Measurements of iron status and survival in African iron overload.

Abstract

Dietary iron overload is common in southern Africa and there is a misconception that the condition is benign. Early descriptions of the condition relied on autopsy studies, and the use of indirect measurements of iron status to diagnose this form of iron overload has not been clarified.
The study involved 22 black subjects found to have iron overload on liver biopsy. Fourteen subjects presented to hospital with liver disease and were found to have iron overload on percutaneous liver biopsy. Eight subjects, drawn from a family study, underwent liver biopsy because of elevated serum ferritin concentrations suggestive of iron overload. Indirect measurements of iron status (transferrin saturation, serum ferritin) were performed on all subjects. Histological iron grade and hepatic iron concentration were used as direct measures of iron status.
There were no significant differences in either direct or indirect measurements of iron status between the two groups. In 75% of these subjects the hepatic iron concentration was greater than 350 micrograms/g dry weight, an extreme elevation associated with a high risk of fibrosis and cirrhosis. Serum ferritin was elevated in all subjects and the transferrin saturation was greater than 60% in 93% of the subjects. Hepatomegaly was present in 20 of the 22 cases and there was only a moderate derangement in liver enzymes except for a tenfold increase in the median gamma-glutamyl transpeptidase concentration. There was a strong correlation between serum ferritin and hepatic iron concentrations (r = 0.71, P = 0.006). After a median follow-up of 19 months, 6 (26%) of the subjects had died. The risk of mortality correlated significantly with both the hepatic iron concentration and the serum ferritin concentration.
Indirect measurements of iron status (serum ferritin concentration and transferrin saturation) are useful in the diagnosis of African dietary iron overload. When dietary iron overload becomes symptomatic it has a high mortality. Measures to prevent and treat this condition are needed.

