Clinical Chemistry 46:6
829 – 836 (2000)
HPLC Analysis of Lipid-derived Polyunsaturated
Fatty Acid Peroxidation Products in Oxidatively
Modified Human Plasma
Richard W. Browne1,2* and Donald Armstrong1
Background: Lipid peroxidation is a prominent mani-
festation of free radical activity and oxidative stress in
biological systems. Diverse methodologies have been
developed that measure a variety of lipid peroxidation
products used as markers of lipid peroxidation processes.
Methods: Hydroxy and hydroperoxy polyunsaturated
fatty acid (PUFA) peroxidation products were analyzed
in human blood plasma by reversed-phase HPLC after
liquid-liquid extraction of total lipids and alkaline hy-
drolysis of lipid esters to liberate free PUFAs. An
isocratic mobile phase containing 1 g/L acetic acid-
acetonitrile-tetrahydrofuran (52:30:18, by volume) over
60 min duration, with ultraviolet absorbance detection
at 236 nm by photodiode array, enabled the resolution
and quantification of 13 regioisomeric hydroxy and
hydroperoxy PUFAs.
Results: As little as 250 ␮L of human plasma was
utilized with an analytical range of 0.033–1.6 ␮mol/L for
each compound. Intra- and interassay CVs for all com-
pounds detected in normal or oxidatively modified
human plasma were 3.2–11% and 4.7–12%, respectively.
Analytical recoveries were 87–103%. Analysis of human
plasma exposed to artificial oxidation with Cu2ⴙ ion and
hydrogen peroxide, a free radical-generating reaction,
showed marked increases in hydroxy and hydroperoxy
PUFA concentrations.
Conclusion: Lipid-derived hydroxy and hydroperoxy
PUFAs may be useful as clinical markers of lipid peroxi-
dation and oxidative stress in the peripheral circulation.
© 2000 American Association for Clinical Chemistry
Departments of 1 Clinical Laboratory Science and 2 Social and Preventive
Medicine, State University of New York at Buffalo, Buffalo, NY 14214.
This work was carried out as part of the doctoral dissertation of Richard W.
Browne.
*Address correspondence to this author at: Department of Social and
Preventive Medicine, State University of New York at Buffalo, 270 Farber Hall,
3435 Main St., Buffalo, NY 14214. Fax 716-829-2979; e-mail rwbrowne@
acsu.buffalo.edu.
Received December 20, 1999; accepted March 28, 2000.
829
Lipids, Lipoproteins,
and Cardiovascular
Risk Factors
The oxidative modification of lipids has been implicated
in the pathogenesis of many diseases (1 ), particularly
atherosclerosis and coronary vascular disease (2 ). Oxida-
tive lipid modifications occur through lipid peroxidation
mechanisms in which free radicals and reactive oxygen
species abstract a methylene hydrogen atom from poly-
unsaturated fatty acids (PUFAs),3 producing a carbon-
centered lipid radical. Spontaneous rearrangement of the
1,4-pentadiene yields a conjugated diene, which reacts
with molecular oxygen to form a lipid peroxyl radical.
Abstraction of a proton from a neighboring PUFA pro-
duces a lipid hydroperoxide (LOOH) and regeneration of
a carbon-centered lipid radical, thereby propagating the
radical reaction. The hydroperoxide moiety of LOOH can
be reduced by glutathione-dependent peroxidases (phos-
pholipid glutathione peroxidase or, after hydrolysis to
free fatty acids, glutathione peroxidase) to an alcohol,
yielding a hydroxy derivative (LOH). LOOH and LOH
represent the primary stable end products of lipid peroxi-
dation (3 ).
Measurement of lipid peroxidation has most often
been accomplished through semiquantitative analysis of
total hydroperoxide content, total conjugated dienes, or
through measurement of LOOH breakdown products
3
Nonstandard abbreviations: PUFA, polyunsaturated fatty acid; LOOH,
lipid hydroperoxide; LOH, lipid hydroxy derivative; MDA, malondialdehyde;
UV, ultraviolet; FAOOH, fatty acid hydroperoxide; FAOH, hydroxy fatty acid;
9-HpODE, 9S-hydroperoxy-10Z,12E-octadecadienoic acid; 13-HpODE, 13S-
hydroperoxy-9Z,11E-octadecadienoic acid; 9-HODE, 9S-10Z,12E-hydroxy-
octadecadienoic acid; 13-HODE, 13S-hydroxy-9Z,11E-octadecadienoic acid;
13-HpOTE, 13S-hydroperoxy-9Z,11E,15Z-octadecatrienoic acid; 15-HpETE, 15-
hydroperoxy-5Z,8Z,11Z,13E-eicosatetraenoic acid; 12-HpETE, 12-hydroper-
oxy-5Z,8Z,10E,14Z-eicosatetraenoic acid; 5-HpETE, 5-hydroperoxy-6E,8Z,
11Z,14Z-eicosatetraenoic acid; 15-HETE, 15-hydroxy-5Z,8Z,11Z,13E-eicosate-
traenoic acid; 12-HETE, 12-hydroxy-5Z,8Z,10E,14Z-eicosatetraenoic acid;
5-HETE, 5-hydroxy-6E,8Z,11Z,14Z-eicosatetraenoic; 5-HETE-ME, (⫾)5-hy-
droxy-6E,8Z,11Z,14Z-eicosatetraenoic acid-methyl ester; 13(R)HODE, 13R-hy-
droxy-9Z,11E-octadecadienoic acid; HDHE, hydroxy-docosahexaenoic acid;
HpDHE, hydroperoxy-docosahexaenoic acid; HIP, hexane/isopropanol (3:2,
by volume); TLC, thin layer chromatography; and MDQ, minimum detectable
quantity.
