ANALYST MAY 1986 VOL. 111 53 1 Indirect Micro-scale Method for the Determination of Desferrioxamine and its Aluminium and Iron Chelated Forms in Biological Samples by Atomic Absorption Spectrometry with Electrothermal Atom isation Pierre Allain Yves Mauras Guilene Beaudeau and Philippe Hingouet La bo ra to ire de Pha rm acolog ie e t Toxic0 log ie Cen tre H ospita lie r Reg iona I e t U n ive rsita ire d 'A ng ers, 49040 Angers Cedex France An indirect micro-scale method is described for the determination of desferrioxamine (DFA) itself and in its iron and aluminium complexed forms (FeA and AIA) in blood plasma and urine. AIA FeA and DFA after its conversion into FeA with an excess of iron are selectively extracted with benzyl alcohol and iron and aluminium are determined in the benzyl alcohol extract by electrothermal atomic absorption spectrometry using partition pyrolytically coated graphite tubes with a cuvette to prevent benzyl alcohol from spreading to the tube extremities.The specificity of the method was assessed and the sensitivity is sufficient for the determination of DFA FeA (44 pg 1-1) and AIA in blood plasma and urine after the administration of desferrioxamine as Desferal. However because of the wide range of concentrations observed in biological samples (from 1 to 300 rng I-' of DFA) the use of two calibration graphs and sometimes a preliminary dilution of the high-concentration samples are required. The use of this indirect micro-scale method for more than 1000 assays without particular problems has allowed the study of DFA pharmacokinetics.Keywords Desferrioxamine determination; aluminium and iron complexes; atomic absorption spectrometry; electrothermal atomisation; biological samples Desferrioxamine (DFA) a well known iron chelator has been used as an aluminium chelator in dialysed patients following the work of Ackrill et al.1 Many papers have reported the efficiency of DFA for Fe and A1 removal in patients but the pharmacokinetics of DFA itself and of its Fe and A1 complexed forms which are referred to as ferrioxamine (FeA) and aluminoxamine (AlA) respectively are not yet well known and there is as yet no suitable method for their determination. The spectrophotometric method of Meyer-Brunot and Keberle2 which has a detection limit of 5 mg 1-1, lacks sensitivity and cannot be adapted to the determination of A1A.High-performance liquid chromatography is a promising method but the results obtained by Cramer et aZ.3 do not allow the determination of DFA A1A and FeA in biological samples although an application of this method to biological samples has recently been described.4 In this paper a micro-scale method consisting in the specific extraction of DFA in complexed forms i.e. FeA and A1A with benzyl alcohol (as used by Meyer-Brunot and Keberle2) is described but instead of using spectrophotometric detec-tion for FeA the concentration of Fe or A1 is measured by electrothermal atomic absorption spectrometry. This indirect method has been adapted to the determination of DFA AIA and FeA at therapeutic concentrations in blood plasma or urine.Experimental The instrument used was a Varian Model 975 atomic absorption spectrometer with a GTA 95 graphite furnace and an autosampler. Partition pyrolytically coated graphite tubes with a cuvette (Varian 63.100008.00) were always used to prevent the spreading of benzyl alcohol to the tube extremi-ties. To improve the benzyl alcohol delivery in the graphite tube 5 1-11 of ethanol taken from the modifier beaker of the programmable sample dispenser were automatically dis-pensed with the blank standards and samples. The main instrument settings are indicated in Table 1. Background correction by a deuterium lamp was always used. Reagents For the conversion of DFA into FeA and AIA iron(II1) chloride solution (0.01 M) and aluminium chloride solution (0.02 M) respectively were used.Instead of using solid sodium chloride as used by Meyer-Brunot and Keberle,* a saturated solution of ammonium nitrate (Prolabo) dissolved in Merck buffer (pH S) giving a solution with a final pH of about 7 was used. DFA as the methanesulphate C26H52N6011S (relative molecular mass 656) was obtained from Ciba Laboratories. Table 1. Main instrument settings for the determination of FeA and AIA Fe (FeA) determination at A1 (AIA) determination at Parameter Low levels High levels Low levels High levels Wavelength/nm . . . . 