JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 369 Determination of Cadmium in Blood Plasma by Graphite Furnace Atomic Absorption Spectrometry M. M. Black* and Gordon S. Fell Department of Clinical Biochemistry, Royal Infirmary, Glasgow G4 OSF, UK John M. Ottaway Department of Pure and Applied Chemistry, University of Strathclyde, 295 Cathedral Street, Glasgow GI IXL, UK A graphite furnace atomic absorption spectrometric (GFAAS) procedure is described for the determination of Cd in rat blood plasma. Blood plasma was deproteinised with nitric acid prior to injection into the furnace. Calibration was with Cd in acidic solution and absorption signals were read in the integration mode. The method has a detection limit of 1.26 pg of Cd, and an imprecision of 1-4% (relative standard deviation) at a plasma concentration of 1 l g 1-1 of Cd.Accuracy was established by quantitative recovery of Cd added to plasma and by intercomparison of results with other atomic spectrometric procedures. Plasma Cd determination and the radioimmunoassay of plasma metallothionein are useful for the study of the mechanism of Cd toxicity in animals. Eventual applications in human occupational medicine could provide a warning of Cd toxicity before irreversible renal damage has occurred. Keywords: Plasma cadmium determination; graphite furnace atomic absorption spectrometry; nitric acid dep ro teinisa tio n; signal integration Occupational and environmental exposure to cadmium is a known health hazard. Cadmium accumulates in body tissues, particularly the liver and kidneys, and has a long biological half-life.' A critical organ effect is seen in the kidney when a level of 100-200 yg g-1 of Cd (wet mass) is reached in the renal cortex.The resultant tubular and glomerular damage is observed as increased urinary excretion of protein, calcium, phosphate and other substances. The clinical consequences are varying degrees of bone disease and an increased tendency to form renal calculi. At present the laboratory monitoring of a workforce exposed to cadium dust or fume relies on the determination of whole blood Cd, urinary Cd and the detection of an excess of low relative molecular mass proteins in urine. These tests when positive indicate that a considerable body burden of Cd has already accumulated. The pathological changes induced are not then reversible, even after removal from further Cd exposure.2 Cadmium in whole blood is present in unexposed people3 at a concentration of ca.1 yg 1-1, cigarette smokers have up to 3 yg 1-1 and a proposed threshold4 for industrial workers is 10 pg 1-1. Studies of the distribution of Cd in blood show that about 90% is in the red cells and less than 10% is bound to plasma proteins. A metabolic model for Cd suggests, however, that the plasma Cd fraction is important in the distribution and uptake of the metal, especially in the transport of Cd from the liver to the kidney. Hitherto, measurement of plasma Cd has not been feasible. In unex- posed people the concentration is very low, probably less than 0.1 yg 1-1, but during environmental or occupational exposure when the whole blood Cd level is 10 yg 1-1 or greater, the plasma Cd level may be above 0.5 yg 1-1.During investi- gations of the effects of chronic Cd toxicity in the rat we have observed that the appearance in blood plasma of detectable amounts of Cd and of metallothionein coincide with early pathological changes in the kidney that may still be revers- ible.5 Metallothionein is measured by a radioimmunoassay6 that determines the protein but does not differentiate between the various metal-containing species (e.g. , between Zn - metallothionein and Cu - metallothionein). In this paper we * Present address, Robens Institute, University of Surrey, Guildford, Surrey, UK. describe the development of a graphite furnace atomic absorption spectrometric (GFAAS) method for the determi- nation of Cd in plasma using commercially available equip- ment.By measuring both plasma Cd and plasma metallo- thionein in samples obtained sequentially from rats dosed with Cd, we were able to show that