Anal. Chem. 1985, 57,2101-2106 (8) Kenyeres, J. S.; Ursu. U. I . J . Polym. Scl., Polym. Chem. Ed. 1980,
18, 275-281. (9) Scoggins, M. W.; Miller, J. W. SOC.Anal. Chem. 1975, 4 7 , 152-154. (10) Muller, G.; Laine, J. P.; Fenyo, J. C. J . Polm. Scl., Polym. Chem. Ed. 1979, 17, 659-872. (11) Mitchell, P. H.; Hester, R. D. 4th Annual Report for Oct 1980-Sept 1981, DOE/BC/10321-5, 1982, 255-274.
2101
(12) Herman, D. P.; Field, L. R.; Abbott, S. J . Chromafogr.Scl. 1981, 19, 470-476. (13) Barth, H. G. J . Chromatogr. Scl. 1980, 18, 409-429.
RECEIVED for review March 18,1985. Accepted May 20,1985.
Simultaneous Determination of Pyridoxal and Pyridoxal 5-Phosphate in Human Serum by Flow Injection Analysis Pilar Linares, M. D. Luque de Castro, a n d Miguel Valcarcel* Department of Analytical Chemistry, Faculty of Sciences, University of CBrdoba, Avda. Medina Azahara, CGrdoba, Spain
The sequential, slmuitaneous, and differential-kinetic determination of pyrldoxai (Py) and pyridoxal 5-phosphate (Py5P) based on the oxidation reaction of these compounds in the presence of cyanlde yielding fluorescent substances is suggested. The sequential method Involves the use of a diverting valve to provlde the suitable carrier for the determination of each of these compounds. The other methods use the native fluorescence of Py and that of the oxidation product of Py5P (simultaneous method) and the fluorescence of both oxldatlon products (differential-kinetlc method), respectively. The calibration curves cover the range of 10-*-10-3 M for both compounds, with sampling rates of 30 sampies/h and relative standard devlatlon values less than f l % . These species have been determined In serum samples at the lo-' M level with an average error of f10-15%.
The determination of BBvitamers and their derivatives is of great interest, especially in clinical chemistry. Among them, the determination of pyridoxal (Py) and ita phosphate (Py5P) has received much attention. The most common technique for the determination of both compounds is fluorometry, involving the use of Zn-glycine ( I ) , formation of hydrazone derivatives (2),direct fluorescence measurements of these substances after a chromatographicseparation (3,4),reductive amination with methyl anthranilate and sodium cyanoborohydride (5) or reaction with cyanide (6) to yield 4pyridoxolactone and 4-pyridoxic acid 5-phosphate (Py and Py5P, respectively) (7). On the basis of this last reaction and by use of flow injection analysis, FIA, in a previous paper (8), we suggested fluorometric methods for the determination of both pyridoxal and pyridoxal 5-phosphate. Both are sensitive methods, with wide linear calibration ranges (2.5 X lo4 to 1.0 X M), good precision (relative standard deviation 0.3%), a sampling rate of 25 samples per hour, and a high tolerance level for species such as NAD,normally an interference in the manual methods (9). The determination of both species (Py and Py5P) in the same samples has not been suggested in any of the above-mentioned papers. At most they reported the determination of one of them (Py), hydrolysis of Py5P, and determination of the sum (IO). Flow injection analysis offers the chance of determining both species in a mixture without a prior separation by several methods based on the different working conditions or difference of completion of the indicator reaction with the use
