Automatic CIN Grades Prediction of Sequential Cervigram Image Using LSTM With Multistate CNN Features

Human papillomavirus (HPV) DNA testing has shown higher sensitivity than cytology for detecting cervical lesions, but it is uncertain whether the higher sensitivity is dependent on the age of the woman being screened. We compared the age-specific performance of primary HPV DNA screening with that of conventional cytology screening in the setting of an organized population-based cervical cancer screening program in Finland. From January 1, 2003, to December 31, 2005, randomized invitations were sent to women aged 25-65 years for routine cervical cancer screening by primary high-risk HPV DNA testing (n = 54 207) with a Hybrid Capture 2 assay followed by cytology triage for women who were HPV DNA positive or by conventional cytology screening (n = 54 218). In both screening arms, cytology results of low-grade squamous intraepithelial lesion or worse triggered a referral for colposcopy. Relative rates (RRs) of detection to assess test sensitivity, specificity, and positive predictive values (PPVs) with 95% confidence intervals (CIs) were calculated for the histological endpoints of cervical intraepithelial neoplasia (CIN) grade 1 or higher (CIN 1+), CIN grade 2 or higher (CIN 2+), and CIN grade 3 or higher (CIN 3+). All statistical tests were two-sided. The overall frequency of colposcopy referrals was 1.2% in both screening arms. Women younger than 35 years were referred more often in the HPV DNA screening vs the conventional screening arm (RR = 1.27, 95% CI = 1.01 to 1.60). The prevalence of histologically confirmed CIN or cancer was 0.59% in the HPV DNA screening arm vs 0.43% in the conventional screening arm. The relative rates of detection for CIN 1, CIN 2, and CIN 3+ for HPV DNA screening with cytology triage vs conventional screening were 1.44 (95% CI = 0.99 to 2.10), 1.39 (95% CI = 1.03 to 1.88), and 1.22 (95% CI = 0.78 to 1.92), respectively. The specificity of the HPV DNA test with cytology triage was equal to that of conventional screening for all age groups (99.2% vs 99.1% for CIN 2+, P = .13). Among women aged 35 years or older, the HPV DNA test with cytology triage tended to have higher specificity than conventional screening. The PPVs for HPV DNA screening with cytology triage were consistently higher than those for conventional screening. In both screening arms, the test specificities increased with increasing age of the women being screening, whereas the highest PPVs were observed among the youngest women being screened. Overall, 7.2% of women in the HPV DNA screening arm vs 6.6% of women in the conventional screening arm were recommended for intensified follow-up, and the percentages were highest among 25- to 29-year-olds (21.9% vs 10.0%, respectively). Primary HPV DNA screening with cytology triage is more sensitive than conventional screening. Among women aged 35 years or older, primary HPV DNA screening with cytology triage is also more specific than conventional screening and decreases colposcopy referrals and follow-up tests.

