copper, is eliminated by adding 1g of thiosemicarbazide per liter, which at a low pH ... mixed, and the tubes were covered with Parafilm, kept at 70. #{176}Cfor 15 ...
CLIN.
CHEM.
26/2,
327-331
(1980)
Improved Direct Specific Determination of Serum Iron and Total
Iron-BindingCapacity Ferrucclo Ceriotti and Giovanni Cerlotti Serum iron is released from transferrin and reduced at pH 1.7 by treating serum with a 10 gIL ascorbic acid solution in 0.1 mol/L HCI. When ferrozine is added to this reagent, it forms a complex with iron that is as intensely colored as at higher pH values, and under these conditions no turbidity is produced. The second major interference, that from copper, is eliminated by adding 1 g of thiosemicarbazide per liter, which at a low pH forms a stable, uncolored complex with copper without affecting the reaction of ferrozine with iron.
AddItIonalKeyphrases: ferrozine/thiosemicarbazide colorimetry
reagent
transferrin
Materials 3- (2-Pyridyl)-5-6-diphenyl-1,2,4-triazine, disodium salt (ferrozine; Aldrich). Ascorbic acid, iron-free (Merck). Thiosemicarbazide (Carlo Erba). HCI, iron-free (Carlo Erba). Glycine (Merck). Iron wire, 99.9% pure (Carlo Erba). Titrisol iron (Merck).
sulfonic
acid
Methods Absorption
Spectroscopy
Equal volumes of serum and of a 200 g/L solution of tnchloroacetic acid containing lOg of ascorbic acid per liter were mixed, and the tubes were covered with Parafilm, kept at 70 #{176}C for 15 mm, and then centrifuged. The supernates were analyzed for iron with a Perkin-Elmer Model 403 apparatus, with air-acetylene flame. ‘rests of reproducibility within the series and between days
Laboratorio
Centrale
di Analisi,
Ospedale Civile
Padova, Italy. Received
April
19, 1979; accepted
Sept. 18, 1979.
di Padova,
35100
recovery
experiments
proved
the reliability
of
Continuous-Flow Determination Sera were analyzed with a SMAC (Technicon) by the Technicon method prescribed for this instrument, which involves detaching the iron from transferrin, dialyzing, and complexing with ferrozine. Although no indications are given by the firm that any inhibitor of the copper reaction with ferrozine is present in the reagents, added copper did not interfere with iron determination in serum by this procedure. Proposed
Direct determinations of serum iron suffer a major drawback: the proteins may produce turbidity because of the high proportion of serum that must be used and because the chromogenic reaction is usually performed at pH’s that are near the isoelectric point of many serum proteins. Usually, surfactants have been used to eliminate turbidity, but not always with complete success, owing to the variable protein composition of sera. Our work started from the observation that when HC1 with or without a reducing agent is added to serum to release iron from its transferrin complex, no turbidity forms, and that, contrary to Stookey (1) but in accordance with Ruutu (2), the chromogenic agent ferrozine will also react with iron at low pH values. However, it reacts also with copper, giving a rather important interference (3-5). Here we report our investigations aimed at finding the best experimental conditions to avoid interference by both proteins and copper.
Atomic
and analytical this method.
Method
Solutions: In a 100-mL volumetric flask, place 100 mg of thiosemicarbazide, 1 g of ascorbic acid, and 100 mg of ferrozine, and dissolve in and dilute to volume with 0.1 mol/L HC1 (reagent 1).
