CLIKICA CHIMICA ACTA 349

ENVIRONMENTAL VARIATION OF ARSENIC LEVELS IN HUMAN BLOOD

DETERMINED BY NEUTRON ACTIVATION ANALYSIS

K. HEYUORN
Isotope Division, Research Establishment RisB, Roskilde (Denmark)
(Received October qth, rgGg)

Arsenic levels in blood plasma and red cells from patients with Blackfoot
disease, a peripheral arteriosclerosis endemic to a small area in Taiwan, were studied
in relation to healthy individuals from the same and other parts of Taiwan and com-
pared with arsenic levels in a control group from Denmark.
Arsenic was determined by neutron activation analysis with radiochemical
separation and re-irradiation yield determina.tion. The precision and accuracy of the
results have been carefully evaluated in order to permit quantitative tests for the
significance of the observed differences.
The results from Taiwan followed a logarithmic normal distribution, and no
difference was found between Blackfoot patients and their healthy family members.
However, their overall arsenic levels were higher than the Taiwan average, presumably
because of arsenic in their drinking water.
Much lower levels were found in Denmark, which geochen~ically belongs to a
soil zone with less arsenic than Taiwan.

INTRODUCTION

While the toxicity of trivalent arsenic compounds has been known for millen-
nia, the possible effects of the minute amounts of natural arsenic present in all living
organisms are largely unknown, and until recently reliable determinations of the
natural levels of arsenic in humans were not available.
Arsenic is generally believed to be a nonessential trace element and does not
appear to accumulate in the body. Levels in human blood would be expected to reflect
the intake of arsenic from food and water, and differences should be observed between
regions with different dietary habits.
In 1967 a small expedition from the Copenhagen University Hospital went to
Taiwan to study an endemic occurrence of peripheral arteriosclerosis, known as
Blackfoot disease, and a large number of blood samples were brought to Denmark for
investigation.

CEin.&him. Acta, 28 (1970) x49-357

350 HEYIWRIi

The composition of the food in Taiwan differs considerably from that in Den-
mark and most other European countries, and, in particular in the endemic area,
water from deep artesian wells was reported to have unusually high levels of arsenic.
Therefore, a number of blood samples from diseased as tvell as healthy individuals
from Taiwan were subjected to neutron actilration analysis for the determination of
arsenic and compared with normal samples of blood from Denma.rk.

Blood samples were taken by venepuncture, and about g ml of blood were

drawn through a stainless-steel needle into a centrifuge tube containing 0.5 ml of
3.83; sodium citrate solution to prevent clotting.
Plasma and red cells were separated by centrifuging, and haemolysis was
brought about by freezing. Scnrm samples were obtained in the same way, omitting
the citrate solution to permit coagulation.
Samples of different preparations of sodium citrate solution used in the collec-
tion of blood samples were included for the determination of blank values.
,411 samples were kept at -zoo until the time of analysis.

Arsenic was determined by neutron activation analysis as arsenic-76 wit11 a

half-life of 26.5 h, in a procedure using multiple carrier addition and re-irradiation
yield determinationr.
Samples of plasma, serum or red cells were irradiated in o.5-dram sealed polo.-
vials for 30 min in the pneumatic tube system of a Danish reactor DR 2 at a thermal
neutron flux of 7.10'~ qz/cm2.sec along with an arsenic standard of 0.5 /“g/ml. Irradia-
tion times exceeding 45 min led to the coagulation of plasma samples with subsequent
handling difficulties.
After rh--20 IIdecay 1.00 ml, or a weighed quantity, was transferred to a beaker
containing a dry carrier mixture of 20 mg of copper and I mg each of arsenic and
antimony oxides. This copper: arsenic ratio corresponds to the mean ratio observed in
the majority of samples analysed.
Sample decomposition with sulphuric and nitric acid was followed by a cup-
ferron scavenging, and the arsenic was precipitated with thioacetamide, redissolved
in ammonium sulphide and transferred to a counting vial.
Samples and standard were counted for 40 min, 24 h after the end of irradiation
in a 3 inch x 3 inch NaI (Tl) scintillation detector connected to a Nuclear Data 16o I?
multichannel analyser at a gain of 6.7 keV/channel.
The samples were re-irradiated for 30 set along with a I mg arsenic standard,
and the chemical yield was determined by counting for a few minutes under exactly
the same conditions as above. The chemical yields were about Soqd,.
The results were calculated as ng/ml, the specific gravity of red cells being
determined separately where weighed samples had to be analysed.

