Sciences ToxSci AdvanceToxicological Access published November 4, 2013
Early prediction of polymyxin-induced nephrotoxicity with next generation urinary kidney injury biomarkers
Manuscript ID: Manuscript Type: Date Submitted by the Author: Complete List of Authors:
Key Words:
Society of Toxicology Specialty Section Subject Area:
Toxicological Sciences TOXSCI-13-0647.R2 Research Article 22-Oct-2013 Keirstead, Natalie; AstraZeneca, Drug Safety and Metabolism Wagoner, Matthew; AstraZeneca, Drug Safety and Metabolism Bentley, Patricia; AstraZeneca, Drug Safety and Metabolism Blais, Marie; AstraZeneca, Drug Safety and Metabolism Brown, Crystal; AstraZeneca, Drug Safety and Metabolism Cheatham, Letitia; AstraZeneca, Drug Safety and Metabolism Ciaccio, Paul; Cubist Pharmaceuticals, Nonclinical Development Dragan, Yvonne; AstraZeneca, Drug Safety and Metabolism Ferguson, Douglas; AstraZeneca, Infection Innovative Medicines Fikes, Jim; AstraZeneca, Drug Safety and Metabolism Galvin, Melanie; AstraZeneca, Drug Safety and Metabolism Gupta, Anshul; AstraZeneca, Infection Innovative Medicines Hale, Michael; AstraZeneca, Infection Innovative Medicines Johnson, Nakpangi; AstraZeneca, Drug Safety and Metabolism Luo, Wenli; AstraZeneca, Discovery Statistics McGrath, Frank; AstraZeneca, Drug Safety and Metabolism Pietras, Mark; AstraZeneca, Drug Safety and Metabolism Price, Sally; AstraZeneca, Drug Safety and Metabolism Sathe, Abhishek; AstraZeneca, Infection Innovative Medicines Sasaki, Jennifer; AstraZeneca, Drug Safety and Metabolism Snow, Debra; AstraZeneca, Drug Safety and Metabolism Walsky, Robert; AstraZeneca, Infection Innovative Medicines Kern, Gunther; AstraZeneca, Infection Innovative Medicines biomarkers < Safety Evaluation, kidney < Systems Toxicology, Risk Assessment
Drug Discovery Toxicology [110]
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
Journal:
Page 1 of 47
1
2
3 4 5
Early prediction of polymyxin-induced nephrotoxicity with next generation urinary kidney injury biomarkers
6
7
Natalie D. Keirstead*1, Matthew P. Wagoner*1, Patricia Bentley1, Marie Blais1, Crystal
8
Brown1, Letitia Cheatham1, Paul Ciaccio1¶, Yvonne Dragan1, Douglas Ferguson2, Jim Fikes1,
9
Melanie Galvin4, Anshul Gupta2, Michael Hale2, Nakpangi Johnson1, Wenli Luo3, Frank
10
McGrath1, Mark Pietras1, Sally Price4, Abhishek G. Sathe2, Jennifer C. Sasaki1, Debra
11
Snow1, Robert L. Walsky2, Gunther Kernψ2
12
13
14
15
1
Drug Safety and Metabolism; 2Infection Innovative Medicines Unit; 3Discovery Statistics,
16
AstraZeneca R&D Boston, 35 Gatehouse Park, Waltham MA 02451, USA; 4Drug Safety and
17
Metabolism, AstraZeneca UK Ltd, Macclesfield, Cheshire, UK
18
¶
19
Lexington, MA 02421, USA
20
*These authors contributed equally to this work.
21
ψ
Current address: Nonclinical Development, Cubist Pharmaceuticals, 65 Hayden Avenue,
Corresponding author: [email protected]
22 23
1
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
24
ABSTRACT
25
Despite six decades of clinical experience with the polymyxin-class of antibiotics, their dose-
26
limiting nephrotoxicity remains difficult to predict due to a paucity of sensitive biomarkers.
27
Here, we evaluate the performance of standard of care and next generation biomarkers of
28
renal injury in the detection and monitoring of polymyxin-induced acute kidney injury in male
29
Han Wistar rats using colistin (polymyxin E) and a polymyxin B (PMB) derivative with
30
reduced nephrotoxicity, PMB nonapeptide (PMBN). This study provides the first
31
histopathological and biomarker analysis of PMBN, an important test of the hypothesis that
32
fatty-acid modifications and charge reductions in polymyxins can reduce their nephrotoxicity.
33
The results indicate that alterations in a panel of urinary kidney injury biomarkers can be
34
used to monitor histopathological injury, with Kim-1 and α-GST emerging as the most
35
sensitive biomarkers outperforming clinical standards of care, serum or plasma creatinine
36
and blood urea nitrogen. To enable the prediction of polymyxin-induced nephrotoxicity, an in
37
vitro cytotoxicity assay was employed using human proximal tubule epithelial cells (HK-2).
38
Cytotoxicity data in these HK-2 cells correlated with the renal toxicity detected via safety
39
biomarker data and histopathological evaluation, suggesting that in vitro and in vivo methods
40
can be incorporated within a screening cascade to prioritize polymyxin class analogs with
41
more favorable renal toxicity profiles.
42
43
44
Keywords:
Polymyxin, colistin, infection, nephrotoxicity, biomarkers, Kim-1, translation,
45
toxicology, drug safety, HK-2, histopathology, immunohistochemistry, AKI, RIFLE
46
47
2
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 47
Page 3 of 47
48
INTRODUCTION
49
Infections caused by multi-drug resistant (MDR) Gram-negative bacteria such as
50
Pseudomonas aeruginosa, Klebsiella pneumoniae and Acinetobacter baumannii have
51
become a serious problem worldwide, and can result in sepsis with high mortality rates
52
(Waterer and Wunderink, 2001). Though the need for new drugs that target MDR-Gram
53
negative bacteria is widely acknowledged, no new treatment options have emerged. It has
54
been predicted that if antibiotics become ineffective, routine surgeries will end in mortalities
55
for 1 in 6 patients (Smith and Coast, 2013). Carbapenem-resistant enterobacteriaceae
56
(CREs) have increased at an alarming rate in the last decade. These multi-drug resistant
57
CREs cause a variety of life-threatening infections killing up to half of all patients who
58
contract them (Smith and Coast, 2013); however, some of these organisms still respond to
59
colistin. Colistin (polymyxin E), is a cationic polypeptide antibiotic that was commonly used
60
until the early 1980s. Clinically, colistin is administered intravenously (IV) as sodium colistin
61
methanesulfonate (CMS), an inactive prodrug, which is then converted in vivo to colistin (Li
62
et al., 2006). When gentamicin and second- and third-generation cephalosporins became
63
available, CMS fell out of favor due to the high incidence of nephrotoxicity (Li et al., 2006,
64
Falagas et al., 2005, Hartzell et al., 2009). However, recent emergence of bacteria resistant
65
to most classes of commercially available antibiotics has resulted in reconsideration of CMS
66
as a last resort for combating Gram-negative sepsis (Hartzell et al., 2009). Despite its
67
nephrotoxic liabilities, CMS is now used as a last-line therapy for MDR Pseudomonas
68
aeruginosa, Acinetobacter baumannii, Klebsiella pneumoniae and Escherichia coli
69
(Landman et al., 2008). Recent studies have reported nephrotoxicity rates as high as 45-
70
55% (Hartzell et al., 2009) using the Risk, Injury, Failure, Loss and End-stage kidney
71
disease (RIFLE) criteria for acute kidney injury (AKI). In the intensive care setting, AKI often
72
is caused or exacerbated by drug-induced nephrotoxicity, contributing to 40-80% of in-
73
hospital mortalities (Chertow et al., 2005). AKI is also associated with high morbidity, with
3
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
74
treatments including dialysis and renal transplant resulting in lower quality of life and
75
tremendous medical costs.