For the blank, prepare the sante solution, omitting ferrozine (reagent 2). Use these solutions within 2 h, because thiosemicarbazide slowly reacts with ascorbic acid at room temperature to form a yellow thiosemicarbazone. To prepare the iron stock-standard solution, containing 1 g/L, dissolve 1 g of the iron wire in 12 mL of concentrated HC1 by gentle boiling, quantitatively transfer the solution to a 1-L volumetric flask, and dilute to the mark with water. Alternatively, the Titnisol solution from Merck can be used. For the working stock solution (100.0 mg/L), dilute the preceding solution 10-fold with a 0.2 mol/L glycine solution in 5 mmol/L HC1 (pH 4.15). Iron working-standard solutions are prepared at desired concentrations from 0.25 to 4.00 mg/L by diluting the working stock solution with the glycine buffer. For routine work a 1.00 mg/L solution is usually used. Procedures: Iron. To 2 mL of reagent 1, add 1 mL of serum, cover with Parafilm, and mix by inversion. Let stand for 10 mm. Mix again and after an additional 10 mm measure the absorbance at 562 nm vs a blank prepared by adding 1 mL of serum to 2 mL of reagent 2. The absorbance is stable for more than 2 h. For the standards, to 2 mL of reagent 1 add 1 mL of the various iron dilutions, mix, and read after 10 mm vs a blank made by adding 1 mL of the glycine solution to reagent 1. The reaction can also be performed in a single cuvette. To do so, add 1 mL of serum to 2 mL of reagent 2 and take a first reading at 562 nm vs water. Then add 100 sL of a 20 g/L solution of ferrozine, mix well by repeated inversions, and after 10 mm again measure at 562 nm the color formed vs a blank made by mixing 2 mL of reagent 2 and 100 fzL of the ferrozine solution with 1 mL of water. Total iron-binding capacity. To 1.5 mL of serum add 1.5 mL of an ammonium ferric sulfate solution in 100 tmol/L HC1, containing 5 mg of iron per liter. Mix by inversion and after at least 10 mm add 150 mg of MgCO3. Mix repeatedly and, after 30 mm, centrifuge. Using the supernate, proceed as for serum, the only difference being that the reagent is made in 0.05 instead of 0.1 mol/L HC1, to account for the halved protein content. CLINICAL
CHEMISTRY,
Vol. 26, No. 2, 1980
327
1OC
9c 8(
Q3
7C 6( 02
50 40 ,
30
0.1
20 10
0.0
i
23456789
f131Q61t25
2.5
pH Fig. 1. Effect of pH on the formation of the Fe2-ferrozine complex The following buffers (0.1 mol/L) were used: from pH 1.1 to 2.9, glyclne HCI; 3.3-3.9 phthalate; 4.2-4.9 acetate; 5.1-7.0 phosphate; 7.4-8.5trls(hydroxymethyl)methylamine. To each buffer, ascorbic acid (10 g/l) and ferrozlne (1 g/L) wereadded. The stock Iron solution was diluted to 1.50 mg/L with each buffer. To 2 ml of reagent was added 1 ml of iron solution and readings were taken at 562 nm vs a reagent blank, at zero time and at 15 and 30 mm. Readingsat 30 mm are shown here. pH was measured with a glass electrode at 25#{176}C after completion of the reaction. The color obtained at the various pH’s Is expressed as percent of the maximum color developed
Results Analytical
Variables
Investigated
Influence of pH on the complexation of Fe2+ with ferrozine (Figure 1). The influence of pH on formation of the colored complex was tested from pH 1.1to 8.5,with useof a standard iron solution of 2.00 mg/L and various buffers at 0.1 mol/L, used in the range of their respective optimum buffering capacities. The pH’s were measured with a pH meter at 25
#{176}C. Rate and intensity
of color formation are rapid and strong pH 1.65 to 4.0 the color develops information is complete in 15 mm at pH 5 and
at low pH values. From
stantly; complex 6, then the rate declines slowly up to pH 7.3, above which there is an abrupt
Complexing
decrease.
of ferrozine
with copper.
