J. clin. Path. (1963), 16, 12

The haemolytic activity of an iron carbohydrate complex
J. FIELDING with the technical assistance of
GILLIAN M. SMITH
From Paddington General Hospital, London

The haemolytic activity of iron-dextran complex is found to be a function of time, temperature, pH, and concentration. The lytic action is enhanced by small amounts of added ferrous sulphate. The lytic action is inhibited by chelating agents such as citrate and sequestrene salts, which bind ionic iron, but not by ferric citrate or ferric sequestrene which do not bind iron. The ionised iron content of iron-dextran is deduced.

SYNOPSIS

The lytic activity of iron-dextran is also inhibited by iron-dextrin and by an iron-sorbitol-citric acid preparation. It is suggested that the iron-sorbitol-citrate molecular complex contains free chelating groups for iron.
The significance of these findings for iron-carbohydrate toxicity and metabolism is briefly discussed. The clinical toxicity of parenteral iron preparations, both intravenous and intramuscular, has been one of the principal problems associated with their use. The toxic manifcstations are varied in kind and tend to form a pattern of reactions characteristic in type for each iron complex.
It is unlikely that a single factor is responsible for all or even most of the observed toxic reactions. Instability of the complex in plasma with possible precipitation in vivo is a likely cause in the case of the saccharated oxide but improbable with the dextran, dextrin, or sorbitol complexes. The rate of reticulo-endothelial uptake from the circulation, the rate of release of dissociated iron from reticuloendothelial cells, dissociation of the complex in the circulation, the intrinsic ionic iron content of the complex, and anaphylactic phenomena have all been postulated. It is clear that each complex needs to be considered separately in detail.
The firm bond between ferric hydroxide and the carbohydrates, from which iron carbohydrate complexes are synthesized, simulates to some extent the binding of inorganic iron to transferrin. The iron in these complexes is almost entirely non-ionized and in consequence comparatively large amounts are tolerated in the circulation without toxic effect. Con-

centrations of 4 mg. iron per 100 ml. as iron-dextrin or iron-dextran are regularly achieved during treatment; this is 50 times the normal concentration of iron in the circulation. After the administration of an iron carbohydrate complex the degree of saturation of the iron-binding capacity rises. The time at which this occurs also varies with the complex: with saccharated iron-oxide it occurs during the first few hours, with iron-dextrin after three hours, with irondextran after two or three days, but with ironsorbitol it is immediate, returning to normal in 20 hours. It appears that all these complexes contain or yield up at some stage diffusible or ionized iron which combines with transferrin. Klopper (1951) suggested that when transferrin was fully saturated excess ionized iron manifested itself in toxic reactions. This is an attractive hypothesis and one which is widely quoted. There have, however, been no reports of the state of transferrin saturation during the actual occurrence of toxic reactions. On the other hand, it may be shown that after injections of iron sorbitol transferrin may be fully saturated without toxic reaction.
Nissim (1954) in a study of toxicity measured several properties, including haemolytic activity of saccharated iron oxide and some other iron compounds. The haemolytic action was assessed by the

Received for publication 7 July 1962.

12

13

The haemolytic activity of an iron carbohydrate complex comparatively crude technique of direct microscopic observation. Golberg (1958) suggested that haemolytic activity may be a measure of the presence of free ferric ion and therefore an index of toxicity. The dark colour and viscosity of iron carbohydrate complexes make measurement of the haemolytic action by conventional haemoglobin measurements difficult and inaccurate but a red cell counting technique using an electronic blood cell counterenables the haemolytic activity of such solutions to be measured simply and accurately (Fielding, 1961). A report is given here of some experimental observations on the haemolytic activity of iron-dextran and its relationship to ionized iron.

