Introduction
Obesity is a worldwide epidemic with a high and increasing prevalence and can result in metabolic syndrome and increased risk of severe diseases such as diabetes, cardiovascular disease, and cancer.
1-3 To decrease the burden of obesity and its sequelae, new and efficient treatments are needed, leading to the necessity for predictive preclinical animal models of obesity.
Rodents, mainly diet-induced obese mice or rats, are used as valuable, first in-line preclinical models for evaluation of efficacy, and often, the efficacy is confirmed in a nonrodent obesity model such as obese nonhuman primates or obese pigs. Pigs share many physiological and anatomical similarities with humans,
4,5 and they easily develop obesity when fed on high-energy diets.
6 The Göttingen minipig (GM) is a small, microbiologically and genetically well-characterized strain, commonly used in both pharmacology and toxicology.
7 The female GM readily develops gross obesity when fed ad libitum on a normal chow diet,
8,9 and the female GM has been used as primary nonrodent obesity model at Novo Nordisk A/S for more than 10 years. The effect of the Glucagon-like peptide-1 receptor agonist liraglutide on food intake and body weight (BW) has been described in this model, indicating its translational potential.
9
The present article describes the occurrence of lipid embolism in 12 obese GMs which either died by acute death or had clinical signs requiring immediate euthanasia. Within a period of 7 years, 8 studies including 128 pigs have been performed with a variety of antiobesity drug candidates. The clinical signs leading to the decision of immediate euthanasia included respiratory distress and/or prolonged anorexia and the affected animals included both vehicle and drug-treated animals. In addition, 3 obese GMs, with clinical observations requiring scheduled euthanasia in the interstudy period, 13 obese drug-naive or previously treated GMs without clinical observations, and 5 lean, age-matched, drug-naive GMs were studied.
Materials and Methods
Animals, Feeding, and Housing
All studies were conducted in accordance with national regulations in Denmark, which are fully compliant with internationally accepted principles for the care and use of laboratory animals and with animal experimental licenses granted by the Danish Ministry of Justice.
In the 7-year period, the obese minipigs were housed either at the experimental animal facility at Copenhagen University (Taastrup, Denmark, Studies 1-4) or at CitoxLAB Denmark (Lille Skensved, Denmark, Studies 5-8). The typical study setup is described below and in
Figure 1.
Female GMs (Ellegaard Göttingen Minipigs A/S, Dalmose, Denmark) were selected at approximately 5 to 6 months of age, and either they were ovariectomized at Ellegaard Göttingen Minipigs A/S or later after arrival to CitoxLAB. Denmark Ovariectomy was performed in order to avoid the large variation in food consumption in relation to the reproductive cycle.
10 Obesity was induced by gradually offering increasing amounts of food approaching ad libitum levels after 9 to 10 months, aiming for a BW of approximately 80 to 85 kg at study start. In the induction period, the pigs were fed SDS minipig diet (Special Diet Services, Essex, United Kingdom), Porcine G+ (Scanbur, Denmark), or Altromin 9023 (Brogaarden, Denmark), but in all cases, the pigs were changed to Altromin 9023 at least 2 months before study start and continued on this diet during the studies. This diet contains 1.5% polyunsaturated fat corresponding to 44% of the total fat content. Domestic water was always available ad libitum. After the induction period, at an age of approximately 15 to 16 months, the pigs entered the first study. In some cases, they were terminated after the first study and in other cases, they were subjected to a suitable washout period and reused in a second study before termination (
Table 1). In the study periods, the pigs were fed ad libitum, and daily food intake was monitored using an automated system (Mpigwin/MP2, Ellegaard Systems/Mbrose, Faaborg, Denmark). In general, the ad libitum fed pigs consumed 1 to 1.5 kg per day, which correspond to 2 to 3 times the recommended amount. In the interstudy period, the pigs were fed restrictedly with 200 to 250 g Altromin 9023 twice daily to limit further development of obesity, and generally the animals had a limited weight loss in this period. In all studies, the obese pigs were allowed ad libitum access to food at least 3 weeks before study start. The lean reference pigs were fed restrictedly with 250 g Altromin 9023 twice daily. Body weight was measured on a large animal scale twice weekly (TW) during the study periods.
