Cardiovascular magnetic resonance imaging for sequential assessment of cardiac fibrosis in mice: technical advancements and reverse translation

Abstract

Cardiovascular magnetic resonance (CMR) imaging has become an essential technique for the assessment of cardiac function and morphology, and is now routinely used to monitor disease progression and intervention efficacy in the clinic. Cardiac fibrosis is a common characteristic of numerous cardiovascular diseases and often precedes cardiac dysfunction and heart failure. Hence, the detection of cardiac fibrosis is important for both early diagnosis and the provision of guidance for interventions/therapies. Experimental mouse models with genetically and/or surgically induced disease have been widely used to understand mechanisms underlying cardiac fibrosis and to assess new treatment strategies. Improving the appropriate applications of CMR to mouse studies of cardiac fibrosis has the potential to generate new knowledge, and more accurately examine the safety and efficacy of antifibrotic therapies. In this review, we provide 1) a brief overview of different types of cardiac fibrosis, 2) general background on magnetic resonance imaging (MRI), 3) a summary of different CMR techniques used in mice for the assessment of cardiac fibrosis including experimental and technical considerations (contrast agents and pulse sequences), and 4) provide an overview of mouse studies that have serially monitored cardiac fibrosis during disease progression and/or therapeutic interventions. Clinically established CMR protocols have advanced mouse CMR for the detection of cardiac fibrosis, and there is hope that discovery studies in mice will identify new antifibrotic therapies for patients, highlighting the value of both reverse translation and bench-to-bedside research.

INTRODUCTION

Cardiovascular magnetic resonance (CMR) imaging is well established as one of the most valuable tools for the diagnosis of heart disease in both clinical and preclinical settings. Improvements in magnetic resonance (MR) scanner hardware and software have rapidly increased the performance, reliability, and accuracy of clinical CMR. CMR provides excellent spatial resolution and tissue contrast, and coupled with its ability to characterize soft tissues, gives it practical advantages over other imaging techniques such as echocardiography and computed tomography both in humans and in mice (1–3). Clinical MR scanners typically have a 1.5 or 3 T magnet with a bore size of ∼60–70 cm and have a typical cine temporal resolution of 30–50 ms and spatial resolution of 1–3 mm, which is sufficient for clinical applications in humans. On the other hand, small animal systems, designed for imaging mice and rats, have a 4.7- to 18-T magnet, a smaller magnet bore size (∼10–40 cm), and stronger gradient systems (3). The temporal resolution (∼10–20 ms) and spatial resolution (∼10–100 µm) of small animal preclinical scanners is generally higher than that of clinical MR scanners (4).

Experimental mouse models are frequently used to study cardiovascular disease because of the availability of genetically modified mice, the ability to mimic various aspects of human cardiac pathology through surgical procedures (e.g., pressure overload and myocardial infarction, MI), as well as the relatively short lifespan (mice are considered adults by 8–10 wk of age), and relatively low housing costs (5). The development of novel MR techniques for the imaging of mouse hearts in vivo can be technically challenging compared with other organs/ex vivo imaging, primarily because of the rapid heart rates (500–600 beats/min) and asynchronous breathing (60–100 breaths/min). Specialist expertise and equipment are required to overcome these challenges. For instance, the higher temporal resolution achievable by small animal preclinical scanners allows for more detailed imaging of cardiac function and structure in small animals, which is important for preclinical research. Despite these challenges, CMR offers the most accurate technique to characterize myocardial tissue and provides unrivaled spatial resolution of heart anatomy in vivo. In addition, the low test-retest variability of CMR enables accurate and repeatable noninvasive assessments of both cardiac function and morphology over time (6).

