Glucagon-like peptide 1–based therapies, collectively described as incretins, produce glycemic benefits in the treatment of type 2 diabetes. Recent publications raised concern for a potential increased risk of pancreatitis and pancreatic cancer with incretins based in part on findings from a small number of rodents. However, extensive toxicology assessments in a substantial number of animals dosed up to 2 years at high multiples of human exposure do not support these concerns. We hypothesized that the lesions being attributed to incretins are commonly observed background findings and endeavored to characterize the incidence of spontaneous pancreatic lesions in three rat strains (Sprague-Dawley [S-D] rats, Zucker diabetic fatty [ZDF] rats, and rats expressing human islet amyloid polypeptide [HIP]; n = 36/group) on a normal or high-fat diet over 4 months. Pancreatic findings in all groups included focal exocrine degeneration, atrophy, inflammation, ductular cell proliferation, and/or observations in large pancreatic ducts similar to those described in the literature, with an incidence of exocrine atrophy/inflammation seen in S-D (42–72%), HIP (39%), and ZDF (6%) rats. These data indicate that the pancreatic findings attributed to incretins are common background findings, observed without drug treatment and independent of diet or glycemic status, suggesting a need to exercise caution when interpreting the relevance of some recent reports regarding human safety.
As the incidence of type 2 diabetes continues to increase worldwide, with a current prevalence of >300 million people (1), the availability of new safe and effective medicines to treat this disease is increasingly important to global health. In particular, there is a need for continued diversification of mechanisms of action that assist patients in safely achieving their glycemic goals while also providing additional potential medical benefits (2,3). Drugs that promote glucagon-like peptide 1 (GLP-1) receptor agonism, either by prolonging the duration of action of endogenous GLP-1 levels through inhibition of dipeptidyl peptidase-4 (DPP-4) or by boosting GLP-1 levels exogenously (GLP-1 analogs), represent such classes of medicines. The marketed drugs in this class (DPP-4 inhibitors: sitagliptin, saxagliptin, vildagliptin, linagliptin, alogliptin; GLP-1 analogs: exenatide, liraglutide), along with a number still in development, have been shown to have improved glycemic control as an adjunct to diet and exercise, with low risk of hypoglycemia and potential for nonglycemic benefits such as body weight loss (4,5). The safety and efficacy of GLP-1–based therapies have been established in multiple phase 3 clinical trials and continue to be monitored as a part of ongoing phase 4 trials and postmarketing event reporting (6). However, several recent clinical and preclinical publications have suggested a link between GLP-1–based therapies (collectively described as incretins, acknowledging that DPP-4 inhibitors may engage with other targets) and pancreatitis and/or pancreatic cancer (7–13). Preclinical studies conducted in small numbers of animals have attributed a variety of pancreatic findings, including ductular metaplasia and exocrine pancreatic degeneration, to incretin-based therapies in normal (euglycemic) rats and rodent models of acute pancreatitis and/or diabetes, including the HIP rat (i.e., rats transgenic for human islet amyloid polypeptide [IAPP]) (7,11,12,14). Conversely, several other publications suggest no correlation between GLP-1–based therapies and pancreatitis/pancreatic cancer (15–28). Most important, the preclinical toxicology and pathology data sets generated by multiple pharmaceutical companies reveal no drug-related adverse effects on the pancreas, including a lack of pancreatic carcinogenicity in rodents after dosing with a number of compounds that are now marketed (29–36). Some authors attributed these apparent discrepancies between studies primarily conducted or sponsored by pharmaceutical companies and those conducted by external research laboratories to a variety of reasons, including, but not limited to, perceived pharmaceutical industry bias, the inappropriateness of the pancreatic tissue region sampled and histologically evaluated, the animal species and models used, and the rigor of the assessment (11,12). However, it must be noted that, prior to the approval of a drug, numerous studies in multiple species with increasing durations of dosing are conducted, and data sets undergo rigorous evaluation for content, quality, and scientific integrity by both sponsor companies and global health authorities; yet, it must be acknowledged that health authorities depend on the data submitted by the sponsoring company and rarely examine the source data (e.g., the histopathology slides from a given study).
