T R A N S F U S I O N A L I R O N OVERLOAD: T H E RELATIONSHIP BETWEEN TISSUE I R O N CONCENTRATION A N D H E P A T I C FIBROSIS I N THALASSAEMIA R. ANTHONY REDON*,MICHAEL BARRY?AND DAVIDM. FLY"$ * Departments of Morbid Anatomy and $ Medicine, The Hospital for Sick Children, Great Ormond Street, London WClN 3JH and the Hospital, London WCIX 8LF
p Department of Medicine, Royal Free
PLATE XXVI HEPATICFIBROSIS is a common finding in all forms of massive iron excess and is generally thought to result directly from heavy deposition of the metal in parenchymal cells. The severity and activity of the process varies considerably from case to case but the pattern of the fibrosis is essentially similar in the various iron-loading disorders (Kent and Popper, 1960). Recently, a similar type of lesion has been produced by chronic massive parenteral iron overload in dogs (Lisboa, 1971), an observation of particular interest as all previous attempts to reproduce the disorder experimentally have met with failure. Although the concept of iron as a relatively low-grade fibrogenic agent is now generally accepted, quantitativeinformation relating the duration and magnitude of iron-loading (which are probably the two most important factors) to the severity of tissue damage does not exist in either human or experimental disease. For various reasons regularly transfused patients with thalassaemia major provide a convenient model for studying these aspects in man: the patients can be identified at an early stage in the iron-loading process and studied prospectively; until recently (Barry et al., 1974) there has been no means of proven value for preventing the accumulation of iron which is regarded as the major factor for their eventual clinical deteriorationand death; the pathological changes found at necropsy closely resemble those of idiopathic haemochromatosis (Witzleben and Wyatt, 1961 ; Howell and Wyatt, 1963); and, finally, the rate of progression of the disease brings it within a time-scale practicable for continuous clinicopathological study. We have investigated the relationship between tissue iron concentration and hepatic fibrosis, assessed morphometrically, in 52 liver specimens obtained from 19 patients with homozygous beta-thalassaemia between 1961 and 1974. Most of the specimens were obtained during the course of a clinical trial of long-term iron-chelation therapy conducted between 1966 and 1973. The children were all maintained on a high-transfusion regime and, as regards the rate of iron accumulation, formed two homogenous groups according to whether or not they received chelation therapy. Received 29 June 1974; accepted 27 July 1974. 1. PATH.-VOL.
116 (1975)
83
84
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLYNN
It has been suggested that splenectomy modifies the distribution of iron within the liver of patients with thalassaemia, increasing the fraction deposited in parenchymal cells and thereby accelerating the process of liver injury (Witzleben and Wyatt, 1961; Berry and Marshall, 1967). We have examined the distribution of stainable iron in the liver in relation both to the splenectomy status of the patients and the severity of the fibrosis. MATERIALS AND METHODS This study is based on a series of 52 specimens of liver obtained from 19 patients with homozygous beta-thalassaemia between 1961 and 1974. Between 1966 and 1973 the children were the subject of a prospective trial of long-term iron-chelation therapy and all but four of the specimens were obtained during this period. The details of the trial and a general description of the results have been given elsewhere (Barry et al., 1974). Ten of the children received intramuscular desferrioxamine on 6 days each week while the remaining nine received no such treatment and acted as controls. The two groups were matched as closely as possible for sex, age, transfusional requirements, and splenectomy status at the start of the trial. All children were maintained on a high transfusion regime so as to keep the haemoglobin concentration between 9 and 15 g/lOO ml. Splenectomy had been performed in nine patients in infancy (at a mean age of 23 mth, range 6-35 mth) and no patient underwent splenectomy during the course of the trial. For the purpose of the present study the liver specimens were divided into two groups. The first group consisted of 32 specimens obtained from children who had never received chelation therapy (23 from the controls and nine from the chelator-treated children immediately before starting desferrioxamine); these specimens are subsequently designated as having been obtained from non-chelator-treated children. The 20 specimens forming the second group came from desferrioxamine-treated patients after periods of treatment ranging from 1 to 6.2 yr and are subsequently designated as having been obtained from chelatortreated patients. Forty-eight of the specimens were percutaneous needle biopsies, three were wedge biopsies, and one was a necropsy specimen. Two children were biopsied on one occasion only, three twice (including a necropsy specimen in one), 12 on three occasions, and two on four occasions. Histological assessment The liver specimens were fixed in buffered form01 saline (PH 7-0) for 24-48 hr before processing and embedding in paraffin wax. Sections 5 p m in thickness were cut from each specimen and stained with Ehrlich’s haematoxylin and eosin, van Gieson’s mixture, Gordon and Sweet’s reticulin stain and Perls’ stain for iron. Histological assessment was performed without knowledge of the origin of the specimen, the clinical details, or the results of chemical analysis. Assessment of hepatic fibrosis. Some increase in hepatic fibrosis was noted in almost all the specimens, ranging from a slight excess of fibrous tissue in the portal tracts to an established nodular cirrhosis. Sections stained for reticulin were examined with a Wild MS stereomicroscope equipped with a drawing tube attachment, using a x25 objective. The outline of each biopsy and the areas of condensed connective tissue (“ fibrosis ”) that it contained were drawn on to paper (fig. 1). Each drawingwas cut out with scissors and weighed to the nearest mg. The areas representing “ fibrosis ” were then cut out, weighed separately, and expressed as a percentage of the total weight. This value was termed the “ fibrosis index ”. Duplicate estimations of the “ fibrosis index ” were made in five specimens; the standard error for a single specimen corresponded to a coefficient of variation of 1.8 per cent. The assessment of fibrosis in needle biopsy specimens is notoriously liable to sampling errors, depending on the amount of fibrosis, the size of the specimen, and the type of needle employed. The tendency for cirrhotic livers to yield fragmented specimens constitutes a particular source of error. At worst, such specimens consist of a few separate nodules of
PLATE XXVI
RISDON,BARRYAND FLYNN
IRONOVERLOAD
FIG. 1.-Camera lucida drawing of a section of a needle liver biopsy. Areas representing condensed connective tissue (“ fibrosis ”) are stippled. Gordon and Sweet’s reticulin stain. x 24.
IRON 0 VERLOAD
85
cored-out parenchyma, each surrounded by a thin and often incomplete rim of fibrous tissue inevitably giving a spuriously low impression of the amount of fibrosis present. However, this limitation did not apply to the present series of biopsies which were notable for their generally large size (sections ranging in length from 10 to 40 mm, mean 15-9 mm) and a lack of fragmentation. To check the effect of specimen size on the assessment of fibrosis, pieces of tissue corresponding in size to needle and wedge-biopsies were cut at random from a formalin-fixed liver obtained at necropsy from a heavily transfused thalassaemic patient with cirrhosis. The table shows that the fibrosis indices showed greater variation in the smaller needle-sized specimens than in the larger wedges. As a further check under more realistic conditions, six TABLE Comparison of ‘‘ fibrosis indices ” in wedge-sized and needle-sized specimens cut from formalin-fixed liver obtained from a heavily transfused thalassaemic with cirrhosis
Wedge-sized specimens
I
34.6 21.8 24.5 23.4 26.1
23-5 25.0 24.6 26.9 28.9
Mean 26.3
1
SD
1
1.90
Needle-sized specimens
26.1 4.49
I
Coefft. 7.25 of var.
