This article summarizes key observations from the peer-reviewed study “Telecytologic diagnosis of cervical smears for triage of self-sampled human papillomavirus–positive women in a resource-limited setting: concept development before implementation.”
Authors: Pierre Vassilakos, Holly Clarke, Micol Murtas, Thomas Stegmüller, Ania Wisniak, Farida Akhoundova, Zacharie Sando, George Enow Orock, Jessica Sormani, Jean-Philippe Thiran, Patrick Petignat
Published: Journal of the American Society of Cytopathology, 2023
DOI: https://doi.org/10.1016/j.jasc.2023.02.001
Cervical cancer remains a major global health challenge, with most deaths occurring in low- and middle-income countries where access to cervical cancer screening is limited. Cervical cancer screening aims to detect abnormal cells and precancerous lesions early to prevent cervical cancer and reduce cervical cancer incidence, particularly in low-income settings where routine cervical cancer screening remains limited. The study notes that primary HPV testing is recommended for women over 30 years in LMICs by the World Health Organization and is also preferred as a primary screening method by the American Cancer Society.
HPV testing has high sensitivity, but HPV testing alone has limited specificity and can lead to unnecessary investigations or overtreatment. For this reason, HPV-positive women need a triage strategy. Cervical cytology is commonly proposed for HPV triage, but conventional cytology can be difficult to implement in settings lacking infrastructure, trained cytotechnologists, and reliable follow-up.
The study evaluated whether telecytology could support Pap smear triage for HPV-positive women in a resource-limited setting. The proposed model was a same-day “test-triage-and-treat” approach, where HPV testing, Pap smear preparation, slide digitization, and remote cytologic diagnosis could be integrated into one screening pathway. Cervical cytology using Pap smear or Pap test was evaluated as a triage approach for identifying abnormal cervical cells and Bethesda categories such as atypical squamous cells of undetermined significance.
Whole slide imaging was central to this concept. Glass slides were digitized into whole slide images using a compact portable Grundium Ocus40 scanner. The scanner used a 12-megapixel image sensor, a 40× objective, and Z-stack acquisition at three focal planes spaced 1 μm apart. The study used manual SurePath liquid-based cytology because it creates a small cell deposit area, which helps reduce scanning time and file size.
To assess operational feasibility, 30 cervical samples were prepared in Geneva using the manual SurePath liquid-based preparation method. Three operators rated the preparation method as very easy. The mean manual preparation time was 20 minutes, and the mean digitization time was 6 minutes per slide.
The resulting whole slide images were rated as good in 18 cases and excellent in 12 cases. Mean estimated cellularity was 6,078 squamous cells per slide. Although 8 slides had cellularity below the 5,000-cell threshold, 6 of these still showed abnormal cells and were considered adequate.
The cost analysis was based on a cervical cancer screening program in Dschang, Cameroon, where about 2,000 women are screened annually and approximately 20% are HPV positive. For 400 HPV-positive women per year, startup costs were estimated at $32,063 and annual operational costs at $2,498. This gave an incremental annual running cost of $6.24 per HPV-positive woman.
The diagnostic performance part of the study used samples from HPV-positive Cameroonian women aged 30 to 49 years. After exclusions, 264 slides were included in the final analysis. Histologic assessment of cervical biopsies and endocervical brushings served as the reference standard.
The overall agreement between whole slide image diagnosis and glass slide diagnosis was 82.6%, with a kappa coefficient of 0.51, indicating moderate agreement. When diagnoses were grouped into three categories, agreement increased to 86.0%, with a kappa coefficient of 0.60. Agreement was highest for grouped diagnoses stratified by ≥CIN3, reaching 90.5%, with a kappa of 0.74.
No statistically significant difference was found in diagnostic accuracy between whole slide imaging and glass slide cytology for detecting ≥CIN2. ROC analysis also showed no statistically significant difference between virtual microscopy and glass slide cytology at the ≥CIN2 or ≥CIN3 thresholds. At the ≥CIN3 threshold, however, WSI showed significantly lower specificity, positive predictive value, and positive likelihood ratio than glass slide cytology.
Screening whole slide images took longer than screening glass slides. Mean screening time was 21 minutes for whole slide images compared with 7 minutes for corresponding glass slides. The authors note that this longer screening time could be a barrier in high-volume laboratories, but may be acceptable in the intended Cameroon setting, where only two or three HPV-positive women per day were expected.
Six cytotechnologists were trained to read digitized slides. Training included a reference atlas, side-by-side comparison of glass slides and whole slide images, and review of Bethesda system categories. After training, cytotechnologists reported that visualization, screening, and annotation on the monitor were easy, although they still preferred glass slides because screening was faster.
The study had important limitations. It was an off-site validation and did not assess the operational time, internet connection strength, or complexity and cost of sending digital slides from Cameroon to remote consultants. The study design also did not allow intraobserver evaluation.
In addition, the number of ≥CIN2 and ≥CIN3 lesions was relatively small. The authors state that future research should include a larger prospective on-site validation study before implementation. They also note that performance should be compared with visual inspection with acetic acid and that combined telecytologic diagnosis with HPV genotyping should be explored.
This study developed a telecytology concept for HPV-positive women in resource-limited settings. Manual liquid-based Pap smear preparation and whole slide imaging were operationally feasible, and virtual microscopy showed diagnostic performance comparable to light microscopy at the ≥ASC-US positivity threshold for detecting ≥CIN2.
These findings contribute to available evidence on digital pathology approaches for cervical screening in low-resource environments and warrant further evaluation of telecytology as an HPV triage tool within same-day cervical cancer screening workflows. However, the authors emphasize that prospective on-site clinical validation is required before broader implementation.
A curated collection of digital pathology studies and references is available on Grundium’s website.
