457 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 Determination of Nickel in Serum of Haemodialysed Patients by Means of Electrothermal Atomic Absorption Spectrometry With Deuterium Background Correction* Marina Patriarcat and Gordon S. Fell Department of Pathological Biochemistry University of Glasgow Royal Infirmary Glasgow G4 OSF UK In this paper an improved method for the determination of Ni in serum by means of electrothermal atomic absorption spectrometry (ETAAS) with deuterium background correction is described. Analysis was performed after a 1 + 1 dilution of the serum samples with a solution containing 1 O/O v/v HN03 and 0.25% v/v Triton X-1 00. Aqueous Ni standard solutions were used for calibration. Sensitivity and accuracy were comparable to those reported for procedures based on Zeeman-effect background corrected ETAAS. The characteristic mass was 14 pg and the average analytical recovery was 101.5+4.8% (n=9).The analysis of Seronorm Trace Element Control Serum yielded a value of 3.21 f 0.1 7 1-19 I-' (n = 10) as compared with the recommended concentration of 3.2 pg I-'. For the Standard Reference Material Bovine Serum SRM 8419 from the National Institute of Standards and Technology a value of 0.46 & 0.05 1-19 I-' (n =4) was determined by this method in agreement with recent findings of other researchers. Improved precision (5.8% within-day 7.5% between- day) and detection limit (0.15 pg I-') in comparison with a previously reported procedure based on ETAAS with deuterium background correction was obtained. Serum Ni levels determined by this method in 25 haemodialysed patients ranged from 0.7 to 4.0 pg I-' [mean & standard deviation (SD) 2.4 f 0.9 pg I-')] and in eight women with normal renal function from 0.3 to 0.9 (mean f SD 0.5 0.2 pg I-').Keywords Nickel; serum; haemodialysis; atomic absorption spectrometry; deuterium background correction Hypernickelaemia has been reported to occur in pathological conditions such as acute myocardial infarction,l- rheumatoid arthritis4 and in subjects undergoing regular haem~dialysis.~-' Nixon et al.' found that the average concentration of Ni in serum of haemodialysed patients was 6.38 & 3.36 pg l-l in comparison with 0.14 k0.09 pg 1-1 in healthy controls. Although the determination of such low levels of Ni in serum required sample preparation and analysis to be carried out in a class 100 environment,' in the experience of the present workers the investigation of the much higher Ni concentrations found in haemodialysed patients can be carried out under less strict conditions.In healthy subjects Ni is rapidly excreted in urine with bile hair and sweat probably playing a minor role." The mechan- ism of Ni clearance from the blood is still unknown although Glennan and Sarkar" have suggested that it could involve an equilibrium between the complexes of Ni with albumin and low relative molecular mass compounds mainly L-hystidine. The observation that patients with chronic renal failure main- tained on haemodialysis have hypernickelaemia indicates that Ni is not completely removed from the body by this treatment and could accumulate in bone and tissues.Nickel is a toxic substance which has been recognized as a carcinogenic agent and a cause of cutaneous and systemic hypersensitivity in man. Acute Ni intoxication and allergic reactions have been reported after dialysis with contaminated fluids.12,13 Hopfer et aL5 highlighted the similarity between a number of disorders observed in patients undergoing long-term haemodialysis and the effects observed in rodents after parenteral administration of NiC1,. These included lipid per~xidation,'~ impaired cellular and humoral imm~nity'~-'~ and hyperpr~lactinaemia.