17-18
percutaneous Menghini needle-biopsy specimens were obtained immediately after death from a patient with idiopathic haemochromatosis from whom an operative wedge-biopsy had been obtained a few weeks previously. The fibrosis indices in the six needle-biopsies ranged from 31.4 to 40.5 per cent. (mean 34.6 per cent.); the fibrosis index in the wedge-biopsy was 36-6 per cent. Assessment of the distribution of stainable iron. In every biopsy examined, stainable iron was present in both parenchymal and reticuloendothelial (RE) cells. Parenchymal siderosis ranged from a moderate excess of iron situated mainly at the periphery of the lobules, to a uniformly heavy deposition throughout the lobules. RE siderosis ranged from occasional deposits in slightly enlarged Kupffer cells to massive accumulations in aggregates of distended Kupffer cells, together with deposits in macrophages in the portal tracts. It proved difficult to grade the severity of iron deposition by direct microscopy, and the results obtained were not reproducible. For this reason representative areas of parenchyma (including peripheral and centrilobular zones), Kupffer cells, and portal tracts, were photographed separately under standard conditions in both black-and-white and in colour. The monochrome photomicrographs and the colour transparencies were coded separately and ranked in order of the severity of the parenchymal siderosis, the mean of the two values thus obtained being taken as the definitive rank for the specimen. The specimens were ranked for RE siderosis in the same manner. The reproducibility of the method was tested by comparing the monochrome rank values with the colour rank values using Spearman’s
86
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLYNN
coefficient of rank correlation (R): the correlation for parenchymal siderosis was 0-93, and for RE siderosis 0-91. Liver iron concentration. Liver iron concentration was determined by the method of Barry and Sherlock (1971). Part of each specimen taken since 1971 was preserved fresh for chemical analysis; earlier specimens had been embedded in paraffin wax and were separated from the block by soaking in xylene before analysis. The precision and reproducibility of these methods have been reported elsewhere (Barry, 1974b).
RESULTS Distribution of stainable liver iron The distribution of stainable iron in the liver was studied in 32 specimens obtained from patients who had never received chelation therapy, including nine Parenchymal siderosis
Reticuloendothelial slderosis
:i -
0
R ~0.73
3 2
Y
u
O
-
16-
a # -
s
O
.
0
-
e
-
0
t-
0
0 0 0
0
12
o
0
0
e e 0 O
- e I
I
I
I
I
e
0
0
8-
I-
0
0
e 0
--
-
0 Oe
..
0
8-
-
e
0
0
16-
o
12-
O
20 -
21
2
-
0
R=0*73
e
I
I
I
I
Liver iron concentration (% dry weight) FIG.2.-Relationships
between reticuloendothelial and parenchymal siderosis (ranked according to intensity) and liver iron concentration in heavily transfused patients with thalassaemia. R = Spearman's coefficient of rank correlation. e Biopsies from patients with the spleen in situ. 0 Biopsies from splenectomised patients.
specimens obtained from chelator-treated patients immediately prior to the introduction of desferrioxamine. Reticuloendothelial siderosis (ranked according to intensity) showed a significant correlation with liver iron concentration, with the number of units of blood transfused, and with the age of the patient respectively (figs. 2-4). Although assessed in an identical manner, parenchymal siderosis showed a less good correlation with these variables ; inspection of the scatter diagrams (figs. 2-4) shows that this variation was partly due to relatively heavier parenchymal iron deposition in splenectomised patients than in subjects with the spleen in situ. Parenchymal siderosis ranking bore a skewed relationship to total liver iron concentration suggesting that parenchymal iron deposition was relatively heavier when liver concentration exceeded 3 per cent. dry weight (fig. 2).
87
IRON OVERLOAD Reticuloendothelial siderosis
";
Parenchymal siderosis
28 -
'b
a
'2
0
0 0
2.4: 20 -
1612-0. -
0 0
16
0
0
b
Y 0
0
0
-
0
2.4
o
32-
0
b
0
8-
R = 0.85 L a
-
I
0
0
O
a a
o
0 0
b
O
b
a
8
R = 0.64
*
I
l
l
I
I
I
1
Number of units transfused FIG.3.-Relationships between reticuloendothelial and parenchymal siderosis and the number of units of blood transfused in heavily transfused patients with thalassaemia. Symbols as in fig. 2.
Reticuloendothelial siderosis
Parenchymal siderosis 0
a b
a
a b
0
-
28
0
b
32-
0
--
b
0
0 0
Ob
24-
-
0
0
O
20-
b 0
0 0
16-
12-
0 0
b b
b
0, b
a-
o o
O
b 0
R = 0.81 I
20
I
I
I-
-
)
0 0 Ib
R =0-69 I
I
I
I
1
a0 120 160 200 240 Age (mth)
FIG.4.-Relationships between reticuloendothelial and parenchymal siderosis and age in heavily transfused patients with thalassaemia. Symbols as in fig. 2.
Relationships between age, liver iron concentration, and hepatic jibrosis Non-chelator treatedpatients. The 32 biopsies in this group were the same as in the previous section. As previously described (Barry et al., 1974) liver iron concentration in the non-chelator treated patients rose progressively during the period of study to attain final values of 4-5 per cent. dry weight; the rate of
88
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLYNN
increase, relative to the amount of blood transfused, declined as the patients became more heavily iron-loaded. Fibrosis was progressive and in six cases had Chelator-treated patients
Non-chelator treated patients
"1
50
r = 0.785
*em
30t
c
* .
A
.
a.
I
am
*=
IOC
c
mi.
I
I
I
I
80
40
120 160 200 240 Age (rnth)
FIG.5.-Relationships between the hepatic fibrosis index and age in heavily transfused patients Biopsies from patients who had not received chelating agents (including with thalassaemia. those obtained prior to the introduction of chelation therapy). @, Biopsies from patients receiving chelation therapy.
Non-chelator treated patients
Chelator-treated patients
50 r = 0.781
/
r = 0-616 30
L10 /
e/
/ e
*, 1
;
m e
,/ :
me
2
3
e
l
1
; 5
Liver iron concentration (% dry weight)
FIG.6.-Relationships between the hepatic fibrosis index and liver iron concentration in heavily transfused patients with thalassaemia. Symbols as in fig. 5. The trend lines were calculated by linear regression analysis with the values for fibrosis index in logarithmic form.
advanced to a stage of severe fibrosis or cirrhosis (corresponding to fibrosis indices>25 per cent.) by the end of the period of study. The severity of the fibrosis was related to age in a normal linear fashion (fig. 5). There was a less
IRON OVERLOAD
89
good correlation between fibrosis and liver iron concentration (r=0.687) but inspection of the scatter diagram suggested that these variables were not associated in a normal linear manner and a much better correlation was obtained by Non-chelator treated patients h U
c
.-M
E
C
0 .E
Chelator-treated patients
5-
r = 0.632
4-
1:
3-
Y
8
r
2-
9 1-
I.
