Performance of a six-methylation-marker assay on self-collected cervical samples - A feasibility study

OBJECTIVES

In high-income countries, a high proportion of cervical cancers is diagnosed in screening non-attendees. One approach to improve screening coverage is to offer self-sampling for human papillomavirus (HPV) testing. However, especially young women are often HPV positive without having a precancerous lesion in need of treatment. To improve the rather low specificity of HPV testing additional markers such as DNA-methylation can be used. The aim of this feasibility study was to examine the performance of the methylation marker assay GynTect®, comprising six methylation markers, on dry self-collected cervico-vaginal samples compared to physician-taken samples.

METHODS

We recruited 89 patients from our colposcopy clinic of whom 87 qualified for the study. The women took a self-sample with the Evalyn-Brush. Afterwards the planned colposcopy was performed and smears for cytology and reference HPV testing were taken as well as a biopsy in cases of abnormalities. Physician-taken and self-collected samples were tested for HPV DNA and were analyzed with GynTect®.

RESULTS

We obtained 95.5 % valid results for the self-collected samples which was very close to the physician-taken samples. Only about half of the self-collected samples were GynTect® positive in comparison to the physician-taken samples. GynTect® scores were significantly lower for self-collected than for physician-taken samples (p = 0.001, paired t-test). The overall concordance for GynTect® results was moderate (kappa 0.394; p < 0.001). For HPV testing we obtained a good concordance (kappa 0.586; p < 0.001). The GynTect® results for the self-collected samples showed a sensitivity for the detection of cervical intraepithelial neoplasia 2 or worse (CIN2+) of 26.1 % (95 %-CI: 0.13-0.46) and a specificity of 95.6 % (95 %-CI: 0.85-0.99), in comparison to a sensitivity of 45.5 % (95 %-CI: 0.27-0.65) and a specificity of 78.3 % (95 %-CI: 0.64-0.88) for the physician-taken samples.

CONCLUSIONS

GynTect® methylation marker testing has a satisfactory amount of valid results on self-collected samples. However, the results of the self-collected samples differed clearly in comparison to the reference samples. To justify an application in screening, a larger study with more cases of high-grade cervical dysplasia and HPV positive patients will be needed.

Association of genomic variants at PAX8 and PBX2 with cervical cancer risk

Cervical malignancy is triggered by human papillomavirus infection but the risk for cervical cancer has a hereditary component. From a recent Genome Wide Association Study meta-analysis, 2q14.1 (PAX8) and 6p21.32 (PBX2) have been proposed as novel cervical cancer susceptibility loci. We investigated the two main signals at these loci in an independent case-control series of 2578 cases with cervical dysplasia or carcinoma and 1483 healthy females. We find significant associations for both variants, rs10175462 at PAX8 and rs2856437 at PBX2, with overall cervical disease (rs10175462: odds ratio [OR] 0.82, 95% confidence interval [CI] 0.74-0.91, P = 2.4 × 10-4 ; rs2856437: OR 1.52, 95% CI 1.14-2.02, P = .004). Both variants showed evidence of association with invasive squamous cervical cancer (rs10175462: OR 0.80, 95% CI 0.68-0.94, P = .006; rs2856437: OR 1.56, 95% CI 1.03-2.36, P = .036) and with high-grade dysplasia (rs10175462: OR 0.79, 95%CI 0.70-0.90, P = 1.9 × 10-4 ; rs2856437: OR 1.58, 95% CI 1.15-2.17, P = .005). A combined analysis of high-grade dysplasia and invasive cervical cancer also showed significant associations for both variants (rs10175462: OR 0.81, 95% CI 0.73-0.91, P = 2.4 × 10-4 ; rs2856437: OR 1.57, 95% CI 1.18-2.10, P = .002). No association was detected for rs2856437 with low-grade dysplasia, while rs10175462 showed weak evidence of association (P = .05). RNA analyses in cervical samples revealed that PAX8 transcripts were upregulated in HPV-positive lesions (P = .008) but this was not observed in the presence of the protective minor allele of rs10175462. The rs10175462 genotype also correlated with reduced levels of the lncRNA PAX8-AS1 (P < .001). Taken together, our results extend the evidence for a link between genomic risk variants at the HLA region (PBX2) with cervical disease and support PAX8 as the first consistent non-HLA cervical cancer susceptibility locus.

Association of genomic variants at the human leukocyte antigen locus with cervical cancer risk, HPV status and gene expression levels

The human leukocyte antigen (HLA) locus on chromosome 6 has been reported to be associated with cervical cancer. We investigated two independent single-nucleotide polymorphisms in a large case-control series of cervical dysplasia and carcinoma that has been newly established by the German Cervigen Consortium, comprising a total of 2481 cases and 1556 healthy females. We find significant associations for both variants, rs9272117 at HLA-DQA1 and rs2844511 at MICA and HCP5, with cervical disease. Both variants showed evidence of association with invasive cervical cancer (rs9272117: OR 0.89, 95% CI 0.79-0.99, P = .036; rs2844511: OR 1.17, 95% CI 1.04-1.31, P = .008) and with high-grade dysplasia (rs9272117: OR 0.78, 95% CI 0.70-0.87, P = 7.1 × 10^-6 ; rs2844511: OR 1.13, 95% CI 1.01-1.26, P = .035), as well as in a combined analysis of both groups (rs9272117: OR 0.83, 95% CI 0.75-0.91, P = 6.9 × 10^-5 ; rs2844511: OR 1.14, 95% CI 1.04-1.26, P = .005). Variant rs2844511, but not rs9272117, also showed modest evidence of association with low-grade dysplasia (OR 1.26, 95% CI 1.04-1.54, P = .019). In case-only analyses, rs2844511 tended to predict HPV status (P = .044) and rs9272117 tended to associate with HPV16 (P = .022). RNA studies in cervical samples showed a significant correlation in the transcript levels of MICA, HCP5 and HLA-DQA1, suggesting extensive co-regulation. All three genes were upregulated in HPV16-positive samples. In stratified analyses, rs9272117 was associated with HLA-DQA1 levels, specifically in HPV-positive samples, while rs2844511 was associated with MICA and HCP5 levels. The risk allele of rs2844511 was required for correlations between MICA or HCP5 with HLA-DQA1. Altogether, our results support 6p21.32-33 as the first consistent cervical cancer susceptibility locus and provide evidence for a link between genetic risk variants, HPV16 status and transcript levels of HLA-DQA1, HCP5 and MICA, which may contribute to tumor immune evasion.