Guidelines for in vivo mouse models of myocardial infarction

Abstract

Despite significant improvements in reperfusion strategies, acute coronary syndromes all too often culminate in a myocardial infarction (MI). The consequent MI can, in turn, lead to remodeling of the left ventricle (LV), the development of LV dysfunction, and ultimately progression to heart failure (HF). Accordingly, an improved understanding of the underlying mechanisms of MI remodeling and progression to HF is necessary. One common approach to examine MI pathology is with murine models that recapitulate components of the clinical context of acute coronary syndrome and subsequent MI. We evaluated the different approaches used to produce MI in mouse models and identified opportunities to consolidate methods, recognizing that reperfused and nonreperfused MI yield different responses. The overall goal in compiling this consensus statement is to unify best practices regarding mouse MI models to improve interpretation and allow comparative examination across studies and laboratories. These guidelines will help to establish rigor and reproducibility and provide increased potential for clinical translation.

INTRODUCTION

Coronary artery disease continues to be a leading cause of morbidity and mortality, with the most significant consequence being an abrupt reduction in blood flow through a major coronary vessel, causing myocardial ischemia, clinically termed acute coronary syndrome (1, 2). Due to various risk factors and other influences, ischemia can range from low flow to no flow, and the period of ischemia can extend from a very short duration in time to permanent occlusion of the vessel. When the reduction in flow is of sufficient duration to cause irreversible damage to the myocardium (>30 min in mice), cardiomyocyte necrosis ensues to generate myocardial infarction (MI) (3). Clinically, the current optimal therapy is reperfusion to restore blood flow through the artery. Approximately 25% of patients will not be reperfused due to many reasons, including not seeking medical attention early enough or lack of success in reperfusion (4). In addition, up to 30% of those given reperfusion therapy will experience no-reflow, which is a state of myocardial tissue hypoperfusion due to impaired microvascular flow that occurs in the presence of a patent epicardial coronary artery (5). This also contributes to adverse cardiac remodeling and heart failure (3).

Although both reperfused and nonreperfused coronary artery occlusion can result in an MI, the underlying pathology, remodeling, and subsequent progression to dysfunction of the left ventricle (LV) can be distinctly different (4, 6–8). The pathophysiological distinction between MI types when induced in animal models is sometimes unintentionally disregarded or ignored. This lack of distinction can cause confusion when evaluating new therapeutic strategies or assessing comparative studies among laboratories. The overall goal of this consensus document is to provide guidance to establish uniformity in terms of MI models in mice. As both reperfused and nonreperfused MI models hold clinical relevance, this perspective is not intended to promulgate the superiority of one MI animal model. Instead, we will provide recommendations on how the cardiovascular physiology research community can standardize approaches and best evaluate primary response variables in the proper context.

There is a wide range of MI animal models that span from rodents to large animals. There are distinct advantages and disadvantages associated with all animal models. Large animals may recapitulate a higher fidelity of the clinical phenotype, whereas rodent models allow for examination of gene effects and offer higher throughput. These guidelines will examine murine models of MI and summarize approaches and biological and physiological response variables that are translationally important. There is a growing awareness in the cardiovascular physiology research community and funding agencies that provide financial support for this area of research for the importance of studies that exhibit strong experimental rigor and reproducibility. The outcome from this consensus document is to put forward a set of uniform strategies for inducing and evaluating reperfused or nonreperfused MI, to establish benchmarks regarding LV remodeling and function that arise as a response to MI, and to provide an extensive reference list that may serve as a useful guide for those interested in designing murine MI studies.

EVALUATION OF CURRENT LITERATURE

To perform a focused assessment of current practices in using MI models in mice and determine where improvements may be needed, we evaluated recent articles published by the American Journal of Physiology-Heart and Circulatory Physiology (AJP-Heart and Circ). The search included all research articles on mouse MI published in 2019 and 2020. Articles were identified from two searches of the journal website (accessed April 7, 2021 by M.L.L.) using the terms: myocardial infarction, mice, reperfusion, heart or myocardial infarction, and coronary artery occlusion. The first search identified 24 original research articles, of which 10 were excluded primarily for not using mice. The second search identified 30 articles, of which an additional 17 articles were included. A third search was performed (accessed April 22, 2021) by a second investigator (K.Y.D.P.) using the terms myocardial infarction, ischemia, and ischemia reperfusion, which identified one additional article that had not been previously included. Of the 32 articles remaining, 7 articles were duplicates and 6 articles were excluded during analysis for not meeting search criteria. This left 19 articles for evaluation (9–27). Four authors evaluated the articles (M.L.L., K.Y.D.P., G.V.H., and Z.K.).

A summary of the evaluation results is provided in Table 1. The use of reperfused versus nonreperfused MI was approximately a 60% reperfused/40% nonreperfused split. None of the articles reported the rationale for why the selected model was used. For reperfused MI studies, both the duration of ischemia and when the hearts were examined (time of reperfusion) were highly variable. Almost every study showed a different duration of ischemia and reperfusion. For the nonreperfused MI studies, there was more uniformity, with most studies evaluating end points at 7 days, 4 wk, or 12 wk, although the rationale for time of end point selected was often not provided.

Table 1.

Evaluation of publications on MI published by the American Journal of Physiology-Heart and Circulatory Physiology in 2019–2020

