Pathogenesis and Treatment of Kidney Disease Bone Marrow Iron, Iron Indices, and the Response to Intravenous Iron in Patients With… [613852]

ORIGINAL INVESTIGATIONS
Pathogenesis and Treatment of Kidney Disease
Bone Marrow Iron, Iron Indices, and the Response to Intravenous Iron in
Patients With Non–Dialysis-Dependent CKD
Simona Stancu, MD, PhD,1,2Ana Stanciu, MD, PhD,2Adrian Zugravu, MD,1Liliana Bârsan, MD,2
Daniela Dumitru, MBiol,2Mariana Lipan, MChem,2and Gabriel Mircescu, MD, PhD1,2
Background: Information about iron stores and their relationship with transferrin saturation (TSAT),
serum ferritin, and the erythropoietic response to iron therapy is scarce in anemic non–dialysis-
dependent patients with chronic kidney disease (CKD). We examined the diagnostic utility of peripheral-
iron indices and the erythropoietic response to intravenous iron as indices of iron store depletion using
bone marrow iron as a reference test in anemic non–dialysis-dependent patients with CKD.
Study Design: Diagnostic test study.
Setting & Participants: 100 anemic (hemoglobin /H1102111 g/dL) patients with CKD stages 3-5, not
receiving epoetin and iron.
Index Tests: TSAT index and serum ferritin level at baseline and increase in hemoglobin level 1
month after 200 mg of iron sucrose daily for 5 days.
Reference Test: Bone marrow iron (assessed using aspiration and Perls’ stain), depleted versus
replete, at baseline.
Measurements: Area under the receiver operating characteristic curve (AUROC), sensitivity, and
specificity of peripheral-iron indices and erythropoietic response to describe bone marrow iron stores.
Results: Bone marrow iron stores were depleted in 48% of patients at baseline. In iron-depleted
versus -replete subjects, mean hemoglobin level, median TSAT index, median serum ferritin level, and
hemoglobin level increase after iron sucrose administration were 8.74 /H110061.1 (SD) versus 9.22 /H110060.9
g/dL (P /H110050.02), 19% (interquartile range [IQR], 15%) versus 28% (IQR, 12%; P/H110210.001), 100 (IQR,
131) versus 220 ng/mL (IQR, 213; P/H110210.001), and 1.2 /H110060.4 versus 0.8 /H110060.3 g/dL (P /H110210.001),
respectively. TSAT, ferritin level, and increase in hemoglobin level AUROCs were similar: 0.75 (95% CI,
0.66-0.85), 0.76 (95% CI, 0.66-0.85), and 0.74 (95% CI, 0.65-0.84), respectively.
Limitations: Bone marrow iron as the index of iron stores.
Conclusions: Half the anemic patients with CKD stages 3-5 had depleted iron stores. Peripheral-iron
indices and erythropoietic response had equivalent, but limited, utility in identifying depletion of bone
marrow iron stores. Use of these indices to indicate depletion of iron stores should be reconsidered.
Am J Kidney Dis 55:639-647. ©2010 by the National Kidney Foundation, Inc.
INDEX WORDS: Anemia; bone marrow iron stores; ferritin; transferrin saturation index; intravenous iron.
Editorial, p. 617
Anemia is common in patients with chronic
kidney disease (CKD) and multifactorial
in origin.1The amount of iron available for
erythropoiesis seems to be an important determi-
nant of hemoglobin level in both hemodialysis
patients and non–dialysis-dependent patients with
CKD despite dissimilar iron status.2-6For ex-
ample, despite lower iron loss in predialysis pa-
tients, iron deficiency was diagnosed using bone
marrow examination in non–dialysis-dependent pa-
tients with CKD in proportions varying from
23%-/H1102290%.7,8In a representative sample of the
US population, less than one-third of anemic pa-
tients with CKD had ferritin levels /H11021100 ng/mL
and transferrin saturation (TSAT) /H1102120% (consid-ered by the National Kidney Foundation’s Kid-
ney Disease Outcomes Quality Initiative
[KDOQI] to reveal iron deficiency), and higher
TSATs were associated independently with higher
hemoglobin levels without a threshold effect.9
Furthermore, the utility of iron indices to predict
From the1“Carol Davila” University of Medicine and
Pharmacy and2“Dr. Carol Davila” Teaching Hospital of
Nephrology, Bucharest, Romania.
Received December 19, 2008. Accepted in revised form
October 9, 2009. Originally published online as doi:10.1053/
j.ajkd.2009.10.043 on January 18, 2010.
Address correspondence to Gabriel Mircescu, MD, PhD,
Calea Grivitei 4, 010731 Bucharest, Romania. E-mail:
gmircescu@hotmail.com
©2010 by the National Kidney Foundation, Inc.
0272-6386/10/5504-0007$36.00/0
doi:10.1053/j.ajkd.2009.10.043
American Journal of Kidney Diseases, Vol 55, No 4 (April), 2010: pp 639-647 639

response to iron therapy was poor in a study by
Van Wyck et al.10However, intravenous (IV)
iron administration resulted in significant im-
provement in anemia in 30%-60% of non–
dialysis-dependent patients with CKD, even with-
out concomitant erythropoiesis-stimulating agent
(ESA) therapy,11-17and IV iron sucrose or feru-
moxytol was shown to be superior to oral iron in
anemic patients with low iron indices.10,11
Taken together, these data suggest that iron
deficiency is frequent in patients with CKD, but
the prevalence of patients with depleted iron
stores and the relationship between peripheral-
iron indices, erythropoietic response to IV iron,
and iron stores are still insufficiently defined.
