Mechanistic Investigations of Test Article–Induced Pancreatic Toxicity at the Endocrine–Exocrine Interface in the Rat

Abstract

Pancreatic toxicity commonly affects the endocrine or exocrine pancreas. However, it can also occur at the endocrine–exocrine interface (EEI), where the capillary network of the islet merges with the capillaries of the surrounding acinar tissue, that is, the insulo-acinar portal system. The goal of this article is to describe a novel, test article–induced pancreatic toxicity that originated at the EEI and to summarize investigations into the mechanistic basis of the injury. This injury was initially characterized by light microscopy in 7/14 day-toxicity studies in Sprague-Dawley (Crl: CD®[SD]) rats with undisclosed test articles. Microvascular injury at the interface resulted in peri-islet serum exudation, fibrin deposition, hemorrhage, inflammation, and secondary degeneration/necrosis of surrounding exocrine tissue. More chronic injury presented as islet fibrosis and lobular atrophy. Direct cytotoxicity affecting the capillary endothelium at the EEI was confirmed ultrastructurally on day 4. Endothelial microparticle and blood flow studies further confirmed endothelial involvement. Similar lesions occurred less frequently in 2 other rat strains and not in the mouse, dog, or cynomolgus macaque. In summary, in vivo and investigative study data confirmed primary endothelial cytotoxicity in the pathogenesis of this lesion and suggested that the lesion may be rat/rat strain–specific and of uncertain relevance for human safety risk assessment.

Introduction

Test article–induced pancreatic toxicity is most commonly reported to affect the endocrine or exocrine pancreas and has only rarely been reported in the literature as arising at their interface at the periphery of the islet. However, the endocrine and exocrine tissue compartments are intimately associated anatomically as well as functionally, as indicated by their interconnected blood supplies. In mammals, much of the blood that perfuses the exocrine pancreas first passes through the islets and then to the surrounding acini via the insulo-acinar portal system. This circulatory pattern allows the acinar cells to rapidly respond to hormonal and other stimuli from the upstream islets resulting in rapid and homeostatic metabolic regulation. However, there are interspecies differences in the extent to which the blood supply to the endocrine and exocrine tissues are interconnected. For example, in the rat, there is a separate direct arterial blood supply to the capillary networks of the acinar lobules as well as the islets, which merge at the insulo-acinar portal system around the islet. In contrast, in the dog and primate, most arterial branches first supply blood flow to the islets and then to the acini (Greaves 2012).

The EEI consists of a close juxtaposition of the exocrine and endocrine cells, which is intermittently traversed by small vessels emerging from the interior of the islet. These efferent vessels are either capillaries or postcapillary venules that arise from the anastomosis of fenestrated capillaries that form a complex glomerulus-like network within the islet. The efferent vessels upon leaving the islet either merge with the capillary bed of the peri-islet acini or drain into tributaries of the portal vein (Bockman 1992; Bockman, Inagaki, and Morohoshi 2005; Bonner-Weir and Orci 1982; Murakami and Fujita 1992; Figure 1). The walls of these small vessels consist of a monolayer of fenestrated endothelial cells enveloped discontinuously by pericytes (Richards, Raines, and Attie 2010).

Figure 1.

Rat islet microvascular anatomy. (A), One or more arterioles enter the islet and after bridging the mantle of non-β cells branch into a network of fenestrated capillaries within the β-cell core that exit the islet or merge into postcapillary venules at the EEI. (B), A normal rat islet stained immunohistochemically for αSMA highlighting pericytes and smooth muscle cells within the walls of the vasculature (200×; hematoxylin counterstain).

Test article–induced changes at the EEI have been only rarely reported, and these findings are mostly limited to the islet and/or its periphery. Busulfan (Myleran^®; Busulfex IV^®), an alkylating antineoplastic agent, was reported to cause peri-islet hemorrhage and fibrosis in the Wistar rat (Kaduk, HuBlein, and Siegfried 1987). Peri-islet hemorrhage, inflammation, fibrosis, and hemosiderin deposition have also been reported in preclinical toxicity studies in the rat with rivaroxaban (Xarelto^®), a Factor Xa inhibitor (Center for Drug Evaluation and Research 2011).

Hemorrhage, fibrosis, and hyperplasia, typically restricted to the islet with limited extension into the surrounding acinar tissue, also occur spontaneously in the SD rat. These microscopic findings are observed most commonly in older (>8-month old), predominantly male SD rats and are likely related to a prediabetic state in this rat strain that is prone to obesity when fed an ad libitum diet (Dillberger 1994; Imaoka, Satoh, and Furuhama 2007). Similar findings, that is, islet degeneration/fibrosis, again typically limited to the islet, also occur in rat models of type 2 diabetes (Nugent, Smith, and Jones 2008).

This article describes a novel, test article–induced, pancreatic lesion in the SD rat, which originates at the EEI and may progress to extensively involve the surrounding acinar tissue. It shares the results of a series of experimental studies conducted to identify, characterize, and investigate possible mechanisms of this toxicity, which was induced by a series of undisclosed test articles.

Materials and Methods

All in vivo studies were conducted in compliance with the U.S. Animal Welfare Act and the Institute for Lab Animal Research (ILAR) Guide for the Care and Use of Laboratory Animals (National Research Council 2011), and the procedures used were reviewed and approved by the Institutional Animal Care and Use Committee. The undisclosed test articles administered in the initial acute toxicity studies were the lead and backup candidates for a drug development program prior to the identification of this pancreatic toxicity. Subsequently, the lead test article was utilized as a tool compound to investigate this toxicity in a time course study and a series of mechanistic studies as described below. The data provided are limited to that most relevant to characterizing the lesion and developing a better understanding of its pathogenesis.

