Alzheimer disease is an insidious and ultimately fatal form of
dementia affecting as many as 1% of the population. The diagnosis can be made
only by exclusion of other causes of encephalopathy or at necropsy, when
characteristic neuropathological changes are found. Its cause is not yet known,
although both genetic and environmental factors are thought to be involved.
Aluminium has been implicated in the pathogenesis of the disorder: animal
studies as early as 1897 showed behavioural changes similar to human dementia
, and many studies have reported increased levels of aluminium in the
affected parts of the brain in Alzheimer disease patients [2,3] even when serum
aluminium levels are within normal limits. Aluminium-induced encephalopathy has
been reported after heavy industrial exposure in subjects with normal renal
function [4,5] and in patients undergoing maintenance haemodialysis for chronic
renal failure, who are exposed to much higher concentrations of aluminium than
normal . Dialysis patients (even if not encephalopathic) show some
similarities in neurochemistry, neuropathology, and cerebral function to
patients with Alzheimer disease [3,7-9].
Epidemiological studies have shown a link between the incidence of
Alzheimer disease and the concentration of aluminium in drinking water
Patients with Down syndrome have similar neurochemical features to
those with Alzheimer disease [11,12] and a predisposition to the disease itself
. In necropsy studies more than 90% of Down syndrome patients aged over 30
years had similar neuropathology to Alzheimer disease patients . These
patients also have higher than normal concentrations of aluminium in their
Although we have shown objective psychomotor impairment in dialysis
patients possibly related to aluminium, the incidence of Alzheimer disease in
these patients is no higher than in the general population -an unexpected
finding if aluminium has a causal effect in the pathogenesis of Alzheimer
disease. In an attempt to resolve this discrepancy we have postulated that the
factor protecting the majority of people from the adverse neurotoxic effects of
aluminium is also protective in dialysis patients (except when they are exposed
to overwhelming amounts).
Experiments on the intestinal absorption of aluminium showed the
importance of its chemical species for bioavailability . The species
distribution may also be important in vivo, affecting the transport of
aluminium to susceptible organs such as the brain. Aluminium binds almost
exclusively to the plasma glycoprotein transferrin  a large transport
protein of approximately 70 kD, which normally transports iron and certain
other metals. The release of metals from transferrin depends on the presence of
transferrin receptors on the cell surface. Once bound to the receptor,
transferrin becomes incorporated into an endosomal compartment within the cell
which has a sufficiently low pH to cause dissociation of the metal from
transferrin. Under normal conditions there is very little non-transferrin-bound
aluminium in the circulation and there are few transferrin receptors in the
brain, so the normal uptake into the brain may be negligible. The transferrin
molecule has spare metal-binding capacity and is found in very large quantities
in the blood. It binds aluminium avidly and because of the excess of
transferrin over aluminium it is difficult under normal circumstances to
saturate the total blood binding capacity. However, any free aluminium in the
circulation - most likely as aluminium citrate - could enter the central
nervous system without hindrance . Thus, any impairment of aluminium
binding to transferrin could give rise to far greater deposition of aluminium
in the brain, thereby increasing the risk of neutotoxic effects.
We have attempted to determine the distribution of aluminium in plasma
from subjects with Alzheimer disease, Down syndrome, stroke dementia, end-stage
renal disease on haemodialysis, and normal subjects. Because there are no
biologically useful radionuclides of aluminium, we used the chemically similar
radioisotope gallium-67 (half-life 78 h) as an analogue for aluminium .
Both aluminium and gallium bind to the same sites on transferrin  and
aluminium does not necessarily compete with iron in binding to transferrin
Patients and methods
Blood (10 ml) was drawn into heparinised containers for plasma
separation from five groups of subjects (table I). The control subjects were
laboratory staff and non-demented, depressed inpatients. In the patients with
Alzheimer disease, the diagnosis was confirmed by standard neuropsychiatric
techniques, as well as (in 5 patients) by the presence of delayed
flash-stimulated visual evoked potentials . All plasma samples were stored
at 4°C for immediate assay of gallium-transferrin binding and subsequent assay
of citrate, iron, ferritin, and transferrin and transferrin subtyping.