Treatment
Various anti-obesity drugs were tested in the model either as monotherapy or as combination therapy, as indicated in
Table 1. All the animals were treated by subcutaneous (SC) injection in the skin behind the ears once daily, TW, or once weekly depending on the compound in question. Various vehicles were used: 8 mmol/L phosphate, 184 mmol/L propylene glycol, 58 mmol/L phenol, pH 7.4 or pH 8.15; 10 mmol/L phosphate, 2% glycerol, pH = 7.6; 50 mmol/L sodium phosphate, 70 mmol/L sodium chloride, 0.05% tween 80, pH 7.4; or 8 mmol/L phosphate, 18 mg/mL propylene glycol, pH 7.4 (see Supplemental Table 1 for details). The scheduled treatment period ranged from 53 to 140 days; in 3 cases, the unscheduled deaths lead to premature termination of the study as specified in
Table 1.
Additional Study-Related Procedures
In some studies, and in the lean and obese reference pigs just before euthanasia, the body composition was determined by dual-energy X-ray absorptiometry scanning (DEXA scanning; see Supplemental Table 2 for details) and a central-venous catheter for stress-free blood sampling during metabolic tests, and exposure profiles were implanted under general anesthesia as described in Supplemental Table 2.
Clinical Evaluation
In the pigs, where clinical signs were observed, a limited clinical examination was performed. Typically, body temperature, respiratory frequency and quality, capillary refill time, heart rate, and in some cases oxygenation were obtained.
Blood Sampling and Plasma Analysis (Baseline Reference Values, Control Animals)
Blood samples for hematology and selected clinical chemistry parameters were obtained in the prestudy period in study 5 (in the anesthesia for baseline DEXA scanning), and clinical chemistry was obtained in the lean and obese reference pigs just before euthanasia to evaluate obesity-related differences in the measured parameters. For hematology, EDTA stabilized blood was taken, and hematological parameters were measured on an ABX Pentra DX120SPS (Horiba ABXdiagnostics, France). For clinical chemistry, blood was taken in plain glass tubes or in EDTA tubes, and concentration of triglycerides, total cholesterol, and albumin in the resulting serum or plasma was analyzed on a Cobas 6000 (Roche, Switzerland) according to the manufacturer’s instructions. Samples were analyzed for C-reactive protein (CrP) using luminescence oxygen channeling immunoassay. Acceptor beads were conjugated with a monoclonal antibody (mAb) toward porcine CrP (AM26061Pu-N, OriGene Technologies GmbH, Germany), donor beads were coated with streptavidin, and a mAb toward human/mouse/porcine CrP (MAB1707, R&D systems, Minnesota) was biotinylated. Five microliter of the calibrators, controls, or ×50 diluted unknown samples in pig plasma were incubated 1 hour at room temperature (RT) with a 15 µL mixture of acceptor beads (79 µg/mL) and biotinylated mAb (15 nmol/L). Then 30 µL streptavidin-coated donor beads (66.67 µg/mL) was added and incubated for 30 minutes at RT. The plates were read in an Envision plate reader at 21°C to 22°C with a filter having a bandwidth of 520 to 645 nm after excitation by a 680 nm laser. The total measurement time per well was 210 milliseconds including a 70 milliseconds excitation time. The assay had a lower and a higher limit of quantification of 20 ng/mL and 1200 ng/mL, respectively, and a pig CrP reference serum (RS-5CrP) was used as calibrator.
Necropsy and Tissue Sampling
Seven animals died unexpectedly, and 5 animals had to be immediately euthanized, these 12 animals are referred to as “unscheduled deaths” (
Table 1). Three obese animals were euthanized via scheduled termination in the interstudy period after a clinical examination before enrollment in the second study, as the animals were judged not to be fit for a second study period. These animals are referred to as “scheduled termination” in
Table 1. All animals had been obese for approximately 6 to 12 months at the time of death. Furthermore, 13 obese drug-naive or previously treated GMs without clinical observations and 5 lean, age-matched, drug-naive GMs were included, these animals are referred to as “obese reference minipigs” and “lean reference minipigs,” respectively. The obese reference minipigs had been obese for approximately 18 months.
The euthanasia was performed by an intravenous injection of the Zoletil mixture described in Supplemental Table 2 followed by exsanguination. A macroscopic examination was performed after opening the thoracic and abdominal cavities and by observing the appearance of the organs and tissues in situ. In studies 3, 4, and 5, 2 to 4 lung samples were fixed in neutral phosphate buffered 4% formaldehyde. In study 7, for the scheduled termination after study 7, and for the obese and lean reference animals, lung samples from left cranial, left caudal, right cranial, right medial, right caudal, and the accessory lobes were included. Tissues from kidney and liver were sampled in all animals. Furthermore, if abnormalities were found at necropsy in tissues/organs, these were sampled as well, which for many animals included samples of necrotic adipose tissue (retroperitoneal and/or intra-abdominal).
