The Growing Importance of Basic Models of Cardiovascular Disease

The past decade has witnessed an explosion in the tools, resources, and publicly available data for human-based discovery implicating multiple new candidate pathways and potential therapeutic targets in cardiovascular disease. In addition to documenting associations between genetic sequence variants or circulating biomarkers and disease or clinically relevant phenotypes, techniques such as Mendelian Randomization (MR) can – with some caveats – help determine whether a candidate is likely part of a causal pathway, contributing to disease pathogenesis, or simply associated with disease. The distinction is crucial as pathways mechanistically important in disease are potential therapeutic targets whereas markers with no functional role, are not.

There are ample examples of these principles but one of the best documented is the strong, quantitative relationship between both positive and negative genetic determinants of circulating LDL levels and the risk of clinically important atherosclerotic coronary artery disease (CAD)^{1, 2}. In contrast, similar analyses of genetic determinants of C-Reactive Protein (CRP), a marker of inflammation which, like LDL, is powerfully predictive of CAD risk, suggest that CRP is not in the causal pathway³. To be clear, as demonstrated in the CANTOS trial⁴, enormous evidence supports the role of inflammation in the pathogenesis of atherosclerotic vascular disease and consequent clinical events. However, CRP itself appears more a reporter of the inflammatory state than a mechanistically important part of this sequence. Advances in bioinformatic approaches combined with the ever-expanding scale of clinical datasets with detailed genetic, phenotypic, and outcome data available have led to a breathtaking acceleration of the pace of discovery with profound implications for our understanding of disease as well as the development of therapeutic approaches.

Based on this remarkable progress, it may be tempting to argue that human evidence, particularly genetic data supporting a causal role, should be the essential foundation for identifying, evaluating, and advancing all therapeutic targets⁵. In this view, the value of in vitro and in vivo biological models may appear diminished. Most basic scientists hope their discoveries will have implications for understanding human biology and disease, and likely feel there is little or no value in discovering a ‘cure’ that works only in animals. The road to failed therapies is littered with interventions that appeared promising in animal models. However, on deeper inspection, some of these encouraging preclinical results involved models with limited track record of predicting translational success or required interventions unrealistic clinically in timing or approach. For example, in the past, considerable effort was devoted to identifying treatments that reduced the development of restenosis in rodent vascular injury models, the vast majority of which failed to translate to large animal models or clinically. The combination of advances in human-based discovery and past disappointments based on animal models understandably leads some to advocate a focus driven by human genetics or genomics with little or no emphasis on nuanced investigation in animal or in vitro models.

In contrast, I would argue that both in vitro and in vivo models continue to play essential roles in our understanding of biology, disease pathogenesis and pathophysiology, and target validation for several reasons. First, genetic datasets for many phenotypes remain incomplete. Previous work has suggested that only a small minority of FDA-approved drugs overlap are implicated by human genetic studies⁶. While this number will undoubtedly continue to expand with the increasing scale and scope of available genetic studies, it is a fitting reminder that the absence of genetic data supporting a pathophysiological role should still not be taken as evidence that such a role is absent. Second, in many ways the explosion of candidates fueled by genetics and genomics has only increased the need for mechanistic models to assess their functional roles, elucidate the responsible mechanisms, and optimize strategies for intervention. This nuanced understanding is often essential to the successful development of new therapies. Finally, genomic and epigenetic studies generally require relevant tissues, which may be readily available for some conditions, such as cancer, but are less so for many cardiovascular phenotypes, particularly early in the course of disease. Thus basic models of cardiovascular disease provide a crucial complement to human discovery efforts whose importance has only expanded with the explosion of putative candidates.

That said, it is important choose the model most appropriate to the goals at hand and recognize the limitations inherent to that model. Often the goal is to mimic a clinical condition with the goal of elucidating mechanisms and validating pathways as therapeutic targets. Of course, no model perfectly recapitulates the nuanced biology of the human cardiovascular system or disease. Moreover, such studies are often performed in young, otherwise healthy animals in contrast to the generally older patients with multiple comorbidities who present clinically with cardiovascular disease. While large animal models often come closer to mimicking human biology, practical considerations including expense, access, and the challenges of genetically manipulating such models, steer most investigators toward smaller, often genetically tractable species, particularly in the early stages of projects, reserving large animal studies for late pre-clinical development. As noted earlier, there are multiple humbling examples where success in small animals has not presaged clinical translation. However, there are also examples where they have. These include the seminal studies of Drs. Marc and Janice Pfeffer along with Dr. Eugene Braunwald documenting the benefits angiotensin-converting enzyme inhibition in mitigating adverse ventricular remodeling after myocardial infarction⁷. These initial studies were performed in rats and ultimately laid a foundation for what became one of the pillars of modern therapy for heart failure with reduced ejection fraction. Recognizing the differences among models in their track record of predicting translational successful as well as the importance of testing clinically relevant interventions can help place the results of experimental models in the appropriate context.

Less commonly, animal models are employed not because they recapitulate human biology but precisely because of the interesting and potentially informative differences they exhibit. Examples include the remarkable regenerative capacity of the zebrafish, including the ability to regrow healthy heart tissue with minimal scarring after cardiac injury. Understanding the basis of this ability and how it was lost later in evolution could provide clues to how it might be at least partially restored. Similarly, the Burmese python’s ability to cyclically grow and shrink its organs – including the heart – after feeding in association with a remarkable increase in its metabolic oxygen consumption may provide insights that help us modulate more common forms of cardiac hypertrophy and regression. Thus we have included discussion of these models not because they resemble human biology but because they display unusual responses that may inform our understanding of more common phenotypes.

