Kidney International, Vol. 66 (2004), pp. 144–156
`
`Parenteral iron nephrotoxicity: Potential mechanisms
`and consequences1
`
`RICHARD A. ZAGER, ALI C.M. JOHNSON, and SHERRY Y. HANSON
`
`Department of Medicine, University of Washington, and the Fred Hutchinson Cancer Research Center, Seattle, Washington
`
`Parenteral iron nephrotoxicity: Potential mechanisms and con-
`sequences.
`Background. Parenteral iron administration is a mainstay of
`anemia management in renal disease patients. However, con-
`cerns of potential iron toxicity persist. Thus, this study was
`conducted to more fully gauge iron toxicologic profiles and po-
`tential determinants thereof.
`Methods. Isolated mouse proximal tubule segments (PTS)
`or cultured proximal tubular [human kidney (HK-2)] cells
`were exposed to four representative iron preparations [iron
`sucrose (FeS), iron dextran (FeD), iron gluconate (FeG), or
`iron oligosaccharide (FeOS)] over a broad dosage range (0, 30
`to 1000 lg iron/mL). Cell injury was assessed by lactate dey-
`hdrogenase (LDH) release, adenosine triphosphate (ATP) re-
`ductions, cell cytochrome c efflux, and/or electron microscopy.
`In vivo toxicity (after 2 mg intravenous iron injections) was
`assessed by plasma/renal/cardiac lipid peroxidation [malondi-
`aldehyde (MDA)], renal ferritin (protein)/heme oxygenase-1
`(HO-1) (mRNA) expression, electron microscopy, or postiron
`injection PTS susceptibility to attack.
`Results. In each test, iron evoked in vitro toxicity, but up to
`30× differences in severity (e.g., ATP declines) were observed
`(FeS > FeG > FeD = FeOS). The in vitro differences paral-
`leled degrees of cell (HK-2) iron uptake. In vivo correlates of
`iron toxicity included variable increases in renal MDA, ferritin,
`and HO-1 mRNA levels. Again, these changes appeared to par-
`allel in vivo (glomerular) iron uptake (seen with FeS and FeG,
`but not with FeD or FeOS). Iron also effected in vivo alter-
`ations in proximal tubule cell homeostasis, as reflected by the
`“downstream” emergence of tubule resistance to in vitro oxi-
`dant attack.
`Conclusion. Parenteral iron formulations have potent, but
`highly variable, cytotoxic potentials which appear to parallel
`degrees of cell iron uptake (FeS > FeG (cid:4) FeD or FeOS). That
`in vitro injury can be expressed at clinically relevant iron con-
`centrations, and that in vivo glomerular iron deposition/injury
`may result, suggest caution is warranted if these agents are to
`be administered to patients with active renal disease.
`
`1See Editorial by Alam et al, p. 457.
`
`Key words: iron dextran, iron sucrose, iron gluconate, oxidant stress,
`heme oxygenase 1.
`
`Received for publication December 23, 2003
`and in revised form January 29, 2004
`Accepted for publication February 11, 2004
`
`C(cid:1) 2004 by the International Society of Nephrology
`
`Administration of parenteral iron has become a main-
`stay for treating anemia in patients with end-stage
`renal disease (ESRD). This practice is required in
`order to offset dialysis-related blood (iron) loss, and the
`need to optimize hematopoietic responsiveness to exoge-
`nous erythropoeitin (Epo) therapy [1–3]. While generally
`regarded as safe, anaphylactic/oid reactions have been
`noted following intravenous iron injection, most com-
`monly but not exclusively, with dextran preparations [4].
`In addition to allergic reactions, each currently employed
`parenteral iron formulation [e.g., iron dextran (FeD), iron
`sucrose (FeS), and iron gluconate (FeG)] has the poten-
`tial to induce oxidative stress [5–9]. For example, when
`administered intravenously, these agents may induce
`free radical generation [10] and lipid peroxidation [5],
`processes which can induce acute endothelial dysfunc-
`tion (e.g., as denoted by perturbed forearm endothelial-
`dependent vasodilation) [10]. Additional support for the
`concept of iron-induced toxicity comes from a recent re-
`port [11] which indicates that clinically achievable con-
`centrations of FeG or FeS can impair polymorphonuclear
`cell (PMN)/transendothelial migration. This could con-
`tribute to infectious complications in dialysis patients.