Ferrozine
also reacts
Thiourea
or thlosemicarbazid.,g,L
Fig. 3. Effect of various concentrations of thiosemicarbazide and of thiourea on formation of the complex between ferrozine and copper
or iron
Copper 26.00 mg/L plus thiourea (A - - - A) or plus thiosemlcarbazide (O-O); copper 2.60 mg/L plus thiosemicarbazide (-); Iron 1.10 mg/L plus thiourea or thlosemlcarbazkie(0-0). The experiments were performed at pH 2.0, wIth use of a 0.1 mol/L glycine buffer
with copper to form a colored complex (1, 3, 4). Although the absorption spectra for the two metals are quite different (Figure 2), interference is observed at 562 nm, the peak absorption for iron. In agreement with Duffy and Gandin (3), we observed for copper an absorbance corresponding to 9.5% of that of iron, on a weight base, at this wavelength. To avoid this interference, it has been suggested that one use bathocuproine and take readings at two wavelengths (4). Thiourea was tried with a good success, reducing color formation by copper to one-tenth (3). In our experiments, besides urea, we also tested thiosemicarbazide, with better results (Figure 3). This compound, even in low concentrations, completely prevents color formation of copper with ferrozine, probably by forming a stable uncolored complex, without affecting its reaction with iron. The affinity of thiosemicarbazide for copper is maximum at the low pH range we used in our reaction (Figure 4). Elimination of copper interference by addition of thiosemicarbazide to the reaction mixture was tested on a series of sera containing various concentrations of iron. In the absence of thiosemicarbazide, interference by copper was
03 U C
02
.0
0
U
.0
400
450
500
550
Wavelength,
600
650
700 2
nm
Fig. 2. Absorption spectra of complexes of ferrozine with copper (26.00 mg/L) (curve 1) and with iron (4.00 mg/L) (curve 2) The reaction was performed in 0.1 mol/L glycine buffer at pH 2.0 in the presence of ascorbic acid (10 gIL) and ferrozine (1 gIL). At 562 nm (the absorbance maximum for the iron complex) the absorbance of the copper complex is 9.5% that of Iron on a weight basis
328
CLINICALCHEMISTRY,Vol. 26, No. 2,
al
1980
3
4
5
6
7
8
pH Fig. 4. Effect of pH on the inhibiting action of thiosemicarbazide formation of the complex between copper and ferrozine Copper, 26.00 mg/L; thiosemicarbazlde. 1 g/L; ferrozlne, 1 gIL; ascorbIc acId, 10 g/L. At pH 7.0, 60% of the complex is formed as compared wIth that In absence of thiosemicarbazide; at pH 2.0itIsonly3.0%
Table 1. Inhibition of Copper Interference
by
Thiosemicarbazide Apparent
carbazlde
addItion
added
PosItIve error, %
0.27 0.28 0.35
33 47 20
0.47
0.37
27
0.49
29
0.74
0.55
34
0.74
0.55
0.84
0.85
0.66 0.66
34 27 28
1.00
0.91
10
1.03 1.10 1.10
0.78 0.88 0.88
34 25 25
1.10
0.94
17
1.16 1.21
1.05 1.07 1.16 1.25
10 11 16 15 9 5.6 5 3.5 8.5 0.0
1.44 1.57
1.44
1.84 1.86
1.74 1.75
2.04
1.97
2.62 2.89
2.41 2.90
Turbidity formation with serum or plasma. Under the conditions used, absorbances of serum blank vs water at 562 nm gave a mean value of 0.037 ± 0.014 (n = 135, range 0.020-0.095), mainly owing to the original color or turbidity of the serum. This blank absorption appeared to be stable for as long as 36 h. A series of 30 sera, especially chosen from among those with an altered protein picture as disclosed by
electrophoresis (cirrhotic, nephrotic, and myelomatous sera), showed the same behavior. Blank readings were somewhat high with icteric sera because of the color contribution of bilirubin, but they were equally stable in time. The same was true for lipemic sera. When plasma was used instead of serum, turbidity
20
0.48
0.013
2.6
19 21
0.76 1.75
0.021 0.024
2.75 1.40
20
0.43
0.020
4.6
21 21
1.00
0.025
2.5
0.051
3.1
1.66
serum sample to Fe2+, giving rise to the same color intensity
mias.
as higher concentrations. Influence of ferrozine concentration. The color intensifies with increasing ferrozine concentration, up to 0.5 g/L. Further increases have no effect. Test of reproducibility within the series and among series.