L aJ
>
I
1
I
I
I
l l l t l l
I
I
80 120 160 200 240
40
80
40
160 200
120
Age (rnth)
FIG.7.-Relationships between liver iron concentration and age in heavily transfused patients with thalassaemia. Symbols as in fig. 5 . The trend lines were calculated by linear regression analysis with the values for age in the logarithmic form. Reticuloendothelial siderosis
32-
28 -
Y
5 L
Parenchymal siderosis
0 0
24 -
0
0
0 0
20-
0
0
0
0
0 0
z -
0
00
aJ
-
z1 2-0
0 0
oo
0
O
R = 085
4 -0,"
-
O
I
I
0
1
I
I
0
0 0
0 0
-
0
VI
G 16-
28 24 20 16 12832-
0
0
0
0
0
0 0
0 0
O O 0 0
0
*O 0 0
4 -0"
-
R = 0.58
0 O
I
I
I
1
f
FIG.8.-Relationships between reticuloendothelial and parenchymal siderosis and hepatic fibrosis index in heavily transfused patients with thalassaemia. Symbols as in fig. 2.
fitting a log normal curve to the data (fig. 6). Although the severity of the fibrosis relative to liver iron concentration increased sharply as the latter exceeded 3.5 per cent. dry weight this was not due to an accelerated rate of iron
90
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLY”
deposition in older patients; the results showed that in fact the reverse occurred, liver iron concentration tending to level off in the older patients at a maximum value of approximately 5 per cent. dry weight (fig. 7). The severity of the fibrosis showed a good correlation with the RE siderosis ranking but not with the parenchymal siderosjs rank (fig. 8). Figure 8 also shows that there was no tendency for fibrosis to be more severe in splenectomised patients than in those with the spleen in situ. Chelator-treatedpatients. There was no significant change in the degree of fibrosis in serial biopsies obtained during the 6-year period of treatment (Barry et aE., 1974). Depending on the initial level, liver iron concentration rose in five patients, fell in two, and remained unchanged in two. The final values tended to level off at a maximum value of approximately 3 per cent. dry weightsignificantly lower, relative to the amount of blood transfused, than in the nonchelator treated patients; the change in liver iron concentration with age followed a similar pattern (fig. ’7). The relationship between the severity of the fibrosis and liver concentration corresponded, within the limited range of the variables, to that observed in the non-chelator treated patients (fig. 6).
DISCUSSION Iron overload in man occurs under four main circumstances: (1) in idiopathic haemochromatosis; (2) in chronic refractory anaemias marked by erythroid hyperplasia and (usually) ineffective erythropoiesis; (3) in chronic oral iron overload, as exemplified by Bantu siderosis; and (4) in transfusional iron overload. In the first three situations iron accumulation results from increased iron absorption,and severe fibrosis or cirrhosis are commonlyfound. In transfusional iron overload, on the other hand, fibrosis is often slight or absent despite equally massive iron deposition. To account for this discrepancy it has been suggested that parenterally administered haemoglobin-iron is less toxic than iron absorbed from the gut, the essential difference being that iron released from effete red cells is mostly held innocuously in reticuloendothelial (RE) cells whereas iron derived from the gut is mainly deposited in parenchymal cells with potentially harmful effect. This distinction is not absolute, however, as the relative severity of parenchymal and RE siderosis in transfusional siderosis is known to vary considerably from case to case, probably reflecting the fact that RE iron tends to be redistributed to the parenchymal cells in time. Current understanding of these aspects has been reviewed by Bothwell and Finch (1962). The role of other factors in the pathogenesis of liver injury have yet to be satisfactorily defined but both the magnitude and duration of iron-loading, and possibly the rate of plasma iron-turnover, are likely to be of particular importance and form the subject of the present discussion. In thalassaemia massive iron excess occasionally develops in subjects who have had few or no transfusions (Whipple and Bradford, 1936; Ellis, Schulman and Smith, 1954; Bannerman et al., 1967; Barry, 1974a) and under these circumstances can be attributed entirely to increased alimentary absorption. In the majority of cases, however, the relative importance of increased absorption and repeated transfusion for iron accumulation is much more difficult to
IRON OVERLOAD
91
ascertain and is further confused by the wide variation in transfusion schedules employed. Witzleben and Wyatt (1961) noted that cirrhosis was a constant but late feature in heavily transfused patients with thalassaemia. They assumed that transfused haemoglobin-iron was in general innocuously deposited in RE cells and attributed the cirrhosis in their patients to the attainment of a critical level of parenchymal cell iron, which they supposed was mainly derived from persistently high alimentary absorption. Although conforming with accepted concepts, the validity of these conclusions may be questioned on two counts: firstly, because it is not possible to determine by direct means what fraction of the iron load, if any, is gut-derived; and secondly, because after the initial stages of iron-loading it is impossible to discern histologically how iron derived from different sources is apportioned between the parenchymal and RE cells. Although increased iron absorption is well documented in thalassaemia and other refractory anaemias associated with marrow hyperplasia, there is evidence of various kinds suggesting that it probably does not contribute significantly to iron-loading in heavily transfused patients : 1. Direct measurements of iron absorption in thalassaemia have shown that the values in transfused children differ but little from those found in controls (Erlandson et al., 1961; Bannerman et al., 1964). This is in accord with generally held concepts that the rate of iron absorption closely reflects erythropoietic activity and that suppression of erythropoiesis by frequent transfusions indirectly checks iron absorption. 2. If continued iron absorption contributes significantly to iron-loading in heavily transfused patients the amount of iron found in the cadavers of such subjects should exceed the amount received as blood in life. However, review of published data reveals that with but few exceptions this applies only in cases that have received relatively few or no transfusions; in the majority of heavily transfused subjects (including thalassaemics) cadaver iron has actually fallen short of the amount transfused in life and, as pointed out by Oliver (1959), the observed discrepancies tend to increase with the number of transfusions (fig. 9). This paradox can be explained by a combination of two factors: firstly, that frequent transfusions indirectly check iron absorption by suppressing erythropoiesis; and secondly, that heavy spontaneous iron loss ensues when iron excess is massive. The alternativepossibility that excessive absorption continues despite transfusions but is more than offset by heavy spontaneous loss seems less likely and conflicts with the observations on iron absorption in thalassaemia. 3. The final piece of evidence stems from the measurements of liver iron concentration made in the present patients. Liver iron concentration is highly correlated with quantitative measurements of total body storage iron in a variety of iron-loading disorders including splenectomised and nonsplenectomised patients with thalassaemia (Barry, 1974b). In the non-chelated children included in this study liver iron concentration was in accordance with the values expected for the amount of blood received until about 100 units had been given ; thereafter the observed values were progressively less than expected for the transfused load, providing further indirect evidence of spontaneous iron excretion in more heavily transfused subjects.
92
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLYNN
From consideration of these facts it is hard to avoid concluding that the iron excess in our patients was largely, if not entirely, derived from the large amounts of blood they had received since infancy and that increased iron absorption was not a significant factor for their iron accumulation. Parenchymal and RE siderosis were assessed separately by ranking the intensity of iron deposition in each compartment on photomicrographsprepared under standard conditions. Iron deposition was at all stages both parenchymal 200
150
h
E
e00
v
c
100 0 I
a
b P W
50
I 200 Transfused iron (grams)
FIG. , 9.-Relationships between cadaver iron and the amount of iron transfused in life in heavily transfused patients. The proportion of iron recovered in the cadaver progressively decreases the more heavily transfused the patient. Based on two personal cases and published data (key references: Howell and Wyatt, 1953; Cappell, Hut.chison and Jowett, 1957; Oliver, 1959; Witzleben and Wyatt, 1961).