Acute Toxicity Studies

Test articles or vehicle were administered once daily by oral gavage (OG) for 7 or 14 days to 6- to 8-week old, male SD rats (n = 4–5/group). The lead test article was administered at dosages of 30, 100, or 500 mg/kg/day for 7 days and at 100 mg/kg/day for 14 days. The backup test article was administered at 40 mg/kg/day for 7 days and at 10, 20, or 40 mg/kg/day for 14 days. Mortality, clinical observations, body weight, food consumption, clinical pathology, organ weights, and macroscopic, microscopic, and toxicokinetic parameters were evaluated. Necropsies were conducted on days 8 and 15.

Time Course Study

The lead test article was administered once daily by OG at 100 or 300 mg/kg/day for 3, 7, or 14 days to 6- to 8-week-old, male SD rats (n = 10/group/necropsy time point). A single group of female SD rats (n = 10) was administered the lead test article at 100 mg/kg/day for 14 days. Mortality, clinical observations, body weight, food consumption, clinical pathology, organ weights, and macroscopic, microscopic, and toxicokinetic parameters were evaluated. Necropsies were conducted on days 4, 8, and 15 for male rats and on day 15 only for female rats. In addition, samples of the pancreas were collected for transmission electron microscopy (TEM) and immunohistochemical analyses.

Histology/Light Microscopy

For acute toxicity studies, one flat-embedded section of the tail of the pancreas was collected at necropsy, fixed in 10% neutral buffered formalin, embedded in paraffin, processed by routine histologic methods, and sectioned at 4 µm. Sections were stained with hematoxylin and eosin (H&E) and examined by light microscopy. For the time course and other mechanistic studies (descriptions to follow), the whole pancreas in 3 sections (head, body, and tail) was evaluated by the same method. Additional histochemical stains (phosphotungstic acid-hematoxylin [PTAH] and Masson’s trichrome [MT]) were performed on select pancreas sections. Light microscopic evaluation was performed by a board-certified veterinary pathologist. Lesion severity scores were graded as follows: 0 = absent, 1 = minimal (∼<20% of tissue area affected); 2 = mild (∼20–40% of tissue affected); 3 = moderate (∼40–60% of tissue affected); 4 = marked (∼60–80% of tissue affected); 5 = severe (∼>80% of tissue affected). When 3 sections of pancreas were examined, the severity score for the most affected tissue section was used for data tabulation.

Immunohistochemistry (IHC)

For both insulin and alpha smooth muscle actin (αSMA) IHC, select pancreatic sections from the time course study were sectioned at 4 µm, placed on charged slides, and air dried. Immunohistochemical staining was performed on a Discovery XT^® automated stainer (Ventana Medical Systems, Tucson, AZ). All solutions used were from Ventana Medical Systems (Tucson, AZ), unless otherwise stated. For the insulin IHC, heat-induced antigen retrieval was performed on the tissue using ethylenediaminetetraacetic acid (EDTA) buffer and a CC1 retrieval setting. Staining was performed using a heat protocol, the endogenous biotin blocking kit, and the basic diaminobenzidine (DAB) detection kit. The primary antibody, a guinea pig anti-human insulin (Cat#A0564, Dako, Denmark), was applied at a dilution of 1:50 for 32 min. The secondary antibody, biotinylated goat anti-guinea pig IgG (Vector Labs, Burlingame, CA), was used at a dilution of 1:400 for 32 min. For the αSMA IHC, no antigen retrieval was necessary, and staining was performed using a heat protocol, the Omni-map anti-rabbit detection system, and Sblock (Ventana Medical Systems, Tucson, AZ). The primary antibody, rabbit anti-human αSMA (Cat#Ab5694, Abcam, Cambridge, MA), was applied at a dilution of 1:100 for 40 min followed by the Omni-map kit for 8 min. For both IHC procedures, the slides were counterstained with hematoxylin for 4 min followed by a bluing reagent (Ventana Medical Systems, Tucson, AZ) for 4 min. After removing the slides from the stainer, they were washed in warm soapy water, dehydrated through graded alcohols, cleared in xylene, and mounted with a synthetic mounting medium.

For morphometric analysis of insulin IHC, representative sections of pancreas were scanned with a Nanozoomer 2.0 HT (Hamamatsu, Japan), image analysis algorithms were developed using Definiens Tissue Studio software (Definiens.AG, Munich, Germany), and insulin expression was quantified on acquired images by measuring the percentage of total tissue area occupied by chromogen and dividing by the total tissue area evaluated, reported as “IHC%.” The lymph nodes and the spaces between lobules were excluded from total tissue area before analysis.

TEM

At scheduled termination of the time course study, rat pancreatic tail samples (n = 2–6 males/time point; dosed with vehicle on day 0 or 100 mg/kg/day of the lead test article for 3, 7, or 14 days) were fixed in a modified Karnovsky’s fixative containing 0.1 M phosphate-buffered 4% formaldehyde (prepared from paraformaldehyde) + 1% glutaraldehyde. Samples from vehicle control or day 3 animals were harvested randomly; whereas samples from days 7 and 14 time points were harvested from macroscopically observable lesions. Select fixed samples were further minced, post-fixed in 0.1 M phosphate-buffered 1% osmium tetroxide, dehydrated in graded ethanol series, and embedded in epoxy resin. Prepared tissue blocks were trimmed, and semi-thin (0.6 µm) sections were prepared from select specimens, mounted to glass slides, stained with 1% toluidine blue, and examined via light microscopy. These semi-thin sections were used to locate islets of interest, and selected blocks were further trimmed to these subregions. Thin sections (∼90 nm) of those subregions were prepared, stained, and examined using a Hitachi H 7100 transmission electron microscope. Digital micrographs (Advanced Microscopy Techniques, Corp.) of representative areas were recorded.

