CLIN.CHEM. 35/5, 874-878 (1989) Determination of trans-Phylloquinone in Children’s Serum Fathl Moussa,1 Luclenne Dutour,’ Jean Ren#{233} Didry,’ and Pierre Aymard2 By optimizing the conditions for determining trans-phylloquinone and its metabolite, K-2,3-epoxide, in serum through a two-step HPLC process combined with fluorometric detection after coulometnc reduction, we have been able to develop a method applicable to small volumes of serum (200 to 500 p.L). The limit of detection (signal-to-noise ratio of 3) was 15 ng/L for trans-phylloquinone, 30 ng/L for K-2,3-epoxide. The trans-phylloquinone concentrations measured by this method in serum from 82 children, ages one to six years, whose results were normal for overall coagulation tests, ranged from 40 to 880 ng/L (median 175 ng/L). We discuss these findings and compare them with vitamin K, (20) values reported for adults. A 3 B k AdditIonal Keyphrases: vitamin K pediatric chemistry reference interval chromatography, reversed phase . fluoromettic and coulometric methods Vitamin K (Figure 1) is required forthe post-translational gamma carboxylation ofglutamic acid residues on several proteins in plasma and tissue.The best known of these vitamin K-dependent proteins are procoagulant Factors II, VII, IX, and X and proteins C and S, allof which are involved in the formation or inhibition of thrombin. The physiological role of the other vitamin K-dependent proteins is less well understood (1, 2). The diagnosis of K hypovitaminoses is currently based on detection of depressed coagulant activity of vitamin Kdependent factors or increased concentrations of their inactive precursors in the circulation (3). However, such tests are not specific (4). Several high-performance liquid-chromatographic (HPLC) methods have been publishedfordetermining concentrationsof vitamin K1(20) in biological media (5-14). However, the mean concentrationsreported for serum of normal adults range from 247 (15) to 2600 ng/L (5). By optimizing the conditions for determination by HPLC, and incorporating fluorometric detection after post-column coulometric reduction (9, 10), we have developed a method that can detect - 15 pg of trans-phylloquinone [the biologically active isomer of vitamin K1 (8)] and -30 pg of its metabolite, K-2,3-epoxide, per milliliter of serum sample. We have applied this method to the study of trans-phylloquinonemia in children. C 3 0 n Fig. 1. Structuralformulas of vitamin K compounds (A) bans.vitaininK1(20) (other synonymsusedare “frans-vitaminK1 or fransphylioquinone’).( bans-vitamin K1(25),synthetic stnicturalanalog of vitamin K,, used as internal standard. ( bans-K,(20)-2,3-epoxide.(1 bans-vitamin I(2(n) or menaquinone-n respiratory infections, tonsillectomy, or adenoidectomy. None ofthem displayedany symptoms ofdigestive or hepatic trouble, and results of their hematological tests [prothrombin time (PT) and activated partial thromboplastin time (APTI’),which evaluatethe total extrinsic clotting system and the total intrinsic clotting system,respectively] were all within normal limits. The blood samples were promptly dispatched to the laboratoryfor centriftigation. The separated serum was immediately frozen and stored at -20 #{176}C until analysis. To plot the calibration curves, we used standards prepared in a commercial lyophilized serum (“Biotrol 3 Taux: Taux faibles”; bioMerieux, Charbonnieres les Bains, France). recurrent MaterIals and Methods Reagents Samples We used two batches of vitamin K1(20) (2-methyl-3phytyl-1,4-naphthoquinone) from different sources (F. Hoffmann-La Roche, Neuilly-sur-Seine, France, and Sigma Chemical Co.,St.Louis, MO). Vitamin K1(25) and K-2,3epoxide were supplied by F. Hoffinann-La Roche. Stock Blood (1-1.5 mL) was sampled from 82 children (34 girls and 48 boys, ages 1-6 years) for various blood testson the first day of their admission to our otorhinolaryngology departments. The children were hospitalized for either ‘Hopita! Trousseau, Laboratoire de Biochimie, 26 Ave du Docteur Arnold-Netter, 75571 Paris Cedex 12, France. 