Full text

Page 1

-.
MEASUREMENTS OF IRON STATUS
AND SURVIVAL IN AFRICAN IRON
OVERLOAD
A Patrick MacPhail, Eberhard M Mandishona, Peter D Bloom,
Alan C Paterson, Tracey A RouauIt, Victor R Gordeuk
Introduction. Dietary iron overload is common in southern
Africa and there is a misconception that the condition is
benign. 'Early descriptions of the condition relied on
autopsy studies, and the use of indirect measurements of
iron status to diagnose this form of iron overload has not
been clarified.
Methods. The study involved 22 black subjects found to
have iron overload on liver biopsy. Fourteen subjects
presented to hospital with liver disease and were found to
have iron overload on percutaneous liver biopsy. Eight
subjects, drawn from a family study, underwent liver
biopsy because of elevated serum ferritin concentrations
suggestive of iron overload. Indirect measurements of iron
status (transferrin saturation, serum ferritin) were
performed on all subjects. Histological iron grade and
hepatic iron concentration were used as direct measures of
iron status.
Results. There were no significant differences in either direct
or indirect measurements of iron status between the two
groups, In 75% of these subjects the hepatic iron
concentration was greater than 350 Ilg1g dry weight, an
extreme elevation associated with a high risk of fibrosis and
cirrhosis. Serum ferritin was elevated in all subjects and the
transferrin saturation was greater than 60% in 93% of the
subjects. Hepatomegaly was present in 20 of the 22 caSes
and there was only a moderate derangement in liver
enzymes except for a tenfold increase in the median
Department ofMedicine, University ofthe Wihvatersrand, Johannesburg
A Patrick MacPhail, BSc, MB BCh, PhD, FCP (SA), FRCP
Eberhard M Mandishona, MB ChB
Peter D Bloom, MD
Deparhnent ofAnatomical Pathology, University ofthe Wihvatersrand and
South African Institute for Medical Research, Johannesburg
Alan C Paterson, MB BCh, PhD
Cell Biology and Metabolism Branch, National Institute ofChild Health and
Human Development, Bethesda, Maryland, USA
Tracey A Rouault, MD
Department ofMedicine, The George Washington University Medical Center,
Washington, DC, USA
Victor R Gordeuk, MD
September 1999, Vo!. 89, No. 9 SAMJ
garnma-glutamyl transpeptidase concentration. There was a
strong correlation between serum ferritin and hepatic iron
concentrations
(r = 0.71, P '= 0.006). After a median follow-up of 19 months,
6 (26%) of the subjects had died. The risk of mortality
correlated significantly with both the hepatic iron
concentration and the serum ferritin concentration.
Conclusions. Indirect measurements of iron status (serum
ferritin concentration and transferrin saturation) are usetul
in the diagnosis of African dietary iron overload. When ~
dietary iron overload becomes symptomatic it has a high
mortality, Measures to prevent and treat this condition are
needed.
5 Afr Med J1999; 89: 96&-972.
African iron overload, first described in 1929/ has not been well
characterised in living subjects. Early studies were based on
autopsy findings and documented an association with chronic
liver disease/''' specifically portal fibrosis and cirrhosis. Later
studies also described associations with diabetes mellitus,'