and RE in distribution, the degree in each compartment showing significant correlations with the age of the patient, with liver iron concentration, and with the amount of blood transfused. Thus it appeared that, in general, iron was distributed at random in both parenchymal and RE cells without preferential channelling into one or other compartment. Superimposed on this general pattern, however, there were two factors clearly predisposing towards relatively high parenchymal siderosis rankings, namely liver iron concentrations in excess of 3 per cent. dry weight and splenectomy. The effect of splenectomy on iron distribution, and the potential dangers thereof, have been previously noted
IRON OVERLOAD
93
(Witzleben and Wyatt, 1961; Berry and Marshall, 1967); however, our results did not suggest that splenectomised patients went on to develop higher total liver iron concentrations than non-splenectomised patients or that they were more likely to develop severe fibrosis. Some degree of fibrosis was present in all but three of the 52 biopsies examined, ranging in severity from a slight excess of fibrous tissue in the portal tracts to an established nodular cirrhosis. The more severe changes resembled those previously described in multiply transfused thalassaemic patients (Witzleben and Wyatt, 1961) and in idiopathic haemochromatosis (Sheldon, 1935) and may therefore be regarded as more advanced than those generally associated with transfusional siderosis. The rate of progression of the fibrosis differed significantly between the nonchelator and chelator-treated patients ; as reported elsewhere (Barry et al., 1974), this was associated with a progressive and significant difference in liver iron concentration. In the non-chelator treated patients fibrosis was progressive, its severity correlating closely both with age and with liver iron concentration. The rate of fibrosis was constant whereas the rate of increase in liver iron Concentration was not constant but occurred more rapidly in the younger patients than in the older and more heavily transfused ones. From consideration of these interrelationships it is apparent that the rate of fibrosis varied with liver iron concentration, proceeding slowly when liver iron concentration was less than 3 per cent. dry weight but accelerating at higher concentrations. Although it is tempting to relate this to the relatively greater parenchymal siderosis observed at high liver concentrations the results failed to bear this out, the severity of the fibrosis in fact showing a better correlation with RE siderosis ranking than with parenchymal ranking. In the chelator-treated patients liver iron concentration levelled off at a maximum value of about 3 per cent. dry weight. Although significantly lower than the final values in the controls this level is well within the range found in idiopathic haemochromatosis (Sheldon, 1935; Barry, 1973) and it was therefore surprising that the severity of the fibrosis showed no appreciable change during the 6-yr period of study; the retardation in fibrosis seemed disproportionate to the reduction in tissue iron level that treatment had achieved. When the results are compared with those for the non-chelator treated patients, however, the explanation becomes clear, namely that chelation therapy held liver iron concentration below the level at which rapid fibrosis occurs. In conclusion, our findings differ in two important respects from those traditionally associated with the pathology of transfusional siderosis-firstly, in the balanced distribution of iron in both parenchymal and R E cells from an early stage of iron-loading, and secondly, in the progressive course of the fibrosis. The distribution of iron may partly depend on the rate of plasma iron turnover, a high turnover conceivably accelerating the redistribution of iron from RE to parenchymal cells. However, we have no data on plasma iron turnover in our patients and the importance of this factor must remain speculative. The progressive and eventually severe fibrosis associated with transfusional iron overload in thalassaemia may largely reflect one special feature of the disease, namely the
94
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLY"
combination of a high rate of iron accumulation with prolonged survival. The expected life-span of approximately 20 yr in regularly transfused thalassaemics (Engle, 1964) contrasts with the mean survival time of 5.5 yr in 17 patients regarded as representative of pure transfusional iron overload in adults (see Table 3 in Bothwell and Finch, 1962). Since, as we have shown, the rate of fibrosis depends on liver iron concentration, it follows that the severity of the fibrosis is a function of both liver iron concentration and time. However, our findings also show that even when iron accumulation is rapid severe fibrosis cannot be expected to develop in less than 10 yr. Thus, we would suggest that the relatively mild fibrosis generally found in patients with transfusional siderosis may reflect the limited survival of these subjects rather than any other factor. We accept the frequent association between cirrhosis and long-continued hyperabsorption of iron in certain hyperplastic erythroid states (including some cases of thalassaemia) but regard this as a self-selecting situation in that such cases have had disease mild enough to permit prolonged survival with little or no need for transfusion. SUMMARY
The interrelationships between liver iron concentration, the duration of iron-loading, and hepatic fibrosis, assessed morphometrically, have been studied in 32 specimens of liver obtained from 19 heavily transfused patients with thalassaemia major whose age ranged from 4 to 19 yr. Similar observations were made in a matched group of thalassaemic patients treated with long-term chelation therapy. The degree of liver damage ranged from very slight increase in fibrous tissue to severe fibrosis and cirrhosis. The severity of the fibrosis was closely correlated both with liver iron concentration and with age. While the relationship between fibrosis and age was linear, both the severity and the rate of fibrosis were exponentially related to liver iron concentration, damage accelerating as liver iron concentration exceeded 3 per cent. dry weight. By producing a modest but significant reduction in liver iron concentration chelation therapy resulted in a disproportionate but predictable retardation in the progression of the fibrosis. The factors affecting the distribution of iron between parenchymal and reticuloendothelial cells were also examined. In general stainable iron was uniformly distributed between parenchymal and reticuloendothelial cells from the early stages of iron-loading. Parenchymal siderosis was relatively heavier in splenectomised patients and in patients with liver iron concentrations above 3 per cent. dry weight than in non-splenectomised patients or patients with liver iron levels of less than 3 per cent. dry weight, but this did not affect the severity of the fibrosis. The relevance of these findings to the traditional concepts of the pathology of transfusional siderosis is discussed. REFERENCES BANNERMAN, R. M., CALLENDER, S. T., HARDISTY,R. M., absorption in thalassaemia. Brit. J. Haemat., 10,490.
AND
SMITH,R. S. 1964. Iron
IRON 0 VERLOAD
95
BANNERMAN, R. M., KEUSCH, G., KREIMER-BIRNEIAUM, M., VANCE, V. K., AND VAUGHAN, S. 1967. Thalassaemia intermedia, with iron overload, cardiac failure, hypopituitarism, and porphyrinuria. Amer. J. Med., 42,476. BARRY,M. 1973. Iron and chronic liver disease. J. Roy. Coll. Phycns. Lond., 8, 52. BARRY, M. 1974a. Unpublished observations. BARRY,M. 19746. Liver iron concentration, stainable iron, and total body storage iron. Gut, 15, 41 1. BARRY, M., AND SHERLOCK, S. 1971. Measurement of liver-kon concentration in needlebiopsy specimens. Lancet, i, 100. BARRY,M., FLY“, D. M., LETSKY,E. A., AND RKSDON, R. A. 1974. Long-term chelation therapy in thalassaemia major: effect on liver iron concentration, liver histology, and clinical progress. Brit. Med. J. ii, 16. BERRY, C. L., AND MARSHALL, W. C. 1967. Iron distribution in the liver of patients with thalassaemia major. Lancet, i, 1031. BOTHWELL, T. H., AND FINCH, C. A. 1962. Pathologic and clinical aspects of iron overload. In Iron metabolism, Churchill, London, chap. 9. CAPPELL, D. F., HUTCHISON,H. E., AND J o m , M. 1957. Transfusional siderosis: the effects of excessive iron deposits on the tissues. J. Path. Bact., 74, 245. ELLIS,J. T., SCHLJLMAN, I., AND SMITH,C. H. 1954. Generalised siderosis with fibrosis of liver and pancreas in Cooley’s (mediterranean) anemia. Amer. J. Path., 30, 287. ENGLE,M. A. 1964. Cardiac involvement in Cooley’s anaemia. Problems in Cooley’s anaemia. Ann. N. Y. Acad. Sci., 19, 694. ERLANDSON, M. E., WALDEN, B., STERN,G., HILGARTNER, M. W., WEHMAN, J., AND SMITH, C. H. 1962. Studies on congenital hemolytic syndromes. IV. Gastrointestinal absorption of iron. Blood, 19, 359. HOWELL,J., AND WYATT,J. P. 1953. Development of pigmentary cirrhosis in Cooley’s anemia. Arch. Path., 55, 423. KENT,G., AND POOPER,H. 1960. Secondary hemochromatosis: its association with anemia. Arch. Path., 70, 623. LISBOA,P. E. 1971. Experimental hepatic cirrhosis in dogs caused by chronic massive iron overload. Gut, 12, 363. OLIVER, R. A. M. 1959. Siderosis following transfusions of blood. J. Path. Bact., 77, 171. SHELDON, J. H. 1935. Haemochromatosis. Oxford University Press. WHIPPLE, G. H., AND BRADFORD, W. L. 1936. Mediterranean disease-thalassaemia (erythroblastic anemia of Cooley). J. Pediat., 9, 279. WITZLEBEN, C. L., AND WYATT,J. P. 1961. The effect of long-survival on the pathology of thalassaemia major. J. Path. Bact., 82, 1.
