CLIN.CHEM.33/2, 214-220 (1987)

214 CLINICALCHEMISTRY,Vol.33, No.2, 1987

Lipoperoxidesin Plasma as Measured by Liquid-ChromatographicSeparationofMalondialdehyde-ThiobarbituricAcid AdductSteven H.Y. Weng,’ Joseph A. KnIght,”3 SIdney M. Hopfe.Y’3 Odette Zaharla,1Charles N. Leach Jr.,4and F. WillIam Sunderman,Jr.”3’3

This assay of plasma lipoperoxidesinvolves hydrolysisindilute H3P04 at 100 #{176}C;complexationof malondialdehyde(MDA), a hydrolysis product,with thiobarbituricacid (TBA);methanol precipitation of plasma proteins;fractionationof theprotein-free extract on a C15column; and spectrophotometncquantification of the MDA-TBA adduct at 532 nm. Thedetection limit was 0.15 &mol of MDA per liter of plasma.Run-to-run precision (CV) averaged 8 to 13%. Analyticalrecovery of MDA after addition of tetraethoxypropane stan-dards to 21 specimens of human or rat plasma averaged98% (SD 7%). Upoperoxideconcentrations(as MDA) aver-aged 0.60 (SD 0.13) .&mol/L in plasma specimens from 41healthy persons and 1.4 (SD 0.3) 1imol/L in plasma speci-mens from 12 control rats. Mean lipoperoxide concentrationswere 1.5 to 2.3 times as great in plasma sampled from ratsone to three days after subcutaneous administration of NiCI2at dosages (250 to 750 &mol per kilogram body wt) previous-ly shown to induce lipid peroxidation in lung, liver, andkidney.

AddItIonalKeyphrases: toxicology nickel rats oxygenfree-radicals referenceinteival

Lipid peroxidation, a general mechanism whereby oxygenfree-radicals induce tissue damage (1-3), has been implicat-ed in diverse pathological conditions, including atheroscle-rosis, aging, rheumatic diseases, cardiac and cerebral isch-emia, respiratory distress syndrome, various liver disorders,irradiation, thermal injury, and toxicity induced by certainmetals, solvents, pesticides, and drugs (4-12). Clinical-chemical tests to assess lipid peroxidation include (a) flue-rometry of lipofuscin-like substances in serum (13); (b)spectrophotometry of coiijugated dienes in lipid extracts ofplasma (14,15); (c) gas chromatography of ethane, pentane,and other alkanes in exhaled breath (16, 17); and (d)hydrolysis of plasma lipoperoxides to form malondialdehyde(MDA; CHOCH2CHO), which reacts with thiobarbituricacid (TBA) to yield a red MDA-TBA adduct, which ismeasured by spectrophotometry or fluorometry, usuallyafter extraction into butanol (7, 18-22).

Bird et al. (23) and Yu et a!. (24) described “high-performance” liquid chromatographic (HPLC) techniquesfor determining MDA in which the MDA-TBA adduct isseparated from interfering substances by chromatographyon a column of octadecyl silica gel. Here we have adapted

Departments of’ Laboratory Medicine and2 Pharmacology, Urn-versity of Connecticut Medical School, Farmington, CT 06032.

30n sabbatical leave from the Department of Pathology, Urnver-sity of Utah Medical School, Salt Lake City, UT 84132.

4Departmenta of Medicine, New Britain General Hospital, NewBritain, CT 06050, and University of Connecticut Medical School,Farmington, CT 06032.

5To whom correspondence should be addressed.#{176}Nonstandardabbreviations: MDA, malondialdehyde; TBA,

thiobarbituric acid; PEP, tetraethoxypropane; HPLC, “high-per-formance” liquid chromatography.

Received October 6, 1986; accepted November 7, 1986.

these HPLC techniques to measurements of plasma lipoper-oxides, after acid hydrolysis of lipoperoxides and formationof the MDA-TBA adduct under the reaction conditions usedby Mthara et al. (25) and Sunderman et al. (26) for analysisof MDA precursors in tissue homogenates. The new HPLCprocedure is more specific, sensitive, and reproducible thanearlier techniques. It provides a practical method for assess-ing lipid peroxidation in both human subjects and experi-mental animals.

Principle

Plasma lipoperoxides are hydrolyzed by boiling in dilutephosphoric acid. MDA, one of the hydrolysis products, isreacted with TBA to form MDA(TBA)2 adduct (24, 27).Plasma proteins are precipitated with methanol and re-moved from the reaction mixture by centrifugation. Theprotein-free extract is fractionated by HPLC on a column ofoctadecyl silica gel, to separate the MDA-TBA adduct frominterfering chromogens. The MDA-TBA adduct is elutedfrom the column with methanol/phosphate buffer and quan-tified spectrophotometrically at 532 nm. Plasma lipoperox-ide concentrations are computed by reference to a calibra-tion curve prepared by assays of tetraethoxypropane (PEP),which undergoes hydrolysis to liberate stoichiometricamounts of MDA (28).

MaterIals and Methods

Apparatus

Polypropylene test tubes, 13-mL capacity, with screw caps(no. 60-541-685; Walter Sarstedt, Inc., Princeton, NJ).

Polypropylene microtubes, 2-mL capacity, with screw capsthat contain 0-ring seals (no. 72-693; Walter Sarstedt, Inc.).

High-speed microcentrifuge, to accommodate 2-mL micro-tubes.