Criteria	Results and Comments
Reperfused vs. Nonreperfused MI	Type: 11 reperfused (58%); 8 nonreperfused (42%)
Rationale provided for model selection	None
Duration of ischemia and reperfusion	Highly variable Ischemia times: 30’ (n = 5), 40’ (n = 1), 45’ (n = 3), 50’ (n = 1), 60’ (n = 1) Reperfusion times: 2.5 h (n = 1), 1 day (n = 2), 2 days (n = 1), 3 days (n = 3), 2 wk (n = 1), 4 wk (n = 3), 35 days (n = 1)
Nonreperfused MI time of evaluation	Focused on four times: 7 day (n = 2), 8 days and 21 days (n = 1); 4 wk (n = 4), 12 wk (n = 1)
Sex, age, strain detailed	Most provide details (only 1 did not report sex, 3 did not report age, 1 did not report strain); 53% used males only; wide variation in ages used (range 7 wk to 1 yr old); 89% used C57/BL6 (82% did not clarify J or N) and 11% used mixed strain
Sample sizes clear	All provided; not always clear the n/group; 9 studies had samples size for at least 1 group in at least 1 experiment that was n < 5, may have power issues with this analysis
Surgery detailed, inclusion/exclusion criteria provided (1-no; 5-very detailed)	Mean = 2.5; highly variable in the information provided; 47% did not confirm occlusion/ischemia (of those that did, 5 used EKG, 3 used visual inspection, 1 used both, and 1 did not state); Of 11 nonreperfused MI, 8 (73%) did not state whether reperfusion was confirmed (of those that did, 1 used EKG and 3 used visual confirmation)
Anesthesia/analgesics reported	79% reported anesthesia use and 37% reported analgesics use; for the 79% who reported what anesthetic was used, 80% used isoflurane. Sevoflurane, ketamine, and pentobarbital each reported n = 1. For analgesic use, 86% of those reporting used buprenorphine. Carprofen was used in one study.
Infarct sizes/area at risk reported	Reported for 74% (n = 14); details variable, with some simply indicating it was done, some missing when it was done; for n = 2, one used Masson’s trichrome staining as a surrogate for infarct sizing and one used hematoxylin & eosin staining as surrogate for area at risk
Cardiac physiology measured	79% performed echocardiography; 94% measured cardiac physiology by echocardiography, MRI, or PV loops; wide variation in time of evaluation and details on methods used
Recommendations from analysis to improve standardization	Provide rationale for model selection (reperfused vs. nonreperfused MI) and end points addressed. Use checklist of minimum details to report in methods and results, including dose, administration route, and frequency and of analgesics before surgery and after surgery. Supply inclusion and exclusion criteria for successful coronary reperfusion or ligation. Consider guidance on optimum times of evaluation depending on questions asked; in experimental design, consider using uniform times for consistent end point assessments to allow cross-study comparisons.

n = 19 publications analyzed. MI, myocardial infarction.

There was overall good reporting of sex, age, and strain used. Recent efforts to include both sexes in rodent studies had an observable effect, with 47% of the studies using male and female mice (53% used males only, no study used females only). In 2018, there was an evaluation of mouse echocardiography articles published in AJP-Heart and Circ between January 1, 2016 and July 21, 2017 (28). In that evaluation, 84% provided details on sex (vs. 95% in the current evaluation), with 21% of studies using male and female mice (55% used males only, 8% used females only). Sample sizes were reported to varying degrees, making it unclear how many mice were used in each group or experiment. Half of the studies included sample sizes for at least one group in at least one experiment of n < 5, which may indicate a potential issue with statistical power (29).

The details provided in the methods to describe the surgical procedure were highly variable, with some articles reporting minimal technical information and others providing extensive details. In terms of rigor, the number of studies that did not confirm ischemia or reperfusion at the time of surgery was high [47% did not confirm ischemia by electrocardiogram (ECG) or blanching and 73% did not confirm reperfusion by resolution of ST segment elevation on ECG), as was the number of studies that did not calculate area at risk (for reperfused MI) or infarct size (for nonreperfused MI). There was adequate reporting of cardiac physiology, with 94% of the studies using either echocardiography, magnetic resonance imaging, hemodynamics by pressure-volume loops, or a combination of these approaches to document effects of MI on organ-level physiology. Based on these analyses, we developed a list of recommendations to focus efforts for improvement in experimental design and reporting on areas where gaps were identified.

REPERFUSED VERSUS NONREPERFUSED MI

Although both reperfused and nonreperfused MI models share an underlying initial ischemic insult, the type of myocardial injury is distinct and the models should not be used interchangeably. When deciding between models, one should first consider what aspect of the myocardial response to ischemic injury is to be studied. This section briefly describes the clinical situations modeled by reperfused and nonreperfused MI, summarizing the key similarities and differences. For a brief review on these two models of MI and which model may be most appropriate for a given study, we point the reader to a recent perspective on this topic (4).

As indicated in the most recent American Heart Association Guidelines for managing patients with ST-elevation MI (STEMI) (30), once a STEMI has been indicated by electrocardiogram, how to achieve reperfusion should be decided within 10 min. Regardless of reperfusion strategy, the goal is to restore flow to avoid cell death as rapidly as possible within 60–90 min (31). Because of increased emphasis on patient education to identify a heart attack and the need for rapid, effective reperfusion, the initial mortality from this acute coronary syndrome has declined (32). Thus, from a clinical perspective, an MI model that simulates a period of coronary occlusion followed by reperfusion is the most representative of the current clinical scenario.

A portion of MI patients are not reperfused; however, for various reasons, including medical decisions (inaccessible anatomy), a delay or no access to care (silent MI, misdiagnosis, or distance from the hospital), or ineffective reperfusion due to thrombotic microvascular obstruction (33). Patients in whom optimal reperfusion (thrombolysis in MI, TIMI 3 flow) is not achieved are at the highest risk of developing heart failure with reduced ejection fraction (HFrEF). Thus, rapid decompensation to HF is best modeled with the nonreperfused MI model. Patients undergoing reperfusion are also at risk of developing HFrEF, albeit at a lower rate (34–36). Thus, there is overlap in the phenotypes of these two animal models in the clinical situation. Plus, it is important to remember that the clinical situation is rarely binary (reperfused or nonreperfused MI), as the severity of both initial insult (lack of perfusion) and efficacy of reperfusion are highly variable, resulting in a gradient of perfusion outcomes in patients (37). In the clinical setting, all patient groups (i.e., those who are successfully and optimally reperfused, those who are not, and for everyone in between) are prone to progressive heart failure and recurrent cardiovascular events (35, 38, 39).

It is important to note that distinct cellular and molecular pathways can be activated in response to ischemia alone versus in response to ischemia and reperfusion. For reperfused MI, the duration of ischemia is an important determinant of the overall tissue response and extent of LV remodeling. Although ischemia tolerance is strain dependent (40), ischemia typically varies from 30–90 min in reperfused MI animal models. Rather than list an exhaustive inventory of differences between reperfused versus nonreperfused MI, which would be at the same time enormous and certainly incomplete, we focus here on the major differences as they apply to animal models (Fig. 1). The most obvious differences arise from the LV remodeling that occurs (43). The reperfused heart shows minimal changes in mass, size, shape, and function for ischemic intervals <45 min (44, 45). The Calvert laboratory and others have found that extended periods of ischemia (60–90 min) before reperfusion can induce large changes in LV dilation and function (46). Indeed, ischemia periods of >60–90 min are considered irreversible, with similar responses to nonreperfused MI (47).