In the process of reviewing publications linking incretins to pancreatitis and pancreatic cancer (7,11,12), particularly photomicrographs of the lesions, veterinary toxicologic pathologists noted that the changes attributed to incretin treatment are frequently observed in placebo control rats and can represent spontaneous or background lesions. Moreover, experience with the HIP rat is limited and requires further characterization to understand the incidence of spontaneous pancreatic lesions in this model.
Accordingly, we hypothesized that some of the pancreatic changes attributed to incretin treatment by academic scientists may be confounded by an incomplete awareness of the background incidence of pancreatic changes in animals and the small number of animals used to make such conclusions. We therefore systematically characterized the incidence of background lesions in several rat strains in this study.
Reported herein are the results of our characterization of spontaneous exocrine pancreatic lesion incidence in the following three rat strains: Sprague-Dawley (S-D; a strain frequently used in pharmaceutical toxicology studies); Zucker diabetic fatty (ZDF; a well-established rat model of type 2 diabetes); and HIP (a more recently developed transgenic rat model of diabetes). The data from these findings support our assertion that the recently published pancreatic findings attributed to incretin therapies are common background findings in rats, in the absence of drug treatment and independent of diet or glycemic status.
Research Design and Methods
Male Crl:CD S-D and ZDF (ZDF-Leprfa/Crl) rats were obtained from Charles River Laboratories (Kingston, NJ). Male RIP-HAT rats (CRL:CD[S-D]-Tg[Ins2-IAPP]Soel; i.e., HIP rats) were obtained from Charles River Laboratories (Wilmington, MA) and originally were developed by Butler et al. (14). A total of 144 male rats were used in this study (72 S-D [36/diet], 36 ZDF, and 36 HIP rats). Male rats were selected because male rodent models of diabetes manifest diabetes earlier than females (14,37).
At study start, S-D and HIP rats were ∼12 weeks old, and ZDF rats were ∼8 weeks old. Younger ZDF rats were included to increase the likelihood of survival through the end of the 4-month study. Rats were pair-housed in suspended stainless steel wire-bottom cages with resting boards and maintained in environmentally controlled rooms (30–70% humidity at 64–79°F) on a 12-h light/dark cycle in an American Association for Accreditation of Laboratory Animal Care–accredited facility.
Rats were fed either a “normal diet” (Diet #5008: Certified 23% Protein Rodent Diet; 6.5% fat; Purina) or a “high-fat diet” (HFD) (Research Diets #D12492: 20% Protein Rodent Diet; 60% fat). Study initiation was defined as the switch from the pretest standard diet to the normal diet or HFD (ZDF rats were maintained on the normal Purina Diet #5008 throughout the study, consistent with the established model). In all cases, access to food and purified water was provided ad libitum.
All rat studies were approved by the Institutional Animal Care and Use Committee at Bristol-Myers Squibb.
Evaluation and End Points
Rats were evaluated over the course of 4 months. There were no pharmacologic treatments in any of the strains. Rats were checked daily for changes in clinical signs, and body weight and food consumption were assessed on a weekly basis. Food consumption was measured in grams, but was converted to calories to normalize data between the different types of food.