2Universite Paris-Sud, Laboratoire de Physiologie, Rue J. B. Clement, 92290 Chatenay-Malabry, France. Received November 21, 1988; accepted February 17, 1989. 874 CLINICALCHEMISTRY, Vol. 35, No. 5, 1989 solutions of the different vitamins-1 g/L, in hexane-were stored at -20 #{176}C in the dark.The standard solutions were diluted in absoluteethanol and stored under the same conditions. The concentration of the standard solutionof vitamin K1 was determined by its absorbance at 248 nm (molar absorptivity = 19.9 mol’cm’ in ethanol). The hexane and acetonitrile were “Uvasol” grade (Merck, Darm- stadt, F.R.G.). The absolute ethanol wa from Carlo Erba, Milan, Italy. The NaC1O4 for analysis came from Merck. All the chemicals were used as obtained, without further purification. An “Oxisorb” cartridge (Messer-Griesheim GMBA, Ivry, France)was used to ifiter the U-grade nitrogen. For the PT test, we used “Thrombomat” thromboplastin with calcium forthe automated system (bioMerieux) and the AP’I’F test (Automated APTF for General Diagnosis; Organon Teknika Corp., Durham, NC 27713). any intake Apparatus Results and Discussion Two HPLC systems are required for this detennination. The first, a normal-phase system, used for the semipreparative step, consisted ofa Model SP 8810 pump (Spectra-Physics, Les Ulis, France) connected to a Model 7125 injector (Rheodyne, Cotati, CA) equipped with a 100-giL loop; the 250 mm x 4 mm silica column packed with 5-pm particles was protected by a 30 mm x 4 mm pre-column (Waters Chromatography, Division of Millipore, St. Quentin en Yveline, France), dry-packed with 30- to 38-pm (diameter) particles of silica (HC Pellosil). The LC-UV Pye Unicam detector (Philips, Bobigny, France) operating at 248 nm was connected to a Servotrace recorder (Sefram, Paris, France). The fractions containing the different K vitamins were collected in a fractioncollector (Microcol TDC 80; Gilson, Procedures Villiers-le-Bel, France). The second, a reversed-phase HPLC system, used forthe analytical step,consisted of the same pump system with a 50-pL injection loop. A 150 mm x 4 mm Novapak (C18, 4pm particles) chromatographic column (Waters) was used for separation tests; a 70 mm x 4.6 mm XL 3-pin octyl cartridge (Beckman, Gagny, France) was used forroutine assays. The post-column coulometric reduction was performed with a Model 5011 cell controlled by a Coulochem 5100A module (both from ESA, Cergy-St. Christophe, France). The fluorescenceof the reduced products was measured in a fluorospectrophotometer (Model RF 530; Shimadzu, Kyoto,Japan) connected to a recorder. of oxygen into the chromatographic system, only tubes and connections were used for its assembly.) Because the applied potential on the two detectors of the cell from ESA was -0.8 V, we did not begin sample injections until the current displayed by the 5100-A module exceeded -8 p.A; this was vital for obtaining optimum efficiency of reduction. For the fluorometer, the excitation and emission wavelengths were set at 320 and 430 nm, respectively. stainless-steel Extraction of lipid fractions. Physical recovery of vitamin K1(20)and K1(25)added to serum exceeded 90% in samples extracted with six volumes of hexane after proteinshad been precipitated with two volumes ofethanol.These results agree with thoseofothers(8). Semi-preparative normal-phase HPLC. Figure 2 illustrates the separation of a mixture of cis-trans isomers of vitamin K1(20) and K1(25) and of vitamin K-2,3-epoxide. The presenceof menaquinones (Figure 1) or vitamins K2 (endogenous vitamin