Fuller application of existing HPV vaccines and test methods can improve cervical cancer control but barriers remain. New technology and improved interventions are needed, especially to improve test acceptance and specificity. Developments in next generation sequencing (NGS) are explored that may radically transform routine molecular diagnostics in 10 to 15 years.
Review of published scientific NGS papers and consideration of recent research in the laboratory of the Wolfson NGS team.
NGS is an ideal method for biomarker discovery and validation and eventually will also be used as a routine test. We have validated a DNA methylation classifier S5 with good performance. Work is underway on an expansion of the S5 biomarker panel by reduced representation bisulfite sequencing on 350 samples from the ARTISTIC study by comparing hrHPV+ women who developed incident CIN2+ versus normal hrHPV+ controls. Our team has shown that a machine learning bioinformatics method (MS-SPCA) can accurately (AUC>0.9) predict CIN2+ years in advance from public NGS data. MS-SPCA has already produced a list of more than 100 potential candidate methylation biomarker genes. Interestingly a gene near the top of the MS-SPCA list is EPB41L3, which is also the independently discovered high performing human gene in the S5 classifier. Our main current aim is to use quasi-genome-wide studies to capture the essence of the very best biomarker genes into small to medium panels. We are also conducting exome sequencing on ARTISTIC samples to identify nucleotide and copy number variations (SNV/CNV) that may reveal CIN2+ molecular signatures in otherwise normal women. Early data have revealed more than 100 high impact SNV that were shared between HPV16 positive CIN3+ and paired HPV16 normal samples taken years earlier from the same women. Corresponding SNV were not found in unmatched normal HPV16 positive women. All women with CIN3+ had changes in HLA-DRB1 several years before the lesions were detected.
NGS is complex and expensive and remains predominantly a research activity. NGS can produce vast amounts of data on samples but improved bioinformatics pipelines are urgently needed. The rapid transformation of NGS data into actionable information could allow clinically meaningful disease risk profiling of individual patients for routine use.