Population-based self-sampling under primary care conditions – a possible approach for cervical cancer screening in Indonesia (IndoCerCa study)

Introduction : In LMIC cervical cancer is a major burden. Screening is mainly based on visual inspection lacking sufficient sensitivity and specificity. For roll-out of colposcopy-based early detection sufficient qualified staff is not available. Several self-sampling device products have been proposed as alternative, but their usability in primary care needs to be proven. Implementation of an HPV-based self-sampling approach in a population-based setting in Indonesia was evaluated. Methods : Four self-sampling devices (2 urine, 2 swab) were applied in a primary care setting covering an entire district in Indonesia with Kulon Progo as pilot region. Cluster randomization was used for comparison of rural and urban areas. HPV-testing was done using standardized and validated PCR-techniques. HPV-positive women and a randomly selected HPV-negative control group underwent colposcopy, PAP smears and biopsies for CIN validation. Results : In 21 primary care units 2056 women (30-55 y) were recruited. Three devices achieved sufficient technical validation (92.1 – 99.7% DNA detection rate). Participant’s test acceptance was 99.1%. HPV-prevalence was 2.6% (urine 2.4%, swab 2.8%). In 29.4% of HPV-positive women high-risk HPV-16/18 were detected. Colposcopy and morphological examination were refused by 3.0% of HPV-positive women and were technically invalid in 5.0%. Pathology revealed NILM in 55.8%, CIN I in 25.0% and CIN II+ in 3.8%. In the control group CIN I was found in 2.0%. This resulted in sensitivity for all CIN/CIN II+ of 27,4%/100,0%, specificity of 98,5%/97,7% with negative (NPV: 97,9%/100,0%) and positive predictive values (PPV: 34,1%/4,5%). Regression analysis confirmed high negative predictive impact of HPV-negativity in women >40 years. Conclusions : Under primary care setting self-sampling-based HPV-testing is accepted. Urine- and swab-based techniques can be applied if the test systems provide technically valid DNA-detection rates. The prevalence was very low and requires further comparison within Indonesia. High NPV of this approach supports its applicability as screening in LMICs. For high-risk lesions PPV is still low suggesting a combination with additional test that are mainly independent from the availability of qualified staff.