, while 31.8% had normal or benign lesions. Among patients with AGC-NOS, 55.2% had normal or benign lesions, and 16.9% had EC. In AEC-FN, 55.6% were diagnosed with AC, and 14.8% with AIS. Lastly, in AGC-FN, 44.9% had EC, and 20.5% had pelvic malignancies. The detailed histopathological of patients with subcategories of atypical glandular cell AC, cervical adenocarcinoma; AEC, atypical endocervical cells; AEC-FN, atypical endocervical cells-favor neoplastic; AEM, atypical endometrial cells; AGC-FN, atypical glandular cells-favor neoplastic; AGC-NOS, atypical glandular cells-not otherwise specified; AIS, cervical adenocarcinoma in situ; EAH, endometrial atypical hyperplasia; EC, endometrial carcinoma; HSIL, high-grade squamous intraepithelial lesions; SCC, squamous cell carcinoma. One specimen was histologically confirmed to be a vaginal squamous cell carcinoma combined with ovarian malignancy. One specimen was histologically confirmed as EC combined with cervical HSIL. The prevalence of precancerous lesions and malignancies was 58.9%, as shown in Fig. 2 . In AEC, 85.2% of cases were ervical lesions, 11.1% were endometrial lesions, and 3.7% were other lesions. In AEM, 3.3% were cervical lesions, whereas 90.0% were endometrial lesions and 6.7% were other lesions. In AGC-NOS, 37.8% were cervical lesions, 38.9% were endometrial lesions, and 23.3% were other lesions. In AEC-FN, 91.3% were cervical lesions, 4.3% were endometrial lesions, and 4.3% were other lesions. In AGC-FN, 19.4% were cervical lesions, 53.7% were endometrial lesions, and 26.9% were other lesions. Histopathology of precancerous lesions and malignancies in the AGC subcategories. AEC, atypical endocervical cells; AEC-FN, atypical endocervical cells-favor neoplastic; AEM, atypical endometrial cells; AGC-FN, atypical glandular cells-favor neoplastic; AGC-NOS, atypical glandular cells-not otherwise specified. Cervical lesion included cervical high-grade squamous intraepithelial lesions (HSIL), cervical squamous cell carcinoma (SCC), cervical adenocarcinoma in situ (AIS), and cervical adenocarcinoma (AC). Endometrial lesions included endometrial atypical hyperplasia (EAH), endometrial carcinoma (EC). Other lesions included pelvic malignancy, uterus malignant mullerian mixed tumor, small-cell neuroendocrine carcinoma of the cervix, SMARCA4-deficient undifferentiated uterine sarcoma, low-grade endometrial stromal sarcomas, vaginal HSIL and vaginal SCC. In this study, a total of 358 patients were tested for hrHPV, as shown in Table 3 . The results revealed that cervical lesions primarily occurred in hrHPV-positive patients. Among these, cervical HSIL/SCC occurred in those who were HPV 16 positive and positive for other hrHPV types excluding 16 or 18/45. This may be attributed to the absence of HPV typing in this study and the differing distribution of HPV subtypes in China compared to other countries ( Bruni et al ., 2023 ). There was a high prevalence of HPV 16 and 18/45 in AIS/AC; however, 6% of cases were HPV-negative. Endometrial lesions and pelvic malignancies primarily occurred in hrHPV-negative patients. Histopathology of precancerous lesions and malignancies in the atypical glandular cells subcategories AC, cervical adenocarcinoma; AEC, atypical endocervical cells; AEC-FN, atypical endocervical cells-favor neoplastic; AEM, atypical endometrial cells; AGC-FN, atypical glandular cells-favor neoplastic; AGC-NOS, atypical glandular cells-not otherwise specified; AIS, cervical adenocarcinoma in situ; EAH, endometrial atypical hyperplasia; EC, endometrial carcinoma; HPV, human papillomavirus; hrHPV, high-risk human papillomavirus; HSIL, high-grade squamous intraepithelial lesions; SCC, squamous cell carcinoma. In this study, among patients aged ≥50 years, the incidence of precancerous lesions and malignancies was significantly higher compared to patients aged <50 years (χ 2 = 45.073, P  < 0.001). In AEC, cervical lesions had a higher incidence, and cervical HSIL/SCC primarily occurred in patients who were HPV16 positive and positive for other hrHPV types excluding 16 or 18/45, whereas AIS/AC primarily occurred in patients who were HPV16 and 18/45 positive. No significant association was found between hrHPV infection and endometrial lesions or pelvic malignancies (χ 2 = 29.813, P <  0.001). In AEM, the primary histopathology was endometrial lesions, with most patients being hrHPV-negative; however, this difference was not statistically significant ( P  = 0.139). In AGC-NOS, cervical HSIL/SCC occurred in patients who were hrHPV-positive, AIS/AC occurred in patients who were HPV16 and 18/45 positive. In contrast, endometrial lesions and pelvic malignancies occurred in hrHPV-negative patients (χ 2 = 55.863, P <  0.001). In AEC-FN, cervical precancerous lesions and malignancies, particularly AIS/AC, exhibited a higher incidence and were closely associated with HPV16 and 18/45 positivity; however, the difference was not statistically significant ( P   = 0.262). In AGC-FN, endometrial lesions and pelvic malignancies occurred in hrHPV-negative patients; however, cervical lesions remained significantly associated with hrHPV infection (χ 2 = 24.324, P  = 0.010). We used multivariate logistic regression analysis to investigate the relationship between age, hrHPV, AGC subcategories, and histopathology. hrHPV negative and AGC-NOS subcategory were considered as baseline. Statistically significant items are detailed in Table 4 . Older patients had a statistically significantly greater risk for EAH/EC [odds ratio (OR): 0.289, 95% confidence interval (CI): 0.154–0.540, P  < 0.001] and pelvic malignancies (OR: 0.101, 95% CI: 0.033–0.315, P  < 0.001). Patients in the AGC-FN (OR: 3.479, 95% CI: 1.018–11.883, P =  0.047) and AEC (OR: 2.773, 95% CI: 1.027–7.486, P =  0.044) subcategories, as well as those who were HPV16 positive (OR: 7.933, 95% CI: 2.050–30.707, P =  0.003) or positive for other hrHPV types excluding 16 or 18/45 (OR: 10.302, 95% CI: 3.853–27.542, P  < 0.001) were identified as significant risk factors for cervical HSIL/SCC. Patients diagnosed AGC-FN (OR: 10.082, 95% CI: 2.651–38.352, P =  0.001) and who were HPV16 (OR: 9.984, 95% CI: 3.211–31.045, P  < 0.001) and HPV18/45 positive (OR: 6.198, 95% CI: 2.203–17.442, P =  0.001) were more likely to develop AIS/AC. The risk factors for EAH/EC were the AGC-FN (OR: 5.824, 95% CI: 2.548–13.311, P  < 0.001) and AEM (OR: 5.186, 95% CI: 2.171–12.389, P  < 0.001) subcategories. Conversely, patients who were HPV18/45 positive (OR: 0.091, 95% CI: 0.010–0.793, P =  0.030) or positive for other hrHPV types excluding 16 or 18/45 (OR: 0.282, 95% CI: 0.102–0.777, P =  0.014) showed no significant association with EAH/EC. In addition, the subcategory AGC-FN was also a risk factor for pelvic malignancy (OR: 4.703, 95% CI: 1.722–12.843, P =  0.003). HPV16 positivity was also a risk factor for other precancerous lesions and malignancies (OR: 10.124, 95% CI: 1.204–85.144, P =  0.033). A possible reason is that HPV positivity is the cause of vaginal lesions; however, the sample size was small and there was a certain bias. Multivariate logistic regression analysis of the relationships between age, human papillomavirus, atypical glandular cells, and histopathology AC, cervical adenocarcinoma; AEC, atypical endocervical cells; AEC-FN, atypical endocervical cells-favor neoplastic; AEM, atypical endometrial cells; AGC-FN, atypical glandular cells-favor neoplastic; AGC-NOS, atypical glandular cells-not otherwise specified; AIS, cervical adenocarcinoma in situ; B, regression coefficient; CI, confidence interval; EAH, endometrial atypical hyperplasia; EC, endometrial carcinoma; HSIL, high-grade squamous intraepithelial lesions; OR, odds ratio; SCC, squamous cell carcinoma.