HPW apparatus. Solvent-delivery module and injectionsystem (Model 112, Beckman Instruments, Inc., Fullerton,CA) with 50-pL sample loop. A guard column, 3.9 mm (i.d.)x 2.3 cm, packed with Bondapak Corasil C18 (37- to 50-pmparticle diameter; Waters-Millipore Corp., Milford, MA). A3.9mm (i.d.) x 30 cm chromatographic column packed with1zBondapak C18 (10-pm particle diameter; Waters-Mffli-pore). The flow rate of the mobile phase is 2 mL/mun.

Spectrophotometric monitor, variable wavelength (Model3000; LDC/Milton Roy Co., Riviera Beach, FL), with 14-pLflow-cell volume, 1-cm optical pathlength, set at 532 nm anda sensitivity of 0.02 absorbance unit (A). Although thevoltage output is nominally 10 mV, we connect the spec-trophotometric monitor to a strip-chart recorder with 1 mVfull-scale range (Model 9176: Varian Corp., Palo Alto, CA),so that the actual recorder sensitivity is 0.002 A full-scale.Recorder chart-speed is 1 cm/mm.

Reagents

Phosphoric acid solution, 0.44 moliL. Dilute 10 mL of“ultra-pure” H3P04 reagent (relative density 1.69, 44.0mo]/L, 850 g/L; J. T. Baker Co., Phillipsburg, NJ) to 1 L with

CLINICALCHEMISTRY,Vol.33, No.2, 1987 215

water (distilled water is used throughout).TBA solution, 42 mmol/L. Dissolve 0.6 g of 4,6-dihydroxy-

2-thiopyrimidine (no. T-5500, molecular mass 144.2 Da;Sigma Chemical Co., St. Louis, MO) in approximately 80mL of water, with stirring on a hot-plate (50-55 #{176}C).Coolthe solution to 25#{176}Cand dilute to 100 mL with water.Stored at room temperature, this reagent is stable for twoweeks.

Ethanol, 400 mL/L solution. Dilute 420 mL of 95% ethanoltolL with water.

TEP standard solutions. In a 25-mL volumetric flask,dilute 50 pL of 1,1,3,3-tetraethoxypropane reagent (anhy-drous, relative density 0.91, molecular mass 220.3 Da,purity 98%, no. T-9889; Sigma Chemical Co.) to the markwith 400 mLIL ethanol solution. Prepare freshly each monthand store at 4#{176}C.For an intermediate standard, pipet 0.5mL of this PEP stock standard solution into a 100-mLvolumetric flask and dilute to the mark with 400 mIJLethanol solution. Prepare freshly each fortnight and store at4#{176}C.To prepare PEP working standard solutions (0.61,1.22,2.43, and 4.86 pmol/L), pipet into four 25-mL volumetricflasks 0.375, 0.75, 1.5, and 3.0 mL, respectively, of PEPintermediate standard solution and dilute the contents tothe mark with water. Prepare these weekly and store at4#{176}C.

Methanol-NaOH solution. Pipet 4.5 mL of 1 molfL NaOHsolution into a 50-mL volumetric flask and dilute to themark with methanol (“HPLC” grade; J. P. Baker Co.).Stored in a polyethylene bottle at 4#{176}C,this is stableindefinitely.

Potassium phosphate buffer, 50 mmol/L, pH 6.8. Dissolve13.6 g of anhydrous KH2PO4 in approximately 1.6 L ofdistilled water and titrate to pH 6.8 with 1 molIL KOHsolution, monitoring constantly with a pH meter. Dilute to 2L with water and ifiter through a polycarbonate ifiter (0.45-pm pore diameter; Waters-Millipore).

Mobile phase. Prepare the HPLC mobile phase daily, justbefore use. Mix 400 mL of HPLC-grade methanol and 600mnL of potassium phosphate buffer solution in a side-armsuction flask; then de-gas by reducing the pressure for 20mm, with vigorous magnetic mixing.

Methanol-water mixtures. Into two side-arm suctionflasks, transfer 400 and 800 mL of HPLC-grade methanoland 600 and 200 mL, respectively, of distilled water that hasbeen ifitered through a polycarbonate ifiter. Before use forHPLC, de-gas these solutions under reduced pressure for 20min, with vigorous magnetic stirring.

Ad4itional materials. We also used nickel chloride(NiCl2 . 3H20; Alfa Inorganics Division, Ventron Corp.,Beverly, MA), quality-control sera of human origin (“Deci-sion Level One Control Serum” from Beckman Instruments,Inc., Brea, CA; “RIA Reference Serum” from Wien Labora-tories, Inc., Succasunna, NJ), six standard reference materi-als (glucose, cholesterol, bilirubin, urea, uric acid, andcreatinine, all from the U.S. National Bureau of Standards,Gaithersburg, MD), and 17 samples of common drugs,obtained from the respective pharmaceutical manufactur-ers. Erythrocyte hemolysate was prepared for interferencetests by mixing 3.5 mL of heparinized whole blood from ahealthy adult with 3 mL of a leukocyte separation medium(“Mono-Poly Resolving Medium,” cat. no. 16-980-49; FlowLaboratories, Inc., McLean, VA), which contains a mixtureof polysaccharide (Ficol-400) and sodium metrizoate (Hy-paque-85) having a density of 1.114 kg/L. After centrifugingthese samples at 300 x g for 30 miii in a swinging-bucket

rotor, we removed the supernatant layers of plasma, leuko-cytes, and platelets. The sedimented erythrocytes werewashed with isotonic NaC1 solution, hemolyzed by additionof five volumes of distilled water, and centrifuged at 2000 xg for 20 mm to sediment erythrocyte envelopes. The concen-tration of hemoglobin in the hemolysate was measured byspectrophotometry (29).