of a very simple manifold and a flow cell in a single detector. The indicator reaction chosen for the development of determinative methods of these vitamers in mixtures has been the above-mentioned oxidation to 4-pyridoxolactone and 4pyridoxic acid 5-phosphate under the catalytic action of cyanide ion. On the basis of this reaction and after checking that both products are suitably former at a pH close to neutrality, but showing maximal fluorescence at very different pH values, a sequential method involving the use of a simple manifold with a diverting valve which provides carriers with cyanide of alternative pH's for the determination of one or the other compound has been applied. The sample injection into each carrier provides peaks in which the contribution of one of the products is maximal and that of the other is minimal or nil. On the basis of the different oxidation rate of Py and Py5P, two methods are suggested with the use of a manifold with splitting of the flow into two channels (with different geometrical and hydrodynamic characteristics) which merge prior to arriving at the detector. At the splitting point the sample is divided into two parts which, due to the characteristics of the two channels, reach the detector at different times, yielding a two-peak fiagram per sample injected. In each of these peaks the contribution of each product to the analytical signal' is different. This reaction-rate difference has been taken advantage of in two different ways providing the measurement of the native fluorescence of one of the compounds (Py), which implies its maximal contribution to the first peak (short reactor and minimal dispersion) and that of the oxidation product of the other (PySP), with which its contribution to the second peak will be higher (longer reaction time). In the differential-kinetic method the working experimental conditions have been optimized in order to make both oxidation reactions feasible, but with the maximal possible rate difference between them. EXPERIMENTAL SECTION Reagents. The stock solutions used included aqueous solutions M), pyridoxal 5-phosphate (1 X M), of pyridoxal (1 X phosphate buffer (0.6 M) at several pH values, 0.9 M hydrochloric acid, and an aqueous 1.000 g/L solution of potassium cyanide. Apparatus. A Perkin-Elmer 650-10 S spectrofluorometer, equipped with a Hellma 176.52-QSflow cell (inner volume 25 pL) and connected to a Radiometer REC-80 recorder, was used. Gilson Minipuls-2 and Ismatec s-840 peristaltic pumps, a Tecator L100-1 injection valve, a Rheodyne 5301 three-way valve, a Tecator TM I11 chemifold,and the accessory instruments Radiometer PHM-82 pH meter and Selecta 38243 thermostat were also used.
0003-2700/85/0357-2101$01.50/00 1985 American Chemical Society
2102
ANALYTICAL CHEMISTRY, VOL. 57, NO. 11, SEPTEMBER 1985
A)
C N-
Buffer
Jt
{
I---
I
Recorder
1 Time-
C NBuffer
T
-
I
I
q3
W
Time
-
Figure 1. FIA manifolds used for the determination of pyridoxal and pyridoxal 5-phosphate together with the recordings obtained in each case: (A) for the sequential determination: (B) for the simultaneous determinations (semikinetic and differential-kinetic methods).
Procedure. To 8 mL of serum, previously centrifuged, 0.7 mL of trichloroacetic acid (TCA) 75% w/v was added; the mixture was stored for 5 min and then centrifuged for 15 min. A 2.5-mL portion of the supernatant was added to a 10-mLvolumetric flask, then 3 mL of H2P04-/HP042-0.6 M buffer (pH 7.40) was added and the mixture ajusted to this pH with NaOH and made up to volume with distilled water. From each type of serum three different samples were prepared: (a) by the above-mentioned procedure; (b) by the same procedure as sample (a) but adding prior to the treatment an accurately measured amount of Py; (c) by the same procedure as sample (a) but adding prior to the treatment an accurately measured amount of Py5P. The blanks were recorded with these samples, but using a carrier lacking cyanide ion.
