Althaia Xarxa Asistencial de ManresaLa Culla, s/n

08240 Manresa (Barcelona), Spain

*Author for correspondence. Fax 34-93-8743859; e-mail hgmlaboratorio@aehh.org.

Effects of Blood-Processing Protocolson Cell-free DNA Quantification inPlasma

To the Editor:Recently, qualitative analysis of cell-free DNA in blood plasma has at-tracted much interest for the diagno-sis of cancer, fetal gender, Rhesus Dstatus, and inherited disorders (1, 2).Other studies have shown that quan-tification of total plasma DNA mayindicate fetal chromosomal aneu-ploidies and pregnancy-associatedcomplications or the presence/recur-rence of cancer (1, 3, 4). Irrespectiveof the type of study (qualitative orquantitative DNA analysis), it is im-portant that cell-free plasma DNA isnot contaminated with cellular DNAthat interferes with analysis and ac-curate quantification.

To prevent cellular DNA contami-nation, Chiu et al. (5 ) emphasizedthe need of preanalytical standard-ization of blood-processing proto-cols. The authors note that after cen-trifugation of blood at a low speed(800g), the amount of isolated DNAfrom plasma is affected by the pres-ence of cells that remain in theplasma fraction. An additional cen-trifugation step (16 000g) or filteringof the plasma is necessary to produceabsolutely cell-free plasma DNA. Be-cause most laboratories, includingours, centrifuge blood at relativelylow speeds (8001500g) and storeplasma without additional treat-ment, these results may have a seri-ous impact on the usefulness of pre-viously collected plasma samples forretrospective DNA analysis/quanti-fication.

To confirm the results of Chiu et al.(5 ), we collected EDTA blood in7-mL Vacutainer Tubes from 18healthy women with uncomplicatedpregnancies, attending the outpa-

tient clinic of the University MedicalCentre Nijmegen, after 21 weeks ofgestation. Blood was centrifuged at800g for 10 min. The plasma super-

natant was carefully removed to0.5 cm above the buffy coat layer,homogenized, and divided into twoportions: fractions A (400 L) and B

Fig. 1. Albumin concentrations in plasma from pregnant (A) and nonpregnant (B) women.(A), albumin concentrations (CE/mL) isolated from plasma samples obtained from 18 pregnant women,collected in 7-mL EDTA-blood Vacutainer Tubes. The tube volume (7.0 mL) and centrifugal forces (800g and16 000g) before plasma collection and storage at 20 C are shown on the x axes. Fraction A, median, 485CE/mL [95% confidence interval (CI), 371207 CE/mL]; fraction B, median, 422 CE/mL (95% CI, 227-1911CE/mL). A two-sided paired t-test showed no significant difference ( 0.27) in albumin concentrationsbetween fractions A and B. (B), albumin concentrations isolated from plasma samples obtained from 10nonpregnant donors, collected in 10- and 7-mL EDTA-blood Vacutainer Tubes. High-speed centrifugation(16 000g) occurred either before storage (fraction B) or after storage at 20 C (fraction C). Fraction A,median, 24 936 CE/mL (95% CI, 1069123 079 CE/mL); fraction B, median, 480 CE/mL (95% CI, 203897CE/mL); fraction C, median, 453 CE/mL (95% CI, 272961 CE/mL); fraction D, median, 682 CE/mL (95% CI,216-2652 CE/mL). A two-sided paired t-test showed a significant difference ( 0.01) in albuminconcentrations between fractions A and B. Differences between other pairs of fractions (B and C, B and D, Cand D) were not significant ( 0.57, 0.25, and 0.28, respectively).

Clinical Chemistry 49, No. 3, 2003 525

(450 L). Fraction B was centrifugedadditionally at 16 000g for 5 min, and400 L of the supernatant was trans-ferred to a new tube. All fractionswere processed within 3 h after col-lection and stored at 20 C for atleast 1 month. After thawing, theDNA in the fractions was extractedexactly according to the protocol ofChiu et al. (5 ). Both fractions A and Bfrom each pregnant woman wereprocessed in the same batches duringthe isolation and quantification pro-cedures. Copies of the albumin genewere quantified with real-time PCRand transformed to cell-equiva-lents/mL of plasma (CE/mL), as de-scribed previously (3 ).