Procedures

Blood collection from humans. Collect the blood samplesby venipuncture into 7-mL evacuated tubes containingEDTA solution as anticoagulant (10.5 mg of K2EDTA in 70pL of water). Centrifuge the blood (900 x g, 20 miii) andremove the supernatant plasma, being careful to avoidcontamination with platelets. Store the plasma at 4#{176}Cfor nolonger than 24 h before analysis.

Blood collection from rats. Induce vasodilation of the tailby placing the rat in a warmed cage (approximately 35 #{176}C)for 10 miii. After transferring the rat to a plastic restrainingcone warmed with a heating pad, excise 5mm from the tip ofthe tail with a scalpel; discard the first drops of blood. Whenthe blood flows freely, collect 0.5 mL in a 0.5-mL microtubecontaining 0.75mg of dry K2EDTA and two plastic beads (2-mm diameter) to facilitate mixing. Centrifuge the microtubeas described for human blood, and pipet duplicate 50-pLaliquots of plasma directly into the 13-mL polypropylenetest tubes for lipoperoxide assay. Discard any samples withvisible hemolysis.

Lipoperoxide hydrolysis and TBA reaction. In each analyt-ical run, assay-in duplicate-reagent blanks, PEP workingstandard solutions, plasma specimens, and a quality-controlspecimen. Pipet 0.75 mL of the phosphoric acid solution intoeach 13-mL polypropylene test tube. Using a piston-dis-placement pipettor, pipet 50 pL of PEP standards, plasmaspecimens, and quality-control specimen into the respectivetubes and vortex-mix. Add 0.25 mL of TBA solution to eachtube. Add distilled water (0.50 mL for reagent blanks, 0.45mL for PEP standards, plasma, samples, and quality-controlsamples) to adjust the final volume to 1.5 mL. Cap the tubestightly and place them in a boiling water bath (100#{176}C)for60 miii, then in an ice-water bath (0#{176}C)until the HPLCanalyses are performed.

The MDA-TBA adduct is stable at acid pH but dissociatesslowly at neutral or alkaline PH; thus the boiled samplesmust be neutralized individually within 10 mm beforeinjection onto the HPLC column, as follows. Pipet 0.5 mL ofeach boiled sample into a polypropylene microtube thatcontains 0.5 mL of methanol-NaOH solution. Cap the tubeand vortex-mix, then centrifuge the plasma and quality-control samples (9500 x g, 5 miii) to sediment the precipitat-ed plasma proteins.

HPLC assay. Equilibrate the HPLC apparatus by pump-ing mobile phase (methanol-phosphate buffer) at 2 mL’munfor at least 30 mm, until the recorder baseline is stable.Inject sequentially 50 pL of each blank, PEP standard,quality control, and protein-free plasma extract, recordingthe absorbance of each effluent at 532 nm for 10 min.Measure the peak height of the MDA-TBA adduct, whichhas an average retention time of 4.2 mm, and prepare acalibration curve by plotting the peak heights of the blankand PEP standard samples. Determine the concentrations ofthe plasma lipoperoxides as MDA (pmolfL) from the calibra-tion curve.

Column regeneration. At the end of each HPLC run,remove the guard column and flush the injection system

216 CLINICALCHEMISTRY, Vol. 33, No. 2, 1987

sequentially with 10 mL of distilled water, 10 mL ofmethanol, 10 mL of nitric acid solution (5 molJL), 10 mL ofdistilled water, and 10 mL of methanol/water (40/60 by vol).Reconnect the HPLC column, without the guard column,and regenerate the column by sequentially pumping metha-nol-water (40/60, 80/20, and 40/60 by vol) for 30 mm each,at 2 mL/mun. Repack the guard column daily, after cleaningthe fits, discs, and fittings by boiling them in 2 moIJL nitricacid solution for 30 mm and then in distilled water for 30miii. Before the next run, re-assemble the HPLC apparatusand inject 10 mL of methanol-water (80/20 by vol) to rinsethe injection port and sample loop, followed by 10 mL ofmobile phase.

With these precautions, the HPLC column should yieldchromatograms with undiminished sensitivity and resolu-tion for more than one month of steady use (>30 samplesper day).

Cleaning plasticware. To clean the 13-mL polypropylenetubes, we treat them with TBA-H3P04 solution, preparedby mixing 300 mL of the dilute phosphoric acid, 100 mL ofthe TBA solution, and 200 mL of distilled water. The tightlycapped tubes are filled, placed in a boiling water bath for 1h, then rinsed several times with distilled water, filled withwater, capped, and returned to the boiling water bath for 1h. The tubes are drained, rinsed, and dried by evaporation.

With these precautions, the peak height for MDA-TBA inchromatograms of reagent blanks should be <4 mm.

Subjects

Our human subjects were 41 asymptomatic, adult resi-dents of central Connecticut (20 men, 21 women, ages 21 to56 years, medical school faculty, staff, and students), whowere not receiving medications (including oral contracep-tives) and who gave no history of heart disease, cancer,diabetes, or other serious illnesses. The subjects were notfasted. Blood was sampled between 08:00 and 11:00 h.

The experimental snimg1s were 59 male Fischer-344 rats(body weight 175-225 g; Charles River Breeding Labora-tories, Inc., North Wilmington, MA), housed in stainless-steel mesh cages and fed Purina rat chow and water adlibitum. The rats were fasted for 17 h before blood collection.Some of the rats received a subcutaneous injection of NiCl2solution (250, 500, or 750 mmol per kilogram body wt) 24,48, or 72 h before the blood was sampled, between 08:00 and09:00 h. Control rats received a 0.2-mL subcutaneous injec-tion of 140 mmol/L NaC1 vehicle solution 24, 48, or 72 hbefore blood collection.

