JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 349 Improved Determination of Cadmium in Blood by Flame Atomic Fluorescence Spectrometry Edet. J. Ekanem,* Charles L. R. Barnardt and John. M. Ottaway Department of Pure and Applied Chemistry, University of Strathclyde, 295 Cathedral Street, Glasgow GI IXL, UK Gordon S. Fell Department of Clinical Biochemistry, Royal Infirmary, Glasgow G4 OSF, UK A procedure is described for the determination of normal levels of cadmium in whole blood using flame atomic fluorescence spectrometry. This method represents an advance on previous work in that better control of the electrodeless discharge lamp source temperature has facilitated greater stability and thus better detection limits. Furthermore, sample pre-treatment by deproteinisation reduces the dilution factor from 1 + 4 to 1 + 1 so that the whole blood detection limit is improved to twice that of aqueous solutions. Keywords: Flame atomic fluorescence spectrometry; electrodeless discharge lamps; blood cadmium determination; protein precipitation The determination of cadmium in human blood has presented two main problems to the analytical chemist. Firstly, the normal levels, which have yet to be agreed upon, are at or close to the limits of detection of most atomic spectrometric techniques.This places an onerous task on the analyst to provide both accuracy and precision for such samples. Secondly, and more important, the matrix is complex and therefore produces difficult interference problems. At present the most widely applied method for the determination of cadmium in blood involves the use of carbon furnace atomic absorption spectrometry (CFAAS), coupled with a sample pre-treatment. Among the many pre-treatments employed are wet ashingl with acids, low-temperature plasma ashing2 and protein precipitation .3.4 In 1979 the development, in this laboratory, of a flame atomic fluorescence spectrometric (FAFS) method for the determination of cadmium in blood and urine was described by Michel et al.5 In this procedure blood samples were haemolysed with Triton X-100, acidified to 0.04 M with hydrochloric acid, mixed and centrifuged.The supernatant liquor was then aspirated into a nitrogen-separated air - acetylene flame supported on a capillary burner6 using a Perkin-Elmer nebuliser and spray chamber. Aqueous cad- mium calibration standards were also acidified to 0.04 M HCl. Before interpolation on the calibration graph, sample results were adjusted by an uptake rate correction factor to compen- sate for viscosity differences between the aqueous standards and blood samples.Blood detection limits of 1.4 pg 1-1 were reported for a 1-s count time. Although this level of sensitivity is adequate for the determination of raised blood cadmium levels in cases of proved intoxication, it is inadequate for determining normal levels of cadmium in blood. A reasonable estimate of the normal levels of cadmium in blood is provided by the lower end of concentration ranges that have been determined for reference populations of unexposed persons. Such a range of 1.1-6.4 pg 1-1 with a mean of 3.1 (k1.5) yg 1-1, obtained by FAFS, has been reported from this laboratory.5 A range of 0.3-7.9 pg 1-1 with a mean of 2 pg 1-1 based on a similar reference population and a range 0.3-6.0 yg 1-1 with a mean of 1.3 yg 1-1 based on a non-smoking population have also been reported using CFAAS .7 Sub-pg 1- 1 blood cadmium levels have frequently * Present address: School of Basic Studies, Ahmadu Bello Univer- i Present address: Department of Chemistry, Glasgow College of sity, Zaria, Nigeria.Technology. Cowcaddens Road, Glasgow, UK. been reported.Sl0 An FAFS procedure with a much lower detection limit than the previously reported 1.4 pg 1-1 is required for the accurate determination of low or normal blood levels. Modification of the blood matrix by protein precipitation has been applied to the determination of cadmium in blood by CFAAS, giving good accuracy for a range of precipitants.11J2 Aim The limit of determination of the previous procedures was hampered by a high aqueous detection limit (0.2 yg l-l), relative to the lower values in the working range, and the need for a five-fold dilution of blood