p Department of Medicine, Royal Free
PLATE XXVI HEPATICFIBROSIS is a common finding in all forms of massive iron excess and is generally thought to result directly from heavy deposition of the metal in parenchymal cells. The severity and activity of the process varies considerably from case to case but the pattern of the fibrosis is essentially similar in the various iron-loading disorders (Kent and Popper, 1960). Recently, a similar type of lesion has been produced by chronic massive parenteral iron overload in dogs (Lisboa, 1971), an observation of particular interest as all previous attempts to reproduce the disorder experimentally have met with failure. Although the concept of iron as a relatively low-grade fibrogenic agent is now generally accepted, quantitativeinformation relating the duration and magnitude of iron-loading (which are probably the two most important factors) to the severity of tissue damage does not exist in either human or experimental disease. For various reasons regularly transfused patients with thalassaemia major provide a convenient model for studying these aspects in man: the patients can be identified at an early stage in the iron-loading process and studied prospectively; until recently (Barry et al., 1974) there has been no means of proven value for preventing the accumulation of iron which is regarded as the major factor for their eventual clinical deteriorationand death; the pathological changes found at necropsy closely resemble those of idiopathic haemochromatosis (Witzleben and Wyatt, 1961 ; Howell and Wyatt, 1963); and, finally, the rate of progression of the disease brings it within a time-scale practicable for continuous clinicopathological study. We have investigated the relationship between tissue iron concentration and hepatic fibrosis, assessed morphometrically, in 52 liver specimens obtained from 19 patients with homozygous beta-thalassaemia between 1961 and 1974. Most of the specimens were obtained during the course of a clinical trial of long-term iron-chelation therapy conducted between 1966 and 1973. The children were all maintained on a high-transfusion regime and, as regards the rate of iron accumulation, formed two homogenous groups according to whether or not they received chelation therapy. Received 29 June 1974; accepted 27 July 1974. 1. PATH.-VOL.
116 (1975)
83
84
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLYNN
It has been suggested that splenectomy modifies the distribution of iron within the liver of patients with thalassaemia, increasing the fraction deposited in parenchymal cells and thereby accelerating the process of liver injury (Witzleben and Wyatt, 1961; Berry and Marshall, 1967). We have examined the distribution of stainable iron in the liver in relation both to the splenectomy status of the patients and the severity of the fibrosis. MATERIALS AND METHODS This study is based on a series of 52 specimens of liver obtained from 19 patients with homozygous beta-thalassaemia between 1961 and 1974. Between 1966 and 1973 the children were the subject of a prospective trial of long-term iron-chelation therapy and all but four of the specimens were obtained during this period. The details of the trial and a general description of the results have been given elsewhere (Barry et al., 1974). Ten of the children received intramuscular desferrioxamine on 6 days each week while the remaining nine received no such treatment and acted as controls. The two groups were matched as closely as possible for sex, age, transfusional requirements, and splenectomy status at the start of the trial. All children were maintained on a high transfusion regime so as to keep the haemoglobin concentration between 9 and 15 g/lOO ml. Splenectomy had been performed in nine patients in infancy (at a mean age of 23 mth, range 6-35 mth) and no patient underwent splenectomy during the course of the trial. For the purpose of the present study the liver specimens were divided into two groups. The first group consisted of 32 specimens obtained from children who had never received chelation therapy (23 from the controls and nine from the chelator-treated children immediately before starting desferrioxamine); these specimens are subsequently designated as having been obtained from non-chelator-treated children. The 20 specimens forming the second group came from desferrioxamine-treated patients after periods of treatment ranging from 1 to 6.2 yr and are subsequently designated as having been obtained from chelatortreated patients. Forty-eight of the specimens were percutaneous needle biopsies, three were wedge biopsies, and one was a necropsy specimen. Two children were biopsied on one occasion only, three twice (including a necropsy specimen in one), 12 on three occasions, and two on four occasions. Histological assessment The liver specimens were fixed in buffered form01 saline (PH 7-0) for 24-48 hr before processing and embedding in paraffin wax. Sections 5 p m in thickness were cut from each specimen and stained with Ehrlich’s haematoxylin and eosin, van Gieson’s mixture, Gordon and Sweet’s reticulin stain and Perls’ stain for iron. Histological assessment was performed without knowledge of the origin of the specimen, the clinical details, or the results of chemical analysis. Assessment of hepatic fibrosis. Some increase in hepatic fibrosis was noted in almost all the specimens, ranging from a slight excess of fibrous tissue in the portal tracts to an established nodular cirrhosis. Sections stained for reticulin were examined with a Wild MS stereomicroscope equipped with a drawing tube attachment, using a x25 objective. The outline of each biopsy and the areas of condensed connective tissue (“ fibrosis ”) that it contained were drawn on to paper (fig. 1). Each drawingwas cut out with scissors and weighed to the nearest mg. The areas representing “ fibrosis ” were then cut out, weighed separately, and expressed as a percentage of the total weight. This value was termed the “ fibrosis index ”. Duplicate estimations of the “ fibrosis index ” were made in five specimens; the standard error for a single specimen corresponded to a coefficient of variation of 1.8 per cent. The assessment of fibrosis in needle biopsy specimens is notoriously liable to sampling errors, depending on the amount of fibrosis, the size of the specimen, and the type of needle employed. The tendency for cirrhotic livers to yield fragmented specimens constitutes a particular source of error. At worst, such specimens consist of a few separate nodules of
PLATE XXVI
RISDON,BARRYAND FLYNN
IRONOVERLOAD
FIG. 1.-Camera lucida drawing of a section of a needle liver biopsy. Areas representing condensed connective tissue (“ fibrosis ”) are stippled. Gordon and Sweet’s reticulin stain. x 24.
IRON 0 VERLOAD
85
cored-out parenchyma, each surrounded by a thin and often incomplete rim of fibrous tissue inevitably giving a spuriously low impression of the amount of fibrosis present. However, this limitation did not apply to the present series of biopsies which were notable for their generally large size (sections ranging in length from 10 to 40 mm, mean 15-9 mm) and a lack of fragmentation. To check the effect of specimen size on the assessment of fibrosis, pieces of tissue corresponding in size to needle and wedge-biopsies were cut at random from a formalin-fixed liver obtained at necropsy from a heavily transfused thalassaemic patient with cirrhosis. The table shows that the fibrosis indices showed greater variation in the smaller needle-sized specimens than in the larger wedges. As a further check under more realistic conditions, six TABLE Comparison of ‘‘ fibrosis indices ” in wedge-sized and needle-sized specimens cut from formalin-fixed liver obtained from a heavily transfused thalassaemic with cirrhosis
Wedge-sized specimens
I
34.6 21.8 24.5 23.4 26.1
23-5 25.0 24.6 26.9 28.9
Mean 26.3
1
SD
1
1.90
Needle-sized specimens
26.1 4.49
I
Coefft. 7.25 of var.