Statistical computations (means, standard deviations, co-efficients of variation, detection limits, and probability esti-mations by the Mann-Whitney test) were performed accord-ing to Sachs (30).

Resufts

HPLC analysis. Optimal conditions for separation of theMDA-TBA adduct were established by evaluating the com-ponents and sources of variation in the HPLC techniques ofBird et al. (23) and Yu et al. (24), including (a) trials ofseveral commercial columns packed with octadecyl silicagel, (b) evaluations of a series of phosphate buffer-methanoland phosphate buffer-acetonitrile mixtures as the mobilephase, in lieu of the water-methanol mixtures used in theoriginal techniques, and (c) varying the phosphate bufferconcentration (from 50 to 200 mmoIIL) and the proportion ofphosphate buffer to organic solvent (from 80/20 to 60/40 byvol) in the mobile phase in order to achieve a narrow,

symmetrical MDA-TBA peak in the chromatogram. Twosatisfactory sets of operating conditions for HPLC fraction-ations were identified: (a) Bondapak C15 (10-pm particlediameter) as the column packing material and a 60/40 (byvol) mixture of phosphate buffer (50 mmol/L) and methanolas the mobile phase, with a solvent flow-rate of 2 mL/min;and (b) Chromegabond MC-C18 (5-pm particle diameter;E.S. Industries, Marlton, NJ) as the column packing materi-al and a 80/20 (by vol) mixture of phosphate buffer (100mmollL) and acetonitrile as the mobile phase, with a solventflow-rate of 1.5 mL/min. Operating condition “'a”was select-ed for routine use. For 22 consecutive daily runs theretention time of the MDA-TBA adduct averaged 4.20 (SD0.28) mm.

Spectrophotometric and fluorometric detectors. In pilotstudies we used a photometric monitor (Model 153; AltexScientific, Inc., Berkeley, CA) with 8-pL flow-cell volumeand 1-cm optical pathlength, set at 546 nm and a recordersensitivity of 0.01 A full scale. The detection limit (as MDA)was 2.2 x 1012 mol per 50 j.tL of sample injected onto theHPLC column (equivalent to 2.7 pmol per liter of plasma),which proved inadequate to detect lipoperoxide concentra-tions in plasma from healthy subjects. To increase thesensitivity, we evaluated a fluorescence detector (Model FS-980; Kratos Analytical Division, Spectros, Inc., Ramsey, NJ)with 5-pL flow cell volume, set at excitation wavelength 525nm, emission wavelength 550 nm, and recorder sensitivity= 0.1 pA full scale. The detection limit (as MDA) was 5.6 x1O-’ mol per 50-pL sample injected onto the HPLC column(equivalent to 0.7 pinol per liter of plasma), which was alsoinsufficient to detect lipoperoxide concentrations in plasmasamples from healthy subjects. Finally, a sensitive spectro-photometric detector was evaluated (Model 3000; LDCIMil-ton Roy) with 14-pL flow-cell volume and 1-cm opticalpathlength, set at 532 nm, and modified by use of a 1-mVstrip-chart recorder to provide 0.002 A full scale. Thedetection limit (as MDA) was 1.25 x 1013 mol per 50-pLsample injected onto the HPLC column (equivalent to 0.15pmol of MDA per liter of plasma), which proved quitesufficient for quantification of plasma lipoperoxide concen-trations in healthy persons. Under the specified operatingconditions, the baseline noise level of the recorder tracingwas 1.6 x iO A and the baseline drift was 1.2 x i0 A/h.illustrative chromatograms are shown in Figure 1.

Reaction with thiobarbituric acid. We established opti-mum reaction conditions for formation of the MDA-TBAadduct and precipitation of plasma proteins by systematical-ly varying the concentration of each ingredient of thereaction mixture, the duration of boiling, and the pH and

T#{149}-‘ q-

L1-L

1dli

Fig. 1. IllustratIve chromatograms for HPLCassay of plasma Ilpoperox-kies, obtained by spectrophotometric monitoring at 532 nmPeakheIght(ordinate) = 0.08 mA/cm; chart speed (absctssa)= 1 cm/mm. A =

reagent blank; B = TEP standard (reang, equaIent MDi#{188}2.43 zmol&); C =

plasma from an untreated rat U = recoverysample, TEP standard (as MDA, 2.43imot&) added to the rat plasma

8-

EC)

-

0-

0

TBA concn, mmol/LAg. 2. Effect of TBA concentration on peak heights of the MDA-TBAadduct, in HPLC assays of TEP standards and a human plasmasamplePeak-heightunits (ordinate)=0.08 mA/cm

4 8

Table 1. Run-to-Run PrecIsIon of Llpoperoxide AssaysQuality-controlserum

Beckman Wln

No. of successiveruns 9 22Lipoperoxide concn (MDA, moI/L)

Mean (SD)Range

1.19(0.15) 2.76(0.22)0.91-1.36 2.40-3.20

CV, % 12.6 8.0

Table 2. AnalytIcal Recovery of TEP Standard Added toPlasma Samples

TEP MDAPlasma added found TEP recovered

Ilpop.roxlde, asMQA, i&mol/L pmol/L emol/L

Human0.45 (0.18)b 1.88 2.33 (0.13)b 1.89 (017)b 100.4 (9.O)l

Ratc1.55 (0.35) 1.88 3.38 (0.27) 1.81 (0.10) 97.4 (4.5)

Human0.48 (0.12) 2.43 2.82 (0.15) 2.34 (0.20) 96.0 (8.6)

Rat1.35 (0.36) 2.43 3.78 (0.33) 2.43 (0.17) 99.0 (7.4)aOverallrecovery of TEP added to 21 plasma samples = 98.1 (SD 7.1)%,

range 84-114%. bMean (SD). en = 6; all others, n = 5 plasma samples.

mide by analyzing plasma specimens from several patientsbeing treated with procainamide. Their chromatogramsconsistently showed dual peaks attributed to procainamideand its major metabolite, N-acetylprocainamide; this unusu-al chromatographic pattern helped us to recognize thissource of drug interference.

