Fig. 1: Workflow of the study design and Consort diagram.
From: Circulating cell-free DNA-based methylation patterns for breast cancer diagnosis
a The breast cancer (BC) methylation panel targets were specifically selected as those areas with differences in methylation between tumour and normal tissue based on The Cancer Genome Atlas (TCGA) data. Forty paired tissue and plasma samples were used to identify a total of 1996 methylation pattern markers. Targeted methylation profiling for cancer/normal classification was performed as follows: LASSO and random forest analyses were applied to a training cohort of 52 normal controls and 108 breast cancer patients to identify a final selection of 26 markers. These 26 markers were applied to a validation cohort of 22 normal controls and 47 breast cancer patients. The final performance of our models was tested in an independent test cohort of 55 normal controls and 49 breast cancer patients. b Participants screened and enrolled. QC quality criteria. Samples in tissue and plasma underwent targeted capture-probe sequencing via high-throughput sequencing with the ultrasensitive AnchorIRISTM assay. Our goal was to select enough controls and malignant plasma, but due to the available clinical information of samples and exclusion criteria, we were not able to identify all assessed samples that entered the analysis. Finally, 40 pairs of breast cancer tissue DNA and matched plasma cfDNA were used for filtering, and 336 plasma cfDNA were randomly assigned to a training group (n = 160), validation group (n = 69) and independent testing group (n = 104).