Oral Glucose Tolerance Test (OGTT)

The purpose of this study was to investigate whether altered insulin/glucose regulation contributed to pancreatic lesion development. Vehicle or lead test article at 100 mg/kg/day was administered by OG daily for 7 days (n = 10 male SD rats/group). After 18 hr of fasting, oral dextrose was administered at approximately Cmax (i.e., 2 hr) after the 4th and 7th doses. Clinical observations, body weights, and clinical pathology parameters were assessed, including serum glucose and insulin concentrations at 30-, 60-, 90-, and 120-min post-dextrose administration. Necropsy was conducted on day 8 with microscopic evaluation limited to the pancreas only.

Biomarker Analyses

Pancreatic amylase (p-amylase) and lipase activities and glucose concentrations were measured using a Siemens Advia 1650 Chemistry Analyzer. Insulin concentrations were measured using the Millipore Milliplex Rat Endocrine kit. During the time course study, p-amylase and lipase activities were evaluated on days 2, 3, 4, 8, 11, and 15. During the OGTT study, p-amylase and lipase activities were evaluated at 2, 2.5, 3, 3.5, 4, 6, and 12 hr after lead test article administration to look at shorter post-dosing intervals. During the endothelial microparticle (EMP) study, glucose and insulin concentrations were measured on day 6 at predose and 1, 3, and 6 hr post dose.

Rat pancreatic lipase immunoreactivity (rPLI) was measured using an enzyme-linked immunosorbent assay (ELISA) according to the manufacturer’s procedure (Cloud-Clone Corp., Houston, TX). Serum samples were prepared by a 1:1,000 dilution using 0.02 M phosphate-buffered saline (PBS; pH 7.0 7.2). Standard curves were calculated using a Log-Log fit (SOFTMAX PRO; Molecular Devices, Sunnyvale, CA). RPLI was measured during the time course study on days 2, 3, 4, 8, 11, and 15.

Acute phase proteins (APPs; α2-macroglobulin [A2 M], α1-acid glycoprotein [A1AGP], and haptoglobin [HG] concentrations) were measured in the plasma using commercially available ELISA assays from Life Diagnostics (West Chester, PA), following the manufacturer’s instructions. To normalize reactivity of A2 M isoforms, sera were digested with trypsin and diluted 1,500-fold before being assayed. Nondigested sera were diluted 15,000-fold and 100,000-fold prior to being analyzed for A1AGP and HG, respectively. The concentration of APP in each diluted sample was determined by comparison of the measured absorbance at 450 nm to that from a standard, and the final serum concentration was corrected for dilution. The concentrations of APPs were determined during the time course study on days 2, 3, 4, 8, 11, and 15.

EMPs were evaluated in a separate in vivo study to determine whether changes indicative of endothelial injury could be observed following test article administration. Vehicle or lead test article at 100 mg/kg/day was administered by OG daily for 7 days (n = 5 male SD rats/group/time point). At 24 hr after the first dose and on day 5, whole blood was collected in sodium citrate tubes, and platelet poor plasma was isolated. Platelet poor plasma was stained with exclusion (CD45 [BioLegend] and CD42d [Becton Dickinson; BD]) and inclusion (CD31, CD106, CD54 [BD], FLK-1 [Abcam], CD146 [R&D Systems], and Von Willebrand factor [Sigma]) markers to identify microparticles originating from endothelial cells. Annexin V (BD), a marker routinely used as a marker of apoptosis, was also included. The microparticles were evaluated by flow cytometry and absolute counts determined by the incorporation of TruCount beads (BD Biosciences, San Jose, CA) in each sample. Necropsy was conducted on day 8 with microscopic evaluation limited to the pancreas only.

Cardiovascular Safety Study

Male Wistar Han rats previously implanted with a blood pressure transmitter were assigned to 2 test article (lead and backup) groups of 8 animals each, which were administered vehicle for 4 consecutive days. After 96 hr, rats were administered either the lead test article at 100 mg/kg/day or the backup test article at 10 or 30 mg/kg/day for 4 consecutive days. Clinical signs and body weights were recorded. Heart rate, mean, systolic, and diastolic blood pressure, body temperature, and relative activity were recorded using PONEMAH P3 Data Acquisition System (Gould Instruments, Inc Valley View, OH, Version 4.8 & 4.9) for ∼1 hr predose and ∼24 hr postdose.

Blood Flow Studies

To explore local vascular blood flow alterations as a possible contributor to lesion pathogenesis, a 4-day oral investigative study in male SD rats (n = 5/group) was performed with 10 and 100 mg/kg/day of the lead test article. Relative blood flow rate of total and endocrine pancreas was evaluated in comparison to control animals using injected microspheres in a similar manner to that described in the literature (Jansson and Hellerstrom 1981). On day 4, fasted rats were anesthetized with isoflurane (∼1 to 3 hr postdose) and a catheter was inserted into the left ventricle of the heart via the right carotid artery. Red fluorescent (580/605 nm) polystyrene microspheres (FluoSpheres^®, Life Technologies, Grand Island, NY) of 10 µm diameter prepared at a concentration of 2,000 beads/g of body weight (in 0.9% NaCl solution containing 0.01% Tween 20) were slowly injected via the catheter into the heart over a 15- to 20-sec period. Immediately following the microsphere injection, the rats were euthanized with an injection of sodium pentobarbital (100 mg/rat) via the catheter. The rats were necropsied, and the pancreas was removed in toto and immersed in Hank’s balanced salt solution. The pancreas was dissected free of other tissue, blotted and weighed (dry weight and volumetric weight), and a section was frozen in optimal cutting temperature compound (OCT).