Bloodsamples for glucose, cholesterol, triglycerides, amylase, lipase, interleukin-6 (IL-6), and monocyte chemotactic protein-1 (MCP-1) were obtained from fasted rats prior to study start and approximately every 2 weeks thereafter, and hemoglobin A1c (HbA1c) was obtained toward the end of the study during week 17. Blood samples for hematology parameters (erythrocyte counts, hemoglobin concentration, hematocrit, mean cell volume, mean cell hemoglobin concentration, erythrocyte distribution width, reticulocyte counts, and total and differential white blood cell counts) and additional clinical chemistry parameters (liver transaminase activity, alkaline phosphatase activity, and total bilirubin, total protein, albumin, globulin, electrolytes, urea, creatinine, and fructosamine levels) were obtained from fasted rats prior to study start, during weeks 5 and 9, and prior to scheduled necropsy in week 17. Urine samples for urinalysis parameters (volume; color and clarity; pH; specific gravity; glucose, ketones, bilirubin, blood, urobilinogen, and protein levels; and microscopic evaluation of sediment) were collected overnight prior to study start and during weeks 8 and 17. Blood samples for coagulation parameters (activated partial thromboplastin time, prothrombin time, and fibrinogen level) were obtained by cardiac puncture from anesthetized rats just prior to necropsy. When possible, blood samples for hematology, clinical chemistry tests, and HbA1c measurement were collected prior to unscheduled necropsy. Hematology parameters were analyzed using an ADVIA 2120i Hematology Analyzer with Multispecies System Software (version 5.9MS; Siemens Healthcare Diagnostics, Deerfield, IL); clinical chemistry parameters, and HbA1c and urine protein levels were analyzed using an ADVIA 1800 Chemistry System (Siemens Healthcare Diagnostics); coagulation parameters were analyzed using an STA-Compact analyzer (Diagnostica Stago, Parsippany, NJ); and urinalysis parameters (except for urine protein measurement, which used a Siemens ADVIA 1800) were analyzed using a Siemens Clinitek Atlas Automated Urine Chemistry Analyzer (Siemens, Tarrytown, NJ). IL-6 and MCP-1 were analyzed using a rat cytokine/chemokine magnetic bead panel kit (EMD Millipore Instrumentation, Billerica, MA) in a Luminex 200 analyzer (Luminex Corporation, Austin, TX). Dunnett t tests, in the context of the ANOVA model, were used to compare mean post-test values to mean baseline values on a group basis. Clinical pathology data were evaluated by a board-certified veterinary clinical pathologist (M.C.P.).
At the end of the 4-month period, rats were killed by isoflurane and subsequent exsanguinations; they were then necropsied, and the entire pancreas was weighed and placed in 10% neutral buffered formalin. After fixation for at least 24 h, the pancreas was trimmed longitudinally to include the head, body, and tail; processed and embedded in paraffin; sectioned at 4 µm; and stained with hematoxylin-eosin (H-E). Additional sections were prepared as needed to ensure the section was representative of the head, body, and tail, or to better evaluate a lesion. The sections were evaluated by a board-certified veterinary toxicologic pathologist (A.M.F.), followed by peer review by a second board-certified veterinary toxicologic pathologist (N. Neef, Bristol-Myers Squibb, New Brunswick, NJ). Histologic findings of the exocrine pancreas were graded as absent, minimal, mild, moderate, marked, or severe. The Society of Toxicologic Pathologists guidelines for histopathology evaluation and histopathology peer review were followed in the conduct of the microscopic examination of the pancreas (38). Additionally, internal standard operating procedures for histopathology data reporting and peer review were followed. The peer review process does not include documentation of differences between the original and reviewing pathologists, but instead the final data interpretation represents a consensus of opinion of the study pathologist and peer review pathologist. An additional informal review of the slides was conducted by one of the authors (S.B.-W.).
Body Weight, Food Consumption, and Mortality
Body weight gain was generally comparable across strains and diets during the first 8 weeks of the study (Fig. 1). During the subsequent 8 weeks, the rate of body weight gain was higher in the S-D rats on the HFD (S-D-HFD) relative to S-D rats on the normal diet (S-D-norm), resulting in an 18% difference in terminal body weight (P ≤ 0.01). HIP rats were 7% heavier relative to the S-D-norm rats (P = NS). There was a plateau in body weight gain in the ZDF rats. The lower absolute body weight for the ZDF rats relative to the other strains (P ≤ 0.01 relative to S-D-norm rats) was due to the younger age of the ZDF rats.
Average calories consumed were initially higher in rats fed the HFD (P < 0.05 compared with S-D-norm), but declined to levels comparable to rats eating the normal diet by the second week (Supplementary Fig. 1). Over the course of the study, food consumption was variable within each strain, trending down with time in the S-D rats (S-D-HFD and S-D-norm rats) and HIP rats and trending up in the ZDF rats (P < 0.05 vs. S-D-norm). This was presumably due to the progressive development of their diabetes status and an attempt to compensate for the calories lost through glucosuria.