This study explores the diagnostic utility of
common peripheral-iron indices and the erythro-
poietic response to IV iron in the diagnosis of
iron deficiency in anemic non–dialysis-depen-
dent patients with CKD who were ESA and iron
naive. Bone marrow iron was used as a reference
test.
METHODS
Study Design
This is a diagnostic study on patients admitted to a single
nephrology department. Study duration was 3 months, with
a run-in period of 2 months and follow-up period of 1 month.
In the run-in period, hemoglobin levels, iron indices, and
estimated glomerular filtration rate (eGFR) were measured.
At the end of the run-in period, blood was drawn for baseline
laboratory tests, and bone marrow aspiration was performed.
Thereafter, 200 mg of iron sucrose was administered IV
daily for 5 days to all patients irrespective of iron status. One
month after the last iron dose, blood was drawn for fol-
low-up assessment. No patient received an ESA before or at
any time during the study.
The index tests were peripheral-iron indices, in other
words, TSAT index and serum ferritin level, and erythropoi-
etic response to IV iron, defined as a /H113501 g/dL increase in
hemoglobin level at assessment versus baseline. The refer-
ence test was bone marrow iron, defined as depleted or
replete (vide infra).
The study was approved by the Helsinki Committee of the
hospital.
Patients
One hundred adults (aged /H1102218 years) were selected from
patients admitted during a 2-year period ( Fig 1).
Selection Criteria
Anemia, defined as hemoglobin level /H1102111.0 g/dL (based
on 3 measurements in the previous 8 weeks with a /H113492.5
g/dL difference between the highest and lowest values) andstage 3-5 CKD, mean eGFR /H1102160 mL/min 1.73 m2(calcu-
lated using the 4-variable Modification of Diet in Renal
Disease Study equation18and based on 3 serum creatinine
measurements in the previous 8 weeks, separated from each
other by at least 1 week) were considered inclusion criteria.
Patients with previous iron or ESA therapy, active infec-
tious conditions, cancer, iron overload (serum ferritin /H11022800
ng/mL), blood transfusions and/or active bleeding within the
preceding 3 months, hemolytic anemia, folate or vitamin B12
deficiency, severe malnutrition (Subjective Global Assess-
ment score of C on a scale of A, B, and C),19or hypothyroid-
ism were excluded. Similarly, individuals with congestive
heart failure (New York Heart Association class III or IV),
severe uncontrolled high blood pressure (systolic blood
pressure /H11022190 mm Hg and/or diastolic blood pressure /H11022115
mm Hg), active liver disease (/H110223 times increase in alanine
or aspartate aminotransferase level), uremic complications
(pericarditis or polyneuropathy), severe hyperparathyroid-
ism (intact parathyroid hormone /H11022800 pg/mL), severe psy-
chiatric disorders, known hypersensitivity to any component
of iron sucrose, and those participating in other clinical trials
were not enrolled.
Laboratory Tests
Bone Marrow Parameters of Iron Status
Bone marrow was collected using aspiration from the
sternum. Smears were prepared from marrow fragments,
colored with Perls’ Prussian blue stain, and interpreted by a
senior hematologist who had no knowledge of patients’ iron
status. Additional slides from aspirate were prepared if
necessary, and at least 9 bone marrow particles were re-
viewed for a final diagnosis.20
The occurrence of siderotic granules in macrophages was
graded on a scale of 0-6.21Patients with a score of 0 or 1
were considered iron deficient; those graded from 2-4, as
Figure 1. Flow chart of patient selection.Stancu et al 640

having normal macrophage iron, and those graded from 5-6,
as having macrophage iron overload.
Erythroblasts with green blue particles on Perls’ stain
were defined as sideroblasts. According to the number of
siderotic granules, sideroblasts were classified as type 0 (0
granules), type 1 (1-3 granules), or type 2 (/H113503 granules).22
Percentages of each type of sideroblast were computed.
The pattern of bone marrow iron distribution was classi-
fied as normal (macrophage iron of 2-4, sideroblasts in
normal percentage), iron deficiency (macrophage iron of 0
or 1, sideroblasts absent or present in a very low percentage),
iron overload (macrophage iron /H113505, sideroblasts in normal
or increased percentage), and anemia of chronic inflamma-
tion (macrophage iron /H113505, sideroblasts absent or present in
very low percentage).
Patients were classified according to bone marrow iron
distribution as iron depleted (bone marrow iron deficiency)
or iron replete (normal iron stores, iron overload, or anemia
of chronic inflammation).