76
Standardized clinical definitions of AKI have been implemented through the use of RIFLE
77
and the AKIN (Acute Kidney Injury Network) criteria (Ricci et al., 2008), with plasma or
78
serum creatinine (Cr) remaining the current standard of diagnosis of AKI. Although Cr is a
79
useful indicator of kidney function in patients with stable chronic kidney disease, it performs
80
poorly in the setting of acute disease, with 25-50% of nephron function lost before detectable
81
elevations in Cr occur (Bellomo et al., 2004). CMS therapy is often stopped upon detection
82
of even minor elevations in Cr as such increases have been correlated with negative
83
outcomes in various populations (Ricci et al., 2008). As a result, CMS therapy may be
84
withdrawn in patients with elevated Cr, even in cases where patients have exhibited a
85
clinical response to therapy.
86
Colistin (polymyxin E) and polymyxin B are cyclic lipodecapeptides, each carrying five free
87
amino groups and, accordingly, five positive charges. These molecules target Gram-
88
negative bacteria by binding to the acidic lipopolysaccharides (LPS), and can also increase
89
the efficacy of other antibiotics by increasing the permeability of the bacterial outer
90
membrane (Falagas et al., 2005). The mechanism of colistin-induced nephrotoxicity is
91
unknown but may relate to its cationic nature and accumulation within proximal tubular
92
epithelial cells (PTECs) of the kidney via endocytosis at the multi-ligand receptor megalin,
93
similar to the mechanism confirmed for aminoglycosides (Lopez-Novoa et al., 2011,
94
Moestrup et al., 1995). Colistin-induced nephrotoxicity is acute and increases with increased
95
dose and duration of therapy, but is usually reversible upon cessation of therapy (Falagas et
96
al., 2005). Cytotoxicity occurs in the cell types in which the drug accumulates, most notably
97
within PTECs (Azad et al., 2013), and may be the result of mitochondrial damage and/or
98
release of reactive oxygen species (Servais et al., 2005, Servais et al., 2008). There have
99
been efforts to create less nephrotoxic polymyxin analogs (Danner et al., 1989, Magee et al.,
4
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 4 of 47
Page 5 of 47
100
2013) by reducing the number of positively charged side chains or removing the fatty-acid
101
tail, which have both been predicted to reduce nephrotoxicity. One such analog, polymyxin B
102
nonapeptide (PMBN), had less acute toxicity than its parent polymyxin but was unfortunately
103
also devoid of direct antibacterial activity (Danner et al., 1989).
104
Recently developed urinary renal injury biomarkers have been shown to be early indicators
105
of renal injury. Several of these markers (Kim-1, albumin, NGAL, TFF3, CLU, RPA-1, total
106
protein (TP), β2-microglobulin and cystatin C) have been qualified by the U.S Food and
107
Drug Administration, the European Medicines Agency and the Pharmaceuticals Medical
108
Devices Agency in Japan for preclinical rodent studies (Bonventre et al., 2010, Dieterle et
109
al., 2010). No literature exists regarding the usefulness of urine acute kidney injury
110
biomarkers for prediction of polymyxin-induced renal toxicity. Suitable nonclinical and
111
translational in vivo markers would allow for a more rapid development of efficacious,
112
polymyxin analogs with reduced nephrotoxic potential. Several investigations describing the
113
relationship between in vitro and in vivo safety assessment of polymyxins have recently
114
been published (Abdelraouf et al., 2012, Magee et al., 2013). We have thus conducted a
115
series of experiments, comparing the nephrotoxicity of colistin sulfate and PMBN in an in
116
vivo rat model of polymyxin-mediated renal toxicity, and an in vitro model of human proximal
117
tubule epithelial cell cytotoxicity. Our results indicate that this multi-tiered screening cascade
118
provides promise in identification of compounds with reduced nephrotoxic liability.
119
120
MATERIALS AND METHODS
121
Antimicrobial Agents. For the in vitro cytotoxicity studies and the in vivo animal studies,
122
colistin sulfate (USP) was purchased from Sigma Aldrich (St. Louis MO), lot number
123
081M1526V. PMBN was made by following the procedure previously reported (Danner et al.,
124
1989); the crude material was purified by reverse phase HPLC. Endotoxin testing was
5
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
125
performed by the Associates of Cape Cod Incorporated (East Falmouth, MA) using the gel-
126
clot method for bacterial endotoxin testing.
127
In vitro cytotoxicity assays. Compounds were diluted directly in Keratinocyte Serum Free
128
Medium supplemented with bovine pituitary extract and epidermal growth factor (Life
129
Technologies, Carlsbad, Ca, Cat # 17005-042). The immortalized proximal tubular cell line
130
human kidney 2 (HK-2) cells (American Type Culture Collection, CRL-2190) were plated at
131
20,000 cells per well in a Corning 96 well tissue culture-treated plate (Sigma, St Louis, Mo,
132
Cat # CLS3595). Cells were cultured at 37oC in 5% CO2 for 7 days with fresh media added
133
every other day to form a confluent monolayer. After 7 days, compound was added directly
134
to cells at ten concentrations in quadruplicate, including a media-only control. After 24 hours,
135
CellTiter-Glo (Promega, G7571) luminescent cell viability assay was used to estimate
136
cellular viability using a VictorLight luminometer (Perkin Elmer, Waltham, MA). TC50s were
137
calculated using non-linear regression with a Hill Slope of -1 and upper and lower
138
constrained values at 100 and 0, respectively, in GraphPad Prism version 5 (La Jolla, CA).
139
Animals. Thirty-three 10-week old male Han Wistar rats weighing 275 to 375g were
140
purchased from Charles River Laboratories (Raleigh, NC). The rats were given a minimum
141
of 72 hours to acclimate before the start of the experiments. Animals were randomly
142
assigned to either vehicle, PMBN or the colistin group and housed 2/cage. Animal
143
identification and conditions of housing, acclimatization, environment, diet and water were in
144
accordance with facility Standard Operating Procedures.
145
Study Design. All animal procedures were conducted in an Association for Assessment
146
and Accreditation of Laboratory Animal Care (AAALAC)-accredited facility under an
147
Institutional Animal Care and Use Committee (IACUC) approval protocol and underwent
148
statistical review to confirm that the study was appropriately powered for the endpoints of
149
interest. In preliminary studies, a total of four different dosage regimens were evaluated for
150
the ability to cause colistin-induced nephrotoxicity. Previous investigative studies from our 6
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 6 of 47
Page 7 of 47
151
laboratory (unpublished results) have shown that cumulative doses of colistin and PMBN up
152
to 39 and 44.75 mg/kg, respectively, administered via IV infusion, did not produce
153
histopathological changes in the kidney. An ascending dose paradigm over 4 days was
154
required to avoid tolerability issues (related to histamine release), but maximum daily doses
155
limited the amount of either colistin or PMBN that could be administered in 7 days. These
156
regimens did not lead to reproducible renal damage based upon clinical pathology (including
157
urinary renal biomarkers) or histopathology, and tolerability issues did not allow for increased
158
IV dosages. Subsequent studies revealed that both compounds were better tolerated when
159
administered via subcutaneous (SQ) injection and resulted in increased plasma exposures.