Reference values were determined for lipoperoxide concen-trations in plasma specimens from 41 healthy adult persons:20 men, 21 women, ages 24 to 56 years. No significantly sex-related differences were observed. Plasma lipoperoxide con-centrations (as MDA) in the aggregate population of 41healthy adults averaged 0.60 .unoI/L (SD 0.13, range 0.39 to0.90 janol/L). Measurements of plasma lipoperoxide concen-trations in patients with atherosclerosis and ischemic heartdisease will be reported separately (Leach CN Jr, et al.,

CLINICALCHEMISTRY,Vol.33, No.2, 1987 217

methanol concentration for protein precipitation. A TBAconcentration of 7 mmol/L in the reaction mixture wasselected on the basis of experiments illustrated in Figure 2,in which chromatographic peak heights of the MDA-TBAadduct obtained with TEP standards and a plasma specimenare plotted vs TBA concentration. A 60-min boiling timewas necessary for maximum yield of MDA-TBA adductfrom plasma samples, although boiling for 30 min sufficedfor the TEP standards. Varying the H3P04 concentration inthe reaction mixture from 0.33 to 0.66 mol/L did not affectthe yield of MDA-TBA adduct from TEP standards orplasma samples, so we selected a H3P04 concentration of0.44 moIJL for routine use.

Calibration curves. Calibration plots (as MDA) werestrictly linear up to 4.86 p.molJL. Based upon measurementsof TEP standards in 13 consecutive analytical runs, the run-to-run coefficient of variation for slopes of the calibrationline was 11%.

Precision and recovery. We evaluated run-to-mn precisionof lipoperoxide assays by analyses of commercial quality-control sera. As indicated in Table 1, replicate daily analy-ses yielded CVs of 12.6% and 8.0% for quality-control serawith lipoperoxide concentrations (as MDA) that averaged,respectively, 1.19 and 2.76 pinol/L. Analytical recovery ofplasma at concentrations of 1.88 or 2.43 pmol/L averaged98.1 (SD 7.1)% (range 88 to 114%), as listed in Table 2.When we increased the plasma sample volume from 50 pLto 100 L, in an attempt to increase analytical sensitivity,this recovery was significantly smaller, ranging from 75 to92%.

Anticoagulants and antioxidant. Plasma specimens col-lected with EDTA as the anticoagulant, as recommended byLee (31), could be validly stored at 4#{176}Cfor at least 24 hwithout significant increase of lipoperoxide concentrationsas compared with assays performed within 1 h after vene-puncture. In contrast, lipoperoxide concentrations were in-creased by 1.5 to 2 times after similar storage of serum or ofplasma specimens collected with heparin or citrate. Addi-tion of EDTA anticoagulant to TEP standards, in the samequantity used for blood collection, did not affect the slope ofthe calibration line. Addition of butylated hydroxytoluenein amounts from 0.75 to 3.0 mg per 1.5 mL of reactionmixture did not affect the yield of MDA-TBA adduct fromplasma samples that were collected with EDTA anticoagu-lant; therefore, butylated hydroxytoluene was not includedin the reaction mixture.

Tests for interference. Addition of glucose, cholesterol,urea, uric acid, creatinine, or bilirubin (in albumin solution)to plasma specimens and TEP standard solutions at theconcentrations listed in Table 3 did not interfere with theassays. Addition of erythrocyte hemolysate to plasma speci-mens caused positive interference in plasma lipoperoxidemeasurements, consistent with previous findings (32).When plasma hemoglobin concentrations were increased to175 mg/L by addition of hemolysate, apparent plasmalipoperoxide concentrations were increased by two- to four-fold.

We added 15 commonly used drugs to plasma specimensand TEP standards at concentrations listed in Table 3 andassayed to test for interference. The only drug that substan-tially interfered was procainamide, which formed a coloredcompound that was eluted from the HPLC column 3.8 mmafter injection. N-Acetylprocainamide also formed a coloredcompound, but because its retention time was 5.5 mm, it didnot interfere. We confirmed the interference by procaina-

NICI2dosage1gmol/kg

0 (controls)250

500

750C

Ref. Date

Serum43

45413546

Plasma87

Presentstudy

1977 Fluorometry

1981 Fluorometry1981 Fluoromotry1983 Spectrophotometry1984 Fluorometry

1985 Spectrophotometry1986 Spectrophotometry1987 HPLC

20 0.61 (0.11)15 0.94 (0.09)41 0.60 (0.13)

Table 3. Tests for Interference In Plasma Llpoperoxide Assays

218 CLINICALCHEMISTRY, Vol. 33, No. 2, 1987

Substnc

GlucoseCholesterolBilirubinUreaUric acidCreatinineHemoglobinAscorbic acidAcetylsalicylic acidAcetaminophenPhenobarbltal

Con added,per titer

16 mmol10 mmol0.1 mmol

36 mmci0.8 mmol0.9 mmol

175 mga0.11 mmol0.55 mmol0.33 mmol0.17 mmol

Substance

TheophyllinePropranololChlordiazepoxideDiazepamAmitriptylineNortnptylineDesiprimineImipriminePhenytoinProcainamideN-Acetylprocainamide

Cone. added,per liter

0.11 mmol1.9 tmol3.3 tmol1.4 mol1.8 Lmol1.9 niol1.9 zmol1.8 niol

80 lLfllOl0.13 mmola0.11 mmol

that interfered inthe assay when added to TEP standardsor humanplasma specimensat the specified concentrations.

manuscript in preparation).Lipoperoxides in plasma of control and NiC12-treated rats.