830 Browne and Armstrong: HPLC Analysis of Lipid Peroxides
(4, 5 ). Each of these techniques has shortcomings. Total
hydroperoxide methods such as iodometry and the ferric
iron oxidation/xylenol orange technique do not measure
the hydroxy derivative, which has been shown to be the
predominant form (6 ), and are subject to interference
from non-lipid peroxides present in either the sample or
the reagents. Total conjugated diene measurements are
not specific for lipid peroxidation because conjugated
dienes are found in compounds such as isolinoleic acid,
which usually is present in human plasma in much higher
concentrations than oxidatively generated conjugated
dienes (7 ). Lipid peroxidation byproduct analyses, such
as the thiobarbituric acid reactive substances test, mea-
sure malondialdehyde (MDA). MDA is a small three-
carbon aldehyde generated from the hydrolysis of certain
LOOHs. However, MDA is not generated exclusively by
breakdown of LOOHs, nor does the thiobarbituric acid
reactive substances test measure MDA exclusively (8 ).
The test is therefore nonspecific. The most detailed anal-
yses achieved to date have been through use of gas
chromatography–mass spectrometry (9 –12 ). However,
the hydroperoxide group of LOOH is not stable at the
temperatures required for gas chromatographic analyses;
therefore, direct measurement is impossible. The typical
strategy has been to analyze the trimethylsilyl derivative
of LOH before and after reduction of the sample with a
strong reducing agent such as sodium borohydride or
triphenylphosphine (9 ). The LOOH concentration is then
calculated as the difference between pre- and postreduc-
tion LOH concentrations. HPLC offers the advantage of
separating LOOHs and LOHs without derivatization.
After HPLC separation, detection schemes that include
ultraviolet (UV) absorbance of the conjugated diene at 234
nm, postcolumn chemiluminescence, and electrochemical
reduction have been described (13–16 ). Although chemi-
luminescence and electrochemical techniques detect
LOOHs at much lower concentrations than UV absor-
bance, they do not detect to LOHs. This is a drawback
because it has been shown that the major lipid peroxida-
tion products in human plasma are the more stable LOHs
(6 ). UV absorbance is able to identify both LOOHs and
LOHs because they both possess a conjugated diene that
absorbs strongly (⑀ ⫽ 23 000 –27 000) near 234 nm. A
general difficulty of chromatography applied to total lipid
peroxidation measurements is that LOOHs and LOHs
occur predominantly esterified in cholesterol esters, phos-
pholipids, and triglycerides, which are difficult to resolve
in a single chromatographic system. To overcome this and
to provide a more accurate estimation of total lipid
peroxidation, alkaline hydrolysis of lipid esters to yield
free fatty acid hydroperoxides (FAOOHs) and hydroxy
fatty acids (FAOHs), followed by HPLC separation and
direct UV detection of the conjugated dienes has been
described (16 ). This method provides a substantial
amount of information regarding the amount of total
lipid peroxidation, as well as the identity of the PUFA
precursor.
FAOOH and FAOH species found in biological mate-
rial can vary in carbon chain length from 18 to 22 carbons
and degree of unsaturation (2 to 6 double bonds). The
position of the oxygenation site relative to the carbon
chain also varies because each pair of double bonds
allows for two theoretical peroxidation sites (17 ). In the
case of linoleic acid, the major PUFA in biological
systems, this produces two possible FAOOH and two
possible FAOH regioisomers (i.e., 9- and 13-hydroperoxy-
octadecadienoic acid and 9- and 13-hydroxy-octadecadi-
enoic acid). As the number of double bonds increases so
does the number of possible regioisomers. Increasing
complexity is added in possible cis, trans (Z, E) isomers
and enantiomers (R, S). Each of the above mentioned
regioisomers can exist in cis-trans or trans-trans configu-
ration, and normal phase HPLC systems have been de-
scribed that resolve these isomeric forms (18 ) as well as
chiral phase systems that resolve enantiomers (19 ).
We have described a method for the separation and
analysis of mixtures of synthetic hydroxy and hydroper-
oxy PUFAs by reversed-phase HPLC (20 ). The present
report improves this methodology to achieve resolution of
regioisomers of linoleic, linolenic, arachidonic, and doco-
sahexaenoic acid FAOOHs and FAOHs. The method is
applied to the analysis of FAOOHs and FAOHs generated
by in vitro oxidation of human plasma by metal-catalyzed
oxidation with the Cu2⫹ ion and hydrogen peroxide. Cu2⫹
has been shown to react with any LOOH already present
in a sample to cleave the LOOH:
Cu2⫹ ⫹ LOOH 3 Cu⫹ ⫹ LOO 䡠 ⫹ H⫹ ⫹ Cu⫹
The Cu⫹ product could then cleave the LOOH or H2O2 in
a reaction analogous to the Fenton reaction:
Cu⫹ ⫹ HOOH 3 Cu2⫹ ⫹ OH 䡠 ⫹ OH⫺
Cu⫹ ⫹ LOOH 3 Cu2⫹ ⫹ LO 䡠 ⫹ ⫺OH
The peroxyl radical (LOO䡠), alkoxyl radical (LO䡠), and
hydroxyl radical (OH䡠) have all been shown to be capable
of initiating and propagating lipid peroxidation (16 ).
