Comparison of Cardiovascular Pathology in Animal Models of SARS-CoV-2 Infection: Recommendations Regarding Standardization of Research Methods

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged as the viral pathogen that led to the global COVID-19 pandemic that began in late 2019. Because SARS-CoV-2 primarily causes a respiratory disease, much research conducted to date has focused on the respiratory system. However, SARS-CoV-2 infection also affects other organ systems, including the cardiovascular system. In this critical analysis of published data, we evaluate the evidence of cardiovascular pathology in human patients and animals. Overall, we find that the presence or absence of cardiovascular pathology is reported infrequently in both human autopsy studies and animal models of SARS-CoV-2 infection. Moreover, in those studies that have reported cardiovascular pathology, we identified issues in their design and execution that reduce confidence in the conclusions regarding SARS-CoV-2 infection as a cause of significant cardiovascular pathology. Throughout this overview, we expand on these limitations and provide recommendations to ensure a high level of scientific rigor and reproducibility.

Introduction to the Virus and the Pandemic

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a positive-sense, single-stranded, enveloped RNA virus that emerged as a human pathogen in late 2019. In March 2020, the World Health Organization declared a global state of emergency for the 2019 Coronavirus Disease (COVID-19) pandemic, which has since resulted in an estimated 18.2 million deaths worldwide.³⁶ Like the majority of emerging infectious diseases affecting human populations in recent history, SARS-CoV-2 is thought to have originated in wildlife.⁶¹ This emergence may have been precipitated by the progressive blurring of human–wildlife–domestic animal interfaces resulting from progressive urbanization, deforestation, disruption of wildlife habitats, and wildlife trafficking.^38,143 Without intervention, these factors are poised to contribute to the emergence of future zoonotic disease.

Pathogenesis of a virus and the consequential lesions depend on the location and expression of host proteins that the virus uses to enter the cell. SARS-CoV-2 enters the host cell via viral spike-protein–mediated fusion at the host’s angiotensin-converting enzyme 2 (ACE2) receptor. The ACE2 receptor is highly expressed in the respiratory tract, myocardium and arterial smooth muscle, gastrointestinal tract, and kidney, paralleling the organ systems affected by the virus.¹⁴ Disease severity in patients with COVID-19 ranges from asymptomatic to critically ill, although the majority of cases are considered mild.⁷⁸ Fever, coughing, respiratory difficulty, loss of smell, muscle aches, fatigue, nausea, vomiting, headaches, and diarrhea are common presenting signs of acute COVID-19 infection.¹⁴ A subset of patients—particularly those with comorbidities including advanced age, obesity, hypertension, and diabetes mellitus—commonly develop more severe clinical manifestations, including acute respiratory distress syndrome and multiorgan failure.⁷⁸ Immune response dysregulation is thought to contribute to the more severe pathogenesis responses in COVID-19, with systemic inflammation and exaggerated proinflammatory cytokine production playing a role in the development of coagulopathy, vasculitis, and multiorgan damage.^60,74,95,154

Pathology of Cardiovascular Disease in Humans with SARS-CoV-2 Infection

This review of SARS-CoV-2 infections focuses on published data from human patient and animal studies that probe effects on the cardiovascular system. According to clinical criteria (for example, elevated serum levels of troponin I, ECG or echocardiography findings), the incidence of cardiac injury among COVID-19 patients ranges from 15% to 42%, depending on patient factors.⁵³ Moreover, cardiac injury is a documented risk factor for in-hospital mortality among COVID-19 patients.¹²⁹ Other viruses, including influenza viruses, also are associated with cardiac events, such as arrythmias,²⁸ and viral infections in general should be viewed as having the potential to exacerbate preexisting cardiac disease.

With regard to the incidence of the SARS-CoV-2 virus infecting cells in the heart, one study of 39 patients who died from COVID-19 found that 61.5% were RT-PCR positive for SARS-CoV-2 RNA in cardiac tissue.⁹¹ However, the investigators could not rule out the possibility that the virus was within the vascular space (viremia) and not within the heart interstitial cells. Similar frequencies of cardiac involvement have been reported in patients who had clinically recovered from COVID-19, with 78% of patients having cardiac inflammation on MRI at time of testing.¹¹⁵ However, these MRI findings were not confirmed by biopsy.

Cardiovascular disease is one of the most common prior comorbidities reported in patients with COVID-19 infection, and cardiovascular disease increases susceptibility to severe clinical disease with COVID-19,^45,132 implicating the heart as a target—either directly or indirectly—in severe SARS-CoV-2 cases. More specifically, SARS-CoV-2 infection has been associated with acute myocarditis, with the CDC reporting that patients with COVID-19 had nearly 16 times higher risk for myocarditis.¹⁶ In addition, young men vaccinated with SARS-CoV-2 mRNA vaccines have a higher incidence of myocarditis.⁷⁷ However, relatively few published cases of COVID-19–associated acute myocarditis were confirmed by autopsy.^{24,31,52,80,109,120}

Although various clinical studies have suggested that heart injury is a common complication of COVID-19,^{4,49,70,101,121,129,164} autopsy studies assessing COVID-induced cardiac disease have been ambiguous with regard to the incidence and types of cardiac injury in patients with COVID-19, and differ in both methodology and criteria for diagnosis.^{10,42,63,66,80,91,116} In particular, variation in the criteria used to render a diagnosis of myocarditis prevents a consensus regarding the frequency of this type of cardiac injury in SARS-CoV-2–infected patients.