Half of the ZDF rats (18 of 36) were killed prior to scheduled necropsy at ∼17–23 weeks of age. The primary clinical findings necessitating euthanasia were swelling of and/or discharge from the prepuce and poor condition associated with the diabetic state (e.g., decreased activity, cold to the touch, and unkempt appearance, including chromodacryorrhea and soiling). Collectively, the gross and microscopic findings were suggestive of an obstructive uropathy, although no cause of obstruction was detectable in the tissues examined.
Diabetes Status and Clinical Chemistries
As expected, evidence of diabetes was observed in all ZDF rats throughout the study with persistently elevated levels of fasting serum glucose (mean ≤16.7 vs. ≤9.2 mmol/L in other strains), plasma fructosamine (mean ≤239.8 vs. ≤ 188.5 μmol/L in other strains), HbA1c (mean 6.5% [48 mmol/mol] vs. highest mean value of 3.9% [19 mmol/mol] in other strains), cholesterol (mean ≤5.1 vs. ≤ 2.3 mmol/L in other strains), and triglycerides (mean ≤13.2 vs. ≤ 0.9 mmol/L in other strains), along with glucosuria (2+ to 3+ glucose) in most of the rats (Table 1 and Supplementary Table 1). There were no significant increases in mean glucose levels in HIP rats, but there were three rats that developed glucosuria (3+ glucose) with accompanying ketonuria (4+ ketones) in one rat, suggesting the development of diabetes in a few rats. The HFD also appeared to be associated with slightly higher levels of plasma fructosamine by week 17 in both the S-D and HIP rats, in most cases without accompanying significant elevations of serum glucose levels relative to baseline. However, there was no relationship of serum glucose or markers of long-term glucose control (fructosamine and HbA1c) with the incidence of exocrine pancreatic changes on either an individual or group basis.
Established (amylase and lipase) and exploratory (MCP-1 and IL-6) markers of pancreatitis were also assessed. The mean lipase and amylase activities at baseline were higher in ZDF rats relative to other strains (means: 104.9 and 2,687.3 units/L, respectively, vs. ≤48.9 and ≤2,106.6 units/L in other strains) and remained higher over the course of the study. There were modest increases relative to baseline in lipase activity in the S-D and HIP strains and amylase activity in all four strains, which were considered part of normal biological variation (Fig. 2). Variability in MCP-1 and IL-6 levels was substantial across time and between strains, and thus was found to be of little value (data not shown). Importantly, there was no correlation of amylase, lipase, MCP-1, or IL-6 levels with microscopic exocrine pancreatic findings on either an individual or strain basis.
Other minor changes in hematology, coagulation, or urinalysis parameters were sporadic, transient, and not biologically significant (data not shown). Moreover, none of these effects correlated with microscopic exocrine pancreatic changes.
Evaluation of the Exocrine Pancreas
Given the variability in age (8 weeks for ZDF rats vs. 12 weeks for other strains at study start) and early euthanasia (ZDF rats), pancreas weights were primarily evaluated relative to body weight to correct (as far as possible) for these factors. Compared with S-D-norm rats, the group mean absolute pancreas weights relative to body weight were consistently decreased in ZDF, S-D-HFD, and HIP rats (−17%, −41%, and −39%, respectively). These decreases in pancreas weights were accompanied by similar decreases in brain weight and absolute pancreas weights. The reason for these decreases was not apparent based on microscopic examination; however, given that exocrine glandular tissue is the major contributor to pancreas weight, this suggests that exocrine volume is reduced in rats fed a HFD and/or as a result of diabetes. None of the rats had macroscopic lesions of the pancreas.