samples, which further raised detection limits.These investigations were carried out in order to overcome these problems by reducing the need for sample dilution by using protein precipitation, and by decreasing the aqueous detection limits by using better temperature control of the source. Experimental Instrumentation The instrumentation used in this work, (summarised in Table 1) has been described in detail previously.5J3314 Radiation from a thermostated microwave-excited cadmium electrode- less discharge lamp (EDL) source ,15716 mechanically modu- - Table 1.Instrument operating conditions Light sources: (1)CdEDL . Broida 21OL cavity (2)Eimac . . . . Modulation. . . . Air - acetylene . . Type. . . . . . . Grating . . . . Wavelength . . Bandpass . . . . Type . . . . . . Photomultiplier type EHTvoltage . . Dutycycle . . . . Measurement period Flame: Monochromator: Photon counter: . . . . . . . . 534 "C (air stream temperature); Microwave power 60 W Operating current varied to match 300 Hz (both sources) corresponds to 180 "C in quartz EDL EDL scatter characteristics Slightly fuel rich, nitrogen-sheathed . . Spex Doublematefl4 . . . . 228.8nm . . 1.0nm Blazed at 300 nm 1200 grooves mm-' .. Ortec Brookdeal 5C1 . . EMI9789QB . . 115OV . . 55% . . I s350 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 lated at 300 Hz, is used to excite cadmium atomic fluores- cence. Samples are aspirated into a nitrogen-separated air - acetylene flame supported on a Perkin-Elmer nebuliser - spray chamber fitted with a capillary burner head. Fluorescence signals were monitored using a double monochromator photon counting system. An intensity- modulated (300 Hz) co-focused xenon-arc (Varian-EMI) continuum source was used for simultaneous background and scatter correction. Signals from the two sources were phase- locked using the synchronous sampling (5C21) module of the photon counter (Brookdeal). Thermostated Environment of EDLs It has been established that the best radiant performance for the cadmium EDL is obtained if the lamp is placed in an environment of appropriate, constant, high temperature.In the p-evious work5 the EDL temperature was monitored with a thermometer placed in the neck of the quartz lamp. A stream of hot air was applied to the outside surface of the lamp, the temperature of which was controlled by a Variac power supply. This approach to thermostating the EDL suffers from two main drawbacks. Firstly, there is a long time lag between any change in EDL temperature and the response of the thermometer, so that short term temperature fluctu- ations would not be noticed. Secondly, the heating power supplied to the air stream was not controlled by changes of the lamp temperature.The radiant output from an EDL is highly responsive to even minor fluctuations in temperature. Such fluctuations result in spectroscopic noise and ultimately in poor detection limits. For these reasons a new type of heater controller was constructed in which the power to the heater was governed by a voltage feedback from a thermocouple placed in the air stream immediately adjacent to the cavity. The placing of a thermocouple probe inside the cavity (closer to the EDL) was precluded because it destabilised the microwave field. This feedback-controlled heater gave a significant improvement in the control of EDL temperature, to within 0.5% at 500°C. Collection and Storage of Blood Blood samples were collected by venepuncture and stored in plastic sample tubes containing anticoagulant (potassium EDTA or lithium heparin).These modes of sample collec- tion17J8 and storage5 have previously been shown to be free from cadmium contamination. The polypropylene centrifuge tubes (Henleys Medical Supplies, London) in which protein precipitation was performed19 were screened for cadmium contamination by shaking the same aliquot of 2 M nitric acid consecutively in a large number of fresh tubes and observing the cadmium atomic fluorescence generated from this wash solution. Blank signal levels were always obtained. Containers and pipettes, used for preparing standards, were soaked overnight in 50% nitric acid, washed out and rinsed with de-ionised water before use. Where necessary, blood samples