Myocarditis is generally defined as the presence of an inflammatory infiltrate with adjacent cardiomyocyte injury (myocyte degeneration or necrosis). Viral infections of the heart are usually associated with a lymphocyte-predominant inflammatory infiltrate.⁸⁷ However, in the absence of known infection, hearts of many elderly patients also contain scattered small collections of tissue lymphocytes without evidence for myocardial injury. Thus, the histologic presence of small aggregates of lymphocytes alone is insufficient for the diagnosis of myocarditis.⁹ Beyond lymphocytic inflammation, autopsy studies of some hearts from patients who died of COVID-19 have revealed microthrombi and inflammation, but necrosis has not commonly been reported.^63,93 In contrast with the high frequency of myocarditis diagnosed based on clinical criteria, if more rigorous criteria are used to diagnose myocarditis (inflammatory infiltrate in conjunction with the presence of myocyte necrosis), the true prevalence of myocarditis is likely relatively low (< 2%). Correlating pathologic findings and clinical data suggests that myocarditis is an uncommon cause of death in patients who die of COVID-19.^{24,31,52,80,109,120} However, because the available literature has not used standardized terminology or methods for diagnosis, additional rigorous autopsy studies might reveal subsets of COVID-19-affected patients who do die as a result of myocarditis and provide insight into this apparently uncommon consequence of COVID-19 infection and its risk factors.

Overview of Spontaneous and Experimental Models of SARS-CoV-2 Infection in Animals

Early in the pandemic, many research facilities worldwide were temporarily shut down, while those that remained operational were focused on SARS-CoV-2. Thus, a great deal of research effort pivoted to SARS-CoV-2 studies. At the time of the writing of this manuscript, a PubMed search using the keyword ‘SARS-CoV-2’ yielded 207,458 publications. This intensive research effort has led to timely establishment of animal models for SARS-CoV-2. These models provided critical insights into SARS-CoV-2 pathogenesis, which were key to the development of appropriate outbreak management and biocontainment strategies, and for investigating therapeutic and preventative measures, including vaccine development.¹⁶¹

Animal models of COVID-19 infection show significant species variation in susceptibility to infection, clinical presentation, viral shedding, and transmission potential. For example, differences in ACE2 receptor structure between animal species determine host susceptibility to SARS-CoV-2 viral entry into cells.¹⁶³ Animal models are important both for understanding the pathogenesis and sequelae of COVID-19 infection and for addressing the applicability of research findings in various animals to the human situation.

Mice (Mus musculus) are one of the most common animals used in biomedical research⁶⁸ due to their well-characterized genome and molecular pathways, and the availability of commercial diagnostic reagents and standardized protocols in mice. However, due to differences in the structure of the murine ACE2 receptor, mice are not naturally susceptible to COVID-19 infection. Mouse-adapted viral strains of SARS-CoV-2 have been developed through a variety of techniques, including genomic (host and virus) modification and viral transfection, thus allowing mice to be utilized as an animal model for COVID-19 infection.^5,40,163 However, in order to study natural infection, researchers have also developed non-transgenic models using other animal species.^30,130

Rather than mice, golden (Syrian) hamsters (Mesocricetus auratus) have emerged as the leading non-transgenic model for studying COVID-19 pathogenesis, due to the natural susceptibility of this species to SARS-CoV-2 infection.^26,58,72,131 In golden hamsters, overt respiratory signs are typically absent, and the primary clinical signs are weight loss, hunched posture, squinting, and lethargy. These clinical signs typically resolve after several days and the infection is cleared, with virus undetectable by 5 to 7 d after inoculation. Gross and histopathologic features of respiratory tract inflammation and pulmonary pathology in hamsters mirror the findings in mild to moderate human cases of COVID-19,^{26,58,72,86,103,108,131} making golden hamsters a useful and reproducible model for research studies. The Roborovski dwarf hamster (Phodopus roborovskii) also shares the same susceptibility to infection after intranasal inoculation,¹⁶² but the dwarf hamster is reported to develop clinical respiratory signs and more prominent alveolar damage and microthrombi.¹⁴⁶ In addition, an inbred hamster substrain, SH101, was recently reported to develop lesions in the brain and liver, thus making this strain a good potential model for COVID-19 systemic disease.¹⁶²

Various nonhuman primate (NHP) species with an ACE2 receptor conformation similar to that of humans are also susceptible to SARS-CoV-2 infection.¹⁶³ Given this susceptibility, as well as anatomic and physiologic similarities to humans, NHPs are important large animal models for COVID-19. Rhesus macaques (Macaca mulatta),^{27,105,126,158} cynomolgus macaques (Macaca fascicularis),^108,112,147 African green monkeys (Chlorocebus sabaeus),¹⁵⁵ pigtail macaques (Macaca nemestrina), and squirrel monkeys (Saimiri spp.) have all been used to study SARS-CoV-2 pathogenesis.⁵⁵ However, despite the development of pulmonary lesions in some species, cardiovascular lesions have rarely been reported, and NHP inconsistently show clinical signs.^{96,104,118,126,130,155} Differential susceptibility of Old World and New World primates has also been reported, with rhesus macaques appearing more susceptible than cynomolgus macaques.⁹²