Cardiac fibrosis is a common characteristic of many genetic and acquired cardiovascular diseases. Fibrosis can broadly be categorized as either focal or diffuse/interstitial based on its distribution (Fig. 1A). Different CMR techniques can be used to assess the different types of cardiac fibrosis (Fig. 1, B–D) and will be discussed in detail later. Focal fibrosis is localized and occurs after significant cardiomyocyte necrosis, for example, in response to MI. Diffuse fibrosis can be characterized by a spread of extracellular collagen throughout the heart in the absence of localized necrotic myocardium (9). The distribution patterns of fibrosis differ between pathologies and disease states. Therefore, detecting and distinguishing between different types of cardiac fibrosis is crucial for several reasons. First, it enables earlier diagnosis and better risk stratification. Second, it may help guide the selection of appropriate therapies, as different distributions of cardiac fibrosis may respond differently to treatments. Finally, it provides a better understanding of the mechanisms that initiate, progress, and resolve cardiac fibrosis, which can lead to the identification of new therapeutic targets and the development of more effective treatments for fibrotic diseases. Imaging techniques to assess cardiac fibrosis are likely to provide more comprehensive diagnostic or prognostic information than the assessment of fibrosis with circulating profibrotic biomarkers or endomyocardial biopsy (9). The focus of this review is to describe the latest applications of CMR in assessing cardiac fibrosis in mice. Although traditional assessments of cardiac geometry and function have been well established [reviewed elsewhere (10)], recent advances in MR technology have improved the feasibility of detecting the presence and extent of fibrosis in the mouse heart. In this review, we will explore the techniques and challenges involved in using CMR to measure fibrosis in mice. In addition, we will discuss the potential implications of this approach in enhancing our understanding of cardiac disease and assessing new therapeutic interventions.

Figure 1.

Overview of different types of cardiac fibrosis and schematics of cardiac magnetic resonance imaging (MRI) techniques in a short-axis view. A: there are two broad types of cardiac fibrosis based on differences in distribution: 1) focal fibrosis and 2) diffuse fibrosis. These types can be further classified as replacement fibrosis, reactive interstitial fibrosis, and infiltrative interstitial fibrosis. B: T1-weighted images without contrast in cardiovascular magnetic resonance (CMR) primarily provide information about the anatomy and function of the heart. These images depict the natural contrast between different tissues based on their T1 relaxation times. Late gadolinium enhancement (LGE) imaging is a specific type of T1-weighted imaging technique that is commonly used to detect and characterize fibrosis in CMR, less sensitive for detecting diffuse fibrosis. C: native/precontrast T1 (Pre-T1)-mapping values in healthy myocardial tissue reflect the inherent T1 relaxation time (fibrosis precontrast T1 values being longer than healthy heart), whereas postcontrast T1 (Post-T1)-mapping values change after the administration of a gadolinium-based contrast agent. In healthy myocardium, postcontrast T1-mapping values typically decrease, resulting in enhanced signal intensity, whereas in cardiac fibrosis, post-T1-mapping values often show delayed and heterogeneous enhancement, with fibrotic areas exhibiting shorter values compared with healthy tissue (fibrosis postcontrast T1 values being shorter than healthy heart). To be noted, native T1 has been shown to be less sensitive to cardiac fibrosis compared with post-T1. Focal cardiac fibrosis can be detected in both Pre- and Post-T1 mappings, LGE, and D: extracellular volume (ECV) mapping. However, the sensitivity of native T1 and LGE is not ideal for detecting diffuse fibrosis. Arrows and red circles indicate the risk area of ischemic myocardial fibrosis. Schematics were generated based on information from Refs. 7 and 8.

PATHOPHYSIOLOGY OF MYOCARDIAL FIBROSIS

Myocardial fibrosis is a process of pathological remodeling that leads to the accumulation of extracellular matrix (ECM) in the myocardium and is an integral component in most cardiovascular diseases, including MI, hypertensive heart disease, cardiomyopathy, valvular heart disease, and ultimately its progression to heart failure (HF; 11–13). Fibrosis is a key contributor leading to cardiac dysfunction by affecting cardiac structure and electrical conduction. The distribution of cardiac fibrosis varies according to the cardiomyopathic triggers or processes (14–16).

Focal Fibrosis

Focal myocardial fibrosis is characterized by the macroscopic distribution of fibrosis in the heart, being typically formed by replacement fibrosis resulting from myocyte death following acute injury (e.g., ischemia/infarction etc.), and is irreversible (17). The death of cardiomyocytes triggers an inflammatory response that leads to fibroblast activation, and deposition of ECM components, resulting in the development of focal cardiac fibrosis (17). Focal fibrosis normally results in localized replacement of cardiomyocytes with collagenous scar tissue to stabilize the injured myocardium, redistributing shear stresses, and preventing cardiac rupture postinjury; it may also present in the later stages of HF in both humans and mouse models (1, 18, 19).