The diagnosis and prevention of cervical squamous cell carcinoma (SCC) have significantly improved owing to the popularization of cervical cancer (CC) screening technologies and the use of the human papillomavirus (HPV) vaccine ( McCredie et al ., 2008 ; Porras et al ., 2020 ). However, the incidence of cervical adenocarcinoma (AC) has steadily increased each year, with cases occurring at progressively younger ages ( Smith et al ., 2000 ; Bray et al ., 2005 ; Baalbergen and Helmerhorst, 2014 ), now accounting for approximately 20–25% of CC cases ( Yamagami et al ., 2017 ). Unlike SCC, partial types of AC have no obvious correlation with high-risk HPV (hrHPV) infection. Therefore, a screening strategy for the early diagnosis of AC is of particular importance. Atypical glandular cells (AGC) are relatively uncommon in cervical cytology. AGC displays abnormal morphology but does not meet the criteria for malignancy. Due to their complex and highly variable morphology, diagnosing AGC remains challenging and lacks reproducibility ( Lee et al ., 2002 ; Kurtycz et al ., 2017 ). In a laboratory survey, fewer than 1% of cervical cytology samples were diagnosed as AGC ( Davey et al ., 2019 ). Although AGCs are relatively rare in cytological examinations, they carry substantial clinical significance. The associated histological diagnoses range from reactivity changes to malignancies, including extrauterine malignancies ( Zhao et al ., 2009 ; Ajit et al ., 2013 ). AGC is associated with the long-term risk of CC, especially AC ( Wang et al ., 2016 ; Norman et al ., 2022 ). The clinical data of patients diagnosed with AGC through cervical cytology at our hospital were retrospectively analyzed. Histological examination and hrHPV test results were reviewed to explore the clinical significance of AGC and the associated risk of precancerous lesions and malignancies. Additionally, the role of AGC in combination with hrHPV detection in patient management was evaluated.

This study showed that cytological diagnosis of AGC has high predictive value for cervical squamous cell and glandular cell lesions, endometrial lesions, and pelvic malignancies, highlighting the importance of cytological testing for women. A more detailed triage procedure based on the AGC subcategory can reduce the possibility of missed diagnoses of reproductive tract tumors. The combination of hrHPV detection and age can serve as an important basis for the further diagnosis and management of AGC. Precise management should be based on the patient’s AGC type, age, and hrHPV test results.

AGC in cytopathologic diagnosis refers to glandular cells with atypia that is more pronounced than inflammatory or reactive changes but not severe enough for a malignant classification. As a specific category, AGC diagnosis is highly subjective, and related lesions can occur in any part of the female reproductive system. Currently, clinicians are in urgent need a more effective method to accurately identify precancerous lesions and malignancy. The 2014 third edition of the TBS provides a detailed description and classification of the sources of various types of glandular cells, which provides guidance for further diagnosis and treatment. This study performed a detailed analysis based on its criteria. In this study, the AGC-NOS subcategory was the most common, accounting for 49.8% of AGC, followed by AGC-FN (19.2%), AEC (13.4%), AEM (10.9%), and AEC-FN (6.7%). The proportion of AGC subclassification was different from that of published literature ( Boyraz et al ., 2017 ; Kawano et al ., 2020 ; Hassan et al ., 2024 ). In this study, the prevalence of precancerous lesions and malignancies was 58.9%, aligning with the previously reported range of 15.3 to 63.6% ( DeSimone et al ., 2006 ; Zhao et al ., 2009 ; Boyraz et al ., 2017 ; Kawano et al ., 2020 ; Cianfrini et al ., 2024 ; Hassan et al ., 2024 ). EC was most common in precancerous lesions and malignancies (24.4%), followed by AC (10.0%), whereas the most common lesions reported by Hassan et al . (2024 ) were cervical HSIL, followed by endometrial lesions. The most common pathological type reported by Wang et al . (2016 ) was AC. In this study, the prevalence of malignancy was 46.3%, with EC accounting for 52.7%, AC for 21.5%, pelvic malignancy for 18.8%, cervical SCC for 3.8%, and other precancerous lesions and malignancies for 3.2%. However, compared to the findings of Schnatz et al ., the incidence of EC (57.6%), AC (23.6%), cervical SCC (5.4%), and other malignancies (6.9%) was slightly higher, whereas the incidence of pelvic malignancies(6.4%) was lower in this study ( Schnatz et al ., 2006 ). The 2019 American Society of Colposcopy and Cervical Pathology recommendations for AGC remain unchanged. For all subcategories of AGC except AEM, colposcopy and endocervical sampling are recommended regardless of HPV results. For all AGC subcategories in patients aged ≥35 years, endometrial sampling is recommended in combination with colposcopy and endocervical sampling ( Perkins et al ., 2020 ). In this study, 42.6% of cases in the AEC subcategory were cervical lesions, 5.6% were endometrial lesions, and 1.9% were pelvic malignancies. These patients may benefit from an initial colposcopy, followed by a hysteroscopy for endocervical and endometrial sampling. In AEM, 