were stored temporarily at 4°C or for longer periods at -22 "C.Reagents Only high-purity (AnalaR) reagents were used and contami- nation checks were made by measuring reagent blanks for cadmium atomic fluorescence. All reagent solutions were prepared in de-ionised water. Aqueous calibration standards were prepared by serial dilution from a 1000 pg 1-1 stock solution. All standards were matrix matched by addition of appropriate amounts of the precipitant. Procedure for Protein Precipitation The protein precipitants considered were hydrochloric acid, chloroacetic acid (CAA) , trichloracetic acid (TCA) and nitric acid. All of these reagents have previously been used as protein precipitants .11912J9 The method used for each was essentially the same. A 2-ml aliquot of blood was dispensed into an equal volume of precipitant solution in a poly- propylene centrifuge tube.The resulting slurry was centri- fuged for 30 s at 3000 rev min-1 and the protein-free supernatant aspirated into the flame. A reagent blank was measured in each instance by replacing the blood with a 2-ml aliquot of de-ionised water. The concentration of each precipitant was optimised to yield the highest cadmium AFS signal from 2 ml of the same blood. The procedure for each was optimised using out-dated blood bank samples. The criteria used to identify the most appropriate reagent were the Cd fluorescence signal, supernatant volume and uptake rate. This procedure was compared with the Triton X-100 - HC1 procedures in terms of the detection limit and the signal obtained from 2 ml of the same blood when treated by each procedure.Analytical Cadmium Recoveries The recovery of cadmium from whole blood following protein precipitation was assessed by adding aliquots of inorganic cadmium standards to a 2-ml blood sample before deprotein- isation. The percentage recovery of total cadmium was acceptable for the range 0-10 pg 1-1 of added cadmium. Precision An assessment of the analytical precision of the method was obtained using a set of blood samples of known cadmium concentration that spanned the working range. Each of ten 2-ml aliquots of the blood samples were deproteinised, centrifuged and analysed, and the over-all precision of the technique was calculated from the signals obtained. Accuracy The accuracy of the results obtained by protein precipitation followed by FAFS was assessed by direct comparison of a set of samples previously analysed by CFAAS in another labora- tory.The accuracy of the FAFS procedure was also tested by the analysis of quality control blood samples from the Supra-Regional Assay Service (organised by Surrey Univer- sity) that had been analysed independently by many labora- tories. Interferences The experimental observations of Michel et al. 5 were con- firmed by the present investigation when nitric acid was used as the protein precipitant. Results and Discussion The concentrations of precipitants yielding the most efficient release of cadmium from the blood were optimised in terms of signal, uptake rate and supernatant volume.19 These con- ditions are summarised in Table 2.The minimum dilution of blood gave a maximum signal, although this resulted in a highly viscous supernatant for both the chloroacetic acids. The variation of cadmium FAFS signal with increasing nitric acid precipitant concentration is shown in Table 3. The optimum release of cadmium from blood is achieved with 2 M nitric acid, which gave the best over-all performance in terms of the highest signal (Tables 3 and 4) with acceptable supernatant volume (Table 5) and an uptake rate equal to that of theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 351 Table 2. Optimisation of reagent concentration Table 6. Sensitivities of procedures for 2-ml aliquots of same samples Concentration Optimum Signal-blank/ Reagent range observed concentration counts Triton X-100 - HCI .. As published5 As published5 106 Nitricacid . . . . 1 . G 4 . 0 ~ 2.0 M 340 C A A . . . . . . 1.0-4.0% 1.5% 101 TCA . . . . . . 1.0-4.0% 1 .O% 94 HCI . . . . . . 