Clinical Chemistry 53:3 489 – 497 (2007)
Endocrimology and Metabolism
HPLC–Atmospheric Pressure Chemical Ionization MS/MS for Quantification of 15-F2t-Isoprostane in Human Urine and Plasma Manuel Haschke,1 Yan Ling Zhang,1 Christine Kahle,1 Jelena Klawitter,1 Magdalena Korecka,2 Leslie M. Shaw,2 and Uwe Christians1*
Background: Quantification of F2-Isoprostanes is considered a reliable index of the oxidative stress status in vivo and is valuable in the diagnosis and monitoring of a variety of diseases. Because of complex and lengthy sample preparation procedures, current chromatography/mass spectrometry and immunoassays are impractical for measuring larger numbers of samples. Thus, we developed and validated a semiautomated highthroughput HPLC tandem mass spectrometry assay for the quantification of F2-Isoprostane F2t in human urine and plasma. Methods: After protein precipitation (500 L methanol/ zinc sulfate added to 500 L plasma), samples were injected into the HPLC system and extracted online. The extracts were then back-flushed onto the analytical column and detected with an atmospheric pressure chemical ionization-triple quadrupole mass spectrometer monitoring the deprotonated molecular ions [M-H]– of 15-F2t-IsoP (m/z ⴝ 3533193) and the internal standard 15-F2t-IsoP-d4 (m/z ⴝ 3573197). Results: In human urine, the assay was linear from 0.025 to 80 g/L and in human plasma from 0.0025 to 80 g/L (r2>0.99). Interday accuracy and precision for concentrations above the lower limit of quantification were 80 samples/day, and has sufficient sensitivity for quantifying 15-F2t-IsoP concentrations in plasma and urine from healthy individuals. It is, thus, suitable for clinical routine monitoring and the analysis of samples from larger clinical trials. © 2007 American Association for Clinical Chemistry
F2-Isoprostanes (F2-IsoPs)3, the stable isomers of prostaglandin F2␣ (PGF2␣), are considered a reliable index of in vivo oxidative stress (1 ). Unlike the enzymatically formed prostaglandins, F2-IsoPs are predominately formed by free-radical– catalyzed peroxidation of arachidonic acid in situ in the phospholipid domain of cell membranes (2 ). After cleavage, presumably by phospholipases, F2-IsoPs circulate in free form in the plasma and are finally excreted in the urine (2, 3 ). The mechanism of formation of F2-IsoPs includes several steps. Initial arachidonoyl radicals undergo endocyclization and are reduced to 4 F-ring regioisomers. Each regioisomer can form 8 racemic diastereomers leading to 64 theoretically possible different F2-IsoPs (4 ). In addition, endoperoxide intermediates can rearrange in vivo to form other types of isoprostanes (5–7 ) or highly reactive ␥-ketoaldehydes (8 ). Other classes of IsoPs can be formed in vivo by free radical catalyzed peroxidation of fatty acids such as eicosapentanoic or docosahexanoic acid (9 –11 ).
1 Clinical Research and Development, Department of Anesthesiology, University of Colorado Health Sciences Center, Denver, Colorado. 2 Department of Pathology and Laboratory Medicine, University of Pennsylvania Medical Center, Philadelphia, PA. *Address correspondence to this author at: Clinical Research and Development, Department of Anesthesiology, University of Colorado Health Sciences Center, 4200 East Ninth Ave., Room UH-2122, Campus Box B-113, Denver, Colorado 80262. Fax 303-315-1858; e-mail [email protected]
Received August 28, 2006; accepted January 3, 2007. Previously published online at DOI: 10.1373/clinchem.2006.078972
3 Nonstandard abbreviations: F2-IsoPs, F2-Isoprostanes; MS, mass spectrometry; MS/MS, tandem mass spectrometry; PGF2␣, prostaglandin F2␣; LPO, lipid peroxidation; SPE, solid-phase extraction; TLC, layer chromatography; IS, internal standard; LLOQ, Lower limit of quantification; ESI, electrospray ionization; and APCI, atmospheric pressure chemical ionization.