Objective: Using methylation characteristics of human genes to construct machine learning predictive models for screening cervical cancer and precancerous lesions. Methods: Human DNA methylation detection was performed on 224 cervical exfoliated cell specimens from the Cancer Hospital of the Chinese Academy of Medical Sciences, Tianjin Central Hospital of Gynecology Obstetrics, Xinmi Maternal and Child Health Hospital of Henan Province, West China Second Affiliated Hospital of Sichuan University, and Heping Hospital Affiliated to Changzhi Medical College collected during April 2014 and March 2015. The hypermethylated gene fragments related to cervical cancer were selected by high-density, high-association, and hypermethylated gene fragment screening and the LASSO regression algorithm. Taking cervical intraepithelial neoplasia grade 2 (CIN2) or more severe lesions as the research outcome, machine learning predictive models based on the random forest (RF), naive Bayes (NB), and support vector machine (SVM) algorithm, respectively, were constructed. A total of 144 outpatient specimens were used as the training set and 80 cervical exfoliated cell specimens from women participating in the cervical cancer screening program were used as the test set to verify the predictive models. Using histological diagnosis results as the gold standard, the detection efficacy for CIN2 or more severe lesions of the three machine learning predictive models were compared with that of the human papilloma virus (HPV) detection and cytological diagnosis. Results: In the training set of 144 cases, there were 34 cases of HPV positivity, with a positive rate of 23.61%. Cytologically, there were 37 cases diagnosed as no intraepithelial lesion or malignancy (NILM), and 107 cases diagnosed as atypical squamous cells of undetermined significance (ASC-US) or above. Histologically, there were 28 cases without cervical intraepithelial neoplasia or benign cervical lesions, 31 cases of CIN1, 18 cases of CIN2, 31 cases of CIN3, and 36 cases of squamous cell carcinoma. Seven hypermethylated gene fragments were selected from 45 genes, and three machine learning prediction models based on the RF, NB, and SVM algorithm, respectively, were constructed. In the validation set of 80 cases, there were 28 cases of HPV positivity, with a positive rate of 35.00%. Cytologically, there were 65 cases diagnosed as NILM and 15 cases as ASC-US or above. Histologically, there were 39 cases without cervical intraepithelial neoplasia or benign cervical lesions, 10 cases of CIN1, 10 cases of CIN2, 11 cases of CIN3, and 10 cases of squamous cell carcinoma. In the validation set, the area under the curve (AUC) values of the RF model, NB model, SVM model, HPV detection, and cytological diagnosis of CIN2 or above were 0.90, 0.88, 0.82, 0.68, and 0.45, respectively. The DeLong test showed that there was no statistically significant difference in the AUC values between the RF, NB, and SVM models (all P＞0.05), and the AUC values of the RF and NB models were higher than that of HPV detection (both P＜0.01), and the AUC values of the RF, NB, and SVM models were higher than that of cytological diagnosis (all P＜0.01). Compared with the NB model, the sensitivity of the RF model was similar (80.65% vs. 77.42%), but the specificity of the NB model was much higher than that of the RF model (93.88% vs. 73.47%). Conclusion: Among the machine learning prediction models for cervical cancer and precancerous lesions constructed based on human DNA methylation, the NB model has good predictive performance for CIN2 and above lesions, and may be used for screening of cervical cancer and precancerous lesions.

Results of a randomized trial on the management of cytology interpretations of atypical squamous cells of undetermined significance

While high-risk human papillomavirus (HPV) is the established cause of cervical intraepithelial neoplasia (CIN), cofactors influencing progression to high-grade disease remain poorly defined. This study aimed to investigate the association between long-term oral contraceptive use and the risk of high-grade CIN (grade 2/3) in a high-risk referral population. A retrospective analysis was conducted on the records of 684 women screened at a specialized gynecologic oncology clinic (2019-2024). Data on oral contraceptive use (≥5 years), high-risk HPV status, age, and smoking history were extracted. The primary outcome was CIN grade. To mitigate bias from sparse data, Firth-penalized logistic regression was used to calculate adjusted ORs (aOR) for the association between long-term oral contraceptive use and CIN grade. Long-term oral contraceptive use was reported by 5.6% of the cohort. After adjusting for confounders, long-term oral contraceptive use was strongly associated with a 16.8-fold increased odds of high-grade CIN (aOR 16.79, 95% CI 3.82 to 73.70, p < .001). However, the wide CI indicates significant statistical uncertainty in the magnitude of the effect. Conversely, oral contraceptive use was associated with significantly lower odds of low-grade CIN (aOR 0.04, 95% CI 0.01 to 0.15, p < .001). In this high-risk population, long-term oral contraceptive use is a significant independent risk factor for high-grade cervical lesions, while paradoxically being associated with a lower risk of low-grade disease. This suggests an important role for hormonal factors in the progression of cervical neoplasia. Clinicians should consider long-term oral contraceptive use a key risk indicator in HPV-positive women and incorporate this modifiable factor into patient counseling.