1.0-4.0 M 2.0 M 119 Table 3. Optimisation of nitric acid concentration HN03 Signal-blank/ concentrationh counts s-* 1 .o 1.5 2.0 2.5 3.0 4.0 274 289 340 3 30 337 339 HN03 signal Triton X-100 Signal-blankkounts s-1 Sample No. Triton X-100 HN03 signal 1 106 340 3.2 2 47 110 2.3 3 164 538 3.3 Table 7. Sample uptake rates Sample type De-ionised water . . . . . . . . . . . . 9.0 1 M nitric acid . . . . . . . . . . . . . . 9.0 2 M HN03 blood deproteinate (1 + 1 dilution) 9.0 Triton X-100 - HC1 blood deproteinate (1 + 4 dilution) . . . . . . . . . . . . 6.8 Triton X-100 - HC1 blood deproteinate (1 + 1 dilution) .. . . . . . . . . . . 5.0 . . Table 4. Comparison of reagent sensitivities Table 8. Typical calibration results Signal-blank/ Sensitivity relative Reagent counts s-l to Triton X-100 TritonX-100 - HCI . . 47 ~ M H N O ~ . . . . . . 110 1.5'IoCAA . . . . . . 56 1.0% TCA . . . . . . 50 2.0M HCI . . . . . . 69 1.3 2.3 1.2 1.1 1.5 Table 5. Recovery of liquid fraction after deproteinisation Supernatant Wet residue Reagent volume/ml volume/ml TritonX-100 - HCI . . 8.0 2.0 2.0 M HNO3 . . . . . . 2.0 2.0 1.5%CAA . . . . . . 1.6 2.4 1.0% TCA . . . . . . 1.6 2.4 aqueous standards (Table 7). It is clear from these results that, even when an uptake rate correction (the factor is the ratio of the aqueous uptake rate to the supernatant uptake rate) is applied to each precipitant solution, as appropriate, nitric acid deproteinisation provides the most efficient release of cad- mium.Both the protein precipitation and Triton X-100 procedures yield two fractions after centrifugation. There is a wet, solid residue of cell protein and a supernatant liquor. In order to perform accurate FAFS analysis for blood cadmium it is necessary that at least 2 ml of supernatant be produced. Attempts to use hydrochloric acid proved unsuccessful as it produced effervescence and foaming when mixed with blood samples. Also, as it yielded inferior FAFS signals and variable supernatant volumes for the same sample, hydrochloric acid was considered unsuitable for protein precipitation. The results in Table 5 illustrate the breakdown of a typical sample when treated with each of the precipitants.In each instance a 2-ml aliquot of sample was treated and converted into supernatant and residue using an equal volume of precipitant, although the Triton X-100 procedure used 1 + 4 volume dilution. Apart from the relatively poor signals obtained with the chloroacetic acids, their supernatant liquids were so viscous that they blocked the capillary burner when aspirated into the flame. For these reasons these acids were not investigated further. This left only nitric acid and the Triton X-100 procedure as suitable methods for blood cadmium determi- nation. The relative sensitivities of the two procedures are compared in Table 6, from which it is clear that the use of nitric acid has distinct advantages over the Triton X-100 procedure.In view of the wide disparity between the dilution Average uptake ratelm1 min-l Parameter Aqueous Standard calibration additions Fittedline . . . . . . y = 141x+ 34 y = 141x+205 Correlation coefficient Blood detection limit Aqueous detection limit ( r ) . . . . . . . . 0.9995 0.9982 (20) . . . . . . . . 0.02 pg 1- ' (20) . . . . . . . . 0.01 pg I - ' factors used by each procedure the two reagents were compared at similar dilutions. It was found that the greater the dilution factor, the closer the results for the two procedures become. The nitric acid deproteinisation, however, yields superior sensitivity at all sample dilutions. The detection limits obtained for both procedures at the various dilutions also follow this trend.Even at the same sample dilution, signal levels obtained with the Triton X-100 procedure are about 35% lower than the corresponding signals obtained following the nitric acid deproteinisation. The liquid fraction resulting from the Triton X-100 treatment is more viscous and yields lower uptake rates than that obtained from nitric acid deproteinisation. When aqueous calibration standards are used for FAFS blood analysis it is essential to compensate for differences in uptake rates between samples and standards. The observed fluctuations in uptake rate associated with variable haemoglobin levels hampered the Triton X-100 procedures in that it necessitated the use of a variable uptake-rate correction factor, which could be sample depen- dent. In