Figure 1.

Key differences between reperfused myocardial infarction (MI) and nonreperfused MI, including differences in infarct sizes, study goals, structural and physiological responses, survival, and cell and molecular signaling. For the reperfused heart, the mid-myocardial section was stained with Evans blue (in vivo) and 1% triphenyltetrazolium chloride (ex vivo) to show the necrotic zone (NEC; white), nonischemic zone (NIZ; blue), and ischemic zone (IZ; red and white) (41). For the nonreperfused heart, the ventricles were sectioned from base (left) to apex (right) and stained with 1% triphenyltetrazolium chloride, which stains viable myocardium red and infarct is white (42). Images are reproduced with permission from the American Journal of Physiology-Heart and Circulatory Physiology.

In reperfused MI, inflammation is accelerated and amplified as leukocytes infiltrate and accumulate in the previously ischemic myocardium through the patent vessel. Infarcted regions in reperfused hearts sometimes present diffuse and patchy necrosis, making it difficult to quantify the true size of the infarct region because surviving myocytes are interspersed. The nonreperfused heart, in contrast, shows extensive LV dilation, infarct wall thinning, and LV dysfunction in response to a substantial amount of cell death (48). Because of considerable wall thinning (∼50% of the infarct wall), the structural and functional signs of HFrEF are rapidly reflected in the nonreperfused MI model. Therefore, rupture-related mortality is only appreciable in the nonreperfused MI model (49). Based on extensive wall thinning and heightened infiltration of leukocytes (neutrophils and monocytes), there is considerable mortality ranging from 10% to 40% due to ischemic events or congestive heart failure-related mortality in the nonreperfused MI model. The mortality range for both reperfused and nonreperfused MI is dependent both on sex (female mice have lower mortality) and on infarct size (50). Overall, if survival outcomes from an investigational therapy are of interest, the nonreperfused MI model is more appropriate (51). Due to accentuated heart failure, multiorgan inflammation and dysfunction (e.g., involving kidney and spleen) are typically more common in the nonreperfused MI (52, 53). These key points provide a means to compare the suitability of the specific models and tailor use to the unique research hypothesis.

MI MODELS IN MICE

The Evolution of the MI Model

The sophistication of MI models in mice have advanced over the last 25 yr. After decades of experimentation in large animals (including nonhuman primates, dogs, cats, pigs, sheep, and goats) and in rats and guinea pigs (54–85), mouse models of reperfused and nonreperfused MI were developed (86) and extensively characterized (6). Although large animal models may better recapitulate the clinical phenotype of ischemic injury regarding indices of cardiac remodeling and arrhythmogenesis, rodent studies, in particular mice, allow for basic mechanistic insight. Through the use of transgenesis, murine studies allow dissection of cellular and molecular mechanisms. In addition, with standardized approaches (the focus of this article), high throughput analyses of a consistent ischemic injury can be achieved.

One initial issue with establishing a protocol for MI surgery in mice was the heterogeneity of the coronary anatomy, which made artery visualization inconsistently difficult (87–89). Despite the technical challenges related to the small size of the animal and the difficulties in visualization of the coronary artery, MI in mice is highly reproducible in laboratories with experienced microsurgeons (1, 90). Because of the rapid progress in the development of genetic targeting approaches in mice and availability of a wide range of reagents, the mouse model of MI has emerged as a tractable tool for mechanistic dissection of cellular events and molecular pathways that may be involved in the injury, repair, and remodeling of the infarcted heart. The first iteration of MI in mice included intubation, followed by a sternal thoracotomy in which approximately three ribs were cut from the sternum and cauterized to gain access to the heart, then ligation of the left coronary artery (LCA) (86). This surgical approach generated a significant injury response that potentially masked early MI-induced inflammation (91).

Over the last 20 yr, significant technical advances have further expanded the range of capabilities of mouse models of MI. First was the development of a closed-chest model of coronary ischemia and reperfusion, in which the coronary artery could be occluded and reperfused 7–10 days after initial surgery through an exteriorized suture, allowing sufficient recovery time such that the study of inflammatory changes following MI could occur without the confounding effects of thoracotomy (91). The closed-chest model of ischemia-reperfusion can also be used to induce repetitive ischemic insults that may recapitulate aspects of chronic ischemic coronary disease (92–94).

Second was the development of a rapid approach for coronary artery ligation surgery by manually exteriorizing the heart outside the chest cavity, which reduced the duration of the surgical procedure by not requiring intubation and significantly decreased perioperative mortality and periprocedural inflammation (91, 95, 96). Even less invasive, ultrasound-guided approaches have been developed in mice, including transthoracic puncture and LCA ligation (97), as well as a novel needle-based electrocautery technique to rapidly coagulate and clot the LCA (98). This method by design does not allow for reperfusion. To date, these procedures have not been widely adopted. Third was later refinement using a minimally invasive intercostal thoracotomy (typically the 4th intercostal space) that does not involve cutting the ribs and was shown to have minimal adverse effects compared with unoperated controls (99). The minimally invasive surgery is recommended to prevent excessive inflammation due to surgery and eliminates the need for sham controls (99).

Additional murine variations on this approach include recurrent MI, which can now be studied in mice (100). As the ischemic event is a direct consequence of advanced atherosclerosis arising from vascular defects (endothelial dysfunction and lipoprotein retention) that are present in humans for years to decades before the MI, this pathological milieu should be considered (101). This model can be accomplished by individually ligating on separate days, first the LCA, followed later by the left circumflex artery distal to the first branching point of the LCA. These are challenging major survival microsurgeries when performed sequentially, and controls should be included to monitor the degree of first infarction, such as blood troponin levels. As another variation, MI has also been reported in the right ventricle using right coronary artery ligation in mice (102). In this model, right ventricle heart failure is induced and is associated with the development of LV diastolic dysfunction. This model provides a means to study mechanisms and treatments of the combination of right and left ventricular dysfunction after MI.