160
Therefore, the administration route was changed to the subcutaneous route with increased
161
frequency of dosing. The nature and extent of kidney damage caused by this regimen was
162
considered representative of colistin-induced nephrotoxicity in human patients (Hartzell et
163
al., 2009).
164
Animals were dosed 4 times daily (QID) for 7 days by SQ injection in the intrascapular
165
region, approximately every 6 hours. Doses for both the colistin and PMBN groups were
166
based on the results of maximum tolerated dose (MTD) studies. At 40 mg/kg of colistin,
167
clinical observations such as decreased motor activity, cyanosis of face and limbs, swollen
168
head and limbs and eyes half shut were recorded out to 4 hours post-dose. At 40 mg/kg of
169
PMBN, animals had swollen muzzles and paws, decreased motor activity, cyanosis and cold
170
extremities; with minimal clinical signs at 4 hours post-dose. Based on these observations,
171
the lower dose of 25 mg/kg of colistin was expected to be well tolerated. The dose level of
172
PMBN remained the same as was used in MTD studies due to the lack of persistence of
173
clinical signs. Control animals were dosed with the vehicle, sterile saline for injection. The
174
colistin-treated group was dosed with 6.25 mg/kg QID (25 mg/kg/day) and the PMBN group
175
was dosed with 10 mg/kg QID (40 mg/kg/day). All groups were also dosed once, 3 hours
176
prior to necropsy, with their respective vehicle or test article. Colistin, rather than CMS, was
177
used in the present study because it is the active antibacterial drug formed in the body after 7
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
178
administration of CMS, reducing the complexity resulting from the simultaneous presence in
179
the body of the prodrug, CMS, and formed colistin.
180
Urine collection. Animals were housed in metabolic cages for 6 hours of a 24 hour cycle.
181
Though longer urine collection times can yield more accurate values, based on previous
182
clinical observations with polymyxin analogs, a 6-hour collection period was used in order to
183
balance animal welfare concerns with data integrity considerations. From the pre-dose time-
184
point through the final necropsy, urine was collected daily for all groups, but only processed
185
for urinalysis (data not shown), urine chemistry and urine kidney injury biomarker analysis on
186
days 0, 2, 4 and 7.
187
Toxicokinetic analysis. Three animals from the colistin and PMBN dose groups were
188
assigned as satellite animals for the 7-day toxicokinetic evaluation. Blood samples were
189
collected from each animal (tail vein venipuncture) on day 1 and day 4 at 2 and 6 hours post-
190
dose and on day 7 at 0.33, 1, 2, 4, 6 & 24 hours. Terminal blood and kidney tissue samples
191
were obtained 3 hours after the final dose on day 8. Blood and kidney samples were also
192
obtained from the interim sample groups of rats (n=3 per treatment group) that were
193
necropsied on day 3, following 2 days of QID dosing. Blood samples from these animals
194
were taken on day 1 at 0.33, 1, 2, 4, 6, 8 & 24 hours and on day 2 at 2 & 24 hours while
195
kidney tissue samples were obtained following termination at 3 hours post-dose on day 3.
196
Satellite and interim sacrifice groups were utilized for toxicokinetic analyses; nephrotoxicity
197
endpoints were only evaluated in the main study animals, where blood sampling only
198
occurred at necropsy.
199
Plasma samples (0.05 mL) were precipitated by the addition of 95/5 acetonitrile/formic acid
200
(0.25 mL) containing fibrinopeptide B (0.1 µg/mL) as an internal standard. Samples were
201
shaken for 5 minutes, and then centrifuged for 5 minutes at 4000 RPM to remove proteins.
202
An aliquot (0.2 mL) was removed, dried under N2 at 40°C., and reconstituted in 95:5:0.1
203
ACN:water:formic acid (0.06 mL). Kidney samples were diluted 1:3 w/v with water, and
8
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 8 of 47
Page 9 of 47
204
homogenized using a tissue miser homogenizer (ThermoFisher Scientific, Waltham, MA) at
205
~30,000 RPM while on ice. The kidney homogenate was then typically diluted 5-fold into
206
plasma and processed as indicated above. Analysis was performed by LC/MS/MS on an
207
Applied Biosystems API5000 triple quadrupole mass spectrometer (AB Sciex LLC, Foster
208
City, CA) operating in the positive ion mode with a Shimadzu NEXERA HPLC inlet system
209
with a Phenomenex reverse phase, 5µ C8(2) column (30x2.0mm). For separation, mobile
210
phase A consisted of water, and mobile phase B consisted of ACN, both containing 0.1%
211
formic acid. Samples (10 µL) were injected onto the LC with a flow rate of 0.6 mL/min
212
starting at 2% B for 0.5 min then ramped to 80% B over 2 min (2.5 min total), the column
213
was flushed for 1 min then returned to 2% B and equilibrated for 1 min. Under these
214
conditions, colistin was monitored by fragmentation at m/z 602>241, PMBN at m/z 786>333,
215
and fibrinopeptide B (IS) at m/z 786>333 with a source temperature of 550°C., a declustering
216
potential of 60 V, and collision energies of 31, 35, and 38 eV respectively.
217
Clinical pathology and urinary kidney injury biomarkers. Routine hematology was
218
evaluated by Bioreliance (Rockville, MD, USA) on samples collected just prior to necropsy,
219
using the Advia 120 Hematology System (Siemens Healthcare, Malvern, PA, USA; data not
220
shown). A full plasma clinical chemistry panel and urine chemistry were evaluated by
221
Bioreliance (Rockville, MD, USA) using the Cobas c311 (Roche Diagnostics, Indianapolis,
222
IN, USA), including plasma and urine creatinine levels. All analytical methods used were fully
223
validated for rat samples. Urine samples were assayed using Meso-Scale Discovery (MSD)
224
electrochemiluminescent immunoassays in technical replicates, with responses evaluated on
225
the Sector Imager 6000 instrument (MSD, Gaithersburg, MD, USA). Kits used were the Rat
226
Kidney Injury Panel 1 (albumin, Kim-1, NGAL/lipocalin-2, osteopontin), the Argutus Acute
227
Kidney Injury Panel (α-GST, GST-µ, RPA-1) and the Clusterin Test Kit (MSD, Gaithersburg,
228
MD, USA), according to the manufacturer's instructions.