this respect the nitric acid deproteinisation procedure has an important advantage as no uptake rate correction is required, as can be seen from Table 7. This, together with the improved sensitivity of the 2 M nitric acid deproteinisation, make it a more suitable matrix modification procedure for the determination of blood cadmium levels by FAFS.Evaluation of the Use of Nitric Acid Deproteinisation for the Determination of Blood Cadmium Levels by FAFS The results obtained by direct analysis using aqueous calibra- tion standards showed good agreement with those obtained by standard additions. A typical example is presented in Table 8, which indicates close agreement between the slopes of aqueous and standard additions calibration graphs. The blood detection limit is expressed as twice the standard deviation of the aqueous signal multiplied by the dilution factor (f).352 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 Recoveries The efficiency of the protein precipitation procedure was assessed in terms of recovery of cadmium following addition of inorganic cadmium to blood samples prior to protein precipi- tation. Despite the difference in nature between the added cadmium and that present in blood, as a protein complex, the results shown in Table 9 indicate efficient release of total cadmium is effected by this procedure, as all the values are close to 100%. Analytical Precision The analytical precisions shown in Table 10 were obtained for the method as a whole, using normal blood samples containing less than 10 pg 1-1 of cadmium.Instrumental precision was measured using aqueous standards. The between-batch precision was assessed by measuring the same blood sample twice a week for 6 weeks. The detection limits are improved considerably (Table 11) by the modified procedure compared with our earlier method. Average values of 0.01 and 0.02 pg 1-1 have been reproducibly obtained for aqueous and whole blood samples, respectively. This rep- resents an almost two orders of magnitude improvement in the blood cadmium detection limit. This improvement has been achieved principally by a more precise EDL temperature control system. This has been further enhanced by the reduced sample dilution allowed by the nitric acid deproteinisation procedure. Good accuracy is indicated in Table 12, which represents a comparison of the FAFS protein precipitation procedure, for 16 samples, with a CFAAS procedure carried out in a different laboratory.The samples numbered 11-16 were provided by the Supra-Regional Assay Service (SAS) quality control service, operated by the Heavy Metals Laboratory at the University of Surrey. The accuracy of the procedure was confirmed by the close correlation (Table 13) between the inter-laboratory means and the FAFS results. Conclusions Nitric acid deproteinisation offers a simple means of matrix modification with minimum sample dilution, by which normal levels of cadmium in whole blood can be accurately deter- mined by FAFS. This procedure is in fact being adapted in this laboratory for the determination of the lower levels of cadmium in plasma and serum.Although FAFS systems of apposite sensitivity are not commercially available, the technique provides a rapid and convenient method for Table 9. Analytical recoveries of cadmium following protein precipitation of whole blood Cadmium concentrationlpg I-* Added Recovered Recovery, O/O 0 1.16 - 2 3.14 99 4 5.56 110 6 7.67 108 8 9.80 108 10 11.40 102 cadmium determinations in both blood and urine as both normal and elevated levels can now be determined with good precision in both matrices. It is of particular value for the rapid screening of samples for Table 11. Detection limits for the determination of cadmium in blood Blood detection Dilution limit/ Reagent Reference factor vgl-' Triton-X100 - HCI . . 