Women living with HIV have an elevated risk for cervical cancer, present earlier, and have more recurrent human papillomavirus (HPV) infections compared with women without HIV. To update WHO recommendations on screening and treatment to prevent cervical cancer, we aimed to identify whether women living with HIV should be screened for cervical cancer at a specific age, the optimal screening interval following a negative cervical screen, and the screening interval following treatment. We conducted a systematic literature review on cervical cancer and HIV by searching MEDLINE, Embase, CENTRAL, the Cochrane Library, and clinical trial registries covering Jan 1, 2012, through to Oct 13, 2019, updating a previous systematic review covering database inception to July, 2012. Included articles reported original data and assessed one or more outcomes related to cervical precancer and cancer in women living with HIV; no restrictions on study design or setting were made. Articles were excluded if they did not include any women living with HIV, or if they reported only HPV genotype prevalence without any other relevant data. Two authors extracted data from any study reporting cervical cancer screening tests and histopathologically confirmed disease outcomes, by age. We analysed summary data on optimal age and screening intervals, providing pooled estimates when possible; we then conducted an individual patient data meta-analysis (IPDMA) to analyse age-specific data on cervical cancer and precancer. Authors of studies with 40 or more women living with HIV and cervical intraepithelial neoplasia (CIN) grade 2+ were invited to submit individual patient data for meta-analysis. Random-effects models were used to calculate predicted probabilities for cervical cancer screening results by age, HIV status, and antiretroviral therapy (ART) status. Of the 304 full-text articles screened, 34 studies from 12 countries, with 128 732 women, including 63 790 women living with HIV, were included in the systematic review. Of 55 studies potentially eligible for the IPDMA, eight studies provided data for 72 350 women, 12 527 of whom were living with HIV, from seven countries (Burkina Faso, Cameroon, India, Kenya, South Africa, Thailand, and the USA). In the IPDMA, the pooled predicted probability of CIN2 or CIN3 among women living with HIV increased from 6·0% (95% CI 0·74-64·1) for ages 15-19 years to 32·4% (8·3-72·7) for ages 20-24 years, 42·1% (16·4-80·2) for ages 25-29 years, 50·3% (16·3-80·0) for ages 30-34 years, 47·0% (16·3-80·0) for ages 35-39 years, 49·0% (16·3-80·2) for ages 40-44 years, 58·1% (17·0-81·5) for ages 45-49 years, and 55·3% (21·0-86·6) for age 50 years and older; invasive cervical cancer was uncommon before 30 years of age. In the systematic review, women living with HIV who had a negative baseline cytology result and negative HPV test had a cumulative incidence of developing CIN2+ that ranged from 0·8% to 5% within 4·2-6·4 years and a cumulative incidence of developing of CIN of any grade up to 10% within 12 years. Also in the systematic review, women living with HIV had high recurrence of CIN2+ following treatment (11-27% by 1 year follow-up, 3-64% by 3 years, and 57% by 10 years). No significant evidence of publication bias was found in the data included in the IPDMA (Egger's test p=0·83). Our data show a clear, age-related increase in CIN2 and CIN3 among women living with HIV, with the highest risk occurring in the 45-49-year age group. Our findings informed WHO recommendations to initiate cervical cancer screening for women living with HIV at age 25 years, with regular screening every 3-5 years. Expanding screening and treatment is necessary to reduce cervical cancer incidence towards its elimination. US Agency for International Development and US President's Emergency Plan for AIDS Relief.

A randomized trial on the management of low-grade squamous intraepithelial lesion cytology interpretations