100

V

50

0

0

8 za Ot

4
6
28

METHODS AND MATERIALS

HAEMOLYSIS Red cell lysis is measured by a cell-counting technique using an EEL electronic blood cell count:r. To
4 ml. volumes of fluid under test, is added 0-02 ml. of a three times washed suspension of human red cells in saline. After incubation the unlysed cells are enumerated in four 0-2 ml. volumes and the mean taken: thus for an original suspension of 5 million red cells per c.mm.
20,000 red cells are counted. The difference in count from a saline control suspension is taken as the measure of haemolysis. Differences of less than + 3 % are ignored.
STANDARD IRON-DEXTRAN HAEMOLYSIS When the effects of various factors on iron-dextran haemolysis had been ascertained as described below, conditions were standardized thus: final concentration of iron-dextran, 5 mg. Fe per ml.; washed red cell suspension, 3 5 to 6-0 million cells per c.mm.; incubation, six hours at 37°C.
References in the text to iron carbohydrate concentration refer to the concentration of elemental iron.
Iron-dextran was used as Imferon (Benger) and irondextrin and iron-sorbitol were used as Astrafer and
Jectofer (Astra-Hewlett) respectively.
Sorensen's phosphate buffers were prepared from
M/15 disodium hydrogen phosphate and M/15 sodium dihydrogen phosphate. To 100 ml. of each pH mixture,
8 g. sodium chloride was added and the solution diluted
1 in 10 to bring it to normal osmolarity.
Mcllvaine's citric acid-phosphate buffers were prepared from M/S disodium hydrogen phosphate and M/10 citric acid. To 100 ml. of each pH mixture, 6 g. sodium chloride was added and the solution diluted 1 in 1O. These buffered salines were tested for haemolytic action before use.
RESULTS

IRON-DEXTRAN, IRON-DEXTRIN, AND IRON-SORBITOL

Figure 1 shows the haemolysis produced by varying concentrations of iron-dextran in six hours at
37°C., whereas iron-dextrin and iron-sorbitol produce no haemolysis, as previously reported (Golberg,
1958; Lindvall and Andersson, 1961; Fielding,
1961).

e

*

10

mg.Fe/ml.

FIG. 1. Iron-dextran (A), iron-dextrin (B), and ironisorbitol (B). Abscissa = concentration of iron coinplex as elemental iron. Red cell suspension 4,580 per 0-2 ml.; incubation six hr.137°C.

CONCENTRATION OF IRON-DEXTRAN The rate of irondextran haemolysis increases with concentration
(Fig. 2). At a concentration of 0-5 %, haemolysis is complete in six hours at 37°C. The reciprocal of the time taken for 50 % haemolysis varies approximately with the logarithm of the concentration.

100'

50'
0.

I-

0
2

3

4

5

6

Incubation t hr. )

FIG. 2. Effect of iron-dextran concentrations, 0 060% to
2-0% Fe, on the rate of haemolysis at 37°C.

EFFECT OF TEMPERATURE Figure 3 shows the rate of haemolysis produced by 0-5 % iron-dextran at 37°C., at room temperature, and at 4°C. At all temperatures there is a latent period of about two hours before perceptible haemolysis is observed, after which the rate of haemolysis increases with temperature. At 37°C. haemolysis is complete in six hours;

J. Fielding and Gillian M. Smith

14

200 C.

/

C

L.I

dextran. At pH 8 a concentration of 0 5 % is inactive but at pH 5-4 haemolysis is virtually complete.
Iron-dextrin and iron-sorbitol were also tested in the range pH 5 4 to 8 at concentrations varying from
006 % to 1 0 % Fe, and again no haemolytic activity was observed.

37C.

100

50

EFFECT OF ETHYLENE

DIAMINE TETRA-ACETIC

ACID

I.

SALTS Sodium, potassium, and calcium sequestrene salts chelate iron, that is, they bind iron in nonionized form, giving up their own metallic radicals.
0~ ~ ~~~~~~~~4C
~
If iron-dextran haemolysis is due to its content of
0
ionized iron it should therefore be inhibited in the presence of such chelating agents. Table II shows the of potassium, calof
I1 resultsand varying concentrationsthe haemolytic acti~4 6 C hr. I lb
2
iron sequestrenes on cium, Incubation iron-dextran. Both the potassium and
FIG. 3. Effect of temperature on 0-5% iron-dextran vity of 0-5 % the calcium salts neutralize the lytic effect of the ironRed cell suspension, 6,200 per 0-2 ml. haemolysis. carbohydrate complex. Ferric sequestrene, which cannot bind inorganic iron, has no effect on the at 4°C. little haemolysis takes place in the first 12 haemolytic action. This supports the view that ironhours, although by 24 hours it is complete. This dextran haemolysis is due to its content of ionized temperature effect contrasts with osmotic haemolysis iron. The concentration of sequestrene required to in which the rate of haemolysis does not vary inhibit haemolytic activity is small: M/320 of either potassium or calcium salt completely inhibits haemoperceptibly in this temperature range. lysis, and concentrations down to 2M x 10-4 exert
EFFECT OF pH Iron-dextran was diluted serially in perceptible inhibitory effects. doubling dilutions through isotonic Sorensen's buffers, and tested for haemolytic activity at 37°C. EFFECT OF ORGANIC SALTS ON IRON-DEXTRAN HAEMOLYSIS The chelating action of sequestrene salts is for six hours.
Table I shows that haemolytic activity increases well known. The citrate ion also has a capacity to with increasing acidity at all concentrations of iron- chelate metals and this is, of course, the basis of the
BLE I pH HAEMOLYTIC ACTIVITY OF IRON-DEXTRAN
Concentration of Iron-dextran (Fe g. per 100 ml.)
2-0

5.4

6
7

8

10

05

0225

0512

006

003

90
20
160
70

80
80
580

290
990

1,160
3,000
4,160

2,250
3,990

4,010

3,950

4,190

4,020

3,840

4,180

4,150

3,180

Red cell counts per 0-2 ml. Incubation six hr./37°C.

