J. TraceE[em. Med. Biol. VoL 16, pp.

221-225 (2002)
http://www.urbanfischer.de/journa[s/jtracee[m
..... ~ ~ 0 ; ~ !
© 2002 by Urban & Fischer Ver[ag

Determination of chromium in cerebrospinal fluid

using electrothermal atomisation atomic absorption
spectrometry
Maria V]ctorina Aguilar*, Carmen Jos6 Mateos and Mafia Carmen Mart~nez Para

Depto. NutritiOn y Bromato[og~a,Facu[tad de Farrnacia, Universidad de A[caL~de Henares, Madrid, Spain

Received February2001 - Accepted Mai 2002

Abstract
A rapid method to determine the chromium content of cerebrospina[ fluid (CSF) samples using electrothermal
atomization.atomic absorption spectrometry (ETA-AAS)with deuterium-arc background correction is described.
The chromium concentration in CSFwas evaluated by the standard addition method. Sample dilution (1 + 1) with
0.25% (m/v) Triton X-IO0 and 4.5% (v/v) HNO3 gave the best combination of sensitivity, reproducibility, and low
blank reading compared with dilution using other solvents. Within-batch reproducibility was 3.2% for 20 CSF
samples, between-batch reproducibility was 4.7%. CSFsamplesfrom 43 healthy volunteers collected in a manner
designed to avoid contamination yielded chromium concentrations of :14.6+ 6.3 ng mL-~.

Key words: electrothermal atomization atomic absorption spectrometry, chromium determination, cere-
brospina[ fluid

Introduction
A number of studies have associated changes in normal
chromium levels with neurological disorders (1) and brain
tumors (2). The association has primarily been established
for serum levels, whereas little information is available for
cerebrospina[ fluid (CSF) (3-4), even though it is a better
indicator of possible neurological alterations, because coL-
lection of samples is harder and more traumatic.
Determination of Cr in biological material is one of the
most difficult analyses to perform. Levels of that eLement
are very low in such biological samples as serum, plasma,
and urine, as well as in CSF, in which it must cross the
blood-brain barrier. This is not the only difficulty that has
*Correspondence to: M.V. Agui[ar, Depto. Nutrici6n y Bro-
matoLog~a, Facultad de Farmacia, Universidad de A[cal~ de
Henares, 28871 Alca[~ de Henares, Madrid, Spain, Phone:
1-8854641, Fax: 1-8854783,
E-mail: mvictorina.agui[ar@uah.es
to be overcome when performing such determinations. In
addition, there is the ubiquitous problem of sample con-
tamination during coLLection and handling and the
absence of reference material
This accounts for the Lack of punished studies on
determinations of Cr in CSE Moreover, the few literature
references that do exist have employed small numbers of
samples. By way of example, EL Yazigi et aL (2) erected
such determinations in patients with brain tumors,
[eukaemia, and other non-neurologicaL alterations. The
mean values reflect a broad spread in the results depend-
ing upon the type of disorder, but sometimes there was
only a single determination (one patient).
PubLished papers have ordinarily employed eLectrother-
maL atomic absorption spectrometry, because it tends to
yietd reliable results, though it can also give rise to
matrix interference problems (5). Interferences can be
avoided in serum and urine samples by employing back-
ground corrections using a tungsten-iodide background

