CLIN. CHEM. 31/1, 24-29 (1985)
Measurement of Aluminum in Serum, Blood, Urine, and Tissues of Chronic
Hemodialyzed Patients by Use of Electrothermal Atomic Absorption
Spectrometry
Patiick
C. D’Haese,’ Frank L. Van de Vyver,1 Frederik A. de Wolff,2Marc E. De Broe”3
We describe different methods of electrothermal atomic
absorption spectrometry with automatic sampling for determining aluminum in human serum, blood, urine, and tissues.
Contamination with Al originating from receptacles and reagents was minimized. Whole-blood Al concentrations were
measured
after hemolysis of EDTA-anticoagulated
blood
samples with Triton X-100. A straightforward method for
tissue destruction was developed. Instrument settings of the
graphite furnace and of the atomic absorption spectrometer
were adjusted so as to obtain close agreement between
direct and standard-additions methodologies. The result is a
reliable direct method appropriate for use with multiple
samples. Tissue Al measurements showed low detection
limits and approximately 100% analytical recoveries. Al concentration in serum of 10 healthy volunteers and of 100
chronic hemodialysis patients were 2.0 (SD 0.4) and 77 (SD
70) g/L, respectively. Blood Al concentrations of 10 controls
and of 100 dialysis patients were 12.1 (SD 1.5) and 79 (SD
70) gIL, respectively. Al concentrations in serum and blood
of 47 chronic hemodialysis patients were not significantly
different. Bone Al concentrations were 8.2 (SD 5.8) tg/g of
fresh tissue for 10 chronic hemodialysis patients without
osteomalacia, as compared with 51 (SD 20) /g
for 10
chronic-dialysis patients with Al-induced osteomalacia.
Additional Keyphrases: liver
/yjne
trace elements
tissue destruction
osteomalacia
standard-additions methods
reference interval
.
Accumulation of aluminum in the tissues of patients on
chronic hemodialysis may cause dysfunction of some organs.
In 1976, Aifrey et al. (1) attributed
the previously described syndrome of dialysis dementia to an increased
aluminum concentration in the brain of their encephalopathic patients. Today, the association between increased
aluminum concentrations in the cerebrum and dialysis
encephalopathy seems reasonably well established (2).
In 1971, Parsons et al. (3) demonstrated abnormally high
aluminum concentrations in the bone of patients on chronic
hemodialysis. More direct evidence for a causal association
between dialysis osteomalacia and aluminum accumulation
in bone has been provided by experimental and human
studies of bone aluminum concentrations and bone histology
(4).
More recently,
microcytic
hypochromic
anemia
has been
‘Department
of Nephrology-Hypertension, University of Antwerp (UIA), University Hospital Antwerp, Edegem, Belgium.
2Toxicology Laboratory, University
Hospital, Leiden, The Neth-
erlands.
3Address correspondence to this author at: Department of
Nephrology-Hypertension,
University of Antwerp, University Hospital Antwerp,
Wih-ijkstraat
10, B-2520 Edegem, Belgium.
Received April 23, 1984; accepted October 5, 1984.
24
CLINICAL CHEMISTRY, Vol. 31, No. 1, 1985
reported in patients with excessive aluminum
accumulation
(5).
Serum
aluminum concentrations may fluctuate as a result of oral (6) or parenteral (7, 8) administration of aluminum-containing compounds. This hampers the use of serum
aluminum values in monitoring tissue aluminum accumulation. In searching for more reliable parameters of aluminum accumulation in patients with severe renal failure, we
have developed accurate procedures for measuring alununum in blood and tissues. Measurements of other trace
metals in blood are widely used to detect intoxication.
Electrothermal atomic absorption spectrometry (ETAAS)
has become the method of choice for measuring aluminum
in biological samples: the required sample size is small, the
graphite furnace is capable of attaining the high temperatures needed, and in comparison with flame analysis the
atoms are in the light path longer, resulting in increased
sensitivity.
Thus ETAAS seems to offer the best combination of sensitivity, simplicity, and low cost.
