The potential of methylated DNA markers for accurate detection of high-risk, HPV-positive cervical lesions

Purpose: Cervical smear cytology exhibits high variability and limited sensitivity in detecting cervical lesions. While the HPV test, recommended by WHO, offers high sensitivity, it has limited specificity. Our goal was to determine whether DNA methylation markers could improve both sensitivity and specificity, thereby enabling more effective triage to colposcopy and biopsy than cytology and HPV testing. Experimental Design: We previously published a 5-gene ( TBXT, MOS, FMN, EDNRB, and ZNF671 ) DNA methylation assay with high sensitivity and specificity for detecting cervical cancer in cytology and histology samples [1]. In the current study, we tested cervical brush samples (n = 222) with known cytology but no available histology that were preserved in CytoLyt, selected and deidentified at the National Health Service screening clinic in Johannesburg, South Africa. Samples were tested for both high-risk HPV variants (Atila Biosystems) and for cumulative 5-gene methylation utilizing Quantitative Multiplex Methylation-Specific PCR (QM-MSP). Statistical tests included the Youden Index for gene-specific thresholds, odds ratio scores for individual genes, Mann-Whitney test, and Receiver Operating Characteristic Area Under the Curve (ROC AUC); a P value ≤ 0.05 was considered significant. Results: Among the 222 cervical brush samples, HPV was prevalent in similar numbers of HSIL (87/95, 92%) and LSIL (47/57, 82%), and in about half of the NIEL (39/70, 56%). Thus, HPV was highly sensitive for the detection of HSIL, but very poorly specific for low-risk LSIL or NIEL. In contrast, among HPV+ samples, to distinguish between high risk HSIL and low risk LSIL/NEIL, the QM-MSP assay showed a high level of sensitivity: 78.16% [CI 68.39 - 85.55] and, in addition, a high level of specificity: 95.35% [CI 88.64 - 98.18], and ROC AUC = 0.890 [CI 83.71 - 94.37]. A highly significant (P<0.0001) difference in methylation levels was observed between HSIL and LSIL/NIELs. In the small set of HPV-negative samples, hypermethylation of the 5-gene panel was observed in 5/8 (63%) HSIL, 0/10 (0%) LSIL, and 1/31 (3%) NIEL, suggesting that QM-MSP may be independent of the HPV status. Conclusions: These findings suggest that the 5-gene panel detected a large proportion of HSIL, while the majority of LSIL and NIEL remained negative. Thus, this assay could serve as an effective triage tool for HPV+ cases, expediting care for women with HSIL while potentially reducing referrals for colposcopy in LSIL and NIEL by up to 95%. This could reduce costs and risks associated with repeated biopsy procedures for low-risk lesions. Reference: [1] DOI: 10.1186/s13148-024-01669-z