61.4% were endometrial lesions (of which 59.1% were EC), 2.3% were cervical lesions, and 4.5% were other precancerous lesions and malignancies. These patients endometrial sampling by hysteroscopy should be performed first, followed by colposcopy if endometrial lesions are ruled out. In AGC-NOS, 17.4% were endometrial lesions, 17.0% were cervical lesions, 9.0% were pelvic malignancies, and 1.5% were other precancerous lesions and malignancies. For these patients, a combined assessment using colposcopy and hysteroscopy, along with additional imaging examinations such as pelvic MRI, may be performed to exclude other pelvic malignancies. In AEC-FN, 77.8% of lesions were cervical (70.4% of which were AIS/AC), while 3.7% were endometrial, and 3.7% were other precancerous lesions and malignancies. These patients may initially undergo colposcopy and endocervical sampling, followed by hysteroscopy for endometrial sampling. In AGC-FN, 46.8% were endometrial lesions, 17.1% were cervical lesions, 20.5% were pelvic malignancies, and 2.6% were other precancerous lesions and malignancies. These patients are initially more suited for hysteroscopy for endocervical and endometrial sampling, followed by colposcopy and additional imaging examinations to exclude other malignancies. These results confirm that the guideline-based classification of AGC sources aids in evaluating lesion sites, acilitating accurate patient management and reducing waiting times in clinical practice. In studies conducted in other regions of China, the average age of patients with AGC was 43.4 ± 11.5 years ( Xiao et al ., 2023 ), lower than the average age in this study. This may be related to the insufficient implementation of CC screening in our region, causing women to undergo screening at a later age. This further highlights the importance of popularization CC screening. Simultaneously, this study found that women aged ≥50 years have a higher risk of EC and pelvic malignancy. However, cervical HSIL/SCC and AIS/AC are more common in younger women. This is consistent with the findings reported by Aitken et al . (2021 ). Castle et al . reported that EC was positively associated with age and negatively associated with hrHPV positivity ( Castle et al ., 2010 ), which is consistent with our findings. Current screening methods have proven highly effective in identifying cervical squamous intraepithelial lesions, however, the detection of cervical glandular cells remains suboptimal and challenging. In this study, the incidence of AIS/AC (13%) was higher than that of cervical HSIL/SCC (9.9%), this is consistent with previous studies ( Wang et al ., 2016 ; Schiffman et al ., 2023 ). This suggests that cervical cytology plays a role in detecting cervical glandular cells. Previous research has also identified AGC as being associated with a long-term risk of CC ( Cheng et al ., 2011 ; Katki et al ., 2013 ; Wang et al ., 2016 ). Additionally, Wang et al . compared the long-term risks of HSIL and LSIL and found that the risk of CC in AGC was lower than in HSIL but higher than in LSIL ( Wang et al ., 2016 ). Moreover, in this study, 35.2% of AGC patients tested positive for hrHPV. Among hrHPV-positive cases, AEC-FN had the highest rate of hrHPV positivity (81.5%), followed by AEC (55.6%), while AEM had the lowest rate (10.5%). Additionally, a high prevalence of cervical lesions was observed in AEC/AEC-FN and endometrial lesions in AEM. Cervical HSIL/SCC was common in patients who were HPV16 positive and positive for other hrHPV types excluding 16 and 18/45. Conversely, AIS/AC was more common in patients positive for HPV16 (37.3%) and 18/45 (31.4%). Zhou et al . (2024 ) reported a higher prevalence of AIS/AC in patients positive for HPV18/45. This suggests that cervical precancerous lesions and malignancies are closely related to hrHPV infection, particularly HPV16 and HPV18/45. This is consistent with the findings of Norman et al . (2022 ). Additionally, in this study, 69.2% of patients with AC were hrHPV positive, with HPV16 and 18/45 being the most common types, consistent with previous research findings ( de Sanjose et al ., 2010 ; Pirog et al ., 2014 ; Schiffman et al ., 2023 ). Therefore, hrHPV testing is essential for the follow-up of patients with AGC. However, hrHPV-negative patients should remain vigilant, as 3.4% are diagnosed with cervical HSIL/ACC and 6.0% with AIS/AC, while EAH/EC and pelvic malignancies are more commonly observed. Therefore, clinical triage should not rely solely on hrHPV results. It is essential to consider the patient’s AGC subtype and clinical symptoms, such as unexplained vaginal bleeding, long-term anovulation, and obesity. AGC subtype and age can help assess the tumor risk, site, and type during follow-up. This study has several limitations. First, histological specimens from some cases were obtained through colposcopy-guided endocervical and endometrial sampling, which may have led to missed lesions. Second, variations in pathologists’ experience contributed to inconsistencies in cytology results. Furthermore, the study’s retrospective design and relatively small sample size introduced selection bias. To mitigate these limitations, a multicenter prospective study with a larger sample size is required for validation.

None. There are no conflicts of interest.