K synthesized by intestinal flora) was not evaluated for lack of calibration reagents. The commercial vitamin K1(20) we used is a mixture of about 12% of cisisomer and 88% of trans-isomer. Similarly, commercial vitamin K1(25) is a mixture of about 33% of cis-isomer and 67% of trans-isomer. A typical chromatographic proffle ofa serum sample is shown in Figure 3. Currently, investigators generally agree on the need fora first, normal-phase HPLC to eliminate the interfering lipids in the serum (8, 11, 12, 14). Haroon et al. (13) proposed a method ofextraction and purification that obviatesthe first chromatographic step; however, their method is tedious and time-consuming and does not separate the cis-trans isomers of vitamin K1. Analytical reversed-phase HPLC. Figure 4 shows the chromatographic profiles of two untreated serum samples (500 pL of serum). The limit of detection of the method for trans-vitamin K1(20) approximates15 ng/L, as determined Procedure After extracting the lipid fractions of the serum, we subjected them to normal-phase HPLC, followed by reversed-phase HPLC. Vitamin K1(25) was used as the internal standard. Extraction of lipid fractions: After adding 2 ng of the internal standard in 50 pL ofethanol to 200 or 500 pL of serum, we precipitated the proteins with two volumes of absolute ethanol. We then mixed the samples with six volumes of n-hexane. After mixing for 15 mm and then centrifuging, we removed the hexane (upper) layer and evaporated it under a stream of nitrogen. Semi-preparative chromatography: We dissolved the residue from the extraction in 100 p.L of mobile phase (hexane/ acetonitrile, 99.85/0.15 by volume) and injected 90 L ofthis intothechromatograph. The fractions containing the transisomers of the internal standard and vitamin K1(20) (and sometimes K-2,3-epoxide) were collected, combined, and evaporated under nitrogen. Analytical chromatography: We dissolvedthe residue from the effluent collected inthe first chromatographic step in 60 pL ofthe second mobile phase (NaC1O4, 5 mmol/L in acetomtrile/ethanol, 95/5 by vol) and injected 50 L of this into the chromatograph. (This mobile phase had been deoxygenating by bubblinga stream ofnitrogen through it; this was maintainedthroughouttheentire operatIon. To prevent I 0 8 16 win Fig.2. Separationof vitamin K1compoundsby normal-phasechromatography Column:250 mm x 4 mm. Silica (5 pm). Mobilephase 1.5 mL per liter ofnhexane,flow rate 1.2 mL/min. Ultraviolet detection (X: 248 nm). Peaks: (A) cia. K1(25), ( cis-K1(20), (C trans.K1(25), (L frans-K,(20), (E) trans-K,-2,3epoxide.Chart scale: 32 x iO absorbanceunit,as indicatedby bar CLINICAL CHEMISTRY, Vol. 35, No. 5, 1989 875 Table 1. Precls Ion of the Method CV, % Av. concn. Serum pool I It Z 12 Fig. 3. Semi-preparative chromatogram of an extract of 0.5 mL of serum Fractionscontaining frans-K1(25),trana.K1(20),and K1 epoxideare indicated. Conditions as in Fig.2 k Fig. 4. Reversed-phase chromatographic analysis of a collected fraction from the silica column Stationaryphase:3-pm octyfcartridge.Mobilephase:acetonithle-ethanol (95/5 by vol) containing5 mmol of sodiumperchiorateper liter. Flowrate:0.8 mL/min. Appliedpotential: -0.8 V. K,,, 430 nm, K,,,, 320 nm.Peaks:(A) trana.K,(20),( nt. 5th., (C) impurityaccompanyingthe nt std.