~'*~' At present electrothermal atomic absorption spectrometry (ETAAS) is the simplest and most reliable technique for the determination of Ni in biological fluids.Analytical procedures using Zeeman-effect background correction (Z-ETAAS)'Y~~~~ have shown better performance in comparison with methods * Presented at the XXVIII Colloquium Spectroscopicum Inter- nationale (CSI) Post-Symposium on Graphite Atomizer Techniques in Analytical Spectroscopy Durham UK July 4-7 1993. 7 On leave from the Laboratorio di Biochimica Clinica Istituto Superiore di Sanita viale Regina Elena 299 00161 Roma Italy. based on ETAAS with deuterium background correction (D,-ETAAS).6 However the determination of Ni in serum by means of D,-ETAAS with instrumentation of more recent design has not been reported. Such information could be useful to routine laboratories for applications in clinical and occupational toxicology.In this paper the performance of a method for the determi- nation of Ni in serum using D,-ETAAS is described and its application to the assessment of serum Ni concentrations in a group of patients with chronic renal failure maintained on haemodialysis is reported. Experimental Instrumentation Determinations of Ni were carried out with a Perkin-Elmer atomic absorption spectrometer Model 1100B with deuterium arc background correction a graphite furnace Model HGA-700 an autosampler Model AS-70 and an Ni hollow cathode lamp. Signals were recorded by a built-in computerized system. Pyrolytic graphite coated furnace tubes were also obtained from Perkin-Elmer. Reagents An Ni stock solution 1 g 1-1 (SpectrosoL grade) Triton X-100 and ultrapure HNO (65% Aristar grade) were all obtained from BDH Poole UK.Nitric acid was further purified by sub-boiling in poly(tetrafluoroethy1ene) bottles. Ultrapure water was obtained by a four-stage purification using ion exchange (Elgastat UHP Elga High Wycombe UK). Working standard solutions containing 0 2.5 5 10 and 20 pg 1-1 of Ni were prepared in 1% v/v HNO,. Contamination Control All plastic-ware (k tubes Pasteurs and AAS cups) were soaked overnight in 20% HNO rinsed thoroughly six times with de-ionized water dried in a laminar flow hood and stored in clean plastic bags until use. Pipette tips were rinsed three times with 20% HNO and de-ionized water before use.458 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 Manipulation of samples was carried out in a laminar flow hood.Subjects Blood samples were obtained from 25 patients with end-stage chronic renal failure [ 18 men and seven women aged from 22 to 78 years mean &standard deviation (SD) 53 & 17 years] who had been treated by haemodialysis three times a week at Glasgow Royal Infirmary for an average of 42+59 months (range > 1-290 months). Dialysis was performed with equipment from Gambro Dialysatoren (Munich Germany) and Fresenius AG (Munich Germany) using capillary flow dialysers with Cuprophan mem- branes (Baxter Healthcare Thetford Norfolk UK CF ST15 membrane surface 0.9m2 and CF ST23 membrane surface 1.25 m2) or cellulose acetate hollow fibre dialysers (Baxter CA 150 membrane surface 1.5 m2) Conventional electrolyte concentrate solutions manufactured by Gambro and Fresenius respectively were diluted 1 + 34 with water purified by reverse osmosis.Serum Ni levels were also measured in a control group of eight women with normal renal function aged from 29 to 73 years (mean -t SD 48 f 19 years) who were receiving total parenteral nutrition (TPN) and were being monitored for essential trace element status. Blood Collection Blood samples were obtained before dialysis directly from the intra-arterial cannula. The first 10 ml of blood were collected for routine analyses another 10 ml aliquot was withdrawn in a plastic tube allowed to clot for at least 1 h and centrifuged at 2500 rev min-l for 10 min. Serum was transferred into a clean 5 ml plastic tube using a poly(propy1ene) pipette and stored at - 20 "C. Post-dialysis samples were obtained with the same procedure from 12 patients (nine men three women aged 53_+21 years range 23-78 average time on dialysis 