0946-672X/02/16/04-221 $15.00/0
222 M.V. Aguitaret at.
correction Lamp or Zeeman-effect background correction
(6). There is a considerable dispute regarding the possible
use of deuterium-arc correction Lamps. According to the
work of Hat[s and FeLl(7), that type of correction is feasi-
ble when there is a balance between the energy levels of
the deuterium-arc lamp and the hoLLowCr cathode and the
atomization temperature is below 2400 °C. In addition,
Baxter et aL (8) have demonstrated that the deuterium
arc can in fact provide adequate background correction at
the 357.9 nm wave[ength of Cr, provided the atomization
temperature is carefuLLyselected.
Sometimes the organic matter has been destroyed by
means of acid mixtures (9) using conventiona[ methods or
microwave digestion (10-11), or by means of enzymatic
methods (12). However, these all involve complicated
sample handling procedures that may entail the risk of
contamination and errors, and such errors are particularly
significant in samples of very smal[ Cr content.
For at[ those reasons, the object of the present study
was to optimize the direct determination of Cr in CSF
using electrothermal atomization atomic absorption spec-
trometry, with a view to develop a fast, simple, and pre-
cise method to determine Cr LeveLsand apply that method
to samples from a healthy population.
Material and methods
Instrumentation
The work was carried out using a Perkin-ELmer mode[ HGA
furnace instaLLed in a Perkin-Elmer mode[ 3110 atomic
absorption spectrometer (Perkin-Etmer, Norwalk, CT, USA)
with deuterium-arc background correction equipped with
a Perkin-Elmer mode[ As 40 autosampLer and an Epson
mode[ LQ-570 recorder. The emission source was an Inten-
sitron hollow-cathode Lamp, and pyrolytic graphite-coat-
ed tubes with L'vov platforms, all from Perkin-Etmer, were
used. The purge gas was argon.
Reagents
Standard chromium solutions were prepared from a i g • L-1
CrCl3 solution (Tritiso[, Merck, Germany). Nitric acid, mag-
nesium nitrate, and Triton X-IO0 were all from Merck.
SoLvent solutions: Triton X-IO0 (0.1-2% m/v) in 4.5%
v/v nitric acid, magnesium nitrate (0.5-1.5% m/v), and
nitric acid (0.2-4.5% v/v) were all prepared using
deionised water.
Calibration solutions were prepared daily in the solvent
solutions tested.
TaMe 1. Instrument settings TaMe 2. Furnace conditions used in the determination of Cr in CSF
Wavelength (nm) 357.9 Cycle
SLitwidth (rim) 0.7
Drying
Lamp current (mA) 10 Ashing (stage 1)
Background correction on Ashing (stage 2)
Atomization
Sample volume (pL) 20 CLeaning
J. TraceE[em.Ned. BioL.16/4 (2002)
SampLes
The Cr content was assessed in CSFsamples from 43 human
volunteers recruited from out-patients attended at the
Neurology Departments of two city hospitals, all of whom
had been shown not to be suffering from any neuroLogicaL
alterations. Informed consent was obtained from each sub-
ject. The foLLowing exclusion criteria were applied: a)
ethanol intake greater than 80 g/day during the preceding
six months; b) a previous history of chronic hepatopathy,
gastrectomy, pancreatic disease, diseases causing maLab-
sorption, and chronic renal failure; c) atypical dietary
habits; d) malnutrition; e) a previous history of severe sys-
temic disease. Lumbar CSF samples were taken from fasted
patients between 8:00 a.m. and 10:00 a.m. ALL specimens
were collected in plastic containers previously treated with
20% (v/v) nitric acid and rinsed with deionised water.
Trau_matic spinal punctures were also excluded from the
study. The CSF samples were immediately frozen (-30 °C)
and protected from exposure to Light until analysis.
Control of contamination
ALLcontainers, glassware, pipettes, micropipette tips, and
autosampLer cups were washed in 20% (v/v) nitric acid for
24 h and then rinsed thoroughly with distiLLed water
before use. Puncture needles were si[iconized, as the sili-
conizing process apparently imparts sufficient hydropho-
bicity to the needle (13) to prevent contact of the CSF
with the metal surface.
Instrumental conditions
The spectrometer settings used are Listed in TabLe 1 below.
In normal operation, to achieve a balance between the
hoLLow-cathode Lamp energy and the deuterium arc ener-
gy, the lamp current was turned down to about 10 mA and
the deuterium arc energy was turned up to its maximum
value.
Results and discussion
Choice of the heating programme
The furnace programme conditions are given in TabLe 2
below.
Conditions for the drying stage were set to achieve
complete drying of the CSF samples without boiling. When
drying was carried out at that temperature, the fumes pro-
duced in the next stage decreased considerably.
Ashing had to be carried out in two stages in order to
eliminate double peak formation. Ashing at 1200 °C
removed residual inorganic salts (14). Fig. 1 depicts the
Temperature Ramp time Hold time Argon flow-rate
(°C) (s) (s) (mk. min"1)
120 6 15 3
1000 10 0 3
1200 1 10 3
2500 0 8 0
2650 1 5 3
 Chromium in cerebrospina[fluid 223