Pancreatic exocrine inflammation (“pancreatitis”) was diagnosed in rats in all groups; additionally, peri-islet inflammation was diagnosed in S-D rats (Table 2). Pancreatitis was morphologically similar in all rat strains irrespective of diet, and was characterized by decreased size of acinar cells (atrophy), acinar cell loss, zymogen granule depletion, single-cell necrosis, interstitial fibrosis, mononuclear or mixed inflammatory cell infiltrates, and a prominently increased presence of duct epithelial cells or duct profiles within the affected areas (Figs. 3 and 4). Peri-islet inflammation was seen in some S-D rats (higher incidence in S-D-norm vs. S-D-HFD rats). This finding consisted of a circumferential pattern of inflammation with fibrosis, mononuclear cell infiltration, hemorrhage, and/or the presence of hemosiderin (data not shown). Some foci of peri-islet inflammation were within or adjacent to foci of pancreatitis, whereas other foci were separate from areas of pancreatitis. Pancreatitis was focal or multifocal, usually in a lobular pattern; randomly distributed; and observed in 72% of S-D-norm, 42% of S-D-HFD, 39% of HIP, and 6% of ZDF rats. The lower level of pancreatitis in the ZDF rats may reflect the younger age of this group, since half of these rats were killed prior to scheduled necropsy. The susceptibility of this lesion did not appear to depend on the presence or degree of diabetes, since fasting serum glucose level was elevated only in the ZDF and a few HIP rats. Foci of pancreatitis were observed in all areas of the pancreas (head, body, and tail). In tissue sections in which the location of exocrine inflammatory lesions could be determined with confidence, 39% were in the head, 26% in the tail, 22% in the body, and 13% in multiple areas of the pancreas. Some foci of focal exocrine atrophy/inflammation occurred adjacent to islets of Langerhans, whereas other foci were separate from islets.
Pancreatic duct features of epithelial stratification/pseudostratification, epithelial papillary projections or cribriform epithelial pattern within the duct lumen, and pancreatic duct glands (PDGs) lined by columnar epithelium were also observed in 7 of 37 S-D-norm and ZDF rats with common pancreatic ducts in the sections examined, and were considered a normal finding in these untreated rats (Figs. 5 and 6).
Nonclinical safety studies in animals are a critical component of the overall risk assessment conducted when developing new drug candidates. Although findings in animals, or lack thereof, are not guaranteed predictors of risks for humans, they can highlight potential hazards that should, when biomarkers exist, be carefully monitored in clinical trials. Conversely, the absence of findings in studies conducted under strict scientific rigor in multiple preclinical species, especially at large multiples of human exposures, would suggest a lower risk and provide a relatively high confidence level for clinical safety. Part of that scientific rigor involves the careful determination of “drug-relatedness,” which involves careful assessment of changes that are over and above normal background findings or biological variation. This is a critical part of the nonclinical safety assessment process, which, if not performed with the highest level of diligence, might otherwise lead to the abandonment of compounds that could have provided significant therapeutic benefit in the treatment of serious diseases. The results presented herein illustrate one such example, where knowledge of background findings in the rat pancreas is critical to the interpretation of potential safety signals associated with incretin-based therapeutics, where conflicting outcomes exist between safety and research studies.
In reviewing recently published literature, we hypothesized that some of the pancreatic changes attributed to incretin treatment may be confounded by the background incidence of pancreatic changes in rats, especially in disease models such as the HIP rat where historical data are limited. Over the course of 4 months, three strains of rats (healthy S-D rats, and two rodent models of type 2 diabetes [ZDF and HIP rats]) were fed standard diet or HFD to assess the development of diabetes and the occurrence of pancreatic changes in the absence of drug treatment. Despite their younger age relative to the other strains, half of the ZDF rats succumbed to their progressing diabetic state and needed to be killed early. Based on clinical chemistry markers (elevated fasting glucose, increased fructosamine, and/or HbA1c) and/or the presence of glucosuria in urine, all ZDF rats developed diabetes, while a few HIP rats were developing diabetes. None of the S-D rats developed diabetes. Although the small proportion of HIP rats that were developing diabetes over the course of this study was a bit unexpected based on available data, it is clear that the age associated with elevated fasting glucose can vary considerably (11,14). However, in the absence of any drug treatment, focal pancreatitis was still observed across all strains, with the greatest incidence seen in S-D rats (up to 70% in both S-D-norm and S-D-HFD rats), followed by HIP rats, and then ZDF rats. This indicates that the incidence and severity of pancreatitis was apparently unrelated to the presence or severity of diabetes in these animals. Additionally, foci of pancreatitis were observed in all areas of the pancreas (head, body, and tail), indicating that susceptibility to pancreatitis is not exclusive to any one region of the pancreas. This contrasts with a previous assertion that adverse findings in the head of the pancreas might be missed by standard evaluation of the pancreatic body or tail alone (12).