Materials and Methods
Unless otherwise indicated, reagents were obtained from
Sigma. All organic solvents were HPLC grade and were
obtained from J.T. Baker. To prevent autooxidation dur-
ing sample processing, dissolved oxygen was removed
from all solvents by ultrasonication under reduced pres-
sure followed by 15 min of sparging with helium. Helium,
oxygen, argon, and prepurified nitrogen were obtained
from Strate Welding Supply.
Bulk quantities of FAOOHs for stability and recovery
studies were synthesized using type 1 soybean lipoxidase
to generate FAOOHs from pure linoleic, linolenic, arachi-
donic, and docosahexaenoic acid (20 ). Bulk FAOHs were
synthesized by reduction of FAOOHs with sodium boro-
hydride in ice-cold methanol followed by chloroform
re-extraction. The isomeric identity of bulk FAOOH and
 Clinical Chemistry 46, No. 6, 2000
FAOH preparations was confirmed against retention time
and absorbance spectra of FAOOHs and FAOHs of
known regioisomeric composition: 9S-hydroperoxy-10Z,12E-
octadecadienoic acid (9-HpODE), 13S-hydroperoxy-9Z,11E-
octadecadienoic acid (13-HpODE), 9S-10Z,12E-hydroxy-
octadecadienoic acid (9-HODE), 13S-hydroxy-9Z,11E-
octadecadienoic acid (13-HODE), 9S-hydroperoxy-10E,
12Z,15Z-octadecatrienoic acid, 13S-hydroperoxy-9Z,11E,
15Z-octadecatrienoic acid (13-HpOTE), 9S-hydroxy-10E,
12Z,15Z-octadecatrienoic acid, 13S-hydroxy-9Z,11E,15Z-
octadecatrienoic acid, 15-hydroperoxy-5Z,8Z,11Z,13E-eico-
satetraenoic acid (15-HpETE), 12-hydroperoxy-5Z,8Z,10E,
14Z-eicosatetraenoic acid (12-HpETE), 5-hydroperoxy-
6E,8Z,11Z,14Z-eicosatetraenoic acid (5-HpETE), 15-hy-
droxy-5Z,8Z,11Z,13E-eicosatetraenoic acid (15-HETE), 12-
hydroxy-5Z,8Z,10E,14Z-eicosatetraenoic acid (12-HETE),
and 5-hydroxy-6E,8Z,11Z,14Z-eicosatetraenoic acid (5-
HETE). These compounds as well as an internal standard
of (⫾)5-hydroxy-6E,8Z,11Z,14Z-eicosatetraenoic acid-
methyl ester (5-HETE-ME) and the R-enantiomer of 13-
HODE (13R-hydroxy-9Z,11E-octadecadienoic acid) were
obtained as pure solutions in ethanol and used as received
(Caymen Chemical).
instrumentation
The equipment used was as follows: a UV-visible record-
ing spectrophotometer (Model 160 U) and HPLC system
were from Shimadzu. The HPLC system consisted of an
LC-7A solvent delivery system, an LPM-600 low-pressure
mixing proportioning valve, a SIL-9A automatic sample
injector, an SPD/M6A UV/VIS photodiode array, and an
online IBM personal computer with Shimadzu Class-VP
chromatograph data processing software. The mobile
phase reservoirs were continuously sparged from a he-
lium degassing system (Kontes Glassware). The reversed-
phase analytical HPLC column was a Supelcosil LC-18
(4.6 mm ⫻ 25.0 cm; 5.0-␮m particle size; 100Å pore size),
with a Supelcoguard LC-18 guard column (4.6 mm ⫻ 2.0
cm; 5-␮m particle size; 100Å pore size; Supelco).
sample preparation
Hexane-isopropanol (HIP) total lipid extracts (21 ) were
prepared by adding 1.0 mL of isopropanol containing 0.1
g/L butylated hydroxytoluene to 0.5 mL of EDTA plasma.
After 2 mL of hexane was added, the vial was perfused
with nitrogen, capped, vortex-mixed for 1 min, and cen-
trifuged for 3 min at 3000g; the upper hexane phase was
collected by aspiration. The extraction was repeated three
times, and the hexane layers were pooled and evaporated
to dryness under nitrogen. To analyze the total lipid
extraction efficiency, the aqueous layer was re-extracted
three times with chloroform-methanol (2:1, by volume).