Diffuse Fibrosis

Diffuse cardiac fibrosis is a more widespread pathological process. There are three major processes for the formation of diffuse cardiac fibrosis, each associated with different underlying conditions, etiologies, and symptomology; these include replacement fibrosis, reactive interstitial fibrosis, and infiltrative interstitial fibrosis (1; Fig. 1A). Replacement diffuse fibrosis can be found in the heart in settings of toxic cardiomyopathies, hypertrophic cardiomyopathy, miscellaneous inflammatory diseases, or chronic renal insufficiency as these diseases lead to diffuse apoptosis within the heart (20–22). In contrast, reactive interstitial fibrosis is a gradual process where an accumulation of ECM deposition and an increase in collagen synthesis is caused by pressure overload and/or cardiomyopathy (22, 23); if the stimulus/cause is treated promptly, it may be reversed (23). Reactive interstitial fibrosis is commonly found in conditions of hypertension, diabetes mellitus, or advanced age (24–26), as well as in the noninfarcted myocardium post-MI (27). The third subtype of fibrosis is infiltrative interstitial fibrosis that is less common and caused primarily by the progressive deposition of glycolipids, for instance, accumulation of insoluble amyloid (amyloidosis) or glycosphingolipids (such as in Anderson–Fabry disease; 1, 28, 29).

BASIC MRI AND CMR

Introduction to MRI Systems and CMR Techniques

A simplified and broad overview of the basic MRI system and formation of CMR images is provided in Fig. 2, and discussed in more detail throughout the review. MRI systems contain three main components (Fig. 2, A and B): 1) main magnet (producing a strong constant/static magnetic field with typical clinical field strengths of 1.5 or 3 T and 4.7–18 T for preclinical systems, referred to as B₀), 2) gradient coils positioned inside the main magnet that generate controlled magnetic field gradients (x, y, z) for spatial encoding, and 3) radiofrequency (RF) coils (used to transmit and receive radiofrequency pulses within the gradient coil, referred to as the B₁ field; Fig. 2B) (30, 31). For cardiac imaging, specialized multichannel RF receiver coils are designed to fit around the chest area to maximize the signal-to-noise ratio (SNR; described further in RF Coils and Pulses) and in doing so, increase the possible spatial resolution of images obtained from the heart (32). These phased-array receiver coil systems are commonly used to achieve high-quality images in shorter acquisition times using a variety of acceleration techniques. The RF/B₁ field [smaller amplitude than B₀ and oscillates at a frequency referred to as the Larmor frequency = gyromagnetic ratio (γ) × B₀] combines with the static magnetic field to generate MR signals that are spatially localized (encoded by the gradient magnetic fields) to create the MR image (Fig. 2C).

Figure 2.

Schematic of basic magnetic resonance imaging (MRI) systems and the formation of cardiovascular magnetic resonance (CMR) images. A and B: MRI systems consist of three main components: 1) main magnet (B₀), 2) gradient coils positioned around the magnet generate controlled magnetic field gradients (x, y, z), and 3) radiofrequency (RF) transmitter coil, used to transmit and receive RF pulses for MRI. C: acquisition parameters, such as echo times (T_E) and repetition times (T_R), along with the chosen imaging sequences influence how T1 and T2 relaxation times contribute to the signal in k-space and ultimately the reconstructed image. K-space is a mathematical representation of the spatial frequency domain in CMR. It involves collecting multiple echoes with varying phase and amplitude encoding gradients to sample different spatial frequency components. The acquired k-space data are then Fourier transformed to reconstruct the final cardiac image in the spatial domain. Various techniques, such as parallel imaging and advanced pulse sequences, are used to accelerate data acquisition and improve imaging efficiency. By using k-space, CMR enables the visualization of the heart’s anatomy, function, and tissue characteristics, facilitating comprehensive cardiovascular assessment.