results of optimization of the second ashing stage. The
ramp time required to reach the ashing temperature was
10 s, and the atomization temperature was held constant
at 2500 °C throughout the tests. Each data point in the
figure represents the mean of several determinations.
The atomization temperature of 2500 °C was shown to
be suitab[e by absorbance measurements at temperatures
of 1800-2600 °C. The response increased until a tempera-
ture of 2500 °C had been reached, after which there was
no further improvement (Fig. 2).
HaL[s and FeLL (15) encountered an interference at
2700 °C, attributable to emissions by the sodium and
potassium present in the matrix. That interference was
reduced by adding a surfactant at 2500 °C. In addition,
VeiL[on et at. (16) reported a sensitivity Loss with pyroLyt-
ic graphite-coated tubes. McAughey and Smith (14) report-
ed that Lowering the atomization temperature from 2700
to 2500 °C improved the reproclucibi[ity and increased the
tube Lifetime. No interference of these types was recorded
in our study.
Fig. 3 depicts the atomic signals obtained using the
optimum temperature programme. EvaLuation of the atom-
ic signal in either the peak height or peak area mode had
no influence on the anaLyticaL results.
0.178-

0.7
0.6
c
..o 0.5 ~ r
~ 0.4
i( :
~ 0.2
,-, 0.1 f !,
0
i
900 950 1000 11O0
Ashing temperature/°C Time (sec) 10.0
Fig. 1. Effect of the ashing temperature on the peak area for a Fig. 3. Absorbance profile of a CSF sample spiked with
solution of 4 ng • mL-~ of chromium added to pooled CSFsamples 60ng • mL-1 of Cr using deuterium arc background correction.
(n = g) diluted as described. The background signaLis aLso shown.

0.7 0,2

0.6
8 o5
~0.4
o
0.3 /
-~ 0.2
~-0.1
0
1 2 3 4 5
1800 2000 2200 2500 2600 Chromium ¢oncentraEons ngfml
Atomization temperature/ oC
[-~-1 ~ -2-~-3 ~ ~ ~5-4~-6 ~--7 ---8 ~-9 ~.-~0 t

Fig. 2. Effect of the atomization temperature on the peak area Fig. 4. CaLibration slopes for chromium in Triton X-IO0 and
for a solution of 4 ng - mL-1 of chromium added to poo[ed CSF 4.5% HNO3 (serie 10) and nine specimens of CSF (1-9) by the
samp[es (n = 9) dituted as described. method of standard addition.

TaMe 3. VaLuesobtained for samples dissolved in different soLvents

SampLe Triton Magnesium nitrate Nitric acid

0.1% 0.25% 2% 0.5% 1% 1.5% 0.2% 1%

BLank 0.057 0.016 0.011 0.034 0.008 0.011 0.063 0.022

SampLe 1 cv 5.3 0.65 72.1 12.9 15.9 2.20 19.2 9.4
a 0.076 0.089 0.046 0.069 0.038 0.033 0.030 0.031
SampLe 2 cv 10.3 9.02 11.1 29.1 12.5 38.2 25.9 0.98
a 0.070 0.089 0.048 0.093 0.037 0.033 0.039 0.027
SampLe 3 cv 9.6 3.71 3.1 66.2 22.3 11.8 12.6 25.5
a 0.068 0.084 0.043 0.146 0.049 0.032 0.033 0.041
SampLe 4 cv 10.6 0.487 8.3 34.9 28.5 37.9 11.7 2.0
a 0.058 0.I04 0.048 0.964 0.036 0.032 0.031 0.026
cv: coefficient of variation (%)
a: slope of regression fine (ng• mL-1)

