Cell-free DNA methylation profiling for lung cancer risk stratification in an LDCT-based clinical cohort
Tereza Senkyrova, Petra Sauer Mikyskova, Anezka Kramarova, Alena Mrkvicova, Michal Jirousek, Eva Kocova, Helena Parova, Vladimir Palicka, Radomir Hyspler, Marcela SlavickovaAbstract
Objectives
Low-dose computed tomography (LDCT) screening enables earlier lung cancer detection but identifies indeterminate pulmonary findings in approximately 10–20 % of screened individuals. This study aimed to identify circulating cell-free DNA (cfDNA) methylation markers associated with lung cancer in an LDCT-based clinical cohort.
Methods
The study included 72 patients with histologically confirmed lung cancer, initially indeterminate LDCT findings, or negative LDCT findings. Exploratory reduced representation bisulfite sequencing of pooled cfDNA identified candidate regions for targeted analysis. Individual samples were analysed using a custom xGen methylation panel. CpG sites were preselected using q<0.05 and an absolute methylation difference >5 %. Marker panels were selected by Elastic Net and entered into multivariable logistic regression models. Discriminatory performance was assessed by receiver operating characteristic analysis and optimism by bootstrap resampling.
Results
Twenty-two CpG sites within the RASSF1A and SLC26A4 regions met the predefined selection criteria. Elastic Net yielded a seven-CpG panel for lung cancer vs. initially indeterminate findings, with apparent and optimism-corrected areas under the curve (AUCs) of 0.855 and 0.773. For lung cancer vs. negative findings, the six-CpG panel yielded AUCs of 0.858 and 0.769, while the seven-CpG panel for lung cancer vs. the combined non-carcinoma group yielded AUCs of 0.787 and 0.688.
Conclusions
The seven-CpG panel distinguishing lung cancer from initially indeterminate findings yielded an optimism-corrected AUC of 0.773. CpG sites within the RASSF1A and SLC26A4 regions warrant external validation to determine whether they can provide complementary information to imaging-based malignancy risk assessment in patients with indeterminate LDCT findings.