A novel method for extracting circulating cell‐free DNA from whole blood samples and its utility in the non‐invasive prenatal test

Abstract Objective We verified a magnetic bead‐based, simple, and fast method for circulating cell‐free DNA (cfDNA) extraction from whole blood samples(CEWB) and characterised its utility in non‐invasive prenatal testing (NIPT). Method We extracted cfDNA from both plasma and whole blood of the patients using CEWB and compared it to that extracted using a Qiagen extraction kit; droplet digital polymerase chain reaction test was used to calculate the fragment size bias. In all, 304 samples were used for NIPT. Results The CEWB group (mean ± standard deviation [SD]: 4.34 ± 0.41 ng/ml plasma) reported less DNA weight yield than the Qiagen group (4.90 ± 0.50 ng/ml plasma). There was no significant difference between the CEWB group and the Qiagen group in the gene fragments (136 bp: p = 0.064 and 420 bp: p = 0.534). In a parallel cohort study to characterise the utility of the CEWB method in NIPT, the treatment group extracted by CEWB showed a sensitivity of 100%, a specificity of 99.65%, and a positive predictive value of 95%. Conclusions This study demonstrated that CEWB achieves an acceptable yield of DNA without contamination from genomic DNA. Subsequent clinical experiments in a parallel cohort indicated its utility for NIPT.


| INTRODUCTION
Circulating cell-free DNA (cfDNA) was discovered in human blood in 1948 1 ; foetal DNA in maternal plasma and serum was discovered in 1997. 2 Since then, cfDNA has become an important method of liquid biopsy, and it is used as a non-invasive screening tool for many diseases, especially solid tumours, and foetal genetic abnormalities. [3][4][5][6][7] Non-invasive prenatal testing (NIPT) uses next-generation sequencing (NGS) technologies using cell-free foetal DNA from maternal plasma to detect certain genetic conditions such as trisomy (T) 21, T18, and T13 during pregnancy. It has a sensitivity and specificity (approximately 99%). Since 2011, massively parallel screening for foetal aneuploidies has become available in more than 60 countries. Increasing use of the cfDNA-based NIPT has created unprecedented challenges in automation in the biotechnology industry. 5 Current cfDNA extraction methods 8,9 mainly include: (I) column-based methods, such as the QIAamp circulating nucleic acid kit [10][11][12][13][14] ; (II) magnetic bead-based methods, such as the NextPrep-Mag™ cfDNA isolation kit 13 ; (III) polymer-mediated enrichment, such as the PME free-circulating DNA extraction kit. 13 However, the preprocessing separation of serum or plasma from whole blood (WB) samples limits the performance of the above cfDNA extraction methods. Pre-processing is a time-consuming method that requires high-speed centrifugation at low or room temperature for 15-30 min, while avoiding contamination of the blood cells during pipetting operations. 2,9,14 The main challenge faced by many engineers is the automation of the separation of serum or plasma costeffectively. This includes intelligent control of the centrifuge for automatic tube insertion, removal, identification of plasma and buffy coat, shorter (<30 min) separation operations, reduced cost, and increased throughput stability. Isolating cfDNA from peripheral blood instead of plasma or serum is not often attempted. Pandoh et al.
reported a high-throughput protocol for isolating tumour cfDNA, 15 but this is complex and requires a long incubation time (>1 h). Here, we introduce a magnetic bead-based, simple, and fast cfDNA extraction method from WB samples, which we have termed CEWB; additionally, we estimate its effect on DNA yield and fragment size bias and characterise its utility for NIPT. This study is one of the first investigations on cfDNA extraction from WB samples. We used Cell storage solutions, which have trace amounts of formaldehyde released from imidazolidinyl urea which coagulates the proteins, immobilises the cells, and prevents them from rupture and genomic DNA (gDNA) release. The cfDNA is then directly bound to blood outside the blood cells by electrostatic force using amino magnetic beads, without the need for a high salt environment as carboxylated magnetic beads and other substances (red blood cells, white blood cells, etc.) are removed by wash buffer. Subsequently, cfDNA is eluted off by the elution buffer in the kit and purified, and fragments are selected using carboxyl magnetic beads to remove impurities and increase the proportion of target cfDNA (e.g., foetal or tumour cfDNA).