248.3 248.3 309.3 309.3 Furnace temperature/"C: Charring . . . . . . 700 700 1000 1000 benzylalcohol/yl . . . . 10 5 30 10 1min-l . . . . . . 0 0.5 0 0 Atomisation . . . . 2400 2400 2500 2500 Injected volume of Argon purge gas flow-rate 532 ANALYST MAY 1986 VOL.111 Eppendorf polypropylene microtubes of 1.5 ml were used for the extraction and polypropylene solvent-resistant cups for the autosampler. Procedure For each sample of plasma or urine to be analysed the following two methods were used. Determination of A1A and FeA already present in biological samples To 100 pl of blood plasma or urine in an Eppendorf microtube, 200 p1 of saturated NH4N03 (pH 7) and 750 pl of benzyl alcohol were added. The tube was shaken for 60 s on a Thermolyne apparatus and centrifuged at 10 000 g for 5 min. Then 500 pl of benzyl alcohol were transferred into a cup of the autosampler for the determination of A1 and Fe by graphite furnace atomic absorption spectrometry.Determination of FeA and DFA The same procedure was used except that 25 pl of 0.01 M iron(II1) chloride were added to convert DFA into FeA [lo0 p1 of plasma 25 pl of 0.01 M iron(II1) chloride 200 p1 of NH4N03 and 750 pl of benzyl alcohol]. Standards for the preparation of calibration graphs were prepared by adding to normal blood plasma or urine known amounts of DFA (1.25, 2.5 5 and 10 mg 1-1 low concentrations graph and 12.5 25 and 50 mg 1-1 high concentrations graph). DFA was then converted into FeA by the addition of 25 pl of 0.01 M iron(II1) chloride solution or into AlA by the addition of 25 pl of 0.02 M aluminium chloride solution. The extraction procedure was used as described. The concentrations of FeA and A1A already present in biological samples were measured directly but the concentra-tion of DFA was calculated by taking the difference between the FeA + DFA concentration (after FeCI3 addition) and the FeA concentration.Results Fig. 1 shows the absorbance signals obtained for DFA after its conversion into FeA or AlA by increasing the amounts of FeCI3 or AIC13 added in a constant volume of 25 pl to a 100-p1 blood plasma sample containing 100 mg 1-1 (152.4 prnol l-1) of DFA. Concentrations of 0.01 M of Fe and 0.02 M of Al, sufficient to obtain maximum signals were used for the I I I I Fig. 2 shows typical calibration graphs obtained after the addition of DFA to normal blood plasma at low concentra-tions (2.5,5 and 10 mg 1-1) and high concentrations (12.5,25, 50 and 100 mg 1-1) and its conversion into FeA or AIA subsequently extracted in benzyl alcohol.The absorbances of low and high concentrations of FeA are approximately the same (Fig. 2) because a lower volume of benzyl alcohol was injected and a flow of argon gas was used to reduce the sensitivity for high concentrations (Table 1). For FeA measurements a curvature of the two calibration graphs is observed above 25 and 2.5 mg 1-1 respectively whereas for AIA measurements the response is almost linear up to 10 mg 1-1. We tried to cover the largest scale of DFA and FeA concentrations found in biological samples using the same extraction procedure but under two different sets of condi-tions (Table 1). However when the DFA or FeA concentra-tions were higher than 25 mg 1-1 a preliminary dilution (1 + 2 to 1 + 10) of the biological sample was necessary.The detection limit calculated as equivalent in concentra-tion to twice the standard deviation of the absorbance signal of a sample of blood plasma without DFA was equivalent to 95 pg 1-1 of DFA. The sensitivity defined as the concentration that produces a 0.0044 absorbance for iron was equivalent to 44 pg 1-1 or 0.07 pmol l-1 of DFA. The reproducibility of the method tested by ten successive assays of the same sample, was 2.5 and 8% for DFA concentrations of 12.5 and 1.25 mg 1-1 respectively. The recovery studied by adding 1.25, 12.5 and 25 mg 1-1 of DFA to different biological samples and measuring their concentrations with respect to the calibration graphs gave an average very close to 100%.The DFA concentrations measured in the blood plasma of patients extended from 1 to 300 mg 1-1 (1.5 to 457 pmol 1-I), depending on the doses given (10-80 mg kg-1 body mass) the route of administration (intravenous or