TABLE II
ACTION OF SEQUESTRENE SALTS ON 0 5 % IRON-DEXTRAN HAEMOLYSIS
Final Dilution of MIJO Sequestrene Salt
No Iron Dextran

0 5 % Fe as Iron-dextran

1/8
K,E.D.T.A.
CaE.D.T.A.
FeNa,E.D.T.A.

1/16

1/32

1/64

5,550
5,640

5,600
5,650

5,580
5,380

3,330
4,890

190

230

210

190

Red cell counts per 0-2 ml. after six hr./37°C. in 05%
Initial count in saline 5,570 per 0-2 ml.

1/128

1/256

11512

1/1,024

1/2,048

1/8

1/2,048

1,300

760
550
90

230
220
100

120
90
120

40
50
40

5,520
5,620

5,490
5,630
5,240

980
180

iron dextran.

5,200

15

The haemolytic activity of an iron carbohydrate complex

anticoagulant action of citrates. Sodium citrate has a similar action on inorganic iron as it has on calcium, and might therefore be expected to have an inhibitory action on iron-dextran haemolysis similar to that of sequestrene salts. The effect of sodium citrate and ferric citrate was therefore investigated in the same way as that of sequestrene salts. Calcium citrate is too insoluble to be tested.
Ferric citrate was without effect. Sodium and potassium citrates were found to have marked inhibitory actions on iron-dextran haemolysis. These salts are, however, strongly alkaline and it was possible therefore that their actions were due to changes in pH. To test this possibility, isotonic
Mcllvaine's citric acid-phosphate buffers at pH 5 4 and pH 6 were serially diluted through Sorensen's phosphate buffers at similar pH levels. The resulting buffered isotonic mixtures gave two series at a constant pH, in which the only significant variable was a serial two-fold dilution of citric acid.
When the haemolytic activity of 05 % iron dextran was tested in these buffers, complete inhibition of lysis resulted in the presence of M/640 citric acid at pH 5 4 and M/2560 at pH 6-0. Thus the citrate effect was not due to pH change alone, but was evidently the result of the ionic iron-binding capacity of the citrate ion.
A number of other organic and inorganic sodium salts were also tested for inhibitory action on irondextran haemolysis. Sodium sulphate, lactate, and salicylate were without effect. Sodium oxalate, fumarate, succinate, fluoride, and cyanide shared varying degrees of protective action. Sodium benzo100

ate and acetate could not be tested since they are haemolytic alone.
The effects of citrate, succinate, and oxalate are shown in Fig. 4, which also includes the action of the sequestrene salts for comparison. The action of sodium citrate is particularly well marked, a concentration of M/1280 giving complete inhibition of
0-5 % iron-dextran haemolysis.
EFFECT OF IRON-DEXTRIN AND IRON-SORBITOL ON

As shown above neither iron-dextrin nor iron-sorbitol have haemolytic activity. Both complexes in the form used here are known to contain citrate and it therefore seemed possible that their citrate content might account for their lack of lytic activity. Furthermore, it seemed possible that these preparations might exhibit chelating activity towards inorganic iron. They were therefore tested in the same way as the sequestrene and citrate salts.
Both complexes showed marked protective effects against iron-dextran haemolysis, the sorbitol complex being more effective than the dextrin (Fig. 5). A dilution of 1 in 80 iron-sorbitol (0-62 mg. Fe per ml.) exerts the same protective effect as M/1,280 sodium citrate. If it is assumed that the whole of the protective effect is due to the citrate content of the complex, the original undiluted iron-sorbitol complex (5-0 %
Fe) would contain M/16 citrate, which corresponds to 12 mg. citric acid per millilitre. It has been shown by continuous flow electrophoresis of iron-sorbitol that the citric acid content moves in two fractions, corresponding to the two main iron fractions in the complex. The total citric acid thus revealed is
IRON-DEXTRAN HAEMOLYSIS

100]

-

A

B

._ c w

u

w
0

50~

50 w ro
-

'a

I2t

I

o-

0-

FIG. 4

2048

.

.

512

.

Final

.