9
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
229
Pathological examination. An interim sampled group of three rats/treatment group was
230
necropsied on day 3, after 2 days of QID dosing to assess any early renal histopathology. In
231
the main study groups, six rats/treatment were necropsied on day 8, following 7 days of QID
232
dosing. After gross examination, the kidneys were weighed, collected, processed and
233
examined microscopically. Sections of liver, lung and heart were also evaluated
234
histologically. Representative kidney sections (left and right) were generated for all animals
235
in each group. Tissues were fixed in 10% neutral-buffered formalin (NBF) for approximately
236
48 hours and subsequently processed, embedded in paraffin, sectioned at 4 µm, mounted
237
on glass slides, deparaffinized, and stained with hematoxylin and eosin (H&E) and periodic-
238
acid Schiff (PAS). Kidney sections were evaluated by light microscopy and graded for the
239
degree of tubular degeneration on a 5-point scale: 0 = no abnormality detected (NAD), 1 =
240
minimal, 2 = slight, 3 = mild, 4 = moderate, 5 = marked or severe findings. The
241
histopathologic grading scheme for degeneration was defined as follows: minimal:
242
occasional degenerate (brightly eosinophilic) cells with pyknotic nuclei; slight: occasional
243
degenerate cells with pyknotic to karyorrhectic nuclei and sloughed cells within tubular
244
lumina (protein casts); mild: small clusters of 2-4 degenerate cells with pyknotic nuclei and
245
protein casts; moderate: larger clusters and chains of degenerate cells, some with complete
246
loss of chromatin, affecting numerous tubules; marked: majority of tubules affected by chains
247
of degenerate cells, or entire tubular segments affected by degeneration.
248
Kim-1 immunohistochemistry (IHC). IHC was carried out using methods previously
249
described (Wadey et al., 2013). Briefly, 4-um thick tissue sections were incubated with Kim-1
250
primary antibody (rabbit immunoglobulin G [IgG], 1:400; R&D Systems, Minneapolis MN). All
251
immunostaining was carried out at room temperature using a Labvision autostainer
252
(Labvision, Fremont, CA, USA). Rabbit and goat isotype controls (Dako, Enmark) were used
253
as negative controls. Images of IHC whole kidney sections were captured using a
254
ScanScope Scanner and analyzed using ImageScope software (Aperio Technologies Inc.,
255
Vista CA). Thresholds were set to detect DAB positivity, and the kidney sections analyzed to 10
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 10 of 47
Page 11 of 47
256
determine the proportion of positive pixels for Kim-1 immunolabeling. Prior to statistical
257
analysis, the proportion of positive pixels was transformed using the Arcsine transformation.
258
A linear regression model using GraphPad Prism version 5.0 (La Jolla, CA), compared Kim-
259
1 urine levels to the proportion of Kim-1 expression in the kidney of each individual rat.
260
Statistical analysis of urine biomarker data. Creatinine correction was calculated for urine
261
chemistry and urine injury biomarker parameters. Creatinine adjustment is an accepted
262
method to normalize urinary biomarker values to account for the wide variation in urine
263
concentration that can be seen within and between individuals over time (Alessio et al.,
264
1985). For the urine injury biomarkers and urine chemistry, mixed models were utilized to
265
account for repeat measurements at different time points pre- and post-dose. Each
266
parameter is log-transformed before modeling and log-transformed fold-change from
267
baseline is modeled. The measure of relationship between urine biomarker and urine
268
chemistry data is based on Pearson’s r correlation coefficients. All statistical analysis was
269
performed using SAS 9.2 software (SAS Institute Inc., Cary, NC, USA). Statistical
270
significance was defined as p0 (’1’). Thus, all positive grades of histopathology (grades 1 to 2) were treated with
295
equal weight for the ROC analysis. Pre-dose and vehicle groups were not included in the
296
second analysis in order to directly compare the predictability of our measured biomarkers
297
with histopathology as the classifier, but in the face of both compounds accumulating within
298
the PTECs. In ROC analysis for these urinary kidney injury biomarkers, we set a low
299
threshold (tendency to “over call”) with a high sensitivity but lower specificity (allowing for a
300
false positive rate of 0.05) because for our purposes in the evaluation of early stage
301
therapeutic compound candidates, it is more important to not miss any cases of even
302
minimal renal injury then to mis-classify a normal animal as having injury.
303
304
RESULTS
305
In vitro cytotoxicity in the HK-2 cell line. Previously published work has demonstrated the
306
sensitivity of the immortalized human proximal tubule kidney cell line (HK-2) to polymyxin-
307
induced cytotoxicity (Vaara and Vaara, 2013). We hypothesized that cytotoxicity assessment
12
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 12 of 47
Page 13 of 47
308
with HK-2 cells could stratify polymyxin derivatives according to nephrotoxic potential. Cell
309
viability upon 24 hour treatment of polymyxin B (PMB), colistin sulfate or PMBN was
310
evaluated (Figure 1). PMBN showed greater than 50-fold less cytotoxicity in the HK-2 cell
311
cytotoxicity assay than its parent compound PMB with TC50 of >1000 µM and 20 µM,
312
respectively. With a TC50 of 70 µM, colistin sulfate consistently showed at least 3-fold less
313
cytotoxicity than PMB.
314
In vivo rat toxicity studies, clinical observations. Supporting previous observations in the
315
literature (Magee et al., 2013, Wallace et al., 2008), IV infusion of PMBN and colistin at
316
doses of 1 mg/kg were not well tolerated, with rats showing many signs commonly ascribed
317
to a histaminergic response, including swollen paws, reduced motor activity and red ears.
318
Subcutaneous dosing of colistin reduced these tolerability issues, and was found to be well
319
tolerated up to 40 mg/kg in Han Wistar rats. Interestingly, the plasma histamine levels noted
320
with subcutaneous administration were not less than that incurred with intravenous dosing
321
(Figure 2), suggesting that the tolerability issues with IV dosing were not solely related to
322
increased histamine levels.
323
Toxicokinetics. For several animals, both colistin and PMBN-dosed, it was not possible to
324
obtain a sufficient blood sample at the 1 and 2 hour time-points following the first dose on
325
day 1. One satellite animal from the colistin dose group died following the first dose, but
326
necropsy results suggested that this was not directly related to treatment, but was due to an
327
underlying transmural stomach ulcer. The measured plasma concentration data for the other
328
colistin and PMBN dosed animals and the corresponding individual model fits are shown in
329
Figure 3. Steady state was observed to be reached by 24 hours and the overall level of inter-
330
individual variability in TK for both colistin and PMBN was relatively low. Derived
331
toxicokinetic parameter values are shown in Table 1.