The OCT blocks were cut into 50-μm sections using a uniform distance between sections, for a total of 5 to 10 sections per block. Indirect immunofluorescence microscopy was used to localize pancreatic islets and vascular endothelial cells in blood flow experiments. Pancreatic islets were labeled with rabbit anti-human CD56 monoclonal antibody (clone MRQ-42, Cell Marque, Rocklin, CA) followed by Alexa Fluor 488-conjugated goat anti-rabbit IgG (H+L) secondary antibody (Life Technologies). Endothelial cells were labeled with mouse anti-rat CD31 (clone TLD-3A12, AbD Serotec, Raleigh, NC) followed by biotinylated horse anti-mouse IgG (Vector Labs) and Alexa Fluor 350-conjugated streptavidin (Life Technologies). The fluorochrome-labeled slides were scanned with an Axio Imager microscope (Carl Zeiss Microscopy GmbH, Gottingen, Germany), and quantitative image analysis of circulating total microspheres and microspheres associated with islets in pancreas sections was performed using CRi inForm™ software.

In Vitro Cytotoxicity

Rat primary pancreatic acinar and islet cells were isolated by perfusion and enzyme digestion. Male SD rats (6–8 weeks of age) were euthanized via CO₂ asphyxiation, and the pancreas was perfused with a digestion buffer through the common bile duct. Once perfused, the pancreas was removed from the abdominal cavity and incubated at 37°C with gentle shaking for approximately 30 min. After incubation, the digested tissue was vigorously shaken in a tube containing cold Hank’s Balanced Salt Solution (with calcium) for approximately 1 min to help disperse the cells from the tissue while stopping the enzyme reaction. The cells were pelleted by centrifugation (800 rpm × 5 min at 4°C), filtered free of debris, and then the islet and acinar cells were separated by density gradient centrifugation in Histopaque 1077 medium (Sigma-Aldrich, St. Louis, MO). The cells were then plated and cultured with complete medium for 3 days, and cell medium was changed daily. On day 4, the cells were treated with various test articles at multiple dose ranges in 0.1% dimethyl sulfoxide–containing medium for 24, 48, and 72 hr. Test article–related cytotoxicity was measured using a fluorometric method quantifying the metabolic capacity of the cells.

Statistical Analysis

For in vivo studies, body weights and body weight gains, food intake, clinical pathology parameters, organ weights including ratios, and APP and rPLI concentrations were analyzed at each scheduled data collection time point for each sex separately. For the OGTT, in addition to the analysis at each sampling time point, glucose and insulin had baseline-adjusted area under the curve calculated and analyzed. For all parameters, trend tests were used to assess the differences between test article–dosed groups and the respective control group. For each parameter of interest, based on an evaluation of historical control data, the parameter was analyzed either parametrically or nonparametrically. “Parametric” parameters were analyzed using the E-bar Square trend test, and “nonparametric” parameters were analyzed via the Chi-bar Square trend test after transforming to ranks.

EMP and blood pressure data were analyzed using an analysis of variance (ANOVA). Insulin immunohistochemical morphometric and blood flow data were analyzed using a Student’s t-test.

Results

Exposure Assessment

For all in vivo studies, exposure to the undisclosed test articles was confirmed. In the acute toxicity studies, exposure increased with increasing dosages, but was less than dose proportional for the lead test article and dose proportional for the backup test article over the range of dosages evaluated. Due to the lack of disclosure of these test articles, the details of these procedures and data are not provided.

Acute Toxicity Studies

In initial acute toxicity studies with the lead and backup test articles, microscopic findings in the tail of the pancreas consisted of rare to few, small foci of degenerative and inflammatory changes affecting islets and less frequently acini as well as the adjacent interstitium. The earliest changes observed in islets were minimal, often focal loss of demarcation of the EEI associated with mixed cell inflammatory infiltrates and serum exudation. Alternatively, minimally affected islets presented with dilated blood-filled spaces at their periphery. These initial findings progressed to circumferential obscuring of the EEI by a mixture of inflammatory cells, serous exudate, hemorrhage, and fibrin deposits. More chronically affected islets were fibrotic and occasionally had adjacent hemosiderin-laden macrophages (Figure 2). Microscopic findings in the acinar tissue variably consisted of focal degeneration, liquefactive necrosis, hemorrhage, mixed cell inflammatory infiltrates, as well as, more chronically, foci of mild acinar atrophy. Clear edema, hemorrhage, and mixed cell inflammation commonly extended into the adjacent interstitium. The most distinctive stage of this lesion was given the colloquial term “exploding islet,” which consisted of large areas of acinar degeneration/necrosis and inflammation at the epicenter of which was an islet free-floating within a lake of serous exudate, fibrin, and hemorrhage. This lesion could encompass part or almost all of an affected lobule (Figure 3A–B). This range of microscopic findings was lumped under the collective diagnostic term of pancreatic degeneration/inflammation of islets and/or acini.

Figure 2.

Progression of light microscopic changes in and around the rat islet (A–D). A, Normal islet. B, Focal disruption of EEI with inflammation and hemorrhage; lead test article at 300 mg/kg/day for 14 days. C, Circumferential peri-islet serum exudation, inflammation, hemorrhage, and vasodilation; lead test article at 100 mg/kg/day for 14 days. D, Islet fibrosis with hemosiderin-laden macrophages; backup test article at 20 mg/kg/day for 14 days (A–C, 200× original magnification; D, 100× original magnification; H&E).

Figure 3.

Rat pancreas. (A), An exploding islet consisting of an islet at the epicenter of a large portion of a pancreatic lobule undergoing acinar degeneration/necrosis after 14 days of dosing with the backup test article at 10 mg/kg/day (50× original magnification; H&E). (B), Higher Figure 3. Continued—magnification view of islet in A showing an atrophied cluster of endocrine cells free-floating within a lake of serum, hemorrhage and fibrin (200× original magnification; H&E); (C), Microvascular anatomy and circulatory pattern of a rat pancreatic lobule. Separate arterial branches supply blood to the islet and acinar capillary beds, which are interconnected at the EEI.

