Analytica Chimica Acta 491 (2003) 231–237

Determination of cadmium in human urine by electrothermal

atomic absorption spectrometry
Cláudio José de Castro Maciel a , Gilmar Machado Miranda b ,
Danielle Palma de Oliveira c , Maria Elisa P.B. de Siqueira c ,
Josianne Nicácio Silveira b , Edna Maria Alvarez Leite b ,
José Bento Borba da Silva d,∗
a Laboratório de Análises Clinicas Humberto Abrão, Belo Horizonte, MG, Brazil
bLaboratório de Toxicologia Ocupacional, Faculdade de Farmácia, Universidade Federal de Minas Gerais,
Belo Horizonte, MG, Brazil
c Escola de Farmácia e Odontologia de Alfenas, Alfenas, MG, Brazil
d Laboratório de Espectrometria de Absorção Atômica, Departamento de Quı́mica, Universidade Federal de Minas Gerais,
Av Antônio Carlos 6627, Belo Horizonte, MG, Brazil
Received 21 April 2003; received in revised form 11 June 2003; accepted 3 July 2003

Abstract
In the present work, a fast acid digestion method was applied to the determination of cadmium in urine samples by
electrothermal atomic absorption spectrometry (ET AAS). The pyrolysis and atomization temperatures were 400 and 1200 ◦ C
with a characteristic mass of 1.3 pg. Quantification was made with matrix matching calibration curves. The method was
validated by the analysis of certified urine samples. Differences were not observed among the levels of the metal in populations
from metropolitan and rural areas. The limit of detection (LOD, k = 3, n = 10) was 0.08 ␮g l−1 . Coefficients of variation
for intra- and interassay studies were 5.5 and 12.7%, respectively. Recovery of spiked samples showed acceptable values.
Significant correlation was observed among sex, with higher values for women. Age also showed an influence on urinary Cd
levels, being higher for individuals over 46 years. No correlation was found between either tabagism or alcoholic beverage
intake and of urinary cadmium levels.
© 2003 Elsevier B.V. All rights reserved.
Keywords: Electrothermal atomic absorption spectrometry; Cadmium; Human urine

1. Introduction tilizers in agriculture, it can be broadly distributed in

soil, water and food and consequently reach the man
Cadmium is a natural constituent of rocks rich in [1]. This metal is also a by-product of lead and zinc
phosphate and due to the use of phosphate-based fer- exploration and this process results in an important
source of environmental contamination by Cd, either
∗ Corresponding author. Present address: Universidade Federal
restricted to the place of the exploration or spread to
de Minas Gerais, Departamento de Quı́mica—Avenida Antônio
Carlos 6627, 31270-901 Belo Horizonte, MG, Brazil.
other areas. Starting in the 20th century the produc-
Tel.: +55-31-34995750; fax: +55-31-34-995700. tion of Cd and its emission in the atmosphere, water
E-mail address: [email protected] (J.B.B. da Silva). and soil has increased in a significant way [2,3].

0003-2670/$ – see front matter © 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0003-2670(03)00820-1
232 C.J. de Castro Maciel et al. / Analytica Chimica Acta 491 (2003) 231–237