Natural and experimental SARS-CoV-2 infections also occur in cats (Felis catus),^6,18,55,111 dogs (Canis lupus familiaris),^128,133 ferrets (Mustela putorius furo),^82,128 farmed American mink (Neovison vison),^43,64 rabbits (Oryctolagus cuniculus),¹⁶³ and other susceptible species.¹²⁷ In zoological collections, SARS-CoV-2 infection has been documented in large cats⁶⁹ and in Western lowland gorillas (Gorilla gorilla gorilla) at 2 zoos (San Diego Zoo¹²⁷ and Zoo Atlanta¹⁶⁵). In addition, wild white-tailed deer (Odocoileus virginianus) are also susceptible to SARS-CoV-2 infection.^{34,62,65,94,110,139} In domestic cats and dogs, myocarditis has been well documented clinically in association with natural SARS-CoV-2 infection, as diagnosed through the presence of acute onset of clinical signs of congestive heart failure, ECG abnormalities, echocardiogram findings, and elevated cardiac troponin I in animals without a prior history of cardiac disease.^29,46 In contrast, little data is available regarding the incidence of cardiac pathology associated with natural SARS-CoV-2 infections in other animal species.

Evaluation of Cardiovascular Pathology in Experimental SARS-CoV-2 Infections

Given the clinical evidence suggesting that the cardiovascular system may be involved during SARS-CoV-2 infections, we conducted a literature review of animal-based studies of COVID-19 to evaluated the effect of infection on the development of cardiovascular pathology. Few SARS-CoV-2 studies in animals have focused primarily on the cardiovascular system. Of those that did address the cardiovascular system, few concluded that cardiovascular pathology was present. Moreover, our review identified limitations in the methodology and interpretations of the animal experiments that did report cardiovascular pathology in SARS-CoV-2 infections of animals. These limitations undermined our confidence in the extent of cardiovascular viral infectivity and subsequent cardiovascular lesions in animals. Misinterpretation in veterinary pathology is a common problem that has been reviewed elsewhere.^{7,17,23,73,145,151} Here, we outline several components of the scientific investigation process that could be improved to increase the rigor and reproducibility of COVID-19 research using animals. In addition, throughout the remaining sections, we expand on some of these components, provide examples, and offer recommendations.

Increasing Rigor and Reproducibility in SARS-CoV-2 Infection in Animals: Common Problems and Examples

Lack of pathology expertise.

One common problem is that research groups may not have consulted with veterinarians and pathologists in the evaluation of study design and study material. Including pathology experts with experience with background lesions of the examined species is critical for interpretation of pathology in animal models of SARS-CoV-2 infection. The veterinary pathology community is aware that myocarditis, necrosis, and fibrosis are background findings in macaques.^25,33 Myocardial lesions, such as degeneration and fibrosis, are also background findings in hamsters,⁹⁷ with myocardial inflammatory cell infiltrate observed as a background lesion in 6% of hamsters.⁹⁸ Experimental groups of hamsters and macaques should be large enough to provide a well-powered background lesion survey. Otherwise, background lesions may be misinterpreted as being caused by the etiologic agent being tested. For example, the authors of one study noted a left atrial thrombus in a golden hamster that died after SARS-CoV-2 infection¹²⁵ but failed to acknowledge the high spontaneous background incidence of atrial thrombi in this species.^97,98,125

Importance of preparing high-quality figures and legends.

The published figures in some reports of animals infected with SARS-CoV-2 are inadequate for critical evaluation, and cardiovascular pathology in particular is often difficult to interpret due to poor image quality. For example, one group reported that virus is present at 3 d after infection only in hearts of golden hamsters with positive RT-PCR findings, but negative for virus based on immunohistochemistry and in situ hybridization, with no positive viral signals at days 5, 7, 18, and 37 after infection.¹³⁵ In the legend and text, the authors described hearts as showing mild focal myocardial degeneration and inflammation compared with the normal control myocardium. However, the provided images showed no apparent cardiovascular lesions in either control or treated animals. Thus, the text and the legend—and ultimately the conclusions—did not accurately reflect the data shown in the figures. In another study, authors reported uniform lethality after experimental SARS-CoV-2 infection in so-called cardiomyopathic J2N-k golden hamsters. However, reporting of the pathology results and cause of death was inadequate, and further information was not given,⁸⁹ thus reducing confidence in these conclusions. In both examples, inappropriate conclusions were made with regard to pathologic findings, and the reviewers and editors apparently lacked the appropriate pathology expertise to evaluate the studies.

Does viral infectivity coincide with cardiovascular lesions?

The majority of SARS-CoV-2 models do show that the virus is present in lung tissue at least one time point during the time course of the study. However, defining viral tropism based on tissue RT-PCR positivity is confounded in animals whose vessels have not been perfused, because virus from cells within the tissue cannot be differentiated from virus present only in the blood.¹⁴² A more rigorous approach is to perfuse an animal with PBS to remove the blood cells from within vessels in organs before performing methods of viral identification such as RT-PCR analysis and genomic sequencing. Many reports have claimed to find viral tropism in a tissue based on evaluation of animals that were not perfused, thus preventing determination of true organ infection.^21,56,57 Other methods of viral identification within organs can also be problematic. For example, positive immunohistochemistry labeling for virus was reported in the myocardium of a single rhesus macaque infected intratracheally, but the figure supporting this claim showed weak specific and high background staining.³⁹ In addition, the study did not illustrate or report myocarditis, cardiomyocyte degeneration or necrosis, thus calling into question conclusions regarding myocardial viral tropism in this one animal. Notably, clearly demonstrating the presence of virus within individual myocardial cells based on positive immunohistochemistry could constitute strong evidence of viral infectivity, even in the absence of cardiac pathology, providing that this was found in multiple animals.