RESULTS AND DISCUSSION Sequential Method for Determination of Pyridoxal a n d Pyridoxal 5-Phosphate. The studies about the optimization of variables for the separate determination of Py and Py5P (8) showed that, although the oxidation products of these two compounds in the presence of cyanide ion are formed under optimum working conditions in a slightly basic medium (pH 7.37-7.43), the maximal fluorescence intensity, If,of these products is obtained at very different pH values, so that for Py its oxidation product (4-pyridoxolactone) shows the maximal analytical signal at its characteristic wavelengths in a slightly basic medium, while 4-pyridoxic acid &phosphate (oxidation product of Py5P) requires an acidic medium for yielding the optimal signal, in which the fluorescence of 4pyridoxolactone is practically nil. On the basis of these conditionsand with the aid of a diverting valve, the sequential method for the determination of these substances has been applied. The FIA scheme for the development of the method is very simple consisting of a cyanide stream at pH 7.40 into which the Py + Py5P sample is injected, a reactor in which the oxidation produds are formed, and a second channel which merges with the main one and carries to it by an alternative turn of the diverting valve (a) a buffered solution of pH 7.47 which allows achieving the determination of Py since when the sample zone reaches the detector it has the suitable pH to make 4-pyridoxolactoneshows its maximal fluorescence and
(b) a 0.9 M hydrochloric solution which allows the sample zone to reach the detector under the suitable conditions for which 4-pyridoxic acid 5-phosphate yields the maximal signal. We should note that in case a the Py5P oxidation product contributes to the analytical signal as does 4-pyridoxolactone in case b, although to a lesser extent. The evolution of the simplex toward increasingly higher values of the injection volume, Vi, reactor length, L1, and temperature showed in the application of the modified simplex method (MSM) to these systems. Thus, the values of these variables were bound between the values shown in Figure 1, the same for the two systems. Hence, taking into account that the characteristic wavelength of the oxidation products were Py-CN- system, A,= 355 nm and A,, = 435 nm, and Py5PCN- system, X, = 325 nm and A,, = 420 nm, only the common A,, and A,, values needed optimizing in order to achieve a similar peak height for equal concentrations of both species in the sample and with minimal contribution to the peak height of the product with lower fluorescence. The optimal values found were A,,(mixture) = 345 nm and A,,(mixture) = 425 nm (Sex= 6 nm, S,, = 7 nm). Resolution of Pyridoxal and Pyridoxal 5-Phosphate Mixtures. The calibration curves of Py and Py5P in isolation have been recorded in both media by using the optimized manifold (Figure lA), finding a linearity range of 1 X to M for the two compounds. The features of these 1X curves as well as the statistical study of each system (ref 11, 5 x 10" M samples prepared under analogous conditions and with triplicate injection) are shown in Table I. The relative standard deviation (RSD) was always less than 1% under the optimum working conditions for the determination of each compound. The applicability of the method has been tested by solving a vast series of samples with different Py/Py5P ratios. Due to the sensitivity of these methods and to the wide concentration ranges covered by the calibration curve, it was necessary to work at several sensitivity values of the spectrofluorometer, so that a different sensitivity was used for each concentration decade. For this reason only the resolution of
ANALYTICAL CHEMISTRY, VOL. 57, NO. 11, SEPTEMBER 1985
2103
Table I. Features of the Optimization, Determination, and Mixtures Resolution of Py and Py5P by the Sequential Method Variables Fixed in the Optimization
M (Py or Py5P) + 50% buffer (HzP0C/HP02-) 0.6 M 1.0 X CN- 300 ppm in buffer (HzPO4-/HPO4*-)0.6 M
sample: carrier:
Variables Optimized 4:
L1
V:
PH
P
method
1.3
600
120.4
7.38
45
MSM
nm
method
43
(HCU
LZC
0.3
0.9
55
A,,
nm
A,,
345
univariant
425
Determination equation Py basic medium:
acidic medium: Py5P basic medium:
acidic medium:
r
RSD
hl = 2634.4 + 637247740(Py) hz = 91.4 + 88054783.3(Py5P)
0.999 0.998
0.38 2.59
hl = -519.9 + 333900664(Py) hz = -3998.9 876466788.5(Py5P)
0.999 0.999
0.76 0.84
+
linear range, M sampling rate, h-I 10-4 - io-*
30
Mixtures Resolution Equations HIM= 637247740.0(Py) + 333900663,O(Py5P) + 2100 HzM= 88054783.3(Py) + 876466788.5(Py5P) - 3900 amt added, M PY 5.00 X 5.00 X 1.00 x 1.00 x 5.00 x 5.00 X 1.00 x 1.00 x 1.00 x 5.00 X 1.00 x 5.00 x “qt
amt found, M Py5P
lom6 10-4 10-6
lo”
10“ 10-6 10“ 10“ 10-7 10-7
5.00 x 1.00 x 1.00 x 5.00 x 5.00 X 1.00 x 1.00 x 5.00 X 5.00 x 5.00 x 5.00 x 1.00 x
(mL/min). * Vi(pL). L (cm).