17-18
percutaneous Menghini needle-biopsy specimens were obtained immediately after death from a patient with idiopathic haemochromatosis from whom an operative wedge-biopsy had been obtained a few weeks previously. The fibrosis indices in the six needle-biopsies ranged from 31.4 to 40.5 per cent. (mean 34.6 per cent.); the fibrosis index in the wedge-biopsy was 36-6 per cent. Assessment of the distribution of stainable iron. In every biopsy examined, stainable iron was present in both parenchymal and reticuloendothelial (RE) cells. Parenchymal siderosis ranged from a moderate excess of iron situated mainly at the periphery of the lobules, to a uniformly heavy deposition throughout the lobules. RE siderosis ranged from occasional deposits in slightly enlarged Kupffer cells to massive accumulations in aggregates of distended Kupffer cells, together with deposits in macrophages in the portal tracts. It proved difficult to grade the severity of iron deposition by direct microscopy, and the results obtained were not reproducible. For this reason representative areas of parenchyma (including peripheral and centrilobular zones), Kupffer cells, and portal tracts, were photographed separately under standard conditions in both black-and-white and in colour. The monochrome photomicrographs and the colour transparencies were coded separately and ranked in order of the severity of the parenchymal siderosis, the mean of the two values thus obtained being taken as the definitive rank for the specimen. The specimens were ranked for RE siderosis in the same manner. The reproducibility of the method was tested by comparing the monochrome rank values with the colour rank values using Spearman’s
86
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLYNN
coefficient of rank correlation (R): the correlation for parenchymal siderosis was 0-93, and for RE siderosis 0-91. Liver iron concentration. Liver iron concentration was determined by the method of Barry and Sherlock (1971). Part of each specimen taken since 1971 was preserved fresh for chemical analysis; earlier specimens had been embedded in paraffin wax and were separated from the block by soaking in xylene before analysis. The precision and reproducibility of these methods have been reported elsewhere (Barry, 1974b).
RESULTS Distribution of stainable liver iron The distribution of stainable iron in the liver was studied in 32 specimens obtained from patients who had never received chelation therapy, including nine Parenchymal siderosis
Reticuloendothelial slderosis
:i -
0
R ~0.73
3 2
Y
u
O
-
16-
a # -
s
O
.
0
-
e
-
0
t-
0
0 0 0
0
12
o
0
0
e e 0 O
- e I
I
I
I
I
e
0
0
8-
I-
0
0
e 0
--
-
0 Oe
..
0
8-
-
e
0
0
16-
o
12-
O
20 -
21
2
-
0
R=0*73
e
I
I
I
I
Liver iron concentration (% dry weight) FIG.2.-Relationships
between reticuloendothelial and parenchymal siderosis (ranked according to intensity) and liver iron concentration in heavily transfused patients with thalassaemia. R = Spearman's coefficient of rank correlation. e Biopsies from patients with the spleen in situ. 0 Biopsies from splenectomised patients.
specimens obtained from chelator-treated patients immediately prior to the introduction of desferrioxamine. Reticuloendothelial siderosis (ranked according to intensity) showed a significant correlation with liver iron concentration, with the number of units of blood transfused, and with the age of the patient respectively (figs. 2-4). Although assessed in an identical manner, parenchymal siderosis showed a less good correlation with these variables ; inspection of the scatter diagrams (figs. 2-4) shows that this variation was partly due to relatively heavier parenchymal iron deposition in splenectomised patients than in subjects with the spleen in situ. Parenchymal siderosis ranking bore a skewed relationship to total liver iron concentration suggesting that parenchymal iron deposition was relatively heavier when liver concentration exceeded 3 per cent. dry weight (fig. 2).
87
IRON OVERLOAD Reticuloendothelial siderosis
";
Parenchymal siderosis
28 -
'b
a
'2
0
0 0
2.4: 20 -
1612-0. -
0 0
16
0
0
b
Y 0
0
0
-
0
2.4
o
32-
0
b
0
8-
R = 0.85 L a
-
I
0
0
O
a a
o
0 0
b
O
b
a
8
R = 0.64
*
I
l
l
I
I
I
1
Number of units transfused FIG.3.-Relationships between reticuloendothelial and parenchymal siderosis and the number of units of blood transfused in heavily transfused patients with thalassaemia. Symbols as in fig. 2.
Reticuloendothelial siderosis
Parenchymal siderosis 0
a b
a
a b
0
-
28
0
b
32-
0
--
b
0
0 0
Ob
24-
-
0
0
O
20-
b 0
0 0
16-
12-
0 0
b b
b
0, b
a-
o o
O
b 0
R = 0.81 I
20
I
I
I-
-
)
0 0 Ib
R =0-69 I
I
I
I
1
a0 120 160 200 240 Age (mth)
FIG.4.-Relationships between reticuloendothelial and parenchymal siderosis and age in heavily transfused patients with thalassaemia. Symbols as in fig. 2.
Relationships between age, liver iron concentration, and hepatic jibrosis Non-chelator treatedpatients. The 32 biopsies in this group were the same as in the previous section. As previously described (Barry et al., 1974) liver iron concentration in the non-chelator treated patients rose progressively during the period of study to attain final values of 4-5 per cent. dry weight; the rate of
88
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLYNN
increase, relative to the amount of blood transfused, declined as the patients became more heavily iron-loaded. Fibrosis was progressive and in six cases had Chelator-treated patients
Non-chelator treated patients
"1
50
r = 0.785
*em
30t
c
* .
A
.
a.
I
am
*=
IOC
c
mi.
I
I
I
I
80
40
120 160 200 240 Age (rnth)
FIG.5.-Relationships between the hepatic fibrosis index and age in heavily transfused patients Biopsies from patients who had not received chelating agents (including with thalassaemia. those obtained prior to the introduction of chelation therapy). @, Biopsies from patients receiving chelation therapy.
Non-chelator treated patients
Chelator-treated patients
50 r = 0.781
/
r = 0-616 30
L10 /
e/
/ e
*, 1
;
m e
,/ :
me
2
3
e
l
1
; 5
Liver iron concentration (% dry weight)
FIG.6.-Relationships between the hepatic fibrosis index and liver iron concentration in heavily transfused patients with thalassaemia. Symbols as in fig. 5. The trend lines were calculated by linear regression analysis with the values for fibrosis index in logarithmic form.
advanced to a stage of severe fibrosis or cirrhosis (corresponding to fibrosis indices>25 per cent.) by the end of the period of study. The severity of the fibrosis was related to age in a normal linear fashion (fig. 5). There was a less
IRON OVERLOAD
89
good correlation between fibrosis and liver iron concentration (r=0.687) but inspection of the scatter diagram suggested that these variables were not associated in a normal linear manner and a much better correlation was obtained by Non-chelator treated patients h U
c
.-M
E
C
0 .E
Chelator-treated patients
5-
r = 0.632
4-
1:
3-
Y
8
r
2-
9 1-
I.
L aJ
>
I
1
I
I
I
l l l t l l
I
I
80 120 160 200 240
40
80
40
160 200
120
Age (rnth)
FIG.7.-Relationships between liver iron concentration and age in heavily transfused patients with thalassaemia. Symbols as in fig. 5 . The trend lines were calculated by linear regression analysis with the values for age in the logarithmic form. Reticuloendothelial siderosis
32-
28 -
Y
5 L
Parenchymal siderosis
0 0
24 -
0
0
0 0
20-
0
0
0
0
0 0
z -
0
00
aJ
-
z1 2-0
0 0
oo
0
O
R = 085
4 -0,"
-
O
I
I
0
1
I
I
0
0 0
0 0
-
0
VI
G 16-
28 24 20 16 12832-
0
0
0
0
0
0 0
0 0
O O 0 0
0
*O 0 0
4 -0"
-
R = 0.58
0 O
I
I
I
1
f
FIG.8.-Relationships between reticuloendothelial and parenchymal siderosis and hepatic fibrosis index in heavily transfused patients with thalassaemia. Symbols as in fig. 2.
fitting a log normal curve to the data (fig. 6). Although the severity of the fibrosis relative to liver iron concentration increased sharply as the latter exceeded 3.5 per cent. dry weight this was not due to an accelerated rate of iron
90
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLY”
deposition in older patients; the results showed that in fact the reverse occurred, liver iron concentration tending to level off in the older patients at a maximum value of approximately 5 per cent. dry weight (fig. 7). The severity of the fibrosis showed a good correlation with the RE siderosis ranking but not with the parenchymal siderosjs rank (fig. 8). Figure 8 also shows that there was no tendency for fibrosis to be more severe in splenectomised patients than in those with the spleen in situ. Chelator-treatedpatients. There was no significant change in the degree of fibrosis in serial biopsies obtained during the 6-year period of treatment (Barry et aE., 1974). Depending on the initial level, liver iron concentration rose in five patients, fell in two, and remained unchanged in two. The final values tended to level off at a maximum value of approximately 3 per cent. dry weightsignificantly lower, relative to the amount of blood transfused, than in the nonchelator treated patients; the change in liver iron concentration with age followed a similar pattern (fig. ’7). The relationship between the severity of the fibrosis and liver concentration corresponded, within the limited range of the variables, to that observed in the non-chelator treated patients (fig. 6).