`While the above evidence suggests potential acute tox-
`icities, the long-term consequences of parenteral iron
`administration remain largely unknown. In this regard,
`it is noteworthy that iron-mediated oxidative stress can
`contribute to both atherogenesis [12–17] and chronic in-
`flammation [18–22], each of which are leading causes of
`morbidity and mortality in ESRD patients [23–25]. Fur-
`thermore, because parenterally administered iron has
`direct glomerular, and as well as tubular access (via per-
`itubular capillaries), it is conceivable that it might con-
`tribute to glomerular and/or tubulointerstitial disease
`progression [26–29]. That intravenous iron + Epo ther-
`apy is currently being administered to pre-ESRD patients
`underscores these concerns.
`Given that parenteral iron therapy is likely to remain
`an integral component of renal disease patient man-
`agement, it is imperative to better define its potential
`cytotoxic effects, and to ascertain whether different toxi-
`city profiles exist amongst currently employed parenteral
`
`144
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1091 - Page 1
`
`

`

`Zager et al: Parenteral iron nephrotoxicity
`
`145
`
`iron formulations. Indeed, in a recent study performed in
`this laboratory using suprapharmacologic iron doses [5], a
`clear gradation of toxicity was apparent amongst four test
`agents [from most to least severe: FeS(cid:4) FeG(cid:4) FeD= Fe
`oligosaccharide (FeOS)]. However, the reason(s) for this
`differential in vitro toxicity, if it might be observed with
`more clinically relevant iron concentrations, whether in
`vivo toxicologic correlates exist, and the nature of un-
`derlying pathogenic mechanisms were not well defined.
`Hence, the present study was undertaken utilizing a num-
`ber of experimental models (freshly isolated mouse prox-
`imal tubules, cultured human proximal tubular cells, and
`in vivo mouse experiments) to gain additional insights.
`
`METHODS
`Proximal tubule segment (PTS) experiments
`Preparation of isolated mouse proximal tubules. Prox-
`imal tubules were isolated from normal CD-1 male mice
`(25 to 35 g) (Charles River, Wilmington, MA, USA), as
`previously described [30]. In brief, the mice were deeply
`anesthetized with pentobarbital (4 to 5 mg intraperi-
`toneally), and the kidneys were resected through a mid-
`line abdominal incision. They were iced, the cortices were
`dissected, and the tissues were subjected to collagenase
`digestion. The tissues were passed through a stainless
`steel mesh, and then viable PTS were collected after
`pelleting through 32% Percoll [30, 31]. The recovered
`tubules were suspended in an experimentation buffer
`consisting of (in mmol/L): NaCl, 100; KCl, 2.1; NaHCO3,
`25; KH2PO4, 2.4; MgSO4 1.2; MgCl2, 1.2; CaCl2, 1.2;
`glucose, 5; alanine, 1; Na lactate, 4; Na butyrate, 10;
`36 kD dextran, 0.6%; and gassed with 95% O2/5% CO2,
`pH 7.44). The final tubule protein concentration was ∼2
`◦
`C
`to 4 mg/mL. Each PTS preparation was rewarmed to 37
`in a heated shaking water bath and divided into four to
`six equal aliquots (1.25 mL) in 10 mL Erlenmyer flasks,
`depending on the needs of individual experiments (see
`below).
`Comparative effects of iron preparations on proxi-
`mal tubule adenosine triphosphate dehydrogenase (ATP)
`content.
`Dose response experiments. The purpose of this
`study was to compare dose-response toxicity effects
`of four test iron preparations. Given that previous
`studies demonstrated that mitochondrial dysfunction,
`as assessed by reductions in tubule ATP production,
`is a sensitive marker of iron toxicity [5, 31], tubule
`ATP concentrations, as well as lethal cell injury [% lac-
`tate dehydrogenase (LDH) release], were chosen as test
`biologic end points. Twelve individual sets of PTS were
`prepared, each was divided into five equal aliquots,
`and these were incubated ×30 minutes in a 37
`◦
`C shak-
`ing water bath in the presence of 95% O2/5% CO2,
`under the following conditions: (1) control incubation;
`
`(2) 1000 lg/mL iron addition; (3) 500 lg/mL iron addi-
`tion; (4) 250 lg/mL iron addition; and (5) 125 lg/mL iron
`addition. Each individual tubule preparation was used to
`test one of the four iron preparations: (1) FeS (Venofer)
`(American Regent, Shirley, NY, USA);
`(2) FeD
`(INFeD) (Watson Pharmaceuticals, Morristown, NJ,
`USA); (3) FeG (Ferrlecit) (Watson Pharmaceuticals);
`and (4) FeOS, an iron preparation currently in clinical
`trials (Pharmacosmos, Copenhagen, DK). In all, each of
`these preparations were tested in three separate dose-
`response experiments. After completing the 30-minute
`incubations, a sample of each aliquot was removed, ade-
`nine nucleotides were extracted in trichloroacetic acid,
`and then the samples were analyzed for ATP by high-
`performance liquid chromatography (HPLC) (nmol/mg
`tubule protein) [32]. An aliquot of each tubule suspension
`was also used to determine % LDH release.