10-6
10-6 10-4 lo” 10” 10-6 10-6 10” 10-7 10-7 10-7 10-7
PY 4.82 X 4.83 X 0.90 x 0.95 X 4.73 x 4.73 x 0.93 X 0.96 X 0.97 X 4.91 x 1.15 x 4.82 x
% error
Py5P
10-4 10“ 10” 10“ lo4 10-7 10-7 10-7
4.83 X 1.05 X 0.98 x 4.90 X 4.93 x 0.92 X 0.96 X 4.79 x 4.48 x 4.59 x 4.68 x 0.97 X
lom6 10-4 10” 104 10” 10-7 10-7 10-7 lo-’
PY
Py5P
-3.6 -3.4 -10.0 -5.0 -5.4 -5.4 -7.0 -4.0 -3.0 -1.8 16.0 -3.6
-3.4 5.0 -2.1 -2.0 -1.4 -8.0 -4.0 -4.2 -10.4 -8.2 -6.4 -3.0
T (“C).
mixtures in the 10.00.1 range ratio was feasible. The errors are higher on both ends of the range, as can be observed from Table I. Nevertheless, in the cases in which the mixtures of these compounds do not fall in this range, the calibration curves may be plotted by using the optimal values for the excitation and emission wavelengths of the species in the lower concentration, thus increasing its analytical signal to the detriment of those of the more concentrated component. Simultaneous Determination of Pyridoxal and Pyridoxal 5-Phosphate. The optimal development of the oxidation reaction of Py and Py5P was shown to occur a t a slightly basic pH (7.40) and high temperature (45 “C) in earlier studies (8).The decrease in the pW and temperature slows down the Py-CN- reaction to a higher extent than the Py5P-CN- one; a phenomenon which properly optimized and used in a configuration such as that in Figure 1B allows the simultaneous determination (which may be called semikinetic to distinguish it from the conventionaldifferential-kinetic one in which both species react, although at a different rate). Since the preliminary assays showed in all cases a higher fluorescence intensity for Py than for the oxidation product of Py5P (each one at its characteristic wavelength), values closer to that of the latter were chosen in order to make the signals more similar: A,, = 330 nm, A, = 430 nm (Sex= 7 nm, S,, = 7 nm). Study of Variables. Although in a configuration such as that shown in Figure 1B where an interrelation between
variables is predictable, the MSM was chosen to optimize the FIA variables (total flow rate, qt, reactor length ratio, L2/L1, injected sample volume, Vi, and carrier and sample pH (the same for both). The selection of the response function, F, yielded the optimum expression
F=
hlPY
(h, and h2 denote the heights of the first and second peaks of the fiagram provided by each system). When the mixture is injected instead of each compound separatelythe response function yields a maximal contribution of Py in the first peak and minimal in the second, whereas the oxidation product of Py5P behaves in just the opposite way. The evolution of the simplex toward the excessive development of hlPYled us to introduce it as denominator in the response function. The results of this study as well as the values of the variables kept constant in this development are listed in Table 11. The rest of the variables (catalyst concentration and temperature) were studied by the univariant method since they were considered independent. The optimum values summarized in Table I1 were also meant to give rise to the minimal development of the Py-CN system, as was shown upon injection of samples of both compounds into a carrier lacking CN- ion. The signal corresponding to Py has the same value
2104