intramuscular) and the time of blood sampling. Discussion DFA is more soluble in water than in benzyl alcohol but FeA and AlA are more soluble in benzyl alcohol than in water. This indirect micro-scale method based on the selective extraction of FeA and A1A with benzyl alcohol and the determination of the metal extracted by electrothermal atomic absorption spectrometry is applicable only if Fe3+ and Al3+ added to the sample are not found in the benzyl alcohol phase in the absence of DFA. After addition of 25 p1 of 0.01 M FeC13 or 0.02 M AlC13 0.6 al 2 0.4 e z Q a 0.2 0 25 50 100 Graph C 0.0025 0.005 0.01 0.02 Concentration of AICIJ or FeCIdM Fig.1. Effect of 25 p1 of different concentrations of A1Cl3 and FeC13 on the conversion of 100 mg 1-1 (152 pmol l-1) of DFA into A FeA and B AlA under the conditions described in the text 1 I 1 J Graphs A and B 2.5 5 10 DFA concentration in blood plasma/mg I-' Fig. 2. Calibration graphs for A AIA; and B and C Fe ANALYST MAY 1986 VOL. 11 1 533 solutions to 100 pl of blood plasma or urine without DFA, FeA and AlA we found at an acidic pH of about 2 very small amounts of Fe and A1 in the benzyl alcohol phase but not at pH 5 or higher. Therefore to guarantee optimum conditions of pH 200 p1 of a buffered solution of NH4N03 at pH 7 were added.Under these conditions all the blanks (plasma urine without DFA AlA and FeA) gave similar absorbance values to those obtained with benzyl alcohol. Moreover we checked that substances such as citrate ascorbate and EDTANa2, added to the sample in high concentrations (250 mg 1-I) did not modify the blank value. Hence to our knowledge DFA is the only molecule present in blood plasma and urine able to transfer Fe and A1 into the benzyl alcohol phase and the method described can be considered as specific. In addition to Fe and Al DFA can form a soluble complex in benzyl alcohol with vanadium. However when present at only very low levels in biological samples vanadium does not interfere in the determination of DFA. However if necessary it is also possible to measure indirectly the concentration of the vanadium DFA complex.It is interesting that human serum transferrin also binds vanadium.5 This indirect micro-scale method with a sensitivity of 44 yg 1-1 for DFA is about 100 times more sensitive than the spectrophotometric macro-scale method of Meyer-Brunot and Keberle2 which is based on an FeA molecular determina-tion at 430 nm in which the smallest amount that can be determined with certainty is 5000 pg 1-1. The sensitivity of this indirect micro-scale method could be improved by increasing the volume of benzyl alcohol injected into the graphite furnace or by automatic multiple injections. In contrast the sensitivity could be decreased by reducing the amount of biological samples by a preliminary dilution by increasing the volume of benzyl alcohol used for extraction and by increasing the argon gas flow-rate during atomisation.Therefore the proposed analytical conditions are a compromise trying to cover the large range of DFA and FeA concentrations found in biological samples after DFA (Desferal) administration. It is necessary to emphasise that the use of graphite tubes with a cuvette is essential to prevent benzyl alcohol from spreading to the tube extremities giving very high background signals and non-reproducible specific signals a problem encountered with other organic solvents.6 We have been using this indirect method for 1 year and more than 1000 biological samples have been analysed without particular problems. The results of the study of desferrioxamine pharmacokin-etics in healthy subjects and patients with renal failure on haemodialysis and with haemochromatosis will be published elsewhere. We are indebted to the Fondation Langlois for its support and Mrs. Laisne for typing the manuscript. References 1. 2. 3. 4. 5. 6. Ackrill P. Ralston A. J . and Day J. P. Lancet 1980,2,692. Meyer-Brunot H. G. and Keberle H. Biochem. Pharmacol., 1967 16 527. Cramer S. M. Nathanael B. and Horvath C. J . Chroma-togr. 1984 295 405. Kruck T. P. A. Kalow W. and Crapper McLachlan D. R., J. Chromatogr. 1985,341 123. Harris W. R. and Carrano C. J. J. Inorg. Biochem. 1984, 22 201. Allain P. and Mauras Y. Anal. Chim. Acta 1984 165 141. Paper A5141 7 Received November 12th 1985 Accepted December 1 Oth 198