Pancreatic degeneration/inflammation of islets and/or acini was minimal to mild and observed in 0/5, 0/5, 2/5, and 1/5 animals dosed with vehicle or the lead test article at 30, 100, or 500 mg/kg/day, respectively, for 7 days or 0/5 and 2/5 animals dosed with vehicle or the lead test article at 100 mg/kg/day, respectively, for 14 days. This lesion was also observed with the backup test article in 0/5, 2/5, 1/5, and 3/5 animals dosed with vehicle or the backup test article at 10, 20, or 40 mg/kg/day, respectively, for 14 days, but not after dosing with vehicle or the backup test article at 40 mg/kg/day for 7 days. Lesions observed were minimal to mild, with the exception of 1 animal dosed with the backup test article at 10 mg/kg/day for 14 days, which was markedly affected with widespread acinar atrophy and islets that varied in appearance from normal to hemorrhagic, inflamed and fibrotic to hyperplastic (Figure 4). Other microscopic findings in the pancreas were decreased zymogen content in acinar cells that, unlike the pancreatic degeneration/inflammation, was dose-responsive in incidence and/or severity.

Figure 4.

Rat pancreas. (A), Vehicle control, normal (12.5× original magnification). (B), Marked acinar atrophy and islets ranging from normal to hemorrhagic, inflamed, and fibrotic to hyperplastic after 14 days of dosing with backup test article at 10 mg/kg/day (12.5× original magnification). (C), Higher magnification view of B showing normal and affected islets with adjacent acinar atrophy and increased ductules (200× original magnification) (H&E).

There were no test article–related clinical observations or mortalities with 7 or 14 days of administration of either the lead or backup test article at dosages up to 100 mg/kg/day or 40 mg/kg/day, respectively. However, administration of the lead test article at 500 mg/kg/day for 7 days was associated with salivation and thin appearance as well as decreased food consumption (−37%) and body weight (−14%) compared with vehicle controls.

Time Course Study

Pancreatic degeneration/inflammation of islets and/or acini was observed light microscopically after 7 and 14 days, but not after 3 days of dosing (Table 1). In general, lesions were observed more frequently compared with the previous acute toxicity studies, likely because 3 sections of pancreas were evaluated as opposed to 1 section in the initial studies. The incidence in males after 7 days of dosing was low to moderate with a slightly higher incidence at 300 mg/kg/day compared with 100 mg/kg/day; whereas the incidence in males after 14 days of dosing was moderate to high with the higher incidence at 100 mg/kg/day compared with 300 mg/kg/day indicating a lack of a consistent dose response. The incidence in males and females after 14 days of dosing at 100 mg/kg/day was very similar suggesting a lack of effect of sex on lesion pathogenesis. The microscopic lesions correlated well with macroscopically observable focal discolorations that were red and/or white, sometimes with a target-like appearance. Lesions were observed most frequently in the tail of the pancreas, which has the greatest density of islets in the rat (Elayat, El-Naggar, and Tahir 1995; Figure 5).

Figure 5.

(A), Anatomical regions of rat pancreas (head, body, and tail). Lesions were most commonly observed in the tail, which has the greatest density of islets. (B), Macroscopic appearance of pancreatic degenerative/inflammatory lesions, which varied from red to white to red and white target-like foci.

Table 1.

Time course study.

Dosing duration		3 Days			7 Days			14 Days
Dosage (mg/kg/day)/Sex		VCM	100M	300M	VCM	100M	300M	VCM	VCF	100M	100F	300M
Diagnosis	Severity	Incidence
Pancreatic, degeneration/inflammation, islets, and/or acini	minimal	–	–	–	–	2/10	1/10	–	–	3/10	1/10	3/10
	mild	–	–	–	–	1/10	4/10	–	–	3/10	2/10	1/10
	moderate	–	–	–	–	–	–	–	–	1/10	3/10	1/10
	marked	–	–	–	–	–	–	–	–	–	1/10	–
	severe	–	–	–	–	–	–	–	–	2/10	–	–
Total incidence		0/10	0/10	0/10	0/10	3/10	5/10	0/10	0/10	9/10	7/10	5/10

Note: – = not observed; VC = vehicle control; M = male; F = female.

Incidences and severities of light microscopic observations of pancreatic degeneration/ inflammation of islets and/or acini in male and female Sprague-Dawley rats after 3, 7 or 14 days of oral dosing with 100 or 300 mg/kg/day of lead test article.

The PTAH stain confirmed the presence of fibrin in the eosinophilic material deposited around islets in the acute stage of the lesion, and the MT stain confirmed the progressively increasing deposition of collagen around hemorrhagic and fibrotic islets. αSMA IHC highlighted pericytes around vessels in normal islets from vehicle control animals and showed initial decreases followed by increased αSMA immunopositivity around islet vessels and subsequently around the islet itself in hemorrhagic and fibrotic islets, respectively, from test article–dosed animals (Figure 6). These findings were consistent with pericyte activation, myofibroblast conversion, and proliferation and further supported primary microvascular injury as an initiating event (Hayden et al. 2007, 2008; Kordes et al. 2012).

Figure 6.

Alpha smooth muscle actin immunohistochemistry. (A), Pericytes are highlighted around normal vessels in control islet. (B), Hemorrhagic islet with multifocally decreased pericyte staining consistent with microvascular injury. (C), Early islet fibrosis with increased pericyte immunopositivity associated with microvasculature. (D), Late stage islet fibrosis showing abundant αSMA immunopositive cells in and around the islet consistent with pericyte activation, proliferation, and myofibroblast conversion after 14 days of dosing with the lead test article at 100 mg/kg/day (200×; hematoxylin counterstain).

Ultrastructural examination of capillaries at the periphery of the islet at or near the EEI sporadically showed an increased frequency of constricted lumina containing entrapped and sometimes misshapen red blood cells on day 4 after 3 days of dosing. There was also occasionally a loss of demarcation of the cell membranes of the entrapped red blood cells as well as the nuclear and cell membranes of degenerating endothelial cells, which were rarely partially enveloped by collagen presumed to be deposited by an adjacent reactive pericyte that had undergone conversion to a myofibroblast. On day 8, extravasated red blood cells and their fragments were visible at and just interior to the EEI, and capillaries were reduced in number between adjacent endocrine cells (Figure 7). At later time points, ultrastructural evaluation confirmed light microscopic observations of peri-islet deposition of coagulated blood and fibrosis.