This study aims to evaluate human papillomavirus (HPV) viral loads as a biomarker for triage into colposcopy and cervical intraepithelial neoplasia grade 2 (CIN2) therapy to reduce the colposcopy referral rate and CIN2 overtreatment in low-resource settings. In 1999, 1997 women aged 35 to 45 years in Shanxi, China, received 6 cervical screenings with pathological confirmation. In 2005, 1461 histologically normal women, 99 with CIN grade 1 (CIN1), and 30 with CIN2 or worse (CIN2+) were rescreened in a follow-up study. Human papillomavirus was detected by Hybrid Capture 2. Viral load, estimated by the ratio of relative light units to standard positive control (RLU/PC), was categorized into 4 groups: negative (<1.0), low (≥1.0, <10.0), moderate (≥10.0, <100.0), and high (≥100.0). We estimated the cumulative incidence of CIN2+ by viral load subgroups and calculated adjusted hazard ratios for CIN2+ using Cox proportional hazards regression. Cumulative incidence of CIN2+ increased with baseline HPV viral loads among normal women and women with CIN1 at baseline (P(-trend) < 0.001). Repeat moderate-high viral load was associated with the highest risk for CIN2+ (adjusted hazard ratio, 188.8; 95% confidence interval, 41.2-864.1). Raising the ratio of relative light units to standard positive control cutoff from 1.0 to 10.0 for colposcopy greatly reduced the referral rate from 18.1% to 12.9%. It also increased the specificity (84.8% vs 90.4%), the positive predictive value (22.5% vs 28.9%), and the positive likelihood ratio (6.4 vs 8.9), yet with loss of sensitivity by 12% (97.6% vs 85.7%). Among women with CIN2 at baseline, 56% regressed to normal, 24% regressed to CIN1, 4% remained CIN2, and 16% progressed to CIN grade 3 or worse. Locales using HPV testing as the primary screening method and lacking high-quality cytology-based screening should consider viral load as an alternative to colposcopy triage for women older than 35 years. Viral loads may also predict CIN2 progression until additional biomarkers become available.

Cervical cancer (CC) screening in women comprises human papillomavirus (HPV) testing followed by cytology triage of positive cases. Drawbacks, including cytology’s low reproducibility and requirement for short screening intervals, raise the need for alternative triage methods. Here we used an innovative triage technique, the WID-qCIN test, to assess the DNA methylation of human genes DPP6, RALYL and GSX1 in a real-life cohort of 28,017 women aged ≥30 years who attended CC screening in Stockholm between January and March 2017. In the analysis of all 2,377 HPV-positive samples, a combination of WID-qCIN (with a predefined threshold) and HPV16 and/or HPV18 (HPV16/18) detected 93.4% of cervical intraepithelial neoplasia grade 3 and 100% of invasive CCs. The WID-qCIN/HPV16/18 combination predicted 69.4% of incident cervical intraepithelial neoplasia grade 2 or worse compared with 18.2% predicted by cytology. Cytology or WID-qCIN/HPV16/18 triage would require 4.1 and 2.4 colposcopy referrals to detect one cervical intraepithelial neoplasia grade 2 or worse, respectively, during the 6 year period. These findings support the use of WID-qCIN/HPV16/18 as an improved triage strategy for HPV-positive women.

Incidence of ocular surface squamous neoplasia in pterygium specimens

Evaluation of women presenting with postcoital bleeding by cytology and colposcopy

Introduction: Cervical cancer is the third most common cancer and the second most frequent cause of mortality from malignancies of genital organs in women, which can be prevented by diagnosis of pre-neoplastic changes in cervix. This study aimed to evaluate the p16INK4a biomarker in different grades of cervical intraepithelial neoplasia (CIN) using immunohistochemistry (IHC) method. Methods: The present cross-sectional study was carried out on the paraffin-embedded blocks of cervical tissue of 100 women with previous histopathological diagnosis of CIN referred to Al-Zahra Hospital, Tabriz, Iran, during 2015-2016. The samples were divided into 4 groups, 31 with normal cervical finding, 30 with low-grade CIN (CIN I), and 39 with high-grade CIN (16 CIN II and 23 CIN III). p16INK4a was investigated on the samples using IHC technique. Data was analyzed by SPSS using chi-square and Mann-Whitney U tests. Results: Thirteen out of 30 (43%), 12 out of 16 (75%) and 23 out of 23 (100%) of the CIN I, CIN II, CIN III were positive for p16INK4a, respectively. None of the normal samples were positive for p16INK4a. Sensitivity of p16INK4a for detection of CIN I, CIN II, CIN III was calculated as 63%, 80% and 100%, respectively. The overall sensitivity of the biomarker for detection of CIN lesions was 76.6% and the specificity was 100% for all CIN grades.Conclusion: The p16INK4A biomarker is a suitable diagnostic tool for high-grade CIN, yet for low-grade ones it has lower sensitivity. p16INK4a can be a helpful tool beside histopathology for diagnosis of CIN lesions.