More important, perhaps, this study has provided an opportunity to compare the observed histologic changes in the pancreas of untreated drug-naive rats to those that have been attributed to GLP-1–based therapies. For example, the histologic changes in these untreated rats are essentially morphologically identical to those described as ductal metaplasia by Matveyenko et al. (11), in which HIP rats were treated with a DPP-4 inhibitor (sitagliptin). In that study, ductal metaplasia in two of eight HIP rats treated with sitagliptin and one of eight HIP rats treated with sitagliptin and metformin (vs. zero of eight untreated HIP control rats) was suggested to indicate a causal relationship with sitagliptin treatment. The current, more extensive, study (n = 36/group) that used untreated rats (including HIP rats on the same HFD) indicates that the incidence of ductal metaplasia observed in sitagliptin-treated rats is similar to the incidence of ductal metaplasia in untreated rats, and therefore may indicate an incidental finding in the study by Matveyenko et al. (11). An additional lesion of acute pancreatitis was described in one of the three affected rats in the sitagliptin-HIP rat study. Although that lesion was large (2 cm) and grossly visible, smaller, but qualitatively similar areas of necrosis, edema, and fibroplasia in exocrine pancreatic tissue were also noted occasionally in the current study using untreated rats. The presence of this lesion in a single rat in the sitagliptin-HIP rat study is insufficient evidence to attribute causality to sitagliptin treatment. Alternatively, the extensive preterminal procedures conducted in the study by Matveyenko et al. (11), which included anesthesia and intra-arterial catheterization (and hence the possibility of thrombi causing end-organ ischemic damage), may also have caused or contributed to the more acute, extensive, and severe lesion in that particular rat.
In a study by Nachnani et al. (7), S-D rats (n = 10/group) were maintained on a standard diet and were administered the GLP-1 analog exenatide for 75 days. Exenatide-treated rats had a statistically significant score for inflammation of the exocrine pancreas compared with the control group. Although the actual incidence of pancreatic inflammation in the two groups was not presented in the article, the inflammatory lesions illustrated were essentially the same as those observed in untreated S-D rats in the current study (Fig. 4). With the low number of rats evaluated in the study by Nachnani et al. (7), and the presence of pancreatic inflammation in both treated and untreated rats, the possibility exists that these findings may be incidental and considered to be within the range of the expected occurrence of pancreatitis in untreated nondiabetic rats. Other findings, such as increased pyknotic nuclei in the exocrine tissue, may reflect nonspecific effects of large reductions in body weight gain (an expected, pharmacologically driven effect of exenatide), which was not controlled for in that and other studies (7,13).
In another publication by Gier et al. (12), expansion of the PDG compartment in S-D rats was interpreted as having a causal relationship with exenatide treatment. In the current study, duct profiles of the common pancreatic duct with a similar appearance as those in the study by Gier et al. (12) (pseudostratification of the epithelium, papillary, and cribriform epithelial projections into the duct lumen, and gland-like structures in the gland wall) were observed in untreated S-D rats, indicating that these structures occur in untreated rats as normal background findings or variations (Figs. 5 and 6). Again, the potential confounding factor of any nonspecific effects of substantially reduced food intake in exenatide-treated rats (e.g., hence, changes in exocrine secretory activity and duct content and flow) was not controlled for in that study.