ANALYTICAL CHEMISTRY, VOL. 57, NO. 11, SEPTEMBER 1985
Table 11. Features of the Optimization, Determidation, and Mixtures Resolution of Py and Py5P by the Semikinetic Method Variables Fired in the Optimization
1.0 X M (Py or Py5P) + 50% buffer (HzPO4-/HP0?-) 0.6 M CN- 200 ppm in buffer (HZPO4-/HP0& 0.6 M
sample: carrier:
Variables Optimized 9t
LZIL1
LO
v,
PH
method
1.9
8.0
0.0
131.0
6.70
MSM
T
(CN-)
25
120
A,,
method
nm
A,,
nm 330
univariant
430
Determination equation
+ +
r
RSD
Py peak 1: peak 2:
hl = 590.8 53122023.9(Py) hz = 90.6 24503377.5(Py)
0.999 0.999
0.36 0.14
Py5P peak 1: peak 2:
hi = 151.8 + 15575675.7(Py5P) hz = -2.6 46017440.7(Py5P)
0.999 0.999
0.58 0.37
+
linear range, m
sampling rate, h-’
10-3 - 10-6
30
Mixtures Resolution HIM= 53122023.9(Py) + 15575675.7(Py5P) + 640 HzM= 24503377.5(Py) + 46017440,7(Py5P) + 80 amt added, M PY ’
1.00 x 10-5
1.00 x 5.00 x 1.00 x 5.00 X 1.00 x 1.00 x 1.00 x 5.00 X 1.00 x 5.00 X 1.00 x
10-6 10-5 10-6
10-4 10” 10” 10“ 10“ 10” 10-5
amt found, M Py5P
PY
1.00 x 10-6 1.00 x 10-4 5.00 X lo-’ 5.00 X lo-’ 1.00 x 10-5 1.00 x 10-4 1.00 x 10” 5.00 X 10” 5.00 X 10” 1.00 x 10” 1.00 x 10” LOO x 104
1.00 x 10-5 1.10x 10-5 4.95 x 10-6 1.12 x 10-5 5.08 X low6 0.99 x 10-4 1.01 x 10-6 0.91 x 10-6 5.26 X 10“ 1.24 X 10” 4.99 x 10” 0.95 x 10-5
in the presence of CN-, while that corresponding to the injection of Py5P becomes zero. Resolution of Py and Py5P Mixtures. The calibration curves of each compound recorded under the optimum working conditions shown in the manifold in Figure 1B are listed in Table 11,where the rest of the determination features are also shown. Note the higher reproducibility of the method (RSD less than 0.5% in all cases) favored by the addition of glycerine to the samples (2% v/v), which increases the sample viscosity making the splitting more even. The sample resolution has been carried out by using the proportional-equation method. The equations used as well as the samples solved are summarized in Table 11,from which species concentrationsabove lo4 M in the mixtures have been omitted. This is due to the fact that at high concentrations, and as in the second peak both substances (Py and oxidation product of Py5P) show fluorescence, a quenching phenomenon occurs, thus yielding high negative errors. In the -lo4 M range the mixtures whose ratio is between 10.0 and 0.1 yield errors less than 10% in all cases. The more different the concentrations of both compounds, the higher these errors, which is due to the fact that the contribution of each of them to the peak height is very different. As in the above method, this drawback may be avoided by choosing suitable values of wavelengths, temperature, etc. to favor the species in lower concentration. Differential-Kinetic Determinatioh of Pyridoxal and Pyridoxal 5-Phosphate. The third method suggested in this paper is based on the same conditions as the previous conditions with regard to manifold characteristics and deter-
% error
Py5P 1.02 x 1.01x 5.03 x 4.88 x 1.01 x 1.05 x 1.00 x 5.18 X 4.20 X 0.89 X 0.89 X 1.02 x
10-5 10-4 10-5 10-5 10-6
10-4 10“ 10”
10” lo* 10“
PY
Py5P
0.0 10.0 -1.0 12.0 1.6 -1.0 1.0 -9.0 5.2 24.0 -0.2 -5.0
2.0 1.0 0.6 -2.4 1.0 5.0 0.0 3.6 -16.0 -11.0 -11.0 2.0