Plcisnisi lipoperoxide concentrations (as MDA) in vehicle-treated control rats averaged 1.4 tmoI/L (SD 0.3, range 0.9to 1.8 mol/L). Mean lipoperoxide concentrations were in-creased by 1.5 to 2.3 times in plasma sampled from rats 24,48, and 72 h after subcutaneous administration of NiC12 atdosages previously shown to induce lipid peroxidation inlung, liver, and kidney (26, 33) (Table 4).

DIscussion

Since 1968, when Yagi et al. (34) first applied the reactionof TEA with MDA and related chromogens to estimatelipoperoxide concentrations in human serum, the assay hasbeen used to study circulating lipoperoxides in a widespectrum of diseases, such as myocardial ischemia andinfarction (35), cerebral ischemia and stroke (8, 18, 19),atherosclerosis (7, 36-38), diabetes meffitus (39, 40), preg-nancy and pre-eclampsia (41), multiple sclerosis (42), mus-cular dystrophy (15), various liver disorders (4, 12,43,44),burns (9), and paraquat poisoning (45). Although assays ofserum lipoperoxides by the MDA-TBA reaction have be-come popular in Japan, they have achieved limited clinicalacceptance elsewhere because of concerns about analyticalspecificity, sensitivity, reproducibility, recovery, specimeninstability, and drug interference.

Our reference values for plasma lipoperoxide concentra-tions in healthy adults are consistent with those ofFrancesco et al. (8) but substantially lower than values forserum reported by previous investigators (Table 5). We

Table 4. Llpoperoxlde Concentrations In Plasma ofNlClrTreated Rats

Plasma llpoperoxld., asMDA, MmOl/LHours

afterInjectIon

No.of

rats

24-72” 12 1.4 (0.3) 0.9-1.824 5 2.2 (0.3) 1.7-2.448 5 2.3 (0.3) 1.9-2.772 6 2.1 (0.4) 1.5-2.624 7 2.1 (0.2) 1.9-2.448 6 3.1 (0.7) 2.4-4.372 6 3.1 (1.4) 1.7-5.324 6 2.3 (0.4) 2.0-3.048 6 3.2 (0.8) 2.2-4.4

#{149}PJIresults for NiCl2-treatedrats differedsignificantly(p <0.01) fromcontrols bythe Mann-Whitney test.

5Pooled data, no significant differences being found In plasma samplesobtainedat 24, 48, or 72 h afterinjection of NaCI vehiclesolution.

eAt thisdosage, all of the ratsdied between 48 and72 h afterinjectionofNICI2.

Mean (SD) Range

attribute the lower reference values to (a) use of plasmainstead of serum, to avoid release of lipoperoxides fromplatelets during blood coagulation (36); (b) acidification withH3P04, as recommended by Mihara et al. (25), instead oftrichloroacetic or phosphotungstic acids, which were used inmost previous assays; and (c) greater specificity, becauseHPLC separation of the MDA-TBA adduct minimizes spec-trophotometric interference. The improved specificity, sensi-tivity, reproducibility, and analytical recovery provided bythe present assay should increase the acceptance of plasmalipoperoxide determinations for clinical research and diag-nosis. We believe that the present method overcomes thedrawbacks of previous techniques and will prove useful as adiagnostic test for lipid peroxidation in a wide spectrum ofdiseases. However, the technique requires stringent precau-tions to prevent hemolysis; to minimize platelet contamina-tion; to remove traces of lipids and detergents from pipets,test-tubes, and HPLC fittings; and to monitor the purity ofanalytical reagents and solvents.

Others (47, 48) have used HPLC analysis with spec-trophotometry at 267 to 270 nm to measure the productionof MDA from polyenoic fatty acids during incubation invitro with hepatic microsomes. Close correlations werenoted between results for MDA as determined by HPLC andby conventional TEA assays. Analytical sensitivity of thedirect HPLC techniques, without prior formation of theMDA-TDA adduct, evidently was insufficient for plasmalipoperoxide analysis, because the reported detection limitsfor MDA were 7 x 10_il mol (47) and 5 x 10_12 mol (48),respectively. For comparisons, the MDA detection limitswere 1.25 x 10_13 mol in the present study, and 1 xmol as obtained by Yu et al. (24) with a double monochroma-tor spectrofluorometric detector. Others (49, 50) described

Table 5. Reference Values for LlpoperoxideConcentratIons In Serum or Plasma of Healthy Persons

MeanIlpopsroxlde

No. concn,asMDA,Method subjects umot/I (and SD)

82 3.42 (0.94)75 9 3.10 (0.62)8 3.74 (0.63)

19 9 3.27 (0.89)94 47.2 (7.0)10 1.7 (0.5)

CLINICAL CHEMISTRY,Vol. 33, No. 2, 1987 219

HPLC techniques for analysis of serum lipoperoxides thatinvolve acid hydrolysis of lipoperoxides and complezation ofMDA with 1,3-diphenyl-2-thiobarbituric acid, extraction ofthe MDA-1,3-diphenyl-2-thiobarbituric acid adduct in ace-tonitrile-pyridine or butane, separation of the adduct byHPLC on a column of LiChrosorb-RP18, with elution inacetonitrile-water, and quantification of it by spectropho-tometry at 537 nni (49) or fluorometry at 548 nm withexcitation at 525 nm (50). Detection limits for this adductwere 1 x 10_12 mol by spectrophotometry (49), 2.4 x 1014mol by fluorometry (50).