DISCUSSION Iron overload in man occurs under four main circumstances: (1) in idiopathic haemochromatosis; (2) in chronic refractory anaemias marked by erythroid hyperplasia and (usually) ineffective erythropoiesis; (3) in chronic oral iron overload, as exemplified by Bantu siderosis; and (4) in transfusional iron overload. In the first three situations iron accumulation results from increased iron absorption,and severe fibrosis or cirrhosis are commonlyfound. In transfusional iron overload, on the other hand, fibrosis is often slight or absent despite equally massive iron deposition. To account for this discrepancy it has been suggested that parenterally administered haemoglobin-iron is less toxic than iron absorbed from the gut, the essential difference being that iron released from effete red cells is mostly held innocuously in reticuloendothelial (RE) cells whereas iron derived from the gut is mainly deposited in parenchymal cells with potentially harmful effect. This distinction is not absolute, however, as the relative severity of parenchymal and RE siderosis in transfusional siderosis is known to vary considerably from case to case, probably reflecting the fact that RE iron tends to be redistributed to the parenchymal cells in time. Current understanding of these aspects has been reviewed by Bothwell and Finch (1962). The role of other factors in the pathogenesis of liver injury have yet to be satisfactorily defined but both the magnitude and duration of iron-loading, and possibly the rate of plasma iron-turnover, are likely to be of particular importance and form the subject of the present discussion. In thalassaemia massive iron excess occasionally develops in subjects who have had few or no transfusions (Whipple and Bradford, 1936; Ellis, Schulman and Smith, 1954; Bannerman et al., 1967; Barry, 1974a) and under these circumstances can be attributed entirely to increased alimentary absorption. In the majority of cases, however, the relative importance of increased absorption and repeated transfusion for iron accumulation is much more difficult to
IRON OVERLOAD
91
ascertain and is further confused by the wide variation in transfusion schedules employed. Witzleben and Wyatt (1961) noted that cirrhosis was a constant but late feature in heavily transfused patients with thalassaemia. They assumed that transfused haemoglobin-iron was in general innocuously deposited in RE cells and attributed the cirrhosis in their patients to the attainment of a critical level of parenchymal cell iron, which they supposed was mainly derived from persistently high alimentary absorption. Although conforming with accepted concepts, the validity of these conclusions may be questioned on two counts: firstly, because it is not possible to determine by direct means what fraction of the iron load, if any, is gut-derived; and secondly, because after the initial stages of iron-loading it is impossible to discern histologically how iron derived from different sources is apportioned between the parenchymal and RE cells. Although increased iron absorption is well documented in thalassaemia and other refractory anaemias associated with marrow hyperplasia, there is evidence of various kinds suggesting that it probably does not contribute significantly to iron-loading in heavily transfused patients : 1. Direct measurements of iron absorption in thalassaemia have shown that the values in transfused children differ but little from those found in controls (Erlandson et al., 1961; Bannerman et al., 1964). This is in accord with generally held concepts that the rate of iron absorption closely reflects erythropoietic activity and that suppression of erythropoiesis by frequent transfusions indirectly checks iron absorption. 2. If continued iron absorption contributes significantly to iron-loading in heavily transfused patients the amount of iron found in the cadavers of such subjects should exceed the amount received as blood in life. However, review of published data reveals that with but few exceptions this applies only in cases that have received relatively few or no transfusions; in the majority of heavily transfused subjects (including thalassaemics) cadaver iron has actually fallen short of the amount transfused in life and, as pointed out by Oliver (1959), the observed discrepancies tend to increase with the number of transfusions (fig. 9). This paradox can be explained by a combination of two factors: firstly, that frequent transfusions indirectly check iron absorption by suppressing erythropoiesis; and secondly, that heavy spontaneous iron loss ensues when iron excess is massive. The alternativepossibility that excessive absorption continues despite transfusions but is more than offset by heavy spontaneous loss seems less likely and conflicts with the observations on iron absorption in thalassaemia. 3. The final piece of evidence stems from the measurements of liver iron concentration made in the present patients. Liver iron concentration is highly correlated with quantitative measurements of total body storage iron in a variety of iron-loading disorders including splenectomised and nonsplenectomised patients with thalassaemia (Barry, 1974b). In the non-chelated children included in this study liver iron concentration was in accordance with the values expected for the amount of blood received until about 100 units had been given ; thereafter the observed values were progressively less than expected for the transfused load, providing further indirect evidence of spontaneous iron excretion in more heavily transfused subjects.
92
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLYNN
From consideration of these facts it is hard to avoid concluding that the iron excess in our patients was largely, if not entirely, derived from the large amounts of blood they had received since infancy and that increased iron absorption was not a significant factor for their iron accumulation. Parenchymal and RE siderosis were assessed separately by ranking the intensity of iron deposition in each compartment on photomicrographsprepared under standard conditions. Iron deposition was at all stages both parenchymal 200
150
h
E
e00
v
c
100 0 I
a
b P W
50
I 200 Transfused iron (grams)
FIG. , 9.-Relationships between cadaver iron and the amount of iron transfused in life in heavily transfused patients. The proportion of iron recovered in the cadaver progressively decreases the more heavily transfused the patient. Based on two personal cases and published data (key references: Howell and Wyatt, 1953; Cappell, Hut.chison and Jowett, 1957; Oliver, 1959; Witzleben and Wyatt, 1961).