Ablation Models

In contrast to traditional ischemia-induced infarcts, cryoablation is used in mice to produce infarcts, with precisely controlled size, shape, and location (103–107). In this approach, a precooled probe or handheld liquid nitrogen probe is applied to the epicardial surface at a specific location for a precise amount of time, resulting in consistent infarcts that are independent of variations in individual coronary anatomy (98). Because of the controlled shape, size, and degree of transmurality of infarcts with this approach, it can be useful for studies of infarct mechanics and long-term LV structural and functional remodeling (107). With potentially lower animal-to-animal variability, cryoablation may increase statistical power to detect treatment effects. Like prolonged ischemia, cryoablation results in widespread necrosis that involves complete cell death in the injured region. Because of the differential modes of cell death, ablative infarcts are less appropriate for studying early mechanisms of infarct-induced inflammation, neovascularization, and scar formation and may be suitable for testing tissue-engineered, stem cell, or other regenerative or rejuvenating therapies (108, 109).

Neonatal MI and Regenerative Medicine Aspects

In contrast to adults, MI in neonatal mice can lead to cardiac regeneration (110, 111). Neonatal cardiac regeneration is reproducible only if the cardiac injury is induced by coronary artery ligation on the first postnatal day after birth (112). Neonatal mice (before P8) have the capability to proliferate myocytes and show self-renewal of resident macrophages (113, 114). The potential importance of macrophages for this response was demonstrated using a targeted cell depletion model, which prevented regeneration of the neonatal myocardium and reduced angiogenesis following MI (115). Cardiac fibroblasts switch from STAT3 to TLR and NF-κB signaling early postnatally, which is also believed to play a role in how a very young versus adult heart will respond to an ischemic event (116, 117). Although neonatal rodents tolerate low temperatures, their body temperature is considerably sensitive to ambient temperature. Thus, the standardized control of body heat is key for reproducible results (118). Initial inhaled isoflurane, combined with induced hypothermia, is advised for anesthesia (119). The reduced blood flow of hypothermia may make it difficult to visualize the LCA, which is critical for consistent degrees of cardiac injury. The length of time that pups are away from the mother can affect maternal cannibalization, and this time should be minimized. Particular attention should be paid to noting sex of the pup. The cardiac metabolic response to ischemia may be distinct in neonates, which may affect degree of ischemic injury and repair (120).

Computational Models Using Mouse Big Data and Artificial Intelligence

One advantage of having uniform evaluation times is the ability to combine results across studies for meta-analysis examination (121). Comparing adult and neonatal MI using a big data approach can yield nonintuitive insights. The Engler laboratory used such an approach to define cross talk mechanisms that occur between macrophages and cardiac fibroblasts in a regenerative versus nonregenerative heart (116). Using genomics data from multiple data sets, they were able to identify key regulators that are changing with age and contributing to LV remodeling after MI injury. In a similar study, Lindsey et al. demonstrated the strength of using a large data set to determine markers of adverse remodeling (122). By querying the mouse Heart Attack Research Tool (mHART) 1.0 (90), a data set collected from a single site laboratory with multiple investigators, the authors were able to identify a phenotype of extreme LV dilation that was not due to infarct size and identify plasma proteins and LV infarct inflammatory cells and genes that explain the differences in dilation. Using mHART in addition to proteomic evaluation of plasma from participants in the Jackson Heart Study, DeLeon-Pennell and colleagues (123) demonstrated a sex-dependent response to MI and development of heart failure. These studies highlight the strength of translating findings by confirming mouse findings in samples from humans.

RECOMMENDATIONS FOR DESIGNING EXPERIMENTS AND REPORTING RESULTS

Experimental Design Details

The MI surgery procedure in mice has been detailed extensively (52, 124–126), and Table 2 provides a summary of recommendations that include consideration for the Animals in Research: Reporting in vivo Experiments (ARRIVE) Guidelines (127). The American Physiological Society has made available guidelines for transparent reporting that it recommends all authors follow (https://www.physiology.org/author-info.promoting-transparent-reporting). We include recommendations to authors to use the ARRIVE guidelines and suggest that journals adopt the ARRIVE or similar reporting guidelines to improve transparency and reproducibility (128). The quality of MI is optimally assessed by measuring infarct size (for nonreperfused MI) and additionally area at risk (for reperfused MI). For detailed staining methodologies, please see Refs. 48 and 129–133. In mice, permanently ligating the coronary artery around 1–2 mm distal to the left atrium will consistently generate large infarcts (35%–60% of total LV) or large areas at risk (47, 134, 135). Small infarcts in the nonreperfused MI model may reflect technical issues in missing the coronary artery, with damage and localized fibrosis resulting from the suture rather than reflecting the intended MI. Ligating more than 3 mm from the atria may result in high MI variability due to branching of the coronary artery (88).

Table 2.

Recommendations for MI studies

Model	Recommendations
All models	• Follow ARRIVE guidelines
Reperfused MI (MI/R)	• Measure both infarct size and area-at-risk • Measure blood pressure and heart rate during ischemia (note that especially the heart rate during ischemia significantly influences the infarct size; if necessary, include appropriate control groups) • Body temperature must be tightly regulated • Use to study impacts on infarct size (cardioprotection) • Use to study interventions that require reperfusion and relationship to clinical scenario of reperfusion • Use to study regeneration, repair, or remodeling • Use MI/R test interventions in a model clinically relevant to the reperfused patient
Nonreperfused MI	• Measure infarct size within 3 days of MI to determine whether the intervention/mutation impacts acute infarct size; differences are highly improbable in mice because of lack of functional collaterals • Use to study regeneration, repair, or remodeling • Use nonreperfused MI to test interventions in a robust remodeling model and to test interventions in a model clinically relevant to the nonreperfused/very late reperfused patient
Cryoinjury	• Use to control size, shape, or location of infarct, but experimental protocol (e.g., contact duration, probe temperature) must be standardized to enhance consistency • Use to achieve uniform cell death in target region, but extent of transmural damage should be determined (early and late)Use for relevance to ablation used in humans • Not ideal for studying pathophysiology of MI

ARRIVE, animals in research: Reporting in vivo experiments; MI/R, myocardial ischemia-reperfusion; MI, myocardial infarction.