Clzn.Chim. Acta, 28 (1970) 349-357

ENVIRONMESTAL VARIATION OF ARSENIC LEVELS 351

ANALYTICAL RESULTS

The arsenic concentrations in three different citrate solutions used in this

study were determined as 1.2, 8.5 and 106 q/ml, the last one fortunately being used
only in a few samples from the largest group. All plasma results are corrected for their
proper citrate blank as well as for the dilution of the sample.
Whole-blood values were calculated on the basis of 572 ml plasma and 428 ml
red cells in IOOO ml of blood.
In the statistical treatment of the results, a single determination was assigned
a standard deviation of 2 ng absolute and 5% relative; in addition, an estimated stan-
dard variation of 20% in the citrate concentration was included in the standard error
of the corrected results so that an influence from high blank values does not jeopardize
the inferences made.

Sawafiles from Dewmnrk

Normal healthy individuals, selected by the Copenhagen University Hospital,
were used as a control group, and results of arsenic determinations in plasma and red
cells are summarized in Table I.

TABLE 1

ARSENIC IN BIJJOD FROM NORMAL DANISH SUBJECTS

Standard WYOY Number of Standard deviation

of mean* samples of distribution

Plasma 2.4 4 0.6 I6 1.9

Red cells 2.7 f 0.9 7 1.3
Whole blood 2.5 i 0.5
* Calculated from precision

The distribution of results is entirely (P > 0.05) accounted for by the precision
of the determinations, and no sample individualization is possible.
The whole-blood average was calculated from the means and their associated
standard errors.

Samples from Taiwan

The endemic Blackfoot area is about 200 km2 and is situated on the west coast
of the island about 200 km south of Taipei. Samples of blood were taken from patients
with Blackfoot disease and from family members without Blackfoot symptoms; in
addition, samples were collected from normal healthy persons in the endemic area.

TABLE II

ARSENIC IN BLOOD PLASMA FROM NORMAL TAIWANESE SUBJECTS

Mean value Number of Standard deviation factor

(nglmtl samples for log distribution

Taipei 1g.0 6 1.37

Endemic area 15.7 II 1.42
Taiwan ‘5.4 17 1.39

Clin. Chim. Acta, 28 (1970) 349-357

352 HEYDORW

Samples from normal healthy persons in Taipei were used as a reference, and
in Table II results of arsenic determinations in plasma from this group are presented
along with those of the corresponding group in the endemic area.
PI:0 significant difference is found between samples from Taipei and those from
the endemic area (P > o.os), and the results can be pooled to represent normal Tai-
wanese subjects. The contribution from precision to sample variance is comparatively
small, and the pooled results are well approximated by a log-normal distribution, as
illustrated in Fig. I.

Y.
90

1
10 15 20 25 ng/ml 10 20 50 100 nglml

Fig. I. Cumulative distribution of arsenic concentrations in blood plasma from normal Taiwanese
subjects.
Fig. 2. Cumulative distribution of arsenic concentrations in blood plasma from Blackfoot families.

Logarithmic means and standard deviation factors are given for plasma in
Table II and for red cells and whole bloodin Table III along with a whole-blood average
calculated from the means of plasma and red cells.

TABLE III
ARSENK IN BLOOD FROM NORMAL TAIWANESE SUBJECTS
__-___I_____-~ __....._
Mean value Number of Standard devzation factov
(v/ml1 samples for log distribution
_..._ __---
Plasma ‘5.4 17 I.39
Red c&s 32.7 5 1.x*

Whole blood 21.6 5 I .07

Average 22.3
._ _~. --.-. __~. _..-

Results of arsenic determinations in plasma from Blackfoot patients and family

members are presented in Table IV. No significant difference is found between the
groups (P > o.og), and the results can be pooled to represent Blackfoot families.

TABLE IV
ARSENICIN BLOOD PLASMA~140~4BLACKFOOT FAMILIES
~ _-.-

Mean value Number of Standard deviation factor

C%;ml) sam$&s forlogd~sty~but~an.
Rlackfoot patients 32-3 33 7-75
Family members 45.2 14 2.01

Blackfoot families 38.1 47 I .90

--. .- ~~.. --..--___ ..-.-.... -

CEiz. Chim. Acta, 28 (1970) 349--_-357

ENVIRONhlEiXTAL VARIATION OF ARSENIC LEVELS 353

The precision is insignificant compared with sample variance, and the two groups
as well as their combination are well represented by log-normal distributions. Loga-
rithmic means and standard deviation factors are given in Table IV, and the distri-
bution of results from the combined groups is plotted in Fig. z.

Results of arsenic determinations in red cells are given in Table V, and again no
significant difference is found between the groups (P >> 0.05). The pooled results are
well represented by a log-normal distribution, and the logarithmic means and standard
deviation factors are given in the table.