Total lipids were analyzed on adsorption thin-layer chro-
matography (TLC) plates (20 ⫻ 20 cm) precoated with
0.25 mm of silica gel G without fluorescent indicator (EM
Laboratories); the plates were developed with hexane-
diethyl ether-acetic acid (80:20:1, by volume). Alkaline
831
hydrolysis of total dried lipid extracts was performed by
dissolving the extracts in 0.95 mL of degassed, absolute
ethanol. NaOH (50 ␮L of a10 mol/L solution) was added,
and the sample was perfused with nitrogen, capped,
heated at 60 °C for 20 min, and neutralized with 30 ␮L of
glacial acetic acid. 5-HETE-ME (100 ␮L of a 1.0 ␮mol/L
solution) was added as internal standard. After the etha-
nol was evaporated under nitrogen, the sample was
dissolved in 1.0 mL of water and extracted twice with 2.0
mL of n-heptane. The upper phase was collected, pooled,
and evaporated under nitrogen, and the residue resus-
pended in 250 ␮L of ethanol. Completeness of the alkaline
hydrolysis was assessed by TLC as described above.
Immediately before HPLC injection, 250 ␮L of water was
added to the samples to ensure efficient mass transfer to
the stationary phase.
hplc analysis
Sample (150 ␮L) was injected into the HPLC system,
eluted isocratically with a mobile phase containing 1 g/L
acetic acid-acetonitrile-tetrahydrofuran (52:30:18, by vol-
ume) over 60 min and monitored at 200 –300 nm by the
photodiode array. It should be noted that a photodiode
array is not necessary for this methodology; a simple UV
detector could be used. Integration of peak areas was
performed at 236 nm with an 8-nm bandwidth. This
wavelength and bandwidth were chosen to encompass
the wavelength of maximum absorbance (␭max) of the
octadecadienoic acids at 234 nm and the eicosatetraenoic
acids at 237 nm. Quantification was based on an external
calibration curve using ethanolic calibrators prepared on
a Shimadzu 160 UV scanning spectrophotometer apply-
ing the ␭max and absorptivity coefficients provided by the
manufacturer. Serial dilutions were made in ethanol-
water (50:50, by volume), and calibration curves were
generated by triplicate injections of each calibrator. Sam-
ple concentrations were interpolated from calibration
curves and corrected for recovery of 5-HETE-ME internal
standard.
FAOH recovery experiments were performed by add-
ing known amounts of calibrators into human plasma.
FAOOH recovery experiments were performed by adding
calibrators into isopropyl alcohol-precipitated plasma.
Reproducibility experiments were carried out on a pool of
normal human EDTA plasma as a low-concentration
control. A separate high-concentration pool was oxidized
with 100 ␮mol/L Cu2⫹ ion for 12 h at 37 °C and then
stabilized by addition of 250 ␮mol/L EDTA followed by
portioning and freezing at ⫺80 °C. Intraassay reproduc-
ibility was estimated by 20 replicates of each control pool
in one batch. Interassay reproducibility was estimated by
duplicate, daily analysis of control pools over a period of
20 days.
Results
Pure bulk preparations of FAOOHs and FAOHs were
obtained by reaction with lipoxidase followed by reduc-
832 Browne and Armstrong: HPLC Analysis of Lipid Peroxides
tion with sodium borohydride. Linoleic and linolenic acid
FAOOHs synthesized in this manner consisted of ⬃95%
13-hydroperoxy isomer, 3– 4% 13-hydroxy isomer, and
1–2% 9-hydroperoxy and -hydroxy isomers. Subsequent
reduction of these preparations produced 95–97% of the
13-hydroxy isomers and small amounts of 9-hydroxy
isomers. The addition of 9-HODE, 13-HODE, 9-HpODE,
and 13-HpODE calibrators confirmed the identity of these
peaks as the cis-trans isomers of these oxidation products.
Pure 13(R)HODE was not resolved from 13(S)HODE, and
the alkaline hydrolysis product of the internal standard
(⫾)5-HETE-ME, which was a mixture of R and S enanti-
omers, gave only one peak on analysis, indicating that
there was no resolution of enantiomers. Arachidonic acid
preparations consisted of comparable purity of the 15-
hydroperoxy and -hydroxy isomers. There were no other
detectable eicosanoid isomers present in lipoxidase prep-
arations. The regioisomer identity of docosahexaenoate
preparations could not be confirmed because of a lack of
commercial calibrators. However, the hydroperoxy doco-
sahexaenoic acid lipoxidase product (HpDHE) as well as
the subsequent hydroxy reduction product (HDHE) each
generated only one chromatographically resolved peak.
These bulk calibrators showed no change in UV absor-
bance or chromatographic characteristics for at least 6
months when stored in degassed ethanol under argon at
⫺80 °C.
Chloroform-methanol extracts of the remaining aque-
ous layer after HIP extraction showed no detectable spots
on TLC, indicating complete extraction of total lipids. Fig.
1 shows the results of TLC analysis of lipid extracts after
alkaline hydrolysis. The lipid esters were completely
hydrolyzed in 20 min in NaOH at a concentration of 0.5
mol/L.
Thomas and Jackson (16 ) reported the conversion of
Fig. 1. Adsorption thin layer chromatogram of total lipid extracts from
human plasma developed with chloroform-ether-acetic acid (80:20:1,
by volume) and visualized by charring with 5 mL/L sulfuric acid
saturated with ceric ammonium sulfate.