mination principle-difference in reaction rate for both systems. However, in this case the optimization of variables is directed to reach the development of both oxidation reactions and maximal rate difference between them. In the systems under study the higher reaction rate corresponds to the Py5P-CN- system imd the lower to the Py-CN- one. Hence, maximal contribution of the Py5P-CN- system and minimal contribution of the Py-CN- in the first peak should be expected; that is, the Py5P-CN- reaction is almost complete and Py-CN- hardly takes place at reactor 1, whereas at reactor 2-long reactor-the Py-CN- reaction will be almost complete, with which the product of the Py5P-CN- system will be dispersed and will have a lower contribution to the first peak. The optimization of variables was very simple thanks to the above studies, because the optimum pHs for development of the reactions were very close (7.43 for Py and 7.30 for Py5P). Hence, an intermediate value (7.38) was chosen. An increase in temperature favors both reactions. Thus, the highest value of this that did not cause trouble to the systems (bubble formation) was chosen. The injection volume chosen was that of the previous method in order to make a comparison between them. The study of the wavelengths, reactor length ratio, and flow rate yielded the optimum values summarized in Table 111. The reactor diameter ratio (key to the adequate splitting of the sample zone (11)) was 0.70/0.35 mm Resolution of Py and Py5P Mixtures. The calibration curves for each species corresponding to both peaks show a straight range 2.5 X lo-’ to 1 X M, their corresponding equations being those shown in Table 111,together with other characteristics of the determination. Note the high precision
ANALYTICAL CHEMISTRY, VOL. 57, NO. 11, SEPTEMBER 1985
2105
Table 111. Features of the Optimization, Determination, and Mixtures bsolution of Py 4nd Py5P by the Kinetic Differential Method Variables Fixed in the Optimization
1.0 X M (Py or Py5P) + 50% buffer (H2PO;/HP02-) 0.6 M CN- 300 ppm in buffer H2P0;/HP0,2-) 0.6 M 131 rL 1.6 mL/min 45 O C
sample: carrier: Vi: qt:
T
Variables Optimized L2IL1
A,,
nm
univariant
432 nm
335 nm
8.33
method
nm
A,
Determination equation
+
r
RSD
Py peak 1: peak 2:
hl = 63.5 137394904(Py) hz = 12.8 + 322355447,9(Py)
0.999 0.999
0.75 0.32
Py5P peak 1: peak 2:
hl = 124.3 + 168316139.7(Py5P) h2 = -45.2 210800499.7(Py5P)
0.999 0.999
0.44 0.37
linear range, M
10-4 - 2.5 x
+
sampling rate, h-l
30
10-8
Mixtures Resolution
HIM= 137394904.0(Py)+ 168316139.7(Py5P) + 187 HZM= 322355447.9(Py) + 210800499.7(Py5P) - 33
amt found. M
amt added, M PY 1.00 x 5.00 x 1.00 x 1.00 x 5.00 X 5.00 x 1.00 x 1.00 x
5.00 X 5.00 x 1.00 x 1.00 x
Py5P 10-5 10-5 10-4 10” 10” 10” 10-5 10-7 lo-’ 10-7 10” 10”
1.00 x 1.00 x 1.00 x 5.00 x 5.00 X 1.00 x 5.00 x 5.00 x 5.00 x 1.00 x 1.00 x 5.00 x
10-5 10-4 10-4 10” 10” 10-5 10” 10-7 10-7 10-7 10“ 10-7
PY 1.10x 5.41 x 0.92 X 1.07 X 5.24 X 5.00 x 1.06 x 1.01 x 5.40 x 5.43 x 1.01 x 1.03 X
%
Py5P 0.98 x 0.98 X 1.06 x 5.30 X 5.13 X 1.04 x 5.00 x 4.98 x 4.50 x 0.90 x 0.98 X 4.51 x
10-5 10-5 lo-’ 10” 10” 10” 10-5 10-7 10-7 10-7 10“ 10”
10-5 lo4 10-4 10” 10” 10-5 10“ 10-7 10-7 10-7 10” 10-7
error
PY
Py5P
10.0 8.2 -8.0 7.0 4.8 0.0 6.0 1.0 8.0 8.6 1.0 3.0
-2.0 -2.0 6.0 6.0 2.6 4.2 0.0 -0.4 -10.0 -10.0 -2.0 -9.8
Table IV. Determination of P y and Py5P in Human Serum amt found, M