The tobacco plant naturally accumulates high levels made by Han [9], revealed that the critical value of
of cadmium in its leaves. As a consequence a signifi- urinary Cd excretion was 15 ␮g l−1 . Both the av-
cant amount of the metal is detected in the cigarettes erage amount of Cd excreted and the incidence of
(1–2 mg per cigarette), variable in agreement with the low-molecular-weight protein in urine were positively
mark and country of production. It is considered that correlated with the total intake of Cd. The study sug-
a person that smokes 20 cigarettes per day absorbs gested that the acceptable daily intake of Cd is 100 ␮g
about 1 ␮g of Cd per day. However, there are consid- per capita or 1.67 ␮g kg−1 body weight.
erable individual variations [3]. The major difficulty in assigning significance to
Far greater exposure to the general population may urine cadmium concentrations or, for that matter, in
result from the application of cadmium-containing establishing a normal range, is due to the problems in
sewage sludge to croplands or from its dumping at measuring ␮g l−1 concentrations of the metal in such
sea. Edible crops as well as a variety of marine or- complex biological matrices.
ganisms are able to effectively concentrate cadmium There are several methodologies in the literature
in their tissues even at relatively low environmental for the determination of cadmium in urine and oth-
concentration [4]. ers biological fluids. Onar and Temizer [10] pro-
Discovered in 1817, cadmium came to be used in posed a method to determine Cd and Pb in urine
industrial scale in the last 50 years in a variety of using differential-pulse anodic stripping voltametry
activities, as in the electric industry and galvanization (DP-ASV). A method for simultaneous determination
due to his non-corrosive property, in the production of of Zn, Cd, Pb and Cu in urine of patients with black-
polyvinyl chloride (PVC), as a stabilizer, in the plastic foot disease using anodic stripping voltametry was
and glass industries, used as a pigment, and in the described [11], while in another study, DP-ASV was
Ni–Cd batteries industry as cathode material, besides used for the simultaneous determination of urinary
taking part in several metallic alloys [5]. Zn, Cd, Pb and Cu concentrations in steel production
Although only about 0.4% of the total amount of workers [12].
ingested cadmium is retained in the human body, it Several methods have been developed using elec-
accumulates in liver and kidney tissues with an esti- trothermal atomic absorption spectrometry (ET AAS)
mated biological half-life of 20–30 years [6]. for Cd determination. Lund and Larsen proposed the
Jarup and Elinder evaluated a group of Swedish determination of Cd in urine by ET AAS, in which the
workers exposed to the metal between 1981 and 1982 metal is electrolyzed onto the thin platinum wire, fol-
(n = 902). The results indicated that the urinary cal- lowed by electrical heating of the wire. In this work,
culi are an important consequence of the renal dam- the proposed methodology was compared with the ex-
age induced by Cd and that the urilithiasis commonly traction of Cd with ammonium pyrrolidine dithiocar-
happens with the tubular proteinuria [7]. bamate (APDC) into methyl isobuthyl-ketone (MIBK)
International Agency of Research on Cancer and flame AAS determination with good agreement
(IARC) considers that there are enough evidences to [13]. Carmack and Evenson [4] determined urinary
classify cadmium as a carcinogenic agent, although Cd by ET AAS with carbon tubes impregnated with
some factors should be considered in this evalua- molybdenum with a limit of detection (LOD) of
tion, such as scarce historical data, limited number 0.05 ␮g l−1 .
of works, tabagism and association with other ele- The microdetermination of lead and cadmium in
ments (nickel, arsenic). Experimental studies in an- blood and urine by ETTAS was accomplished using
imals are being carried out to evaluation the risk of extraction into MIBK after chelation by APDC. Blood
lung, prostate, kidney, testicle and bladder cancer [3]. is hemolysed by addition of deionized water. The pH
Recently, it has been demonstrated that glutathione of urine is fixed by a pH 5 buffer [14]. Slavin and
(GSH) is directly related with the protection mecha- co-workers [15] used ET AAS with Zeeman back-
nisms of nephrotoxicity of cadmium [8]. ground correction in the determination of Cd in urine
An investigation on the effects of cadmium ex- using (NH4 )2 HPO4 and HNO3 as chemical modifier
posure on human health, made in five provinces of with a LOD of 0.4 ␮g l−1 of urine, while De Broe
China in the population living near Cd industries and co-workers [16], determined Cd in urine by ET
 C.J. de Castro Maciel et al. / Analytica Chimica Acta 491 (2003) 231–237 233

AAS with deuterium correction using the L’Vov plat- Table 1

form treated with ammonium molybdate using matrix Temperature program for Cd determination in human urine by ET
matching calibration curves with a LOD of 0.4 ␮g l−1 AAS without chemical modifier

of urine. Step Temperature (◦ C) Ramp (s) Hold (s) Ar flow rate

(ml min−1 )
In the present study, a methodology using fast acid
digest of urine samples to determine cadmium in urine 1 85 5.0 – 3.0
2 95 20.0 – 3.0
by ET AAS was investigated. With such methodol- 3 120 10.0 – 3.0
ogy, they were certain and compared the levels of the 4 400 5.0 – 3.0
metal in urine of two groups of people without his- 5 400 – 15.0 3.0
tory of occupational exposure to the Cd: 115 people 6 400 – 15.0 0
resident in an industrial area and 156 residents in a 7 1200 0.7 – 0
8 1200a – 7.0 0
small town in the rural area. The influence of sex, age, 9 2400 2.0 – 3.0
tabagism and alcoholic intake in the urinary Cd was
a Reading in this step.
also investigated.