37f20 months range 8-67 months).Blood samples from the TPN subjects were taken using a plastic intravenous (IV) cannula. Procedure Aqueous standards and serum samples were diluted 1 + 1 with a solution containing 1% v/v HNO and 0.25% v/v Triton X-100. A volume of 5 0 ~ 1 was injected into the furnace. All standards and serum samples were analysed in duplicate. The instrumental conditions were wavelength 232.0 nm; slit 0.2 nm; integration time 4 s; and lamp current 15 mA. Signals were measured as peak area. The graphite furnace programme is reported in Table 1. Concentrations of Ni in the samples were obtained by comparison with a calibration curve obtained from the absorbance of the aqueous standard solutions.Table 1 Graphite furnace temperature programme Temperature/ Step "C 1 100 2 150 3 200 4 1200 5 2600 6 2700 Ramp time/ Hold time/ 1 1 50 5 30 5 80 50 0 4 1 3 S S Gas flow/ Read/ ml min" s 300 - 300 - 300 - 300* - 300 0 -0.5 - * Gas flow = 0 for the last 5 s in this step. Additional Analysis Serum albumin was determined by the Bromocresol Green method with an Olympus automatic analyser. Results and Discussion Analytical Performances of Method The plot of absorbance versus Ni concentration was linear within the range 0-20 pg I-'. Using a prolonged ashing time the background signal observed during the analysis of serum samples was maintained below 0.150 A s and could be managed efficaciously by the deuterium background corrector (Fig.1). A long ramp time was found to be necessary in order to avoid the build-up of carbonaceous residues. Under these conditions matrix interferences were reduced and a calibration graph obtained with aqueous Ni solutions could be used for quantifi- cation but only when absorbance was measured as peak area. The plots of peak area versus Ni concentration obtained with either aqueous or serum-based standard solutions were parallel (Fig. 2). No significant difference was observed between the concentrations of 38 serum samples within the range 1.5-4.5 pg l-' determined using both aqueous and serum- based calibration standards (paired-data t-test average differ- ence -0.06+0.13 pg 1-l). On the contrary the peak-height response measured for the determination of equal amounts of Ni was higher for serum-based standards than for aqueous solutions (regression line equations aqueous solutions y = 1.3 x 10-3+6.30 x 10-3x r2= 1.000; serum-based standards y = 22.7 x The detection limit (three times the standard deviation of ten replicate measurements of the blank) was 0.15 pg lel equivalent to 0.3 pg I-' in the undiluted sample.The character- + 6.67 x 10-3x r2 = 0.999). 0.22 u) 0.20 E 2 si .$ 0.18 0.16 P X 4 0.14 0.12 I I I I I 40 60 80 100 120 140 Ashing time/s Fig. 1 Background signal observed during the analysis of a serum sample obtained from a haemodialysed patient using increasing total ashing time. All other furnace conditions are as in Table 1 -$ 0.10 ; 0.08 s 0 % 0.06 5 0.04 E 0.02 u 01 - 0 2 4 6 8 1 0 Ni concentration/pg I-' Fig.2 Calibration graphs for A aqueous solutions y = 0.2 x lop3 + 7.783 x 10-3x r2 =0.998; and B serum-based standards y=24.9x 10-3+7.823 x x r2=1.000JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 459 istic mass (mass of analyte in pg that yields a signal of 0.0044 A s) was 14 pg. Precision was evaluated as the relative pooled standard deviation of replicate measurements of serum samples. To determine within-day precision 35 serum samples (average concentration 3.6+ 1.8 pg 1-l) were analysed in duplicate on the same day. Between-day precision was evaluated from the replicate analysis of 14 serum samples (average concentration 3.5 & 1.7 pg 1-') carried out on two different days. The pooled standard deviation (PSD) was calculated in both cases with the following formula i = n where xil and xi2 are the first and second measurement on the ith sample and n is the number of samples.The relative