`pH control experiment. Because FeS stock solution
`has a pH of approximately 10.5, and because a high pH
`can induce cytotoxicity [33], a control for the above FeS
`additions was conducted with an equivalent amount of
`sucrose (300 mg/mL) with its pH adjusted to 10.5 by
`1 N NaOH addition. Aliquots from four sets of tubules
`were incubated either under control conditions or with
`62.5 lL of the alkaline sucrose solution (equivalent
`to the volume of the 1000 lg/mL FeS dosage). After
`30-minute incubations, ATP concentrations and % LDH
`release were assessed.
`Effects of low dose FeS and FeG on tubule ATP
`concentrations. The above dose-titration experiments
`indicated that FeS and FeG had the greatest suppres-
`sive effects on tubule ATP concentrations, with reduc-
`tions being apparent at the lowest test concentration
`(125 lg iron/mL) (see Results section). The following ex-
`periment ascertained whether ATP reductions could be
`induced by even lower iron concentrations (i.e., within
`clinically achievable plasma iron concentrations). Four
`sets of tubules were prepared, each was divided into
`five equal aliquots, and incubated ×30 minutes as fol-
`lows: group 1, control conditions; groups 2 and 3, with
`30 or 60 lg/mL FeS iron; and groups 4 and 5, with 30 or
`60 lg/mL FeG iron. ATP levels and LDH release were
`then assessed.
`
`In vivo mouse experiments
`Assessment of lipid peroxidation following intravenous
`iron treatment. The following experiments were under-
`taken to ascertain the relative degrees of lipid peroxida-
`tion induced by three representative test iron compounds:
`FeD, FeS, and FeG. These three compounds were selected
`because they manifested the greatest differential toxic-
`ity in the above described proximal tubule experiments
`(see Results section). Mice (N = 18) were placed in non-
`traumatic restraining cages, and they were injected via
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1091 - Page 2
`
`

`

`146
`
`Genes
`
`Mouse
`GAPDH
`Mouse
`HO-1
`
`Zager et al: Parenteral iron nephrotoxicity
`
`Table 1. Mouse primers for quantitating heme oxygenase 1 (HO-1) mRNA in renal cortex
`
`Primer sequences
`
`(cid:8)
`5
`(cid:8)
`5
`(cid:8)
`5
`(cid:8)
`5
`
`-CTG CCA TTT GCA GTG GCA AAG TGG-3
`(cid:8)
`-TTG TCA TGG ATG ACC TTG GCC AGG-3
`(cid:8)
`-AAC ACA AAG ACC AGA GTC CCT CAC-3
`-CAA GAG AAG AGA GCC AGG CAA GAT-3
`
`(cid:8)
`
`(cid:8)
`
`Polymerase chain reaction conditions
`◦
`◦
`C – 45 sec;
`94
`C – 45 sec; 57
`◦
`72
`C – 45 sec; 28 cycles
`◦
`◦
`94
`C – 45 sec; 57
`C – 45 sec;
`◦
`72
`C – 45 sec; 28 cycles
`
`Product size
`
`437 bp
`
`288 bp
`
`Primer sequences used for quantitating HO-1 mRNA in mouse renal cortex 4 hours following intravenous iron treatment (see text). Glyeraldehyde-3-phosphate
`dehydrogenase (GAPDH) was quantified as a “housekeeping gene.”