Histopathological Evaluation
The fixed tissues were trimmed, embedded in paraffin, and approximately 3-µm thick sections were stained with hematoxylin and eosin. The tissue sections were evaluated by the study pathologist using a light microscope. The severity of vacuolation in lung and kidney was assigned one of the following 5 grades: minimal (very few), mild (few), moderate (moderate number), marked (many), and severe (extensive number).
For demonstration of lipid, formalin-fixed tissue was soaked in 30% glucose overnight, embedded in OCT compound (Optimal Cutting Temperature, VWR, Denmark), and frozen on dry ice. Cryosections were stained with Oil Red O (Sigma-Aldrich, Denmark), counterstained with hematoxylin, and mounted with Faramount (Dako-Agilent, Denmark). Oil Red O staining was performed on lung tissue from 4 unscheduled deaths, 3 scheduled terminations, and 1 obese and 2 lean reference GMs.
The composition of lipid in the lipid emboli was evaluated by acetone treatment of cryosections from lung tissue. The sections were treated with 99.9% acetone (VWR, Denmark) for 30 minutes at either 4°C, RT, or 37°C followed by Oil Red O staining.
Immunohistochemistry
Immunohistochemistry (IHC) was performed to demonstrate macrophages with an antibody toward CD68 (rabbit antimouse CD68, Abcam, United Kingdom), lymph vessels with an antibody toward LYVE1 (rabbit antimouse LYVE1, Abcam), and blood vessels with an antibody toward von Willebrand factor (VWF; rabbit antihuman VWF, Abcam). In short, deparaffinized sections were blocked with peroxidase-blocking solution (Dako-Agilent, Denmark) followed by blocking in 3% bovine serum albumin (BSA) in tris-buffered saline (TBS; Ampliqon, Denmark). The sections were incubated 30 minutes with primary antibody diluted 1:250 (anti-CD68), 1:2500 (anti-LYVE1), or 1:8000 (anti-VWF) in TBS, 3% BSA, followed by incubation with BrightVision-HRP (Immunologic, VWR, Denmark) for 30 minutes and diaminobenzidine (DAB) + substrate chromogen system (Dako-Agilent, Denmark) for 5 minutes. Slides were washed in TBS, 0.01% Tween 20 between incubations. Finally, slides were counterstained with hematoxylin and mounted in Pertex. All incubations were performed at RT. Double staining, IHC anti-CD68/Oil Red O, was performed as described above except that cryosections were used, and incubation with DAB + chromogen was followed by Oil Red O staining.
Transmission Electron Microscopy
The tissue was processed for transmission electron microscopy (TEM) as described previously.
11 In short, the formalin-fixed tissues were refixed in cooled 0.5% glutaraldehyde and 2% paraformaldehyde and postfixed in 2% OsO
4 in cacodylate buffer (Ampliqon, Denmark) for 1 hour at RT followed by a wash in cacodylate buffer and staining with 0.5% uranyle acetate (Sigma-Aldrich, Denmark) before processing to epon (Ax-Lab, Denmark) blocks using increasing gradients of ethanol and increasing concentrations of epon in propylene oxide (Sigma-Aldrich, Denmark). The presence of lipid was verified in 1 µm semithin sections stained with 1% toluidine blue (Merck, Denmark), before preparing sections for the electron microscope. Ultrathin sections (∼70 nm) cut with a diamond knife (Diatome 3 mm) collected at copper grids, mesh 200, were examined in a Tecnai G2 Bio TWIN FP 5018/41 transmission electron microscope (FEI, the Netherlands). Images were obtained with a Veleta CCD Camera (Olympus-SIS, Germany) using Microscope User Interface version 4.6.4 TEM Imaging & Analysis version 4.7 SP3 software.
Statistical Analysis
Body weight, fat percentages, and blood parameters from the lean and obese reference groups were compared using Student
t test. In addition, triglycerides, total cholesterol, and albumin concentrations from obese pigs with pulmonary embolism of moderate, severe, or marked degree were compared to the plasma values from obese pigs without pulmonary embolism (unscheduled death and obese reference pigs) using Student
t test. All statistical analyses were done in GraphPad Prism version 8.0.0 for Windows, GraphPad Software (San Diego, California;
www.graphpad.com).
Discussion
In the present article, the occurrence of fatal pulmonary lipid embolism in obese GM is described. Overall, 5 cohorts of obese GM were included in 8 studies with different anti-obesity treatments. In these 8 studies, marked to severe pulmonary lipid embolism was seen in 10 of 12 unscheduled deaths (acute death or immediate euthanasia), and mild to marked lipid embolism was observed in 3 GMs with scheduled termination in the interstudy period due to clinical observations. Furthermore, 7 of the 13 obese reference animals that were terminated and evaluated in the poststudy periods had variable degrees of pulmonary lipid embolism ranging from minimal to marked in severity. Pulmonary lipid embolism was not detected in any of the lean aged-matched reference animals.