Another example of inadequate documentation of virus infectivity is the description of cardiovascular lesions without demonstrating the presence of virus. In one study in rhesus macaques,¹²⁶ the authors reported varying levels of edema and inflammatory cell infiltration in multiple tissues, including the heart, on day 3 after inoculation, and diminished injury on day 6, but the figures provided do not match the descriptions and thus do not support the conclusions. Importantly, that particular study¹²⁶ provided no documentation of the presence of viral antigen or genome, and the authors failed to consider background inflammation. Another study in the golden hamster diagnosed myocarditis and myocardial microthrombi, but supporting figures were not included, nor was evidence of viral infection of the heart.⁵⁰ Demonstrating viral infectivity in an animal model at some point after inoculation should be a standard for all published work but is not always included. A rigorous experimental approach, following the guidelines of a modernized Koch’s postulates (etiologic agent A causes lesion B) would include isolation or identification of virus from cells within the lesions that are attributed to the virus.

The current body of literature on SARS-CoV-2 in animals likewise does not sufficiently account for the pathogenicity concept of linking viral tissue tropism as either cause or effect of tissue injury by the virus. A question that should always be considered is whether observed lesions are caused by the virus itself, or whether the lesions are caused by a secondary biologic process, such as cytokine release syndrome (a phenomenon of hyper-inflammation involve elevation of various cytokines that is believed to contribute to increased myocardial oxygen consumption, endothelial dysfunction, and suppressed cardiac function).^102,141,159 Questions that should also be addressed in the literature are whether the virus present in the tissue with pathology, and whether the virus is present in a tissue in a feasible time frame for virus-induced cytopathology.

Confirmation of cardiac hypertrophy in SARS-CoV-2 infection studies.

Cardiomyocyte cell death stresses the remaining cardiomyocytes of the heart pump. In response, surviving cardiomyocytes can become enlarged (that is, hypertrophic). Documentation of cardiac hypertrophy in the context of SARS-CoV-2 infection can therefore be used to support potential effects of the virus on the cardiovascular system. However, hypertrophy is common in aging animals and can be a background lesion, not caused by SARS-CoV-2 infection. A standard method used in cardiovascular research with rodents defines cardiac hypertrophy at necropsy by dividing the heart weight (in milligrams) by tibia length (in millimeters).¹⁵⁷ In NHP, heart weight as a proportion of body weight and myocyte cross-section measurements are used to assess hypertrophy. However, the heart weight:body weight ratio is not as accurate as heart weight:tibial length, given that body weight can vary due to the species, genetics, diet, and body condition. Thus, conclusions of heart hypertrophy in studies of SARS-CoV-2 infection that did not use the heart weight:tibial length method should be interpreted with caution, and overall changes in body weight should be evaluated as a potential confounder. According to the ventricular mass:body mass ratio, ventricular hypertrophy was present at 7 and 14 d after SARS-CoV-2 challenge in infected compared with unchallenged golden hamsters.¹¹⁷ However, the authors did not correct for the fact that weight loss after SARS-CoV-2 infection is a consistent feature of the disease in this species, confounding their interpretation (that is, tibial length does not change despite weight loss).²⁶ This conclusion—that hearts develop hypertrophy after SARS-CoV-2 infection—is not supported by the data and was apparently a result of a misunderstanding of the experimental model.

Other methods also can lead to incorrect diagnosis of cardiac hypertrophy. Some studies have misinterpreted ratios of ventricular space to ventricle wall thickness based on the assumption that ventricular space reduction, as characterized by thickening of the ventricular wall and interventricular septum, implies hypertrophy.¹¹⁷ However, the methods used to make these conclusions must be analyzed carefully. When a heart ceases to contract after euthanasia or natural death, it stops in either systole or diastole. Rigor mortis and fixation can affect these features of the heart. Due to this variability, scientists who are not experienced with anatomic pathology may interpret systole as heart hypertrophy or confuse a heart that stops in diastole with dilated cardiomyopathy.¹²⁴

Determining chronicity of lesions in the cardiovascular system with SARS-CoV-2 infection.

Because interstitial collagen deposition reflects prior or chronic inflammation in the heart, some studies use special stains to identify cardiac collagen deposition. For example, Masson trichrome staining has been used to identify perivascular collagen in a study of SARS-CoV-2 infection in golden hamsters, but the reported timeline of collagen deposition was not consistent with the acknowledged time course for collagen deposition after myocardial injury.¹³⁵

When fibrosis is a component of a lesion, cardiomyocyte injury should also be assessed because it typically precedes fibrosis. Serum troponins are commonly used as biomarkers of cardiomyocyte damage, and troponin I is an excellent biomarker for cardiomyocyte damage in rodents and NHP.¹¹ Appropriate control subjects should always be used to determine the baseline levels of troponin release, but in at least one study in golden hamsters, troponin levels were compared only between treated and untreated infected animals, with no uninfected controls.¹³⁵ That particular study presented no evidence for viral infection of the heart, and intraluminal clotted blood in the heart of a hamster that recovered was misidentified as “mild focal myocardial degeneration” when compared with the normal control myocardium,¹³⁵ thus suggesting that a pathologist was not consulted for data interpretation.