osteoporosis and scurvy: and cardiomyopathy.'" Hepatic iron
deposition in African iron overload occurs in parenchymal cells,
Kupffer cells and portal macrophages.' This is in contrast to
Caucasian hereditary haemochromatosis, in which condition
iron deposition is mainly parenchyma!.1O In both conditions
widespread deposition of iron may occur ID other organs.3
The development of African iron overload has been
associated with the consumption of traditional beer rich ID iron
leached from the steel containers used in brewing.n.nRecent
studies in southern Africa provide strong evidence that there is
a genetic predisposition to African iron overload.!"·l5
Understanding of the condition has been compromised by the
widely held belief, based on the prominence of macrophage
iron, that African iron overload is a benign condition and that
alcohol plays the major role in causing hepatic disease.
Despite the fact that African iron overload has been
recognised for more than 65 years and is still common in rural
Africa in the 1990s," it is not clear which indirect measurements
of iron status correlate with chemical hepatic iron concentration,
the gold standard in characterising iron overload." The purpose
of this study was to bring together the clinical and pathological
features of iron overload seen in subjects who were recruited as
part of a larger family study aimed at establishing a genetic
factor in African iron overload,l5 and to examine the hypothesis
that mortality correlates with the severity of iron overload.
MATERIALS AND METHODS
Study subjects
Twenty-two subjects aged 25 - 84 years with iron overload
confirmed on liver biopsy, were studied between 1993 and 1996

Page 2

I
If----
Table I. Traditional beer consumption, hepatic histology and measurements of iron stal:us in 22 subjects with African iron overload
Hepatic iron
Selection
Subject Age consump-Hepato-Kupffer
criteriaNo.
(yrs)
Sex
tion (1)
cytescells
Lifetime beer
Serum
ferritin
Hepatic
iron
(I1mo1/g)
Hepatic
iron
index
Macro-
phages
Hepatic histology
Fibrosis
TransferrinVitamin C
(l1g/]O'wbc)
Cil'l'hosis saturation (%) (Iig/!)
Status
Liver
biopsy
1
2
3
4
5
6
7
8
9
55
56
84
84
71
55
80
65
67
69
51
56
65
69
F
68544
316]6
141960
12090
186150
16992
]0'176
28440
31584
6528
14808
7488
30912
6048
3
3
4
3
3
4
3
4
3
3
3
3
3
3
3
3
4
3
4
3
3
4
4
3
3
3
3
3
3
3
4
3
3
3
4
4
3
3
3
4
3
3
4
1
4
4
1
1
1
1
4
1
1
1
3
3
1
0
1
1
1
0
1
0
1
1
0
0
0
0
105
81
91
99
98
101
96
86
96
104
93
94
95
92
38483
1799]
6914
6716
5074
4968
4307
4016
3091
2938
2459
2062
] 900
1331
428
557
7.8
9.9
12.3
7.5
9.8
7.8
4.6
0.7
9.9.
2.5
21.4
3.7
4.0
12.4
14.1
16.8
6.2
13.3
9.4
Dead
Dead
Dead
Alive
Dead
Alive
Alive
Alive
Dead·
Alive
Alive
Alive
Alive
Alive
M
M
M
M
M
M
M
M
M
M
M
F
M
1035
627
694
173
612
577
361
2.7
9.1
8.4
7.1
10
11
12
13
14
4687.2
Raised
ferritin
serum
15
16
17
18
]9
20
21
22
65
70
69
75
70
54
25
60
M
F
M
M
F
M
F
M
41472
67200
67160
147840
17280
84000
3
3
3
3
2
3
2
2
3
3
3
4
2
1
2
1
4
3
3
3
1
2
1
1
2
3
3
4
1
2
1
1
1
0
0
1
0
0
0
0
92
93
87
103
35
3]
108
67
13108
3655
3009
2718
2 ]62
1180
1026
773
725
438
558
275
82
217
11.2
6.3
8.1
3.7
1.2
4.0
6.4
9.0
12.1
7.0
13.6
18.0
44.7
25.4
Dead
Alive
Alive
Alive
Alive
Alive
Alive
Alive
72
1584