In rats with lipid peroxidation induced by NiC12, the time.course and magnitude of hyperlipoperoxidemia found in thisstudy are consistent with previous data showing above-normal concentrations of TEA chromogens in homogenatesof lung, liver, and kidney (26, 33), increased excretion ofethane and ethylene in expired breath (51), and increasedconcentration of conjugated dienes in hepatic microsomallipids (33). Thus, the present findings demonstrate theapplicability of the plasma lipoperoxide assay for detectionof lipid peroxidation induced by xenobiotic toxicity in experi-mental animals.

This study was supported by grants from the Northeast UtilitiesCompany, the National Institute of Environmental Health Sciences(ES-01337),and the U.S. Dept. of Energy (EY-03 140).

References1. Slater TF. Free-radical mechanisms in tissue injury [Review].Biochem J 1984;222:1-15.2. Halliwell B, Gutteridge JMC. The importance of free radicalsand catalytic metal concentrations in human diseases [Review].Molec Aspects Med 1985;8:89-193.3. Sevanian A, Hochstein P. Mechanisms and consequences of lipidperoxidation in biological systems [Review]. Ann Rev Nutr1985;5:365-90.4. Comporti M. Biologyof disease: lipid peroxidation and cellulardamage in toxic liver injury [Review]. Lab Invest 1985;53:599-623.5. Pryor WA. Free radical involvement in chronic diseases andaging the toxicity of lipid hydroperoxides and their decompositionproducts [Review]. In: Finley JW, Schwass DE, eds. Xenobioticmetabolism: nutritional effects. Washington, DC: American Chemi-cal Society, 1985:77-96.6. Halliwell B, Gutteridge ,JMC, Blake D. Metal ions and oxygenradical reactions in human inflammatory joint disease. PhilosTrans R Soc London1985;B311:659-71.7. Ledwozyw A, Michalak J, Stepien A, Kadziolka A. The relation-ship between plasma triglycerides, cholesterol, total lipids and lipidperoxidation products during human atherosclerosis. Clin ChimActs 1986;155:275-84.8. Francesco V, Cesare A, Luigi I, Stefano F, Andrea G, FrancescoB. Malondialdehyde-like material and beta-thromboglobulin plas-ma levels in patients suffering from transient ischemic attacks.Stroke 1985;16:14-6.9. Hiramatsu M, Izawa Y, Hgihara M, Nishigaki I, Yagi K. Serumlipid peroxide levels of patients suffering from thermal injury.Burns 1984;11:111-6.10. Sunderman FW Jr. Metals and lipid peroxidation [Review].Acts Pharmacol Toxicol 1986;59(suppl 7):248-55.11. Sunderman FW Jr. Biochemical indices of lipid peroxidation inoccupational and environmental medicine. In: Colombi A, EmmettEA, Foa V, Maroni M, eds. Chemical monitoring of occupationaltoxicity, Chichester, U.K.: Ellis Harwood, Ltd., 1987 (in press).12. Suematsu T, Abe H. Liver and serum lipid peroxide levels inpatients with liver diseases. In: Yagi K, ed. Lipid peroxides inmedicine and biology. New York: Academic Press, 1982:285-93.13. Tsuchida M, Miura T, Mizutani K, Aibara K. Fluorescentsubstances in mouse and human sera as a parameter of in vivo lipidperoxidation. Biochim Biophys Acts 1985;834:196-204.