Haschke et al.: HPLC–Atmospheric Pressure Chemical Ionization MS/MS
In vitro several IsoPs including 15-F2t-IsoP (8-isoPGF2␣/iPF2␣-III; for nomenclatures see Supplemental Data Table 1), the most extensively studied F2-IsoP isomer, possess biological activity such as nonspecific vasoconstriction (12–15 ), bronchoconstriction (16, 17 ), and modulation of platelet function (13, 18, 19 ). It remains unresolved whether the effects observed in vitro at high concentrations are also relevant in vivo at physiological concentrations, which are considerably lower. Since F2-IsoPs were first characterized in humans (20 ), increasing evidence has indicated that they provide a specific and sensitive assessment of lipid peroxidation (LPO) in vivo. Although F2-IsoPs are not a major product of LPO, their characteristics favor F2-IsoPs as specific markers of LPO. F2-IsoPs have been measured with gas chromatography/mass spectrometry (GC/MS) (20 –22 ), GC-tandem MS (MS/MS) (23 ), liquid chromatography MS (LC/MS) (3, 24 –27 ), and immunoassays. GC-MS assays require extensive sample preparation, including solid-phase extraction (SPE), thin layer chromatography (TLC), and derivatization reactions to protect the polar groups (28 ). Commercial immunoassays require extensive sample purification (SPE/TLC) and are susceptible to cross-reactivity because F2-IsoPs and their metabolites share a 1,3-syn-hydroxycyclopentane ring, which is the major determinant of antigenicity (28 ). Comparison of F2-IsoP measurement in urine by immunoassay and GC-MS has revealed considerable inconsistencies (29 ) attributable in part to the fact that the peak observed in many GC-MS assays after standard SPE and TLC sample purification is comprised of more than 1 F2-IsoP isomer (28 ). All LC-MS methods published to date require SPE for sample purification (3, 24 –27 ). We report here the development and validation of an HPLC-MS/MS assay for the quantification of 15-F2t-IsoP with automated online extraction. Manual sample preparation is limited to a simple protein precipitation step.
Materials and Methods materials HPLC grade methanol and water, formic acid 88%, and zinc sulfate were purchased from Fisher Scientific. 15-F2tIsoP (8-iso-PGF2␣/iPF2␣-III), 15-F2t-IsoP-d4 (3,3,4,4-d4-15F2t-IsoP, IS, ⱖ98% d4), 15(R)-F2t-IsoP, 9␤,11␣-15-F2t-IsoP, prostaglandin F2␣ (PGF2␣), 15(R)-PGF2␣, 5-trans-PGF2␣, 9␣,11␤-PGF2, 9␤,11␣-PGF2, and 5-trans-9␤,11␣-PGF2 (Supplemental Data Fig. 1) were purchased from Cayman Chemical.
calibrators and quality control samples Stock solutions (1 g/L) of all compounds were prepared in methanol and stored in polypropylene screw-top tubes at – 80 °C. Working solutions for quality control and calibration were prepared by dilution of the stock solutions with methanol and were also stored in polypropylene tubes. Plasma and urine samples used for assay
development and validation were obtained from healthy volunteers and pooled human plasma from the local blood bank. Blood samples were collected into EDTA tubes containing no other additives. Samples from healthy volunteers and patients were collected during various clinical trials. All protocols were approved by the local institutional review boards at both institutions. All study participants provided written informed consents and studies were conducted in full compliance with the principals of good clinical practice as set forth in the International Conference on Harmonization Harmonized Tripartite Guidelines (version April 1996), the US Code of Federal Regulations (21 CFR 50, 54, 56, and 312) and the principles stated in the Declaration of Helsinki (version 11, October 2000). The use of blood bank samples for assay validation and quality control was institutional review board exempt (Colorado Multiinstitutional Review Board).
sample preparation and protein precipitation The protein precipitation solution (methanol/0.2 mol/L ZnSO4, 7:3, v/v) contained the internal standard (IS) 15-F2t-IsoP-d4 at 4 g/L; 500 L protein precipitation solution was added to an equal volume of plasma or urine. After vortex-mixing for 1 min and centrifugation (13 000g, 10 min, 4 °C), the supernatant was transferred into an HPLC vial and placed in the autosampler at 4 °C. Creatinine concentrations in urine were measured with a Beckman Synchron LX® system using the Jaffe rate method.
automated online extraction and hplc conditions The HPLC system consisted of 2 G1312A binary pumps, 2 G1322A vacuum degassers, a G1329A/G1330A thermostated autosampler and a G1316A thermostated column compartment (all Agilent 1100 series) with an integrated 6-port Rheodyne column switching valve, as shown in Supplemental Data Fig. 2. Sciex API4000 or API5000 triple quadrupole mass spectrometers were used as the detector. The HPLC system and the mass spectrometers were controlled by the Analyst software (Applied Biosystems). We injected 500 L of the samples onto a 4.6 ⫻ 12.5–mm Eclipse XDB-C8 5-m extraction column (Agilent Technologies) with a mobile phase of 30% methanol and 70% 0.1%-formic acid, flow rate 5 mL/min. After 1 min, the switching valve was activated and the analytes back-flushed from the extraction column onto a Phenomenex Synergi Hydro-RP 80Å, 3.0 ⫻ 250 –mm, 4-m column filled with polar endcapped-C18 material. We used methanol and 0.1% formic acid, flow rate 0.6 mL/min, with the following gradient: 0 –1 min 63% methanol; 1.1–10 min 63%–98%. The analytical column was kept at 98% for 1 min and then reequilibrated to the starting conditions. After 9 min the column-switching valve was switched back into the extraction position and the extraction column reequilibrated to the starting conditions; the total run
Clinical Chemistry 53, No. 3, 2007
time between injections of 13 min. Both columns were maintained at 60 °C.