332
Clinical pathology and renal biomarker results. Hematology was collected and evaluated
333
on the day of necropsy. Relevant observed changes were restricted to a 3.5-fold increase in 13
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
334
eosinophils in rats administered 6.25 mg/kg QID colistin sulfate, consistent with clinical
335
observations of an allergic hypersensitivity where affected animals demonstrated swollen
336
limbs and muzzles, red ears, half-shut/shut eyes, decreased motor activity and general
337
distress. A full plasma clinical chemistry panel was also evaluated; blood urea nitrogen
338
(BUN) and Cr were mildly increased in colistin-treated animals by 1.2-fold and 1.5-fold,
339
respectively (see Table 2). Renal injury urine biomarkers and urine chemistry were
340
measured at pre-dose (day 0) and on days 2, 4 and 7 post-dose. As shown in Figure 4, the
341
urinary concentration or activity of several biomarkers (Kim-1, α-GST, albumin, NGAL,
342
clusterin, GST-µ, osteopontin, RPA-1) was elevated in a time-dependent manner in the
343
colistin-treated group. Among these biomarkers, albumin, Kim-1, α-GST, clusterin, GST-µ,
344
NGAL, and osteopontin showed early and significantly large alterations in concentration 2
345
days post-treatment (47-fold, 17-fold, 16-fold, 15-fold, 13-fold, 12-fold, and 4-fold increases,
346
respectively, when compared to vehicle-treated rats, p
Early prediction of polymyxin-induced nephrotoxicity with next generation urinary kidney injury biomarkers
Manuscript ID: Manuscript Type: Date Submitted by the Author: Complete List of Authors:
Key Words:
Society of Toxicology Specialty Section Subject Area:
Toxicological Sciences TOXSCI-13-0647.R2 Research Article 22-Oct-2013 Keirstead, Natalie; AstraZeneca, Drug Safety and Metabolism Wagoner, Matthew; AstraZeneca, Drug Safety and Metabolism Bentley, Patricia; AstraZeneca, Drug Safety and Metabolism Blais, Marie; AstraZeneca, Drug Safety and Metabolism Brown, Crystal; AstraZeneca, Drug Safety and Metabolism Cheatham, Letitia; AstraZeneca, Drug Safety and Metabolism Ciaccio, Paul; Cubist Pharmaceuticals, Nonclinical Development Dragan, Yvonne; AstraZeneca, Drug Safety and Metabolism Ferguson, Douglas; AstraZeneca, Infection Innovative Medicines Fikes, Jim; AstraZeneca, Drug Safety and Metabolism Galvin, Melanie; AstraZeneca, Drug Safety and Metabolism Gupta, Anshul; AstraZeneca, Infection Innovative Medicines Hale, Michael; AstraZeneca, Infection Innovative Medicines Johnson, Nakpangi; AstraZeneca, Drug Safety and Metabolism Luo, Wenli; AstraZeneca, Discovery Statistics McGrath, Frank; AstraZeneca, Drug Safety and Metabolism Pietras, Mark; AstraZeneca, Drug Safety and Metabolism Price, Sally; AstraZeneca, Drug Safety and Metabolism Sathe, Abhishek; AstraZeneca, Infection Innovative Medicines Sasaki, Jennifer; AstraZeneca, Drug Safety and Metabolism Snow, Debra; AstraZeneca, Drug Safety and Metabolism Walsky, Robert; AstraZeneca, Infection Innovative Medicines Kern, Gunther; AstraZeneca, Infection Innovative Medicines biomarkers < Safety Evaluation, kidney < Systems Toxicology, Risk Assessment
Drug Discovery Toxicology [110]
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
Journal:
Page 1 of 47
1
2
3 4 5
Early prediction of polymyxin-induced nephrotoxicity with next generation urinary kidney injury biomarkers
6
7
Natalie D. Keirstead*1, Matthew P. Wagoner*1, Patricia Bentley1, Marie Blais1, Crystal
8
Brown1, Letitia Cheatham1, Paul Ciaccio1¶, Yvonne Dragan1, Douglas Ferguson2, Jim Fikes1,
9
Melanie Galvin4, Anshul Gupta2, Michael Hale2, Nakpangi Johnson1, Wenli Luo3, Frank
10
McGrath1, Mark Pietras1, Sally Price4, Abhishek G. Sathe2, Jennifer C. Sasaki1, Debra
11
Snow1, Robert L. Walsky2, Gunther Kernψ2
12
13
14
15
1
Drug Safety and Metabolism; 2Infection Innovative Medicines Unit; 3Discovery Statistics,
16
AstraZeneca R&D Boston, 35 Gatehouse Park, Waltham MA 02451, USA; 4Drug Safety and
17
Metabolism, AstraZeneca UK Ltd, Macclesfield, Cheshire, UK
18
¶
19
Lexington, MA 02421, USA
20
*These authors contributed equally to this work.
21
ψ
Current address: Nonclinical Development, Cubist Pharmaceuticals, 65 Hayden Avenue,
Corresponding author: [email protected]
22 23
1
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
24
ABSTRACT
25
Despite six decades of clinical experience with the polymyxin-class of antibiotics, their dose-
26
limiting nephrotoxicity remains difficult to predict due to a paucity of sensitive biomarkers.
27
Here, we evaluate the performance of standard of care and next generation biomarkers of
28
renal injury in the detection and monitoring of polymyxin-induced acute kidney injury in male
29
Han Wistar rats using colistin (polymyxin E) and a polymyxin B (PMB) derivative with
30
reduced nephrotoxicity, PMB nonapeptide (PMBN). This study provides the first
31
histopathological and biomarker analysis of PMBN, an important test of the hypothesis that
32
fatty-acid modifications and charge reductions in polymyxins can reduce their nephrotoxicity.
33
The results indicate that alterations in a panel of urinary kidney injury biomarkers can be
34
used to monitor histopathological injury, with Kim-1 and α-GST emerging as the most
35
sensitive biomarkers outperforming clinical standards of care, serum or plasma creatinine
36
and blood urea nitrogen. To enable the prediction of polymyxin-induced nephrotoxicity, an in
37
vitro cytotoxicity assay was employed using human proximal tubule epithelial cells (HK-2).
38
Cytotoxicity data in these HK-2 cells correlated with the renal toxicity detected via safety
39
biomarker data and histopathological evaluation, suggesting that in vitro and in vivo methods
40
can be incorporated within a screening cascade to prioritize polymyxin class analogs with
41
more favorable renal toxicity profiles.
42
43
44
Keywords:
Polymyxin, colistin, infection, nephrotoxicity, biomarkers, Kim-1, translation,
45
toxicology, drug safety, HK-2, histopathology, immunohistochemistry, AKI, RIFLE
46
47
2
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 47
Page 3 of 47
48
INTRODUCTION
49
Infections caused by multi-drug resistant (MDR) Gram-negative bacteria such as
50
Pseudomonas aeruginosa, Klebsiella pneumoniae and Acinetobacter baumannii have
51
become a serious problem worldwide, and can result in sepsis with high mortality rates
52
(Waterer and Wunderink, 2001). Though the need for new drugs that target MDR-Gram
53
negative bacteria is widely acknowledged, no new treatment options have emerged. It has
54
been predicted that if antibiotics become ineffective, routine surgeries will end in mortalities
55
for 1 in 6 patients (Smith and Coast, 2013). Carbapenem-resistant enterobacteriaceae
56
(CREs) have increased at an alarming rate in the last decade. These multi-drug resistant
57
CREs cause a variety of life-threatening infections killing up to half of all patients who
58
contract them (Smith and Coast, 2013); however, some of these organisms still respond to
59
colistin. Colistin (polymyxin E), is a cationic polypeptide antibiotic that was commonly used
60
until the early 1980s. Clinically, colistin is administered intravenously (IV) as sodium colistin
61
methanesulfonate (CMS), an inactive prodrug, which is then converted in vivo to colistin (Li
62
et al., 2006). When gentamicin and second- and third-generation cephalosporins became
63
available, CMS fell out of favor due to the high incidence of nephrotoxicity (Li et al., 2006,
64
Falagas et al., 2005, Hartzell et al., 2009). However, recent emergence of bacteria resistant
65
to most classes of commercially available antibiotics has resulted in reconsideration of CMS
66
as a last resort for combating Gram-negative sepsis (Hartzell et al., 2009). Despite its
67
nephrotoxic liabilities, CMS is now used as a last-line therapy for MDR Pseudomonas
68
aeruginosa, Acinetobacter baumannii, Klebsiella pneumoniae and Escherichia coli
69
(Landman et al., 2008). Recent studies have reported nephrotoxicity rates as high as 45-
70
55% (Hartzell et al., 2009) using the Risk, Injury, Failure, Loss and End-stage kidney
71
disease (RIFLE) criteria for acute kidney injury (AKI). In the intensive care setting, AKI often
72
is caused or exacerbated by drug-induced nephrotoxicity, contributing to 40-80% of in-
73
hospital mortalities (Chertow et al., 2005). AKI is also associated with high morbidity, with
3
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
74
treatments including dialysis and renal transplant resulting in lower quality of life and
75
tremendous medical costs.