Results of tests of reproducibility on sera at three different iron concentrations are shown in Table 2. Addition experiments. Progressive increases in concentration from 0.25 to 2.00 mg/L were obtained by adding standard iron solutions to two pools of serum. As shown in Figure 5, the two addition curves have the same slope as the standard curve and intercept on the ordinate at the points of zero addition. The original iron concentration of sera can therefore be determined by extrapolating to the abscissa. Beer’s law is followed up to 5.00 mg/L, and this is also true if common grating or interference filter photometers are
used. Comparison
with Other Methods
We compared
the present
method
both with methods in-
09
U C 0
formed,
sometimes even a precipitate, but this dissolved rapidly on shaking. Absorbances were somewhat higher than with serum (mean: 0.110; range 0.025-0.335), but they were stable at least up to 1 h. In a few samples with high initial absorbance, a little decrease was noticed after 3 h owing to enlargement of the suspended particles, but shaking also restored the initial absorbance in these cases. Influence of HCI concentration. We tested the effect of various HC1 concentrations on color development by adding two volumes of HC1 per volume of serum as described under Methods. With concentrations of 0.3 and 0.2 mol/L, and a final pH of 1.1 and 1.3, the color was pale and developed slowly; with 0.1 mol/L (pH 1.7-1.9) the color developed rapidly,
reaching a maximum within 5 mm, and then remained stable. were obtained
%
molfL HC1. With 125 mmol/L HC1, color formation was pale and slow, and turbidity was often present. A thorough mixing by inversion, without shaking, appeared to be important for rapid and complete color development. Influence of ascorbic acid concentration. A concentration of ascorbic acid of 2.50 g/L suffices to reduce all the iron in a
constantly present, producing very important positive errors, especially at mean and low iron and high copper concentrations (Table 1). This combination of circumstances is common in severe diseases such as malignartt lymphomas and leuke-
the same results
cv,
Among-series
0.63
1.35
±SD
Within-series
0.36 0.41 0.42
Practically
Fe , mg/L
Iron concn, mg/I
ThiosemlNo
Table 2. Repr oducibility of the Assay n
with 0.75 and 0.5
0)
4
IRON, mg/L Fig. 5. Standard curve and addition experiments The standardcurve (0-0) islinear to5.00mg/L with simple instruments (Spekol-Zeiss, Jena, Germany).Two series of addition experiments (and V-V) are reported. The original absorbances of sera are coincident with the points of IntersectIon of the curves with the ordinate (‘I
tMIt”AI
(‘UIAIQTDV
‘,I,-J
OC
Mn
0
iom
9Q
-J
Table 3. Results for Serum Iron by Various Methods Compared for 26 Sera a Iron, mg/L 01 serum
I-
By At Absorp.
mg/L
Fig. 6. Results obtained by the present method compared with those by atomic absorption spectroscopy n = 102, r = 0.9944,a = -1.7789, b = 1.0275 deproteinization and with other direct methods. Atomic absorption spectroscopy was chosen for comparison according to the procedure outlined above, because it is undoubtedlyfree from interference by copper. According to Blair and Diehl (5), bathophenanthroline gives a 3% interference, on a weight base, with this cation. We assessed the iron concentration of the serum pools to be used as standards by determining their absorption after adding iron solutions of increasing concentrations and extrapolating to the abscissa, as described above for the colonmetric method. Parallel experiments by atomic absorption spectroscopy and by colorimetry gave superimposable results. Iron was determined in 100 sera that included a wide range of concentrations by the SMAC, by atomic absorption spectroscopy, volving
and by the present method. A good correlation was found between results by our method and those by atomic absorption spectroscopy for the entire range of concentrations (Figure 6). Also, the correlation with SMAC appeared to be good (n = 100; r = 0.9962; a = +12.1887; b = 0.8927); however, at concentrations below 0.60 mg/L the SMAC values were usually and randomly lower, while values for concentrations exceeding 1.60 mg/L tended to be higher than by the other two methods. The intermediate values were coincident for all the three methods. In the low and high range, values by atomic absorption spectroscopy and the direct method agreed better with the clihical picture than did those by the SMAC. In 26 additional samples we compared the present method both with SMAC and atomic absorption spectroscopy and with two commercial direct procedures (DR 1 and DR 2), both based on the method of Lauber (6), but probably differing in the tensioactive agent, which is Teepol in the kit DR 1 and an unspecified tensioactive in kit DR 2. The results are shown in Table 3. With kit DR 1 intense turbidity is frequently observed that in some cases increases steadily, giving erroneously high values. With kit DR 2, the solutions usually remain clear, but in occasional samples a rapidly increasing turbidity forms that prevents any possibility of reading (sample 1); even slight hemolysis (sample 3) interferes sharply to give unreliable results; furthermore, the values found are usually lower than those by the other methods, possibly owing to incomplete release of iron from transferrin at the pH used (pH 5.6).