| Participants
A total of 366 pregnant women with a gestation period between 12 and 24 weeks were enrolled from 2020 to 2021. We confirmed 20 NIPT-positive samples with foetal T21, T18, or T13 by karyotyping and/or chromosomal microarray analysis (CMA). A total of 282 NIPTnegative samples were followed up 3 months after delivery and were found to be true negative. All participants provided written informed consent before blood collection, and the study was approved by the Institutional Review Board (IRB) of the BGI (NO. BGI-IRB 21008). All WB samples collected in EDTA tubes were used for plasma isolation or cfDNA extraction using CEWB within 8 h.
We used the following exclusion criteria: a gestation period of <12 + 0 weeks; one spouse with a chromosomal abnormality; treatment for an abnormality, such as stem cell therapy (within 1 year) or exogenous DNA treatment (within 4 weeks); a foetal ultrasound indicating structural abnormalities; a family history of genetic disorders or a high risk of genetic disorders in the foetus; a combination of malignant tumours during pregnancy (except for benign uterine myoma); or a multiple pregnancy.

| CfDNA extraction from WB samples
The cfDNA was extracted from WB samples by modifying the magnetic bead extraction kit-Whole Blood Cell-Free DNA Extraction Kit (cat #CFDNAWBB50, Jiashan Zhijian Tech Co., Ltd., Zhejiang, China).
There are two types of magnetic beads in the kit. Type 1 beads are amino magnetic beads which have a superparamagnetic silica matrix and an active amino group. These beads can bind to DNA based on electrostatic force without a high salt environment and can bind to free DNA in blood directly when added to blood. Type 2 magnetic beads are carboxyl magnetic beads which have a superparamagnetic silica matrix and an active carboxyl group. When the outer surface of the magnetic beads with carboxyl functional group is modified in the purification buffer system containing polyethylene glycol (PEG), high salt ions, etc., the DNA is adsorbed by forming an ionic DNA-salt 1174ions-carboxyl bridge. This binding is reversible; the ionic bridge is dissolved in TE (Tris-EDTA) buffer without PEG and salt ions to obtain purified DNA. This is the type commonly used for DNA purification and can also be used for fragment selection.
The two types of magnetic beads were prepared and incubated for 10 min at room temperature. Type 1 magnetic beads (30 μl) were mixed with 100 μl wash buffer from the kit, the supernatant was discarded, and then 30 μl wash buffer was re-added; 0.1� volume of Cell storage solution (the main ingredient is imidazolidinyl urea) from the kit was added to 1 ml WB before tapping and mixing (not required for plasma samples). The blood sample was mixed with the type 1 magnetic beads, incubated at room temperature for 5 min, centrifuged at low speed for 5 s, placed on a magnet for 2 min, then the blood was discarded. The sample was then mixed with 500 μl wash buffer, centrifuged at low speed for 5 s, placed on a magnet for 30 s, and the supernatant was discarded. This procedure was repeated twice. We then mixed 40 μl of elution buffer for 4 min and aspirated 40 μl of the supernatant. The above supernatant was mixed with 50 μl type 2 magnetic beads, incubated at room temperature for 5 min, centrifuged at low speed for 5 s, placed on a magnet for 5 min, supernatant was discarded, and for cfDNA fragment size enrichment and selection in the NIPT experiments, the supernatant was mixed with 20 μl type 2 magnetic beads, incubated at room temperature for 5 min, centrifuged at low speed for 5 s, placed on a magnet for 5 min, and the supernatant was aspirated into a new 1.5 ml centrifuge tube, mixed with 30 μl type 2 magnetic beads, incubated at room temperature for 5 min, centrifuged at low speed for 5 s, placed on a magnet for 5 min, and the supernatant discarded. The type 2 magnetic beads were cleaned twice with 500 μl of 75% ethanol for 5 min after the ethanol evaporated. The ultrapure water or TE buffer (≥20 μl) was mixed for 4 min, and the supernatant was aspirated into a new 1.5 ml centrifuge tube. Figure 1 illustrates the working principle of the cfDNA extraction method for WB samples.