Using nationwide Danish registries we examined the long-term risk of cervical cancer in women diagnosed with cervical intraepithelial neoplasia grade 3 (CIN3) (including adenocarcinoma in situ (AIS)) on the cone compared to women with a normal cytology test. Initially, we identified women born 1918-1990, who were recorded as living in Denmark between January 1, 1978 and December 31, 2012. From the Pathology Data Bank information on CIN3 on the cone, margins status, histological type of CIN3 and cervical cytology results was extracted. Cox proportional hazard model was used to estimate the relative risk of subsequent cervical cancer. We included 59,464 women with CIN3 on the cone and 1,918,508 women with a normal cytology test. Overall, women diagnosed with CIN3 had a higher risk of subsequent cervical cancer compared to women with normal cytology (HR = 2.06; 95%CI: 1.81-2.35). Analyses according to time since conization showed elevated risks in all time periods, and 25 years or more after conization the relative risk was significantly increased (HR = 2.56; 95%CI: 1.37-4.77). Twenty years or more after conization, also women with negative margins had an increased relative risk (HR = 2.49; 95%CI: 1.12-5.57). In addition, the long-term relative risk of cervical cancer varied with the different histological types of CIN3 and was highest for AIS (HR = 7.50; 95%CI: 1.87-30.01, 10-14 years after conization). In conclusion, women diagnosed with CIN3 on the cone have a long-lasting increased risk of cervical cancer even when the margins on the cone are negative.

Purpose: Cervical cancer disproportionately affects women in in low- and middle-income countries, where it accounts for 85% of cases and 90% of deaths globally. Although WHO’s 90/70/90 cervical cancer elimination strategy aims to treat 90% of cervical precancers by 2030, treatment access remains limited. In Kenya, only 5% of hospitals provide precancer treatment, and just 30% of women in 2022 received it. Innovative methods, such as self-administered topical therapies, could help bridge this treatment gap. Artesunate, derived from the antimalarial artemisinin, has shown promise against HPV in lab studies. In a Phase I U.S. trial, self-administered intravaginal artesunate led to 63% disease regression of cervical intraepithelial neoplasia grade 2/3 (CIN2/3), and a Phase II is underway. We are conducting a Phase I trial in Kenya (NCT06165614) to evaluate its feasibility as CIN2/3 therapy. Methods: Eighteen women with CIN2/3 will self-administer intravaginal artesunate for 5 days on weeks 1, 3, 5, and 7. The primary objective is safety; secondary outcomes include adherence, histologic regression (CIN1 or lower), and acceptability. Treatment response is evaluated by colposcopy and biopsy at week 14. Those with persistent CIN2/3 undergo excision, while those with regression to CIN1 or less will be monitored. The trial began in March 2024. Results: Seventeen women (mean age of 39.9 years) are enrolled; 15 (88.2%) are HIV-positive and on antiretroviral therapy, with a mean CD4 count of 627.8 cells/mm3. Most participants (58.8%) are married and have at least a primary education (64.7%). All reported at least one grade 1 adverse event (AE), commonly vaginal discharge (60%) and headache (53.3%). Six grade 2 AEs (perineal or pelvic pain, headache) were reported, with no grade 3 or 4 AEs. Adherence, assessed by self-report and returned pessary covers, was excellent. Of 15 participants completing week 14 colposcopy visits, 11 (73.3%) had persistent CIN2/3 or worse and underwent excision or hysterectomy, while 4 (26.6%) regressed to CIN1 or less and are under observation. Study visits continue, with samples stored for HPV genotyping and microbiome testing. Conclusion: Early U.S. studies of intravaginal artesunate showed potential efficacy for CIN2/3. However, preliminary results from our ongoing Phase I study in Kenyan women, most of whom are living with HIV, suggest lower response rates. Further analyses will explore the impact of HPV genotype and microbiome. Citation Format: Chemtai Mungo, Cirilus Ogollah, Gershon Rota, Kate Sorgi, Elizabeth Bukusi, Jennifer H. Tang, Lisa Rahangdale, Jackton Omoto. Phase I Study of Self-Administered Artesunate Pessaries for Treatment of Cervical Precancer in HIV- Positive and HIV-Negative Women in Kenya [abstract]. In: Proceedings of the 13th Annual Symposium on Global Cancer Research; 2025 Sep 16. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2025;34(12_Suppl):Abstract nr 68.

