Comparison of 7 Methods for Extracting Cell-Free DNA from Serum Samples of Colorectal Cancer Patients

To the Editor:

The presence of cell-free DNA of tumor origin in serum or plasma of cancer patients (1) has triggered numerous studies to explore the diagnostic and prognostic potential of circulating DNA. This DNA, present only in minute concentrations in plasma or serum, is highly fragmented (2), a condition that often leads to substantial loss of DNA of small sizes in the course of DNA isolation. The lack of consensus regarding which extraction method is better for the efficient capture of such DNA might be partially responsible for the large disparities in the literature, which are reflected in reports of total concentrations of plasma or serum DNA alone (3) or DNA integrity measurement as a diagnostic or prognostic tool. The ability to detect mutated v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS) DNA in serum has been reported to vary with the chosen DNA isolation methods (4).

We evaluated in parallel 7 isolation approaches (Table 1⇓ ) by extracting cell-free DNA from 12 pooled sera obtained from 67 colorectal cancer patients and grouped on the basis of TNM tumor staging [tumor extent (T), spread to lymph nodes (N), and metastasis (M)].1 The approaches we evaluated involved diverse strategies for DNA isolation. The experimental set-up involved DNA isolation from a 2-mL aliquot of serum in duplicate followed by DNA quantification by the fluorescent Quant-iT dsDNA HS assay (Invitrogen) and a Taqman real-time PCR (rPCR) technique on the cadherin 1, type 1, E-cadherin (epithelial) (CDH1) gene. To exclude false results in the CDH1 amplification due to various PCR inhibitors present in DNA extracts, the isolated DNA was further quantified by Taqman rPCR on bisulfite-converted DNA because the procedure of bisulfite-conversion of DNA removes many PCR-inhibitory components such as proteins, EDTA, and ethanol (EpiTech bisulfite kit, Qiagen). A fragment devoid of the CpG site of the actin, beta (ACTB) gene was the target for this round of rPCR. All 3 assays for DNA quantification were carried out in duplicate, and the DNA/gene amount in samples was interpolated by reference to corresponding standard curves generated by using 5 to 6 serial dilutions of a DNA standard. Appropriate blanks were included in each run. For rPCR, a positive control sample was run in each plate to control the interplate variation.

The 7 extraction methods showed remarkable differences in the recovery of DNA from serum (Table 1⇑ ). The phenol-chloroform procedure (PCI-glycogen), sodium iodide method (NaI method), and QIAamp DNA blood kit generated significantly higher yields of DNA, assessed by fluorescent measurement, than the other 4 methods (all P < 0.05). Assessed by rPCR targeting on CDH1 (amplicon size: 68 bp), the NaI method was ranked top in the list, and statistical significance (P < 0.05) was achieved in all pairwise comparisons except with the PCI-glycogen approach. The latter, however, outperformed only the guanidine-resin procedure as determined by application of the Dunnett T₃ test for pairwise multiple comparisons (Table 1⇑ ). Because of the insufficient balance of DNA in the other 2 samples, we carried out the second round of rPCR on bisulfite-converted DNA in 10 pooled samples. Among 3 methods evaluated, the NaI method exhibited the most abundant gene copy numbers of ACTB (amplicon size: 115 bp), with the median of ACTB being 7 times higher than that generated by the QIAamp DNA blood kit, although the difference did not reach the statistical significance (Table 1⇑ ).

The higher recovery of DNA obtained with the NaI and PCI glycogen procedures was also revealed on the agarose gel, which showed much stronger DNA signals along each of 2 lanes (electrophoretic image available on request). Interestingly, substantial amounts of small DNA fragments were recovered with these 2 methods, a result that was not achievable with the other 5 protocols (Table 1⇑ ). Furthermore, the size of these small fragments appeared to correspond to that of nucleosomal DNA, i.e., approximately 180–220 bp or its multiples. Because circulating DNA is highly fragmented, any isolation method that favors capture of fragmented DNA from serum or plasma will be useful for a variety of downstream applications in modern clinical and translational research laboratories. These include efficient detection of mutations and capture of DNA methylation markers from serum or plasma.

In comparing the PCI-glycogen approach and the NaI method, we found that the latter was not only superior to the former in terms of DNA quantity, as assessed by 2 rounds of rPCR, but also was simpler, more rapid, and less costly (data available on request).

To summarize, the results of the present study, which involved 7 isolation methods performed via quantification of serum DNA by 3 assays and examination of fragment sizes of DNA isolated by electrophoresis, indicated that the NaI procedure consistently revealed better performance. Therefore, this procedure appears to be a suitable method for cell-free DNA extraction for many downstream applications in cancer research.