Importance of differentiating vessel inflammation from leukocyte transmigration.

In rodents, pulmonary arteries are easily delineated from pulmonary veins by the smooth muscle media of the arteries, in contrast to the cardiac (striated) muscle media of the pulmonary veins.⁸¹ This morphology helps differentiate these vessels, which could be targets of inflammation or serve as a conduit for leukocytes to enter the lung during infection. Leukocyte entry into the lung from the bloodstream is important for combating infectious disease and is classically described as occurring via transmigration of leukocytes through the thin walls of capillaries and postcapillary venules.¹ Movement of leukocytes through these vessels can be easily misinterpreted as inflammation.

The host response of leukocyte paving and transmigrating from small pulmonary arteries and veins to alveolar spaces is another important process that is an expected host response to infection in the lung parenchyma. In numerous published reports, this process of leukocyte paving and transmigration has been mischaracterized as vasculitis or “endothelialitis” (sometimes called “endothelitis”), despite the clear absence of viral antigen or genomic DNA in the affected vessels.^{12,41,59,88,108,144}

Typical of most reports of so-called endothelialitis or vasculitis in golden hamsters, one of the most detailed investigations involved histology, immunohistochemistry, and transmission electron microscopy evaluated pulmonary vessels at 1, 3, 6, and 14 d after intranasal inoculation of hamsters with SARS-CoV-2.³ The study reported that SARS-CoV-2 infected hamsters showed endothelial hypertrophy, endothelialitis, and vasculitis, with inflammation mainly consisting of macrophages with lower numbers of T-lymphocytes and neutrophils infiltrating the vascular walls and the perivascular region at 3 and 6 d after inoculation. The authors defined endothelial hypertrophy as endothelial cells bulging into the vascular lumen, endothelialitis as macrophages and lymphocytes directly beneath within or the endothelial cell layer, and vasculitis as inflammatory neutrophils, macrophages, and lymphocytes in small (< 100-µm diameter) and medium (100- to 200-µm diameter) vessels (artery or vein was not specified). Vasculitis was reported as early as 1 d after inoculation, peaking at 3 d, and “resolved” at 14 d.³ However, our review of the provided images found no convincing evidence of an actual inflammatory process targeting the blood vessels in these hamsters; instead, the images were consistent with inflammatory cell movement through vessels walls and into the lung parenchyma during an infection. Pathologists vary on the diagnostic criteria used to differentiate vasculitis and leukocyte vascular transmigration. Providing the criteria for this diagnosis within the publication would allow readers to understand the rationale that supports this diagnosis, especially given the differences of opinions in this regard.

To differentiate between the process of normal transmigration of inflammatory cells from an infection with vasculitis, pathologists must consider the following questions: 1) does the response resolve? 2) does the response occur early after infection at a time when inflammatory cell recruitment to the lung is expected, and 3) do the vessels show a morphologic change at a later time point after viral infection? The previously mentioned report³ included no descriptions or images of damage to the vessel walls, such as medial muscle cell necrosis, hemorrhage, or insudation of plasma proteins (fibrinoid necrosis). Moreover, positive medial immunostaining for complement, fibrin, or (hamster) immunoglobulin was not described, and later time points were not reported to show replacement fibroplasia, medial or intimal hypertrophy, or loss of elastic laminae. Rather, in this example,³ the authors solely reported the complete disappearance of detectable pathologic changes at the end of the investigation period of 14 d after inoculation, when neither the vascular wall nor lumen showed significant numbers of inflammatory cells. Indeed, a transcriptomic study of SARS-CoV-2–infected hamster lungs found endothelial cell expression of chemoattractant chemokines, Icam1, and Vcam1,¹⁰⁶ thus supporting the fact that this transmigration process is part of the body’s response to clear the infection, and vessels are left unharmed.

Another important point worth discussing is determining the causal relationship of virus infectivity in a cell and the cytopathic change in the cells that make up the vessel. In the study discussed earlier,³ the authors noted a wide zone of absent immunostaining for SARS-CoV-2 nucleoprotein in and surrounding these vessels. Based on the Chapel Hill consensus conference on vasculitides:⁷⁵ the so-called inflamed vessels are clearly uninfected in this model,³ therefore, this outcome cannot be considered an infectious vasculitis per se. Because the so-called inflammation is present at 24 to 72 h after inoculation, it cannot reasonably be interpreted as an adaptive immunity–mediated process, unless all of these hamsters uniformly had preexisting reactive B or T lymphocytes or both). Rather, a genuine autoimmune disease develops weeks after a viral trigger. Furthermore, any autoimmune vasculitis triggered by the virus would continue unabated in the absence of treatment after resolution of infection, given that these processes are not self-limited. Notably, in this example,³ the virus was cleared, and no chronic lesions were ever found in vessels.