In addition to monitoring the LV for visual changes at the time of surgical MI induction (i.e., blanching and akinesis), ECG monitoring for ST segment elevation to confirm ischemia is also recommended. Cardiac troponin I or other cardiac injury markers can also be measured in plasma or serum as surrogates for infarct size. Technical ischemia success should be confirmed for both MI models by confirming ST segment elevation and QTc prolongation. These alterations should normalize with timely reperfusion (136), and a lack of ECG changes during reperfusion can indicate permanent ligation (perhaps due to vessel damage). Telemetry can also be useful to monitor postsurgical intervention ECG in conscious and unrestrained mice. Noninvasive two-dimensional short and long-axis echocardiography can be used to estimate the degree of infarction by akinesis, wall thickness, and chamber dimensions (24). Cardiac output with a deformed LV free wall can be measured by Doppler echocardiography of pulmonary flow (125). In vivo infarct size can be noninvasively measured with late gadolinium enhancement MRI and independently corroborated by histology (42, 69). For terminal endpoints, LV pressure-volume catheterization is useful to determine load-dependent and load-independent conditions of systolic and diastolic function (137, 138). For nonreperfused MI, infarct sizes ≤30% (% infarct to total LV size) are typically excluded because of the absence of remodeling. If included, small and large infarcts should be grouped separately to reduce possible type II statistical errors. A necropsy is strongly recommended for all mice that die prematurely to determine the cause of death and assess early deaths due to technical issues for improvement. Perioperative death within 24 h of MI in nonreperfused mice is usually due to surgical errors (e.g., excessive bleeding or lung injury) or very large infarct sizes and arrhythmia. In mice with nonreperfused MI, postoperative deaths typically occur during days 3–7 after MI and are most often due to cardiac rupture, acute heart failure, or arrhythmias. Table 3 lists suggested inclusion and exclusion criteria for reperfused and nonreperfused MI experiments. We recommend that studies be randomized and blinded, with operators remaining blinded during treatment, data acquisition (e.g., functional diagnostic by echocardiography), and analysis (139–141).

Table 3.

Inclusion and exclusion criteria for reperfused and nonreperfused MI experiments

	Inclusion Criteria	Exclusion Criteria
Reperfused	Confirmed ligation by electrocardiogram (ECG, ST segment elevation) and visual inspection (blanching and akinesis downstream of ligation) Confirmed reperfusion by ECG (resolution of ST segment elevation) and visual inspection (return of coloration and movement)	Failure to confirm ischemia or reperfusion Inconsistent ischemia period Death due to a technical error
Nonreperfused	Confirmed ligation by ECG (ST segment elevation), visual inspection (blanching and akinesis downstream of ligation), or echocardiography assessment after ligation (wall motion abnormalities and reduced ejection fraction) Infarct size ≥30% measured in the acute phase (i.e., MI days 1–3)	Perioperative death (within 24 h of ligation, which generally indicates a technical error) Infarct size <30% Ejection fraction >40% or fractional shortening >20% at MI day 1

These criteria are indicated for control wild-type mice, and adjustments may need to be considered for mice that are genetically modified or treated. MI, myocardial infarction.

When selecting optimal time points for evaluation, know that timing is everything. We noticed from our evaluation that there was a wide range of options for both duration of ischemia and time of end point assessment. For reperfused MI studies, ischemia times ranged from 30 to 60 min and evaluation time ranged from 2.5 h to 35 days after reperfusion, with no uniformity across studies. For nonreperfused MI studies, the evaluation time ranged from 7 days to 4 wk to 12 wk and showed more uniformity across studies. The optimal design of studies using mouse models of reperfused or nonreperfused MI requires an understanding of the temporal sequence of the events associated with injury and repair (1, 142–147).

Traditional concepts on the time course of ischemic myocardial injury, derived from seminal studies in the canine model, suggest that cardiomyocyte death progresses with increased duration of the ischemic insult as a wavefront of necrosis from the more susceptible subendocardium to the less vulnerable subepicardium (148, 149). A similar wavefront of cardiomyocyte death likely occurs in the mouse model of coronary occlusion (6). This wavefront leads to expansion of the infarct zone with increasing duration of ischemia, from a thin area of dead cardiomyocytes in the subendocardium and mid-myocardium after 30 min of coronary occlusion to a more extensive infarct that becomes transmural throughout the area at risk after 4–6 h of ischemia (1). Studies using the reperfused MI model to dissect pathways regulating cardiomyocyte survival in response to injury typically use ischemia intervals between 45 and 60 min, thus allowing evaluation of interventions that may either accelerate or delay the wavefront of death. Shorter intervals of ischemia (<45 min) are associated with very small infarcts and may hamper documentation of cardioprotection or cardiac repair interventions. With long coronary occlusion intervals (>2 h), the infarct involves almost the entire area at risk, precluding evaluation of interventions that may accentuate injury. Some investigators have used varying times of ischemia to differentiate genotypes with increased susceptibility from those with protective genotypes (e.g., 30 vs. 60 min), and shorter durations can be useful to reveal interventions that increase cardiac susceptibility to injury or dysfunction (150–153). The scientific question to be addressed is the driving aspect of the experimental design, and the exact time of ischemia or reperfusion should be determined based on that question. As a general rule for reperfused MI, starting out with ischemia of 45 min is recommended. For evaluation times for both reperfused and nonreperfused MI, any time in the first week is appropriate to determine how the inflammatory and healing responses are progressing, with days 1, 3, 5, and 7 being common times of evaluation in the current literature. For chronic evaluation, 4 wk is a common end point, with 2, 8, and 12 wk also being represented (12, 13, 154–162).