ms/ms analysis The HPLC system was interfaced with the mass spectrometer with an atmospheric pressure chemical ionization (APCI) source. Nitrogen (purity: 99.999%) was used as collision-activated dissociation gas. The mass spectrometer was run in the negative multiple reaction monitoring (MRM) mode. The declustering potential was set to ⫺70 V, the entrance potential (EP) to ⫺5V, the interface to 400 °C, and the collision energy to 36 eV. The first quadrupole (Q1) was set to select the deprotonated molecular ions [M-H]– of 15-F2t-IsoP (m/z⫽ 353) and 15-F2tIsoP-d4 (IS, m/z⫽ 357), and the 3rd quadrupole (Q3) to select the characteristic product ions of 15-F2t-IsoP (m/z 193) and 15-F2t-IsoP-d4 (IS, m/z 197). Peak area ratios obtained from MRM mode of the mass transitions for 15-F2t-IsoP (m/z 353 3 193) and 15-F2t-IsoP-d4 (IS, m/z 357 3 197) were used for quantification.
calibration and quality control samples Calibration and QC samples were prepared by enriching plasma or urine with 15-F2t-IsoP. To account for endogenous 15-F2t-IsoP, the ratio of endogenous 15-F2t-IsoP peak area divided by the IS peak area of unenriched matrix was subtracted from area ratios of enriched samples (corrected analyte area/IS area ratio). The calibration curves (0, 0.0005, 0.001, 0.0025, 0.005, 0.01, 0.025, 0.04, 0.1, 1, 10, 20, 40, 80 g/L, n ⫽ 6 per concentration) were constructed with nonweighted linear regression. Concentrations in plasma were calculated as nanograms per liter, in urine as nanograms per gram creatinine. QC samples were prepared in human urine or plasma at concentrations of 0.04 g/L, 0.1 g/L, 20 g/L, and 40 g/L 15-F2t-IsoP.
matrix effects/ion suppression To detect ionization efficiency changes attributable to coeluting matrix substances, we tested urine and plasma from 10 different healthy volunteers and aliquots of 2 different pooled human plasma samples. After protein precipitation, samples were extracted online and backflushed onto the analytical column as described above. We infused 15-F2t-IsoP (10 mg/L dissolved in H2O/ methanol, 7:3, v/v) postcolumn via a tee at 10 L/min using a syringe pump. The extent of ion suppression was established by monitoring the intensity of the ion currents in MRM mode (m/z ⫽ 3533 m/z ⫽ 193) at the retention times of analyte and IS (30 ).
validation procedures The assay was completely validated according to the FDA Center for Drug Evaluation and Research guidelines for bioanalytical method validation (31 ). Acceptance criteria. The performance of the assay was considered acceptable if the precision (CV%) at each
concentration was ⱕ15% for intraday and interday variability, the expected result was within 15% of the assigned concentration for both intra- and interday variability, and the calibration curve showed a correlation coefficient r2 of 0.99 or better. Lower limit of quantification. The lower limit of quantification (LLOQ) was determined as the lowest concentration of the calibration curve consistently yielding results within 20% of the nominal concentration and a precision ⱕ20%. Precision and recovery. The intraday precision, interday precision, and recovery were determined by analysis of QC samples containing 0.04 g/L, 0.1 g/L, 20 g/L, and 40 g/L 15-F2t-IsoP (n ⫽ 6/concentration). Samples were extracted and analyzed on 3 different days (n ⫽ 6/ concentration and day). Extraction Recoveries. The extraction recoveries were determined by comparing the signals for 15-F2t-IsoP obtained after extraction of QC samples (n ⫽ 6) with the signals of extracted matrix enriched with the respective concentrations of 15-F2t-IsoP after the extraction procedure. Stability studies. We tested stability in urine and plasma during 3 freeze-thaw cycles at 0.04, 0.1, 20, and 40 g/L (n ⫽ 3). Samples were kept frozen at ⫺80 °C and thawed at room temperature. Within-batch stability in fresh plasma was tested at 0.1 g/L. Samples were kept at ⫺80 °C, ⫺20 °C, ⫹4 °C, or at room temperature. After 4, 8, 12, 24, 48, and 168 h samples were extracted, analyzed, and compared with freshly prepared samples. Dilution integrity. Dilution integrity was established with freshly prepared urine and plasma samples enriched with 40 g/L 15-F2t-IsoP. Dilutions (1:1, 1:10, 1:100, and 1:400, n ⫽ 6) were made with fresh unenriched urine or plasma. Deviations from the nominal concentrations after dilution were calculated. Carryover effect. Carryover was assessed by analyzing urine or plasma samples enriched with 15-F2t-IsoP at the upper limit of quantification (80 g/L, n ⫽ 3) followed by blank methanol samples. Cross validation. Random spot urine samples obtained from 30 healthy volunteers were divided into aliquots, shipped on dry ice, and stored at ⫺80 °C until analysis at the University of Colorado Health Sciences Center and at the University of Pennsylvania Medical Center. Investigators at both laboratories were blinded to the concentrations in the samples. Samples were analyzed in duplicate and quantified with an abbreviated standard curve (0, 0.001, 0.0025, 0.005, 0.01, 0.025, 0.04, 0.1, 0.25, 0.5, and 1 ng/L, n ⫽ 3/batch). Correlations were analyzed by linear regression (SPSS 14.0).