`
`the tail vein with either 2 mg of iron (N = 4 of each of the
`above iron preparations), or with a sham tail vein saline
`injection (N = 6). The mice were then released from
`the restrainers and, 90 minutes later, they were deeply
`anesthetized with pentobarbital, as above. The abdomi-
`nal cavities were opened, a plasma sample was obtained
`from the inferior vena cava, and then one kidney per
`animal was resected. The thorax was opened and the
`heart was removed. The tissues were placed on an iced
`plate. A piece of renal cortex and of cardiac apex were
`resected, the tissues rinsed in iced saline to remove con-
`taminating blood, and then ∼75 mg of renal cortex or
`heart tissue were homogenized in 1 mL of iced phosphate-
`buffered saline (PBS) containing 25 mmol/L desferriox-
`amine (DFO) to chelate any free iron which may have
`been generated during this process. Samples of tissue ho-
`mogenates (200 lL) were then assayed for malondialde-
`hyde (MDA) concentrations by the thiobarbituric acid
`method [34]. Tissue MDA concentrations were expressed
`as nmol/mg tissue protein. Plasma samples (200 lL), to
`which 25 mmol/L DFO was added, were also assayed for
`MDA with values being expressed as nmol/mL.
`Parenteral
`iron effects on renal ferritin and heme
`oxygenase-1 (HO-1) expression. The following experi-
`ments were conducted to ascertain whether, and to what
`degree, the four test iron preparations impact renal corti-
`cal homeostasis, as assessed by the potential induction of
`ferritin and HO-1 proteins (redox-sensitive indicators).
`To this end, mice received every other day tail vein in-
`jections of 2 mg iron, administered as either FeS (N =
`5), FeG (N = 5), INFeD (N = 5), or FeOS (N = 4).
`Each group of mice had their own simultaneous control
`groups which received equal volume tail vein saline in-
`jections. Approximately 24 hours following the last of the
`three injections, the mice were anesthetized with pento-
`barbital, the kidneys were removed, and the cortices were
`dissected on an iced plate.
`Western blotting. The above noted renal cortical tis-
`sue samples were extracted for protein and probed by
`Western blot for ferritin and HO-1, using previously de-
`scribed general methodologies [35]. In the case of ferritin,
`25 lg of protein extract were electrophoresed through
`a 12% Bis-Tris acrylamide Nupage gel (Invitrogen Life
`Technologies, Carlsbad, CA, USA) and probed with goat
`antiferritin antibody (catalog number SC-14416) (Santa
`Cruz Biotechnology, Santa Cruz, CA, USA), according
`
`to manufacturer’s instructions. For HO-1 detection,
`50 lg of protein extract was electrophoresed as described
`above, and probed with rabbit anti-HO-1 antibody (cat-
`alog number SC-10789) (Santa Cruz Biotechnology) as
`the primary antibody as per manufacturer’s instructions.
`Secondary detection of the anti-ferritin and anti-HO-1
`antibodies was performed with either horseradish per-
`oxidase (HRP)-labeled donkey antigoat IgG (catalog
`number SC-2020) (Santa Cruz Biotechnology) for fer-
`ritin or with HRP-labeled donkey antirabbit IgG (cata-
`log number NA 934) (Amersham-Pharmacia, Piscataway,
`NJ, USA) for HO-1. Detection was by enhanced chemi-
`luminescence (ECL Kit) (Amersham-Pharmacia). West-
`ern blot semiquantitative analysis was performed by band
`optical density scanning. Nonspecific secondary antibody
`staining was ruled out by the fact that the secondary anti-
`body, in the absence of the primary antibody, did not iden-
`tify the relevant protein bands (ferritin, ∼25 kD; HO-1,
`∼32 kD). Equal protein loading/transfer was verified by
`India ink staining. A positive control consisted of renal
`cortical protein samples from mice 18 hours postinduc-
`tion of glycerol induced-acute renal failure (which up-
`regulates both HO-1 and ferritin) [36].
`HO-1 mRNA expression following iron treatment.
`Stress induced changes in tissue ferritin concentra-
`tions are largely determined by posttranslational events,
`whereas HO-1 expression is regulated via oxidant stress-
`induced HO-1 gene transcription [37]. Therefore, to gain
`further insights into relative degrees of iron-induced oxi-
`dant stress, mice were injected with either FeD, FeG, FeS,
`or FeOS, as noted above (N = 4 to 6 per group). Controls
`consisted of ten mice subjected to tail vein saline injec-
`tions. Four hours later, the mice were anesthetized with
`pentobarbital, and the kidneys resected. The renal corti-
`cal tissues were immediately placed into TRIzol reagent
`(Invitrogen Life Technologies) and total RNA was ex-
`tracted according to the manufacturer’s instructions. The
`final RNA pellet was brought up in RNase-free water to
`an approximate concentration of 3 mg/mL.