pooled standard deviation was then obtained expressing the calculated PSD as a percentage of the average concentration of the n samples analysed on each occasion. Within-day precision was 5.6% and between-day precision was 7.5%. The average recovery of known amounts of Ni (2.5 5.0 and 10.0 pg 1- ') added to a serum sample was 101 + 4.8% (n = 9). The analysis of Seronorm Trace Element Control Serum (Nycomed AS Diagnostics Oslo Norway 3.2 pg 1-l) and Standard Reference Material (SRM) 8419 Bovine Serum (National Institute of Standards and Technology Gaithersburg MD USA 1.8 pg 1-l) by this method gave average values of 3.21 k0.17 (n= 10) and 0.46k0.05 pg 1-1 (n = 4) respectively.A similar discrepancy with the certified value of the SRM 8419 has been reported by two other groups of (Table 2) who suggested that the recommended value of 1.8 pg 1-1 could be in error. According to the data in Table2 where the performances of this and other ETAAS methods are summarized the pro- posed method compares well with Zeeman ETAAS procedures in terms of characteristic mass precision and accuracy. It shows better precision and detection limit than those previously reported for D,-ETAAS using older instrumentation,6 although Drazniowsky et aL6 reported a much lower value for the characteristic mass. On the other hand the detection limit is higher than those reported for procedures that apply Zeeman- effect background correction and attempts to improve the sensitivity using a multiple injection failed because of the unmanageable increase in background signal.Therefore the normal concentrations of Ni in serum of unexposed subjects cannot be determined by this method. However the measurement of such low levels requires specialized facilities for the control of contamination as described by Nixon et al.,' and is confined to a limited number of research centres. The proposed method using widely available instrumen- tation can be applied by routine laboratories to monitor Ni exposure in clinical and occupational toxicology. Ni Levels in Serum of Haemodialysed Subjects and Controls The average serum Ni concentration measured by this method in 25 patients maintained on haemodialysis was 2.4 & 0.9 pg 1-l (range 0.7-4.0 pg 1-l).In comparison serum Ni levels in eight women who were receiving TPN but had normal renal func- tion ranged between 0.3 and 0.9 pgl-' (meanfSD 0.5 kO.2 pg 1-l). The distribution of the observed values is reported in Fig. 3. Although the control group could in theory be exposed to Ni present as a contaminant in nutrient solutions Berner et ~ 1 . ~ ~ have observed that the daily intake of Ni received by patients maintained on TPN is comparable to the amounts reported to be absorbed through the gastrointestinal tract in healthy subjects. This amount is rapidly eliminated by the kidneys provided that the renal function is not affected.The values observed for serum Ni concentrations in the TPN Serum Ni intervals (upper limit)/pg I -' Fig. 3 Distribution of the serum Ni concentrations in haemodialysed patients (filled bars) and controls (open bars) Table 2 Analytical performance of this method compared with those of previously reported procedures Parameter Background correction Sample pre-treatment Injection Calibration Detection limit*/pg 1-' Characteristic masst/pg Precision Within-day RSD(%)f Between-day RSD(Y0) Seronorm (3.2 pg 1-') SRM 8419 (1.8 pg 1-') Average recovery (YO) Range Accuracy Reference 20 Zeeman Deproteinization Single Aqueous standards 0.05 27.9 3.8 8.1 - 97 _+ 2.7 94-103 6 Deuterium Dilution (1 + 1) Single Serum-based standards 0.9 6.3 6.3 34.9 - - 99+6 86-118 21 Zeeman Dilution (1 + 1) Single Serum- based standards 0.09 13 2.93 & 0.34 0.48 f 0.04 - 8 Zeeman Dilution ( 3 + 1) Multiple Serum-based standards 0.06 11 3.2 - 3.30 k 0.23 0.50 - This method Deuterium Dilution (1 + 1) Single Aqueous standards 0.15 14 5.6 7.5 3.21 k0.17 0.46 + 0.05 101.5k4.8 92-107 ~~ ~~~ ~ ~ ~ ~ ~~ * Three times the standard deviation of the blank value except for ref.10 where the minimum Ni concentration detectable with 95% confidence 7 Mass of Ni which gives a signal of 0.0044 A s. $ RSD relative standard deviation. limits is reported.460 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 Table 3 Reported concentrations of Ni in serum of haemodialysed patients and controls (mean+SD/pg 1-') Reference Controls n Patients n Hopfer et al.ref. 5 1985 0.3 f 0.2 30 5.4k2.1 65 Drazniowsky et al. ref. 6 1985 1.0 (0.6- 1.4)* 71 7.4 (6.0-9.1)* 16 Wills et al. ref. 7 1985 0.44 k 0.1 8 18 3.71 1.54 28 Hopfer et al. ref. 9 1989 0.6 & 0.3 22 7.0 k 2.4 30 Nixon et al. ref. 8 1989 0.14 f 0.09 38 6.38k0.18 27 This work 1993 0.5 f 0.2 8 2.4 k 0.9 25 ~~ ~~~ * Median (lower - upper quartile). Table 4 Reported concentrations of Ni (meanfSD/pg 1-') in serum of haemodialysed patients before and after a dialysis treatment Reference Hopfer et ul. ref. 5 1985 Hopfer et al. ref. 9 1989 This work 1993 Pre-dial ysis Post -dialysis Difference n P* 6.2+ 1.8 7.2 2.2 1.Of 1.1 40 <0.01 3.9 f 2.0 5.2 f 2.5 1.3 f 1.0 9 <0.01 3.0-t 1.3 3.7 & 1.3 0.7 f 0.3 10 <0.01 7.0 f 2.4 8.5 k 2.8 1.5 f 1.3 30 ~ 0 . 0 1 2.2 f 0.9 3.0 & 0.9 0.8 k 0.9 12 <0.01 * Versus pre-dialysis values by paired-data t-test.subjects are comparable to those reported for healthy volun- teers by other workers except for Nixon et aE.* (Table3). However because Ni concentrations lower than 0.3 pg 1-1 are not detectable with this method the average value of Ni levels in the control group could be slightly overestimated. Serum Ni levels did not appear to correlate with the length of dialysis. The average serum Ni concentrations observed in this study for sub-groups of patients maintained on dialysis for less than 10 months (n=9) 10-49 months (n= lo) 50-88 months (n = 5) and 290 months (n = 1) were 2.2 _+ 1.1 2.5 f0.9 2.6f0.3 and 1.9 pg 1-' respectively. Other workers5.* have also reported that in their groups of haemodialysed patients only those who had started the dialysis treatment less than 13 months ago had slightly lower serum Ni concentrations com- pared with the others.The serum Ni levels observed for haemodialysed subjects in this study are lower than those reported by other (Table 3). This could reflect the improvement in the purity of water and electrolyte concentrate solutions now used. The Ni concentration in samples of dialysis fluid collected just before and after the dialyser from five dialysis sets was lower than the detection limit of the proposed method (0.45 pg 1-l for samples diluted 1 + 2). Hopfer et aL5 observed a significant reduction in serum Ni levels of a group of haemodialysed patients six months after the reduction of the Ni content of the dialysis fluid (from 0.82 to 0.53 pg1-l) owing to the introduction of a new reverse osmosis system for purification of the water.The analysis of serum samples obtained on the same day from 12 patients before and after dialysis yielded average Ni concentrations of 2.2 f 0.9 and 3.0 f 0.9 pg l-l respectively ( p < 0.01 paired-data t-test). The average increase in serum Ni concentrations after a single treatment of dialysis was only slightly lower than those reported by Hopfer and co-worker~~.~ (Table 4) despite the lower serum Ni concentrations observed in the present group of subjects. On the contrary the increment of serum albumin in post-dialysis specimens (pre-dialysis value 41 * 3.5 g 1-l; post-dialysis value 45 _+ 5.6 g I-l; average differ- ence 4 * 5 g l-') owing to haemoconcentration was compar- able to the values of 10 and 8% observed by Hopfer and co-worker~.