As with most recently published methods for serum
aluminum measurement (9), the methods presented here for
serum, blood, urine, and tissue aluminum determination
share the meticulous attention to avoid contamination, an
optimal drying and a two-stage ashing cycle, and some prior
dilution of the samples. Adjustment of time and temperature settings of the graphite furnace resulted in good
agreement between results by the standard-additions
and
direct methods. A straightforward
method for tissue destruction was developed, which is applicable both for dry
weight and wet weight measurements.
Materials and Methods
Sampling
Venipuncture.
For venipuncture we used 10-mL sterile
syringes
(Monovette;
Sarstedt,
NUmbrecht,
F.R.G.)
equipped with 18-gauge syringe needles (Terumo Europe
N.y., Haasrode, Belgium).
Serum. After coagulation
and centrifugation,
the serum
was transferred to 5-mL polystyrene tubes (Biolab, Limal,
Belgium) with tight-fitting
polyethylene caps (Biolab) by
means of a “Finnpipette” (Labsystems, Helsinki, Finland)
with disposable tips (Labsystems).
Blood. Blood samples were transferred immediately after
venipuncture into 5-mL tubes containing potassium EDTA
as anticoagulant (Brunswick, Division of Sherwood Medical
Ltd, Ballymoney,
N. Ireland).
Biopsies.
Transiliac
bone biopsies were taken approximately 2 cm beneath the dorsal from the anterior superior
iliac spine with a 7-mm-diameter Bordier-Meunier
trephine.
Tissues. Liver-tissue was sampled at autopsy by means of
plastic knives. After weighing, the samples were transferred
to polystyrene tubes with polyethylene
caps (Biolab).
Urine. Twenty-four-hour
urine specimens were collected
in 2-L plastic bottles (Sarstedt). After thorough mixing, 5
mL of each specimen was transferred
to i0-mL
tubes with tight-fitting
caps (Biolab).
polyethylene
polystyrene
Equipment
The analysis of serum was done in three laboratories,
each equipped with a different type of instrument from
Perkin-Elmer Corp., Norwalk, CT 06856.
Laboratory 1 used the Model 305 B atomic absorption
spectrometer, together with an HGA 74 graphite furnace.
An AS-i automatic sampler was used to dispense 10 L of
the sample solutions. The spectrometer output was recorded
with a Model 56 recorder. The graphite tubes were Uncoated.
Laboratory
ratories:
2 used the
Model
2380
atomic
absorption
spectrometer together with the HGA-500 graphite furnace,
an AS-40 autosampler to inject 20 tL of the sample, and a
Model 561 two-pen recorder together with a PRS-10 printer
sequencer. Zirconium-coated graphite tubes were used.
Laboratory 3 used a Model 372 atomic absorption spectrometer, together with an HGA-500 graphite furnace, and
an AS-40 autosampler to inject 20 L of the sample. The
results were plotted by use of a Model 023 recorder and a
PRS-10 printer. The graphite tubes were uncoated.
Aluminum concentrations in whole blood, urine and tissues were only determined in laboratory 3.
Materials
and Reagents
for Analysis
Precautions
to avoid aluminum contamination.
All materials coming in contact with the samples were tested as
potential sources of aluminum. Materials showing detectable aluminum (1 tg/L or more) after five-day contact with
doubly distilled water and nitric acid (100 mL/L) were
discarded. No glassware was used. Doubly distilled water
was used throughout.
Materials. Materials used were either 5-or i0-mL polystyrene tubes, 50-mL Teflon volumetric flasks with tightfitting Teflon caps (Brand GmbH, Wertheim, F.R.G.), quartz
Kjeldahl flasks (VEL, Leuven, Belgium), digestion bombs
(Uniseal, Haifa, Israel), Teflon tubes (VEL), and polystyrene
sample cups (Biolab). The automatic pipettes used were a
500-L “Assipette” (Igny, Paris, France), a 0-200 .tL and a
1-5 mL “Finnpipette”
(Labsystems), and a 50-pL Lancer
pipette (Lancer, Sherwood Medical Industries, St. Louis, MO
63103), all with disposable polypropylene pipette tips.
Reagents. Concentrated nitric acid (Suprapur 441; Merck,
Darmstadt, F.R.G.) and concentrated Triton X-100 (BDH
Chemicals Ltd., Poole, U.K.) were used to prepare a 20 mL/L
nitric acid and a 2 g/L Triton X-100 solution. These solutions
and an undiluted acetic acid solution (UCB, Leuven, Belgium) were stored in 250-mL polypropylene flasks.