There were no test article–related mortalities and minimal test article–related clinical signs (i.e., salivation in both sexes at ≥100 mg/kg/day). However, a slight decrease (−4% on day 14) in group mean body weight was observed in males at 300 mg/kg/day and slight decreases in group mean food consumption were observed in females (up to −12% on day 3) at 100 mg/kg/day and in males (up to −15% on day 11) at 300 mg/kg/day compared with vehicle controls, which were sometimes transient and may have been due, at least in part, to interim fasting for collections of biomarker specimens.

Figure 7.

Ultrastructural changes in capillaries near the EEI. (A), Normal capillary at islet periphery in vehicle control animal on day 0 (2,500×). On day 4 after 3 days of administration of lead test article at 100 mg/kg/day: (B), Constricted capillary with entrapped degenerating red blood cell and early endothelial degeneration (2,000×) and (C), Pericyte activation with collagen deposition (*) around degenerating endothelial cell (2,000×). On day 8 after 7 days of administration of lead test article at 100 mg/kg/day: (D), Extravasated red blood cells and their fragments (arrows) are scattered at EEI (1,000×). Ac = acinar cell, En = endothelial cell, Is = islet endocrine cell, P = pericyte.

Glucose/Insulin Metabolic Studies

The glucose/insulin curve evaluated on day 6 of dosing did not demonstrate any test article–related findings. The OGTT conducted after 4 and 7 days of dosing did not clearly demonstrate a prediabetic state (i.e., test article–related increases in glucose and/or insulin) as an initiating event. Insulin IHC showed a similar pattern of diffuse moderate to strong immunopositivity of insulin-producing β-cells throughout the core of the unaffected islets in vehicle control and test article–dosed animals that received the lead test article for 3, 7, or 14 days at 100 mg/kg/day. This indicated that altered insulin production or release was unlikely to be an early initiating event in the pathogenesis of this lesion. However, partial decreases and diffuse increases in insulin immunopositivity of islet endocrine cells were observed in hemorrhagic and fibrotic islets, respectively, suggesting that altered insulin production and/or release occurred secondary to lesion development. Morphometric analysis of group mean insulin intensity and group mean insulin IHC% also did not demonstrate an effect of test article (data not shown).

Biomarker Analyses

Test article–related increases were not clearly demonstrated for any of the traditional biomarkers of pancreatic injury (i.e., p-amylase, lipase, or rPLI) or inflammation (i.e., APPs) evaluated at any time point (data not shown); therefore, they were determined to not be good serological biomarkers for this type of test article–induced pancreatic toxicity. Interestingly, test article- and dose-related decreases in p-amylase activity were consistently observed across studies, which in some cases had a microscopic correlate of decreased zymogen content in acinar cells and were inconsistently associated with decreased body weight and/or food consumption.

Test article–related increases in EMPs were observed 24 hr after the first dose and after 5 days of dosing. FLK-1⁺ CD106⁺ EMP counts were increased 12% at 24 hr and 25% (p < .01) on day 5 when comparing group mean to control mean. FLK-1 is a marker specific for endothelial cells, while CD106 is a marker specific for activation of endothelial cells. By light microscopy, 8 of the 10 test article–dosed animals had mild to marked pancreatic degeneration/inflammation of islets and/or acini on day 8; however, there was limited correlation between light microscopic lesion development and EMP increases on an individual animal basis.

Cardiovascular Safety Study

The lead test article at 100 mg/kg/day and the backup test article at 30 mg/kg/day produced biologically relevant increases in mean, systolic, and diastolic blood pressure over a 24-hr assessment period. Increases in mean blood pressure for the lead and backup test articles were up to 16 mmHg and 9 mmHg, respectively.

Blood Flow Study

While there were no detectable blood flow changes in the total pancreas between groups, there was a slight increase in blood flow through the islets with high individual animal variability at 10 and 100 mg/kg/day. However, the differences between vehicle and test article–dosed groups were not statistically significant (data not shown).

In Vitro Cytotoxicity

The lead test article and several affected backup test articles had an inhibitory concentration 50 (IC₅₀) of <1 μM for rat islet cells, but not rat acinar cells, in the in vitro cytotoxicity assay suggesting a possible contribution to lesion pathogenesis of primary islet cell toxicity. However, in contrast, one test article that was associated with rat pancreatic toxicity had an IC₅₀ of >30 μM for islet cell toxicity suggesting that this was not a key factor in lesion development. Lesion morphology also did not support primary islet cell toxicity as the initiating event. Although technically challenging, selective isolation and culture of rat islet endothelial cells is feasible (Suschek et al. 1994); however, this specific cell type was not evaluated in our in vitro cytotoxicity assay within the context of these studies.

Discussion and Conclusion

This novel test article–induced degenerative and inflammatory lesion in the rat pancreas, which originated at the EEI and affected islets and acini presented a formidable derisking hurdle due to its lack of literary precedent, infrequent and variable appearance in tissue sections, lack of a consistent dose response or biomarker, and uncertain pathogenesis and relevance to human safety. There was no conclusive evidence that the drug target was present at the site of injury suggesting that this was possibly an off-target toxicity with narrow safety margins. Given that both the lead and the backup test articles were affected, there was also a need to develop an effective screening tool to rapidly identify unaffected test articles. In vitro cytotoxicity assays with islet or acinar cells were not able to consistently identify test articles that caused the rat pancreatic toxicity, and the backup test article did not produce the light microscopic lesion until after 14 days of dosing. Therefore, a 14-day in vivo toxicity study in the male SD rat with 3 sections of pancreas examined per animal was utilized for screening of subsequent candidate test articles. In parallel with test article screening, additional studies were conducted to better characterize the lesion, determine the possible mechanisms of toxicity, attempt to identify a potential biomarker, and evaluate the possibility that this was a species-specific lesion that was unlikely to translate to humans in regard to safety risk.