Haschke et al.: HPLC–Atmospheric Pressure Chemical Ionization MS/MS
Fig. 1. MS/MS spectrum of 15-F2tisoprostane. cps, counts per second. amu, atomic mass units.
Results MS and MS/MS spectra were recorded after direct infusion of 15-F2t-IsoP and 15-F2t-IsoP-d4 into the source of the mass spectrometer via a syringe pump. Under APCI conditions in negative ion mode the deprotonated molecular ions ([M-H]⫺) of 15-F2t-IsoP at m/z ⫽ 353 and the IS 15-F2t-IsoP-d4 at m/z ⫽ 357 were generated. For 15-F2tIsoP the product ion at m/z ⫽193 was the most abundant ion (Fig. 1), and thus the transition m/z ⫽ 3533m/z ⫽ 193 was selected for quantification. The most abundant product ion detected for 15-F2t-IsoP-d4 was at m/z ⫽ 197. Based on this the transition m/z ⫽ 3573m/z ⫽ 197 was selected for quantification of the IS. No ion suppression was detected for most of the urine and for all the tested plasma samples (Figs. 2A and 2B). However, 2 of 10 urine samples showed a loss of signal
Fig. 2. Matrix effects. Extracted ion chromatograms of ion suppression experiments in APCI mode with urine (A) and plasma (B) from healthy individuals. The bar marks the retention time range of 15-F2t-isoprostane and the IS 15-F2t-isoprostane-d4. Extracted ion chromatograms of ion suppression experiment with urine sample that cause loss of signal intensity by a factor of 3 in APCI mode (C) and by a factor of 10 in ESI mode (D).
intensity at the retention times of 15-F2t-IsoP and IS by a factor of ⬃3 (Fig. 2C). 15-F2t-IsoP eluted with a mean (SD) retention time of 8.30 (0.12) min and the IS with a retention time of 8.27 (0.13) min. In urine, the LLOQ was 0.025 g/L (with API5000) and linearity was up to 80 g/L (0.231x ⫹ 0.011, r2 ⫽ 0.9971). With the API4000 detector, the assay was linear from 0.04 (LLOQ) to 80 g/L (y ⫽ 0.204x⫺0.098, r2 ⫽ 0.9995, Fig. 3A). In plasma, the LLOQ was 0.0025 g/L (with API5000) and linearity up to 80 g/L (y ⫽ 0.976x – 0.002, r2 ⫽ 0.9912). With the API4000, the linear range was 0.01 (LLOQ) to 80 g/L (y ⫽ 1.180x – 0.075, r2 ⫽ 0.9996, Fig. 3B). At the tested concentrations of 0.04, 0.1, 20, and 40 g/L, intraday recoveries in urine and plasma were 93.8%–106.4% and intraday precisions ⱕ6.8% (Table 1A).
Clinical Chemistry 53, No. 3, 2007
Interday recoveries were 91.1%–106.2%, and interday precisions ⱕ8% (Table 1B). The mean (SD) absolute recovery of 15-F2t-IsoP after protein precipitation of urine and plasma was 96.7% (8.1%) and 100.2% (9.5%), respectively. 15-F2t-IsoP was stable in urine and plasma for at least 3 freeze-thaw cycles (Figs. 4A and 6B). Samples were considered stable if they contained analyte concentrations not statistically different from the corresponding baseline samples as tested by ANOVA in combinations with Dunnett’s post hoc test. In addition, the mean concentrations had to be within 15% of baseline. In fresh human plasma, the concentration of 15-F2t-IsoP was stable for at least 168 h when stored at ⫹4 °C or below. At room
Fig. 3. Linearity of the assay in human urine (A) and human plasma (B). Data is presented as mean (SD); n ⫽ 18 for each concentration.
Table 1. A. Intraday recovery and imprecision. Intraday recovery, %
Intraday precision, %
Native 0.04 g/L 0.1 g/L 20 g/L 40 g/L
N/A 97.9 103.0 104.7 106.4
N/A 102.1 100.7 93.9 93.8
2.2 6.7 4.5 1.7 1.9
5.0 4.0 2.1 3.0 3.0
B. Interday recovery and imprecision. Interday recovery, %
Interday precision, %
Native 0.04 g/L 0.1 g/L 20 g/L 40 g/L
N/A 93.9 98.6 105.3 106.2
N/A 99.6 100.8 93.7 91.1
5.4 6.0 6.4 1.3 1.5
7.8 5.6 5.1 2.6 2.1
N/A: not applicable. The native plasma and urine samples were pools from 5 healthy individuals. The plasma sample contained 0.012 g/L and the urine sample 0.2 g/g creatinine 15-F2t-IsoP.
Fig. 4. Freeze-thaw stability of 15-F2t-isoprostane in human urine (A) and human plasma (B). Data are presented as mean (SD); n ⫽ 3 per concentration and cycle.
