Comparison of Two Digital PCR Systems for the Analysis of Liquid Biopsy Samples of Patients Affected by Lung and Colorectal Cancer

Comparison of Two Digital PCR Systems for the Analysis of Liquid Biopsy Samples of Patients Affected by Lung and Colorectal Cancer

Abstract
Background and aims Non-invasive techniques and highly sensitive technologies are available for tumor molecular characterization at the diagnosis or during therapy, both for lung and for colon cancer patients. Molecular evaluation using liquid biopsy based on circulating free DNA (cfDNA) offers the opportunity to track the genomic evolution of tumors, and is often used to assess tumor molecular profile, i.e., EGFR or RAS alterations. Our study aimed to compare the results of two different digital PCR (dPCR) platforms for detecting mutations in cfDNA.
Methods Plasma samples from patients with lung and colorectal cancer collected as per routinary procedures have been tested. CfDNA was extracted from 3 ml of plasma using the QIAamp Circulating Nucleic Acid Kit (Qiagen, Valencia, CA). Samples were screened on the droplet digital PCR (BioRad®, Hercules, CA) and on the dPCR Qiacuity (Qiagen, Valencia, CA) using the ddPCR Mutation Assay on both instruments (BioRad®, Hercules, CA).
Results A total of 42 samples were analyzed, of which 20 were Non-Small Cell Lung Cancer (NSCLC) samples of patients’ carrying an EGFR or a KRAS mutation on tissue at diagnosis, and 22 were samples of colorectal cancer (CRC) patients, of which 10 patients were carrier of a RAS mutation on tissue at diagnosis and 12 were RAS wild-type. EGFR mutation detection was 58.8% (10/17) for ddPCR and 100% (17/17) for dPCR, with moderate agreement between the two techniques (κ= 0.54; 95% CI, 0.37-0.71). The detection rate for RAS and BRAF mutations was 68% (17/25) for ddPCR and 88% (22/25) for dPCR, with moderate agreement (κ= 0.37; 95% CI, 0.09-0.65).
Conclusions The present comparison, even if with small numbers, showed moderate agreement between dPCR and ddPCR. The discordancy observed may be related to several factors, such as the threshold needed to define a positive signal or sampling effect. These factors potentially explain many of the differences in the cfDNA data derived from different platforms.