Abstract 4272: Improving measurement of assay variation and Limit of Detection (LOD) of ctDNA assay by using validation samples contrived with DNA-free human plasma

The usage of PCR-based or NGS-based liquid biopsy assays for circulating tumor DNA (ctDNA) detection is growing rapidly in clinical labs. However, inconsistent results are often observed especially when assays from different developers are compared. One reason for the discrepancy may be a lack of cost-effective evaluation samples based on true human plasma without which the performance parameters such as Limit of Detection (LOD) are often measured incorrectly. To confirm the necessity of using human plasma as sample matrix, the following study was conducted. We have developed a proprietary technology that specifically removes DNA from plasma without affecting the composition of the plasma. This overcomes the scarcity of patient sample and uncertainty of endogenous cfDNA from an unknown patient sample. Based on DNA-depleted plasma, patient-like ctDNA validation samples were made by quantitatively spiking in a ctDNA fragment, BRAF V600E, together with wild type genomic DNA fragments. Linearity and sensitivity of allele frequency (AF%) determination were performed in qPCR assay. Matrix effect on LOD measurement was studied among ctDNAs made with TE buffer, BSA-based “synthetic plasma,” and human plasma at 10ng/ml pre-spiked cfDNA level in each matrix. The effect of different extraction methods on LOD measurement was compared using three different commercially available cfDNA extraction kits. Characterization of DNA-depleted plasma, including measurement of protein, lipids, and exosome content showed comparable results before and after depletion. cfDNAs spiked into the plasma were stable for at least 135 days at 2-8 °C. Number of copies of DNAs was recovered linearly. Replicate standard deviations (SD) of measured ctDNAs spiked in DNA-depleted plasma were equivalent to those spiked in real patient sample but significantly higher than those spiked in “synthetic plasma” or TE buffer. For samples in the same matrix, SDs of AF% dropped as the concentration increased. The drop was especially sharp at levels near the LOD. The LOD was much higher for samples in plasma (AF 1.83%) than for samples in buffer (AF 0.29%). For a positive sample with AF at 0.5%, the buffer-based was detected 100% of the time, whereas the true positive rate for plasma-based was only 69%. Significant difference in ctDNA recovery, replicate SD, and LOD were observed with different extraction methods. The data demonstrated that plasma factors influence ctDNA assay through cfDNA extraction. Vali