76
Standardized clinical definitions of AKI have been implemented through the use of RIFLE
77
and the AKIN (Acute Kidney Injury Network) criteria (Ricci et al., 2008), with plasma or
78
serum creatinine (Cr) remaining the current standard of diagnosis of AKI. Although Cr is a
79
useful indicator of kidney function in patients with stable chronic kidney disease, it performs
80
poorly in the setting of acute disease, with 25-50% of nephron function lost before detectable
81
elevations in Cr occur (Bellomo et al., 2004). CMS therapy is often stopped upon detection
82
of even minor elevations in Cr as such increases have been correlated with negative
83
outcomes in various populations (Ricci et al., 2008). As a result, CMS therapy may be
84
withdrawn in patients with elevated Cr, even in cases where patients have exhibited a
85
clinical response to therapy.
86
Colistin (polymyxin E) and polymyxin B are cyclic lipodecapeptides, each carrying five free
87
amino groups and, accordingly, five positive charges. These molecules target Gram-
88
negative bacteria by binding to the acidic lipopolysaccharides (LPS), and can also increase
89
the efficacy of other antibiotics by increasing the permeability of the bacterial outer
90
membrane (Falagas et al., 2005). The mechanism of colistin-induced nephrotoxicity is
91
unknown but may relate to its cationic nature and accumulation within proximal tubular
92
epithelial cells (PTECs) of the kidney via endocytosis at the multi-ligand receptor megalin,
93
similar to the mechanism confirmed for aminoglycosides (Lopez-Novoa et al., 2011,
94
Moestrup et al., 1995). Colistin-induced nephrotoxicity is acute and increases with increased
95
dose and duration of therapy, but is usually reversible upon cessation of therapy (Falagas et
96
al., 2005). Cytotoxicity occurs in the cell types in which the drug accumulates, most notably
97
within PTECs (Azad et al., 2013), and may be the result of mitochondrial damage and/or
98
release of reactive oxygen species (Servais et al., 2005, Servais et al., 2008). There have
99
been efforts to create less nephrotoxic polymyxin analogs (Danner et al., 1989, Magee et al.,
4
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 4 of 47
Page 5 of 47
100
2013) by reducing the number of positively charged side chains or removing the fatty-acid
101
tail, which have both been predicted to reduce nephrotoxicity. One such analog, polymyxin B
102
nonapeptide (PMBN), had less acute toxicity than its parent polymyxin but was unfortunately
103
also devoid of direct antibacterial activity (Danner et al., 1989).
104
Recently developed urinary renal injury biomarkers have been shown to be early indicators
105
of renal injury. Several of these markers (Kim-1, albumin, NGAL, TFF3, CLU, RPA-1, total
106
protein (TP), β2-microglobulin and cystatin C) have been qualified by the U.S Food and
107
Drug Administration, the European Medicines Agency and the Pharmaceuticals Medical
108
Devices Agency in Japan for preclinical rodent studies (Bonventre et al., 2010, Dieterle et
109
al., 2010). No literature exists regarding the usefulness of urine acute kidney injury
110
biomarkers for prediction of polymyxin-induced renal toxicity. Suitable nonclinical and
111
translational in vivo markers would allow for a more rapid development of efficacious,
112
polymyxin analogs with reduced nephrotoxic potential. Several investigations describing the
113
relationship between in vitro and in vivo safety assessment of polymyxins have recently
114
been published (Abdelraouf et al., 2012, Magee et al., 2013). We have thus conducted a
115
series of experiments, comparing the nephrotoxicity of colistin sulfate and PMBN in an in
116
vivo rat model of polymyxin-mediated renal toxicity, and an in vitro model of human proximal
117
tubule epithelial cell cytotoxicity. Our results indicate that this multi-tiered screening cascade
118
provides promise in identification of compounds with reduced nephrotoxic liability.
119
120
MATERIALS AND METHODS
121
Antimicrobial Agents. For the in vitro cytotoxicity studies and the in vivo animal studies,
122
colistin sulfate (USP) was purchased from Sigma Aldrich (St. Louis MO), lot number
123
081M1526V. PMBN was made by following the procedure previously reported (Danner et al.,
124
1989); the crude material was purified by reverse phase HPLC. Endotoxin testing was
5
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
125
performed by the Associates of Cape Cod Incorporated (East Falmouth, MA) using the gel-
126
clot method for bacterial endotoxin testing.
127
In vitro cytotoxicity assays. Compounds were diluted directly in Keratinocyte Serum Free
128
Medium supplemented with bovine pituitary extract and epidermal growth factor (Life
129
Technologies, Carlsbad, Ca, Cat # 17005-042). The immortalized proximal tubular cell line
130
human kidney 2 (HK-2) cells (American Type Culture Collection, CRL-2190) were plated at
131
20,000 cells per well in a Corning 96 well tissue culture-treated plate (Sigma, St Louis, Mo,
132
Cat # CLS3595). Cells were cultured at 37oC in 5% CO2 for 7 days with fresh media added
133
every other day to form a confluent monolayer. After 7 days, compound was added directly
134
to cells at ten concentrations in quadruplicate, including a media-only control. After 24 hours,
135
CellTiter-Glo (Promega, G7571) luminescent cell viability assay was used to estimate
136
cellular viability using a VictorLight luminometer (Perkin Elmer, Waltham, MA). TC50s were
137
calculated using non-linear regression with a Hill Slope of -1 and upper and lower
138
constrained values at 100 and 0, respectively, in GraphPad Prism version 5 (La Jolla, CA).
139
Animals. Thirty-three 10-week old male Han Wistar rats weighing 275 to 375g were
140
purchased from Charles River Laboratories (Raleigh, NC). The rats were given a minimum
141
of 72 hours to acclimate before the start of the experiments. Animals were randomly
142
assigned to either vehicle, PMBN or the colistin group and housed 2/cage. Animal
143
identification and conditions of housing, acclimatization, environment, diet and water were in
144
accordance with facility Standard Operating Procedures.
145
Study Design. All animal procedures were conducted in an Association for Assessment
146
and Accreditation of Laboratory Animal Care (AAALAC)-accredited facility under an
147
Institutional Animal Care and Use Committee (IACUC) approval protocol and underwent
148
statistical review to confirm that the study was appropriately powered for the endpoints of
149
interest. In preliminary studies, a total of four different dosage regimens were evaluated for
150
the ability to cause colistin-induced nephrotoxicity. Previous investigative studies from our 6
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 6 of 47
Page 7 of 47
151
laboratory (unpublished results) have shown that cumulative doses of colistin and PMBN up
152
to 39 and 44.75 mg/kg, respectively, administered via IV infusion, did not produce
153
histopathological changes in the kidney. An ascending dose paradigm over 4 days was
154
required to avoid tolerability issues (related to histamine release), but maximum daily doses
155
limited the amount of either colistin or PMBN that could be administered in 7 days. These
156
regimens did not lead to reproducible renal damage based upon clinical pathology (including
157
urinary renal biomarkers) or histopathology, and tolerability issues did not allow for increased
158
IV dosages. Subsequent studies revealed that both compounds were better tolerated when
159
administered via subcutaneous (SQ) injection and resulted in increased plasma exposures.