J. TraceE[em.Med. BioL 16/4 (2002)

224 M.V. Aguilar eta[.
SoLvent
Four CSF samples were prepared for analysis by dilution
using nitric acid only (0.2%, and 1% v/v), or using Triton
X-IO0 (0.1%, 0.25%, and 2% m/v) or magnesium nitrate
(0.5°1o, 1%, and 1.5°1o m/v) dissolved in 4.5°1o (v/v) nitric
acid. Table 3 sets out the values obtained for each of the
solvents alone and for the diluted samples.
The best combination of sensitivity, reproducibility, and
low blank value was obtained using the 0.25% (m/v) Tri-
ton X-100. The 0.25% (m/v) Triton X-100 solution in
4.5% (v/v) nitric acid was therefore used for all subse-
quent measurements.
Calibration
The concentrations of Cr in nine CSF samples were deter-
mined with reference to the Cr calibration solutions pre-
pared using the Triton X-100 in 4.5% nitric acid solvent
solution and by the standard addition method. The slopes
obtained for the different CSF samples varied by no more
than 17.33% (Fig. 4). The absorbance for the endogenous
sample was subtracted. The different slopes obtained for
Cr in the solvent mixture of Triton X-100 in nitric acid
indicated that there was a marked CSF matrix effect.
Sensitivity
The characteristic mass in pg of Cr required for an average
absorbance area of 0.0044 A.s. was 3.4 pg for the aqueous
standard in the solvent Triton X-100 in nitric acid. That
value showed a good agreement with published values
(17-19).
Precision and detection threshold
The method's precision was checked by determining Cr in
20 CSF samples several times a day for 15 days. Within-
day variation (coefficient of variation) was _+ 3.2%, and
the day-to-day variation was _+4.7%.
The detection threshold calculated as twice the stan-
dard deviation for the blanks was 11.8 pg, and the detec-
tion threshold calculated as twice the standard deviation
for readings at the lowest absorbance level was 1.22 pg.
Analytical recovery
There are no certified reference materials for CSF. For that
reason, the recovery was determined by adding Cr
(10 ng • mL-1) to pooled CSF samples (n = 9). Good recov-
eries were obtained using the Triton X-100-based solvent.
The percentage of recovery was 96.9 _+ 2.9%.
Concentration of Cr in CSF
The samples from the 43 volunteers had Cr concentra-
tions of 14.6 + 6.3 ng • mL-1. Those values were slightly
higher than the values reported by certain other
researchers, such as El Yazigi (2), who recorded control
group values of 6.6 ng. mL-1. A comparison with the
findings reported by other groups is more difficult,
because their Cr determinations were performed using
neutron activation analysis and was therefore expressed
as mass percent of ashed or dried CSF (20). Other inves-
tigators have sometimes reported the concentration only
as "traces" (21-22).
J. TraceElern.Med. Biol. 16/4 (2002)
Conclusions
The proposed method revealed itself to be suitable for
determining chromium in CSF, yielding both good analyti-
cal parameter values and affording other advantages, such
as reduced sample handling, Low wear of the pyrolytic
tubes, and the possibility to use instruments equipped
with deuterium arc background correction, making it use-
fu[ as a routine method of determination that can be used
by many laboratories.
The method proposed, tested here on samples of CSF
collected from individuals without neurological alter-
ations, is equally suitable for use with samples from
patients who do present neurological disorders.
References
1. Jim~nez-Jim~nez FJ, Molina JA, Aguilar MV, Arrieta FJ,
3orge-Santamaria A, Cabrera-Va[divia F, Ayuso-Pera[ta L,
Rabasa M, V~zquez A, Garcia-A[bea E, and Martinez Para ME
(1995) Serum and urinary manganese levels in patients with
Parkinson's disease. Acta Neural Scand 91:317-320
2. El-Yazigi A, Martin CR, and Siqueira ED (1988) Concentra-
tions of chromium, caesium and tin in cerebrospinal fluid of
patients with neoplasms, leukaemia or other noncerebral
malignancies, and neurological diseases. Clin. Chem. 34:
1084-1086
3. Kuang P, Wu W, and Lang S (1993) Trace elements and
ischemic cerebral vascular disease. Ann. NY. Acad. Sci. 15:
340-341
4. Shi MT (1991) Determination of multiple chemical elements
in CSF in Parkinson disease after intracerebral autotrans-
p[antion of the adrenal medulla. Chung Hua Wai Ko Tsa Chih
29:129-132
5. Dube P (1988) Determination of chromium in human urine
by graphite furnace atomic absorption spectrometry with
Zeeman-effect background correction. Analyst 113:
917-921
6. Jackson KW, and Lu S (1998) Atomic absorption, atomic
emission, and flame emission spectrometry. Anal. Chem. 70:
363R-383R
7. Halls DJ, and Fell GS (1988) Faster determination of chromi-
um in urine by electrothermal atomisation atomic absorp-
tion spectrometry. 3. Anal. At. Spectrom. 3:105-109
8. Baxter DE, Litt[ejohn D, Ottaway JM, Fell GS, and Halls DJ
(1986) Determination of chromium in urine by probe etec-
trothermat atomisation atomic emission spectrometry using
a low-resolution monochromator. J. Anal. At. Spectrom. 1:
35-39
9. Lugowski SJ, Smith DE, McHugh AD, and Van Loon 3C (1991)
Determination of chromium, cobalt and molybdenum in syn-
ovial fluid by GFAAS. J. Trace E[em. Electrolytes Health Dis.
5:23-29
i0. Alvarado 3, Moreno R, and Cristiano AR (1991) Determina-
Lion of Cd, Cr, Cu, Pb and Zn in human semen by graphite
furnace atomic absorption spectrometry after microwave
sample dissolution. J. Trace Elem. Electrolytes Health Dis. 5:
173-180
11. Gal[o P, Serpe L, and BrambiIla G (1997) Determination of
chromium in feeds by automated microwave digestion and
atomic absorption spectrometry. J. AOACInt. 80:956-960
12. Offenbacher EG, Dowling Hi, Rinko CJ, and Pi-Sunyer F
(1986) Rapid enzymatic pre-treatment of samples before
determining chromium in serum or plasma. Clin. Chem. 32:
1383-1386