the major Cd species in plasma is Cd - metallothionein, and that this is toxic to the kidney.5 If a similar mechanism occurs in humans, then the determi- nation of plasma Cd together with plasma metallothionein could be used in occupational medicine to provide an early indication of Cd toxicity before irreversible changes have occurred. Experimental Apparatus A Perkin-Elmer Model 2280 atomic absorption spectrometer was used, combined with a Perkin-Elmer HGA-500 graphite furnace equipped with a Perkin-Elmer AS-1 autosampler and a Perkin-Elmer Model 56 chart recorder.Argon was used as the purge gas. Measurements from the readout systems were manually entered into a DEC Professional 350 computer for prep- aration of calibration graphs, calculation of cadmium concen- trations and statistical analyses. Preparation of Samples and Standards Whole blood (5-10 ml) was collected from killed rats by abdominal aorta puncture in plastic tubes with heparin as anticoagulant. Plasma was obtained by centrifugation and transferred into plastic containers for storage at 4-10°C. All steps of this procedure were checked for possible Cd contamination. Plasma was deproteinised with equal volumes of acid (10% V/V Aristar nitric acid), in acid-washed plastic tubes. Samples were then allowed to stand €or 15 min at 4°C prior to centrifugation. (Refrigeration prior to centrifugation was found to give clearer supernatants.) Standards were prepared370 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 Table 1. Instrumental conditions Stage Temperature/"C Ramp/s Hold/s 120 1 20 Dry . . . . . . . . Ash . . . . . . . . 400 1 20 Atomise . . . . . . 1400 1 10GS,t= 1Os* Clean . . . . . . 2750 1 5 Injection volume . . . . . . . . 20 p1 Lamp current . . . . . . . . . . 4 mA Wavelength . . . . . . . . . . 228.88 nm Band pass . . . . . . . . . . . . 0.7 nm Operating mode . . . . . . . . . . Peak area Deuterium background correction Standard graphite tube * GS = gas stop; t = integration period. 0.3 rw Peak + area y v) a, C m $ 0.1 2 1.2 a, C m + a 0.1 2 I I I I I w 150 250 350 450 550 650 750 0- I TemperaturePC Fig.1. Effect of increasing ashing temperature on the peak-height and peak-area signals for Cd (3 pg 1-I) in deproteinised rat plasma (5% V/V nitric acid) that contained between 0 and 3 yg 1-1 of Cd in 5% V/V nitric acid in plastic containers. Any samples found to have Cd concentrations greater than 3 pg 1-1 were diluted with 5% V/V nitric acid and re-analysed. Analytical Procedure Dilution of plasma with either water or dilute Triton X-100 solution (0.1%) was ineffective in reducing the non-atomic background absorption to a level (<0.7 A) that could be compensated for by the deuterium arc background correction system. An alternative was to remove the major portion of the organic matrix, which caused smoke during the atomisation stage, before injection of the sample into the furnace.This was achieved by deproteinisation with acid as described above. Nitric acid also acts as a matrix modifier by reducing halide interferences. It is preferred to phosphate matrix modifiers because it can be obtained in a pure form with a low level of Cd contamination. Results and Discussion The optimum instrumental conditions are shown in Table 1. An optimum atomisation temperature was found for peak-height (1900 "C) and peak-area (1400 "C) measurements using Cd standards prepared in 5% V/V nitric acid. Peak- height measurements were obtained from the chart recorder whereas peak-area measurements were obtained from the digital display of the spectrometer. These values were used to optimise the ashing temperature.Fig. 1 shows an ashing profile for deproteinised plasma. 