Extracts were subjected to alkaline hydrolysis for 20 min at 60 °C at varying
concentrations of NaOH. Lane 1, no NaOH; lane 2, 0.03 mol/L NaOH; lane 3,
0.06 mol/L NaOH; lane 4, 0.125 mol/L NaOH; lane 5, 0.25 mol/L NaOH; lane
6, 0.5 mol/L NaOH; lane 7, calibration mixture of cholesterol linoleate (CE),
␣-tocopherol (VIT E), triolein (TRIG), linoleic and arachidonic acids (FFA), choles-
terol (CHOL), phosphatidyl choline (PC), and monoglycerides (MG).
small amounts of FAOOHs to FAOHs during alkaline
hydrolysis of lipid esters, and the reaction of FAOOHs
with strong alkali has been described in detail by Gar-
dener et al. (22 ). To account for possible artifacts in our
analysis, we studied the effect of alkaline hydrolysis on
the stability of FAOOH calibrators. A mixture of FAOOH
calibrators in ethanol (13-HpODE, 13-HpOTE, 15-HpETE,
and HpDHE) was split into two equal parts. One was
analyzed directly by HPLC, whereas the other was sapon-
ified as described above, neutralized, and then analyzed.
Fig. 2 shows a comparison of these two chromatograms.
Alkaline hydrolysis produces an 8 –12% loss of each
FAOOH, which was recovered as the corresponding
FAOH. Three-dimensional diode array analysis showed
no change in the absorbance spectra of the FAOOHs or
the FAOHs after alkaline hydrolysis.
Autooxidation of PUFAs during processing or analysis
of biological samples is a concern with all lipid peroxida-
tion methodologies. To assess possible lipid peroxidation
during sample processing, we prepared a mixture con-
taining 12.5 mg each of linoleic, arachidonic, linolenic,
and docosahexaenoic acids in 1.0 mL of ethanol. 5-HETE-
ME was added to the solution. The solution was then
split, and half was diluted 1:2 in water and immediately
subjected to HPLC using a mobile phase of 1 g/L acetic
acid-acetonitrile-tetrahydrofuran (41:41:18, by volume).
The photodiode array monitored the PUFAs at 215 nm
and the LOOHs and LOHs at 236 nm. The second split
was evaporated to dryness under nitrogen to remove the
ethanol, redissolved in 1.5 mL of 0.1 g/L butylated
Fig. 2. Comparison of HPLC chromatograms of FAOOH calibrators
before (A) and after (B) alkaline hydrolysis in 0.5 mol/L ethanolic
NaOH.
Peaks: 1, 13-HpOTE; 2, 13-HpODE; 3, 15-HpETE; 4, HpDHE; 5, 13S-hydroxy-
9Z,11E,15Z-octadecatrienoic acid; 6, 13-HODE; 7, 15-HETE; 8, HDHE.
 Clinical Chemistry 46, No. 6, 2000 833

hydroxytoluene in isopropyl alcohol-water (2:1, by vol-
ume), processed as described for plasma, and analyzed
under the same HPLC conditions. Fig. 3 shows a compar-
ison of these two chromatograms. This mobile phase
allowed the simultaneous determination of native PUFAs,
FAOOHs, and FAOHs within 60 min, although there was
considerable loss of resolution of the FAOOH and FAOH
regioisomers. We noted that there were substantial
amounts of LOOHs and LOHs in pure commercial prep-
arations of PUFAs. Approximately 90% of the PUFAs,
FAOOHs, and FAOHs were recovered after sample pro-
cessing, and there was no significant increase in any
FAOOH or FAOH relative to the corresponding PUFA.
The most notable difference was the presence of a sizeable
peak at 17.5 min, which corresponded to 5-HETE and was
attributable to alkaline hydrolysis of the 5-HETE-ME
internal standard during processing.
To estimate the possible complexity of chromatograms
generated from biological samples and to assess the
ability of the method to respond to a known mechanism
of induced oxidative damage, plasma was exposed to
CuSO4-H2O2. Normal human plasma was obtained, and
baseline FAOOH and FAOH concentrations were deter-
mined. The specimens were incubated with 100 ␮mol/L
CuSO4 and 0.3 mL/L H2O2 at 37 °C for 2 h, and aliquots
removed at 30, 60, and 120 min for FAOOH/FAOH
analysis. Fig. 4 shows the increase of plasma FAOOH and
FAOH concentrations as a function of time. The linoleic
acid oxidation products 13-HODE and 9-HODE were the
only quantifiable compounds in normal plasma and con-
stituted the major peaks in all analyses. 9-HpODE was
detected at baseline but was below the limit of quantifi-
cation. 9-HpODE was the first increased FAOOH and was
quantifiable at 30 min, 13(S)HpODE was quantifiable by
60 min as were the arachidonic acid products
15(S)HpETE, 12(S)HETE, and 5(S)HETE. These products
continued to rise in proportion to each other as oxidation
continued.
After systematic changes in the mobile phase, the
maximum resolution of all calibrators with acceptable
elution time was achieved using an isocratic mobile phase
consisting of 1 g/L acetic acid-acetonitrile-tetrahydrofu-
ran (52:30:18, by volume). A chromatogram of all of the
commercial calibrators is shown in Fig. 5. Of the 15
FAOOH and FAOH isomers that we synthesized or
purchased, 13 peaks were resolved by the HPLC condi-
tions described above. Among the calibrators, co-elution
occurred between 12-HETE and HpDHE (peak 11) and
between 12-HpETE and 5-HETE (peak 12). 15-HpETE and
15-HETE (peaks 6 and 10), which were not fully resolved
from the down slope of 9-HpODE and 9-HODE, respec-
tively, were integrated by a manual peak splitting func-
tion of the chromatography software.