The light microscopic appearance of the pancreatic lesions, especially the most subtle and presumably early initiating events, consisted of a localized obscuring of the EEI by an accumulation of blood components (i.e., serum, fibrin, red and white blood cells) at 1 pole of the islet suggesting a localized compromise of vascular integrity. Given that the efferent vessels at the EEI consist of thin-walled fenestrated capillaries and/or post-capillary venules (depending upon islet size) that are more prominent at the pole of the islet opposite the entrance point for the afferent arteriole (Bonner-Weir and Orci 1982), these are the vessels that were most likely compromised and leaking these blood components. The ultrastructural findings of increased capillary constriction with degeneration of entrapped red cells seemed in contradiction to the light microscopic findings indicating increased vascular permeability; however, the ultrastructural degenerative changes observed in the capillary endothelial cells that ultimately led to red blood cell extravasation and vessel loss were consistent with this interpretation of the light microscopic changes. In some affected islets, the light microscopic finding of peri-islet vasodilation appeared to be an early and possibly initiating event for lesion development, which may have been related to a possible test article–related increase in islet blood flow.

The “exploding islet” stage of the lesion appears to have been related to the microvascular anatomy and circulatory pattern of the rat pancreatic lobule (Figure 3C). Within each lobule, the endocrine and exocrine tissue each have their own arterial blood supplies. The exocrine branch of the arterial supply provides blood flow to the capillary bed of the tele-islet acinar tissue; whereas the endocrine branch of the arterial supply provides blood flow to the islets and the peri-islet acinar tissue via the insulo-acinar portal system (Murakami and Fujita 1992; Wayland 1997). Microvascular injury originating at the EEI would be expected to disrupt the outflow of blood from the islet into the surrounding acinar tissue and the extent of ischemic tissue injury would be related to the extent and caliber of the vessels injured, with the periphery of the lobule being the last portion to be affected.

Although this specific test article–induced lesion has not been previously reported in the literature, several rat pancreatic lesions with similar features have been published and may provide insights into this lesion’s pathogenesis. The lesion morphology, the similar lesions in the literature, and the data obtained during this series of investigative studies supported three major mechanistic hypotheses for this lesion, that is, glucose/insulin dysregulation, hypertension, and direct cytotoxicity affecting the microvasculature. One or more of these mechanisms were considered likely to be responsible for the primary injury to the microvasculature at the EEI, possibly exacerbated by increased islet blood flow, that subsequently led to secondary acinar degeneration/necrosis and atrophy.

Islet hemorrhage, fibrosis, and hyperplasia have been reported to occur spontaneously in SD and other strains of rats (Dillberger 1994; Imaoka, Satoh, and Furuhama 2007), suggesting that this test article–related lesion may be an acceleration of this common background finding. It may be caused by a primary test article–induced dysregulation of glucose/insulin metabolism, that is, β cell dysfunction or increased peripheral insulin resistance. The spontaneous lesion has a male sex predilection, has an earliest reported age of onset of 12 weeks but occurs more frequently with increasing age, and is likely related to a prediabetic state in this rat strain that is prone to diabetes and obesity when fed an ad libitum diet (Dillberger 1994; Imaoka, Satoh, and Furuhama 2007). However, this hypothesis was not supported because, unlike the spontaneous lesion, the test article–induced lesion had no sex predilection, was associated with extensive degeneration/necrosis/atrophy of peri- and tele-islet acinar tissue in addition to the islet changes, and the outcomes of the metabolic studies (i.e., glucose/insulin curve, OGTT, insulin IHC) did not indicate a primary alteration in glucose/insulin metabolism as an initiating event in this lesion’s pathogenesis. Although, secondary islet dysfunction would likely be a long-term consequence of this test article–induced pancreatic lesion due to the islet fibrosis and likely eventual decrease in functional islet cell mass. Rat models of type 2 diabetes also show a similar range of islet degenerative changes, the specifics of which are dependent upon the model, its pathogenesis, and the stage of the disease process evaluated (Nugent, Smith, and Jones 2008). Interestingly, in the prediabetic stage of the Otsuka Long-Evans Tokushima fatty (OLETF) rat model of type 2 diabetes, increased islet blood flow with concurrent normoglycemia and normoinsulinemia, all of which were demonstrated with our test article–induced pancreatic lesion, have been associated with the later development of islet microvascular injury and pathology (Iwase et al. 2002).

Fibrosis and pericyte activation/proliferation at the EEI have been observed ultrastructurally in the Ren2 rat model of hypertension (Hayden et al. 2007) suggesting that the hypertension observed with the first three test articles evaluated in this series may have contributed to lesion development. However, the Ren2 rat model does not develop light microscopically observable changes in or around the islet, but this may be due to the animal’s limited lifespan (M. R. Hayden, personal communication, August 1, 2011). Although the lead and backup test articles produced significant increases in blood pressure, the third test article evaluated in cardiovascular safety studies caused only a minimal increase but still developed the pancreatic lesion putting into question the potential relevance of this mechanism. It is also important to note that the cardiovascular safety studies were conducted with Wistar Han rats, which were not evaluated microscopically for the pancreatic toxicity that was observed in SD rats in the other studies conducted. In addition, subsequent test articles that caused pancreatic toxicity were not evaluated for cardiovascular safety. Spontaneously hypertensive rats do develop pancreatitis; however, unlike this test article–induced pancreatic lesion, the resulting inflammatory, degenerative, and fibrotic disease process is focused on the acinar tissue and not the islet, is associated with arteriosclerosis, and in its later stages is associated with decreased pancreatic and islet blood flow (Onizuka et al. 1994; Iwase et al. 2001).

