Fig. 4

Detection of multiple genes from DNA enriched by SEEMLIS. Methylated DNA was enriched by MBD2-MBD capture from DNA extracted from serially diluted LNCaP cells (n = 4 per dilution), 1000 (n = 16) and 100 (n = 8) patient-derived WBCs, and serially diluted LNCaP cells spiked into 1000 patient-derived WBCs (n = 4 per dilution). Enriched methylated DNA was pre-amplified with probes to the indicated genes (excluding LINE1). Pre-amplified DNA was diluted 1:5 and qPCR was performed with the same probes, including LINE1. A For each gene, ROC curves for WBC samples of 1000 cells and LNCaP samples of 1000, 100, or 10 cells were created. Area under the curve (AUC) with 95% confidence interval is indicated. Optimal threshold (OT) values determined by Youden’s J statistic are listed with their associated sensitivity and specificity values. Detection limit was calculated using the slope of the best fit line of Ct values plotted against cell input D. B MI was calculated by delta Ct relative to a max cycles value (MCV) of 35 (GSTP1), 33 (RASSF1, APC, RARB), or 45 (LINE1) with all undetected samples set to the corresponding MCV for analysis. Optimal threshold as determined by ROC curve is shown as a dotted line. Each dot represents an individual sample taken from a pool of cells at the indicated concentration. C Ct values for LINE1 versus Ct values for each gene were plotted against each other for LNCaP samples of 1000, 100, 10, and 1 cell(s). A simple linear regression was performed to determine the line of best fit and 95% confidence interval for that line (shaded region). R and R² values are listed for the correlation. Each gene was significantly correlated to LINE1 values with p values < 0.0001. D Ct values were averaged for each cell input and plotted against the cell input values. A semi-log nonlinear fit was performed to determine the line of best fit and the slope of the line of best fit, which are indicated in parentheses for each gene. All error bars represent standard error of the mean (SEM)