In the current study, it should be noted that inflammatory lesions of the exocrine pancreas were more frequent and severe in nondiabetic S-D rats compared with the diabetic models evaluated. Nondiabetic S-D rats are routinely used in many pharmaceutical toxicology studies. This provides reassurance that routine toxicology studies provide a sensitive model to detect any drug-induced or drug exacerbation of background lesions of this nature, compared with the untreated ZDF or HIP rats evaluated here. This is further supported by the consistent lack of pancreatitis or other adverse pancreatic changes in toxicology studies using S-D rats, ZDF rats, and human IAPP mice with sitagliptin, liraglutide, and/or exenatide (22–24,26,27).
Although the cause of these background exocrine pancreatic findings is unknown, the lobular distribution of these changes may suggest a vascular (ischemia) or ductular (obstruction) component. Similar morphologic changes in the exocrine pancreas have been reported in mice after pancreatic duct ligation (39,40).
While histologic evaluation of the pancreas should ideally include evaluation of the head, body, and tail, our comparison of lesion incidence did not suggest significant differences in lesion distribution between the various regions of the pancreas. The number of animals evaluated is critical to ensure the observed background lesions are representative of the strain examined, and should be favored over the evaluation of multiple sections from fewer animals. Additionally, establishing and maintaining an adequate historical database for each strain evaluated is necessary for proper evaluation and interpretation. With these data, simple statistical methods can be used to determine the correct number of animals needed to ensure that an investigative study is adequately powered. Experienced pathologists familiar with the appearance of spontaneous lesions, together with peer review, should be used to ensure accuracy and consistency of data recording and terminology.
In conclusion, the various histological effects ascribed to GLP-1–based therapies in various academic publications are commonly seen in untreated rats, with the highest incidence of spontaneous lesions occurring in nondiabetic S-D rats. Our observations do not refute the potential for pancreatic safety issues with incretin therapies. However, they clearly suggest that studies asserting such a causal relationship need to be interpreted with caution unless studies are sized accordingly and interpreted by veterinary toxicologic pathologists familiar with the background pathology in the animals being studied. Such data also need to be interpreted with a good comprehension of the impact of nonspecific effects (e.g., reduced food consumption) on the incidence of any histological features or other end points used to make an assessment. The systematic assessment reported here also indicates that the S-D rat frequently used for pharmaceutical toxicology studies, and that the method for histological evaluation is suitable for the detection of both de novo as well as exacerbated background pancreatic changes.
See accompanying article, p. 1174.
Acknowledgments. The authors thank Mary McManus, Natasha Neef, Carol Gleason, Holly Burr, Joey Simms, Henning Jonassen, and the Drug Safety Evaluation staff at Bristol-Myers Squibb (New Brunswick, NJ) for their assistance with the conduct and evaluation of this study.
Duality of Interest. This work was supported by Bristol-Myers Squibb. At the time the work for the manuscript was planned and executed, all authors were employed by Amylin Pharmaceuticals, LLC (a wholly owned subsidiary of Bristol-Myers Squibb), Bristol-Myers Squibb, or the Joslin Diabetes Center. Some authors held stock in Bristol-Myers Squibb. S.B.-W. did not receive financial or other compensation for her histological review and input on the manuscript. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. K.D.C. oversaw the conduct of the study, and assisted in the study design, interpretation, and writing of the manuscript. A.M.F. and M.C.P. evaluated the histopathology or clinical pathology, interpreted the study findings, and wrote the manuscript. S.B.-W. reviewed the histopathology and reviewed the manuscript. R.S.M., E.J., and M.J.G. contributed to the strategic and scientific objectives, and reviewed and edited the manuscript. D.R. reviewed the manuscript. T.P.R. provided strategic input into the study design and writing of the manuscript. K.D.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Prior Presentation. Parts of this study were presented in abstract form and as a poster at the European Association for the Study of Diabetes 49th Annual Meeting, Barcelona, Spain, 23–27 September 2013; as a poster at the National Institute of Diabetes and Digestive and Kidney Diseases–National Cancer Institute Workshop on Pancreatitis–Diabetes–Pancreatic Cancer, Bethesda, MD, 12–13 June 2013; and in abstract form at the American College of Toxicology Annual Meeting, San Antonio, TX, 3–6 November 2013.