128

Dilutions

*

*

32

I

*

8

FIG. 5

0o0o9

I

9

o-638

0-15

FIG. 4. Inhibition of 0-5% iron-dextran haemolysis, six hr./37°C. Abscissa = final dilution of sodium citrate; B, CaE.D.T.A.; C, K2E.D.T.A.; D, sodium succinate; E, sodium oxalate.
FIG. 5. Inhibition of 05% iron-dextran haemolysis, six hr.137°C. Abscissa sorbitol (A) or iron-dextrin (B).

0-62

25

mg. Fe/ml.

=

MIJO salts of A,

final concentration of added iron-

J. Fielding and Gillian M. Smith

16

based on the inhibitory effects of chelating salts.
Furthermore, ionic iron as ferrous sulphate exhibits only agglutination without haemolysis. It seemed possible that the agglutination by ferrous sulphate might be modified in the presence of the large ironcomplex molecule.
The effect of adding small amounts of ferrous sulphate to iron-dextran was therefore tested. Table
IV shows that ferrous sulphate increases the haemolytic titre of iron-dextran. This haemolysis is also inhibited by sequestrene salts and by citrate.
By measuring the inhibition titre of calcium sequestrene on iron-dextran haemolysis with and without added ferrous sulphate, the haemolytic equivalent of iron-dextran in terms of ferrous sulphate may be calculated. From such an evaluation it appeared that iron-dextran (as Imferon) contains the equivalent of N/200 ferrous sulphate. A similar titration, using sodium citrate as inhibitor, gives the equivalent value of N/1 60 ferrous sulphate. Thus this preparation of iron-dextran, which contains 5 % elemental iron, contains between 0-014 % and 00230o ionized iron, that is, about 1/300th part of the iron present is ionized.

approximately 31 mg. per millilitre. It seems likely, therefore, that part of the citrate content of ironsorbitol (as Jectofer) determines the non-haemolytic character of iron-sorbitol. No similar data are available for iron-dextrin, but since it is also known to contain citrate, similar considerations may apply.
If it is accepted that the ionized iron content of iron-dextran causes haemolysis, a similar direct action of ionized iron in the form of inorganic salts might be expected. Since inorganic ferric salts are strongly acid in solution and cause acid haemolysis, the action of the ferric ion in this form cannot be ascertained. On the other hand, the chelated ferric organic salts, such as ferric citrate and ferric sequestrene, should have no direct haemolytic or agglutinating action on red cells, and this was found to be so.
Freshly prepared solutions of ferrous sulphate made up to isotonic strength with saline have an agglutinating action on red cells, without haemolysis. The agglutination produced by weak concentrations is easily broken up by shaking and the cells may be counted. Ferrous sulphate agglutination is inhibited by the presence of sodium and calcium sequestrene and by sodium citrate, but not by ferric sequestrene or ferric citrate. Table III illustrates the effect of calcium sequestrene. Thus ferrous sulphate agglutination is inhibited by the same agents as inhibit iron-dextran haemolysis.
EFFECT OF IRON SALTS ON RED CELLS

DISCUSSION

The experiments described demonstrate that irondextran (Imferon) has a haemolytic action which is a function of pH, concentration, temperature, and time. This haemolysis is inhibited by potassium and calcium sequestrenes but not by ferric sequestrene.
It is also inhibited by some sodium salts, such as

EFFECT OF FERROUS SULPHATE ON IRON-DEXTRAN

HAEMOLYSIS The evidence that iron-dextran haemolysis is due to ionized iron is thus far indirect and

TABLE III
AGGLUTINATING ACTION OF FeSO4 AND ITS INHIBITION BY

CaE.D.T.A.
3 mnl. Dilutions of M/300 FeSo4

1/8

114

1132

1/16

1/64

1/256

11128

1,1512

Plus I ml. saline
Plus 1 ml. M/160 CaE.D.T.A.
, , degrees of agglutination, observed after adding 0 02 ml. red cell suspension containing 4 x

10'

cells c.mm.

TABLE IV
INTERACTION OF IRON-DEXTRAN HAEMOLYSIS AND FERROUS SULPHATE AGGLUTINATION
2 ntl. Dilutions of I% Iron-dextran

1/2
Plus 2 ml. saline
Plus 2 ml. M/640 FeSO4 in saline

114

118

1/16

1/32

1/64

1/128

11256

60

140

520

2,520

3,230

4,090

4,250

4,350

120

40

30

20

170

190

440

1,070

Red cell counts/0-2 ml. 37°C./six hr.
- no agglutination; 2 degrees of agglutination.