Peripheral Parameters of Iron Status
Serum ferritin and serum transferrin were measured using
immunoturbidometric methods (Good Biotech, www.good-
biotech.com; Giesse Diagnostics, www.giessediagnostics.
com) on an Olympus AU400 autoanalyzer (Olympus Diag-
nostica GmbH, www.olympus.diagnostica.com). Total se-
rum iron-binding capacity was calculated as serum trans-
ferrin level multiplied by 1.25. Serum iron was assessed
using a colorimetric method (Giesse Diagnostics). TSAT
index was calculated as the percentage of serum iron from
total serum iron-binding capacity.
C-Reactive Protein
C-Reactive protein (CRP) was assessed using a high-
sensitivity latex immunoturbidometric method (Giesse Diag-
nostics).
Statistical Analyses
Data are presented as mean /H11006standard deviation, 95%
confidence interval (CI), or median and interquartile range
(IQR) according to their distribution.
tTest (paired or not), Mann-Whitney, /H92732, Wilcoxon, or
McNemar test was used to compare groups, as required.
Receiver operating characteristic (ROC) curves23-25were
created using bone marrow iron status (depleted vs replete)
as the dichotomous variable and ferritin level, TSAT, and the
difference from baseline in hemoglobin levels 1 month after
iron administration as continuous variables. Odds ratio (OR)
for a positive test, area under the ROC curve (AUROC),
sensitivity, and specificity were calculated at various cutoff
values.26
Multivariable binary logistic regression was used to ana-
lyze relationships between bone marrow iron (replete vs
depleted) and relevant parameters.
Considering a probability of 50% to attain a /H113501 g/dL
increase in hemoglobin level and a presumed significant differ-
ence among groups of 10%, at least 27 patients per group
would be required for a 95% probability and power of 95%.
Statistical analyses were performed using Analyse-it (Ana-
lyse-it Software Ltd, www.analyse-it.com) and SPSS (SPSS
Inc,www.spss.com) packages.RESULTS
One hundred patients were enrolled. Median age
was 62 years (range, 24-84 years), and 55% were
/H1102260 years old. There was a slight male preponder-
ance (55%). Vascular nephropathies and primary
glomerulonephritis were the main causes of CKD,
and diabetic nephropathy was rare (43%, 36%, and
3%, respectively). Median eGFR was 14 mL/min
1.73 m2(IQR, 10), with most patients in CKD
stage 4 or 5 (28% and 56%; Table 1).
Inflammation was highly prevalent because
60% of patients had a serum CRP level /H1102210
mg/L, and the median CRP level of the cohort
was 15 mg/L (IQR, 24). Nutritional status was
good: 65% of patients had a Subjective Global
Assessment score of A, mean body mass index
was 24 /H110064 kg/m2, and mean serum albumin
level was 3.7 /H110060.5 g/dL.
Anemia was relatively severe: mean hemoglo-
bin level was 8.99 g/dL, and 70% of patients had
a hemoglobin level /H1102110 g/dL.
Bone Marrow Stores and Peripheral-Iron Indices
Iron deficiency was diagnosed in 48% of pa-
tients. Normal iron distribution was found in
31%, and iron overload in 18%, but only 3% had
a pattern suggestive of anemia of chronic inflam-
mation (Fig 2).
Only 17% of patients had iron indices sugges-
tive of iron deficiency, in other words, TSAT
/H1102120% and ferritin level /H11021100 ng/mL (Table 1).
To evaluate the diagnostic utility of peripheral-
iron indices for bone marrow iron status, ROC
curves were plotted using bone marrow iron
depletion as the reference. In the entire group,
there were similar moderate relationships between
bone marrow iron deficiency and peripheral-iron
indices. Compared with the indeterminate value
of 0.5, AUROCs were significantly higher for
TSAT and ferritin levels: 0.75 (95% CI, 0.66-
0.85) and 0.76 (95% CI, 0.66-0.85), respectively
(Fig 3).
To obtain estimations of clinically relevant
thresholds for TSAT, ferritin, and their combina-
tions, ORs of a positive index test, sensitivity,
and specificity were calculated at different cutoff
levels (Table 2). As listed in Table 2, the tradeoff
between good sensitivity and specificity is evi-
dent. Several cutoff values for TSAT and fer-
ritin yielded good or even excellent specificity,Iron Indices in Anemia of Chronic Kidney Disease 641

but sensitivity generally was poor. Only fer-
ritin level /H11021175 ng/mL provided both sensitiv-
ity (0.71; 95% CI, 0.56-0.83) and specificity
(0.71; 95% CI, 0.57-0.83) that were statistically
distinct from the null value, equivalent to ran-
dom chance (Table 2).
Inflammation had a low influence on both
bone marrow iron and peripheral-iron indices
because no relations were found among serum
CRP levels and these parameters, and iron indi-
ces’ predictive values for iron stores were similar
when arbitrary selected CRP cutoff values of 5and 10 mg/L were used for stratification (data not
shown).
Bone Marrow Iron and Erythropoietic Response
to IV Iron Therapy
One month after IV administration of 1,000 mg
of iron sucrose, 49% of patients had a /H113501 g/dL
increase in hemoglobin level and 26% attained a
hemoglobin level of 11 g/dL. Average increase in
hemoglobin levels was 1.0 /H110060.4 g/dL (Table 1).