all aqueous solutions. The nitric acid used to solu-

2. Experimental bilize the urine sample was from Merck (Darmstadt,
Germany, No. 7587956). Cadmium standard solutions
2.1. Apparatus were prepared from a 1000 mg l−1 stock solution from
Merck made by dilutions of a Titrisol ampoule in ni-
The integrated absorbances were obtained in an tric acid 3% (v/v). Palladium nitrate 1.5 mg l−1 was
atomic absorption spectrometer, SpectrAA 220 from from Merck (No. B9366989 710) while the monoba-
Varian (Victoria, Australia), equipped with a graphite sic ammonium phosphate was from Reagen (Rio de
furnace, an autosampler EL 98013384-2C, and the Janeiro, RJ, Brazil, No. 910228).
polarized Zeeman background correction, all from
Varian. Hollow cathode lamp for Cd from Varian (part 2.3. Samples and sample treatment
no. 56-101-008-00), operating at 4.0 mA, was used.
Unless otherwise, the stated manufacturer recom- The urine samples, the standard solutions in wa-
mended conditions were employed. Argon 99.999% ter and urine and the blanks were previously treated
from Air Liquid (Contagem, MG, Brazil) was used with the modified method described by Voloder [19],
as the purge gas. Pyloric graphite coated tubes with 2.5 ml of sample (urine), Cd standard solution and
Forked platform (Varian, part no. 63-1000-23-00 and Milli-Q water were transferred to screw glass tubes
63-1000-24-00, respectively) were used. The sample and added 800 ␮l of concentrated HNO3 . Next, they
and standard solutions, 10 ␮l, was introduced onto were placed in a boiling water bath for 10 min, cooled
the platform at 40 ◦ C. In all the studies with chemical and the volume completed to 10 ml with deionized
modifier (monobasic ammonium phosphate or palla- water. Certified urine samples from BioRad (US Bio-
dium nitrate) a volume of 5 ␮l was used. The final Rad Laboratories, Anaheim, USES) level 1 (Code
concentration of the modifiers were of 2 and 5% for 4001) and level 2 (Code 4051) were analyzed.
monobasic ammonium phosphate and 1.5 mg ml−1 for For the studies of determination of the levels of Cd
palladium nitrate. The temperature program was opti- in urine samples from a population without history
mized through pyrolysis and atomization temperature of occupational exposure to Cd, male and female
curves, and is shown in Table 1. adults aged between 15 and 60 years resident in the
metropolitan area of Belo Horizonte, MG, Brazil
2.2. Reagents and solutions (n = 115) and in the city of Alfenas, MG (rural
area), Brazil (n = 156) were selected, using a tox-
All reagents employed were of analytical grade. icological protocol (personal data such as age, sex,
Deionized water (18.2 M cm) from a Milli-Q Sys- profession, tabagism, frequency of intake of alco-
tem (Millipore, Bedford, MA) was used to prepare holic beverages, diet, presence of diseases, workplace
234 C.J. de Castro Maciel et al. / Analytica Chimica Acta 491 (2003) 231–237

information, among others). The exclusion criteria For the recovery studies, a pool of urine, spiked with
used in the selection of individuals were: age under Cd 0.5 and 3.0 ␮g l−1 (five solutions of each concen-
15 and over 60 years, or reports of several patholo- tration) was solubilized as described after subtracting
gies such as lupus erythematosus, renal and breathing the blank (same urine pool without spike).
alterations.
Urine samples (medium jet) were collected at ran-
dom in polyethylene flasks and immediately acidified 3. Results and discussion
with concentrated HNO3 p.a. (pH ≤ 2.0) before de-
termination of cadmium content. 3.1. Pyrolysis and atomization temperature curves