0.3 vl 73 s 8 0.2 a, C (0 + s 0.1 2 <\ Peak height I I I 0 1 5 10 HN03, Yo 3.2 a, f m 0.1 e s n 4 0 Fig. 2. Effect of increasing nitric acid concentration on the sensitivity of the measurement of Cd (3 pg 1-1) in pooled rat plasma samples Peak heighl n 0 5 10 HN03, "10 Fig. 3. Effect of increasing the concentration of nitric acid on the precision of measurement of Cd (3 pg 1-1) in pooled rat plasma samples The purpose of the ashing stage is to remove matrix constituents that may cause interference effects. Using the peak-area mode, temperatures of 450-550 "C can be utilised without losses of Cd. These temperatures are too low to have any effect on the removal of inorganic interferents, so an ashing temperature of 400°C was chosen as giving the best precision [0.97% relative standard deviation (RSD)].The optimum acid concentration is the lowest acid concen- tration that gives the best reproducibility and the lowest background signal. A low acid concentration is also preferred in order to increase graphite tube lifetimes. The effect of increasing nitric acid concentration on sensitivity is shown in Fig. 2. The rapid decrease in sensitivity for peak-height measurements is believed to be due to spreading effects, as nitric acid has a low surface tension and contributes to increased spreading as the droplet dries out during the drying stage and the residual nitric acid is effectively concentrated. Spreading results in lower peak- height values becaus? there is a temperature gradient along the graphite tube, which results in Cd being atomised at different times.Peak-area measurements are not affected so much, provided that the signal produced remains within the integration period, Fig. 3 shows that the optimum acid concentration on the basis of precision is a final nitric acid concentration of 5% V/V. The background absorbance was measured using the cadmium non-absorbing line of 226.5 nm (Fig. 4). A sharp decline in non-atomic background absorbance is found with increasing nitric acid concentration up to 5% V/V. At this concentration the background signal is reduced to a level which is well within the capability of the deuterium arc background correctionJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 37 1 v) U C s 4 - v) ; 3 - m e $ 2 - a a 1 - - 0 Peak height - Peakarea a, C (0 f 0.5 2 Q I ' ' 0 O ' il 5 10 HN03, O h Fig. 4. Effect of increasing acid concentration on the non-atomic background absorbance observed during measurement of Cd (3 pg I - l ) in pooled rat plasma samples (Cd line at 226.5 nm) 0.4 v) 0.3 U s a) 0.2 f 2 0.1 a cn 0 C m 0 B 0 1 2 3 Cdipg I-' Fig. 5. Comparison of the gradients of calibration graphs for (A) Cd added to deproteinised plasma (standard additions), (B) Cd standards in water and (C) Cd standards in dilute acid (5% V/V nitric acid) system. The accuracy of the plasma cadmium method was determined by standard additions plots, recovery experiments and the use of inter-laboratory comparisons.No certified quality control samples were available. As nitric acid causes a depression in the absorbance signal for Cd, standards were prepared in acid of the same concentration as used for the samples. Using the peak-area mode, the gradients of the standard additions plots were found to be similar to those obtained using standards prepared in 5% V/Vnitric acid (Fig. 5 ) . These were equal within the limits of imprecision of the measurement. With peak-height measurements there was a poorer corre- lation between the gradients of standard additions plots and those for standards in 5% V/V nitric acid. Recovery experiments were performed on pooled plasma, which was placed in acid-washed containers and spiked to give Cd concentrations of 0.9, 1.8 and 2.5 pg 1-1.Each sample was deproteinised and analysed in duplicate six times (Table 2). Recoveries of added Cd were excellent. To examine further the accuracy of the procedure, a comparison was made between the two laboratories involved in this work, using three different techniques for the determination of plasma Cd in five samples (Table 3) using a pool of rat plasma, obtained as described, from rats dosed with Cd in drinking water. Flame atomic fluorescence spectrometry (AFS) involved deproteinisation of the plasma with nitric acid, but required only aqueous Cd standards for preparing calibration graphs. It Table 2. Recovery experiments on cadmium in spiked plasma Cadmium Concentration added/ obtained/ Recovery Pg I - ' pg I-' ( n = 6) kSD, O/o 0 0.22.5 k 0.042 - 0.9 1.15 +_ 0.046 102.7 rt 5.1 99.7 rt 