~,~ Although statistically significant the increase in Ni in the present group of subjects was very variable and did not correlate with the increment of serum albumin or with pre- dialysis serum Ni values.In addition two patients had reduced serum Ni levels after dialysis despite the rise in albumin concentrations. A large variability could also be observed from the data reported by Hopfer and c o - ~ o r k e r s ~ . ~ (Table 4). In contrast when individual post-dialysis serum Ni levels were corrected for haemoconcentration using the serum albumin values the average concentration was 2.7 kO.7 pg 1-1 and the increase versus pre-dialysis values (mean f SD 0.5 k 0.8 pg 1-I) was lower and not significant by the paired-data t-test.Despite the improvement in the purity of dialysis fluids hypernickelaemia although moderate still occurred in the present group of haemodialysed patients. Olerud et a2.13 have demonstrated in vitro that Ni can be absorbed into the blood from dialysis fluids against a concen- tration gradient owing to the high affinity of albumin and other plasma proteins for Ni. However the exchange of Ni in vitro during dialysis appears to be variable and dependent on factors that have not yet been completely clarified. Conclusions A method has been described for the determination of Ni in serum by means of D,-ETAAS that can be applied in clinical and occupational toxicology. The levels of Ni in serum of patients maintained on haemodialysis were lower than those previously reported and could reflect the 'clean-up' of haemo- dialysis systems.The authors acknowledge the assistance of the Staff of the Renal Unit Ward 12 Glasgow Royal Infirmary for the collection of blood samples from haemodialysed patients. References Howard J. M. Clin. Chem. 1980 26 1515. Khan S. N. Rahman M. A. and Samad A. Clin. Chem. 1984 30 644. Leach C. N. Jr. Linden J. V. Hopfer S . M. Crisostomo M. C. and Sunderman F. W. Jr. Clin. Chem. 1985 31 556. Milling Pedersen L. and Molin Christensen J. Acta Pharmacol. Toxicol. 1985 59 Suppl. VII 392. Hopfer S . M. Linden J. V. Crisostomo M. C. Catalanatto F. A. Galen M. and Sunderman F. W. Jr Truce Elem. Med. 1985 2 68. Drazniowsky M.Parkinson I. S. Ward M. K. Channon S. M. and Kerr D. N. S. Clin. Chim. Acta 1985 145 219. Wills M. R. Brown C. S. Bertholf R. L. Ross R. and Savory J. Clin. Chim. Acta 1985 145 193. Nixon D. E. Moyer T. P. Squillace D. P. and McCarthy J. T. Analyst 1989 114 1671. Hopfer S. M. Fay W. P. and Sunderman F. W. Jr. Ann. Clin. Lab. Sci. 1989 19 161.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 461 10 11 12 13 14 15 16 Niebor E. Tom R. T. and Sandford W. E. in Nickel and its Role in Biology. Series Metal Ions in Biological Systems eds. Sigel H. and Sigel A. Marcel Dekker New York 1988 vol. 23 Glennan J. D. and Sarkar B. Biochem. J. 1982 19 847. Webster J. D. Parker T. F. Alfrey A. C. Smythe W. R. Kubo H. Neal G. and Hull A. R. Ann. Intern. Med. 1980 92 631. Olerud J. E. Lee M. Y. Uvelli D. A. Goble G. J. and Babb A. L. Arch. Dermatol. 1984 120 1066. Giardini 0 Taccone-Gallucci M. Lubrano R. Ricciardi- Tenore G. Bandino O. Silvi I. Paradisi C. Mannarino O. Citti G. Elli M. and Casciani C. U. Clin. Nephrol. 1984,21 174. Donnelly P. K. Shenton B. K. Alomran A. M. Francis D. M. A. Proud G. and Taylor R. M. R. Proc. Eur. Dial. Transplant. Assoc. Eur. Renal Assoc. 1983 20 297. Graham J. A. Miller F. J. Daniel M. J. Payne E. A. and Gardiner D. E. Enuiron. Res. 1978 16 77. ch. 4 pp. 91-122. 17 Smialowicz R. J. Rogers R. R. Riddle M. M. and Stott G. A. Enuiron. Rex 1984 33 413. 18 Mastrogiacomo I. DeBesi L. Serafini W. Zucchetta P. Romagnoli G. F. Saporiti E. Dean P. Ronco C. and Adami A. Nephron 1984 37 195. 19 Clemons G. K. and Garcia J. F. Toxicol. Appl. Pharmacol. 1981 61 343. 20 Sunderman F. W. Jr. Crisostomo M. C. Reid M. C. Hopfer S. M. and Nomoto S. Ann. Clin. Lab. Sci. 1984 14 232. 21 Andersen J. R. Gammelgaard B. and Reinert S. Analyst 1986 111 721. 22 Berner Y. N. Shuler T. R. Nielsen F. H. Flombaum C. Farkouh S. A. and Shike M. Am. J. Clin. Nutr. 1989 50 1079. Paper 310671 OF Received November 8 1993 Accepted December 28 1993