Page 3

ORIGINAL ARTICLES
as part of a larger genetic study on iron overload based at
Shongwe Hospital, Mpumalanga.15Fifteen of these patients
were identified as index cases when they presented to hospital
with hepatomegaly and other features of liver diseases
(abdominal pain, ascites, haematemesis, hepatic encephalopathy
and/or peripheral oedema), and the diagnosis of iron overload
was made on liver biopsy. A further 8 subjects, from 2 of the
above patients' families, underwent diagnostic liver biopsy
when they were found to have elevated serum ferritin
concehtrations suggestive of iron overload. A medical history
was taken and physical examination was performed on each
subject. The volume of traditional beer consumed was estimated
by direct questioning of each subject. A typical indigenous clay
pot or udzivo, employed as a measure when selling beer, was
used to assess the amount imbibed each day. The total lifetime
beer consumption (in litres) was calculated from the daily or
weekly consumption and the number of years spent drinking.
Fasting morning blood samples were collected by venipuncture
on 2 consecutive days.
Collection and analysis of traditional beer samples
Forty-eight samples of traditional beer that were ready for
consumption were collected from homes in the Shongwe
community and frozen immediately at -70°C. The iron
concentrations in the beer supernatants were measured after
acid digestion using bathophenthroline disulphonate at 535 nm
absorbance. The alcohol concentration in the beer was
determined by ultraviolet spectroscopy.t·
Analysis of liver biopsy specimens
A direct estimate of iron status was obtained by grading the
liver biopsy histologically and by measuring the chemical iron
concentration. Hepatocyte iron was graded histologically on
sections stained with Perl's reagent using the method of Scheuer
and colleagues.l9Kupffer cell iron and portal tract macrophage
iron were each assessed separately. Iron grades were assigned
according to the consensus of four observers (ACP' VRG, APM,
EMM) observing the specimen simultaneously and blinded to
the results of the indirect measurements of iron status. Fibrosis
and cirrhosis were assessed on the basis of haematoxylin and
eosin, Masson's and reticulin stains. The hepatic non-haem iron
concentration was measured on dewaxed samples prepared
from the histology blocks,'" and the hepatic iron index {Jlmol
iron/g liver tissue (dry weight)/age)21 was calculated.
Indirect measures of iron status
Serum iron and total iron binding capacity were measured
using methods recommended by the International Committee
for Sta.n.dardisation in Haematology.= The transferrin
saturation was calculated by dividing the serum iron by the
total iron binding capacity and multiplying by 100, with a
maximum of 100%. The unsaturated iron binding capacity was
the difference between the total iron binding capacity and the
.
September 1999, Vo!. 89, No. 9 SAMJ
serum iron, with a minimum of O. The serum ferritin
concentration was measured by enzyme-linked immunosorbent
assay (ELlSA)." The ratio of serum ferritin to aspartate
aminotransferase (AST) measured on day 1 was calculated
because this ratio reflects hepatic iron stores in the setting of
acute alcohol consumption, shortly after stopping and after up
to 3 weeks' abstention from alcoho!.25.26 The mean of two results
measured on blood samples drawn on consecutive days was
used in the indirect measurements of iron status.
Haematological and biochemical tests
The full blood count was performed using an automated cell
counter (Sysmex K 1 000, CA Milsch, Kobe, Japan). Serum
concentrations of lactate dehydrogenase (LDH), AST, alanine
aminotransferase (ALT), garnma-glutamyl transpeptidase
(-yGT), bilirubin, glucose and s
automated clinical chemistry analyser. Erythrocyte
sedimentation rate (ESR) was determined using the Westergren
method. C-reactive protein was measuredby
immunoturbidimetric assay (Boehringer Mannheim, Germany).
The reticulocyte count was determined microscopically with
peripheral blood smears stained with supra vital stain. Hepatitis
Bviral markers (hepatitis B surface antigen (HBsAg), surface
antibody (anti-HBs), core antibody (anti-HBcore» were
measured by radio-immunoassay (RIA) (Abbot Diagnostics,
Chicago, USA), and antibody to hepatitis C virus (anti-HC) was
measured by ELlSA (Murex Diagnostics, Temple Hill, Kent,
England). White blood cell ascorbic acid concentration was
determined according to the method of Denson and Bowers."
odi~ were measured using an
Statistical methods
Continuous variables were compared with the Mann-Whitney
U-test and categorical variables were compared with the Fisher
exact or the Pearson chi-square test. Linear regression was used
to examine relationships between variables. Hepatic iron stores
do not correlate well with serum ferritin concentration at levels
greater than 4 000 Ilg/l.'" This study includes 3 subjects (see
Table l) in whom serum ferritin was greater than 10 000 IlgII
and in whom the serum ferritin values are randomly scattered
in the data. As regression analysis of the data identifies these
values as outliers (studentised residual absolute values greater
than 2), these values were not included in the comparison of
direct and indirect measures of iron status. The Spearman
correlation coefficient was used to analyse indirect measures of
iron status according to the histological grades of hepatic iron.
Cox proportional hazards models were used to analyse the
influence of hepatic iron and serum ferritin concentrations on
time to death.
RESULTS
Iron-related measurements
Table I lists the iron-related features and liver biopsy findings in
each of the subjects. The iron-related features are summarised in
a