% recovery
sample
Py, Py5P added, M
PY
Py5P
1
0 2.5 X 0 2.5 X 0 2.5 X 0 2.5 X 10” 0 2.5 X
2.07 X 4.57 x 10-8 2.09 X 4.22 X lo-* 1.50 X 4.17 X 6.50 X 8.70 X 0.80 X 2.50 X
2.48 X 5.65 X 3.95 x 10-8 6.50 X 9.40 X 11.20 x 10-8 1.60 X 3.67 X
2
3 4 5
of the suggested method (RSDless than 0.8% in all cases) partially due to the presence of glycerine in the samples (2 % VIVI.
Mixture resolution is subject to the same conditions of the semikinetic method, the errors found being of about the same order (see Tables I1 and 111). Determination of Pyridoxal and Pyridoxal 5-Phosphate in Human Serum. Due to the low concentration of Py and Py5P in the human serum the sequential method has been chosen to be applied to the determination of these compounds in this type of sample because of its lower determination limit. The pretreatment of serum sample requires only the deproteinization with trichloroacetic acid (TCA) which at high concentration partially destroys Py and Py5P. Therefore a
2.48 X
PY
Py5P
100
I27
85
102
107
72
88
83
68
99
previous study was carried out on the TCA concentration effecting deproteinization without destroying the analytes. The native fluorescence from Py and Py5P and especially from pyridoxic acid makes it necessary to record blanks by injecting the corresponding samples into a carrier with the same characteristics, but without cyanide ion. The standard addition method was used to calculate the Py and Py5P concentrations as explained under Experimental Section in a similar way to that used for other authors for determination of these species (9). The results of the determination of the analytes on five samples of different sera are shown in Table IV, where the acceptable recovery percentages obtained, taking into account the type of analytes and their concentration, can be observed. The application of the standard addition method to check the validity of the
2106
Anal. Chem. 1985. 57. 2106-2109
suggested method is supported by the fact that if variable amounts of P y and/or Py5P are added to several samples,the difference in A& is the same between samples with or without serum if the same amount of Py and/or Py5P has been added, with an accuracy of 1%.
Registry No. Pyridoxal, 66-72-8;pyridoxal 5-phosphate, 5447-7; 4-pyridoxolactone,4753-19-9; 4-pyridoxic acid 5-phosphate, 954-27-8.
CONCLUSIONS
(1) Maeda, M.; Ikeda, M.; Tsujl, A. Chem. Pharm. Bull. 1978, 24, 1094-1 097. (2) Uno, T.;Nakano, S.; Taniguchi, H. Jpn. Ana/ysf 1971, 20, 1117-1123. (3) Yamada, M.; Saito, A.; Tamura, 2. Chem. Pharm. Bull. 1988, 74, 482-487. (4) Hakanson, R. J . Chromatogr. 1884, 73, 263-265. (5) Chauhan, M. S.; Dakshlnamurti, K. Anal. Biochem. 1979, 96. 426-432. (6) Bonavita, V. Arch. Biochem. Biophys. 1880, 88, 366-372. (7) Ohishl, N.; Furui, S. Arch. Biochem. Siophys. 1968, 728, 606-610. (8) Linares, P.; Lugue de Castro, M. D.: Valcarcel, M. Anal. Lett. 1984, 78 (BI), 67. (9) Takanashi, S.; Tamura, 2. J . V/faminol. 1970, 76, 129-131. (IO) Fernandez, A.; Gomez-Nieto, M. A.; Luque de Castro. M. D.; Valcarcel, M. Anal. Chim. Acta 1884, 765, 217-227. (11) Takanashi, S.; Matsunaga, I.; Tamura, 2. J . Vitaminoi. 1870, 16, 132-136. (12) Lelnert, J.; Simon, I.; Hoetzel. D. Inf. J . Vitam. N u h . Res. 1983, 53, 156-165.