| CfDNA extraction from plasma samples
We used two cfDNA extraction methods for plasma samples. Plasma samples were first obtained by high-speed centrifugation of 1 ml WB samples, 16 and then cfDNA was extracted using either a columnbased method on a QIAamp circulating nucleic acid kit (cat #55114, Qiagen, Hilden, Germany) or a magnetic bead-based method on a MGIEasy Circulating DNA Isolation Kit (MGI Tech Co., Ltd, Shenzhen, China). Both kits were used following the manufacturer's instructions.

| Calculation of the DNA weight yield
cfDNA concentration was measured using a Qubit dsDNA HS Assay Kit (cat# Q32854, Thermo Fisher Scientific Inc., Waltham, MA, USA), and DNA weight was calculated as concentration � volume.

| Preparation of simulated samples containing tumour reference material
gDNA was extracted from NCI-H1975, a human lung adenocarcinoma cell line with a 77.7% mutation rate for EGFR-T790M (ATCC, Maryland, USA). gDNA extracted from immortalised cell lines (originally donated by healthy individuals), was ultrasonically fragmented using a Coviras LE220 (Agilent, Santa Clara, CA, USA) under the following parameters: 40 cycles at peak power: 500, duty factor: 21, and burst: 500; and then measured using an Agilent 2100 bioanalyzer. Next, the fragmented tumour DNAs were proportionally prepared and mixed into two groups: a 5% EGFR-T790M group and a 0.5% EGFR-T790M group. Both groups were validated with droplet digital polymerase chain reaction (ddPCR).

| The ddPCR assays
The primers and probes for the short β-actin gene fragment (136 bp) and longer β-actin fragment (420 bp) have been previously recorded in the literature. 16 The following primers were used: For EGFR- The PCR conditions were as follows: pre-denaturation at 98°C for  produced. A binary hypothesis strategy was developed for detection, using the bioinformatics pipelines for T21, T18, and T13. 17,18 A Tscore >4 for Chr21/Chr18/Chr13 was classified as trisomy.

| Statistical analyses
Statistical analyses were performed using Excel 2010 and R software (version 3.6.0). The boxplot, violin plot, and dot-plot diagrams were implemented by the 'ggplot2' R library. The p-value was calculated using the student's t-test between two groups. p < 0.05 was considered statistically significant. The 95% confidence interval (CI) was calculated using the Wilson CI. Results are presented as mean � standard deviation (SD) and coefficient of variation (CV).

| DNA weight yield and fragment size bias
The WB samples from 22 pregnant women were divided equally into two groups. CEWB and the Qiagen kit were used to extract DNA from 1 ml WB and from plasma isolated from 1 ml WB, respectively. Subsequently, the samples from both groups were examined with Qubit to calculate the DNA weight yield, and ddPCR was used to calculate the copies per mL of WB for the 136 bp and 420 bp fragment lengths.
As shown in Figure 2 and Table S1, the Qiagen-plasma group had higher yield of DNA (4.90 � 0.50 ng/ml; 10.17%) than the CEWBplasma group (4.34 � 0.41 ng/ml; 9.55%), although there was similar repeatability and stability. There was a significant difference between the two methods (p = 0.015).
The copy number of the 136 bp fragment obtained by CEWB was significantly different from that obtained by Qiagen (p = 0.064), while the copy numbers of the 420 bp fragment were not significantly different between the two extraction methods (p = 0.534; Figure 3); therefore, indicating that the differences in DNA extracted by two kits were mainly due to the differences in the amount of short fragments (cfDNA) obtained.