The method performance characteristics for all ana-
lytes detected in the high- and low-concentration plasma
pools are shown in Table 1. The minimum detectable
quantity (MDQ) was defined as a peak signal greater that
3 SD above the baseline noise. For all linoleic acid- and
linolenic acid-derived FAOOHs and FAOHs, which gave
sharp peaks and were well resolved, the MDQ was 5–7
pmol on column. The MDQs of the arachidonic acid-
derived FAOOHs and FAOHs were lower because they
absorb more strongly at 236 nm (⑀ ⫽ 27 000) than the
linoleates or linolenates (⑀ ⫽ 23 000). The MDQs of

Fig. 3. Comparison of HPLC scans of PUFA calibration mixtures before and after sample processing.
(A), PUFAs before processing at 215 nm; (B), FAOOH and FAOH before processing at 236 nm; (C), PUFAs after processing at 215 nm; (D), FAOOHs and FAOHs after
processing at 236 nm. See text for HPLC conditions. ⴱ indicates 5-HETE derived from alkaline hydrolysis of 5-HETE-ME [internal standard (I.S.)].
834 Browne and Armstrong: HPLC Analysis of Lipid Peroxides
Fig. 4. HPLC scans of FAOOH and FAOH compounds generated during incubation of human plasma with 100 mmol/L Cu2⫹-0.3 mL/L H2O2 at 37 °C.
15-HpETE and 15-HETE were 5.1 and 5.2 pmol in pure
solution but were three times higher in a mixture with
9-HpODE and 9-HODE, which if present in higher con-
centrations, mask these peaks. Each analyte produced a
linear response between the MDQ and 250 pmol, where
the peak width began to overlap adjacent peaks. Linear
regression of all calibration curves produced correlation
coefficients (r) ⬎0.99.
The CVs were ⬍5% for 9- and 13-HODE, which were
the only peaks detected in the low-concentration control
pool. The high-concentration pool was inherently more
complex, and the CVs for all peaks were between 8.4%
and 12%, with most of the variability arising from the
extraction process because replicate injections of a single
extract showed CVs ⬍1%. The moderate reproducibility is
also likely to be a function of the reduction of FAOOHs to
FAOHs during alkaline hydrolysis, which was evident in
the low recovery of the FAOOH compounds (87–93%).
Fig. 5. HPLC chromatogram of FAOOH and FAOH
calibrators.
Peaks: 1, 13S-hydroxy-9Z,11E,15Z-octadecatrienoic acid;
2, 13S-hydroxy-6Z,9Z,11E-octadecatrienoic acid; 3, 9S-
hydroxy-10E,12Z,15Z-octadecatrienoic acid; 4, 13-HODE;
5, 9-HODE; 6, 15-HETE; 7, HDHE; 8, 13-HpODE; 9,
9-HpODE; 10, 15-HpETE; 11, 12-HETE and HpDHE; 12,
12-HpETE and 5-HETE; 13, 5-HpETE; 14, internal standard
5-HETE-ME.
Discussion
PUFA oxidation is the most extensively studied compo-
nent of oxidative damage to biological systems, and many
studies have been conducted that suggest a role for lipid
peroxidation in the pathogenesis of many diseases (1 ).
Most analytical measurements of lipid peroxidation have
used nonspecific methods for total FAOOH breakdown
products and have not directly measured the primary
oxidized metabolites, i.e., FAOOH and FAOH. The
method described here represents the first direct isomeric
analysis of FAOOHs and FAOHs in human plasma.
Furthermore, seven distinct isomers of linoleic and ara-
chidonic acid FAOOHs and FAOHs were measured,
thereby defining the major plasma components of total
lipid peroxidation measurements.
Once isolated, plasma total lipids were completely ex-
tracted by the HIP extraction system. We have compared
HIP extraction with chloroform-methanol and ether-eth-
 Clinical Chemistry 46, No. 6, 2000 835

Table 1. Method performance characteristics for individual FAOOH and FAOH isomers.
CV, %

Intraassay Interassay
Retention Linear
Pead IDa time, min range, pmol MDQ, pmol Recovery, % Highb Lowc High Low
I3(S)HODE 22.15 5–250 6.7 98–102 9.6 3.4 11 5.0
9(S)HODE 24.95 5–250 6.2 97–102 8.6 3.2 10 4.7
15(S)HETE 25.64 15–250 15.3 98–103 10 ND 12 ND
13(S)HpODE 29.56 5–250 6.5 92–96 10 ND 12 ND
9(S)HpODE 31.25 5–250 6.7 93–97 9.0 ND 9.9 ND
15(S)HpETE 31.94 15–250 14.9 87–95 11 ND 12 ND
12(S)HETE 33.52 5–250 5.1 97–101 8.1 ND 8.3 ND
5(S)HETE 38.41 5–250 5.2 92–99 7.1 ND 8.1 ND
a
ID, identification; ND, not detected.
b
High-concentration pool.
c
Low-concentration pool.