14. Dormandy TL. Free radical activity and diene conjugation inman. In: Poli G, Cheeseman KH, Dianzani MU, Slater TF, eds. Freeradicals in liver injury. Oxford, U.K.: IRL Press, Lt& 1985:167-73.15. Hunter MIS, Mohamed JB. Plasma antioxidanta and lipidperoxidation products in Duchenne muscular dystrophy. Clin ChimActs 1986;155:123-32.16. Wade CR, van Rij AM. In vim lipid peroxidation in man asmeasured by the respiratory excretion of ethane, pentane, and otherlow-molecular-weight hydrocarbons. Anal Biochem 1985;150:1-7.17. Wispe JR, Bell EF, Roberta RJ. Assessment of lipid peroxida-tion in newborn infants and rabbits by measurements of expiredethane and pentane: influence of parenteral lipid infusion. PediatrHas 1985;19:374-9.18. Satoh K. Serum lipid peroxide in cerebrovascular disordersdetermined by a new colorimetric method. Cliii Chim Acts1978;90:37-43.19. Santos MT, Valles J, Aznar J, Vilches J. Determination ofplasma malondialdehyde-like material and its clinical applicationin stroke patients. J Cliii Pathol 1980;33:973-.6.20. Yagi K. Assay for serum lipid peroxide level and its clinicalsignificance. In: Yagi K, ed. Lipid peroxides in biology and medi-cine. New York: Academic Press, 1982:223-42.21. Schimke I, Papies B. Some methodological aspects of theestimation of thiobarbituric acid-reactive substances in plasma. ZMed Lab Diagn 1986;27:71-5.22. Kirkpatrick DT, Guth DJ, Mavis RD. Detection of in vivo lipidperoxidation using the thiobarbituric acid assay for lipid hydroper-oxides. J Biochem Toxicol 1986;1:93-104.23. Bird HP, Hung SSO, Hadley M, Draper H}I. Determination ofmalonaldehyde in biological materials by high-pressure liquidchromatography. Anal Biochem 1983;128:240-4.24. Yu LW, Latriano L, Duncan S, Hartwick HA, Witz G. High-performance liquid chromatographic analysis of the thiobarbituricacid adducts of malonaldehyde and trans,tmns-muconaldehyde.Anal Biochem 1986;156:326-.33.25. Mihara M, Uchiyama M, Fukuzawa K. Thiobarbituric acidvalue on fresh homogenate of rat as a parameter of lipid peroxida-tion in aging, Cd4 intoxication, and vitamin E deficiency. BiochemMed 198023:302-11.26. Sunderman FW Jr, Marzouk AB, Hopfer SM, Zaharia 0, ReidMC. Increased lipid peroxidation in tissues of nickel chloridetreated rats. Ann Cliii Lab Sci 1985;15:229-36.27. Nair V, Turner GA. The thiobarbituric acid test for lipidperoxidation. Structure of the adduct with malondialdehyde. Lipids1984;19:804-5.28. Gutteridge JMC. The use of standards for malonyldialdehyde.Anal Biochem 1975;69:518-26.29. Harboe M. A method for determination of hemoglobin inplasma by near-ultraviolet spectrophotometry. Scand J Cliii LabInvest 1959;11:66-70.30. Sachs L. Applied statisti . a handbook of techniques. 2nd ed.New York: Springer-Verlag, 1984:1-707.31. Lee DM. Malondialdehyde formation in stored plasma. Bio-diem Biophys Res Commun 1980;95:1663-72.32. Gilbert ITS, Stump DD, Roth EF Jr. A method to correct forerrors caused by generation of interfering compounds during eryth-rocyte lipid peroxidation. Anal Biochem 1984;137:282-6.33. Donskoy E, Donskoy M, Forouhar F, et al. Hepatic toxicity ofnickel chloride in rats. Ann Cliii Lab Sci 1986;16:108-17.34. Yagi K, Nishigaki I, Ohama H. Measurement of serum TBA-value. Vitamins 1968;37:105-12.35. Aznar J, Santos MT, Vafles J, Sala J. Serum malondialdehyde-like material (MDA-LM) in acute myocardial infarction. J ClinPathol 1983;36:712-5.36. Szczeklik A, Gryglewski RJ, Domagala B, et al. Serum lipopro-teins, lipid peroxides and prostacydlin biosynthesis in patients withcoronary heart disease. Prostaglandins 1981;22:795-807.37. Loeper J, Coy J, Emerit J, Rozenstajn L, Jeny C, Bedu 0. Studyof fatty acids and lipid peroxidation during the course of humanatherosclerosis. Sem Hop Paris 1983;59:1657-60.

220 CLINICAL CHEMISTRY, Vol. 33, No. 2, 1987

38. Zany C, Coy J, Fechner J, Leaper J, Emerit J. Study ofperoxidation in human and experimental atheroma by measure-ments of plasma malondialdehyde. Presse Med 1986;15:1131-3.39. Sato Y, Hotta N, Sakamoto N, Matsuoka 5, Ohishi N, Yagi K.Lipid peroxide levels in plasma of diabetic patients. Biochem Med197921:104-7.40. Nishigaki I, Hagihara M, Tsunekawa H, Maseki M, Yagi K.Lipid peroxide levels of serum lipoprotein fractions of diabeticpatients. Biochem Med 198125:373-8.41. Maseki M, Niahigaki J, Hagihara M, Tomoda Y, Yagi K. Lipidperoxide levels and lipid content of serum lipoprotein fractions ofpregnant subjects with or without pre-eclampsia. Clin Chim Acts1981;115:155-61.42. Hunter MIS, Nlemadim BC, Davidson DLW. Lipid peroxida-tion products and antioxidant proteins in plasma and cerebroepinalfluid from multiple sclerosis patients. Neurochem Res1985;10:1645.-52.

43. Suematsu T, Kamada T, Abe H, Kikuchi S, Yagi K. Serumlipoperoxide level in patients suffering from liver disease. ClinChim Acts 1977;79:267-70.44. Suematsu T, Matsumura T, Sato N, et al. Lipid peroxidation inalcoholic liver disease in humans. Alcohol Cliii Exp Has 1981;5:427-30.45. Yasaka K, Ohya I, Mataumoto J, Shiramizu T, Sasaguri Y

Acceleration of lipid peroxidation in human paraquat poisoning.Arch Intern Med 1981;141:1169-71.46. Viinikka L, Vuori J, Ylikorkala 0. Lipid peroxides, proatacy-cliii, and thomboxane A2 in runners during acute exercise. Med SdSport Exer 1984;16:275-7.

47. Esterbauer H, Slater TF. The quantitative estimation by highperformance liquid chromatography of free malonaldehyde pro-duced by peroxidizing microsomes. IRCS Med Sci 1981;9:749-50.

48. Bull AW, Marnett U. Determination of malondialdehyde byion-pairing high-performance liquid chromatography. Anal Bio-chem 1985;149:284-90.49. Nakashima K, Ando T, Nakamizo T, Akiyama S. Developmentand application of organic reagents for analysis. V. High-perform-ance liquid chromatographic determination of lipoperoxides inbiological fluids with 1,3-diphenyl-2-thiobarbituric acid. ChemPharm Bull 1985;32:5380-4.50. Ohsawa K, Yoshimura Y, Tanaka H, Okajima F, Tamura K,!maeda K. High performance liquid chromatography-fluorometricdetermination of lipid peroxides in serum with 1,3-diphenyl-2-thiobarbituric acid. Anal Sci 1985;1:473.-6.51. Knight JA, Hopfer SM, Reid MC, Wong SHY, Sunderman FWJr. Ethene (ethylene) and ethane exhalation in Ni[IJJ-treated rats,using an improved rebreathing apparatus. Ann Cliii Lab Sci1986;16:386-.394.