(a) frans-K1(20)concentration= 140ng/L. (b) eans-K1(20) concentration= 730 ng/L K, ng/L Within-run (n = 5) Between-run (n = 10) 1 100 2 250 8.9 6.6 3 400 - 7.1 4 750 5.1 - Although vitamins o of vit. 12.7 - this does not separate the cis-trans isomers, K1 and K2 are easily separated. The fluorescent naphthohydroquinone, obtained by reduction of the nonfluorescent naphthoquinone (9), affords a high degree ofspecificity and sufficient sensitivity for the study of small volumes ofserum. Post-column reductioncan be done eitherchemically (13,14) orcoulometrically (9-12). Chemical derivitization is difficult to set up, and the void volumes it generates seem prejudicial to the sensitivity of the method; reported (13, 14) detection limits are all >50 ng/L. Coulometric reduction is easier to use, cleaner, and especially more specific, because it is governed by the applied potential. However, the method proposed by Langenberg and Tjaden (9, 10) has been criticized (13, 14). Van Haard etal. (11) improved themethod ofLangenberg and Qjaden (9) by introducing a first, normal-phase chromatographicstep,but they used the same mobile phase for the analytical step, thereby retainingthe problems of adsorptiononto the working electrodes. In fact,when we ourselvesinitially attempted to use the mobile phase proposed by these authors (9, 11)-NaClO4, 5 mmol/L, in methanol/water, 92.5/7.5 by vol-we observed a lack of reproducibility, related to a gradual declineinthe detector’s response with the number ofinjections. This lackof reproducibility is,in fact,ascribable to the passivationof the working electrodes by adsorption of the reduction products, because regeneration of the electrodes is accompanied by improved sensitivity, as we recorded during a series of determinations. Moreover, the decrease in sensitivity observed at voltages under -0.5 V [20% loss according to Langenberg and 1’aden (10)1 was undoubtedly caused by amplification of the phenomena of adsorption, which would tend to limit reduction of the vitamin K molecule. It was for these reasons that we decided to reject the mixture they (10) used in favor of a mobile phase with an acetonitrile base. According to Langenberg and Tjaden,the background noise produced by acetomtrile/water is higher than that produced by methanol/water (9). In our opinion, this is not caused by the acetomtrile; conversely, the water added to the medium may accountforthe phenomena ofadsorptionand passivationofthe electrodes (16). The mobilephase we propose in our methods consistsofa mixture ofNaC1O4, 5 mmol/L, in acetonitrile/ethanol (95/5 by vol) and offers the following advantages: with a signal-to-noise ratio of 3 fora zero standard. The linearity of the method was studied by supplementing three 500-giL aliquots of Biotrol control serum with 50, 150, and 375 pg of vitamin K1, respectively, and analyzing them. Within these limits, the calibration curve determined by linear regression gives the following equation: y = 0.61x + 5.92 ng/L (r = 0.997, n = 3). Analytical recovery after extraction and the first chromatography exceeded 70% as measured in relation to the internal standard. The precision of the method was studied by use of four pools of serum with 500-pL samples (Table1). Published methods indicate the wide use of silica C18. 876 CLINICALCHEMISTRY, Vol. 35, No. 5, 1989 #{149} As acetonitrile is less viscous than methanol, it is more efficient for chromatographic separation. #{149} From the polarographic point of view, under the same experimental conditions (electrodes, supportingelectrolyte, etc.), thepotential fields thatcan be used aremore cathodic for acetonitrile than for methanol (15). We verified thiswith the porous graphite electrode we used. (The reference electrode is not specified by the supplier.) Figure 5 shows the current-voltage curves obtainedunder the same operating conditionsfor the two mixtures considered. Use of our acetonitrile mixture minimized background noise and achieved optimum efficiency of the working electrodes. C,,) (2) 20 o 8 8 12 16 20 20 -in 10 (b) A -0.5 -1.0 - 2,0 volt, Fig. 5. Coulometricscans of mobile phases: (A) methanol-water(92.5/ 7.5 by vol) containing 5 mmol of sodium perctilorate per liter, (B) acetonitille-ethanol (95/5 byvol)containing5 mmolof sodium perchlorate per liter 0 8 8 12 16 