Stock 1 g/L standards of aluminum chloride (J. T. Baker
Chemical Co., Phillipsburg, NJ 08865) and aluminum nitrate (BDH Chemicals Ltd.) were used to prepare the
standards in laboratory 1 and 3. Laboratory 2 used a selfprepared 1 g/L standard of aluminum chloride in water. For
dilutions,
water was used in which no aluminum
was
detectable.
Analysis
respectively, to 250 pL of a fresh serum pool having an
aluminum concentration of <50 ig/L.
In addition, laboratories 1 and 3 used a standard-additions method for reference. The standard-additions protocol
was as follows: Of each specimen of serum, 200-L aliquots
were pipetted into each of three sampling cups. Then, 400
1.L of water was added to the first, 200 L of water and 200
L of 200 g/L aluminum standard to the second, and 400
L of aluminum standard to the third. This resulted in a
threefold
dilution of the serum in each sampling cup,
equalizing any eventual matrix effects.
Spectrometer settings were comparable in the three labo-
Procedures
Serum. The serum samples were diluted threefold with
water. All three laboratories used a direct method. Standards for establishing the standard curves were aqueous in
laboratories 1 and 3, and serum-based in laboratory 2. In
laboratories 1 and 3, aliquots of a 1 g/L stock solution of
aluminum were diluted in water to yield aqueous working
standards of 10, 20, 50, and 100 g/L. In laboratory 2,
serum-containing
standards were prepared by adding 500
j.iL of 0, 20, 50, and 100 tgfL aqueous aluminum standards,
wavelength
309.3 urn; slit width
0.7 rim; signal
mode, peak height; hollow-cathode lamps used at a current
of 25 mA; no background correction; and purge gas, argon.
Oxygen was used as alternate gas in laboratory 2. Read
cycles were 10 s, 8 s, and 6 s, in laboratories 1, 2, and 3,
respectively.
Blood. For hemolysis we used a 2 g/L solution of Triton X100. The direct method consisted of fourfold dilution of the
blood samples in the Triton X-100. Aliquots
of a 1 g/L
aluminum stock solution were diluted in Triton X-iOO to
yield aqueous calibration
standards: 10, 20, 50, and 100
LgiL.
Furthermore,
was performed.
the following
standard-additions
protocol
Of a blood sample, 2#{174}-L aliquots were
transferred
to each of three sampling cups. Then 600 pL of
Triton X-100 was added to the first, 400 L of Triton X-100
and 200 /LL of 200 pgfL aluminum standard to the second,
and 200 pL of Triton X-100 and 400 L of aluminum
standard to the third-a
fourfold dilution
of the blood in
each sampling cup.
Urine. The low solubility product of aluminum phosphate,
the form presumed present in urine, may cause precipitation. To overcome this problem, we added pure acetic acid to
dissolve the aluminum phosphates (10). This resulted in
lower CVs. No direct method was used and the standardadditions protocol was as follows. Of a urine specimen, 500z.L aliquots were pipetted into each of three sampling cups.
Subsequently,
1 mL of water and 75 zL of pure acetic acid
were added to the first; 500 j.tL of water, 500 L of
aluminum standard (200 g/L), and 75 L of pure acetic
acid to the second; and 1 mL of aluminum standard, and 75
L of pure acetic acid to the third.
Tissue samples. All containers (Kjeldahl flasks and Teflon
containers) were leached before use in concentrated nitric
acid until aluminum was undetectable.
Three different destruction methods were tested. All three
are applicable both for dry-weight and wet-weight measurements.
Method 1: The biopsy specimen was quantitatively
transferred to a quartz Kjeldahl flask and 5 mL of pure nitric acid
was added. The flask was then electrically heated to 200#{176}C.
Destruction was considered complete when the solution
became clear. After evaporation of the remaining liquid, 1
mL of concentrated nitric acid was added. The destruction
liquid was quantitatively transferred to a Teflon volumetric
flask and adjusted to 50 mL with water. This solution was
then transferred to two 10-mL polystyrene tubes and stored
at 4 #{176}C.