2.4. Statistical analysis Pyrolysis and atomization temperatures were op-

timized by pyrolysis and atomization temperature
The statistical analysis was accomplished with curves in the digested urine sample with 4.0 ␮g l−1
the aid of the software SigmaStat for Windows 2.0, of Cd. Graphite tubes were tested with and with-
from Jandel Corporation. The analytical data were out Forked platform. However the best results were
evaluated by the non-parametric tests of Mann– obtained with the use of the platform, in which the
Whitney, Kruskal–Wallis, by multiple comparison of sample is concentrated in a limited reaction area.
Student–Newman–Keuls (SNK) and by Dunn’s and The Forked platform delays the vaporization of the
Spearman correlation. For all of the analyses, the sample until the tube reaches a high stabilization
significance criterion was P < 0.05. temperature, thus minimizing vapor phase interfer-
The limit of detection (␮g l−1 ) was calculated us- ences. Some chemical modifiers were also tested
ing the equation LOD = 3 × sB /b, where sB is the without sample mineralization: monobasic ammo-
standard deviation of 10 measurement of the blank (10 nium phosphate from 2 to 5% and palladium nitrate
aliquots of Milli-Q water treated with nitric acid as 1.5 mg ml−1 . However, as the best results were ob-
previously described)—the phrase made in triplicate tained through the mineralization of the sample with
and using the medium values and b is the slope of the concentrated HNO3 , and without modifier, we opted
matrix matching calibration curve. for this approach.
The precision of the method was evaluated by the Table 1 shows the temperature program for the de-
coefficient of variation (COV) intra- and interassay termination of Cd in urine, while Fig. 1 shows the py-
using analytical solutions of Cd in urine pool in con- rolysis and atomization temperature curves obtained
centrations of 0.1 and 1.0 ␮g l−1 , submitted to the de- using the Forked platform in the digested urine with
scribed solubilization. 4.0 ␮g l−1 of Cd and without chemical modifier. For
To evaluate the precision intraassay of the analyt- the pyrolysis temperature studies without modifier, a
ical method, six aliquots of each concentration were temperature of 1200 ◦ C was used in the atomization.
analyzed in the same day. In the study of the interas- The pyrolysis temperature of 400 ◦ C was chosen
say precision, the analytical solutions of Cd in urine in function of the characteristics of the peaks, the
pool were analyzed in five replicates for five consec- best absorbances values and the lower background.
utive days. The best absorbances were obtained at some temper-
The Cd concentration range was between 0.1 and atures lower than 400 ◦ C (200 and 250 ◦ C), however
4.0 ␮g l−1 . A Cd standard solution with a concentra- the peaks were not homogeneous. The atomization
tion of 4.0 ␮g l−1 and blanks were submitted to solu- temperature of 1200 ◦ C was chosen in function of the
bilization as described, and the calibration curves for characteristics of the peaks obtained and the accept-
each matrix were prepared automatically and in seven able absorbance values. At temperatures lower than
replicates. The Cd concentrations in the calibration 1200 ◦ C higher absorbances were obtained, however,
curves were 0.1; 0.2; 0.4; 0.8; 1.6; 3.2 and 4.0 ␮g l−1 . the peaks were not homogeneous and were wide.
Absorbance was read three times and the respective The absorption signals produced in the atomization
medium values were used for the construction of the stage generated a sharp peak, which is directly related
calibration curves. to cadmium concentration in the sample.
 C.J. de Castro Maciel et al. / Analytica Chimica Acta 491 (2003) 231–237 235

0,60

0,55

0,50
Integrated absorbance, s

0,45

0,40

0,35

0,30

0,25

0,20

0,15
200 400 600 800 1000 1200
o
Temperature, C

Fig. 1. Pyrolysis and atomization temperature curves for urine sample with 40 pg of Cd.

Pruszowska et al. [15], standardized higher temper- Table 2

atures, 700 and 1600 ◦ C for pyrolysis and atomization, Cd determination in certified urine samples (␮g l−1 ) in two levels
(n = 5)
respectively. In their work, the L’vov platform with
using (NH4 )2 HPO4 and HNO3 as chemical modifiers Certified Determined Certified Determined
was used. Such fact justifies the discovery of higher value 1 value 1 value 2 value 2
temperatures, since they stabilize the analyte, conse- 6.2 ± 1.2 5.8 ± 0.1 11.4 ± 2.3 11.8 ± 1.1
quently elevating the pyrolysis and atomization tem-
peratures.
Being analyzed certified samples of urine in two
concentration levels with the proposed method the re-
3.2. Analytical results
sults was in agreement for the two levels as it is ob-
served in Table 2.
Because the matrix matching calibration and the
The precision of the analytical method expressed
standard addition curves were parallel, the calibration
by the coefficient of variation intra- and interassays is
was made with matrix matching (5 ␮l of urine in all
presented in Table 3.
of points of the calibration curve). In all the studies,
The coefficient of variation interassays with lower
the matrix matching calibration curves presented an r
Cd concentrations presented a considerably higher
higher than 0.99.
The method presented a linear answer for Cd con-
Table 3
centrations in the range from 0.1 to 4.0 ␮g l−1 . This Coefficients of variation intra- and interassay observed in the
range is appropriate for the objective of this work, be- determination of urinary Cd
cause it allows the determination of basal Cd values,
Cd (␮g l−1 ) COV intraassay (%) COV interassay (%)
and with the appropriate sample dilution of, occupa-
tional exposure levels. The limit of detection of the 0.1 8.0 14.0
method obtained was of 0.08 ␮g l−1 , with a quantifi- 1.0 2.9 11.4
Average 5.5 12.7
cation limit (QL) of 0.10 ␮g l−1 .
236 C.J. de Castro Maciel et al. / Analytica Chimica Acta 491 (2003) 231–237