The clinical course of the condition varied quite a lot, from severe respiratory distress and cyanosis to no obvious clinical signs despite marked pulmonary lipid embolism in one obese reference animal. Since the acute deaths occurred when the animals were not observed, respiratory distress or other clinical signs could have been present in the period from last observation to death. Also, it should be noted that more subtle clinical signs can be hard to evaluate in these grossly obese, sedentary minipigs that spend much of their times lying down. Reluctance to get up and to exercise could be a first sign of pulmonary lipid embolism in the absence of other clinical signs like lameness but could also just be related to the obesity. Marked to severe pulmonary lipid embolism was observed in all 7 animals that were found dead in the present studies. Cyanosis was observed in 2 of these animals before death and further 3 animals that were euthanized experienced clinical signs of abdominally forced respiration and cyanosis. Severe pulmonary lipid embolism was observed in all 5 animals presented with cyanosis showing the correlation between the lipid embolism and the cause of respiratory distress and death in these animals.
Macroscopically, there were no findings in the lungs of the pigs with pulmonary lipid embolism except for small dark red areas described in a few animals corresponding to the histopathological findings of focal hemorrhage. It has been suggested that the hemorrhage can be caused by local toxic action of free fatty acids (FFAs) that are released from the lipid emboli due to action of endothelial lipases that act on the lipid emboli.
13
Microscopically, the lipid embolism presented as a thickening of the alveolar septa due to accumulation of lipid in the lung capillaries. The degree of lipid embolism varied between lung lobes in the individual animals with a tendency to be most severe in the caudal lobes. In the most affected animals, the majority of the alveolar septa in the tissue sample were involved with alveolar septal capillaries being totally occluded with lipid globules. Therefore, it is very likely that these large quantities of lipid caused direct mechanical obstruction of the pulmonary vasculature and lead to impaired gas exchange and hypoxia, observed as severe respiratory distress, and eventually to the death of the animals. Aggregation of macrophages was observed in the alveoli of some animals, and these macrophages could contain large droplets of lipid. However, aggregation of macrophages in alveoli is a common background finding in minipigs
14 and was also observed in the lean control animals of this study.
In addition to pulmonary lipid embolism, lipid embolism was also observed in glomerular capillaries of kidney suggesting that the lipid emboli are transported in the blood and trapped in the capillaries of lung and kidney.
Interestingly, all the animals with pulmonary lipid embolism also had widespread lipogranulomatous inflammation in the retroperitoneal and/or visceral adipose tissue. Very similar macroscopic and histopathological changes in form of lipid embolism in lung and kidney as well as lipogranulomatous inflammation in retroperitoneal/visceral adipose tissue have previously been observed in one pet Vietnamese potbellied pig.
15 Recently, Renner et al also described lipogranulomatous inflammation in retroperitoneal and/or visceral adipose tissue in several obese high-fat diet fed GM,
16 with no involvement of SC adipose tissue. These authors observed lipid in the lungs of the animals as well; however, in contrast to the present study, it was detected within the lymphatic vessels. In the present study, localization of the lipid in pulmonary capillaries was confirmed by TEM, whereas no lipid was detected in lymphatic vessels by IHC. The co-occurrence of lipogranulomatous inflammation with the fat embolism indicates a potential involvement of the lipogranulomatous inflammation in the pathogenesis of lipid embolism in the present study. Likely, necrosis of adipocytes will cause leakage of triglycerides, and the finding of multinucleated giant cells in hepatic vessels in 2 acutely death animals could indicate that components from the lipogranulomatous inflammation enter the blood circulation.
Three animals that had only been treated with vehicle, 2 unscheduled deaths and 1 obese reference pig, displayed either severe or marked pulmonary fat embolism. In addition, similar macroscopic and histopathological findings have been observed in the pet Vietnamese potbellied pig.
15 These observations confirm the fact that the condition is not directly caused by the given treatments, although there was a predominance of unscheduled deaths in the treated groups. The condition, however, seems to be related to prolonged periods of reduced food intake together with the presence of lipogranulomatous inflammation in the retroperitoneal and/or visceral adipose tissue. The reduced food intake was, in most cases, an expected pharmacological effect, and with the relatively higher number of pigs receiving active treatment compared to vehicle treatment, this can at least partly explain the preponderance of cases in the treated groups. In some instances, the anorectic effect seemed to be exaggerated by anesthesia in relation to DEXA scanning, and in one case (a vehicle treated animal), the anorectic period lasted only for one day just before acute death and had no obvious reason. However, it can be assumed that severe lipid embolism in itself will cause reduced food intake due to poor clinical condition.