Comparing iron-depleted versus iron-replete
patients in erythropoietic response, proportionsTable 1. Patient Characteristics According to Bone Marrow Iron
CharacteristicAll
(N/H11549100)Iron Depleted
(n/H1154948)Iron Replete
(n/H1154952) P(test)a
Age
Mean/H11006SD (y) 62/H1100613 59.6 /H1100613.7 63.5 /H1100611.9 0.1 (t)
/H1102260 y (%) 55 46 63 0.02 (/H92732)
Men (%) 55 52 58 0.6 (/H92732)
Primary renal disease (%)
Glomerulonephritis 36 33 40 0.5 (/H92732)
Interstitial nephropathies 10 10 17 0.3 (/H92732)
ADPKD 8 8 8 0.8 (/H92732)
Vascular disease 43 45 31 0.1 (/H92732)
Diabetic nephropathy 3 4 4 0.7 (/H92732)
eGFR (mL/min/1.73 m2) 14 [10] 15.5 [13.6] 14.0 [14] 0.8 (Mann-Whitney)
CKD stage (%)
3 18 10 21 0.1 (/H92732)
4 28 44 23 0.02 (/H92732)
5 56 46 56 0.2 (/H92732)
Body mass index
Mean/H11006SD (kg/m2) 24/H110064 23.6 /H110063.7 25.2 /H110064.4 0.0 (t)
18.5-25 kg/m2(%) 46 46 44 0.6 ( /H92732)
SGA score A (%) 65 62 67 0.7 (/H92732)
Serum albumin
Mean/H11006SD (g/dL) 3.7/H110060.5 3.6 /H110060.5 3.7/H110060.5 0.4 (t)
/H110214 g/dL (%) 75 77 73 0.6 (/H92732)
C-Reactive protein
Median [IQR] (in mg/dL) 15 [24] 15 [24] 15 [22] 0.8 (Mann-Whitney)
/H1102210 mg/L (%) 60 62 60 0.9 (/H92732)
Hemoglobin
Mean/H11006SD (g/dL) 8.99/H110061.0 8.74 /H110061.1 9.22/H110060.9 0.02 (t)
/H1102110 g/dL (%) 70 75 68 0.2 (/H92732)
TSAT
Median [IQR] (in %) 23 [17] 19 [15] 28 [12] 0.001 (Mann-Whitney)
/H1102120% (%) 67 50 17 /H110210.001 ( /H92732)
Serum ferritin
Median [IQR] (in ng/mL) 176 [221] 100 [131] 220 [213] /H110210.001 (Mann-Whitney)
/H11021100 ng/mL (%) 31 48 15 /H110210.001 ( /H92732)
TSAT/H1102120% and ferritin /H11021100 ng/mL (%) 17 33 2 /H110210.001 ( /H92732)
Erythropoietic response to iron
/H9004Hemoglobin /H113501 g/dL (%) 49 65 35 /H110210.001 (McNemar)
Mean/H9004hemoglobin /H11006SD (g/dL) 1.0/H110060.4 1.2 /H110060.4 0.8/H110060.3 /H110210.001 (t)
Hemoglobin /H1102211 g/dL (%) 26 31 21 /H110210.001 (McNemar)
Abbreviations: ADPKD, autosomal dominant polycystic kidney disease; CKD, chronic kidney disease; eGFR, estimated
glomerular filtration rate; IQR, interquartile range; SD, standard deviation; SGA, Subjective Global Assessment; TSAT,
transferrin saturation index.
aIron depleted versus iron replete.Stancu et al 642

of patients with either a /H113501 g/dL increase in
hemoglobin level (65% vs 35%) or reaching a
hemoglobin level /H1135011 g/dL (31% vs 21%), as
well as average increases in hemoglobin levels
(1.2/H110060.4 vs 0.8 /H110060.3 g/dL) were higher in
iron-depleted patients (Table 1). The AUROC of
an increase in hemoglobin concentration after
iron administration using bone marrow iron (de-
plete vs replete) as a reference test was similar to
AUROCs of the 2 peripheral-iron indices: 0.74
(95% CI, 0.65-0.84; Fig 3). ORs for a 1.5-, 1.0-,
and 0.5-g/dL increase in hemoglobin level were
70.00 (95% CI, 12.16-2,803.89), 8.60 (95% CI,
3.41-27.82), and 0.33 (95% CI, 0.09-0.97) in
iron-depleted versus iron-replete patients, respec-
tively. A 1.0-g/dL increase in hemoglobin level
had 65% sensitivity (95% CI, 49-78) and 65%
specificity (95% CI, 51-78) in correctly describ-
ing bone marrow iron depletion (Table 3).
Inflammation did not seem to influence the
erythropoietic response because proportions of
patients with CRP level /H1102210 mg/L were similar
between responders and nonresponders (62% vs
60%; OR, 0.61; 95% CI, 0.33-1.17; P/H110050.1).