To measure the binding of gallium to transferrin, gallium nitrate
(Ga[NO3]3, Aldrich Chemical Company) and 1-5 µCi 67Ga (Amersham
International) were added to 1800 µl plasma to a final concentration of 10
µg/l. This concentration is similar to that of aluminium in the serum of
normal individuals  and those with Alzheimer disease . The plasma was
incubated at 37°C for 1 h in a shaking water bath (60 oscillations/mn), then
applied to a Sephadex G75 column (60 x 1.5 cm; Amicon Wright) at a flow rate of
1 ml/min. The column was eluted upwards with 25 mmol/l "tris"-HCl, 100 mmol/1
sodium chloride buffer at pH 7.4, and 50 fractions of 6 ml were collected.
Recovery of 85-90 % was achieved and after each run any remaining gallium was
eluted by means of 5 ml 0.1 % EDTA solution. Each fraction and standards of a
known volume were counted for 67Ga activity by measuring gamma
emissions over the energy range 50-420 keV (LKB Compugamma Counter;
Pharmacia-LKB). The percentage binding of gallium was calculated from the
recovered radioactivity. No difference in gallium distribution was seen when a
smaller quantity of gallium (0.15 µg/l) was added. We examined the possible
effect of sample storage on gallium binding and found no difference between
samples stored at 4°C or - 20°C, nor any effect of duration of storage over 1
week in either set of samples. Since we were concerned with relative
differences in binding, absolute levels of gallium binding are not necessarily
comparable with other human studies because the assay conditions, such as
incubation times, differ.
Plasma citrate levels (normal range below 140 µmol/l) were measured
by an enzymic method (Boehringer Mannheim)  by means of the unextracted
protocol of Borland et al  modified for the 'Multistat III'
microcentrifugal analyser (Instrumentation Laboratories Ltd), plasma iron
(normal range 500-1680 µg/l) by the guanidine/'FerroZine' method (JT Baker
BV), and plasma ferritin (normal range 4-330 µg/l) by radioimmunoassay (Becton
Plasma transferrin levels were measured by immunonephelometry on the
Technicon Instruments DPA1. The normal range is 1.96-4.58 g/l. Transferrin iron
saturation is the ratio between plasma iron and total iron-binding capacity
(itself derived from the transferrin concentration, 1 mole being able to bind 2
moles of iron), and is normally between 20 and 40 %. Subtyping was carried out
by a modification of an isoelectric focusing procedure . Before
electrophoresis, one volume of plasma was mixed with four volumes of rivanol (3
mg/ml). The mixture was saturated with ferric iron by the addition of iron
nitrilotriacetate. The precipitated proteins were removed by centrifugation. 5
µl transferrin-containing supernatant was used for isoelectric focusing on
polyacrylamide gels containing 3 % (by volume) LKB ampholine pH 4-6 and 2 % LKB
ampholine pH 5-7. The gels (250 mm x 120 mm x 1.5 mm) were run for 3 h at 25 mA
and a constant power of 25 W. The transferrin bands were detected by staining
with coomassie blue. The genotype frequency in the general population is
transferring-(1) 60.8 %, transferrin(2-1) 24.9 %, transferrin-(2) 2.6%,
transferrin-(3-1) 9.4%, transferrin-(3-2) 1.9%, and transferrin-(3) 0.4%.
Group data are given as mean and SEM, with ranges where appropriate.
Differences between grouped data were assessed by the Mann-Whitney U test, and
correlations between variables by linear regression analysis . A two-tailed
significance value of p < 0.05 was taken as significant.
TABLE I. - COMPOSITION OF STUDY GROUPS
|Group||n||Sex (M/F)||Mean (SEM) age
|Control ||22||11/11||27.5 (4.0)
|Alzheimer disease ||13||3/10||75.9
|Down syndrome* ||12||5/7||38.4
|Stroke dementia ||10||9/1||76.6
|*Non-demented, all trisomy 21, **No evidence of
A bimodal distribution of gallium was observed in all test and control
subjects, the two peaks eluting in similar fractions in all experiments. The
first peak, a high-molecular-weight gallium-protein species (fractions measured
for protein by the Lowry assay), corresponded to that of transferrin; the
low-molecular-weight species (less than 4 kD) is as yet unidentified, although
we presume it is citrate since that is the major species of
non-transferrin-bound aluminium in man .