160
Therefore, the administration route was changed to the subcutaneous route with increased
161
frequency of dosing. The nature and extent of kidney damage caused by this regimen was
162
considered representative of colistin-induced nephrotoxicity in human patients (Hartzell et
163
al., 2009).
164
Animals were dosed 4 times daily (QID) for 7 days by SQ injection in the intrascapular
165
region, approximately every 6 hours. Doses for both the colistin and PMBN groups were
166
based on the results of maximum tolerated dose (MTD) studies. At 40 mg/kg of colistin,
167
clinical observations such as decreased motor activity, cyanosis of face and limbs, swollen
168
head and limbs and eyes half shut were recorded out to 4 hours post-dose. At 40 mg/kg of
169
PMBN, animals had swollen muzzles and paws, decreased motor activity, cyanosis and cold
170
extremities; with minimal clinical signs at 4 hours post-dose. Based on these observations,
171
the lower dose of 25 mg/kg of colistin was expected to be well tolerated. The dose level of
172
PMBN remained the same as was used in MTD studies due to the lack of persistence of
173
clinical signs. Control animals were dosed with the vehicle, sterile saline for injection. The
174
colistin-treated group was dosed with 6.25 mg/kg QID (25 mg/kg/day) and the PMBN group
175
was dosed with 10 mg/kg QID (40 mg/kg/day). All groups were also dosed once, 3 hours
176
prior to necropsy, with their respective vehicle or test article. Colistin, rather than CMS, was
177
used in the present study because it is the active antibacterial drug formed in the body after 7
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
178
administration of CMS, reducing the complexity resulting from the simultaneous presence in
179
the body of the prodrug, CMS, and formed colistin.
180
Urine collection. Animals were housed in metabolic cages for 6 hours of a 24 hour cycle.
181
Though longer urine collection times can yield more accurate values, based on previous
182
clinical observations with polymyxin analogs, a 6-hour collection period was used in order to
183
balance animal welfare concerns with data integrity considerations. From the pre-dose time-
184
point through the final necropsy, urine was collected daily for all groups, but only processed
185
for urinalysis (data not shown), urine chemistry and urine kidney injury biomarker analysis on
186
days 0, 2, 4 and 7.
187
Toxicokinetic analysis. Three animals from the colistin and PMBN dose groups were
188
assigned as satellite animals for the 7-day toxicokinetic evaluation. Blood samples were
189
collected from each animal (tail vein venipuncture) on day 1 and day 4 at 2 and 6 hours post-
190
dose and on day 7 at 0.33, 1, 2, 4, 6 & 24 hours. Terminal blood and kidney tissue samples
191
were obtained 3 hours after the final dose on day 8. Blood and kidney samples were also
192
obtained from the interim sample groups of rats (n=3 per treatment group) that were
193
necropsied on day 3, following 2 days of QID dosing. Blood samples from these animals
194
were taken on day 1 at 0.33, 1, 2, 4, 6, 8 & 24 hours and on day 2 at 2 & 24 hours while
195
kidney tissue samples were obtained following termination at 3 hours post-dose on day 3.
196
Satellite and interim sacrifice groups were utilized for toxicokinetic analyses; nephrotoxicity
197
endpoints were only evaluated in the main study animals, where blood sampling only
198
occurred at necropsy.
199
Plasma samples (0.05 mL) were precipitated by the addition of 95/5 acetonitrile/formic acid
200
(0.25 mL) containing fibrinopeptide B (0.1 µg/mL) as an internal standard. Samples were
201
shaken for 5 minutes, and then centrifuged for 5 minutes at 4000 RPM to remove proteins.
202
An aliquot (0.2 mL) was removed, dried under N2 at 40°C., and reconstituted in 95:5:0.1
203
ACN:water:formic acid (0.06 mL). Kidney samples were diluted 1:3 w/v with water, and
8
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 8 of 47
Page 9 of 47
204
homogenized using a tissue miser homogenizer (ThermoFisher Scientific, Waltham, MA) at
205
~30,000 RPM while on ice. The kidney homogenate was then typically diluted 5-fold into
206
plasma and processed as indicated above. Analysis was performed by LC/MS/MS on an
207
Applied Biosystems API5000 triple quadrupole mass spectrometer (AB Sciex LLC, Foster
208
City, CA) operating in the positive ion mode with a Shimadzu NEXERA HPLC inlet system
209
with a Phenomenex reverse phase, 5µ C8(2) column (30x2.0mm). For separation, mobile
210
phase A consisted of water, and mobile phase B consisted of ACN, both containing 0.1%
211
formic acid. Samples (10 µL) were injected onto the LC with a flow rate of 0.6 mL/min
212
starting at 2% B for 0.5 min then ramped to 80% B over 2 min (2.5 min total), the column
213
was flushed for 1 min then returned to 2% B and equilibrated for 1 min. Under these
214
conditions, colistin was monitored by fragmentation at m/z 602>241, PMBN at m/z 786>333,
215
and fibrinopeptide B (IS) at m/z 786>333 with a source temperature of 550°C., a declustering
216
potential of 60 V, and collision energies of 31, 35, and 38 eV respectively.
217
Clinical pathology and urinary kidney injury biomarkers. Routine hematology was
218
evaluated by Bioreliance (Rockville, MD, USA) on samples collected just prior to necropsy,
219
using the Advia 120 Hematology System (Siemens Healthcare, Malvern, PA, USA; data not
220
shown). A full plasma clinical chemistry panel and urine chemistry were evaluated by
221
Bioreliance (Rockville, MD, USA) using the Cobas c311 (Roche Diagnostics, Indianapolis,
222
IN, USA), including plasma and urine creatinine levels. All analytical methods used were fully
223
validated for rat samples. Urine samples were assayed using Meso-Scale Discovery (MSD)
224
electrochemiluminescent immunoassays in technical replicates, with responses evaluated on
225
the Sector Imager 6000 instrument (MSD, Gaithersburg, MD, USA). Kits used were the Rat
226
Kidney Injury Panel 1 (albumin, Kim-1, NGAL/lipocalin-2, osteopontin), the Argutus Acute
227
Kidney Injury Panel (α-GST, GST-µ, RPA-1) and the Clusterin Test Kit (MSD, Gaithersburg,
228
MD, USA), according to the manufacturer's instructions.
9
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
229
Pathological examination. An interim sampled group of three rats/treatment group was
230
necropsied on day 3, after 2 days of QID dosing to assess any early renal histopathology. In
231
the main study groups, six rats/treatment were necropsied on day 8, following 7 days of QID
232
dosing. After gross examination, the kidneys were weighed, collected, processed and
233
examined microscopically. Sections of liver, lung and heart were also evaluated
234
histologically. Representative kidney sections (left and right) were generated for all animals
235
in each group. Tissues were fixed in 10% neutral-buffered formalin (NBF) for approximately
236
48 hours and subsequently processed, embedded in paraffin, sectioned at 4 µm, mounted
237
on glass slides, deparaffinized, and stained with hematoxylin and eosin (H&E) and periodic-
238
acid Schiff (PAS). Kidney sections were evaluated by light microscopy and graded for the
239
degree of tubular degeneration on a 5-point scale: 0 = no abnormality detected (NAD), 1 =
240
minimal, 2 = slight, 3 = mild, 4 = moderate, 5 = marked or severe findings. The
241
histopathologic grading scheme for degeneration was defined as follows: minimal:
242
occasional degenerate (brightly eosinophilic) cells with pyknotic nuclei; slight: occasional
243
degenerate cells with pyknotic to karyorrhectic nuclei and sloughed cells within tubular
244
lumina (protein casts); mild: small clusters of 2-4 degenerate cells with pyknotic nuclei and
245
protein casts; moderate: larger clusters and chains of degenerate cells, some with complete
246
loss of chromatin, affecting numerous tubules; marked: majority of tubules affected by chains
247
of degenerate cells, or entire tubular segments affected by degeneration.