`Reverse transcription (RT) and polymerase chain re-
`action (PCR) were performed using the 1st-Strand Syn-
`thesis Kit for RT-PCR (Ambion Inc., Austin, TX, USA),
`as previously described in detail [38]. The specific primers
`for HO-1 and glyceraldehyde-3-3-phosphate dehydroge-
`nase (GAPDH) were designed with 50% to 60% GC
`composition (see Table 1). The similarity in annealing
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1091 - Page 3
`
`

`

`Zager et al: Parenteral iron nephrotoxicity
`
`147
`
`temperature, but dissimilarity in PCR products, enabled
`a multiplexed reaction whose products were analyzed
`by agarose gel electrophoresis and ethidium bromide
`staining. cDNA bands were visualized and quantified by
`densitometry with a Typhoon 8600 scanner (Amersham
`Pharmacia Biotech). HO-1 cDNA bands were expressed
`as ratios to the simultaneously obtained GAPDH cDNA
`bands, the latter used as a housekeeping gene.
`Renal histology. To assess whether parenteral iron
`treatment might induce structural renal alterations, mice
`which were subjected to the above intravenous iron treat-
`ment protocols (2 mg iron every other day × 1 week;
`N = 2 for each test agent) or to sham saline injections
`(N = 3). One day following the last injection, the kid-
`neys were harvested, and prepared for either light or
`electron microscopy. For light microscopy, a midline slice
`of kidney (cortex to papilla) was fixed in 10% formalin
`and 4 l paraffin-embedded sections prepared and stained
`with hematoxylin and eosin. For electron microscopy,
`1 mm cubes of renal cortex were fixed by immersion in
`1/2 strength Karnovsky’s fixative. Tissue sections were cut
`and evaluated by transmission electron microscopic anal-
`ysis, as previously described [39]. At least four glomeruli
`from two different kidneys were extensively examined by
`electron microscopy.
`
`Cultured proximal tubular [human kidney (HK-2)]
`cell experiments
`Cytotoxicity and cellular loss of cytochrome c. The fol-
`lowing experiment was undertaken to further ascertain
`relative degrees of iron-mediated cytotoxicity, as assessed
`by % LDH release and extracellular cytochrome c re-
`lease (a marker of mitochondrial damage) [31]. To these
`ends, immortalized human proximal tubular (HK-2) cells
`were cultured in T-75 flasks with keratinocyte serum-free
`medium (K-SFM) and passaged by trypsinization every
`5 to 6 days, as previously described [40]. For experimen-
`tation, the cells were seeded into 18 T-25 flasks. After
`an overnight postseeding recovery period, the cells were
`divided into six groups of three flasks each: (1) control
`cells (N = 3); (2) incubation with 100 lg/mL FeS iron;
`(3) incubation with 100 lg/mL FeG iron; (4) incubation
`with 100 lg/mL FeD iron; (5) incubation with 100 lg/mL
`FeOS iron; and (6) a second group of control incubated
`cells. The cells were maintained under routine culture
`conditions for 3 days. At the completion of the incuba-
`tions, % LDH release was determined. Then, the cells
`which remained attached to the flasks were recovered
`by scraping with a cell scraper, and washed with Hanks’
`balanced salt solution (HBSS), and pelleted. The pellets
`were photographed with a digital camera. Then, cell pro-
`tein extracts were prepared and probed for cytochrome
`c by Western blotting [31]. An equal amount of protein
`(8 lg) from each cell sample was applied.
`
`Electron microscopic analysis of iron effects on HK-2
`cell morphology. The following experiment was under-
`taken to ascertain the effect of the four test iron prepa-
`rations on HK-2 cell morphology. To this end, a 6-well
`Costar plate was seeded with HK-2 cells and allowed to
`grow to near confluence. One well each was subjected to
`the following conditions: (1) control incubation; (2) incu-
`bation with 100 lg FeS iron; (3) incubation with 100 lg
`FeG iron; (4) 100 lg FeD iron; (5) 100 lg of FeOS iron;
`and (6) additional control culture. After an 18-hour incu-
`bation, the cell culture medium in each well was removed,
`and then a mixture of 1 part of 1/2 strength Karnovsky’s
`fixative/1 part fresh culture medium was added to the
`adherent cells. The cells were allowed to fix overnight.
`After dehydration and alcohol fixation, groups of cells
`were randomly lifted off the plates by applying small resin
`blocks to the monolayers. These blocks were then cut
`and processed for transmission electron microscopy, as
`previously described [41].