In logistic regression analysis, a model in-
cluding age, sex, initial hemoglobin level,
TSAT and ferritin levels, and assessment ver-
sus baseline difference in hemoglobin levels,
CRP levels, and eGFR explained 34% of the
variance in bone marrow iron. Only age, initial
hemoglobin level, TSAT, and the difference in
hemoglobin levels made significant contribu-
tions (Table 4).DISCUSSION
This is one of the largest studies comparing
the diagnostic utility of peripheral-iron indices
and erythropoietic response to IV iron therapy
for iron stores using bone marrow iron as refer-
ence in non–dialysis-dependent patients with
CKD. The study confirmed the high prevalence
of iron-depleted bone marrow stores in even
non–dialysis-dependent patients with CKD, and
the important proportion of responders to iron
infusion. However, erythropoietic response did
not perform better than peripheral-iron tests as
an indicator of decreased bone marrow iron
stores.
Few data are available about iron balance and
the prevalence of iron deficiency in non–dialysis-
dependent patients with CKD.1The reported
prevalence of bone marrow iron deficiency in
non–dialysis-dependent patients with CKD var-
ies widely from /H1102290%7to 40%4and even 23%.8
In our data, using semiquantitative assessment,
48% of patients had both iron-depleted bone
marrow macrophages and erythroblasts deprived
of siderotic granules, highly suggestive of iron
deficiency.
The reported iron overload prevalence in dialy-
sis patients with CKD on IV iron and ESA
Figure 3. Sensitivity and specificity of transferrin satu-
ration index (TSAT), serum ferritin (sFerritin) level, and
increase in hemoglobin level (/H9004Hb) to detect depletion of
bone marrow iron stores (denoted by BM iron /H110050) in 100
non–dialysis-dependent patients with chronic kidney dis-
ease assessed using receiver operating characteristic
curves. Areas under the curves do not differ.
Figure 2. Results of bone marrow examination in 100
anemic non–dialysis-dependent patients with chronic kid-
ney disease.Iron Indices in Anemia of Chronic Kidney Disease 643

therapy varied from 3% (bone marrow iron)27to
66% (liver iron stores)28and is a matter of
debate.29Increased bone marrow iron was seen
in 18% of our iron- and ESA-naive anemic
patients. How can bone marrow iron overload be
explained in these anemic iron-naive patients?
Inflammation is one possibility because 60% of
our patients had a serum CRP level /H1102210 mg/L,
but only 3% had a bone marrow pattern sugges-
tive of anemia of inflammation. Hepcidin inter-
vention is another possibility. Increased hepcidin
levels are found in both patients with inflamma-
tion and patients with CKD, and hepcidin is
believed to “lock” the iron in the reticuloendothe-
lial system and decrease iron absorption.30,31
Most of our patients were iron depleted, but
abnormal bone marrow iron distribution sugges-
tive of iron lock was noticed in only a minority
of cases despite the impressive prevalence of
inflammation. Therefore, hepcidin effect on iron
absorption seems to prevail its effect on iron
distribution.
Ferritin level and TSAT are widely used to
evaluate iron status in hemodialysis patients;1,2however, these tests are not highly accurate in
predicting depletion of iron stores.3Examining
relationships between bone marrow iron and
peripheral indices in peritoneal dialysis patients,
Fernández-Rodríguez et al27found the highest
AUROC for ferritin and computed specificity
and sensitivity of 75% at a cutoff value of 121
ng/dL. More recently, Rocha et al32reported a
close correlation between ferritin level and bone
marrow iron evaluated using histomorphometric
analyses in hemodialysis patients. However, cut-
off values were not assessed.32
However, in patients with non–dialysis-depen-
dent CKD without ESA therapy, the body of
evidence is insufficient to guide anything other
than opinion with regard to the interpretation of
iron test results.1,2For example, using KDOQI
criteria, such as TSAT /H1102120% or serum ferritin
level/H11021100 ng/mL, Fishbane et al33found abnor-
mal iron indices in /H1101160% of participants with
CKD in the National Health and Nutrition Exami-
nation Survey (NHANES) 1988-2004, a very
high prevalence that made the utility of these
criteria doubtful.Table 2. Diagnostic Utility of TSAT and Serum Ferritin for Detecting Bone Marrow Iron Depletion Using Different
Cutoff Values
Parameter Odds Ratio (95% confidence interval)aSensitivity Specificity
TSAT/H1102115% 2.30 (1.33-4.10) 0.42 (0.28-0.57) 0.88 (0.77-0.96)
TSAT/H1102120% 1.79 (1.06-3.09) 0.50 (0.35-0.65) 0.83 (0.70-0.92)
TSAT/H1102125% 0.91 (0.54-1.53)b0.71 (0.56-0.83) 0.60 (0.45-0.73)
Ferritin /H1102175 ng/mL 2.47 (1.42-4.46) 0.40 (0.26-0.55) 0.90 (0.79-0.97)
Ferritin /H11021100 ng/mL 1.91 (1.13-3.32) 0.48 (0.33-0.63) 0.85 (0.72-0.93)
Ferritin /H11021175 ng/mL 1.09 (0.66-1.79)b0.71 (0.56-0.83) 0.71 (0.57-0.83)
TSAT/H1102115% & ferritin /H1102175 ng/mL 3.92 (2.10-7.86) 0.27 (0.15-0.42) 0.98 (0.90-1.00)
TSAT/H1102120% & ferritin /H11021100 ng/mL 3.19 (1.79-5.99) 0.33 (0.20-0.48) 0.98 (0.90-1.00)
TSAT/H1102125% & ferritin /H11021175 ng/mL 1.87 (1.10-3.25) 0.48 (0.33-0.63) 0.83 (0.70-0.92)
Abbreviation: TSAT, transferrin saturation index.
aMcNemar test to evaluate the effect of different cutoff values and combination of positive index test results on the correct
identification of bone marrow iron (depleted vs replete).
bP/H110220.5.