Interference from Bilirubin, Hemoglobin, and Lipemia Intense bilirubinemia and lipemia gave high blank readings, CIA
1.’I
IMI(’AI
(‘U&AIOTDV
II
0
kI
)
IGQA
DRC 0.11
SMAC
AA
0.04
0.10
0.15
0.17
0.24
0.34
0.16 0.36
0.33 0.40 0.44
0.48 0.40 0.45
0.50 0.39 0.44
0.51 0.55 0.59
0.52 0.55 0.60
0.52 0.55
0.39 0.50
0.48
0.58
0.55
0.50
0.66 0.73 0.79
0.70 0.72 0.70
0.75 0.72 0.66
0.55 0.67 0.78
0.53 0.60 0.75
0.82 0.87
0.84
0.83 0.87
0.72 0,83
0.60
0.88
0.91 0.94 0.98 1,08 1.20 1.28 1.40 1.50 1.63
0.93 0.90 0.96 1.10 1.25 1.27 1.40 1.45 1.55 1.75 2.10 2.38
0.97 0.92 0.98 1.08 1.28 1.26 1.38 1.45 1.52 1.80 2.17 2.39
0.83 0.89
1.S0
2.13 2.41
OR1 0,70b
DR2
Unreadableb
0.17 0.20 0.45#{176} Unreadabled 0.45 0.40 0.45#{176} 0.35 0.45 0.45
0.40
0.70
0.75 0.75 1,33b 1.00 1.05 0.85 1.28#{176} 1.10 1.22 1.12 200b 1.25 1.34 1.25 172b 1.40 1.67 1.55 1.94 1.75 2.28 2.2
Contlnuous-flow (SMAc. Technicon); AA, atomic absorption pectroscopy; DRC, present method; DR 1 and DR 2, commercIal direct methods. b Formation of rapidly increasing turbidity. C
Formationof turbidity.
d
Siit
hemolysls.
but the direct from
iron determinations
were unaffected
the parallel results with atomic absorption
as judged
spectroscopy
and SMAC. A slight positive moglobin
started
interference (0.010 A) prom heto appear at 2.0 g of hemoglobin per liter.
Discussion the colored complex of ferrozine with iron forms at a pH as low as 1.65, the assay,can be done at a pH range where, in the presence of a reducing agent, iron is rapidly and completely released from transferrin. The optimum pH is reached simply by exploiting the buffering action of the serum proteins on dilute HC1, with formation of soluble proteinates. This pH is far from the isoelectric points of the serum proteins, and the solutions therefore remain clear, eveil without added surfactants. At this pH, thiosemicarbazide redcts specifically and completely with copper to form a stable uncolored complex, without affecting the reaction, of ferrozine with iron. Interference by copper, which could produce very important positive errors at low and normal iron concentrations, is thus eliminated. In comparison with comnionly used direct methods, blank values are usually low, depending on the original color or turbidity of the serum under test, and are always stable in time. Hemoglobin interferes little, even at rather high concentrations. Avoding the use of surfactants and of concentrated buffers makes it easier to avoid contamination with iron from reagents. Because completely
The accuracy of the method is good, as might be expected after elimination of interferences from copper and from variable turbidity. The simplification of the procedure not only accelerates the determination but also favors good reproducibility. The possibility of doing a differential reading in a single cuvette may decrease both the amount of serum and the number of manipulations required. The method, with minor modifications, is easily amenable to automation (F. Ceriotti, P. Bonvicini, and G. Ceriotti, in preparation).
References 1. Stookey, L. L., Ferrozine. iron. Anal. Chem. 42, 779-781
CLIN.CHEM. 26/2,
331-334
A new spectrophotometric (1970).
reagent for
R., Determination of iron and unsaturated in serum with ferrozine. Clin. Chim. Acta
2. Ruutu,
capacity
iron binding 61, 229-232
(1975).
J. R., and Gandin, J., Copper interference in the determination of iron in serum using ferrozine. Clin. Biochem. 10, 122-123 (1977). 4. Manasterski, A., Weiner, L. M., and Zak, B., Spectrophotometric study of the reaction of 3-(2-pyridyl)-5,6-bis(4-phenylsulfonic acid)-1,2,4-triazine with iron and copper. Microchem. J. 16, 245-252 (1971). 3. Duffy,
5.