anol methodologies. Ether-ethanol was found to be un-
suitable because of the epoxides that form in ether and
catalyze in vitro peroxidation of the sample. Chloroform-
methanol-extracted lipids are partitioned into the lower
phase, are difficult to aspirate, and contain considerable
water after drying. HIP extraction is favorable because no
water is present after drying and the degree of toxicity of
the solvents is much lower (19 ). The addition of butylated
hydroxytoluene and the removal of dissolved oxygen
from the solvents by sonication under reduced pressure
and helium sparging prevents any detectable autooxida-
tion of lipids during processing.
Total esterified PUFAs were completely hydrolyzed in
20 min by the alkaline hydrolysis procedure described,
which produces an 8 –12% conversion of FAOOHs to the
corresponding FAOHs. This reduction of FAOOH to
FAOH is higher than that reported by Thomas and
Jackson (16 ), who described a 6 – 8% conversion. This
reduction becomes less significant at lower concentrations
of base, but as the base concentration is decreased, the
amount of time required to completely hydrolyze the
sample increases, leading to the potential for autooxida-
tion.
The HPLC technique described can identify up to 13
distinct regioisomers of FAOOHs and FAOHs, constitut-
ing the most detailed HPLC separation of these com-
pounds reported to date. The method has a low detection
limit, detecting as little as 5 pmol, and is linear up to 250
pmol. Based on the procedure described, this converts to
a linear range of 0.033–1.6 ␮mol/L, which encompasses
the range of concentrations reported in normal human
plasma by Wilson et al. (12 ). Analysis of the method
performance characteristics indicated acceptable impreci-
sion, with CVs of 7–12%, and recoveries ⬎90%. It should
be noted that FAOOH recovery experiments require that
the FAOOH solution be added to the sample only after
the sample is precipitated with isopropyl alcohol. Direct
addition of FAOOH to intact plasma leads to a reduction
of 25–50% of the FAOOH to the corresponding FAOH,
depending on the time between addition of the FAOOH
solution and alcohol precipitation. This phenomenon is
most likely attributable to the activity of endogenous
peroxidases such as glutathione peroxidase. Precipitation
of the sample before the addition of exogenous FAOOH
inactivates these enzymes and prevents the reduction of
the FAOOH.
The in vitro oxidation of plasma demonstrated small
but detectable FAOH concentrations at baseline and mea-
surable increases in FAOOH and FAOH concentrations
with time. The linoleic acid oxidation products 13-HODE
and 9-HODE constituted the major peaks in all analyses.
9-HpODE was the only FAOOH visible at baseline but
was below the MDQ. 9-HpODE and 13-HpODE were
quantifiable by 60 min of oxidation. Teng and Smith (18 )
reported the inability to separate the cis-trans, trans-trans
regioisomeric HODEs (13ZE, 13EE, 9EZ, and 9EE), the
cis-trans HpODES (9EZ and 13ZE), and trans-trans
HpODEs (9EE and 13EE) under similar chromatographic
conditions. We have clearly demonstrated the separation
of the four cis-trans regioisomers 9-EZ-HODE, 13-ZE-
HODE, 9-EZ-HpODE, and 13-ZE-HpODE, with each cal-
ibrator giving a single resolve peak. This improvement
may be attributable to the longer analytical column (25
cm), slightly higher tetrahydrofuran concentration, five-
fold lower acetic acid concentration, or a combination of
all these factors. However, because of the lack of suitable
calibrators, we were unable to confirm the presence or
absence of all trans (EE) regioisomers in Cu2⫹-H2O2-
oxidized plasma. This is an acknowledged limitation of
the method.
The arachidonic acid products 15-HpETE, 12-HETE,
and 5-HETE were detectable by 60 min of oxidation.
These products continued to rise in proportion to each
other as oxidation continued. Nonenzymatic oxidation of
arachidonic acid can also give rise to 8-, 9-, and 11-
hydroperoxy and -hydroxy regioisomers in addition to
the ones discussed here. Based on the elution pattern of
15-, 12-, and 5-HpETE and -HETE, we observed that the
836 Browne and Armstrong: HPLC Analysis of Lipid Peroxides
retention time of the particular isomer shortened as the
oxygenation site moved farther from the carboxylic acid
end of the acyl chain. This was also true for the linoleic
and linolenic FAOOHs and FAOHs. The farther the
negative charges of the hydroperoxy or hydroxy group
are from the negative charge of the carboxylic acid group,
the more independent they become, giving an increas-
ingly polar character to the isomer and shortening the
retention time in reversed-phase systems. We hypothesize
that the regioisomers not characterized here, 11-, 9-, and
8-HETE and -HpETE, would elute in this order between
the 12- and 5-regioisomers, where there is a moderate
stretch of baseline. However, no peaks were detected in
this region on analysis of Cu2⫹-H2O2-oxidized plasma,
and these isomers were either not present or present in
amounts below the detection limit of the method.
There were no peroxidation products detected that
were attributable to linolenic or docosahexaenoic acid.
These findings result from the small percentage that these
PUFAs constitute of total plasma PUFAs. After 2 h of
oxidation, FAOOH and FAOH isomers of linoleic and
arachidonic acids account for ⬍1% of the corresponding
unoxidized PUFAs, which make up ⬎85% of total plasma
PUFAs. Unoxidized linolenic and docosahexaenoic acids
constitute ⬍5% of the total PUFAs, and their oxidation
products are well below the MDQ of this method. These
findings are in agreement with both Wilson et al. (12 ) and
Spitellar (6 ), who reported that C20 and C18 fatty acids,
especially linoleic acid, are the major precursors to plasma
FAOOHs and FAOHs.