Table 3. Relationship Between Increase in Hemoglobin After Intravenous Iron Therapy and Bone Marrow Iron Using
Different Cutoff Values
Parameter Odds Ratio (95% confidence interval)aSensitivity Specificity
1.5 g/dL 1hemoglobin 70.00 (12.16-2,803.89) 0.19 (0.09-0.33) 0.96 (0.87-0.99)
1.0 g/dL 1hemoglobin 8.60 (3.41-27.82) 0.65 (0.49-0.78) 0.65 (0.51-0.78)
0.5 g/dL 1hemoglobin 0.33 (0.09-0.97) 1.00 (0.93-1.00) 0.15 (0.07-0.28)
aMcNemar test to evaluate the effect of different cutoff values of positive index test results on the correct identification of
bone marrow iron (depleted vs replete).Stancu et al 644

In our data, only 17% of anemic patients had
indices suggestive of iron deficiency; however,
TSAT and ferritin level seemed to reasonably
predict bone marrow iron status because in ROC
analysis, TSAT and ferritin AUROCs were
/H1101175%. However, examining various cutoff val-
ues, almost none of the selected cutoff values for
TSAT, ferritin, and their combination rendered
both significant sensitivity and specificity. Only
ferritin at a cutoff value of 175 ng/mL yielded
significant sensitivity and specificity, both esti-
mated at 71%, which are moderate test character-
istics to inform a clinical decision.
In an epidemiologic-based analysis of the rela-
tionship between iron status tests and hemoglo-
bin level difference from a reference value, a
threshold of 25 ng/mL was found for only fer-
ritin, and a linear continuous relationship was
found in the case of TSAT.9Moreover, our re-
sults are similar to data presented by Kalantar-
Zadeh et al,5in which semiquantitative bone
marrow iron assessment also was used as refer-
ence. One explanation of these discrepancies
could be the skewed distribution of our cohort
toward advanced stages of CKD (65% in CKD
stage 5), unlike the distribution reported in
NHANES-based analysis (14.5% in CKD stage
5). Recently, in a cross-sectional study of a large
database, the prevalence of inflammation, de-
fined as CRP level /H110223 mg/L, increased from
55% to 74% in patients with CKD stage 4-5,34
close to levels observed in our patients (60%).
Because ferritin also is an acute-phase reac-
tant,6,35,36inflammation could set the ferritin
level cutoff at higher values in our cohort.
In our data, only TSAT had some predictive
value for iron-replete bone marrow (10% in-
crease/unit). Recently, high TSATs were associ-
ated with lower mortality and more rapid de-crease in renal function in non–dialysis-dependent
patients with CKD.37Clearly, studies with ad-
equate power are needed to elucidate the appro-
priate iron index ranges for safe and efficient iron
therapy.
Because of these uncertainties, some have
suggested administration of iron irrespective of
bone marrow iron status. After IV iron sucrose
administration, a significant hematologic re-
sponse was noted in the entire cohort: 49% of
patients had a /H110221 g/dL increase in hemoglobin
level, and 26% attained a hemoglobin level of 11
g/dL, which is in accord with previous studies
reporting that 30%-50% of nondialysis depen-
dent CKD patients reached the target hemoglo-
bin level after IV iron therapy only7,11,12and
with more recent studies in which IV ferumoxy-
tol was used.11,14-17In our data, erythropoietic
response depended on iron stores because propor-
tions of patients with a /H110221 g/dL increase in
hemoglobin level (65% vs 35%) and a final
hemoglobin level /H1135011 g/dL (31% vs 21%) were
higher in those with depleted bone marrow iron
stores. Furthermore, the average increase in he-
moglobin level was higher in iron-depleted pa-
tients (1.2 /H110060.4 vs 0.8 /H110060.3 g/dL).
Unfortunately, the erythropoietic response and
peripheral-iron indices had similar AUROCs, a
finding that suggests their equivalent and only
moderate utility in identifying patients with de-
pleted iron stores. Moreover, both iron-depleted
patients responding to iron therapy and patients
with replete iron who do not respond were identi-
fied correctly in 65% of cases, which is similar to
values for iron indices. This is in line with results
of multivariable analysis, in which higher initial
hemoglobin and TSAT values and a lower in-
crease in hemoglobin level were more likely
associated with replete iron stores. Thus, in thisTable 4. Relationship Between the Index Test (bone marrow iron) and Reference Tests in a Model of Multivariable Binary
Logistic Regression Analysis
Variablesa/H9252 Standard Error Wald df P e/H9252(95% confidence interval)
Age (y) 0.05 0.02 5.40 1 0.02 1.05 (1.01-1.09)
Hemoglobin (g/dL) 0.70 0.28 6.45 1 0.01 2.01 (1.17-3.45)
Transferrin saturation (%) 0.01 0.03 10.40 1 0.001 1.10 (1.04-1.17)
/H9004Hemoglobin (g/dL) /H110022.00 0.75 7.06 1 0.008 0.14 (0.03-0.59)
Constant /H110029.38 3.38 7.71 1 0.005 0.00
Note: Cox and Snell R2/H110050.34;/H92732/H1100541.42; P/H110210.001.
aVariable(s) entered in step 1: sex, age, initial hemoglobin level, C-reactive protein level, transferrin saturation index,
serum ferritin level, glomerular filtration rate, and change in hemoglobin level (assessment vs baseline difference).Iron Indices in Anemia of Chronic Kidney Disease 645

study, response to iron perfusion was not a highly
reliable indicator of decreased iron stores and
performed similarly to peripheral-iron indices.