248
Kim-1 immunohistochemistry (IHC). IHC was carried out using methods previously
249
described (Wadey et al., 2013). Briefly, 4-um thick tissue sections were incubated with Kim-1
250
primary antibody (rabbit immunoglobulin G [IgG], 1:400; R&D Systems, Minneapolis MN). All
251
immunostaining was carried out at room temperature using a Labvision autostainer
252
(Labvision, Fremont, CA, USA). Rabbit and goat isotype controls (Dako, Enmark) were used
253
as negative controls. Images of IHC whole kidney sections were captured using a
254
ScanScope Scanner and analyzed using ImageScope software (Aperio Technologies Inc.,
255
Vista CA). Thresholds were set to detect DAB positivity, and the kidney sections analyzed to 10
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 10 of 47
Page 11 of 47
256
determine the proportion of positive pixels for Kim-1 immunolabeling. Prior to statistical
257
analysis, the proportion of positive pixels was transformed using the Arcsine transformation.
258
A linear regression model using GraphPad Prism version 5.0 (La Jolla, CA), compared Kim-
259
1 urine levels to the proportion of Kim-1 expression in the kidney of each individual rat.
260
Statistical analysis of urine biomarker data. Creatinine correction was calculated for urine
261
chemistry and urine injury biomarker parameters. Creatinine adjustment is an accepted
262
method to normalize urinary biomarker values to account for the wide variation in urine
263
concentration that can be seen within and between individuals over time (Alessio et al.,
264
1985). For the urine injury biomarkers and urine chemistry, mixed models were utilized to
265
account for repeat measurements at different time points pre- and post-dose. Each
266
parameter is log-transformed before modeling and log-transformed fold-change from
267
baseline is modeled. The measure of relationship between urine biomarker and urine
268
chemistry data is based on Pearson’s r correlation coefficients. All statistical analysis was
269
performed using SAS 9.2 software (SAS Institute Inc., Cary, NC, USA). Statistical
270
significance was defined as p0 (’1’). Thus, all positive grades of histopathology (grades 1 to 2) were treated with
295
equal weight for the ROC analysis. Pre-dose and vehicle groups were not included in the
296
second analysis in order to directly compare the predictability of our measured biomarkers
297
with histopathology as the classifier, but in the face of both compounds accumulating within
298
the PTECs. In ROC analysis for these urinary kidney injury biomarkers, we set a low
299
threshold (tendency to “over call”) with a high sensitivity but lower specificity (allowing for a
300
false positive rate of 0.05) because for our purposes in the evaluation of early stage
301
therapeutic compound candidates, it is more important to not miss any cases of even
302
minimal renal injury then to mis-classify a normal animal as having injury.
303
304
RESULTS
305
In vitro cytotoxicity in the HK-2 cell line. Previously published work has demonstrated the
306
sensitivity of the immortalized human proximal tubule kidney cell line (HK-2) to polymyxin-
307
induced cytotoxicity (Vaara and Vaara, 2013). We hypothesized that cytotoxicity assessment
12
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 12 of 47
Page 13 of 47
308
with HK-2 cells could stratify polymyxin derivatives according to nephrotoxic potential. Cell
309
viability upon 24 hour treatment of polymyxin B (PMB), colistin sulfate or PMBN was
310
evaluated (Figure 1). PMBN showed greater than 50-fold less cytotoxicity in the HK-2 cell
311
cytotoxicity assay than its parent compound PMB with TC50 of >1000 µM and 20 µM,
312
respectively. With a TC50 of 70 µM, colistin sulfate consistently showed at least 3-fold less
313
cytotoxicity than PMB.
314
In vivo rat toxicity studies, clinical observations. Supporting previous observations in the
315
literature (Magee et al., 2013, Wallace et al., 2008), IV infusion of PMBN and colistin at
316
doses of 1 mg/kg were not well tolerated, with rats showing many signs commonly ascribed
317
to a histaminergic response, including swollen paws, reduced motor activity and red ears.
318
Subcutaneous dosing of colistin reduced these tolerability issues, and was found to be well
319
tolerated up to 40 mg/kg in Han Wistar rats. Interestingly, the plasma histamine levels noted
320
with subcutaneous administration were not less than that incurred with intravenous dosing
321
(Figure 2), suggesting that the tolerability issues with IV dosing were not solely related to
322
increased histamine levels.
323
Toxicokinetics. For several animals, both colistin and PMBN-dosed, it was not possible to
324
obtain a sufficient blood sample at the 1 and 2 hour time-points following the first dose on
325
day 1. One satellite animal from the colistin dose group died following the first dose, but
326
necropsy results suggested that this was not directly related to treatment, but was due to an
327
underlying transmural stomach ulcer. The measured plasma concentration data for the other
328
colistin and PMBN dosed animals and the corresponding individual model fits are shown in
329
Figure 3. Steady state was observed to be reached by 24 hours and the overall level of inter-
330
individual variability in TK for both colistin and PMBN was relatively low. Derived
331
toxicokinetic parameter values are shown in Table 1.
332
Clinical pathology and renal biomarker results. Hematology was collected and evaluated
333
on the day of necropsy. Relevant observed changes were restricted to a 3.5-fold increase in 13
Downloaded from http://toxsci.oxfordjournals.org/ at Grand Valley State Univ on November 9, 2013
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Toxicological Sciences
Toxicological Sciences
334
eosinophils in rats administered 6.25 mg/kg QID colistin sulfate, consistent with clinical
335
observations of an allergic hypersensitivity where affected animals demonstrated swollen
336
limbs and muzzles, red ears, half-shut/shut eyes, decreased motor activity and general
337
distress. A full plasma clinical chemistry panel was also evaluated; blood urea nitrogen
338
(BUN) and Cr were mildly increased in colistin-treated animals by 1.2-fold and 1.5-fold,
339
respectively (see Table 2). Renal injury urine biomarkers and urine chemistry were
340
measured at pre-dose (day 0) and on days 2, 4 and 7 post-dose. As shown in Figure 4, the
341
urinary concentration or activity of several biomarkers (Kim-1, α-GST, albumin, NGAL,
342
clusterin, GST-µ, osteopontin, RPA-1) was elevated in a time-dependent manner in the
343
colistin-treated group. Among these biomarkers, albumin, Kim-1, α-GST, clusterin, GST-µ,
344
NGAL, and osteopontin showed early and significantly large alterations in concentration 2
345
days post-treatment (47-fold, 17-fold, 16-fold, 15-fold, 13-fold, 12-fold, and 4-fold increases,
346
respectively, when compared to vehicle-treated rats, p