Method 2: In this method the tissues were destroyed in
Teflon containers. Two kinds of Teflon containers were used:
Teflon tubes and digestion bombs consisting of a stainlesssteel holder, wherein a Teflon container is fitted. The holder
was tightly closed by means of a stainless-steel screw. The
sample was quantitatively
transferred into the Teflon contamer and 1 mL of concentrated nitric acid was added. The
whole was placed in an oven for 3 to 4 h at 80 to 90 #{176}C;
this
CLINICAL CHEMISTRY, Vol. 31, No. 1, 1985
25
resulted in a clear digest. Just as with method 1, this liquid
was quantitatively
transferred
to a Teflon volumetric flask,
diluted to 50 mL with water, then transferred to two 10-mL
polystyrene
tubes and stored at 4 #{176}C.
Method 3: The same procedure was followed as in method
1, except a perchloric/sulfuric/nitric
acid mixture (1/1/3 by
vol) was used instead of concentrated nitric acid.
Furnace programs for serum, blood, urine, and tissues are
outlined in Table 1.
JWW
Results
Serum
The calibration
curve was linear to at least 200 tg/L.
Good precision was achieved only with a two-step drying
procedure (see Table 1). With more rapid drying, spattering
may occur. Like other authors (9) we observed that reproducibility is ameliorated by adding a “clean-up” stage (i.e.,
the second atomization
stage, see Table 1), lasting a few
seconds, at the end of the analytical cycle. Figure 1 illustrates the good reproducibility
of the results obtained for
repetitive injections of serum containing 250 pg of aluminum per liter. All values for serum aluminum were determined in duplicate. For 50 serum samples with aluminum
concentrations
ranging
from
18 to 900 pg/L,
the
peak
heights corresponding to both injections were inter-compared. The linear correlation coefficient (r) was 0.998, andy
= 0.99x
0.2 pg/L was the equation for the best-fitting line.
This close agreement suggests the absence of carryover.
Between-run CVs (six runs) were 2.9, 2.9, 1.7, and 7.9% for
serum aluminum
concentrations of 19, 93, 224, and 582
pg/L, respectively.
With argon as the purge gas, a high sensitivity was
obtained. The apparatus of laboratory 3 provides facilities
for adjustment of the argon flow. By using an argon flow of
-
Fig. 1. Recorder tracing obtained for aluminum analysis: consecutive
injectionsof 15 aliquots prepared from the same serum sample(250 g
of Al per liter)
20 mlJmin instead of 300 mL/min,
the sensitivity was
improved 10-fold. This observation prompted us to evaluate
the relationship between argon flow and absorbance (Figure
2).
Detection
limit. The amount of aluminum yielding a 1%
absorption signal (0.0044 absorbance) was 20 pg, which
compares favorably with the lowest values found in the
literature (9). The detection limit was low, even when a
short ramp time, 1 s, was used at the atomization step.
Accuracy.
In laboratories 1 and 3 the standard-additions
method was used for reference in order to establish a
reliable direct method. In fact, time and temperature
settings of the heated graphite
atomizers
(Table 1) were
Table 1. Instrument Settings of the Heated Graphite Atomizer
Stages
Dry
SERUM in laboratory 1:
temp, #{176}C
2
Bum
RefIll
1
2
1
2
260
1400
2700
2700
10
300
7
300
-
2600
-
0
8
10
2600
0
4
300
2700
2700
1
1580
120
10
6
300
300
20
6
300
120
15
15
300
300
5
5
300
1530
25
10
300
2700
1
6
20
2700
100
15
15
300
700
5
5
300
1400
32
10
300
2700
1
6
20
2700
100
700
10
10
300
1500
30
10
300
2700
2700
90
ramp time, s
hold time, s
purge gas flow, mL/min
SERUM in laboratory 2:
temp, #{176}C
30
30
300
300
140
500
20
10
5
300
65
10
300
1
5
02 mi.
1
10
300
temp, #{176}C
100
120
ramp time, s
hold time, s
purge gas flow, mL/min
15
15
300
10
10
purge gas flow, mLfmin
SERUM in laboratory 3:
380
-
110
ramp time, s
hold time, s
Atomize
Ash
1
300
1400
1
30
300
700
2
-
300
-
BLOOD:
temp, #{176}C
ramp time, s
hold time, s
purge gas flow, mL/min
300
URINE:
temp, #{176}C
ramp time, s
hold time, s
purge gas flow, mLfmin
TISSUE:
temp, #{176}C
ramp time, 5
hold time, s
purge gas flow, mLimin
26
10
10
300
CLINICAL CHEMISTRY, Vol. 31, No. 1, 1985
6
300
1
6
20
6
300
Accuracy.