The mean (SEM) percentage binding of gallium to transferrin was
similar in control (17.1 [1.58]%) and haemodialysis subjects (17.6 [2.38]%) and
patients with stroke dementia (19.8 [1.45]%). The Alzheimer disease and Down
syndrome groups showed significantly lower binding (7.92 [1.11]% and 6.92
[0.70]%, respectively) than the control group and there was no significant
difference in binding between these two groups (see accompanying figure). There
was no significant difference in age between the Alzheimer and stroke dementia
groups, and there was no correlation between age and gallium-transferrin
binding, either overall or in any of the groups. All citrate concentrations
were within the normal range, and there were no significant differences among
The transferrin concentration was lower in the Alzheimer group than in
the controls (table II), but the difference did not reach significance
(p<0.07), whereas the difference between the Down syndrome and control
groups was significant. However, the stroke dementia and haemodialysis groups
had much lower transferrin concentrations (below the normal range), despite
having normal gallium binding. Iron and ferritin levels (table II) were
generally higher in patients with Alzheimer disease or Down syndrome than in
control subjects, but because of the wide normal range this difference was not
significant. Patients with stroke dementia had significantly lower iron and
higher ferritin levels than controls, but the ferritin level was not
significantly different from that of me Alzheimer disease group. There were no
correlations between levels of transferrin, citrate, iron, or ferritin and
Transferrin iron saturation (table II) was higher in the Alzheimer
disease group and much higher in the Down syndrome group than in the control
group, whereas the stroke dementia group had normal saturation. The results of
transferrin subtyping showed a slightly different distribution of genotype from
the general population in the combined group of Alzheimer disease and Down
syndrome patients: 50% were transferrin-(1) homozygotes and 50%
transferrin-(2-1) heterozygotes. However, this difference did not reach
significance, possibly owing to the small numbers.
Table II. - Plasma transferrin, iron and ferritin
concentrations and transferrin iron saturation
(µg/l)||Ferritin (µg/l)||Transferrin saturation (%)
(273)||63 (11)||39.0 (7.4)
|Alzheimer disease ||10||2.45
(0.20)||2056 (770)||133 (35)||58.9 (20.2)
|Down syndrome ||14||2.44 (0.13)*||2604
(509)||124 (40)||81.6 (14.2)**
|Alzheimer + Down ||24||2.44
(0.11)*||2358 (437)||127 (27)||71.4 (11.9)*
|Stroke dementia ||10||1.74
(0.14)***||865 (230)*||164 (46)*||33.4 (8.9)
|*p < 0.05; **p < 0.01; ***p < O.OOO1 for
difference from control group. ND = not done.
Percentile plots (percentiles 10, 25, 50, 75, 90)
of gallium-transferrin binding in Alzheimer disease. Down syndrome, stroke
dementia, and control groups.p values refer to differences between
patients and controls.
Our results show that significantly lower amounts of gallium are bound
to plasma transferrin in patients with Alzheimer disease and Down syndrome than
in normal subjects or patients with stroke dementia. Consequently, greater
amounts of gallium are present as a low-molecular-weight complex. De Wolffet al
 observed a similar component in rat plasma and identified it as a citrate
complex. In comparative experiments the two gallium species eluted in similar
positions in both rat and human plasma . We did not find any differences in
plasma concentrations, of citrate. Although transferrin levels were slightly
lower in the Alzheimer disease and Down syndrome groups than in the controls,
this difference should not affect the capacity to bind gallium, both the stroke
dementia and dialysis groups had a greater reduction in transferrin
concentrations (often found in chronically ill people) but had normal gallium
binding. This result is corroborated by the finding that transferrin saturation
by iron was greater in the Alzheimer disease and Down syndrome groups than in
the controls or stroke dementia patients, suggesting separately that there is
an abnormality of transferrin function in both groups. All subjects had normal
iron status: none had low iron stores, as measured by ferritin, or ferritin
levels in the haemochromatosis range (several thousand (µg/l). The transferrin
genotype distribution was slightly different from normal in the Alzheimer
disease and Down syndrome group. The reduced binding of gallium, and by
implication aluminium, to transferrin must be due to a functional defect of
transferrin, to a very high-affinity, low-molecular-weight substance (much less
likely in view of the transferrin iron saturation results), or to a third
factor which changes the equilibrium between these two species. The high
transferrin iron saturation may be secondary to the greater number of available
binding sites owing to reduced occupation by gallium, aluminium, or other
metals, or a primary factor leading to a reduction in the capacity of
transferrin to bind gallium, aluminium, and other metals. In either case the
result is consistent with a substantial change in transferrin function.