20 20 0 0 8 12 -in Figure 6 shows the fluorescence curves depending on the potential. For vitamin K1 the maximum signal is obtained at -0.8 V; for epoxide, at -1.2 V. Contrary to what happens in the methanol/water mixture (10), there is no decrease in the fluorescence intensity at the more cathodic potentials, the adsorption phenomena being negligible. In the methanol/water mixture, the half-wave potential (E’) approximates -0.3 V; in the acetonitrile/ethanol mixture, -0.6 V. This cathodic shift is due to the lower proton-donor capacity ofthe acetonitrile/ethanol mixture. #{149} Adding ethanol to acetonitrile increases the detector’s response twofold. This is mainly because ethanol, by increasing the solubility ofvitamin K, decreases its capacity ratio; also, acetonitrile is an aprotic solvent, and the ethanol acts as proton donor to facilitate the reduction of the naphthoquinone. There is sufficient final concentration of supporting electrolyte forthe reduction current to be governed essentially by the diffusion phenomena. Figure 7, a and b, shows the results obtained with a Novapak column (C18, 4 jim) at two different voltages. The stationary phase is highly selective, entailing big retentionvolumes and overly long analysis times. We obviated this by using a C8 column Fluorescence (B) (A) 0.1 0.5 0.8 1.0 1.2 volt. Fig. 6. Relationship between the fluorescence intensity and the potential appliedto the electrodes of the electrochemicalcell (A) K1(20);( K1(20)-2.3-epoxide Fig. 7. Chromatogramsof a mixtureof (A) vitamin K1 epoxide,(B) K1(20),and (C) K1(25),the internalstandard (a)stationary phase:Novopak C18column, flowrate1.6 mL/min, applied potential -0.8 V. (b) Same, but applied potential-1.2 V. Retention volume(V,) ofpeakA = 5.9OmL, V,ofB= 11.62mL,and VrOf C= 37.5OmL(C)Stationaryphase:3pm octyl cartridge, flow rate 0.8 mL/min,appliedpotential-1.2 V. V, of A = 1.90 mL, V, of B = 2.60 mL, and V, of C = 5.20 mL D is an unknown impurity accompanying the vitamin K1(25). Mobile phase: acetonitnle-ethanol(95/5 by vol) containing 5 mmolof NaCIO4per liter to reduce the hydrophobiclinkages between the stationary phase and the solutes. Figure 7c shows the results obtained with a fast column. The method of determination proposed by MummahSchendel and Suttie (12) is similar to that of van Haard et al. (11) and ours; however, these authors used a mobile phase with an ethanol base (too viscous). Moreover, their method is not very practicable, because they used radioactive vitamin K1 as the internal standard. trans-Phylloquinone in Children’s Serum Figure 8 shows trans-phylloquinone concentrations, categorized by age and sex, for the 82 children studied. Because of the small amount of serum available, we did not study K2,3-epoxide. The distribution (cf thehistogramin Figure 9) followsa normal logarithmic rule, as confirmed by Shapiro’s test. This agrees with the results observed in adults by Lambert et al. (14). The observed concentrations in serum ranged from 40 to 880 ngfL. When we compared the median values for the 48 boys (196 ngIL) and the 32 girls (149 ng/L), we found no meaningful sex-related difference (t = 1.51, 80 degrees of freedom). Apparently, there is no correlation between age and the concentrations of trans-phylloquinone in serum. The significant F-value for 1 and 80 degrees of freedom for numerator and denominator, respectively, at the 5% level of significance was 3.9, and we obtained an Fvalue of 0.79. The overall median was 175 ng/L (95% limits for one measurement: 33 to 920 ng/L; 95% limits for the median: 140 to 210 ngfL). Although close to those reported by Shearer et al. (6) and Lambert et al. (14), our values are lower than those indicated for adults (14). Remember that our values were obtained for children hospitalized for ear, nose, or