As for serum, time and temperature settings of
the heated graphite atomizer (Table 1) were adjusted until
the aqueous standard curves of the direct method were
parallel to the standard-additions
curves. For 45 blood
samples with aluminum content ranging from 23 to 490
we compared results by the two methods. The linear
correlation
coefficient was r = 0.994 and y = 0.96x + 0.1
pg/L was the equation for the best-fitting line (y standard
additions method; x = direct method). Minor differences
were comparable to the CVs of each analytical procedure.
1.00
=
Clinical
results.
Blood aluminum
concentrations
were
12.1 (SD 1.5) j.tg/L for 10 healthy controls and 79 (SD 70)
pg/L for 100 chronic hemodialysis patients. Blood (y) and
C
serum (x)
050
aluminum
hemodialyzed
values
patients.
The
were compared in 47 chronic
linear
correlation
coefficient
between the two was r = 0.97, andy = 0.81x + 11 pg/L was
the equation for the best-fitting line. The values for serum
and blood from hemodialysis patients were not significantly
different.
025
Urine
Precision. The between-run
and 4.6% for urine aluminum
78 pg/L, respectively.
100
200
3O0
Argon flow (rnL/rnin)
Fig. 2. Absorption signal as a function of argon gas flow during
atomization of an aqueous 200 tg/L aluminum standard
adjusted so as to yield identical values for aqueous standards and for serum-based samples with the same alumi-
num concentration. Thus, when the parameters of Table 1
are used, the aqueous calibration curves of the direct
method parallel those of the standard-addition curves. Analyses for aluminum in the same serum samples, performed in
the three laboratories using different instruments and different ETAAS procedures, showed good agreement. Linear
correlations
were r = 0.98, y = 0.91x + 8 pg/L, n = 30
between results in laboratories
1 (x) and 2 (y), and r = 0.98, y
=
0.92x + 8 pg/L, n = 14 between results in laboratories 1
(x) and 3 (y). In laboratory 3 a number of serum samples
were obtained from an international quality-control organization (the Institute
of Industrial
and Environmental
Health and Safety, University of Surrey, Guildford, Surrey
GU2 5XH). We compared our results (x) with the means of
those found in the other laboratories participating in the
quality-assurance
scheme (y): r = 0.99, y = 0.996x + 0.0025
pmol/L, n = 12. Our results (12 samples) differed by 0.34 SD
from the means of the values from the other laboratories.
Results obtained with aluminum
nitrate and aluminum
chloride standards were always equivalent.
Clinical results. Serum aluminum concentrations were
2.0 (SD 0.4) pg/L for 10 healthy controls and 77 (SD 70) pg/L
for 100 unselected chronic-hemodialysis patients.
Blood
Anticoagulation.
Before collecting blood samples we studied the aluminum content of different anticoagulants. Sodium citrate and lithium heparmn were rejected because of
their high aluminum content, and fluoride oxalate had to be
discarded because of irreproducible
signals at aluminum
analysis. Brunswick
potassium EDTA-containing
tubes
showed minor aluminum
contamination
and blood samples
collected therein yielded reproducible aluminum
signals.
Consequently,
we used only these tubes for blood collection.
Hernolysis.
By using Triton X-100 for hemolysis, we could
do the drying in one stage, and the digestion time required
was shorter than for serum.
CVs (six runs) were 5.9, 4.1,
of 3.9, 46, and
concentrations
Sensitivity
and detection limit. The addition of acetic acid,
which was required to avoid precipitation of aluminum
phosphates, decreased the absorbance by about a third. The
detection
pg.
limit
for urine
aluminum
determinations
was 26
Accuracy. Instrument settings were adjusted as for serum
and blood (Table 1). Because absorbance was affected by
acetic acid concentrations, the standard-additions
method
was used in all instances.
