100 Chapter 4 0.563 and 0.550 for self-collected samples versus clinician-collected samples, 0.459 and 0.359 for self-collected samples versus tissue specimens and 0.524 and 0.429 for clinician-collected samples versus tissue specimens, respectively). The methylation levels of ASCL1 and LHX8 in self-collected samples were lower compared to those in paired clinician-collected cervical samples (both p values < 0.0001). DISCUSSION In this study, we showed that methylation levels of ASCL1 and LHX8 in hrHPV-positive self-collected samples correlated with underlying disease severity and were significantly higher in women with CIN3+ than in control women with no evidence of disease. The CIN3+ sensitivity of the ASCL1/LHX8 marker panel on hrHPV-positive self-collected samples was 73.3% (95% CI 63.9-82.6%) at a specificity of 61.1% (95% CI 56.9-65.4%). The triage performance of ASCL1/LHX8 methylation analysis on self-collected samples was somewhat lower than that on clinician-collected cervical samples (relative sensitivity for CIN3+ detection 0.95, 95% CI 0.82-1.10 and relative specificity 0.82, 95% CI 0.75-0.90). Our data indicate that the methylation marker panel ASCL1/LHX8 constitutes a feasible direct triage method for the detection of CIN3+ in hrHPV-positive women participating in routine screening by self-sampling. The advantage of DNA methylation analysis as a triage test is the use of an objective, non-morphological assay, with a high reproducibility 32, directly applicable to self-collected samples. Importantly, about half of the women with ≥ASC-US on the paired clinician-collected cervical sample taken for cytology triage, and ~75% of the women with histological samples of CIN3+ could have been directly referred for colposcopy, without recall for clinician-collection, after ASCL1/LHX8 methylation analysis on the hrHPV-positive self-collected sample. In our study, the performance of the bi-marker panel ASCL1/LHX8 did not differ from that of HPV16/18 genotyping. Of interest, methylation positivity rate did not differ across the various genotypes, supporting the continued value of host-cell DNA methylation markers in the post-vaccination era 33. HPV16/18 genotyping and ASCL1/LHX8 methylation analysis were to a certain extent complementary in line with previous findings 16, 34. Although comparison with cytology on the paired clinician-collected sample must be done with caution due to the fact that the hrHPV-positive women in the IMPROVE study were managed based on cytology, HPV16/18 genotyping in combination with the bi-marker panel ASCL1/LHX8 increased CIN3+ sensitivity to a level that did not differ from cytology triage on a clinician-collected cervical sample, though at the cost of a decreased specificity.
RkJQdWJsaXNoZXIy MjY0ODMw