Not all descriptions SARS-CoV-2 infections of hamsters have described pulmonary endothelialitis or vasculitis.^{26,48,114,119,131,135} A detailed ultrastructural study of the golden hamster SARS-CoV-2 model did not document viral infection of the endothelium or endothelial cell damage.¹⁵⁶ A number of papers with complete and thorough necropsy examinations of infected hamsters have likewise failed to note any systemic cardiovascular lesions.^26,103,131

Similar vessel interpretations have been published regarding the pathology of NHP with SARS-CoV-2 infection. For example, endothelialitis was reported in 4 rhesus macaques infected with SARS-CoV-2² and in the lung endothelium of 4 Cambodian-origin cynomolgus and 4 Chinese-origin rhesus macaques, with mononuclear and neutrophil infiltrates in the intima of many vessels, vessel edema, and focal hemorrhage.⁸⁴ As for the hamster studies discussed above, these publications provided no evidence documenting viral targeting of the endothelium.

In contrast, pulmonary endothelialitis, vasculitis, and thromboemboli were not observed in a comparative review of SARS-CoV-2 lung lesions in 4 NHP species (rhesus macaques, pigtail macaques, African green monkeys, and squirrel monkeys).³² Numerous other studies of SARS-CoV-2–infected rhesus, cynomolgus, and pigtail macaques; African green monkeys; and marmosets document absence of lesions or viral infection in the cardiovascular system, including pulmonary vessels.^{15,19,37,76,92,99,104,105,118,123,137,149,150,152} Multiple studies of SARS-CoV-2 in ferrets similarly report no cardiovascular infection or lesions.^{51,82,83,122,148,160}

Evidence of coagulopathies in SARS-CoV-2 animals.

Viral infection of endothelial cells and inflammation of vessels are important in the pathogenesis of SARS-CoV-2. Coagulopathy is a logical result of these events. COVID-19–associated coagulopathy has been studied in human patients, with documented involvement of venous, arterial, and microvascular systems.¹³ One human study proposed that hypercoagulability may enhance SARS-CoV-2 viral entry,⁷⁹ and thromboprophylaxis may improve outcomes for critically ill COVID-19 patients.⁹⁰ However, reports of thrombosis are rare in animal studies. Macroscopic or microscopic pulmonary thrombosis were absent in lung tissue from SARS-CoV-2 infected ferrets, and mass spectroscopy demonstrated stable plasma coagulation factors, suggesting that ferrets do not develop thrombotic sequela in response to SARS-CoV-2 infection.⁸⁵ Unfortunately, that study did not perform functional testing of coagulation (for example, prothrombin time, activated partial thromboplastin time).

A hamster study evaluating SARS-CoV-2 infectivity reported “virus infection-induced vasculitis,” despite no detection of viral antigen in the perivascular tissues or endothelium in the lungs of infected hamsters at any time point.¹⁰⁷ The authors did note a coagulopathy in infected hamsters, as determined by prolonged prothrombin times—based on human point-of-care strip tests that were not validated for use in hamsters—and by phosphotungstic acid–hematoxylin staining of the lung, without actual demonstration or reporting of any actual intravascular thrombi.¹⁰⁸ Salient details regarding blood collection (for example, site, anticoagulant) were not reported, and the collection of blood after cervical dislocation (and release of tissue factor) may be a confounding feature.

Recommendations

Since the start of the COVID-19 pandemic, a focus on SARS-CoV-2 research and the development of animal models has generated a large body of research that has undoubtedly contributed toward clinical benefits for COVID-19 patients. However, multiple opportunities remain to improve rigor and reproducibility in animal studies of SARS-CoV-2. Although clinical patient studies have demonstrated a concern for immune-mediated disease as a sequela of SARS-CoV-2 infection and subsequent development of vasculitis and coagulopathy, the current literature reveals sparse evidence of these pathologic changes in animals. We recommend the following practices for conducting and interpreting future studies of SARS-CoV-2 in animals (Table 1).

Table 1.

Summary of recommendations related to research on cardiovascular pathology in animal models of SARS-CoV2 infections