Nwnbev of Standavd deviafinn factor

samplrs .fov log dzstnbution

Blackfoot patients 5 1.69

I’amily members 6 1.80

Blackfoot families II I.70

Rveragc
-- ___~

Corresponding results for arsenic in whole blood are given in Table VI for the
two groups as well as their combination, along with a whole-blood average for Black-
foot families calculated from the means of plasma and red cells.
No variation with age and sex could be observed.

EVALUATION OF RESULTS

Reported values for arsenic levels in normal human blood are in disagreement
by almost 3 orders of magnitude, even within recent years. Bowen has reviewed the
literature until 1962, and in his monograph3 from 1966 a whole-blood value of 490
ng/ml is given.
An evaluation of the possible errors associated with the present results is thus
highly desirable, and an attempt has been made to quantify the overall precision and
accuracy of measurement.

Precision
The analytical precision was determined by multiple analyses of the same
samples. A total of 39 determinations were carried out on 18 different samples, cover-
ing the range 22200 ng/ml and including plasma, serum and red cells.
354 HEYDORS

By calculation of the variance of the differences between determinations and

corresponding mean values, as well as the variance of the ratios of these differences to
the mean values, the standard deviation of a single determination was estimated at
&z ng absolute and 506 relative.
The contribution from counting statistics was insignificant with an nveragt
of < 0.5 ng.

The sampling precision can be estimated by comparing results for plasma and
serum from the same person which were available in a few cases. In Table VII plasma
results corrected for citrate addition are shown together with corresponding results
for serum. The variance of their differences is in very good agreement with that ex-
pected from their analytical precision, and variations in the sampling procedure can
be neglected.

Accuracy
The analytical accuracy has been investigated previously’, and interference
from other elements in the blood is completely negligible in comparison with the
overall precision.
The results for the control group shown in Table I are in complete agreement
ENVIROSMESTAL VARIATION OF ARSENIC LEVELS 355

(P > 0.05) with results reported by Brune et aL4 for normal human whole blood and
by Giovannetti et ah5 for plasma.
In the absence of certified biological standard material, dried kale powder
prepared by Bowen and analysed by several laboratories was included in this study.
An arsenic concentration of 118 +_ 4 rig/g of dry material was found as the mean of
six determinations, and this result is consistent (P > 0.05) with the results given by
Bowen’.
The overall accuracy of the results within the concentration range of interest
may thus be taken as consistent with the overall precision, and in all statistical cal-
culations each determination has been weighted according to a standard deviation of
z ng absolute and 5”,/, relative.

Distrihtion
The distribution of results in each group was assumed to be either gaussian or
logarithmic normal, the choice being made by means of a chi-square test.
The adequacy of the chosen log-normal distribution for the results from Taiwan
is supported by the close agreement between the whole-blood mean values, calculated
from corresponding red cells and plasma determinations, in Tables III and VI, and
those calculated from the means of the total number of determinations in the two
groups.

The levels of arsenic found in blood plasma as well as in red cells from normal
subjects are much higher in Taiwan than in Denmark.
The most important difference between the diets in Taiwan and in Denmark
is the low consumption of animal food products in Taiwan, products which contribute
less than 100/o to the daily calorie intakes. The arsenic content in a large number of foods
has been measured by Schroeder and Balassas, and on this basis it can be estimated
that the daily intake of arsenic in Taiwan should not exceed twice the Danish average.
The observed differences in blood arsenic levels thus cannot be attributed solely
to dietary differences but must reflect a higher overall level of arsenic in food and
water from Taiwan as compared with Denmark which may be ascribed to geochemical
differences.
According to Vinogradovlo, Denmark belongs to the northern latitudinal podsol
zone, in which a disproportionately small content of arsenic is found in the soil, about
I ppm or less. On the other hand, the lateritic soil of Taiwan is of comparatively re-
cent volcanic origin, related to soils from contemporary volcanic areas of similar
geological history, such as Japan with average arsenic concentrations of about 20
PPm.
Arsenic in whole blood from Japanese subjects was determined by Iwataki and
Horiuchi”, who found a logarithmic mean concentration of 66 rig/g by a polarographic
method. Dietary differences between Japan and Taiwan are small; but the higher
arsenic concentrations in Japan are consistent with geochemical considerations.
The logarithmic-normal distribution of the results from Taiwan corresponds
with that of the results from Japan, and the difference between the reported value of
2.09 for the standard deviation factor for arsenic concentrations in whole blood” and