`
`Combination in vivo/in vitro experiments
`Intravenous iron injection with subsequent in vitro anal-
`ysis of cytoresistance. A feature of acute sublethal renal
`tubular injury is the subsequent emergence of partial cell
`resistance to further attack [42–45]. In particular, iron-
`mediated injury induces resistance to further oxidative
`damage [36, 42]. Hence, the goal of this experiment was
`to ascertain whether parenteral iron administration can
`induce sublethal proximal tubular injury, and that this
`prior injury is denoted by the emergence of cytoresis-
`tance to subsequent iron-mediated tubular attack. To this
`end, four mice were injected with 2 mg of FeS via the tail
`vein (0.1 mL). Four mice subjected to equal saline tail
`vein injections served as controls. The mice were then
`provided with free food and water access (preliminary
`data indicated that no difference in food intake/body
`weight resulted from the iron injection). Eighteen hours
`postinjections, they were anesthetized with pentobarbi-
`tal, the kidneys resected, and cortical proximal tubules
`were isolated, as above. The eight preparations (four
`postiron injection; four postsaline injection) were each
`divided into five equal tubule aliquots as follows: (1) con-
`trol incubation (95% O2/5% CO2); (2) hypoxic incuba-
`tion (95% N2/5% CO2); (3) exposure to 100 lmol/L an-
`timycin A (a mitochondrial inhibitor); or (4) addition of
`25 lmol/L ferrous ammonium sulfate (iron), complexed
`to the siderophore hydroxyquinoline (FeHQ), permitting
`iron to gain intracellular access [42]. After completing
`15-minute incubations under each of these conditions,
`the extent of lethal cell injury was gauged by % LDH
`release. The results for the control and Venofer pretreat-
`ment groups were compared.
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1091 - Page 4
`
`

`

`Zager et al: Parenteral iron nephrotoxicity
`
`10
`
`0123456789
`
`ATP, nmol/mg protein
`
`FeOS
`
`FeD
`
`FeG
`
`FeS
`
`NS
`
`<0.0001
`
`0
`
`125
`
`250
`Fe, µg/mL
`
`500
`
`1000
`
`0123456789
`
`148
`
`ATP, nmol/mg
`
`NS
`
`No Fe
`FeG
`FeS
`<0.02
`
`<0.05
`
`<0.01
`
`0
`
`60
`
`0
`60
`µg/mL
`
`30
`
`30
`
`Fig. 2. Proximal tubular segment adenosine triphosphate (ATP) con-
`centrations with low dose (30 and 60 lg/mL) iron sucrose (FeS) and iron
`gluconate (FeG) exposures. Both drugs caused significant ATP depres-
`sions at the 60 lg/mL concentration. Each drug also tended to depress
`ATP concentrations even at the 30 lg/mL dosage, but only the FeS
`result achieved statistical significance (P < 0.02 vs. controls).
`
`Each of the other test compounds (FeD, FeG, and
`FeOS) failed to raise % LDH release above control val-
`ues (≤13%) even with application of 1000 lg/mL iron
`concentrations.
`pH controls for high dose Venofer additions. Addition
`of alkaline sucrose solution (pH 10.5), did not reproduce
`FeS cytotoxic effects. First, it tended to raise, rather than
`lower, tubule ATP concentrations (pH 10.5, 8.4 ± 0.6;
`controls, 8.0 ± 0.5 nmol/mg protein). Second, % LDH
`release was 14 ± 1% with alkaline sucrose incubation,
`compared to 41 ± 3% with the 1000 lg/mL FeS addition.
`Third, even the highest test dose of FeS (1000 lg/mL)
`had only a small effect on tubule suspension pH, raising
`it from 7.44 to 7.8. Lesser amounts of FeS addition had
`no discernible pH effect.
`ATP concentrations and LDH release with “low dose”
`(30 and 60 lg/mL) iron concentrations. As shown in
`Figure 2, even when added in a 30 or 60 lg iron/mL
`dose, FeS still caused statistically significant reductions
`in tubule ATP concentrations, compared to co-incubated
`control tubules. FeG also lowered ATP at these two iron
`concentrations, but the result was statistically significant
`only at the 60 lg/mL concentration (Fig. 2). None of these
`incubations caused a significant increase in LDH release
`(range for controls and iron compounds, 9% to 11%).