With these concepts in mind, we have invited experts to share their perspectives on a broad range of cardiovascular models. Our hope is that these reviews will serve as a resource for our community, providing an authoritative reference for cardiovascular investigators interested in learning more or potentially using these models. In addition, we have asked each of the authors to share insights based on their wealth of their experience as to the advantages and disadvantages of these models as well as their translational relevance. We begin with two reviews related to sex-differences in cardiovascular disease, an important and historically understudied area that was the subject of a prior Compendium in Circulation Research⁸. The first review by Reue and Wiese discusses powerful tools available to illuminate the mechanisms responsible for sex differences in cardiovascular disease, including enabling investigators to parse the relative contributions of hormonal and chromosomal effects⁹. The authors describe cardiovascular conditions with well-documented sex-based differences in prevalence or presentation and highlight some of the mechanistic insights garnered to date, acknowledging that much remains to be learned in this area. Cardiovascular disease is the most common cause of pregnancy-related death in women and in the next review, Arany et al discuss models of pregnancy-related cardiovascular diseases including preeclampsia and peripartum cardiomyopathy¹⁰. These models have provided some surprising new insights into the biology of these conditions and stimulated ongoing trials of new therapeutic approaches.

The next four reviews describe tools and systems used by many investigators across diverse model systems. The first by Thomas and colleagues describes the use of human induced pluripotent stem cells (iPSCs) to generate cardiomyocytes and other cardiac cell types that can be studied in isolation in vitro or in structured multi-cellular organoid systems¹¹. This approach has the advantage of using human cells that can be derived from patients and the authors thoughtfully discuss the potential advantages and limitations of current systems as well as advising us on how best to select the right model for a particular application. Gonzalez-Rosa then discusses zebrafish cardiac models, describing both the unique advantages of this system and its limitations¹². As noted above, the regenerative capacity of the adult zebrafish heart is remarkable but is only one of several features that have stimulated interest in this model. Genome editing is briefly discussed in these reviews as it has enhanced the power of both iPSC-derived cellular models and zebrafish studies. Liu and Olson¹³ then provide a more comprehensive review of genome editing to generate in vivo cardiovascular disease models as well as the potential and challenges in developing this as a therapeutic strategy for cardiovascular disease. An important aspect of evaluating many animal models is the assessment of in vivo cardiac function. Sosnovik and Scherrer-Crosbie¹⁴ discuss advanced imaging approaches – including echocardiography, magnetic resonance and positron emission tomography (PET) – to visualization of left ventricular function in small animal models.

The next four reviews illustrate the utility of animal models for investigating four different disease processes. Gisterå and colleagues discussed models of atherosclerotic vascular disease¹⁵. They provide a valuable perspective on legacy and current models across a broad range of species, and discuss in detail murine models, including the advantages and disadvantages of actively used strains. Two reviews on heart failure follow. The first, by Pilz et al¹⁶ discusses large and small animal models of heart failure with reduced ejection fraction (HFrEF). The second, by Roh and colleagues, is focused on heart failure with preserved ejection fraction (HFpEF)¹⁷. Given our still-limited understanding of HFpEF pathophysiology and the absence of any therapies to date that reduce mortality in this high-risk and growing population, the hope is that these models can help. Finally, Blackwell and colleagues provide a detailed review of animal models of arrhythmia¹⁸. The authors provide a useful review of electrophysiological principles followed by discussion of models relevant to genetic arrhythmia syndromes, genetic cardiomyopathies, and acquired arrhythmias.

The last two reviews in the series address conceptual issues relevant to multiple models. Given the relentless, energy-consuming work that the heart must perform, it is no surprise that cardiac metabolism has long been a topic of great interest. Bugger et al discuss the tools available to study both genetic and acquired models of dysregulated metabolism¹⁹. Of course, some abnormalities may not be apparent in unperturbed animals at rest. Many of the reviews touch on surgical or other interventions that can be used to induce pathophysiologically relevant stress that elicits more subtle cardiac abnormalities. In the last review, Hastings and colleagues discuss exercise as a physiological stress that can also elicit phenotypes not apparent at rest²⁰. They also discuss exercise training protocols that can serve to elucidate both the impact of specific interventions on cardiovascular adaptations to training and as a platform for discovery of pathways mediating the benefits of exercise. While many of the reviews examine what goes wrong in models of cardiovascular disease, we believe there is also value in understanding the mechanisms that keep the heart healthy, using the exercised heart as a model, and determining whether these can be exploited to prevent and/or treat disease.

Of course, there are many useful models that could not be included due to space contraints and even where models are discussed often only an overview could be provided. We apologize to those whose work or favorite models we were unable to include. We also acknowledge that the insights shared, particularly regarding advantages and disadvantages of individual models, necessarily reflect the experience and perspectives of the authors, which may differ from those of the reader. Nevertheless, we hope these reviews will provide a useful starting place and conceptual framework as well as a reference resource for investigators seeking to learn more about particular models and determine which might be best suited to their own needs.