Three FIA methods which permit the determination of Py and Py5P over a wider concentration range are proposed. The sequential method has a low determination limit due to the fact that this configuration does not involve splitting of the sample plug, as is the case with the other two configurations, allowing its application to serum samples. This determination has the following features: The pretreatment of the sample is very simple, in contrast to other nonchromatographic methods in the literature for determination of both compounds (10). The sampling rate (30 samples per hour) is much faster than for other methods suggested for the determination of one of these substances in biological samples (32 samples per day (12)). The recovery is similar to the values found in the literature (9, 11).
LITERATURE CITED
RECEIVED for review February 25,1985. Accepted April 29, 1985.
Characterization of a Bovine Serum Reference Material for Major, Minor, and Trace Elements Claude Veillon,* Susan A. Lewis, Kristine Y. Patterson, Wayne R. Wolf, and James M. Harnly Beltsville Human Nutrition Research Center, U S . Department of Agriculture, Beltsville, Maryland 20705
Jacques Versieck, Lidia Vanballenberghe, and Rita Cornelia Department of Internal Medicine and Institute for Nuclear Sciences, University of Ghent, Ghent, Belgium
Thomas C. O’Haver Department of Chemistry, University of Maryland, College Park, Maryland 20742
A bovine serum pool was collected, homogenized, and allquoted under carefully controlled condltlons to mlnlmlze contamination wlth trace elements. The composRlon was establlshed for metals present at hlgh levels (Na, K, Ca, Mg), low levels (Fe, Cu, Zn),and trace levels (AI, Co, Cr, Mn, Mo, NI, Se, V), by the authors and 12 collaboratlng laboratorles, uslng several analytlcal methods. Excellent agreement was obtalned In almost all cases, permtltlng the authors to suggest “recommended” concentrations and estlmated uncertalntles for these 15 metals. Matrlx components, physical properties, and trace dement levels of bovine serum are very slmllar to those of human serum. The AI, Co, and Mn levels are slightly hlgher, the Mb level substantially hlgher, and the Se level lower than those usually observed In human serum. This material will be made available to the sclentlflc community through the Natlonal Bureau of Standard’s Office of Standard Reference Materlals as Reference Material 8419.
There is currently a great deal of interest in determining trace elements in biological samples. In addition to the 15 or so trace elements known or believed to be essential to man
and animals and thus of interest to nutritionists, medical practitioners, and veterinarians, several others are of interest from the standpoint of toxicology. In investigating the role of trace elements in human nutrition, diseases, and toxicology, one is usually limited to readily accessible samples such as hair, nails, perspiration, saliva, urine, feces, and the various blood components (whole blood, plasma, serum, platelets, red cells, etc.). These various biological materials can present some serious challenges during the analytical procedures. Serum, for example, contains high levels of both proteins and inorganic salts. Often the organic matter must be destroyed (without contaminating the sample) and the remaining salts either removed or compensated for if they cause any matrix effects during the analysis. Numerous methods and procedures have been used to measure trace elements in these biological materials. However, in many cases, the reported values for the same element in the same substances vary over a wide range. This is often indicative of sample contamination, but in many cases it is also due to inaccuracy of the analytical method. Many reported methods and procedures do not include adequate verification of their accuracy (I). One means of accuracy verification is to analyze the same sample or material for the same element by two independent
0003-2700/85/0357-2106$01.50/00 1985 American Chemical Society