`
`In vivo experiments
`MDA levels following parenteral iron treatment. As
`shown in Figure 3A, each of the iron compounds induced
`statistically significant plasma MDA increments, rising
`well above the upper 95% confidence limit (shown by hor-
`izontal line) for normal plasma MDA values. The plasma
`MDA increase was ∼2× as great with FeD, compared
`
`Fig. 1. Proximal tubular segment adenosine triphosphate (ATP) con-
`centrations following 30-minute incubations with four test iron prepara-
`tions: iron dextran (FeD), iron oligosaccharide (FeOS), iron gluconate
`(FeG), and iron sucrose (FeS). ATP concentrations are presented as
`nmol/mg tubule protein. FeD and FeOS caused only minimal ATP de-
`clines, and these were apparent at only the 1000 lg/mL iron concentra-
`tion. In contrast, steep ATP declines were observed with both FeS and
`FeG, with the degree of ATP reductions being statistically greater with
`FeS vs. FeG (P < 0.0001; all dose-paired comparison). Standard error
`bars are not shown for clarity sake, but were all <0.4 nmol/mg protein.
`
`Calculations and statistics
`All values are presented as means ± 1 SEM. Statistical
`comparisons were made by paired or unpaired Student
`t testing, as per the nature of the experiment. If multiple
`comparisons were made, the Bonferroni correction was
`applied.
`
`RESULTS
`Isolated tubule experiments
`Proximal tubule ATP concentrations in response to 125
`to 1000 lg/mL iron exposures. As shown in Figure 1,
`each of the test iron preparations caused dose depen-
`dent reductions in tubule ATP concentrations. The FeOS
`and FeD curves did not significantly differ, and statisti-
`cally significant ATP reductions were observed at only
`the highest tested concentration (1000 lg/mL of iron;
`P < 0.04 vs. their respective controls). In contrast, FeS
`and FeG each induced steep dose-response curves, clearly
`much more severe than those observed with either FeOS
`or FeD. FeS caused the most severe ATP depressions,
`with ∼50% greater ATP losses being observed vs. their
`corresponding FeG results (P < 0.0001 in an overall com-
`parison between paired concentrations).
`% LDH release with the 125 to 1000 lg/mL iron dosage
`range. The 95% confidence band for % LDH release for
`control tubules was 8% to 13%. In the above-described
`30-minute titration experiments, only FeS raised % LDH
`release above this normal range, but this was observed
`only at the two highest test concentrations (41 ± 3%,
`14 ± 1%, and 13 ± 1% with 1000 lg/mL, 500 lg/mL, and
`250 lg/mL iron doses, respectively) (data not shown).
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1091 - Page 5
`
`

`

`Zager et al: Parenteral iron nephrotoxicity
`
`149
`
`B
`
`2.5
`
`C
`
`2.5
`
`FeD FeG FeS
`
`2
`
`1.5
`
`1
`
`0.5
`
`0
`
`MDA, nmol/mg protein
`
`FeD FeG FeS
`
`2
`
`1.5
`
`1
`
`0.5
`
`0
`
`MDA, nmol/mg protein
`
`FeD FeG FeS
`
`A
`
`12
`
`10
`
`02468
`
`MDA, nmol/mL
`
`Fig. 3. Malondialdehyde (MDA) concentrations in plasma (A), renal cortex (B), and apical myocardium (C) following intravenous iron dextran
`(FeD), iron gluconate (FeG), or iron sucrose (FeS) injection. Whereas FeD caused the greatest plasma MDA increase, it failed to induce renal
`cortical lipid peroxidation (horizontal lines are the upper limit of the 95% confidence interval for normal plasma or tissue MDA levels). Minimal
`cardiac lipid peroxidation resulted, and it was significant only with FeS treatment. Thus, these results indicate that while intravenous irons can cause
`in vivo lipid peroxidation, the degrees to which they do so are both compound, and target tissue, dependent.
`
`<0.03
`
`<0.0001
`
`<0.01
`
`Control
`
`FeD
`
`FeOS
`
`FeG
`
`FeS
`
`1300
`
`1100
`
`900
`
`700
`
`500
`
`300
`
`100
`
`Ferritin, densitometry units
`
`Fes
`
`C
`
`Fes
`
`C
`
`Fes
`
`C
`
`FeG
`
`C
`
`FeG
`
`C
`
`Fig. 4. Western blotting of renal cortex for tissue ferritin after the
`1 week, every other day, iron treatment protocols. Only minimal fer-
`ritin expression was seen in control kidney samples (C). Iron sucrose
`(FeS) treatment clearly increased tissue ferritin expression. A lesser,
`but still significant, increase in ferritin was apparent following iron glu-
`conate (FeG) treatment. In contrast, neither iron dextran (FeD) nor
`iron oligosaccharide (FeOS) caused any ferritin increments (not de-
`picted; see Fig. 5).