In adult mammals, repair of the infarcted heart is dependent on the formation of a collagen-based scar and requires sequential infiltration of the infarcted heart with phagocytic leukocytes, as well as activation of reparative myofibroblasts and vascular cells (163, 164). These cellular events can be studied using either reperfused or nonreperfused MI model. Because inflammation and repair are accelerated upon reperfusion, early time points need to be studied when using the reperfused MI model (56). In nonreperfused MI, the inflammatory response is initiated early upon ischemic injury and peaks between days 1 and 3 after coronary occlusion. Transcription of proinflammatory cytokine and chemokine genes exhibits an early peak that is followed by transient neutrophil infiltration peaking at MI days 1–3 (131, 165, 166). This is overlapped by monocyte recruitment (166–168), prolonged accumulation of macrophages (113, 169, 170), and activation of α-smooth muscle actin (α-SMA)-expressing myofibroblasts at MI days 3–7 (171). As the scar matures (beginning around MI day 7), infarct myofibroblasts transition to matrifibrocytes, which are specialized fibroblast-like cells that do not express α-SMA but may be involved in scar maintenance (172). Infarct neovessels acquire a coat comprised of mural cells, which may stabilize the vasculature and inhibit angiogenesis (144, 173, 174). Lymphatics also help to modulate inflammation resolution after MI (175). Healing of the infarct zone is associated with remodeling of the noninfarcted remote myocardium, as viable cardiomyocytes become hypertrophic and the chamber progressively dilates, presumably through the activation of matrix-degrading proteases and matricellular proteins in the cardiac interstitium (176). Optimal study of chronic dilative remodeling of the infarcted heart requires assessing chamber dimensions and function for at least 4 wk after coronary occlusion. Ultimately, the experimental question will drive the time point(s) selected. We recommend 28 days for a universal chronic time point, as this is the most common time of chronic evaluation in the literature (12, 13, 154–161). For exercise protocols that start at 4 wk after MI, the 12-wk time point is often used, and we refer to recent guidelines on using exercise protocols (177–179). The major reason for conforming times is to evaluate reproducibility of findings. In addition, compiling datasets across studies and teams that follow the same protocols will allow the use of big data tools to provide conserved observations, generate new hypotheses, and assess variability across laboratories.

Anesthesia and Analgesics

The majority of studies evaluated (80%) used the volatile anesthetic isoflurane for the MI surgery while postoperative analgesics varied. Several of these agents may influence the biology of the MI through effects on inflammation that should be considered (180). For example, isoflurane is often used as a volatile anesthetic that can systemically reduce levels of inflammatory cytokines and impact diastolic function, heart rate, and body temperature; therefore, prolonged exposure is not recommended (137, 181). The anesthetic ketamine can exhibit anti-inflammatory properties that directly ameliorate ischemia-reperfusion injury. Anesthetics and analgesics can also affect myocardial oxygen consumption and influence infarct size. Buprenorphine, one of the most used analgesics in MI models, has also been associated with suppression of inflammation in rodents (182). Carprofen is a nonsteriodal anti-inflammatory drug, and pretreatment of carprofen in a mouse nonreperfused MI model impairs the initiation of the inflammatory phase and delays resolution response to promote development of the cardiorenal syndrome (180, 183). The analgesic meloxicam can reduce infarct size during cerebral injury (184). Lidocaine has well-known anti-inflammatory properties and is rarely used in mice (184). Because anesthesia and analgesics are necessary components of the surgical protocol, it is important to consider the possible influence these agents may have on results and to consistently monitor for respiration, heart rate, and body temperature to ensure time and exposures are applied similarly across comparison groups.

Sex as a Biological Variable

In our evaluation, 47% of studies included both males and females, up from 21% found in an evaluation done in 2018 for studies using echocardiography (28). For nonreperfused MI, female mice survive better and have a lower incidence of rupture in nonsurvivors (15, 50, 123, 185). Women who present with MI are not reperfused as quickly as men and have lower rates of guideline-directed medical care (32). Unadjusted 5- and 10-yr MI mortality was reported to be higher in women than men, and sex-dependent differences became less prominent when corrected for age, comorbidities, and treatments (186). A recent study from Spain reported that women with STEMI had higher mortality than men, underwent invasive intervention procedures less often than men, and had higher in-hospital mortality when admitted for STEMI (187). Another recent study from India similarly reported worse in-hospital outcomes in women because they did not receive guideline-directed management (188), whereas women undergoing percutaneous coronary intervention showed better outcomes (lower all-cause death) compared with men (189). In patients with type 2 diabetes, the risk of death following MI was lower in women than men (186–190). When systems were implemented to remove sex disparities in care, the differences in outcomes resolved (191), indicating the increased MI mortality in females is not due to underlying biological differences. Therefore, our understanding of MI responses across sexes/genders is still rudimentary and future studies on this important topic are warranted (192).

Because MI is prevalent in humans regardless of sex or gender, we recommend using both male and female mice. The National Institutes of Health requires that consideration be given to sex as a biological variable, and AJP-Heart and Circ has recently published an editorial on journal expectations for use of sex and gender (192). There was a lack of consensus by the writing team regarding whether sexes or genders should be combined for analysis. Using both sexes in a combined analysis will allow identification of pathways and targets that affect both sexes equally. At the same time, the best way to formulate a therapeutic approach for sexes may entail identifying similarities and differences with tailored intervention by sex/gender. Evaluating by gender in humans and sex in mice allowed the elucidation of the role of LXR/RXR signaling in female versus male neutrophil responses to MI (123). Initial concerns that female mice may show high variability has not stood up in the data; in fact, female mice often have lower standard deviations than male counterparts (90, 193–195). Although male and female mouse hearts remodel after MI, the exact underlying mechanisms may vary. Furthermore, even if no overt sex differences are observed with MI in wild-type mice or control settings, there could be sex differences in response to specific treatments, particularly when interactions with G protein-coupled receptors or other hormone-related receptors are involved (196–199) Together, the existing data indicate a strong reason to evaluate MI remodeling in both sexes/genders.

If there is no a priori rationale to segregate or bias to one sex, then equal numbers of both sexes should be included under the assumption of nonvariation (e.g., n = 8, 4 M/4 F). The aggregated data should be reported with indication within the group as to sex of each individual. If there is an a priori rationale to segregate (e.g., evaluating LV rupture as a primary outcome of interest), this should be stated and cited in the methods to justify separate sex analysis. If a mouse breeding colony is being maintained for study enrollment, there is an ethical responsibility to ensure both sexes are responsibly used to avoid the unnecessary generation of animals (200, 201).