Page 4

ORIGINAL ARTICLES
Table 11. Exposure to traditional beer and measurements of iron
status in 22 subjects with African iron overload (continuous
variables are medians and ranges)
the subjects admitted to drirIking commercial 'Western-type'
alcohol occasionally, mostly in the form of beer, while the
consumption of spirits was very low.
Table IT. There were no significant differences in the age, se>.
distribution, lifetime traditional beer consumption, chemical
hepatic iron concentration or biochemical measurements of iron
status between those subjects selected on the basis of iron
overload on liver biopsy and those selected on the basis of
raised serum ferritin who subsequently had a liver biopsy.
Significant differences were encountered in the hepatic iron
grading on histological examination (see 'Liver histology and
iron measurements' below).
Exposure to dietary iron
The exposure to traditional beer varied from a calculated
lifetime consumption of 72 litres to over 150 000 litres and there
was also a wide variation in the amount of traditional beer
claimed to be consumed per session, ranging from 1 to 14 litres
per session. An inverse correlation was found between the
estimated amount of traditional beer consumed per day and the
serum sodium concentration, as a marker of haemodilution,
lending some credence to these claims (r = 0.59, P = 0.003). The
mean (SD) alcohol concentration of 48 samples of home-brewed
beer collected from the subjects' households was 3.2% (0.4%)
with a mean iron content of 46 (17) mg/I. In terms of exposure
to alcohol and iron this translates into a median of 96 g of
alcohol and 138 mg of iron per drirIking day. Sixty per cent of
Indirect measurements of iron status
The transferrin saturation was greater than 60% in 93% of the
subjects and was mirrored by the low median unsaturated
binding capacity. In three-quarters of the subjects the transferrin
saturation was greater than 90%. Seventy-seven per cent of the
subjects had a total iron binding capacity lower than normal, a
reflection of low plasma transferrin concentration characteristic
of iron overload. The serum ferritin was markedly elevated in
all subjects. In 3 of the subjects, 1 of whom had an
hepatocellular carcinoma, the serum ferritin concentration was~
greater than 10000 J.1g/I. The ferritin/AST ratio was greater· -=-
than 15 in all but 1 case, suggesting that alcohol alone was not
the cause of the high ferritin values. Nineteen of the 22 cases
(86%) had white cell vitamin C concentrations below the normal
range.
Liver histology and iron measurements
The histological grade of hepatocyte (parenchymal) iron was
above the normal range in all study subjects: grade 2 in 3, grade
3 in 16 and grade 4 in 3 (Table I). The distribution of iron in the
liver varied from heavy iron deposits (4) in portal macrophages
with moderate parenchymal iron to massive parenchymal iron
(4) and little portal macrophage iron (I). In all instances,
Kupffer cell iron was present, with a gradation of mild (1) to
heavy (4). The group selected on liver biopsy had significantly
higher iron grades in hepatocytes, Kupffer cells and
macrophages than the group selected on the basis of raised
serum ferritin concentration (P = 0.013, 0.025 and 0.025,
respectively). Fibrosis was present in all cases and cirrhosis was
detected in 9 (41%) of the needle biopsies, but this figure may
be an underestimate because the detection of fibrosis and
cirrhosis is unreliable in needle biopsies. There was no
significant difference in the degree of fibrosis or the prevalence
of cirrhosis between the hoVO groups. While 2 of the subjects (1
in each group) had mild steatosis, no other features of liver
damage attributable to alcohol were seen. In 16 subjects Lhere
was sufficient biopsy material available for the direct chemical
measurement qf hepatic iron. The median hepatic iron
concentration was 512 J.1mol/g dry weight (range 82 - 1 035) and
in 12 cases (75%) concentrations were greater than 360 J.1mol/g
dry weight, the level associated with high risk for the
development of cirrhosis in both African iron overload' and
Caucasian hereditary haemochromatosis.'l In all but 1 subject,
the hepatic iron index was greater than 1.9, the level used to
distinguish between iron-loading due to alcohol and that
caused by hereditary haemochromatosis.2l.29.30 No s t ~ t i s t i c a l l y
significant difference in iron concentration or hepatic iron index
was detected between the hoVo groups.
;,,20
<30
0-1
20-45
Normal
range
250 -400
20 -400
< 15
100-300
Result
3 (2 - 4)
66 (25 - 84)
5 (23)
14 (O - 197)
93 (31 -100)
7.6 (l.2 - 12.3)
512 (82 - 1 035)
212 (113 - 304)
3 050 (773 - 38 483)
56 (5 -193)
29 676 (72 - 186 150)
40 (2 - 65)
3 (0.4 - 14)
Characteristic
Age (years)
Female subjects (N (%»
Estimated traditional beer
consumption
Lifetime total (1)
Years of exposure
Amount consumed per day (I)
Measures of iron status
Direct measurements
Hepatocellular iron grade
{median (range»
Hepatic iron concentration
{JImol/g dry weightY
Hepatic iron index (JImol/g
dry weight/years)
Indirect measurements
Transferrin saturation (%)
Unsaturated iron binding
capacity (JIg/dl)
Total iron binding capacity
(JIg/dl)
Serum ferritin (JIgfl)
Ferritin/AST ratio (JIg/U)
Vitamin C (JIg/IO"white blood
cells)9.7 (0.7 - 44.7)
'N~
16.