Complications	Recommendations
The scarcity of comparative pathologists involved in the design and analysis of COVID-19 animal research models.	A veterinary pathologist should be included on the team of authors to provide expertise in the interpretation of pathologic findings from a comparative pathology background. The inclusion of veterinary pathologists in comprehensive reviews of animal models would ensure that the literature supports appropriate interpretation with regard to the validity of pathologic findings.
Animals have background (baseline) lesions that can occur naturally in a particular species. Background lesions must be identified and not misinterpreted as an outcome of the experimental model.	All investigators using animals in biomedical research should collaborate with a pathologist who has the appropriate expertise to interpret the histologic lesions. Organizations such as the American College of Veterinary Pathologists (https://www.acvp.org) may be contacted for assistance in locating a pathologist in your region. Furthermore, the Dean’s Office at a regional veterinary or medical school can provide suggestions regarding collaborating veterinarians and pathologists in your area. In addition, sample size must be maximized whenever possible to determine the type and level of background lesions and account for this potential confounding variable.
Some publications that use SARS-CoV-2-inoculated animals provide low-resolution figures that may lack markers (for example, arrowheads), thereby interfering with critical evaluation by readers. Also, photomicrographs are not always consistent with the lesions described in the figure legends or the text.	When reporting histopathologic lesions, figures should be provided to illustrate the findings. These images have resolution sufficient to allow readers to make their own interpretation. Images should have visual aids (e.g., arrowheads) to indicate the relevant lesions. If possible, a veterinary pathologist should perform the histologic interpretation.
In many cases, conclusions of cardiac hypertrophy in experimental SARS-CoV-2 infection are inconsistent in methodology, calling into question the validity of the conclusions.	When cardiac hypertrophy is suspected, animals should undergo perfusion with a potassium–cadmium chloride solution; the heart will then consistently arrest in diastole, enabling consistent comparison of cardiac size between treatment groups. In addition, heart weights and tibia lengths should be recorded at necropsy; the ratio of these parameters can be used to assess for the presence of cardiac hypertrophy across different treatment groups. After 48-h of fixation in 10% buffered formalin, the heart should be sectioned at the midpapillary level, processed, paraffin-embedded, sectioned at 4 to 5 µm, and stained with either hematoxylin and eosin or with wheat germ agglutinin to outline the cardiomyocyte cross section.
Leukocyte transmigration can be confused for vasculitis on histopathologic evaluation. Vascular lesions and microthrombi are often misinterpreted or over-interpreted, and the definitions of vasculitis and coagulopathy lack standardization.	Involvement of a pathologist in the interpretation of histopathologic lesions would enhance appropriate differentiation of vasculitis and leukocyte transmigration. When vascular lesions and microthrombi are reported, authors should consider that microthrombi can also occur after phlebotomy, venous catheter placement, or the use of euthanasia solution. The histologic presence of a thrombus does not necessarily mean that coagulopathy was present. Functional testing of coagulation (for example, prothrombin time, activated partial thromboplastin time, D dimers) should be included in studies with coagulopathy diagnoses. In addition, the definitions of vasculitis and coagulopathy should be standardized. In the meantime, authors should provide definitions of vasculitis and coagulopathy in their publications.
Some animal models require a power analysis to determine the sample size needed to provide statistical significance for a pre-determined effect size and to allow determination of background pathology.	Whenever possible and feasible, a power analysis should be performed to determine the sample size needed for an adequately powered experiment. The incidence of background lesions should be considered in the power analysis.
Reproducibility of animal models can be complicated by inadequate reporting of methods, particularly with regard to scoring lesions.	Accurate and thorough details of the methods used, including systems used to score lesions, should always be reported to improve reproducibility.161 Standard recommendations such as the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines have been published to aid investigators.
The focus on lung pathology in animal studies of SARS-CoV-2 infection results in a lost opportunity for data collection in studies that fail to investigate the heart and vasculature. This failure makes it impossible to determine whether the lack of cardiac pathology in COVID-infected animals is due to the absence of this complication or lack of investigation.	The heart and vasculature (and perhaps other tissues) should be investigated (or at least retained for future evaluation) of histopathologic lesions and the presence of SARS-CoV-2 viral genome.
Failure to remove blood from the heart via vascular perfusion can confound the ability to differentiate the presence of the virus within cardiomyocytes compared with the blood cells within heart chambers. Molecular methods of detection of SARS-CoV-2 viral genomic DNA in tissue samples cannot be a reliable indicator of viral replication in those tissues where blood contamination is present.	Tissues should be processed after performing vascular perfusion to remove blood cell contamination. When potassium-cadmium chloride solution is not available, PBS perfusion should be performed. Vascular perfusion will enable detection of SARS-CoV-2 viral genome in tissue samples to be more representative of viral replication within those tissues. Molecular methods of SARS-CoV-2 detection in tissues should be correlated with histopathologic findings (for example evidence of viral replication within tissue). The methods of documenting viral tropism within tissues must be standardized.
Direct comparisons between the pathology findings of animals and humans are complicated by the variability in the time course of the disease in these populations.	The different time courses of infection between human and animal SARS-CoV-2 infection must be acknowledged as a potential confound of direct comparisons between species.

First and foremost, the scarcity of comparative pathologists has been a persistent issue.^44,138 The inclusion of veterinary pathologists in the design and analysis of animal-based COVID-19 research would ameliorate many of the concerns that we have addressed and would facilitate the implementation of many of our other recommendations. In addition, the inclusion of veterinary pathologists in comprehensive reviews of animal models would ensure that the literature is being appropriately interpreted with regards to the validity of pathologic findings (or lack thereof), which could influence the design of future studies.

Background lesions are histologic findings that, although abnormal, occur naturally in a particular species; familiarity with these lesions in the examined species is essential to avoid interpreting them as results of the experimental model. In addition, in initial studies, sample size must be maximized to the extent possible in order to allow determination of the level of background lesions and account for this potential confounding variable. For example, myocardial degeneration and fibrosis occur as background lesions in hamsters, whereas myocarditis, necrosis, and fibrosis are background lesions in macaques. Veterinary pathologists have unique training and familiarity with lesions in a variety of species, whereas physician pathologists or other MDs may lack sufficient experience with the species to recognize these findings, and may thus overinterpret or misinterpret the presence or significance of these lesions as related to infection. In species where background lesions have been previously characterized, historical controls can be used to reduce the number of animals used in a study. However, this requires that prior studies have rigorously evaluated and documented background lesions. On the other hand, diet and infectious disease prevalence could vary from year to year. Thus, controls should be included and evaluated at the same time as the experimental groups in order to control for environmental changes.