In conclusion, the HPLC methodology described here is
sensitive and specific for the primary products of lipid
peroxidation and facilitates the use of these products as
markers of free radical-mediated lipid damage in the
peripheral circulation.
This work was supported, in part, by Grant S-96-7 from
the Mark Diamond Graduate Research Fund of the State
University of New York at Buffalo.
References
1. Mylonas C. Kouretas D. Lipid peroxidation and tissue damage
[Review]. In Vivo 1999;13:295–309.
2. Steinberg D. A critical look at the evidence for the oxidation of LDL
in atherogenesis [Review]. Atherosclerosis 1997;131(Suppl):
s5–7.
3. Porter N, Wolfe R, Weenan H. The free radical oxidation of
polyunsaturated lecithins. Lipids 1979;15:163–7.
4. Armstrong D, Browne R. The analysis of free radicals, lipid
peroxides, antioxidant enzymes and compounds related to oxida-
tive stress as applied to the clinical chemistry laboratory. In:
Armstrong D, ed. Free radicals in diagnostic medicine. Advances in
experimental and medical biology, Vol. 366. New York: Plenum
Press, 1994:43–58.
5. Jiang Z, Woollard A, Wolff S. Lipid hydroperoxide measurement by
oxidation of Fe2⫹ in the presence of xylenol orange. Comparison
with the TBA assay and iodometric method. Lipids 1991;26:
853– 6.
6. Spitellar G. Linoleic acid peroxidation-the dominant lipid peroxida-
tion process in low density lipoprotein-and its relationship to
chronic disease [Review]. Chem Phys Lipids 1998;95:105– 62.
7. Jahreis G. Cancer inhibiting fatty acids in milk and beef. Ernaehr-
Umsch 1997;44:168 –72.
8. Janero DR. Malondialdehyde and thiobarbituric acid-reactivity as
diagnostic indices of lipid peroxidation and peroxidative tissue
injury. Free Radic Biol Med 1990;13:341–90.
9. Jira W, Spitellar G, Schramm A. Increase in hydroxy fatty acids in
human low density lipoproteins with age. Chem Phys Lipids
1996;84:165–73.
10. Jira W, Spitellar G, Richter A. Increased levels of lipid oxidation
products in low density lipoproteins of patients suffering from
rheumatoid arthritis. Chem Phys Lipids 1997;87:81–9.
11. Jira W, Spitellar G. Strong increase in hydroxy fatty acids derived
from linoleic acid in human low density lipoproteins of atheroscle-
rotic patients. Chem Phys Lipids 1998;91:1–11.
12. Wilson R, Smith R, Wilson P, Shepard MJ, Riemersma RA.
Quantitative gas-chromatography-mass spectrometry isomer-spe-
cific measurement of hydroxy fatty acids in biological samples and
food as a marker of lipid peroxidation. Anal Biochem 1997;248:
76 – 85.
13. Zhang JR, Scherch HM. Hall ED. Direct measurement of lipid
hydroperoxides in iron-dependent spinal neuronal injury. J Neuro-
chem 1996;66:355– 61.
14. Yamamoto Y. Chemiluminescence-based high-performance liquid
chromatography assay of lipid hydroperoxides. Methods Enzymol
1994;233:319 –24.
15. Matsushita S, Terao K, Yamada K, Shibata SS. Specific detection
of lipid hydroperoxides using HPLC-EC method. In: Simic MG, ed.
Oxygen radicals in biology and medicine. New York: Plenum Press,
1988:163–74.
16. Thomas CE, Jackson RL. Lipid hydroperoxide involvement in
copper-dependent and independent oxidation of low density li-
poproteins. J Pharm Exp Ther 1991;256:1182– 8.
17. Porter N. Autooxidation of polyunsaturated fatty acids: initiation,
propagation, and product distribution. In: Vigo-Pelfrey C, ed.
Membrane lipid oxidation. Boca Raton, FL: CRC Press, 1990:33–
62.
18. Teng JI, Smith LL. High performance liquid chromatography of
linoleic acid hydroperoxides and their corresponding alcohol de-
rivatives. J Chromatogr 1985;350:445–51.
19. Hawkins DJ, Kuhn H, Petty EH, Brash AR. Resolution of enanti-
omers of hydroxyeicosatetraenoate derivatives by chiral phase
high-pressure liquid chromatography. Anal Biochem 1988;173:
456 – 62.
20. Browne R, Armstrong D. Separation of hydroxy and hydroperoxy
polyunsaturated fatty acids by high pressure liquid chromatogra-
phy. In: Armstrong D, ed. Methods in molecular biology, free
radicals and antioxidant protocols. Totowa, NJ: Humana Press,
1998:147–55.
21. Atsushi H, Radin NS. Lipid extraction of tissues with a low-toxicity
solvent. Anal Biochem 1978;90:420 – 6.
22. Gardner HW, Simpson TD. Hamberg M. Mechanism of linoleic acid
hydroperoxide reaction with alkali. Lipids 1996;31:1023– 8.