Page 5

- ;-
i
j
20 30 40 50 60 70 80 90 100
Per cent
Hepatomegalyliiif55::....
Tuberculosis
Ascites
Cardiomegaly
Diabetes
Back pain
Joint pain
Pigmented
Scurvy
Table ill. Haematology and biochemistry in 22 subjects with
African iron overload (continuous variables are medians.and
ranges)
*Values in the left-hand column are nonnal values.
Haemoglobin (g/dl)
Reticulocytes (%)
Erythrocyte sedimentation
rate (mm/h)
C-reactive protein
(mg/I mean)
Glucose (mmol/l)
Sodium (mEq/l)
Bilirubin (mmol/I)
Aspartate aminotransferase
(VII)
Alanine aminotransferase
(V/I)
Gamma-glutamyl transpeptidase
(V/I)
Viral hepatitiS markers
Hepatitis A antibody
(N (%) positive)
Hepatitis B core antibody
(N (%) positive)
Hepatitis Bsurface antibody
(
(%) positive)
Hepatitis Bsurface antigen
(N (%) positive)
Hepatitis C antibody
(N (%) positive)
13.3 (10.5 - 15.3)
0.8 (0.3 - 1.9)
12.0 - 18.0'
<2.0
49 (4 - 114)
< 20
2 (0 - 174)
4.8 (2.3 - 14.0)
138 (121 - 146)
8 (2 - 147)
,;2
< 6.6
135 -145
<25
63 (14 - 405)
<35
41 (7 - 172)
<35
234 (55 - 1 836)
<35
22 (100.0)
negative
14 (64)
negative
13 (59)
negative
1 (5)
negative
0(0)
negative
Fig. 2. Clinical features seen in 22 subjects with iron overload.
normal in all cases. Hepatitis virus markers showed that all but
1 of the subjects had been exposed to the hepatitis A virus,
while none had evidence of hepatitis C infection. Eighteen
(78%) had evidence of past exposure to hepatitis B virus, but
only 1 (5%) was positive for hepatitis B surface antigen.
subject showed evidence of chronic viral hepatitis on liver
histology.
0
7000
•
6000
'a,
-3
:§
·C
5000
4000
~
E
:::>
ID
(/)
3000
•
•
2000
•
1000
•
Correlation between direct and indirect
measurements of iron status
Serum ferritin concentration correlated significantly with the
hepatic iron concentration (r =0.71, P =0.006) (Fig. 1). Similarly,
serum ferritin correlated significantly with the histological
grading of hepatocellular (rho =0.48, P =0.03), Kupffer cell (rho
= 0.70, P = 0.007) and portal macrophage iron (rho = 0.54, P =
0.03). The relationship of hepatic iron concentration with
transferrin saturation (r =0.51) and unsaturated iron binding
capacity (r = 0.55) were also significant (P < 0.04).
Fig. 1. Plot ofserum ferritin and hepatic iron concentration in 13
subjects with African iron overload. Three subjects with serum
ferritin concentrations greater than 10 000 )./g/l were excluded (see
methods).
o
200
Hepatic iron concentration (Ilmol/g)
400 600800 1 0001 200
Clinical features and other laboratory measurements
Clinical features and other laboratory measurements are shown
in Fig. 2 and in Tables I - Ill. Apart from the erythrocyte
sedimentation rate (see below), there were no significant
differences in these features between those subjects selected on
the basis of iron overload on liver biopsy and those selected on
the basis of raised serum ferritin. The most common clinical
finding was hepatomegaly, which was present in 91% of the
subjects. A quarter of these subjects had ascites. More than half
of the subjects complained of joint pain. Only 1 subject had
clinical evidence of scurvy despite low white blood cell levels of
vitamin C « 20 I-lg/IOS white cells) in 19 subjects (86%). One-
fifth had a history of treated pulmonary tuberculosis. Two
subjects had a fasting blood glucose greater than 6.6 mmol/I
and 3 had clinical evidence of cardiomegaly. The median
haematological measurements were within the normal range.
The erythrocyte sedimentation rate was elevated above normal
in most of the subjects, with the group selected on the basis of
iron overload on liver biopsy having a significantly higher
mean erythrocyte sedimentation rate (61 (29) and 32 (20) mm in
1 hour, respectively, P = 0.01). Abnormalities in liver function
tests were common, particularly the yGT, which was above
September 1999, Vo!' 89,o. 9 SAMJ
z