There are some limits of our study. Assess-
ment of iron using staining of bone marrow was
made on an aspirate, not a biopsy specimen, and
some blood dilution could interfere with the
result. Additionally, the accuracy of the micro-
scopic evaluation had been questioned at both
ends of the classic grading scale because results
depend critically on the number of particles
reviewed.14Precautions to overcome these prob-
lems were taken: a minimum of 9 particles were
reviewed in each case. Nevertheless, the robust-
ness of our data is supported by the significant
relationships found between bone marrow iron
and both peripheral-iron indices and erythropoi-
etic response. Although the number of partici-
pants is one of the largest reported, it still is too
small for final conclusions, the proportion of
patients with diabetes was low, and the small
number of patients with iron overload did not
allow a better definition of this group. Moreover,
most patients were in stage 5 CKD, anemia was
relatively severe, and inflammation was highly
prevalent. Although the above mentioned charac-
teristics may influence the results, these charac-
teristics are closer to those of patients seen in
day-to-day practice.
In conclusion, in a cohort of nonselected ane-
mic non–dialysis-dependent patients in advanced
CKD stages with a high prevalence of inflamma-
tion, bone marrow iron stores were depleted in
about half. Peripheral-iron indices and erythropoi-
etic response to an iron perfusion had equivalent,
but limited, usefulness in identifying bone mar-
row iron store status. Use of these indices to
indicate depletion of iron stores should be recon-
sidered.
ACKNOWLEDGEMENTS
Support: This research was supported in part by a grant
from the Anemia Working Group–Romania.
Financial Disclosure: None.
REFERENCES
1. Fishbane S. Iron management in nondialysis-depen-
dent CKD. Am J Kidney Dis. 2007;49:736-743.
2. National Kidney Foundation. KDOQI Clinical Prac-
tice Guideline and Clinical Practice Recommendations for
Anemia in Chronic Kidney Disease: 2007 Update of hemo-
globin target. Am J Kidney Dis. 2007;50:471-530.3. Fishbane S, Kowalski EA, Imbriano LJ, Maesaka JK.
The evaluation of iron status in hemodialysis patients. JA m
Soc Nephrol. 1996;7:2654-2657.
4. Tessitore N, Solero GP, Lippi G, et al. The role of iron
status markers in predicting response to intravenous iron in
haemodialysis patients on maintenance erythropoietin. Neph-
rol Dial Transplant. 2001;16:1416-1423.
5. Kalantar-Zadeh K, Hoffken B, Wunsch H, Fink H,
Kleiner M, Luft EC. Diagnosis of iron deficiency anemia in
renal failure patients during the post-erythropoietin era.
Am J Kidney Dis. 1995;26:292-299.
6. Kalantar-Zadeh K, Kalantar-Zadeh K, Lee GH. The
fascinating but deceptive ferritin: to measure it or not to
measure it in chronic kidney disease? Clin J Am Soc Neph-
rol. 2006;1(suppl 1):S9-18.
7. Gotloib L, Silverberg D, Shostak A. Iron deficiency is
a very common cause of anemia in chronic kidney insuffi-
ciency and can often be corrected with IV iron. J Nephrol.
2006;19:161-167.
8. Rahman MM, Dutta PK, Hoque M, et al. Evaluation of
iron status by bone marrow iron stain and correlation with
serum profile in chronic kidney disease (CKD). J Bang-
ladesh Coll Physicians Surg. 2007;25:117-120.
9. Hsu C-Y , McCulloch CE, Curham GC. Iron status and
hemoglobin level in chronic renal insufficiency. J Am Soc
Nephrol. 2002;13:2783-2786.
10. Van Wyck DB, Roppolo M, Martinez CO, Mazey
RM, McMurray S. A randomized, controlled trial comparing
IV iron sucrose to oral iron in anemic patients with nondialy-
sis-dependent CKD. Kidney Int. 2005;68:2846-2856.
11. Qunibi CW, Bailie G. Comparison of intravenous
iron sucrose to oral iron in the treatment of anemic patients
with chronic kidney disease not on dialysis. Nephron Clin
Pract. 2005;100:c55–62.
12. Silverberg DS, Iaina A, Peer G, et al. Intravenous iron
supplementation for the treatment of the anemia of moderate
to severe chronic renal failure patients not receiving dialysis.
Am J Kidney Dis. 1996;27:234-238.