Previous studies with gallium and citrate have shown that a
lipid-soluble species is formed between the two, and that there is enhanced
transport across the intestinal membrane , as seen with aluminium . It
seems likely that a low-molecular-weight aluminium citrate species in plasma
would move across the blood-brain barrier. Transferrin-bound aluminium
transport is limited, since it is mediated by brain transferrin receptors 
which are not increased in number in Alzheimer disease compared with
age-matched controls . Similarly, levels of transferrin in both grey and
white matter are not abnormal in Alzheimer disease. Since Alzheimer disease and
Down syndrome patients show similar neuropathology, it is possible that both
have a predisposition to aluminium accumulation. Patients with stroke dementia,
whose dementia is due to repeated episodes of cerebrovascular occlusion and who
have very different neuropathological changes in the brain, had normal binding
characteristics, as would be expected.
The finding that the few haemodialysis patients studied did not have
the gallium-transferrin binding defect, despite slightly low total transferrin
levels, may solve the question of why such patients do not show a high
incidence of Alzheimer disease. Because transferrin has such a large capacity
to bind aluminium normally, these patients remain reasonably well detoxified,
and acute aluminium toxicity develops only when levels of exposure rise above
the saturation point for transferrin or other binding proteins. In addition, we
know little of the interaction between aluminium and silicon, which is present
in the plasma of dialysis patients in much greater amounts than normal . It
is possible that the insoluble aluminium silicate complex which may form in the
presence of non-transferrin-bound aluminium also helps to prevent
neurotoxicity. This possibility is also pertinent to healthy people whose diet
and drinking water contains substantial amounts of silicon .
In conclusion, we have found a substantial difference in the plasma
distribution of gallium, and aluminium by implication, between patients with
Alzheimer disease or Down syndrome and control subjects or those with stroke
dementia. The study indicates a mechanism by which aluminium may accumulate in
the brains of individuals with defective transferrin, thereby leading to
neurotoxic changes seen in Alzheimer disease and Down syndrome. We suggest that
people without this defect would be protected from the neurotoxic effects of
aluminium. The presence of the defect could be used as a diagnostic test for
Alzheimer disease and may prove to be useful in screening for susceptibility to
the disorder, although we are investigating other, simpler, techniques to show
The exact nature of the functional defect of transferrin remains
unclear and we are undertaking studies to elucidate it further. Once the
molecular defect is identified, it should be possible to explore its genetic
basis, which would open the door to future preventive and therapeutic
1. von Döllken. Ueber die Wirkung des Aluminiums mit besonderer
Berücksichtigung der durch das Aluminium verursachten Läsionen im
Centralnervensystem. Archiv Exp Pathol Pharmacol 1897; 40:98-111.
2. Peri DP. Relationship of aluminium to Alzheimers disease. Environ
Health Perspect 1985, 63; 149-53.
3. Crapper McLachlan DR, Lukiw WJ, Kruck TPA. New evidence for an
active role of aluminium in Alzheimer's disease. Can J Neural Sci 1989;
4. Spofforth J. Case of aluminium poisoning. Lancet 1921; i:1301.
5. McLaughlin AIG, Kazantzis G, King E, Tearc D, Porter RJ, Owen R.
Pulmonary fibrosis and encephatepathy associated with the inhalation of
aluminium dust. Br J Industr Med 1962; 19:253-63.
6. Alfrey AC, LeGendre GR, Kaehny WD. The dialysis cnccphalopathy
syndrome. Possible aluminium intoxication. N Engl J Med 1976; 294:184-88.
7. Altmann P, Al-Salihi F, Butter K, et al. Serum aluminium levels and
erythrocyte dihydropteridine reductase activity in patients on hemodialysis N
Engl J Med 1987; 317:80-84.
8. Scholtz CL, Swash M, Gray A, Kogeorgos J, Marsh F. Neurofibrillary
neuronal degeneration in dialysis dementia: a feature of aluminium toxicity.
Clin Neuropathol 1987; 6:93-97.
9. Altmann P, Dhanesha U, Hamon C, Cunningham J, Blair J, Marsh F.
Disturbance of cerebral function by aluminium in haemodialysis patients without
overt signs of aluminium toxicity. Lancet 1989; ii:7-12.