13. Ottaway JM, and Fe[[ GS (1986) Determination of chromium
in biological materia[s. Pure & App[. Chem. 58:1707-1720
14. McAugheyJJ, and Smith NJ (1987) Automated direct deter-
mination of chromium in blood and urine by e[ectrotherma[
atomic absorption spectrometry. Anal Chim. Acta. 193:
137-146
15. Hal[s DJ, and Fell GS (1986) The problem of background cor-
rection in the determination of chromium in urine by atom-
ic absorption spectrometry with e[ectrotherma[ atomisation.
J. Anal At. Spectrom. 1:135-139
16. Veil[on C, Patterson KY, and Bryden NA (1984) Determina-
tion of chromium in human serum by e[ectrotherma[ atomic
absorption spectrometry. Anal Chim. Acta. 164:67-76
17. Lagarda MJ, A[onso de Armi~o V, and Farr6 R (1991) The use
of direct determination of chromium in human urine by
e[ectrotherma[ atomic absorption spectrometry in diabetic
patients. J. Pharmaceutica[ & Biomedica[ Analysis 9:
191-194
Chromium in cerebrospinal fluid 225
18. Paschal DC, and Bailey GG (1991) Determination of chromium
in urine with graphite furnace atomic absorption spectroscopy
using Zeeman correction. Atomic Spectroscopy 12:151-154
19. Miller-lhli NJ, and Greene FE (1992) Graphite furnace atom-
ic absorption method for the determination of chromium in
foods and biological materials. J. AOACInt. 75:354-359
20. Shevaga VN (1971) Molybdenum and chromium levels in the
blood and spinal fluid of patients with brain tumors. Vrach
De[o (USSR) 6:97-99
21. Goody W, Williams TR, and Nicholas D (1974) Spark-source
mass spectrometry in the investigation of neurological dis-
ease, I. Multie[ement analysis in blood and cerebrospina[
fluid. Brain 97:327-336
22. Goody W, Hamilton El, and Williams TR (1975) Spark-source
mass spectrometry in the investigation of neurological dis-
ease, II Element levels in brain, cerebrospina[ fluid and
blood: some observations on their abundance and signifi-
cance. Brain 98:65-70
J. TraceELem.Med. BioL 16/4 (2002)