Time of Day Considerations

Circadian rhythm is essential for healthy cardiovascular physiology, and occurrence of MI exhibits a diurnal rhythm with increased incidence of patients presenting early in the morning compared with any other time of the day (202, 203). In studies when MI was induced in mice during wake versus sleep hours, higher mortality was reported in the sleep-time MI cohort (204). The Steffens laboratory has demonstrated that mice show larger infarct size and greater neutrophil infiltration and more adverse remodeling when coronary ligation is performed in the evening, compared with the morning (204). In addition, the time of MI-triggered expression of a unique set of genes with higher inflammatory and cytokine expression following sleep-time MI (205). Diurnal fluctuation in blood monocyte count has been reported in humans (206). Differences are reported to be more relevant to permanent MI, whereas reperfused MI may not exhibit a circadian rhythm (207). Alternating group randomization during the operational window is recommended (i.e., do not work on all males, followed by all females, nor a series of controls, followed by a series of treatment groups). We recommend that similar time of day (e.g., 8:00 AM to 12:00 PM or 11:00 AM to 3:00 PM) be used across your experimental protocol and reported in methods, such that group differences influenced by circadian rhythms are mitigated.

Comorbidities

In addition to considering biological variables (i.e., age, sex, and circadian rhythm), the presence of comorbidities that replicate the human scenario more closely have been used with MI models in mice. At 3–6 mo of age, mice are homogenous in physiological maturation, including body weight (208–210), and MI has been reported in adult mice ranging in age from 6 wk to 36 mo (186–190). In this context, it is relevant to note that increased age in humans reduces sex-specific associations between risk factors and MI (211). Mouse MI models have been used in various experimental contexts that mimic different clinical scenarios to enhance our understanding of the role of comorbidities in the response of the heart to injury.

In addition to aging (212–215), other comorbidities examined in MI models include dietary interventions with or without aging (216, 217), preexposure to lipopolysaccharide from Porphyromonas gingivalis (218, 219), preexisting atherosclerosis (220), and other comorbid conditions prevalent in patients with coronary disease that may contribute to coronary events (41, 221). In both MI models, coexistence of morbidities such as obesity, metabolic syndrome, lifestyle factor modifications (sleep, stress, exercise), or aging increases the translational relevance of the study, as these are often associated with MI in humans. MI in diabetic or hyperlipidemic mice can provide information on the impact of common comorbid conditions on ischemia-induced cell death, repair, remodeling, and fibrosis, thus contributing to our understanding of the highly heterogeneous spectrum of clinical responses (41, 222). In addition, other comorbidities can be induced by MI itself to worsen outcomes; one example is cardiac dyssynchrony (223).

To summarize our recommendations for experimental design, we suggest using mice of similar age, of both sexes, and performing MI surgery and tissue collection at a consistent time of day for all mice, preferably within a 4-h window. All these factors should be considered when designing, analyzing, and reporting experimental studies.

Reporting Details

Over the last decade, there has been an increasing recognition for the necessity of established standard operating procedures and guidelines related to the rigor and reproducibility of reported scientific data. Several reports highlighted the lack of reproducibility in preclinical animal research, particularly from findings presented in high-profile journals (224–228). Lack of rigor or reproducibility affects the rate of basic research discovery and the success of clinical translation (225). To avoid biased, false, or incorrect reporting, rigor and reproducibility must be included within the study design and reporting. Table 4 provides a checklist for reporting MI study results that give consideration for rigor and reproducibility.

Table 4.

Reporting checklist, modified from the Animals in Research: reporting in vivo Experiments (ARRIVE) Guidelines

Item	Details
Abstract: report of relevant details	• Provide summary of research background, rationale, objectives • Give details to the experimental approach, report animal species and the animal model used
Ethical statements	• Institutional Animal Care and Use Committee approval, Guide for the Care and Use of Laboratory Animals, welfare assessments and interventions
Animal description, housing, husbandry, animal care, and monitoring	• Species, strain, source, age (mean and range), sex, genotypes, body weight (mean and range), health/immune status, housing • Type (specific pathogen free), cage type, bedding material, number of cage companions, light-dark cycle, room • Temperature and humidity, food type, food, and water access • Report interventions/steps to reduce pain, suffering and distress and report animal monitoring (if necessary) • Report anesthesia and analgesic use and pain monitoring; surgical procedure details and monitoring, method of euthanasia • Report expected and unexpected adverse events
Study design	• Define model used • Define groups; matching, randomization, and blinding protocols; order of treatment and assessment of groups; method to confirm model success • Report inclusion/exclusion criteria (e.g., hemodynamics or lower limit for infarct size) • Report daily monitoring to record time and cause of death • Define timing of perioperative and postoperative phases for survival analysis, flow chart for complex designs
Interpretation and translation	• Comment on study limitations, limitations to the animal model used • Comment on the relevance to human myocardial infarction
Experimental procedures	• Define area examined (e.g., infarct, remote, both regions) • Define positive and negative controls; for drugs: formulation, dose, site, and route of administration • Record and report the recorded data (e.g., electrocardiogram, heart rate, and anesthesia depth) • Report time of day performed
Sample sizes	• Report sample size calculation • Report number of animals used per group for each experiment • List any animals or samples removed from analysis with reason
Variables measured and data access	• Report primary and secondary end points • Provide a statement describing if and where data are available
Statistics	• Methods used for each analysis, test for assumptions • Report a measure of variability (e.g., means ± standard deviation or median and range)

For each study, a hypothesis should be stated (as appropriate), and inclusion and exclusion experimental criteria should be predefined to determine true outliers before experiments are initiated (229). Details on the rationale for specific ischemia and reperfusion times for reperfused MI must be provided in the methods. Finally, sample size and outcome measures should be determined a priori (230, 231). Power analyses are warranted to determine the sample size that can detect a relevant treatment effect based upon the primary outcome measure evaluated. Several statistical programs allow for these calculations and, in general, include four basic components: 1) type I error (α, commonly fixed at 0.05), which measures the probability that the differences found are likely to happen; 2) type II error (β, commonly set at 0.20), which accounts for false negatives and represents the power of a study (i.e., a β of 0.20 represents 0.80 power or 80% probability of avoiding a false negative conclusion); 3) the smallest effect of interest, which represents the minimal difference between the studied groups that the investigator wishes to detect (often referred to as the minimal difference physiologically/biologically relevant or effect size); and 4) variability, for which preliminary or published data can be used to estimate the standard deviation of a given outcome variable (232). For more details on use of statistics, please see the guidelines Statistical Considerations in Reporting Cardiovascular Research (29).