Direct cytotoxicity affecting either the endothelial cells and/or pericytes of the microvasculature at the EEI was our third major mechanistic hypothesis. Peri-islet hemorrhage, fibrin deposition, and inflammation in the rat pancreas, indistinguishable from the acute stage of our test article-induced lesion, were induced by contrast enhanced diagnostic ultrasound (CEDUS) that resulted in mechanical disruption of capillaries (Miller et al. 2011). Although the CEDUS was administered uniformly to the entire organ, the most common site of the vascular injury was at the periphery of the islet extending into the adjacent acinar tissue and interlobular septa. This pattern of lesion distribution resembled that of our test article–induced lesion and suggested that these regions of the pancreatic microvasculature are particularly susceptible to injury and leakage. This may be related to the greater frequency and size of fenestrations in the capillaries within and at the periphery of the islet compared with the adjacent acinar tissue as well as the outward direction of the blood flow from the islet to the surrounding acini via the insulo-acinar portal system (Richards, Raines, and Attie 2010).

Peri-islet hemorrhage and fibrosis in the rat pancreas have been reported with oral administration of Busulfan, an alkylating antineoplastic agent. The exact mechanism of pathogenesis is not known; however, local damage to the fenestrated capillaries in and around the islets possibly exacerbated by Busulfan deposition in pericapillary pericytes has been postulated. The microvascular injury resulted acutely in hemorrhage and secondary hyperglycemia and chronically in islet fibrosis, atrophy, and hypoglycemia (Kaduk, HuBlein, and Siegfried 1987). Interestingly, Busulfan (Myleran; Busulfex IV), which has been in use in human medicine since 1959, has not been reported in prescribing information as having a safety risk for humans related to the Busulfan-induced pancreatic toxicity in the rat indicating that this lesion may be unique to the rat and not translatable to humans. However, possible interspecies differences in pharmacokinetics and metabolism would need to be taken into account to make this determination.

Peri-islet hemorrhage, inflammation, fibrosis and hemosiderin deposition have been reported in the rat with intravenous administration of rivaroxaban (Xarelto), a Factor Xa inhibitor, in association with microscopic evidence of hemorrhage in other tissues indicative of a pharmacologically mediated hypocoagulable state (Center for Drug Evaluation and Research 2011). In our in vivo toxicity studies in the rat, light microscopically observable hemorrhage was observed only in the pancreas suggesting that our lesion was not attributable to a systemic compromise in clotting ability. Additionally, systemic alterations in coagulation state were not observed; however, evaluation of coagulation-related parameters was limited to platelet count and fibrinogen concentrations.

Results from our investigative studies that further supported primary microvascular injury as the cause for our test article–induced lesion included test article–related increases in EMPs that increased over time from the 24 hr to the 5-day time point. The specific EMP phenotype that was elevated was consistent with endothelial activation. Although considered preliminary, these findings are similar to the results obtained from studies with known perpetrators of vascular injury such as the phosphodiesterase (PDE)-4 inhibitor, CI-1044 (Sokolowski et al. 2011). However, it cannot be determined whether these changes in EMPs are a primary or secondary effect of the test article. This was the sole potential serologic biomarker identified in our series of studies; however, the utility of this biomarker is limited by its inherent lack of tissue specificity unless we were able to utilize islet endothelium specific markers. Even so, a pancreatic microvascular origin for the increased EMPs appears most likely, given the lack of light microscopic evidence of vascular injury in other tissues evaluated in previous acute toxicity studies. Ultrastructural, histochemical, and immunohistochemical demonstration of pericyte activation (i.e., myofibroblast conversion), proliferation, and fibrosis associated with the islet microvasculature, and the EEI further supported our mechanistic hypothesis of primary microvascular injury.

Preclinical studies in other species (mouse, dog, and cynomolgus macaque; details not provided) were conducted with the lead test article. These studies ranged from 7 days to 1 month in duration and failed to demonstrate the test article–induced pancreatic lesion observed in the rat. Interestingly, another novel pancreatic lesion was observed in the mouse that consisted of centrilobular coagulative necrosis, which was minimal, focal, and low in incidence. In another mechanistic study exploring the possible contribution of a rat strain–specific susceptibility to pancreas-specific inducible venular leakage (Majno et al. 1987; Desemone et al. 1990), it was determined that both the Piebald Virol Glaxo (PVG) and Brown Norway (BN) rat strains were less susceptible than the SD rat to our test article–induced pancreatic toxicity. Although the pancreas-specific inducible venular leakage mechanistic hypothesis was disproven, these findings indicated a difference in sensitivities between rat strains to developing our test article–induced pancreatic lesion.

In conclusion, a novel pancreatic toxicity at the EEI was observed in male and female SD rats after 7 and/or 14 days of oral daily dosing with undisclosed test articles. The light microscopic lesion originated focally as inflammation, hemorrhage, serum exudation, and/or vasodilation at the EEI, which eventually became circumferential and progressed chronically to islet fibrosis or extended outward into the surrounding acinar tissue as degeneration/necrosis/atrophy. Lesion morphology, test article–related increases in EMPs, and ultrastructural/histochemical/immunohistochemical changes supported microvascular and, in particular, endothelial injury as the initiating event in this lesion’s pathogenesis. Increased islet blood flow may have also contributed to this outcome. The 14-day in vivo toxicity study in the male SD rat with 3 sections (head, body, and tail) of the pancreas examined was utilized as a screening tool to identify unaffected test articles. No selective biomarker was identified that could detect lesion onset and progression. This lesion did not occur in the mouse, dog, or cynomolgus macaque with up to 1 month of dosing and was less frequently observed in the PVG and BN rats compared with SD rat indicating that the lesion may be rat-specific or possibly rat strain–specific. The relevance of this test article–induced pancreatic toxicity in the rat to human safety risk assessment at this time is uncertain.