Haschke et al.: HPLC–Atmospheric Pressure Chemical Ionization MS/MS
temperature, the concentration of 15-F2t-IsoP was stable for at least 48 h; after 168 h the measured concentration increased to 189.2% (26.6%) of the nominal concentration, mainly because of to autooxidation of arachidonic acid resulting in in vitro formation of 15-F2t-IsoP. To confirm stability of endogenous 15-F2t-IsoP in nonenriched samples from healthy individuals, we tested 6 plasma and 6 urine samples, stored for 24 h at 4 °C or room temperature, and during 3 freeze-thaw cycles. The 15-F2t-IsoP concentrations in plasma relative to immediately analyzed controls were (all n ⫽ 6): after 24 h at 4 °C, 95.0% (1.9%); after 24 h at room temperature, 89.4% (8.0%); after 1 freeze-thaw cycle, 95.3% (4.6%); after 2 freeze-thaw cycles, 96.8% (4.0%); and after 3 freeze-thaw cycles, 106.2% (4.0%). The 15-F2t-IsoP concentrations in urine relative to immediately analyzed controls were (all n ⫽ 6): after 24 h at 4 °C, 102.3% (3.4%); after 24 h at room temperature, 102.6% (3.7%); after 1 freeze-thaw cycle, 95.9% (2.3%); after 2 freeze-thaw cycles, 97.0% (4.5%); and after 3 freeze-thaw cycles, (117.7%) (2.1%). The results showed that except after 3 freeze-thaw cycles in urine, 15-F2t-IsoP was stable in native plasma and urine samples at the low concentrations found in healthy persons. Recovery was not affected by dilution of urine or plasma samples. The mean recoveries for dilutions of 1:1,
1:10, 1:100, and 1:400 of samples enriched with 40 g/L in urine and plasma were between 100.7%–99.9% (urine) and 100.2%–104.1% (plasma). Carryover was not observed after analysis of samples with an 15-F2t-IsoP concentration of 80 g/L Analysis of urine and plasma samples from healthy subjects showed the endogenous 15-F2t-IsoP peak clearly separated from adjacent peaks (Figs. 5A and 5B). Neighboring peaks were identified by enriching urine or plasma from healthy persons with commercially available isomers of 15-F2tIsoP and PGF2␣. All of the tested isomers showed the ion transition m/z ⫽ 3533193. With the exception of the 2 9␤-isomers of 15-F2t-IsoP and PGF2␣ that do not occur naturally (9␤,11␣-15-F2t-IsoP and 9␤,11␣-PGF2), all isomers were chromatographically separated from 15-F2t-IsoP (Figs. 5C and 5D). Cross-validation of urine samples from 30 healthy volunteers showed good correlation of results obtained by the 2 laboratories (r2 ⫽ 0.95, P ⬍0.001, Supplemental Data Fig. 3). Concentrations of 15-F2t-IsoP in urine of healthy persons (n ⫽ 16) ranged from 55 to 348 ng/g creatinine. In plasma from the same healthy persons, free 15-F2t-IsoP was detectable in all the samples (Fig. 5B), with concentrations ranging between 3–25 ng/L (n ⫽ 16).
Fig. 5. Representative ion chromatograms of extracted urine and plasma samples. Representative ion chromatograms of urine (A) and plasma from healthy subjects (B) with endogenous 15-F2t-isoprostane peak (arrow). Enlarged section (7 to 10-min) of representative ion chromatograms of urine (C) and plasma from healthy persons (D). Peak 1: 15(R)-F2t-isoprostane, peak 2: 15-F2t-isoprostane, peak 3: 9␣,11␤-prostaglandin F2, peak 4: unidentified, peak 5: 15(R)-prostaglandin F2␣, peak 6: 5-trans prostaglandin F2␣, peak 7: prostaglandin F2␣, peaks 8 and 9: unidentified. cps, counts per second.