| DNA quantification and purity analysis from the tumour reference material
The DNA of 5% or 0.5% theoretical ratio tumour reference material was mixed with the normal WB samples as simulated samples, divided equally into two portions, extracted using the CEWB method from 1 ml WB and the magnetic bead-based kits from plasma isolated from 1 ml WB as controls, respectively. Subsequently, we measured the mutation ratio with ddPCR. Figure 4, Table S4, and Table S5 show an increase in the detected mutation ratio from the tumour reference material when compared to the theoretical ratio as well as a decrease and instability in the detected mutation ratio from the simulated samples when compared to the detected mutation ratio only from the tumour reference material. There are many similarities between DNA extracted by the CEWB method and controls. What stands out is the promising purity of DNA extracted by the CEWB method.

| A parallel cohort study to characterise the utility of CEWB for NIPT
Of the 304 samples, 302 were successfully used to construct a library (99.34%). The remaining two samples were discarded. Figure S2 shows the DNA peaks plot of sequencing library PCR product. It shows that we have obtained the cfDNA fragment correctly and that the sequencing library was constructed without gDNA contamination.

| DISCUSSION
There have been continual demands for automation of experimental operations in molecular laboratories, especially in clinical testing centres that use mature testing technology. For laboratorydeveloped tests using NGS, automation in library construction, sequencing, bioinformatics, and report management has been gradually realised. [19][20][21] However, there has been a continuous demand for automation in sample processing. Automated sample processing systems are complicated, expensive, and not suitable for application in small and medium-throughput laboratories. Therefore, we proposed CEWB without pre-processing separation of serum or plasma from WB samples and aimed to create a cost-effective, simple, small, all-in-one machine that combines CEWB with downstream systems.
The current automated cfDNA extraction process takes approxi-  The DNA extracted using CEWB is inferior in quality to that extracted using the Qiagen kit, probably due to the intrinsic nature of these two extraction methods, CEWB being magnetic bead-based and the Qiagen kit being column-based. This is consistent with previous studies that have demonstrated that the column-based method produces more DNA weight yield than the magnetic bead-based method. 22 This study did not detect any evidence of contamination from gDNA ( Figure S1 27,28 We found that the use of supernatant blood instead of WB extracted by CEWB can effectively increase the cfDNA yield (Table S7). This could be explained by blood cells gradually settling down, retaining only plasma and a small number of blood cells in the supernatant if the peripheral blood is left standing for hours.
Our results have many practical applications. In principle, CEWB can be applied to cfDNA research in nucleic acid molecular diagnostics. In addition to blood, tumour cfDNA has been found in spinal fluid, urine, saliva, and faeces. [29][30][31][32] The amino magnetic beads of CEWB could directly bind and separate cfDNA from liquids and therefore, it can be used for spinal fluid, urine, saliva, and faecal samples where cfDNA may be directly extracted without the time-consuming high-speed centrifugation step. This technology also has the potential to be used to automate sample processing systems combined with sample scanning and coding function modules. The ideal end result is an automatic all-in-one machine that combines all processes involved in cfDNA extraction from peripheral blood.

| CONCLUSION
We proposed a cfDNA extraction method from WB samples, termed as CEWB, and demonstrated that it achieves 4.34 � 0.41 ng/ml plasma DNA yield with promising purity without contamination from T A B L E 1 The performance of the clinical test in NIPT a with different extraction methods

Method
Control (extracted by magnetic bead-based method from plasma) Treatment (extracted by CEWB b ) gDNA. Subsequent clinical experiments indicated its utility for NIPT. This approach can be applied to non-invasive liquid biopsy in nucleic acid molecular diagnostics and in automated sample processing systems to eventually realise the goal of an automatic all-in-one machine in a cost-effective, simple, and compact package.

ACKNOWLEDGEMENT
We acknowledge the volunteers who participated in our study.