1.8 1.8 2.02 kO.034 102.0 * 1.9 2.5 2.73 k 0.049 Table 3.Inter-laboratory comparison of different techniques for the determination of cadmium in blood plasma Deprotenisation + Sample No. tube wall AAS AFS Probe AAS 1 2.9 3.4 3.8 2 1.7 2.2 4.0 3 1.5 2.0 2.2 4 3.8 3.2 3.8 5 2.6 2.8 3.5 is believed that this technique is an accurate and interference- free procedure.7 The third method used was probe atomisa- tion atomic absorption spectrometry.8 For this technique plasma was diluted 5-fold with water and 20 pl of the solution were injected on to the probe and then dried. During the ashing stage 2 p1 of 50% Aristar nitric acid were added. The probe was then removed from the furnace while the furnace was raised to isothermal conditions.8 After 10 s, the probe was re-inserted in the furnace.Calibration was effected with aqueous standards and signals were obtained as peak-height measurements on a chart recorder. Overall, the GFAAS method with deproteinisation gave slightly lower values than the AFS technique. The results obtained by the experimental probe atomisation method are higher than those given by other techniques, but most values were within k0.5 yg 1-1 of each other, which is reasonable at such low concentrations. Using the GFAAS method with deproteinisation the calibration graph was linear for Cd levels up to 3 pg 1-1. The Cd concentration giving an absorbance of 0.0044 A is 0.048 P8 I-' (0.96 pg). A detection limit of 0.063 yg 1-l (1.26 pg) of Cd was calculated from seven replicate analyses of plasma samples, using the convention of taking twice the standard deviation of the signal variation, using plasma samples with a Cd concentration of around 1.0 pg 1-1.The performance of the method was checked on a day-to-day basis by two methods: (1) using a pool of plasma as a control and (2) duplicate analyses of samples. Pooled plasma was prepared by spiking 100 ml of plasma with cadmium to give a concentration between 1 and 2 yg 1-1. This plasma was stored at 4 "C with 1% sodium azide as an antibacterial agent. Measurements of the cadmium concentration in this pool were 1.76 k 0.04 pg 1-1. Results deviating more than &lo% from this value required samples within a batch to be repeated. Duplicate samples were also compared from batch to batch as a check on the stability of the control plasma. The determination of plasma Cd could offer a more sensitive means of detecting impending Cd toxicity than the present laboratory procedures.This possibility should be investigated in workers known to be exposed to Cd and being carefully monitored. Initially, parallel measurement of plasma metallothionein would also be desirable to confirm the presence of Cd - metallothionein and to exclude the possibility of Cd contamination of the plasma samples prior to analysis.372 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 One of us (M. M. B.) was in receipt of a grant funded by the 3. 4. 5. Greater Glasgow Health Board (Unit 1 East). The authors acknowledge the assistance of P. R. Stahpit and S. K. Giri in carrying out Cd determinations by atomic fluorescence and probe atomic absorption methods. References 6. 7 . 1. 2. Friberg, L., Elinder, C.-G.. Kjellstrom, T., and Nordberg, 8. G. F., Editors, “Cadmium and Health: A Toxicological and Epidemiological Appraisal, Volume I, Exposure, Dose and Metabolism,” CRC Press, Boca Raton, FL, 1985. Roels, H., Djubang, J., Buchet, J.-P., Bernard, A., and Lauwerys, R., Scand. J . Work Environ., 1982, 8, 191. Mcintosh, M. J . , Moore, M. R., Goldberg, A , , Fell, G. S . , Halls, D. J., and Cunningham, C., Ecol. Dis., 1982, 1, 177. Rogenfelt, A . , Elinder, C. G., and Jarup, L. Znt. Arch. Occup. Environ. Health, 1984, 55, 43. Aughey, E., Fell, G. S . , Scott, R., and Black, M., Enuiron. Health Perspect., 1984, 54, 153. Mehra, R. K., and Bremner, I., Biochem. J . , 1983, 213,459. Michel, R. G., Hall, M. L., Ottaway, J. M., and Fell, G. S . , Analyst, 1979, 104, 491. Littlejohn, D., Cook, S., Durie, D., and Ottaway, J . M., Spectrochim. Acta, Part B , 1984, 39, 295. Paper 5611 7 Received March 12th, 1986 Accepted April 24th, 1986