Clinical Chemistry 53, No. 3, 2007
Discussion The automated column-switching technique and on-line sample extraction reduced manual sample preparation to a simple protein precipitation step, eliminating the need for laborious sample purification and avoiding potential sources of analytical imprecision in existing methods (3, 23–27, 32 ). Increased use of MS detection has led to a revival of the use of column-switching techniques for automated online sample preparation, used for many years for HPLC analyses (33 ). To date, SPE has been used for sample preparation in all published LC-MS methods for the analysis of 15-F2t-IsoP (3, 23–27 ). Only one method employed automated SPE, but the extraction was carried out offline with an automated SPE workstation (27 ). Matrix effects should be assessed during development and validation of an analytical method (34 ). Among the atmospheric pressure ionization techniques, electrospray ionization (ESI) has been reported to be particularly susceptible to matrix effects (35, 36 ). When we evaluated ESI during method development, all tested urine samples caused severe ion suppression, resulting in a drop of signal intensity by an order of magnitude around the retention of our analytes of interest (Fig. 2D). After the ionization mode was switched to APCI, ion suppression was not detected in any of the plasma samples and was detected in only 2 of 10 tested urine samples. These findings are in line with observations by King et al. (35 ), who reported that plasma samples prepared by acetonitrile protein precipitation caused much more severe ion suppression in ESI than APCI. Interestingly, for 15-F2tIsoP, ion suppression was observed for urine but not plasma, independent of the ionization mode. The fact that matrix effects can alter ionization efficiency in both ESI and APCI highlights the importance of the use of a stable isotope-labeled analyte as IS (37 ). All previous LC-MS methods for 15-F2t-IsoP used negative ion mode ESI, but none were tested for matrix effects (3, 24 –27 ). Even if a stable isotope-labeled analyte is used as the IS (37 ), the loss of signal intensity will negatively affect the LLOQ. For compounds such as 15-F2t-IsoP, which normally occur at very low concentrations, and urine samples that may be subject to varying degrees of dilution, the LLOQ may not be low enough to cover the lower concentration range of the analyte. Method development was mainly carried out using an API4000 mass spectrometer. During our study, an API5000 mass spectrometer became available. Compared with the API4000, the API5000 mass spectrometer improved the LLOQ by a factor of 1.6 in urine and 4.0 in plasma. This improvement is critical for plasma, for which the concentrations in healthy persons were 3–25 ng/L. At these concentrations the API4000 mass spectrometer was not sufficiently sensitive. For IsoPs, there are 4 regioisomers, each theoretically comprised of 16 isomers (4 ). The regioisomers differ by the length of the alkyl side chains and/or by the position
of the 3rd hydroxyl group on the side chain and can, therefore, be distinguished by monitoring typical daughter ions generated in MRM mode. Within a given group of regioisomers (e.g., 15-F2-IsoP), however, all the isomers show the same fragmentation pattern and thus must be separated chromatographically to allow for specific quantification. Only 3 of the 16 theoretically possible isomers in the 15-F2-IsoP group are commercially available; 15-F2tIsoP, 15(R)-F2t-IsoP, and 9␤, 11␣-15-F2t-IsoP were tested together with PGF2␣ and 5 PGF2 isomers for chromatographic separation (Fig. 5). The only 2 compounds that coeluted with 15-F2t-IsoP were the 2 9␤-isomers of 15-F2tIsoP and PGF2␣. 9␤, 11 ␣-compounds cannot occur naturally (5–7, 38 –39 ). However, theoretical considerations suggest that half of the 15-F2t-IsoP formed will be 9␤, 11␤ and half will be 9␣, 11␣. It must be noted that our study was not designed to determine reference intervals. Our intention was to demonstrate that the 15-F2t-IsoP concentrations found in urine and plasma of healthy persons were within the range of the reliable response of our assay. It is not surprising, however, that in this situation in which multiple isomers must be separated chromatographically for reliable quantification, the normal values for 15-F2-IsoP in biological samples reported in the literature differ widely. Reported mean (SD) normal values measured by LC-MS/MS in urine of healthy volunteers range from 250 (220) ng/g creatinine (27 ) to 1110 (450) ng/g creatinine (3 ). The method by Liang et al. (27 ), which reported the lowest normal values, also achieved the best chromatographic separation of 4 isomers [15(R)F2t-IsoP, 15-F2t-IsoP, 15(R)-PGF2␣, PGF2␣] and one unknown peak. The above-mentioned isomers were separated in the same sequence by our method; however, it allowed for additional separation of 9␣, 11␤-PGF2, which eluted between the 15-F2t-IsoP and 15(R)-PGF2␣ peaks. The naturally occurring PGF2 isomer can be identified in urine and plasma (Fig. 5, C and D, peak 3). Separation of 9␣, 11␤-PGF2 is crucial because it elutes shortly after the 15-F2t-IsoP peak, but is not an isoprostane. The superior chromatographic separation achieved by our method most likely explains why our measured values of 15-F2tIsoP in urine of healthy volunteers are lower than the values reported in the literature for comparable analytical methodologies. The selection of the analytical column during method development proved to be critical to achieve the desired chromatographic separation.
Support for this study was provided by the National Institutes of Health grants R01 HL071805, R01 DK065094, P30 DK048520 (U.C.), AG-024904 (M.K. and L.M.S.) and by the Merck Company Foundation.