throat trouble, and whose diets in the days preceding the sampling were not monitored. Also, all these chilCLINICALCHEMISTRY, Vol. 35, No. 5, 1989 877 conc. of methods of determination of PIVKA-ll, a “proteininduced in vitamin K absence,”will undoubtedly make it possible to define vitamin K requirements better by studying the correlation between the concentrations of this protein in serum and those of vitamin K. This is the objective we are pursuing in our laboratory. (nq/L) U I I U o) U #{149} I 0 #{149} 13 00 p U We thank Christine Beri-ivin fortranslation U ‘> #{149}U.)O0 0.) 0 #{149} U #{149} #{149} 0’’#{149} 6i > 0 20 0 00 #{149} U #{149} .3 0 40 00 ) 70 *131 (oocthsl I 0 50t between serum frans-K1(20)concentrationsin 82 childrenaccording age and sex Fig. 8. Relationship number 25 30 25 20 ‘5 References 1. Suttie JW. The metabolic role of vitamin K. Fed Proc Fed Am Soc Exp Biol 1980;39:2730-5. 2. Goodman LS, Gilnian AG, Murad F. Pharmacological basis of therapeutics. 7th ed. New York: Macmillan 1985;1582-6. 3. Kries RV, Shearer MJ, Gobel U. Vitamin K in infancy. Eur J Pediatr 1988;147:106-12. 4. Lane PA, MD, Hathaway WE. Vitamin K in infancy. J Pediatr 1985;106:351-9. 5. Lefevere MF, De Leenheer AP, Clayes AE. High performance liquid chromatographic assay of vitamin K in human serum. J Chromatogr 1979;186:749-62. 6. Shearer MJ, Rahim S, Stimniler L. Plasma vitamin K1 in mothers and theirnewborn babies. Lancet 1982;ii:460-3. 7. Pietersma De Bruyn ALJM, Van Haard PMM. Vitamin K1 in the newborn. ClinChim Acts 1985;150:95-101. 8. Takami U, Suttie JW. High pressure liquid chromatographic reductive electrochemical detection analysis of serum transphylloquinone. Anal Biochem 1983;133:62-7. 9. Langenberg JP, Tjaden UR. Determination of (endogenous) vitamin K1 in human plasma by reversed phase high performance liquid chromatography using fluorometric detection after post- column electrochemical reduction. J Chromatogr 1984;305:61-72. 10. Langenberg JP, Tjaden UR. Improved method for the determination of vitamin K1 epoxyde in human plasma with electrofluorometric reaction detection. J Chromatogr 1984;289:377-85. 11. Van Haard PMM, Engel R, Pietersma De Bruyn AL.JM. Quantitation of trans-vitamin K1 in small serum samples by offline 10 0.i tOO 200 300 400 500 600 700 multidimensionalliquidchromatography. 000 900 (ng/t) 1986;157:221-30. 12. Mummah-Schendel centrations dren studied had normal valuesforPT and APTF, even with serum vitamin K concentrations of 40 ngfL. However, these two tests (PT and APTT’) are not sensitive enough to detect a Clin Chem 1986;32:1925-9. K hypovitaminosis; nor should we overlook the contrary to what is observed in infants, menaquin- ones may play a role in children and adults. At present an isolated concentration of trans-phylloquinone in serum is difficult to interpret. The recent elaboration 878 CLINICALCHEMISTRY, Vol. 35, No. 5, 1989 Clin Chim Acta LL, Suttie JW. Serum phylloquinone conadult population. Am J Clin Nutr Fig. 9. Frequencydistribution curves forserum trans-K1(20) concentrations in 82 children moderate fact that, and Valerie Jour- dam forsecretarial assistance. 0. in a normal 1986;44:686-9. 13. Haroon Y, Bacon DS, Sadowski JA. Liquid chromatographic determination ofvitaminK1 in plasma with fluorometric detection. 14. Lambert WE, De Leenheer AP, Baert FJ. Wet chemical postcolumn reaction and fluorescence detection analysis ofthe reference internal of endogenous serum vitamin K1(20). Anal Biochem 1986;158:257-61. 15. Bard AJ. Electro-analytical chemistry: a series of advances. Vol 3. New York: Marcel Dekker, 1969:65-110. 16. Chao F. Transfert d’#{233}lectrons et reactions superficiellea. In: Ecole d’#{233}lectrochimie du C.N.R.S., Reactions electrochimiques applications. Vol 1. Lea Houches: C.N.R.S., 1978:59-85.