Blair, D., and Diehl, H., Preparation, characterization and application of bathophenanthroline disulfonic acid and bathocuproine
disulfonic acid, water soluble reagents for iron and copper. Talanta 7, 163-174 (1961). 6. Lauber, K., Bestimmung von Serumeisen und eisenbindung Kapazitaet ohne Enteiweissung. Z. KIm. Chem. 3, 96-99 (1965).
(1980)
Automated Multiple Flow-Injection Analysis in Clinical Chemistry: Determination of Albumin with Bromcresol Green B. W. Renoe,1 K. K. Stewart,4 G. R. Beecher,4 M. R. Wills,2 and J. Savory3 We describe an adaptation of automated multiple flowinjection analysis instrumentation to an analysis for albumin in serum. The bromcresol green reaction was used to test the utility of the system. The approach yielded albumin results with excellent sensitivity, no measurable carryover, a relative standard deviation of less than 1 %, good correlations with published procedures, and no measurable interferences. The simplicity and flexibility of the instrumentation and its performance integrity, as indicated by the analytical results, make this a viable clinical chemical tool. Unsegmented continuous-flow analysis, commonly known as flow-injection analysis, was first described by two independent groups of workers (1-4) and has been recently reviewed by Betteridge (5). In this reaction system, the reagents and samples flow continuously (5), without bubble-segmentation. The use of small-bore tubing (0.5 mm i.d.), low-volume fittings and flow cells, and high linear-flow rates gives excellent mixing and maintains sample integrity. Readings return to baseline between samples, at rates greater than 120 samples per hour, with no loss of precision, which is in contrast to the conventional continuous-flow system of Skeggs (6). With the use of the automatic sampling system of Skeggs and a sample-insertion valve to inject measured volumes of sample into the reaction system, an automated version of flow-injection analysis-automated multiple flow-injection analysis (AMFIA) (7)-is possible and appears to offer a viable alternative for automation in the clinical chemistry laboratory. Few applications of flow-injection analysis have been made
for clinical chemical analyses. A procedure for determining CO2 in plasma was described recently (8); and some feasibility
studies for serum phosphate and chloride (9); serum potassium, sodium, and nitrate (10); and serum glucose (11) have been published. In evaluating AMFIA for serum analysis, we decided that the best test would be the assay of albumin in serum bromcresol green (BCG) reaction (12). The specificity procedure has been questioned (13), but interferences globulins can be eliminated by reading the absorbance a few seconds after initiating the reaction (14). The
min/BCG sample bility
reaction is very sensitive and thus volumes, a requirement we felt would
Nutrition
Institute,
Human
Nutrition
Center,
of Agriculture, Beltsville, MD 20705. Received ,July 5, 1979; accepted Nov. 2, 1979.
U.S. Department
of this from within albu-
requires small test the flexi-
of AMFIA.
Materials and Methods Apparatus The AMFIA instrument used has been described previously by Stewart et al. (3, 15). A flow diagram for the determination of albumin is shown in Figure 1. The timing coil after the mixing “T” can be used to control the reaction time. This albumin determination had a reaction time of 14s. The sample size was 2 tL (the minimum volume of the slide valve used for sampling). Flow rates through the system were: BCG reagent, 67 zL/s, and saline carrier, 44 1zL/s. The standards, serum pools, and samples were analyzed at 120 samples per hour. The temperature used throughout was 23 #{176}C. An interference filter with a 10-nm bandpass at 600 nm was used in the colorimeter.
Reagents Bromcresol
Departments of Pathology and Chemistry, 2 Pathology and Internal Medicine, and Pathology and Biochemistry, University of Virginia Medical Center, Charlottesville, VA 22908.
by the
and Standard5 green reagent.
This reagent
contained,
per liter,
Mention of a trademark or proprietary product does not constitute a guarantee or warranty of the product by the U.S. Department of Agriculture, and does not imply its approval to the exclusion of other products that may also be suitable. (‘I
IMI(’AI
1’I4AIQTDV
InI
0
Mn
0
1000
CCI