10. Martyn CN, Barker DJP, Osmond C, Harris EC, Edwardson JA, Lacey
RF. Geographical relation between Alzheimer's disease and aluminium in drinking
water. Lancet 1989; i:59-62.
11. Yates CM, Simpson J, Mafoney AF, et al. Alzheimer like cholinergic
deficiency in Down's syndrome. Lancet 1980; ii:979.
12. Blair JA, Morar C, Hamon CGB, et al. Tetrahydrobiopterin
metabolism in the central nervous system. In: Pfleiderer W, Wachter H, Curtius
HC. Biochemical and clinical aspects of pteridines. Vol 3: Proceedings of Third
Winter Workshop on Pteridines, Feb 1984. Berlin: Walter de Gruyter,
13. Wisniewski K, Wisniewski H, Wen G. Occurrence of neuropathological
changes and dementia of Alzheimer's disease in Down's syndrome. Ann Neurol
14. Williams RS. Age related changes in Down's syndrome brain and the
cellular pathology of Alzheimers disease. In: Swaab D, Fleirs E, eds. Progress
in brain research. Amsterdam: Elsevier, 1986: 49-67.
15. Farrar G, Morton AP, Blair JA. The intestinal speciation of
gallium: possible models to describe the bioavailability of aluminium. In:
Bratter P, Schramel P, eds. Trace element analytical chemistry in medicine and
biology, vol 5. Berlin: Walter de Gruyier, 1988:343-47.
16. Trapp GA. Plasma aluminium is bound to transferrin. Life Sci 1983;
17. Farrar G, Blair JA. A proposed model to account for the
accumulation of aluminium in Alzheimers brain. In: De Silva L, ed. Trace
elements in health and disease. New Delhi: Hamdard Nagar Press (in press).
18. Farrar G, Morton AP, Blair JA. The intestinal absorption and
tissue distribution of aluminium, gallium and scandium: a comparative study.
Biochem Soc Trans 1987; 15:1164-65.
19. Tomimatsu K, Donovan JW. Spectroscopic evidence for perturbation
of tryptophan in Al(III) and Ga(III) binding to ovotransferrin and human
transferrin. FEBS Lett 1976; 71;299-302.
20. Martin RB, Savory J, Brown S, Bertholf RL, Wills MR. Transferrin
binding ofAl3+ and Fe3+. Clin Chem 1987; 33:405-07.
21. Wright CE, Harding GFA, Orwin A. Presenile dementia-the use of the
flash and pattern VEP in diagnosis. Electroencephahgr Clin Neurophys 1984;
22. Altmann P, Butter KC, Plowman D, Chaput de Saintonge DM,
Cunningham J, Marsh FP. Residual renal function in hemodialysis patients may
protect against hypcraluminemia. Kidney Int 1987; 32:710-13.
23. Shore D, Wyatt RJ. Aluminium and Alzheimers disease. J Nerv Ment
Dis 1983; 171:553-59.
24. Mollering H, Gruber W. Determination of citrate with citrate
lyase. Anal Biochem 1966; 17:369-76.
25. Borland WW, Ferguson JC, Dryburgh PJ. A fast automated method for
measuring serum and urine citrate. Ann Clin Biochem 1989; 26:286-88.
26. Thymann M. Identification of a new serum protein polymorphism as
transferrin. Human Genet 1978; 43:225-29.
27. Feldman DS, Gagnon J. Stat View 512 + Statistical Software.
California: Brainpower Inc, 1986.
28. Dc Wolff FA, van Ginkel MP, Brandsma A, van der Voet GB.
Association of aluminium with plasma and liver constituents, Clin Chem
29. Slanina P, Frech W, Ekstrom LG, Loof L, Slorach S, Cedergreen A.
Dietary citric acid enhances absorption of aluminium in antacids. Clin Chem
30. Candy JM, Edwardson JA, Faircloth R, et al. 67Ga as a marker for
aluminium transport in rat brain. J Physiol 1987; 391:34P.
31. Morris CM, Corn, JA, Moshtaghie AA, et al. Transferrin and
transferrin receptors in normal and Alzheimers disease. Biochem Soc Trans 1987,
32. Dobbie JW, Smith MJB. Urinary and serum silicon in normal and
uraemic individuals. CIBA Found Symp 1986; 121:194-213.