Clin. Chim. ,4cta, 28 (1970) 349-357

356 HEYl)OKN

the present value of 1.70 for Taiwan may not be significant (P = 0.05). Lenillan and
SmithI noted that for essential elements the standard deviation factor was almost
always less than 1.6, while nonessential elements generally showed a standard devia-
tion factor exceeding z. They found standard deviation factors between 2.3 and 4 for
arsenic in normal tissue but gave no results for blood. Obviously arsenic is not essen-
tial at the levels observed in Taiwan and Japan, but it is not inconceivable that the
variations could be smaller in regions where individual differences in food habits
have less opportunity to develop than in Scotland.
In addition, a highly significant (P < 0.001) difference is observed betbveen
plasma levels in Blackfoot families and in normal subjects from Taiwan, while the
corresponding difference for red cells is probably significant (0.01 < 1’ < 0.05).
h’o difference was found between Blackfoot patients and other members of
their family, and the higher arsenic levels in Blackfoot families must be caused by
differences in arsenic intake and not by metabolic anomalies connected with the Black-
foot disease.
The diet in the endemic area has been studied by Yang and HlackwellH, who
found a very low fat intake compared with the Taiwan average and a high intake of
carbohydrates, mainly from sweet potatoes. In addition, it has been observed’s that
families with Blackfoot-afflicted members generally get their water from deep artesian
wells, although shallow wells and tap water arc now preferred. Water from deep wells
in the endemic area averaged 0.8 ppm of arsenic”, while water from otlicr sources
contained < 0.15 ppm. \I?th an estimated daily water intake between I and 3 1 (ref. t;),
the additional arsenic intake from deep wells may well be responsible for the higher
levels observed in blood from Hlackfoot families.
The chemical form of arsenic in the water as well as in blood is not known, but
no adverse effect on the organism would normally be expected, even if the water
contained all its arsenic in the trivalent form 15. Arsenic in human blood is probably
bound to proteins in plasma and red cells, but no pronounced preference for erythro-
cytes is found, such as is observed in rats Ifi. Corresponding results for arsenic in plasma
and red cells are plotted against each other in Fig. 3 together with lines correspond-
ing to a simple mean ratio of I .7 and a logarithmic mean ratio of 2.0.
Summing up, it appears that the observed differences between blood arsenic
levels in Denmark and in Taiwan are predominantly caused by the environment and
onlv to a limited extent by differences in diet.

ACKNOWLEI)GEMlXNTS

I am indebted to Dr. Paul Astrup of the Copenhagen l’niversity Hospital for

initiating this study and making his samples available for analysis.
Skilful assistance was given by Mrs. Birgit Lund and Mr. I. Funck Hansen in
the separation and counting of samples for analysis, and by Miss Randi Beier in the
statistical calculations.

I K. HEYDORS, Nuclear .4ctiuation Trchniques in thr LzfeSciences,

Intern.
‘It. Irnergy .\gc%cy,
Vienna, 1967. p. 179.
L H. J. RI. BOWEN, AERE-R 4190 (1963).

Clin. Chim. .4cta, 28 (1970) 349-35 7

GNVIRONMENTAL VARIATION OF ARSENIC LEVELS 357

3 1% J. M. ROWEN, Trace Elements in ~~~~ke?~~st~y, Academic Press, London, 1966.

_t U. RRUNE, K. SAMSAHL AND P. C?.WESTER, Clin. Ckim. Acta, 13 (1966) 2%.
; S. GIOVANNETTI, Q. MACGIORE AND R. MALVANO , Nuclear Act&t& ‘Techraiques iw the Life
Sc.‘ences, Intern. At. Knergy Plgency, Vienna, 1967, p. grr.
o H. J, M. Rowm~, in I'.U'. SHALLIS, Proc. S.A.C. Cwzf,, Nottingham, 1964, W. Heffer and Sons,
Cambridge, 1965, p. 25.
7 H. J. M. ROWER, Analyst, 92 (rgG7) 124.
S TSU-HSING YAKG AXD R. Q. BLACKWELL, Fovmosan Sci., rg (1961) 101.
9 H. A. SCHROEDER AND J. J. RALASSA, .I_Chronic Diseases, rg (1966) 85.
IO .%. P. ~INOGRADOV, ~euc~z~~n~st~~iifNave n& Disf~seu Chemical Elements i?t Soils, Consultants
Bureau, New York, 1959.
II N. IW'ATAKI AND I<. ~~~&UCHI, Osaka City Iv&d.,]., 5 (1959) 209.
IL J, bl. .4. LENJHAN AKIJ H. SMITH, IVuclear Activatio+z Techniqzrrs in tlw Life .%encrs. Intern.
At. Energy Agency, Vienna, 1967, p. 601.
.,

Clin. Ckim. Acta, 28 (1970) 349-357