The Selfie study: Cervical precancer detection using novel human papillomavirus biomarkers.

10533 Background: Human papillomavirus (HPV) causes the majority of cervical cancers worldwide. Dual stain cytology can detect HPV oncogenic activity through biomarkers p16/Ki-67 in cervical samples. Dual stain is an acceptable triage strategy for HPV-positive results from clinician-collected samples, reducing the number of low-risk individuals sent for colposcopy or treatment. However, the ability of dual stain to triage self-collected vaginal specimen is poorly understood. The Selfie Study is an observational study to assess the diagnostic accuracy of different biomarkers for cervical precancer on self-collected samples. Here, we evaluated dual stain cytology in clinician and self-collected specimens for detection of cervical precancer. Methods: Individuals with a cervix ages 25-69 years undergoing cervical cancer screening, colposcopy, or treatment at George Washington University (GWU) and Sarasota Memorial Hospital (SMH) were included in this study (August 2020-August 2024). Participants were instructed to perform self-collection prior to clinician-collection during the clinic visit. Demographics and clinical outcomes of dual stain on paired cervicovaginal samples and presence of cervical intraepithelial neoplasia 2 or worse (CIN2+) were recorded and assessed using descriptive statistics. Differences in sensitivity and specificity of dual stain to detect CIN2+ between the two collection methods were assessed using McNemar’s test. Results: A total of 548 participants enrolled in Selfie. Paired dual stain results were available for 411 participants (411/548, 75%), of which 24 participants were CIN2+ (24/411, 5.8%). Missing results were due to inadequate/absent staining for one or both collection methods. Dual stain was positive in 63 clinician-collected samples (63/411, 15.3%) and in 39 (39/411, 9.5%) self-collected samples. Overall percent agreement between collection methods was 89.3% (367/411). Percent positive agreement was 39.7% (29/73). Dual stain was 83.3% (20/24) sensitive to detect CIN2+ in clinician-collected samples and 45.8% (11/24) sensitive in self-collected samples (83.3% versus 45.8%, p = 0.003). Conclusions: Dual stain on self-collected samples had significantly lower sensitivity for detection of CIN2+ compared to clinician-collected samples. Additional research is needed to evaluate whether the performance is more comparable for CIN3+ endpoints. Our study highlights the challenges of conducting triage assays from self-collected specimens. Clinical trial information: NCT04423679 .