Comparison of protocols for extracting circulating DNA and RNA from maternal plasma.

To the Editor: We compared 2 column-based and an automated magnetic bead separation protocol for extracting circulating DNA and RNA from maternal plasma. We obtained peripheral blood from pregnant women at Prince of Wales Hospital, Hong Kong, and King’s College Hospital, London, with informed consent and institutional ethics approval. Thirty second-trimester samples (median gestational age, 17.6 weeks) were divided into 2 aliquots for maternal plasma DNA extraction by either a QIAamp Mini Kit (Qiagen), with 800 L of plasma applied per column and DNA elution into 50 L of deionized water, or a MagNA Pure Total Nucleic Acid Large Volume Isolation Kit on the MagNA Pure LC instrument (Roche Diagnostics) (1 ) with DNA elution into 50 L of elution buffer. -Globin and SRY concentrations were quantified by real-time PCR assays (2 ). With 5 L of plasma DNA per PCR (2 ) for samples from women with male fetuses, SRY amplification was observed in 15 (100%) columnmethod samples and in 10 (67%) automated-protocol samples. The globin DNA concentration was significantly higher (P 0.001, Wilcoxon signed-rank test; SigmaStat Ver. 3.0; SPSS) in the column extractions [median, 456 copies/mL; interquartile range (IQR), 276–744 copies/mL] than the automated protocol [median (IQR), 223 (169–350) copies/mL]. With 10 L of plasma DNA extracted by the automated protocol per PCR (3 ), we detected all but 1 SRY-positive cases. Median (IQR) SRY concentrations for plasma DNA extracted by the column and automated methods were 40 (28–65) and 4 (1–9) copies/mL, respectively; median -globin concentrations were 5257 (793–14 213) and 909 (239–5728) copies/mL, respectively. Maternal plasma SRY concentrations were significantly lower with the automated protocol (P 0.002, Wilcoxon), but -globin concentrations showed no significant difference (P 0.087, Wilcoxon). Maternal plasma was treated with Trizol LS (Invitrogen) to prevent RNA degradation (4 ), and 2 protocols for RNA extraction were compared. Trizol LS (4 mL; Invitrogen) was mixed with plasma (3.2 mL) before storage at 80 °C. Trizol-preserved plasma was thawed and mixed with 0.8 mL of chloroform and centrifuged at 12 000g for 15 min at 4 °C. The upper aqueous layer was transferred into new tubes as 2 aliquots. RNA was extracted from 1 aliquot with 1 mL of the aqueous layer (equivalent to an original plasma volume of 727 L) by the MagNA Pure Kit and instrument and eluted into 50 L of elution buffer. We added 1 volume of 700 mL/L ethanol to 2.2 mL of the aqueous layer (equivalent to an original plasma volume of 1.6 mL) from the other aliquot and applied it to an RNeasy Mini Kit minicolumn (Qiagen) as described previously (5 ). Oncolumn DNase digestion was performed with 40 L of the enzymebuffer mixture from the RNase-Free DNase Set (Qiagen). -Globin DNA, to assess DNA contamination, was detected with realtime PCR (2 ) in 3 of 10 samples in both protocols; concentrations were 98, 135, and 3419 copies/mL for the automated protocol and 30, 54, and 31 copies/mL for the column protocol. For another 10 samples, -globin DNA was not detected in any samples after we used on-column DNase digestion with 80 L of enzymebuffer mixture from the RNase-Free DNase Set (Qiagen) for the columnbased protocol and postextraction DNase treatment with a DNase I reagent set (Invitrogen) for the automated protocol. We used these methods to assess human chorionic gonadotropin -subunit ( hCG) mRNA concentrations in 40 first-trimester maternal plasma samples (median gestational age, 13 weeks) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in 20 of these samples, using intron-spanning reverse transcription-PCR assays as described previously (5, 6). The absence of DNA contamination was further confirmed with the AmpliTaq Gold enzyme (Applied Biosystems) for the hCG assay without the reverse transcription step. Median (IQR) hCG mRNA concentrations for the column-based and automated protocols were 447 (94–1478) and 893 (111–1606) copies/mL, respectively. Median GAPDH mRNA concentrations were 17 260 (9376–35 396) and 47 003 (37 704–102 520) copies/mL for the column-based and automated protocols, respectively. mRNA concentrations for both hCG (P 0.014, Wilcoxon) and GAPDH (P 0.001, Wilcoxon) were significantly higher with the automated protocol. Plasma DNA from the automated extraction was less concentrated than that from column extraction, whereas the contrary was true for plasma RNA. Circulating DNA is nonparticulate, whereas RNA circulates in association with particulate matter (5–7), and different extraction methods may favor the isolation of certain physical forms of plasma nucleic acids. We demonstrated that the observed concentrations of circulating nucleic acids differ depending on the processing and analysis methods (7 ). This study also reveals that careful evaluation of RNA extraction protocols and assay designs is necessary to avoid DNA contamination.