13. Mircescu G, Gârneat ¸a˘L ,C a ˘pus¸a˘ C, Ursea N. Intrave-
nous iron supplementation for the treatment of anemia in
pre-dialyzed chronic renal failure patients. Nephrol Dial
Transplant. 2006;21:120-124.
14. Spinowitz BS, Kausz AT, Baptista J, et al. Ferumoxy-
tol for treating iron deficiency anemia in CKD. J Am Soc
Nephrol. 2008;19:1599-1605.
15. Blaustein DA, Schwenk MH, Chattopadhyay J, et al.
The safety and efficacy of an accelerated iron sucrose dosing
regimen in patients with chronic kidney disease. Kidney Int
Suppl. 2003;87:S72-77.
16. Provenzano R, Schiller B, Rao M, Coyne D, Brenner
L, Pereira BJG. Ferumoxytol as an intravenous iron replace-
ment therapy in hemodialysis patients. Clin J Am Soc Neph-
rol. 2009;4:386-393.
17. Han L, Wang SX. Intravenous iron sucrose in Chi-
nese hemodialysis patients with renal anemia. Blood Purif.
2008;26:151-156.
18. Levey AS, Greene T, Kusek JW, et al. A simplified
equation to predict glomerular filtration rate from serum
creatinine [abstract]. J Am Soc Nephrol. 2000;11:0828A.
19. Canada-USA (CANUSA) Peritoneal Dialysis Study
Group. Adequacy of dialysis and nutrition in continuousStancu et al 646

peritoneal dialysis: association with clinical outcomes. JA m
Soc Nephrol. 1996;7:198-207.
20. Hughes DA, Stuart-Smith SE, Bain BJ. How should
stainable iron in bone marrow films be assessed? J Clin
Pathol. 2004;57:1038-1040.
21. Rath CE, Finch CA. Sternal marrow hemosiderin: a
method for the determination of available iron stores in man.
J Lab Clin Med. 1948;33:81-86.
22. Hansen H, Weinfeld A. Hemosiderin estimations and
sideroblasts counts in the differential diagnosis of iron
deficiency and other anemias. Acta Med Scand. 1959;165:
333-356.
23. Altman DG, Bland JM. Statistics notes. Diagnostic
tests 1: sensitivity and specificity. BMJ. 1994;308:1552.
24. Altman DG, Bland JM. Statistics notes. Diagnostic
tests 2: predictive values. BMJ. 1994;309:102.
25. Altman DG, Bland JM. Statistics notes. Diagnostic
tests 3: receiver-operating characteristics plots. BMJ. 1994;
309:188.
26. Zweig MH, Campbell G. Receiver-operating charac-
teristic (ROC) plots: a fundamental evaluation tool in clini-
cal medicine. Clin Chem. 1993;39:561-577.
27. Fernández-Rodríguez AM, Guindeo-Casasús MC,
Molero-Labarta T, et al. Diagnosis of iron deficiency in
chronic renal failure. Am J Kidney Dis. 1999;34:508-513.
28. Canavese C, Bergamo D, Ciccone G, et al. Low-dose
continuous iron therapy leads to a positive iron balance and
decreased serum transferrin levels in chronic haemodialysis
patients. Nephrol Dial Transplant. 2004;19:1564-1570.
29. Hörl WH. Adjunctive therapy in anaemia manage-
ment. Nephrol Dial Transplant. 2002;17:56-59.30. Malyszko J, Malyszko JS, Pawlak K, et al. Hepcidin,
iron status, and renal function in chronic renal failure,
kidney transplantation and hemodialysis. Am J Hematol.
2006;81:832-837.
31. Swinkels DW, Wetzels JFM. Hepcidin: a new tool
in the management of anaemia in patients with chronic
kidney disease? Nephrol Dial Transplant . 2008;23:2450-
2453.
32. Rocha LA, Barreto DV , Barreto FC, et al. Serum
ferritin level remains a reliable marker of bone marrow iron
stores evaluated by histomorphometry in hemodialysis pa-
tients. Clin J Am Soc Nephrol. 2009;4:57-61.
33. Fishbane S, Pollack S, Feldman HI, Joffe MM. Iron
indices in chronic kidney disease in the National Health and
Nutritional Examination Survey 1988-2004. Clin J Am Soc
Nephrol. 2009;4:57-61.
34. Chonchol M, Lippi G, Montagnana M, Muggeo M,
Targher G. Association of inflammation with anaemia in
patients with chronic kidney disease not requiring chronic
dialysis. Nephrol Dial Transplant. 2008;23:2879-2883.
35. Kalantar-Zadeh K, Rodriguez RA, Humphreys MH.
Association between serum ferritin and measures of inflam-
mation, nutrition and iron in haemodialysis patients. Neph-
rol Dial Transplant. 2004;19:141-149.
36. Ford BA, Coyne DW, Eby CS, Scott MG. Variabil-
ity of ferritin measurements in chronic kidney disease;
implications for iron management Kidney Int. 2008;75:
104-110.
37. Kovesdy CP, Estrada W, Shahram A, Kalantar-Zadeh.
Association of markers of iron stores with outcomes in
patients with nondialysis-dependent chronic kidney disease.
Clin J Am Soc Nephrol. 2009;4:435-441.Iron Indices in Anemia of Chronic Kidney Disease 647