1/512

1/1,024

0

4.300
2,540

4,210

4,280

The haemolytic activity of an iron carbohydrate complex citrate, lactate, and fumarate, which form poorly ionized iron compounds, but not by ferric citrate.
Haemolysis is thus inhibited by iron-binding chelating agents but not by ferric sequestrene or ferric citrate which cannot bind iron. Small amounts of ferrous sulphate added to iron-dextran increase the haemolytic titre and this increased activity is also inhibited by chelating agents. Ferrous sulphate alone has an agglutinating but no haemolytic action on red cells, which is also inhibited by the same reagents.
It appears, therefore, that in the presence of the large molecule of the iron-dextran complex the agglutinating action of ferrous sulphate becomes a haemolytic one. These results support the hypothesis that iron-dextran haemolysis is due to its content of ionized iron. It should be noted that the concentrations required for haemolysis in vitro are not attained in vivo during therapy.
As measured by the inhibitory effect of calcium sequestrene and of sodium citrate, iron-dextran, which contains 5 % w/v elemental iron, contains the equivalent of 0-014 % to 0-023 % ferrous iron. Thus about 1/300th part of the iron content of Imferon is ionized. Assuming therefore that the haemolytic action of iron-dextran is due entirely to its content of ionized iron, 5 ml. Imferon (= 250 mg. Fe) contains at the most 1 mg. ionic iron. This is far short of the amount required to saturate transferrin ironbinding capacity, and the ionized content of this complex cannot therefore account for the toxic effects sometimes observed soon after intravenous injection in man.
Since both iron-dextrin and iron-sorbitol have neither agglutinating nor haemolytic action over a wide range of pH and concentration, it appears that they contain no free ionic iron. On the contrary, they appear to contain an excess of chelating bonds, since they are both capable of neutralizing the haemolytic effect of iron-dextran in the same way as chelating salts. If it is assumed that the anti-haemolytic activity is due solely to citrate, the activity may be expressed as a citric acid concentration, and amounts in the case of iron-sorbitol to 12 mg. per ml.
Analysis of iron-sorbitol (as Jectofer) by continuous flow electrophoresis (Lindvall, 1962) shows that its citric acid content is about 31 mg. per ml., and moves in two fractions corresponding to two iron fractions. The minor fast-moving fraction is dialysable, and its citric acid content is about 13 mg. per ml. In this fraction, iron and citric acid are

17

present in equivalent proportions, and it is therefore unlikely that the citric acid in this fraction is capable of binding further iron. Thus the anti-haemolytic activity may reside in the larger, slow-moving, nondialysable fraction, consisting of iron-sorbitol-citric acid of mean molecular weight about 5000. The citric acid content of this fraction amounts to about
18 mg. per ml.
Laurell (1951), in discussing the physiological transfer of iron from the vascular to the extravascular and cellular compartments, shows that transferrin does not leave the circulation, and postulates that iron enters cells as a relatively small molecular weight complex after dissociation from transferrin. Charley,
Rosenstein, Shore, and Saltman (1960) have shown that chelating agents, such as citrate and E.D.T.A., enhance the uptake of iron from serum transferrin by rat liver slices and homogenates. They suggest that naturally occurring low molecular weight chelators are active in the movement of iron out of the circulation from transferrin to tissue receptors.
It is possible, therefore, that low molecular weight chelated iron, permeable to capillary endothelium, may have important physiological functions. If this were so, it could be anticipated that the utilization of the iron-sorbitol-citric acid complex, which contains its iron in this form and remains capable of binding further ionic iron, would follow different and perhaps more physiological pathways than nondiffusible iron complexes of high molecular weight.
The present results suggest that the view of iron carbohydrate toxicity, which considers only the release of ionic iron and the saturation of transferrin. is probably too simple a picture. The availability of chelating bonds other than transferrin may well modify the status of iron in the circulation, and it becomes possible to visualize transferrin saturation without necessarily incurring the risk of ionic toxicity. REFERENCES

Charley, P., Rosenstein, M., Shore, E., and Saltman, P. (1960).Arch.
Biochem., 88, 222.
Fielding, J. (1961). Brit. med. J., 2, 279.
Golberg, L. (1958). In Iron in Clinical Medicine, edited by R. 0.
Wallerstein and S. R. Mettier, p. 77. Univ. California Press,
Berkeley.
Klopper, A. (1951). Lancet, 1, 531.
Laurell, C. B. (1951). Blood, 6, 183.
Lindvall, S. (1962). Personal communication.
, and Andersson, N. S. E. (1961). Brit. J. Pharmacol., 17, 358.
Nissim, J. A. (1954). Ibid., 9, 103.