5 5 1.14 2.0 M nitric acid .. . . This work 5 0.05 2.0~nitricacid . . . . Thiswork 2 0.02 Table 12. Comparison of FAFS with CFAAS Cadmium concentration/ Pi3 1- Sample No. FAFS CFAAS 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 15.4 8.0 10.0 28.1 0.8 7.6 9.9 9.2 9.6 6.6 24.2 14.9 3.6 31.4 11.2 18.6 15.1 7.8 9.3 27.9 0.7 6.1 8.0 7.5 9.0 4.8 23.8 15.7 3.3 27.3 8.0 16.2 Regression results: CFAAS results is The regression equation for the correlation between FAFS and FAFS = 1.02CFAAS -+ 0.91 The standard deviation of scatter20 of the FAFS results about this line is 1.32 pg I-'. Table 13. Inter-laboratory comparison of FAFS results Cadmium concentration/ Clg1-l Sample No. FAFS SAS means 11 24.2 24.9 12 14.9 17.6 13 3.6 2.9 14 31.4 32.4 15 11.2 10.5 16 18.6 18.1 Regression results: and the Supra-Regional Assay Service (SAS) is The standard deviation of scatter20 of the FAFS results about this line is 1.28 pg 1-1.The regression equation for the correlation between FAFS results FAFS 1.02SAS + 0.00 Table 10. Measurement precisions Precision (RSD), Yo Cadmium level/ Sample Pg I-' Total Instrumental Between-batch - - Blood (1) . . . . . . 1.8 6.6 Blood (2) 7.8 1.9 - Aqueous(1) . . . . . . 5.0 - 3.1 - Aqueous(2) . . . . . . 10.0 - 1.8 - . . . . . . 7.4JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 353 which results are urgently needed and for population surveys, as about 300 samples can be analysed daily. The blood detection limit of 0.02 pg 1-I is ideally suited to the determination of cadmium at the sub-pg 1-1 level. The advantages of the FAFS method with nitric acid deproteinisation include the following: ( i ) lower blood levels can be determined more precisely; (ii) analytical sample solutions have the same uptake rate as calibration standards, hence no uptake rate correction factor is required; (iii) the scatter of incident radiation by sample particles in the flame is negligible; background correction for scatter is therefore virtually unnecessary, but has still been used in this work to take account of any gross variations in matrix from sample to sample; (iv) only one reagent is involved, reducing the risk of con tamination.The authors thank the Scottish Home and Health Department for the award of a postdoctoral fellowship (to C. L. R. B.). They also acknowledge the continued support of the Eastern District, Greater Glasgow Health Board.The thanks are extended to Messrs. M. Porter and G. Brown of the Chemistry Workshop, Strathclyde University, for the construction of the heater and also to Mr. G. Henderson of the Biochemistry Workshop, Glasgow Royal Infirmary, for the design and construction of the heater controller. References Gorusch, T. T., “The Destruction of Organic Matter,” Pergamon Press, Oxford, 1970. Carter, G. F., and Yeoman, W. B., Analyst, 1980, 105, 295. Stoeppler, M., Brandt, K., and Rains, T. C., Analyst, 1978, 103, 714. Stoeppler, M., and Brandt, K . , Fresenius 2. Anal. Chem., 1980, 300, 372. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Michel, R. G . , Hall, M. L., Ottaway, J. M., and Fell, G. S . , Analyst, 1979, 104, 491. Aldous, K. M., Browner, R. F., Dagnall, R. M., and West, T. S., Anal. Chem., 1970,42, 939. Fell, G. S . , Chem. Brit., 1980, 16, 323. Friberg, L., Piscator, M., Nordberg, G., and Kjellstrom, T., “Cadmium in the Environment,” Second Edition, CRC Press, Cleveland, OH, 1974. Sharma, R. P., MacKenzie, J. M., and Kjellstrom, T., J . Anal. Toxicol., 1982, 6, 135. Subramanian, K. S., and Meranger, J. C., Clin. Chem., 1981, 27, 1866. Einarsson, O., and Lindstedt, S., Scand. 1. Clin. Lab. Invest., 1969, 23, 367. Baily, P., and Kilroe-Smith, T. A., Anal. Chim. Acta, 1975,77, 29. Michel, R. G., Sneddon, J., Hunter, J. K., Ottaway, J. M., and Fell, G. S . , Analyst, 1981, 106, 288. Sthapit, P. R., Ottaway, J. M., and Fell, G. S . , Analyst, 1983, 108, 235. Michel, R. G., Coleman, J., and Winefordner, J. D., Spectro- chim. Acta, Part B, 1978, 33, 195. Michel, R. G., Ottaway, J. M., Sneddon, J., and Fell, G. S., Analyst, 1978, 103, 1204. Fell, G. S . , Ottaway, J. M., Hussein, F. E. R., Michel, R. G . , and Hall, M. L., in Brown, S. S., Editor, “Clinical Chemistry and Chemical Toxicology of Metals,” Elsevier North-Holland, Amsterdam, 1977, p. 367. Fell, G. S., Ottaway, J. M., and Hussein, F. E. R., Br. J. Znd. Med., 1977,34, 106. Ekanem, E. J . , Barnard, C. L. R., Ottaway, J. M., and Fell, G. S., Talanta, 1986,33, 55. Davies, 0. L., “Statistical Methods in Research and Produc- tion,” Third Edition, Oliver and Boyd, London, 1967, p. 150. Paper 5619 Received February 24th, 1986 Accepted May 6th, 1986