All investigators using animals in biomedical research should collaborate with a pathologist who has the appropriate expertise to provide interpretation of the histologic lesions and, where necessary, consult on the adequacy of the prior literature with regards to historical controls. Although veterinary pathologists with the appropriate experience are scarce and geographically dispersed,^8,22 resources are available to facilitate consultations and collaboration between the medical and veterinary research fields. Organizations such as the American College of Veterinary Pathologists may be contacted for assistance in locating a pathologist in your region (https://www.acvp.org). Furthermore, the Dean’s Office at your regional veterinary or medical school may be able to provide suggestions regarding collaborating veterinarians and pathologists in the area. Some universities have core facilities for pathology and phenotyping animal models.

When reporting histopathologic lesions, authors should provide figures that clearly illustrate the findings, and these images must have adequate quality and resolution to allow readers to make their own interpretation. Figure legends must correspond to the lesions shown in the images. Ideally, images should contain visual aids (for example, arrowheads) that point out relevant lesions to readers. To ensure that the figure images and legends are appropriate, a veterinary pathologist should perform the histologic interpretation. Understandably, not every scientific journal will have a veterinary pathologist on the editorial staff or available to serve as a reviewer. Including a veterinary pathologist as an author reduces the likelihood of publication of inconsistent or inaccurate figures.

When cardiac hypertrophy is suspected, animals should be perfused with a potassium–cadmium chloride solution; this procedure will consistently stop the heart in diastole, thus facilitating comparison of hearts among treatment groups.¹³⁶ When this cannot be done, perfusion with PBS at least allows testing for SARS-CoV-2 viral genome in tissue samples without blood cell contamination. At necropsy, heart weights and tibia lengths should be recorded; the ratio of these parameters can then be used to assess the presence of cardiac hypertrophy across various treatment groups. After 48-h fixation in 10% buffered formalin, the heart should be sectioned at the midpapillary level, processed, paraffin-embedded, sectioned at 4 to 5 µm, and stained with either hematoxylin and eosin or with wheat germ agglutinin to outline the cardiomyocyte cross section. Microscopic evaluation of the heart can then be used to measure and compare cardiomyocyte cross sectional contours across treatment groups.⁵⁴

When vascular lesions and microthrombi are reported, readers should be aware that microthrombi can occur after phlebotomy, venous catheter placement for anesthesia or euthanasia, following release of tissue factor associated with cervical dislocation, and with the use of euthanasia solution in NHP. It is important to recognize that the histologic presence of a thrombus does not necessarily mean that coagulopathy was present. Functional testing of coagulation (for example, prothrombin time, activated partial thromboplastin time, D dimers) should be included in studies with coagulopathy diagnoses to support or rule-out confounders that can cause thrombosis. In addition, the definitions of vasculitis and coagulopathy should be standardized. Reviewers and journal editors should be aware of these potential misunderstandings, particularly from authors without the background to understand the differences between vasculitis and the normal process of vessel transmigration. Editorial boards of journals should be expanded to include pathologists and veterinarians. In addition, funding agencies can organize workshops involving experts who can design a standardization scheme for infectious disease studies involving vessels and coagulopathy.

Whenever possible and feasible, we recommend performing a power analysis to accurately determine the sample size needed for an adequately powered experiment, and we propose that this analysis consider the incidence of background lesions.^47,140 We recognize that large animal models, such as macaques, require more resources and are associated with a greater financial overhead than other species, such that a large sample size may not be feasible. However, background pathology should still be acknowledged, and a board-certified veterinary pathologist (or a veterinary pathologist with sufficient experience) should consider background lesions when interpreting the pathology. Published guidelines for pathology review are available for various species, and may serve as a resource when the use of historical controls is to be considered.^{11,20,35,67,71,124,134,153}

Accurate and thorough details of the methods, including scoring systems used for lesions, should be included to improve reproducibility.¹⁰⁰ Standard recommendations such as the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines have been published to aid investigators in reporting necessary experimental conditions to ensure full transparency and replicability.^112,113

In addition, the majority of SARS-CoV-2 animal models in the literature principally evaluate lung tissues. By failing to investigate the heart and vasculature or by inappropriately processing these tissues (for example, vascular perfusion), important biologic information is lost, making it impossible to determine whether the lack of cardiac pathology in animals infected with SARS-CoV-2 results from the absence of this condition, or from inadequate investigation and rigor. A similar principle exists with regard to standardizing the methods of documenting viral tissue tropism; for example, failure to remove blood from the heart via perfusion confounds the differentiation of viral presence in cardiomyocytes as opposed to its presence within blood cells in the heart chambers.

Finally, inherent issues exist in making direct comparisons between experimental findings from animals and autopsy data from human cases of COVID-19, principally with regard to the time course of the disease. For example, at the time of death, human patients vary in the presence of comorbidities, ventilator injury, and severity of lung injury. Animals used in research typically lack preexisting comorbidities and have a similar background of genetic and environmental conditions. In addition, animals are euthanized at a predetermined experimental endpoint, thus standardizing the time course of the disease across the cohort, or potentially are euthanized sooner, depending on the clinical course and whether a humane endpoint becomes necessary. Although long-term effects of COVID-19 have been documented in human patients, the constraints of animal models make long-term sequelae difficult to evaluate. These differences between humans and animal models are important potential confounders of direct comparisons between species.