variation, probably due to either the contamination of Table 5

the glasses or the conditions of the graphite tube due Values of Cd (␮g l−1 ) in urine of individuals not exposed to the
metal at work as a function of sex
to the number of cycles of atomization.
Female (n = 144) Male (n = 115)

3.3. Recovery study Average ± s 0.38 ± 0.33 0.35 ± 0.33

Maximum value 1.87 1.53
The recovery varied from 89.0 ± 3.8 for Cd in the s: standard deviation.
concentration of 0.5 ␮g l−1 and 92.8 ± 1.1 for a con-
centration of 3.0 ␮g l−1 . This recovery values were Another study was made to compare Cd urinary
considered appropriate. levels in men and women (Table 5).
Spearman correlation test was positive for the con-
3.4. Influence in the Cd urinary concentrations and centration of urinary Cd and sex. The values of the
kind of area that people live, sex, age, tabagism and metal in the samples of the female group were signif-
alcoholic beverage intakes icantly higher than the those of the male one. This re-
sult is in agreement with the data of Buchet et al. [20]
The concentration of urinary Cd in the population and also of Jarup et al. [3], which described that in
resident in the metropolitan area varied between 0.03 general women display higher urinary Cd concentra-
and 1.56 ␮g g−1 of creatinine, with a medium value of tion than men due to the greater intestinal absorption
0.29 ± 0.26 ␮g g−1 of creatinine and a median equal of Cd as a result of a smaller level of corporal iron.
to 0.21 ␮g g−1 of creatinine. In the urban non-polluted For the age related study, the group was divided
area, the range of metal in urine samples was from 0.08 into three age subgroups, 15–30, 31–45, 46–60 years
to 2.03 ␮g g−1 of creatinine, with a medium value of (Table 6).
0.36 ± 0.27 ␮g g−1 of creatinine and a median equal The Spearman test showed a positive correlation
to 0.24 ␮g g−1 of creatinine. between urinary Cd and age (P = 0.019), while the
Theoretically, it was expected that the values of Cd multiple comparison test (Student-Newman–Keuls)
for the population resident in the polluted metropoli- demonstrated a significant difference only between
tan area (group 1) should be higher than those of the the 15–35 and 46–60 years age ranges.
non-polluted one (group 2). In practice, the values Roels et al. [21] also found a positive correlation
in both areas were quite similar probably due to the between urinary Cd and age in the two control groups
significant difference in individuals’ medium ages of studied with ages under and over 50 years, while Hotz
groups 1 and 2, which will be discussed later. Another et al. [22] did not find any solid interaction between
possible explanation can be the presence of Cd in the urinary Cd and age in the study of low levels of Cd
fertilizers, thus increasing the exposure of the individ- environmental exposure in Belgium.
uals of group 2. No correlation was found between tabagism and Cd
Considering that the regional differences did not urinary values, test of Mann–Whitney (P > 0.05).
influence the Cd basal concentration significantly in Alcoholic beverage intake did not show any signifi-
the populations studied further analyses were car- cant influence on Cd urinary levels. The Spearman test
ried out considering the two groups as a single one did not reveal a correlation between the parameters
(Table 4).
Table 6
Table 4 Values of Cd (␮g l−1 ) in urine as a function of age
Determination of urinary Cd in population (n = 259) resident in Age group (years)
Minas Gerais, Brazil
15–30 31–45 46–60
Cd (␮g l−1 ) (n = 116) (n = 89) (n = 54)
Average ± s 0.37 ± 0.33 Average ± s 0.33 ± 0.33 0.39 ± 0.34 0.41 ± 0.31
Maximum value 1.87 Maximum value 1.87 1.28 1.24
s: standard deviation. s: standard deviation.
 C.J. de Castro Maciel et al. / Analytica Chimica Acta 491 (2003) 231–237 237