The quantification results forplasma fractions A and B from 18pregnant women are shown in Fig.1A. Although the median number ofcell-equivalents in fractions B (422CE/mL) correlated well with the re-sults of Chiu et al. (5 ), surprisinglywe detected no significant amountsof additional DNA (from cellular or-igin) in fractions A. Obviously, in oursamples, all cellular DNA had beenremoved from the plasma by centrif-ugation at 800g. After careful reanal-ysis of the protocol of Chiu et al., wecould identify only a single differ-ence with our procedure: we col-lected blood in 7-mL EDTA collec-tion tubes instead of 10-mL tubes.

To demonstrate the possible effectof this difference, we performed anadditional experiment in which wecollected blood from 10 nonpregnantdonors into both 10- and 7-mL tubes.Blood and plasma were processed asdescribed before. In these samples,after DNA isolation, contaminatingcellular DNA was still present in theplasma fraction of the 10-mL collec-tion tubes after centrifugation at 800g(Fig. 1B, fraction A). It was removed

after additional centrifugation at16 000g, confirming the results ofChiu et al. (Fig. 1B, fraction B). More-over, plasma DNA concentrationsderived from the plasma from 7-mLtubes (Fig. 1B, fraction D) were iden-tical to the DNA concentrations incell-free plasma (Fig. 1B, fraction B),corroborating our previous resultsshown in Fig. 1A.

From these results, we concludedthat the amount of blood collected(i.e., the size of the blood collectiontube) affects the presence of contam-inating cells in the plasma fraction.Because these cells were present inplasma both from pregnant women(5 ) and nonpregnant individuals(our results), leukocytes are the mostlikely explanation for the contami-nating cells in the 10-mL blood col-lection tubes.

In parallel, we questioned whetherthese contaminating cells in plasma(Fig. 1B, fraction A) could be re-moved after storage at 20 C. Wehypothesized that as a result of thefreeze-thaw step, contaminating cellslyse, but their nuclei remain intactand can still be eliminated by high-speed centrifugation. Indeed, high-speed centrifugation of fraction Aplasma samples after storage com-pletely eliminated the cellular con-tamination (Fig. 1B, fraction C). Theefficiency of elimination was identi-cal to that of high-speed centrifuga-tion before storage (Fig. 1B, fractionB).

We conclude that the efficiency ofremoval of contaminating cellularDNA from plasma is dependent onmany preanalytical factors, includingcentrifugal force, amount of collectedblood, and pipetting efficacy. Mostimportantly, we have shown that ir-respective of the protocol used forplasma collection, contaminating

cells that remained in the plasma canbe removed after storage (20 C) ofsamples. Therefore, frozen plasmacollections can be used retrospec-tively for DNA analysis when sub-jected to additional centrifugation at16 000g after thawing.

References1. Lo YMD. Fetal DNA in maternal plasma: biology

and diagnostic applications. Clin Chem 2000;46:19036.

2. Anker P, Mulcahy H, Chen XQ, Stroun M. Detec-tion of circulating tumour DNA in the blood(plasma/serum) of cancer patients. Cancer Me-tastasis Rev 1999;18:6573.

3. Swinkels DW, de Kok JB, Hendriks JCM, Wieger-inck E, Zusterzeel PLM, Steegers EAP. Hemoly-sis, elevated liver enzymes, and low plateletcount (HELLP) syndrome as a complication ofpreeclampsia in pregnant women increases theamount of cell-free fetal and maternal DNA inmaternal plasma and serum. Clin Chem 2002;48:6503.

4. Sozzi G, Conte D, Mariani L, Lo Vullo S, Roz L,Lombardo C, et al. Analysis of circulating tumorDNA in plasma at diagnosis and during follow-upof lung cancer patients. Cancer Res 2001;61:46758.

5. Chiu RWK, Poon LLM, Lau TK, Leung TN, WongEMC, Lo YMD. Effects of blood-processing pro-tocols on fetal and total DNA quantification inmaternal plasma. Clin Chem 2001;47:160713.

Dorine W. Swinkels1*

Erwin Wiegerinck1

Eric A.P. Steegers2

Jacques B. de Kok1

1 Department of Clinical ChemistryUniversity Medical Centre Nijmegen6500 HB Nijmegen, The Netherlands

2 Department of Obstetricsand Gynaecology

University Hospital Rotterdam3000 CA Rotterdam, The Netherlands

*Address correspondence to this au-thor at: Department of Clinical Chemis-try/564, University Medical CentreNijmegen, PO Box 9101, 6500 HB Nijme-gen, The Netherlands. Fax 31-243541743;e-mail D.Swinkels@akc.umcn.nl.

526 Letters