Clinical results. For 10 patients with severe renal failure
(creatimne clearance <15 mL/min) not yet requiring dialysis therapy, the 24-h urinary excretion of aluminum was 34
(SD 21) pg before and 48 (SD 39) pg after the oral administration of a single 7-g dose of aluminum hydroxide (Aludrox#{174}
gel, Wyeth).
Tissues
Precision. Three different methods for tissue destruction
were evaluated. Methods 1 and 2, both of which include
digestion with nitric acid, were accepted for use. The third
digestion procedure was rejected because of poor reproducibility.
Detection limit. The least detectable amount of aluminum
in the 20-giL injection volume was 20 pg. Thus, there is a
detection limit of 50 ng for the 50 mL of digest, corresponding to 50 ng/g for a tissue sample with 1.0 g wet weight.
Accuracy.
Aqueous standards were subjected to the first
two digestion procedures. Analytical recoveries of aqueous
aluminum-SO,
100, and 200 pg/L, respectively-were
102,
103, and 99% for method 1, and 98, 99, and 98% for method
2. To evaluate the linearity of the absorbance as a function
of tissue sample weight, we determined tissue aluminum in
20 liver samples and in 12 transiliac bone specimens obtained at autopsy in one single chronic hemodialysis patient. For the liver, 18 samples were examined, ranging in
weight from 0.54 to 1.95 g. Twelve were digested by method
1 and six with method 2. The linear correlation between
sample weight and absorbance was r = 0.98. For the bone,
six samples were digested with method 1, and method 2 was
applied to the other 6. Observed bone aluminum levels were
9.2 (SD 1.1) pg/g for the first, and 9.4 (SD 2.6) pg/g for the
second method.
To rule out interference with the assay of different ions
commonly present in tissues, we brought these ions into a 20
mL/L nitric acid solution (50 mL) and added aluminum
CLINICAL CHEMISTRY, Vol. 31, No.1, 1985
27
standard (10 pL, containing 10 pg of aluminum). These
solutions were measured as they were digests. Table 2
indicates that none of these ions have any significant effect.
Clinical results. Bone aluminum concentrations, determined for 10 chronic hemodialysis patients without histological osteomalacia, were 8.2 (SD 5.8) pg/g of fresh tissue.
For 10 chronic-dialysis patients with aluminum-induced
osteomalacia, bone aluminum values were 51 (SD 20) pg/g.
Discussion
Contamination.
Contamination
with aluminum arising
during blood sampling and from handling devices may be a
most important factor in the wide variation of values
reported (11). It is striking that reported normal values for
serum aluminum go up with the increasing complexity of
the procedure before the actual detection. In the procedures
presented here, we attempted to minimize contamination
by
minimizing
manipulations.
The automatic injection structure, made of Teflon material, enhances precision and
obviates the contamination
that can occur when plastic
pipette tips are used in combination with nitric acid.
Blank measurements
could be performed before each
assay, to ascertain the absence of detectable contamination
originating
from the containers (12). Our serum aluminum
values obtained for healthy volunteers compare well with
the lowest values found in the literature (13), suggesting
that contamination and background interference was effectively absent. Given the stability of the standard solutions
in the receptacles used for collecting serum samples, these
low values cannot be the result of absorption of Al to the
container walls.
The procedures for tissue aluminum
determination
involve less manipulation than do those described by other
authors (14), and these procedures are applicable both for
dry-weight
and wet-weight measurements. In the latter
case, weight determination is more precise and less time
consuming.
The aluminum
concentrations of trabecular and cortical
bone may differ (1). Nevertheless, we have opted to measure
whole transiliac bone biopsies obtained with a standardized
technique, in order to avoid contamination in preparing the
different components.
Precision. The methods we present are simple and precise.
Their precision is documented by the stable signal obtained
with biological samples and the low inter-assay CV. Furthermore, measurements of the same samples in the three
laboratories using slightly different methods yielded comparable results.