Lower Anogenital Squamous Terminology (LAST) standardization recommended p16INK4a immunohistochemistry (p16 IHC) for biopsies diagnosed morphologically as cervical intraepithelial neoplasia (CIN) grade 2 (CIN2) to classify them as low-grade or high-grade squamous intraepithelial lesions (HSILs). To describe the relationships of p16 IHC and other biomarkers associated with cervical cancer risk with biopsy diagnoses. A statewide, stratified sample of cervical biopsies diagnosed by community pathologists (CPs), including 1512 CIN2, underwent a consensus, expert pathologist panel (EP) review (without p16 IHC results), p16 IHC interpretation by a third pathology group, and human papillomavirus (HPV) genotyping, results of which were grouped hierarchically according to cancer risk. Antecedent cytologic interpretations were also available. Biopsies were more likely to test p16 IHC positive with increasing severity of CP diagnoses, overall (Ptrend ≤ .001) and within each HPV risk group (Ptrend ≤ .001 except for low-risk HPV [Ptrend < .010]). All abnormal grades of CP-diagnosed biopsies were more likely to test p16 IHC positive with a higher HPV risk group (Ptrend < .001), and testing p16 IHC positive was associated with higher HPV risk group than testing p16 IHC negative for each grade of CP-diagnosed biopsies (P < .001). p16 IHC-positive, CP-diagnosed CIN2 biopsies were less likely than CP-diagnosed CIN3 biopsies to test HPV16 positive, have an antecedent HSIL+ cytology, or to be diagnosed as CIN3+ by the EP (P < .001 for all). p16 IHC-positive, CP-diagnosed CIN1 biopsies had lower HPV risk groups than p16 IHC-negative, CP-diagnosed CIN2 biopsies (P < .001). p16 IHC-positive, CP-diagnosed CIN2 appears to be lower cancer risk than CP-diagnosed CIN3. LAST classification of "HSIL" diagnosis, which includes p16 IHC-positive CIN2, should annotate the morphologic diagnosis (CIN2 or CIN3) to inform all management decisions, which is especially important for young (<30 years) women diagnosed with CIN2 for whom surveillance rather than treatment is recommended.

We investigated telomerase expression in cervical lesions and its diagnostic value for cervical cancer and the precancerous lesions. The subjects were 100 women who underwent cervical biopsy, comprising 16 normal cervix, 61 cervical intraepithelial neoplasia (CIN) and 23 invasive cervical cancer. Telomerase activity was detected by the telomerase repeat amplification protocol using a non-radioisotope system, and quantitated using a DNA image analyzing system. Telomerase was detected at 18.8% (3/16) in normal cervix, 32. 0% (8/25) in CIN I, 50.0% (3/6) in CIN II, 60.0% (18/30) in CIN III, and 91.3% (21/23) in invasive cervical cancer. Semiquantitation of telomerase activity was significantly higher in the subjects with invasive cancer than in those with CIN or a normal cervix (p<0.05). However, telomerase activity was not significantly different between the grades of CIN. Telomerase was detected in subjects with CIN, suggesting that telomerase expression is an early event in cervical carcinogenesis. Quantitation of telomerase activity may be helpful for the diagnosis of cervical cancer.