Haschke et al.: HPLC–Atmospheric Pressure Chemical Ionization MS/MS
References 1. Roberts LJ, Morrow JD. Measurement of F-2-isoprostanes as an index of oxidative stress in vivo. Free Radical Bio Med 2000;28: 505–13. 2. Morrow JD, Awad JA, Boss HJ, Blair IA, Roberts LJ. Non-cyclooxygenase-derived prostanoids (F2-isoprostanes) are formed in situ on phospholipids. Proc Natl Acad Sci U S A 1992;89:10721–5. 3. Li HW, Lawson JA, Reilly M, Adiyaman M, Hwang SW, Rokach J, et al. Quantitative high performance liquid chromatography tandem mass spectrometric analysis of the four classes of F-2-isoprostanes in human urine. Proc Natl Acad Sci U S A 1999;96:13381– 6. 4. Morrow JD, Harris TM, Roberts LJ. Noncyclooxygenase oxidative formation of a series of novel prostaglandins: analytical ramifications for measurement of eicosanoids. Anal Biochem 1990;184: 1–10. 5. Morrow JD, Minton TA, Mukundan CR, Campbell MD, Zackert WE, Daniel VC, et al. Free radical-induced generation of isoprostanes in-vivo: evidence for the Formation of D-Ring and E-Ring Isoprostanes. J Biol Chem 1994;269:4317–26. 6. Morrow JD, Awad JA, Wu AP, Zackert WE, Daniel VC, Roberts LJ. Nonenzymatic free radical-catalyzed generation of thromboxanelike compounds (isothromboxanes) in vivo. J Biol Chem 1996; 271:23185–90. 7. Chen Y, Morrow JD, Roberts LJ. Formation of reactive cyclopentenone compounds in vivo as products of the isoprostane pathway. J Biol Chem 1999;274:10863– 8. 8. Brame CJ, Salomon RG, Morrow JD, Roberts LJ. Identification of extremely reactive gamma-ketoaldehydes (isolevuglandins) as products of the isoprostane pathway and characterization of their lysyl protein adducts. J Biol Chem 1999;274:13139 – 46. 9. Nourooz-Zadeh J, Halliwell B, Anggard EE. Formation of a novel class of F-3-isoprostanes during peroxidation of eicosapentaenoic acid (EPA). Adv Exp Med Biol 1997;433:185– 8. 10. Nourooz-Zadeh J, Liu EHC, Anggard EE, Halliwell B. F-4-isoprostanes: A novel class of prostanoids formed during peroxidation of docosahexaenoic acid (DHA). Biochem Biophys Res Commun 1998;242:338 – 44. 11. Roberts LJ, Montine TJ, Markesbery WR, Tapper AR, Hardy P, Chemtob S, et al. Formation of isoprostane-like compounds (neuroprostanes) in vivo from docosahexaenoic acid. J Biol Chem 1998;273:13605–12. 12. Takahashi K, Nammour TM, Fukunaga M, Ebert J, Morrow JD, Roberts LJ, et al. Glomerular actions of a free radical-generated novel prostaglandin, 8-epi-prostaglandin-F2-␣, in the rat: evidence for interaction with thromboxane-A2 receptors. J Clin Invest 1992; 90:136 – 41. 13. Morrow JD, Minton TA, Roberts LJ. The F2-isoprostane, 8-epiprostaglandin-F2-␣, a potent agonist of the vascular thromboxane endoperoxide receptor, is a platelet thromboxane endoperoxide receptor antagonist. Prostaglandins 1992;44:155– 63. 14. Lahaie I, Hardy P, Hou X, Hassessian H, Asselin P, Lachapelle P, et al. A novel mechanism for vasoconstrictor action of 8-isoprostaglandin F-2 ␣ on retinal vessels. Am J Physiol-Reg I 1998;43: R1406 –R16. 15. Audoly LP, Rocca B, Fabre JE, Koller BH, Thomas D, Loeb AL, et al. Cardiovascular responses to the isoprostanes iPF(2 ␣)-III and iPE(2)-III are mediated via the thromboxane A(2) receptor in vivo. Circulation 2000;101:2833– 40. 16. Kang KH, Morrow JD, Roberts LJ, Newman JH, Banerjee M. Airway and vascular effects of 8-epi-prostaglandin-F2-␣ in isolated perfused rat lung. J Appl Physiol 1993;74:460 –5. 17. Banerjee M, Kang KH, Morrow JD, Roberts LJ, Newman JH. Effects of a novel prostaglandin, 8-epi-PgF2-␣, in rabbit lung in situ. Am J Physiol 1992;263:H660 –H3.