(P > 0.05) and no significant difference was found [2] W.H. Porter, T.P. Moyer, Clinical toxicology, in: C.A.
in the levels of Cd between individuals who ingested Burtis, E.R. Ashwood (Eds.), Tietz Textbook of Clinical
alcoholic beverages (Mann–Whitney test, P > 0.05). Chemistry, second ed., W.B. Saunders Company, Philadelphia,
1994.
[3] L. Jarup, M. Berglund, C.G. Elinder, G. Nordberg, M. Vather,
J. Work Environ. Health 24 (1998) 1.
4. Conclusions [4] G.D. Carmack, M.A. Evenson, Anal. Chem. 51 (1979) 907.
[5] D.B. Tonks, Cádmio, in: A.J. Pesce, L.A. Kaplan (Eds.),
Quı́mica Clı́nica: métodos, Buenos Aires, Panamericana,
The use of fast acid digestion of urine samples was
1990.
shown to be appropriate for the determination of Cd [6] G.F. Nordberg, Effects and Dose–Response Relationships of
urinary levels by ET AAS. The results obtained agree Toxic Metals, Elsevier, New York, 1976.
with those of certified samples in two concentration [7] L. Jarup, C.G. Elinder, Br. J. Ind. Med. 50 (1993) 598.
levels. The quantification in all the cases was made [8] Z.A. Shaikh, J.B. Northup, P. Vestergaard, J. Toxicol. Environ.
Health 57 (1999) 211.
with matrix matching calibration curves. An LOD
[9] C. Han, Biomed. Environ. Sci. 1 (1988) 323.
of 0.08 ␮g l−1 , with satisfactory inter- and intraas- [10] A.N. Onar, A. Temizer, Analyst 112 (1987) 227.
says COVs were obtained. Recovery of spiked urine [11] T.C. Pan, C.J. Horning, S.R. Lin, T.H. Lin, C.W. Huang,
samples showed acceptable results. Concentrations Biol. Trace Element Res. 38 (1993) 233.
in individuals without history of occupational expo- [12] C.J. Horn, Analyst 121 (1996) 1511.
[13] W. Lund, B.V. Larsen, Anal. Chim. Acta 81 (1976) 319.
sure to Cd from the two different areas did not show
[14] P. Allain, Y. Mauras, Clin. Chim. Acta 91 (1979) 41.
any significant difference. A significant correlation [15] E. Pruszdowska, G.R. Carnick, W. Slavin, Clin. Chem. 29
was observed between sexes, with higher values for (1983) 477.
women. Age also had an influence on urinary Cd, [16] P.C. D’Haese, L.V. Lamberts, L. Liang, F.L. Van de Vyver,
being associated to higher values for individuals over M.E. De Broe, Clin. Chem. 37 (1991) 1583.
[17] K. Voloder, Clin. Chem. Acta 25 (1969) 373; B. Rudolf
46 years. No correlation was found between either
(Ed.), A Handbook of Decomposition Methods in Analytical
tabagism or the ingestion of alcoholic beverages and Chemistry, T&A Constable Ltd., Edinburg, 1979, p. 201.
the levels of urinary Cd. [18] J.P. Buchet, R. Lauwerys, H. Roels, A. Bernard, P. Bruaux,
F. Claes, G. Ducoffre, P. Plaen, J. Staessen, A. Amery, P.
Lijnen, L. Thijs, D. Rondia, F. Sartor, A.S. Remy, L. Nick,
References Lancet 336 (1990) 699.
[19] H.A. Roels, R. Lauwers, A. Bernard, J.P. Buchet, Br. J. Med.
[1] M.O. Amdur, J. Doull, C.D. Klaanen, Casarett and Doll’s 48 (1991) 365.
Toxicology: the Basic Science of Poison, fourth ed., Pergamon [20] P. Hotz, J.P. Buchet, A. Bernard, D. Lison, R. Lauwerys,
Press, New York, 1996. Lancet 354 (1999) 1508.