For transparent reporting, we recommend researchers adhere to the ARRIVE Guidelines (127). These guidelines were initially developed in 2010 to improve the reporting of in vivo research to ensure a comprehensive and transparent description; thus, better enabling others to evaluate and reproduce the study. The guidelines were recently updated, and information was reorganized to facilitate better use in practice (233).

DISCUSSION

The approach selected (reperfused vs. nonreperfused MI) will depend on the questions being asked. Likewise, the inputs examined, outputs measured, and time(s) of evaluation are dependent on study goals. There is no one gold standard to use for experimental design of MI studies. The focus instead should be on the design that most appropriately addresses the specific hypothesis being tested. Both reperfused and nonreperfused MI models are well suited to study repair and remodeling. Reperfused MI is mandatory for studying cardioprotection, and ischemia must be of sufficient severity and duration to induce ischemic injury and remodeling. Consideration should be given to the question being asked and to prospectively plan the study design around that question. This includes determining the best controls, randomization, blinding schemes, and analysis using appropriate statistics before experiments commence.

MI models are essential for translational research. Only under in vivo conditions can the effect of an intervention by adequately studied to assess the full interaction among different cell and molecular components to influence overall LV remodeling or healing after MI. Several examples highlight the translational potential of animal models. A classic example of a bench to bedside translation is the use of ACE inhibitors as standard treatment of MI; these studies were initially undertaken in a rodent permanent occlusion MI model by Janice Pfeffer and colleagues (234). There are clear limitations to clinical translation that include the small size of the mouse, which makes arrhythmias difficult to detect; different healing patterns (frequent LV rupture after nonreperfused MI); the predominant use of young adult animals without comorbidities; a lack of genetic variations in the mouse strains used; and cleaner working conditions (lack of regular infections in animal facilities and specific diets). Another caveat that must be considered is that in the murine models described, there is no inciting thrombotic event causing myocardial ischemia, which makes it difficult to determine the impact of platelet activation and release of numerous inflammatory mediators that contribute to microvascular obstruction. Add to this list the potential variability in experimental conditions across laboratories and differences can be observed between preclinical research and the actual clinical practice. The last point can be avoided and is a prime motivation for the writing of this article.

In conclusion, these Guidelines provide best practice details to assist investigators in the planning and executing studies involving MI in mice. This consensus is part of ongoing efforts by AJP-Heart and Circ to provide guidelines articles that help to improve the rigor and reproducibility of cardiovascular physiology studies (28, 29, 47, 179, 235, 236).

GRANTS

This work was funded by National Institutes of Health Grants GM115458 and HL137319 (to M. L. Lindsey); AA027625, HL136737, and HL147570 (to J. A. Kirk); HL076246, HL085440, and HL149407 (to N. G. Frangogiannis); HL111600 (to C. M. Ripplinger); HL122309 (to E. B. Thorp); HL147844, GM127607, and HL078825 (to S. P. Jones); HL132989 and HL144788 (to G. V. Halade); HL145817 (to K. Y. DeLeon-Pennell); HL130972 (to F. G. Spinale); HL127442 (to R. J. Gumina); DK115213 and HL136915 (to J. W. Calvert); and HL152297 (to L. E. C. Brás); Biomedical Laboratory Research and Development Service of the Veterans Affairs Office of Research and Development Grants BX000168 (to F. G. Spinale), BX000505 (to M. L. Lindsey), and BX003922 (to K. Y. DeLeon-Pennell); American Heart Association Innovator Project IPA35260039 (KYD-P); United States Department of Defense Grant PR181464 (to N. G. Frangogiannis); Natural Sciences Engineering Research Council Grants RGPIN-05520 and RGPIN-04878 (to K. R. Brunt); Canadian Institutes of Health Research Grants PJT-37522 and PJT-153306 (to Z. Kassiri) and PJT-421341 and PJo-413883 (to K. R. Brunt); Canadian Foundation for Innovation Grant 31708 (to K. R. Brunt); Heart and Stroke Foundation of Canada Grants G-19–0026282 (to Z. Kassiri) and G-16–00014524 (to K. R. Brunt); Heart and Stroke Foundation of New Brunswick (to K. R. Brunt); American Heart Association Grants 18AIREA33960311 (to L. E. C. Brás) and TPA2035490150 (to D. P. Del Re.); National Health and Medical Research Council of Australia Grant APP1158013 (to R. H. Ritchie); and New Brunswick Health Research and Innovation Foundations (to K. R. Brunt).

DISCLAIMERS

The content is solely the responsibility of the authors and does not necessarily represent the official views of any of the funding agencies or the American Physiological Society. All authors have reviewed and approved the article.

DISCLOSURES

M. L. Lindsey, K. R. Brunt, J. A. Kirk, P. Kleinbongard, C. M. Ripplinger, and Z. Kassiri are editors of AJP-Heart and Circ. Given their roles, none had any involvement in the peer review of this article outside of being authors and had no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to Jason Carter.

AUTHOR CONTRIBUTIONS

M.L.L., J.A.K., and P.K. conceived and designed research; M.L.L., L.E.d.C.B., and K.Y.D.-P. analyzed data; M.L.L., L.E.d.C.B., and K.Y.D.-P. interpreted results of experiments; M.L.L., K.R.B., J.A.K., P.K., J.W.C., L.E.d.C.B., K.Y.D.-P., D.P.D.R., N.G.F. S.F., R.J.G., G.V.H., S.P.J., R.H.R., F.G.S., E.B.T., C.M.R., and Z.K., drafted manuscript; M.L.L., K.R.B., J.A.K., P.K., J.W.C., L.E.d.C.B., K.Y.D.-P., D.P.D.R., N.G.F. S.F., R.J.G., G.V.H., S.P.J., R.H.R., F.G.S., E.B.T., C.M.R., and Z.K., edited and revised manuscript; M.L.L., K.R.B., J.A.K., P.K., J.W.C., L.E.d.C.B., K.Y.D.-P., D.P.D.R., N.G.F., S.F., R.J.G., G.V.H., S.P.J., R.H.R., F.G.S., E.B.T., C.M.R., and Z.K. approved final version of manuscript.