18. Pratico D, Smyth EM, Violi F, FitzGerald GA. Local amplification of platelet function by 8-epi prostaglandin F-2 ␣ is not mediated by thromboxane receptor isoforms. J Biol Chem 1996;271:14916 – 24. 19. Minuz P, Andrioli G, Degan M, Gaino S, Ortolani R, Tommasoli R, et al. The F-2-isoprostane 8-epiprostaglandin F-2 ␣ increases platelet adhesion and reduces the antiadhesive and antiaggregatory effects of NO. Arterioscl Throm Vas 1998;18:1248 –56. 20. Morrow JD, Hill KE, Burk RF, Nammour TM, Badr KF, Roberts LJ. A Series of prostaglandin-F2-like compounds are produced in vivo in humans by a noncyclooxygenase, free radical-catalyzed mechanism. Proc Nat Acad Sci U S A 1990;87:9383–7. 21. Gopaul NK, Anggard EE, Mallet AI, Betteridge DJ, Wolff SP, Nouroozzadeh J. Plasma 8-epi-Pgf(2-␣) levels are elevated in individuals with non-insulin-dependent diabetes mellitus. Febs Letters 1995;368:225–9. 22. Bachi A, Zuccato E, Baraldi M, Fanelli R, Chiabrando C. Measurement of urinary 8-Epi-prostaglandin F2alpha, a novel index of lipid peroxidation in vivo, by immunoaffinity extraction/gas chromatography-mass spectrometry: basal levels in smokers and nonsmokers. Free Rad Biol Med 1996;20:619 –24. 23. Tsikas D, Schwedhelm E, Suchy MT, Niemann J, Gutzki FM, Erpenbeck VJ, et al. Divergence in urinary 8-iso-PGF(2x) (iPF(2x)-III 15-F-2t-IsoP) levels from gas chromatography tandem mass spectrometry quantification after thin-layer chromatography and immunoaffinity column chromatography reveals heterogeneity of 8-iso-PGF(2x):-possible methodological, mechanistic and clinical implications. J Chromatogr B 2003;794:237–55. 24. Waugh RJ, Murphy RC. Mass spectrometric analysis of four regioisomers of F-2-isoprostanes formed by free radical oxidation of arachidonic acid. J Am Soc Mass Spectr 1996;7:490 –9. 25. Ohashi N, Yoshikawa M. Rapid and sensitive quantification of 8-isoprostaglandin F-2 ␣ in human plasma and urine by liquid chromatography-electrospray ionization mass spectrometry. J Chromatogr B 2000;746:17–24. 26. Murai Y, Hishinuma T, Suzuki N, Satoh J, Toyota T, Mizugaki M. Determination of urinary 8-epi-prostaglandin F-2 ␣ using liquid chromatography-tandem mass spectrometry: increased excretion in diabetics. Prostaglandins Other Lipid Mediat 2000;62:173– 81. 27. Liang YL, Wei P, Duke RW, Reaven PD, Harman SM, Cutler RG, et al. Quantification of 8-iso-prostaglandin-F-2 ␣ and 2,3-dinor-8isoprostaglandin-F-2 ␣ in human urine using liquid chromatography-tandem mass spectrometry. Free Radical Bio Med 2003;34: 409 –18. 28. Lawson JA, Rokach J, FitzGerald GA. Isoprostanes: formation, analysis and use as indices of lipid peroxidation in vivo. J Biol Chem 1999;274:24441– 4. 29. Proudfoot J, Barden A, Mori TA, Burke V, Croft KD, Beilin LJ, et al. Measurement of urinary F-2,-isoprostanes as markers of in vivo lipid peroxidation - A comparison of enzyme immunoassay with gas chromatography/mass spectrometry. Anal Biochem 1999; 272:209 –15. 30. Muller C, Schafer P, Stortzel M, Vogt S, Weinmann W. Ion suppression effects in liquid chromatography-electrospray-ionisation transport-region collision induced dissociation mass spectrometry with different serum extraction methods for systematic toxicological analysis with mass spectra libraries. J Chromatogr B 2002;773:47–52. 31. Center for Drug Evaluation and Research. Guidance for Industry. Bioanalytical Method Validation. US Food and Drug Administration 2001; http://www.fda.gov/cder/Guidance/4252fnl.pdf (accessed January 19, 2007). 32. Pratico D, Barry OP, Lawson JA, Adiyaman M, Hwang SW, Khanapure SP, et al. IPF2 ␣-I: An index of lipid peroxidation in humans. Proc Natl Acad Sci U S A 1998;95:3449 –54.
Clinical Chemistry 53, No. 3, 2007
33. Huber R, Zech K. In: Selective Sample Handling and Detection in High- Performance Liquid Chromatography. R.W. Frei and K. Zech, eds. J. Chromatogr. Library, Elsevier, Vol. 39A, Amsterdam, 1988, pp 81–144. 34. Annesley TM. Ion suppression in mass spectrometry. Clin Chem 2003;49:1041– 4. 35. King R, Bonfiglio R, Fernandez-Metzler C, Miller-Stein C, Olah T. Mechanistic investigation of ionization suppression in electrospray ionization. J Am Soc Mass Spectr 2000;11:942–50. 36. Mei H, Hsieh YS, Nardo C, Xu XY, Wang SY, Ng K, et al. Investigation of matrix effects in bioanalytical high-performance liquid chromatography/tandem mass spectrometric assays: appli-
cation to drug discovery. Rapid Commun Mass Spectrom 2003; 17:97–103. 37. Schuhmacher J, Zimmer D, Tesche F, Pickard V. Matrix effects during analysis of plasma samples by electrospray and atmospheric pressure chemical ionization mass spectrometry: practical approaches to their elimination. Rapid Commun Mass Spectrom 2003;17:1950 –7. 38. Miller WL, Sutton MJ. Relative biological-activity of certain prostaglandins and their enantiomers. Prostaglandins 1976;11:77– 84. 39. Yin H, Porter NA. New insights regarding the autooxidation of polyunsaturated fatty acids. Antioxid Redox Signal 2005;7:170 – 84.