Pulmonary lipid embolism is also seen in humans, and the clinical manifestation is referred to as fat embolism syndrome (FES).
17 Much like in the pigs, human FES may range from subclinical FES with no obvious clinical signs to subacute FES with respiratory insufficiency, petechiae, and cerebral involvement to fulminant FES, with severe respiratory failure and death although mortality solely due to respiratory insufficiency in fat embolism patients is uncommon.
17-19 The condition may be due to tissue trauma caused by, for example, bone fracture, bone surgery, bone marrow harvesting, or liposuction; to hyperlipemic parenteral alimentation or; more rarely, to metabolic conditions such as pancreatitis, fat necrosis of the omentum, diabetes, hepatic steatosis, and panniculitis.
17,20-23 In relation to trauma and hyperlipemic parenteral nutrition, large lipid particles enter the blood stream directly from the injured tissue/bone or via intravenous infusion and get captured in the pulmonary microcirculation. For the metabolic causes, there seems to be a circulating metabolic/inflammatory component that leads to disruption of the normal fine emulsion of fat in the blood stream, thereby causing the lipid to aggregate into larger spherules of fat, which will get stuck in the capillaries of, for example, the lung, kidney, and brain.
21,24,25 For instance, excessive FFA in the plasma may lead to increased formation of enlarged and unstable very low-density lipoprotein particles that may coalesce into fat droplets of embolic size.
23,26,27 In vitro, several substances or conditions may alter the physiological state of fat emulsions, for example ether, protamine, histamine, 10% glucose, pancreatic extract, bile, surface-active agents, extracts of necrotic muscle, variation in pH,
25,28,29 and acute-phase proteins like CrP.
24,30 Whether such factors are needed to trigger the coalescence of lipid particles in vivo is not completely known, but it seems very likely.
22,28,29 Importantly, hyperlipidemia is not a prerequisite for the condition to occur.
20,29
It can be hypothesized that the low food intake seen in most of the unscheduled deaths in the present study has led to mobilization of FFAs from the very extensive fat depots
31,32 and that these FFAs caused coalescence of fat droplet
19,26-28,33 as described above. This lipid aggregation could occur as a result of mere overloading of the systemic circulation with FFA, in combination with high levels of, for example, the acute-phase protein, CrP, that is released in response to various stimuli, for example, acute infection, inflammatory conditions, obesity, cancer, myocardial infarction, trauma, and so on
25,34-38 or in combination with other proaggregating circulating factors. In this case, increased CrP levels could be related not only to the obesity
39,40 but also to the extensive lipogranulomatous inflammation in the abdominal adipose tissue, potentially causing a systemic acute phase response. Unfortunately, in the present study, blood samples were only obtained from 8 of the animals with clinical signs, and FFA and CrP were not measured in these animals. High concentrations of triglycerides were, however, seen in 4 of these 8 animals, and respiratory distress and marked to severe pulmonary lipid emboli were observed in all of these 4 animals.
The cause of the lipogranulomatous inflammation in the adipose tissue is not known, but it has been suggested that the fat cells undergo necrosis due to hypoxia in the center of the fat depot.
39,41 Thus, fat may be released directly into the systemic circulation from these ruptured necrotic fat cells, further aggravating the lipid embolism. The combination of lipogranulomatous inflammation and pulmonary lipid embolism has also been observed in a horse secondary to yellow fat disease (steatitis).
42 Yellow fat disease has also been observed in cats, mink, and pigs and can be induced experimentally by feeding a diet high in polyunsaturated fat combined with low vitamin E.
43-45 Yellow fat disease is characterized by inflammation and necrosis of adipose tissue, both visceral/abdominal and SC fat, and deposition of ceroid pigment in macrophages and adipocytes. Although ceroid was not observed in the affected abdominal adipose tissue in the present study, involvement of nutritional factors such as vitamin E and polyunsaturated fat in the pathogenesis cannot be excluded as the high food intake could result in a relative high intake of polyunsaturated fat.
In conclusion, fatal pulmonary lipid embolism occurred in obese GM included in pharmacological anti-obesity studies and seemed to be related to the presence of lipogranulomatous inflammation in the abdominal adipose tissue. The data from the present study suggest that the pulmonary lipid embolism is not directly related to the test items but rather caused by a combination of factors present in the obese minipig model. This is supported by a previous report in a pet Vietnamese potbellied pig.