Characterization of Plasma Cell-Free DNA Integrity Using Droplet-Based Digital PCR: Toward the Development of Circulating Tumor DNA-Dedicated Assays

Abstract

Background: Cellular-cell free-DNA (ccfDNA) is being explored as a diagnostic and prognostic tool for various diseases including cancer. Beyond the evaluation of the ccfDNA mutational status, its fragmentation has been investigated as a potential cancer biomarker in several studies. However, probably due to a lack of standardized procedures dedicated to preanalytical and analytical processing of plasma samples, contradictory results have been published. Methods: ddPCR assays allowing the detection of KRAS wild-type and mutated sequences (KRAS p.G12V, pG12D, and pG13D) were designed to target different fragments sizes. Once validated on fragmented and non-fragmented DNA extracted from cancer cell lines, these assays were used to investigate the influence of the extraction methods on the non-mutated and mutated ccfDNA integrity reflected by the DNA integrity index (DII). The DII was then analyzed in two prospective cohorts of metastatic colorectal cancer patients (RASANC study n = 34; PLACOL study n = 12) and healthy subjects (n = 49). Results and Discussion: Our results demonstrate that ccfDNA is highly fragmented in mCRC patients compared with healthy individuals. These results strongly suggest that the characterization of ccfDNA integrity hold great promise toward the development of a universal biomarker for the follow-up of mCRC patients. Furthermore, they support the importance of standardization of sample handling and processing in such analysis.

Keywords: DNA integrity index; apoptosis; cancer biomarker; circulating cell-free DNA; circulating tumor DNA; necrosis; picoliter-droplet digital PCR.

Figure 1

Workflow of the integrity study. (A) Development of the DII using the PLACOL cohort (n = 11 samples analyzed). DNA hasbeen quantified using, respectively, 60, 86, 164, and 302 bp amplicons. Several combinations have been tested to determine the optimal DNA integrity index. Finally, the 302/60 bp ratio turned out to be the most clinically relevant. (B) Samples from two prospective cohorts of patients presenting a mCRC were included, AGEO RASANC (n = 67) and PLACOL (n = 12) cohort with a total of 81 metastatic CRC samples. Circulating cell-free DNA integrity of 34 patients (AGEO RASANC cohort) and 12 patients (PLACOL cohort) was analyzed by ddPCR, and ccfDNA size profiling of 33 patients (AGEO RASANC cohort) was analyzed by BIAbooster System. As the preanalytic conditions (blood collection tubes, plasma ccfDNA extraction kits) are different for these two prospective cohorts of patients, the corresponding control groups were included; 8.5 ml blood per tube from 59 healthy subjects are used as negative controls and split into two groups (control groups A and B). Thirty-four healthy blood samples were collected in cell-free DNA BCT from Streck, and circulating DNA has been extracted with Maxwell RSC ccfDNA Plasma Kit from the plasma samples associated, and 25 blood tubes were sampled in EDTA tubes and circulating DNA purified by the QIAmp® Circulating Nucleic Acid Kit (Qiagen) from plasma samples associated. Among control A and B healthy subjects' samples, DNA integrity of 49 plasma was analyzed by droplet-based digital PCR, and ccfDNA size profiling of 10 plasma was analyzed by BIAbooster System.

Figure 2

Design and validation of ddPCR assays for ccfDNA integrity analysis with the use of fragmented and non-fragmented cell line DNA. Droplet-based digital PCR duplex assays amplifying different sizes of KRAS wild-type (WT) and mutant (MT) fragments (60, 86, 184, and 302 bp) were designed and optimized (A). The LOB for the 302-, 164-, 86-, and 60-bp KRAS assays were, respectively, 0, 0, 3, and 1, irrespective of the mutation status. Fragmented (F) and non-fragmented (NF) DNA from LoVo (B1, C1) and SW620 CRC line (B2, C2) were used. DNA integrity index (DII) were analyzed using 302 bp amplicon (corresponding to the long fragments) and 60, 84, and 184 bp amplicons as short fragments (B1, B2). Percentage of KRAS-mutated alleles for F and NF DNA of each cell lines are given in (C1, C2). Nevertheless, PCR efficiency of the primers has not been determined by qPCR for each amplicon size on fragmented and unfragmented DNA.

Figure 3

Characterization of ctDNA by droplet-based digital PCR using assays targeting increasing amplicon sizes. Concentration of plasma ctDNA (A1) and percentage of KRAS-mutated alleles (A2) in mCRC patients and DNA integrity indexes (DII) calculated with the use of 302 bp (B) or 164 bp amplicon (C) (corresponding to the long fragments) by ddPCR. KRAS WT and MT ddPCR duplex assays amplifying different sizes of fragments (60, 86, 164, and 302 bp) were used for measuring ctDNA concentration and percentage of KRAS-mutated alleles (A1,A2) for 11 mCRC patients (one patient excluded since no 302 bp MT fragments were detectable). DII were calculated based on the amplification of sequences with different lengths as described in the section MATERIALS AND METHODS. For DII calculation, 60 bp amplicons were used as the short fragments and either 302 bp (B) or 164 bp amplicon (C) as the long fragments. Mann–Whitney U-test was used for statistical significance analysis. *p < 0.05; **p < 0.01.

Figure 4

Comparison of the influence of ccfDNA extraction methods on the resulting ccfDNA fragment concentrations (A) and DNA integrity (B) by ddPCR. Plasma from 10 healthy individuals were used in this analysis (see workflow in Supplementary Figure 1). Maxwell RSC ccfDNA Plasma Kit (extraction kit 1) (Promega) and QIAmp® Circulating Nucleic Acid Kit (extraction kit 2) (Qiagen) were used for ccfDNA extraction. DII were calculated for ddPCR using 302 bp amplicon as the long fragment and 60 bp amplicon as the short fragment (n = 8, two individuals were excluded as no ccfDNA was available) (B). Wilcoxon test was used for significance analysis. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 5

Comparison of plasma ccfDNA integrity and ccfDNA size profiling between healthy individuals and metastatic CRC patients. As ccfDNA from different patients' groups were prepared with different preanalytic conditions (blood collection tubes and ccfDNA extraction kits), DII comparative analysis was performed for each patient group samples using healthy subject samples collected and treated with the same procedure (see workflow Figure 2). To calculate DII by ddPCR, 302 bp amplicon was used as long fragment and 60 bp amplicon as short fragment (A1: AGEO RASANC cohort and A2: PLACOL cohort). Two samples (one from AGEO RASANC study and one from PLACOL study) have been excluded due to an absence of detection of droplets bearing DNA-mutated fragments of 302 bp or more. To investigate different size distributions by the BIAbooster System, fragments with size ranging from 110 to 239 bp were used as long fragments and fragments with size ranging from 75 to 110 bp as short fragments (B1,B2: AGEO RASANC cohort only). Mann–Whitney U-test was used for significance analysis. *p < 0.05; **p < 0.01; ***p < 0.001; ***p < 0.0005. Fragmentation of KRAS WT and MT alleles within ccfDNA for mCRC patients is shown in (A1,A2). Mutant-allele frequency determined with the DII assay was compared with the one determined previously using other assays for the RASANC samples [BPER NGS, Bachet et al. Ann Oncol. (25)] and PLACOL samples [ddPCR targeting 86 bp amplicon, Garlan et al. CCR. (26)]. Linear regression model is given in Supplementary Figure 6.