Sensitivity and detection limits. The methods are highly
sensitive
and have low detection limits. To some extent, this
is due to the absence of noise at the given instrument
settings. With argon as the purge gas, a higher sensitivity
was obtained than with nitrogen. Like other metals, aluminum may form nitride compounds during the atomization
Amt of ion added to 50 mL
-c
of dIgestion fluId, mg
-
Mg2
Fe3
Salt usedL
-
HPO42
4.0
0.14
3.0
3.0
150.0
MgCI2
FeCI3
NaCI
K2SO4
(NH4)2HP04
Ca2
250.0
CaCO3
Na
Recovery, %
100
98
100
100
99
102
98
The amounts of the ions tested correspond approximately to quantities
physiologically present in 1 g of bone. Al concn: 200 g/L.
b Interference by the counter-ion of the salt evidently is excluded simultaneously.
C In this instance no supplementary
ions were added to the digestion fluid.
28
CLINICAL CHEMISTRY, Vol. 31, No. 1, 1985
(15). These nitrides remain
temperatures,
causing lower absorbances.
Reportedly (16), interruption of the gas flow during atomization increases sensitivity of ETAAS for many metals, including aluminum. A pronounced decrease of the purge-gas
flow during atomization also improved sensitivity considerably. A short ramp-time at the atomization step lowered the
detection limit. However, neither argon flow nor ramp time
should be decreased to zero if rapid wastage of the graphite
tubes is to be avoided. The detection limit for tissueat high
aluminum measurement was 20 pg for a 2O-L injection
with our method, as compared with 150 pg reported with
other methods (14). These detection limits are sufficiently
low for clinical use.
Accuracy. Our methods of analysis for aluminum in
serum and blood offer good accuracy. The availability
of a
rapid, reliable direct method is essential to our laboratory,
which has a large turnover of clinical samples. This motivated us to establish a direct method yielding results equivalent to those by the standard-addition
methods, and adjustment of the instrument settings of the graphite furnace did
result in a close agreement. This contraindicates
any interference by matrix (9,17-19).
The blood aluminum values we
obtained for healthy volunteers are comparable to those
reported by other authors (20), who used inductively
coupled
plasma atomic emission spectrometry.
Although a biological Standard Reference Material with
reported aluminum values is needed, considerable other
evidence presented here suggests that our procedure gives
accurate
results.
when tissu& samples of different weights from the
same patient were analyzed, the values found for all were
similar. The greater variation observed with bone biopsies
than with liver samples may be ascribed to a more homogeneous distribution of aluminum throughout the liver (21).
Secondly, two somewhat
different
digestion
methods
First,
yielded similar results.
Finally,
after correcting for the differences between dry
and wet weight (3), our values for liver and bone aluminum
of chronic hemodialysis
patients are similar to those reported by others, who used other destruction
techniques and
other analytical
methods.
Two matrix phenomena, which may jeopardize the accuracy of aluminum
analysis, deserve careful consideration.
First, losses may occur during the char cycle when volatile
aluminum
compounds are present. Because aluminum
chlo-
ride sublimes at 178 #{176}C,
and because chloride is the predominant anion in serum and urine, low absorbances for these
matrices should be suspect (17). However, the agreement
between the direct and the standard-additions
method and
the invariable atomic absorption signals obtained with
chloride and nitrate aluminum standards indicate there is
no loss of aluminum.
This is achieved by cautiously selecting the temperature scheme to be used in the graphite
furnace.
A second problem
encountered
in aluminum
ETAAS is the possible effect of various ions (14). That such
Table 2. Interference of Ions
Iontested
steps in the presence of nitrogen
stable
interferences are absent
that addition of relatively
no significant effect.
Nearly 100% analytical
ous standards undergoing
tion procedure. Thus, in
observed no losses with
contrary, tissue aluminum
was suggested by demonstrating
large amounts of these ions had
recovery was observed with aquethe steps of the nitric acid digescontrast to other authors (14), we
nitric acid destruction. On the
may be underestimated if EDTA
is used in tissue extraction (22).
We are most grateful to Prof. R. E. Van Grieken for revising the
manuscript and to Ms. A. Grootveld and Mr. E. Snelders for expert
secretarial work. This study was supported inpartby a grant from
the “Dienst voor de Programmatie
(DPWB)” of the Belgian Government,
and by a “Krediet san navorsers”
National Fund for Scientific Research
van het Wetenschapsbeleid
Contract no. 82-87/47 (MDB),
1983, 1984 of the Belgian
(NFWO) 1.5.744.84N(MDB).