and RE in distribution, the degree in each compartment showing significant correlations with the age of the patient, with liver iron concentration, and with the amount of blood transfused. Thus it appeared that, in general, iron was distributed at random in both parenchymal and RE cells without preferential channelling into one or other compartment. Superimposed on this general pattern, however, there were two factors clearly predisposing towards relatively high parenchymal siderosis rankings, namely liver iron concentrations in excess of 3 per cent. dry weight and splenectomy. The effect of splenectomy on iron distribution, and the potential dangers thereof, have been previously noted
IRON OVERLOAD
93
(Witzleben and Wyatt, 1961; Berry and Marshall, 1967); however, our results did not suggest that splenectomised patients went on to develop higher total liver iron concentrations than non-splenectomised patients or that they were more likely to develop severe fibrosis. Some degree of fibrosis was present in all but three of the 52 biopsies examined, ranging in severity from a slight excess of fibrous tissue in the portal tracts to an established nodular cirrhosis. The more severe changes resembled those previously described in multiply transfused thalassaemic patients (Witzleben and Wyatt, 1961) and in idiopathic haemochromatosis (Sheldon, 1935) and may therefore be regarded as more advanced than those generally associated with transfusional siderosis. The rate of progression of the fibrosis differed significantly between the nonchelator and chelator-treated patients ; as reported elsewhere (Barry et al., 1974), this was associated with a progressive and significant difference in liver iron concentration. In the non-chelator treated patients fibrosis was progressive, its severity correlating closely both with age and with liver iron concentration. The rate of fibrosis was constant whereas the rate of increase in liver iron Concentration was not constant but occurred more rapidly in the younger patients than in the older and more heavily transfused ones. From consideration of these interrelationships it is apparent that the rate of fibrosis varied with liver iron concentration, proceeding slowly when liver iron concentration was less than 3 per cent. dry weight but accelerating at higher concentrations. Although it is tempting to relate this to the relatively greater parenchymal siderosis observed at high liver concentrations the results failed to bear this out, the severity of the fibrosis in fact showing a better correlation with RE siderosis ranking than with parenchymal ranking. In the chelator-treated patients liver iron concentration levelled off at a maximum value of about 3 per cent. dry weight. Although significantly lower than the final values in the controls this level is well within the range found in idiopathic haemochromatosis (Sheldon, 1935; Barry, 1973) and it was therefore surprising that the severity of the fibrosis showed no appreciable change during the 6-yr period of study; the retardation in fibrosis seemed disproportionate to the reduction in tissue iron level that treatment had achieved. When the results are compared with those for the non-chelator treated patients, however, the explanation becomes clear, namely that chelation therapy held liver iron concentration below the level at which rapid fibrosis occurs. In conclusion, our findings differ in two important respects from those traditionally associated with the pathology of transfusional siderosis-firstly, in the balanced distribution of iron in both parenchymal and R E cells from an early stage of iron-loading, and secondly, in the progressive course of the fibrosis. The distribution of iron may partly depend on the rate of plasma iron turnover, a high turnover conceivably accelerating the redistribution of iron from RE to parenchymal cells. However, we have no data on plasma iron turnover in our patients and the importance of this factor must remain speculative. The progressive and eventually severe fibrosis associated with transfusional iron overload in thalassaemia may largely reflect one special feature of the disease, namely the
94
R. ANTHONY RISDON, MICHAEL BARRY AND DAVID M. FLY"
combination of a high rate of iron accumulation with prolonged survival. The expected life-span of approximately 20 yr in regularly transfused thalassaemics (Engle, 1964) contrasts with the mean survival time of 5.5 yr in 17 patients regarded as representative of pure transfusional iron overload in adults (see Table 3 in Bothwell and Finch, 1962). Since, as we have shown, the rate of fibrosis depends on liver iron concentration, it follows that the severity of the fibrosis is a function of both liver iron concentration and time. However, our findings also show that even when iron accumulation is rapid severe fibrosis cannot be expected to develop in less than 10 yr. Thus, we would suggest that the relatively mild fibrosis generally found in patients with transfusional siderosis may reflect the limited survival of these subjects rather than any other factor. We accept the frequent association between cirrhosis and long-continued hyperabsorption of iron in certain hyperplastic erythroid states (including some cases of thalassaemia) but regard this as a self-selecting situation in that such cases have had disease mild enough to permit prolonged survival with little or no need for transfusion. SUMMARY
The interrelationships between liver iron concentration, the duration of iron-loading, and hepatic fibrosis, assessed morphometrically, have been studied in 32 specimens of liver obtained from 19 heavily transfused patients with thalassaemia major whose age ranged from 4 to 19 yr. Similar observations were made in a matched group of thalassaemic patients treated with long-term chelation therapy. The degree of liver damage ranged from very slight increase in fibrous tissue to severe fibrosis and cirrhosis. The severity of the fibrosis was closely correlated both with liver iron concentration and with age. While the relationship between fibrosis and age was linear, both the severity and the rate of fibrosis were exponentially related to liver iron concentration, damage accelerating as liver iron concentration exceeded 3 per cent. dry weight. By producing a modest but significant reduction in liver iron concentration chelation therapy resulted in a disproportionate but predictable retardation in the progression of the fibrosis. The factors affecting the distribution of iron between parenchymal and reticuloendothelial cells were also examined. In general stainable iron was uniformly distributed between parenchymal and reticuloendothelial cells from the early stages of iron-loading. Parenchymal siderosis was relatively heavier in splenectomised patients and in patients with liver iron concentrations above 3 per cent. dry weight than in non-splenectomised patients or patients with liver iron levels of less than 3 per cent. dry weight, but this did not affect the severity of the fibrosis. The relevance of these findings to the traditional concepts of the pathology of transfusional siderosis is discussed. REFERENCES BANNERMAN, R. M., CALLENDER, S. T., HARDISTY,R. M., absorption in thalassaemia. Brit. J. Haemat., 10,490.
AND
SMITH,R. S. 1964. Iron
IRON 0 VERLOAD
95
BANNERMAN, R. M., KEUSCH, G., KREIMER-BIRNEIAUM, M., VANCE, V. K., AND VAUGHAN, S. 1967. Thalassaemia intermedia, with iron overload, cardiac failure, hypopituitarism, and porphyrinuria. Amer. J. Med., 42,476. BARRY,M. 1973. Iron and chronic liver disease. J. Roy. Coll. Phycns. Lond., 8, 52. BARRY, M. 1974a. Unpublished observations. BARRY,M. 19746. Liver iron concentration, stainable iron, and total body storage iron. Gut, 15, 41 1. BARRY, M., AND SHERLOCK, S. 1971. Measurement of liver-kon concentration in needlebiopsy specimens. Lancet, i, 100. BARRY,M., FLY“, D. M., LETSKY,E. A., AND RKSDON, R. A. 1974. Long-term chelation therapy in thalassaemia major: effect on liver iron concentration, liver histology, and clinical progress. Brit. Med. J. ii, 16. BERRY, C. L., AND MARSHALL, W. C. 1967. Iron distribution in the liver of patients with thalassaemia major. Lancet, i, 1031. BOTHWELL, T. H., AND FINCH, C. A. 1962. Pathologic and clinical aspects of iron overload. In Iron metabolism, Churchill, London, chap. 9. CAPPELL, D. F., HUTCHISON,H. E., AND J o m , M. 1957. Transfusional siderosis: the effects of excessive iron deposits on the tissues. J. Path. Bact., 74, 245. ELLIS,J. T., SCHLJLMAN, I., AND SMITH,C. H. 1954. Generalised siderosis with fibrosis of liver and pancreas in Cooley’s (mediterranean) anemia. Amer. J. Path., 30, 287. ENGLE,M. A. 1964. Cardiac involvement in Cooley’s anaemia. Problems in Cooley’s anaemia. Ann. N. Y. Acad. Sci., 19, 694. ERLANDSON, M. E., WALDEN, B., STERN,G., HILGARTNER, M. W., WEHMAN, J., AND SMITH, C. H. 1962. Studies on congenital hemolytic syndromes. IV. Gastrointestinal absorption of iron. Blood, 19, 359. HOWELL,J., AND WYATT,J. P. 1953. Development of pigmentary cirrhosis in Cooley’s anemia. Arch. Path., 55, 423. KENT,G., AND POOPER,H. 1960. Secondary hemochromatosis: its association with anemia. Arch. Path., 70, 623. LISBOA,P. E. 1971. Experimental hepatic cirrhosis in dogs caused by chronic massive iron overload. Gut, 12, 363. OLIVER, R. A. M. 1959. Siderosis following transfusions of blood. J. Path. Bact., 77, 171. SHELDON, J. H. 1935. Haemochromatosis. Oxford University Press. WHIPPLE, G. H., AND BRADFORD, W. L. 1936. Mediterranean disease-thalassaemia (erythroblastic anemia of Cooley). J. Pediat., 9, 279. WITZLEBEN, C. L., AND WYATT,J. P. 1961. The effect of long-survival on the pathology of thalassaemia major. J. Path. Bact., 82, 1.