`
`to either FeG or FeS treatment. However, in striking
`contrast to plasma (where FeD caused the greatest MDA
`increases), in renal cortex, MDA increments resulted
`from only FeG and FeS, but not FeD injection (Fig. 3B).
`The heart was relatively resistant to iron-mediated lipid
`peroxidation, as only FeS caused any increase in cardiac
`MDA values (above the 95% confidence limits) (Fig. 3C).
`Thus, in composite, these MDA results indicate that while
`all of the test iron compounds can evince lipid peroxida-
`tion, the degree to which this occurs depends on the par-
`ticular tissue target (e.g., plasma, kidney, or heart) and
`the particular drug (FeS, FeG, or FeD) involved.
`Renal cortical ferritin and HO-1 protein expression.
`Slight ferritin expression was seen in control renal cor-
`tical tissue samples (Fig. 4). FeG and FeS each increased
`
`Fig. 5. Western blot densitometric analysis of renal cortical ferritin ex-
`pression. Neither iron dextran (FeD) nor iron oligosaccharide (FeOS)
`induced any change in renal cortical ferritin levels, as assessed by West-
`ern blotting. In contrast, iron gluconate (FeG) and iron sucrose (FeS)
`each raised renal cortical ferritin levels, compared to the control tis-
`sue samples (P < 0.01 and P < 0.0001, respectively). The increase was
`significantly greater with FeS vs. FeG (P < 0.03).
`
`ferritin levels, rising ∼2× and ∼4× over control values,
`respectively (Fig. 5). The increase was statistically greater
`with FeS, in comparison to FeG treatment (P < 0.03).
`Neither FeD nor FeOS caused any discernible ferritin
`increase (i.e., above control values).
`In contrast to ferritin, none of the treatments caused
`any clearly discernible change in HO-1 protein expres-
`sion (data not shown). In contrast, renal cortex obtained
`18 hours postinduction of rhabdomyolysis-induced acute
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1091 - Page 6
`
`

`

`Zager et al: Parenteral iron nephrotoxicity
`
`<0.0001
`
`<0.0001
`
`<0.0001
`
`injury, be observed. However, with FeS treatment, occa-
`sional empty vacuoles were observed in proximal tubular
`cells (suggesting prior iron trafficking through proximal
`tubular cells).
`
`150
`
`1.2
`
`1
`
`0.8
`
`0.6
`
`0.4
`
`0.2
`
`0
`
`HO-1/GAPDH mRNA
`
`NS
`
`Control
`
`FeD
`
`FeS
`
`FeG
`
`FeOS
`
`Fig. 6. Heme oxygenase- 1(HO-1) mRNA levels in renal cortex 4 hours
`following intravenous iron dextrose (FeD), iron sucrose (FeS), iron glu-
`conate (FeG), or iron oligosaccharide (FeOS) injection. Excepting FeD,
`all of the iron preparations induced an approximate doubling of HO-1
`mRNA levels, consistent with the induction of oxidative stress in renal
`cortex.
`
`renal failure showed prominent HO-1 induction (seen at
`32 kD), confirming the adequacy of the employed West-
`ern blot analysis.
`Renal cortical HO-1 mRNA expression. As shown
`in Figure 6, by 4 hours postinjection, FeS, FeG, and
`FeOS each caused an approximate doubling of HO-
`1 mRNA levels (P < 0.0001 for each vs. controls). In
`striking contrast, FeD caused no change in renal corti-
`cal HO-1 mRNA, with values being virtually identical
`to control values. Of note, these results paralleled the
`above-described MDA values, whereas both FeG and
`FeS caused an approximate 30% increase in renal cor-
`tical MDA, FeD did not raise renal MDA levels.
`Renal histology. There was no definitive evidence of
`renal histologic injury (Figs. 7 to 9), as discerned by light
`microscopy (which was the reason for undertaking the
`electron microscopy analyses). FeS, and to a lesser ex-
`tent FeG (but not FeD or FeOS) did induce histologic
`damage, as assessed by electron microscopy. The most
`notable change was glomerular iron accumulation, tak-
`ing the form of electron dense aggregates which were
`most prominent in the mesangium and in endothelial
`cells. Additionally, occasional large iron deposits could
`be found in podocyte cell bodies (Fig. 8B). The foot pro-
`cesses remained well preserved. Glomerular endothelial
`deposits were associated with moderate to marked en-
`dothelial cell swelling, observed in ∼40% of glomeru-
`lar capillary loops (for example,