References
1. Alfrey AC, LeGendre GR, Kaehny WD. The dialysis encephalopathy syndrome. Possible aluminum intoxication. NEngl J Med 294,
184-188 (1976).
2. Parkinson IS, Ward MK, Kerr DNS. Dialysis encephalopathy,
bone disease and anaemia: The aluminium intoxication syndrome
during regular haemodialysis. J Clin Pathol 34, 1285-1294(1981).
Review.
3. Parsons V, Davies C, Goode C, et al. Aluminium in bone from
patients with renal failure. Br Med J iv, 273-275 (1971).
4. Ellis HA, McCarthy JH, Herrington J. Bone aluminium in
haemodialysed patients and in rats injected with aluminium chloride: Relationship to impaired bone mineralisation. J Clin Pathol
32, 832-844 (1979).
5. Elliott HL, Macdougall Al, Fell GS. Aluminium toxicity syndrome. Lancet i, 1203 (1978).
6. Kaehny WD, Hegg AP, Alfrey AC. Gastrointestinal absorption
of aluminum from aluminum-containing
antacids. N Engi J Med
296, 1389-1390 (1977).
7. Cannata JB, Briggs JD, Junor BJR, et al. Effect of acute
aluminium overload on calcium and parathyroid-hormone metabolism. Lancet i, 501-503 (1983).
8. Van de Vyver FL, Van Waeleghem JP, De Broe ME, et al. Water
treatment and dialysis dementia. Lancet ii, 1106 (1982).
9. BertholfRL, BrownS, RenoeBW, et al. Improved determination
of aluminum in serum by electrothermal atomic absorption spectrophotometry. Clin C/tern 29, 1087-1089 (1983).
10. Lefrancois H, Haguenoer JM, Gamot AP. Determination de
l’aluminium dam les liquides biologiques par spectrophotometrie
d’absorption atomique en four de graphite. Innovation Technolog
Biol Med 3, 459-467 (1982).
11. Versieck J, Cornelis R. Measuring aluminum levels. N Engi J
Med 302, 468 (1980).
12. Kostyniak PJ. An electrothermal atomic absorption method for
aluminum analysis in plasma: Identification of sourcesof contamination in blood sampling procedures. J Anal Toxicol 7, 20-23
(1983).
13. Cornelis R, Schutyser P. Analytical problems related to Aldetermination in body fluids, water and dialysate. Contrib Nephrol
38, 1-11 (1984).
14. LeGendre GR, Alfrey AC. Measuring picogram amounts of
aluminum in biological tissue by flameless atomic absorption
analysis of a chelate. Clin C/tern 22, 53-56 (1976).
15. Sturgeon RE, Chakrabarti CL, Langford CH. Studies on the
mechanism of atom formation in graphite furnace atomic absorption spectrometry. Anal Chem 48, 1792-1807 (1976).
16. McDermott JR, Whitehill I. Determination of aluminium in
biological tissue by flameless atomic absorption spectrometry. Anal
Chim Acta 85, 195-198 (1976).
17. Gorsky JE, Dietz AA. Determination of aluminum in biological
samples by atomic absorption spectrophotometry with a graphite
furnace. Clin Chem 24, 1485-1490(1978).
18. Smeyers-VerbekeJ, Verbeelen D, Massart DL. The determination of aluminum in biological fluids by means of graphite furnace
atomic absorption spectrometry. Clin Chim Acta 108,67-73 (1980).
19. Gardiner PE, Ottaway JM, Fell GS, Halls DJ. Determination
of
aluminium in blood plasma or serum by electrothermal atomic
absorption spectrometry. Anal C/tim Acta 128, 57-66 (1981).
20. Allain P, Mauras Y. Determination of aluminum in blood,
urine, and water by inductively coupled plasma emission spectrometry. Anal C/tern 51, 2089-2091 (1979).
21. De Broe ME, Van de Vyver FL, Bekaert AB. Correlation of
serum aluminum values with tissue aluminum concentration.
Cont rib Nephrol 38, 37-46 (1984).
22. Stevens Ed. Electrothermal atomic absorption determination
of
aluminum in tissues dissolved in tetramethylammonium hydroxide.Clin C/tern 30, 745-747 (1984).
CLINICAL CHEMISTRY, Vol.31,No. 1,1985 29