The MR signal typically originates from the hydrogen nuclei (single proton) within the tissue of interest (e.g., free water, blood, lipid molecules, etc.). Each proton generates a small magnetic field, the direction of which is random under basal conditions; an intrinsic property of hydrogen nuclei/single protons (Fig. 3A). However, when subjected to a static magnetic field, the protons align themselves along the field lines (Fig. 3B). This alignment leads to a property called precession, characterized by the spinning motion of protons around the magnetic field’s direction. A reference coordinate system (x, y, z) defines the magnetic field direction, with the z-axis in parallel to B₀. At equilibrium, the net magnetic field (M) from the protons is aligned along B₀ (positive z-axis) with the value M₀ (Fig. 3B). To examine tissue parameters, the system needs to be excited by applying an RF pulse (Fig. 3C). Resonance occurs when a RF pulse is applied to the protons at their precessional frequency, causing them to absorb energy and alter their alignment, which results in the net magnetization moving from the B₀ field alignment (z-axis) toward the x- and y-axes (from M₀ to M_xy; the direction of the RF pulse is referred to as B₁, Fig. 3C). The greater the energy from the RF pulse, the greater the angle (flip angle) the net magnetization moves from the B₀ field. The ability to control the flip angle is important in MRI, as it affects the resulting image contrast and signal intensity. Different pulse sequences and imaging goals may require specific flip angles to optimize the desired image characteristics. Once the desired flip angle is reached, the RF pulse is turned off. The new net magnetization has two components: 1) the z component (M_z)/longitudinal component and 2) the x-y component (M_xy)/transverse component (Fig. 3D). After the RF pulse has ceased, the protons start returning to their original/equilibrium state during which the MRI scanner detects the emitted energy/signals that can be spatially localized to generate the imaging data. This process of “relaxation” consists of two simultaneous processes: T1 relaxation and T2 relaxation (Fig. 3D). T1 relaxation represents the recovery of magnetization in the longitudinal plane (z plane), and T2 relaxation represents the decay of magnetization in the transverse plane (x-y plane; 30). T1 and T2 relaxation properties highlight signal differences from different tissue/organs based on water and fat content but also allow for assessment of tissue characterization, for example, fibrosis.

Figure 3.

Basic principles of magnetic resonance imaging (MRI). A: protons in the body without a magnetic field. B: when an external magnetic field B₀ (green arrow) is applied, protons tend to align with the direction of B₀. The sum of the protons is referred to the magnetization vector (M₀, blue arrow). C: if a second magnetic field [B₁, radiofrequency (RF) pulse] orthogonal to B₀ is applied, it is possible to tilt M₀ of 90° along the x-y direction (M_xy, smaller green arrow). D: when B₁ (RF pulse) is switched off, M_xy returns to the equilibrium through two processes: T1 and T2 relaxation. T1 relaxation is defined as the time needed to achieve 63% of the original longitudinal magnetization. Purple curve and the orange curve represent tissues with short and long T1 values, respectively. T2 relaxation is defined as the time to dephase up to 37% of the original value. Purple curve and the orange curve represent tissues with short and long T2 values, respectively. T2* relaxation represents the combined effect of T2 relaxation (intrinsic tissue effects, spin-spin relaxation) and the effect of magnetic field nonuniformities (extrinsic effects). Schematics were generated based on information from Ref. 33.

T1 is also referred to as “spin lattice” relaxation where “lattice” refers to the surrounding material. For each proton to move back into equilibrium (parallel to B₀) it must exchange its energy with the surrounding material/lattice. In fat, the protons have an efficient spin-lattice exchange, short T1, and generate the brightest signal. Whereas, in tissues with high water content, T1 is much longer because it takes longer for water protons to relax, and the signal is much weaker.

T2 is also referred to as “spin spin” relaxation because transverse relaxation is largely due to the intrinsic field caused by adjacent protons. When the RF pulse is first stopped, the protons rotate together in a coherent fashion, and are in the same direction as each other within the x-y plane, i.e., the proton spins are “in phase.” However, due to interactions between neighboring protons, and local static variations (inhomogeneities/nonuniformities) in the applied magnetic field, there is loss of coherence, and the magnetic spins of the protons move “out of phase” (also referred to as dephasing). The combined effect of T2 relaxation (intrinsic tissue effects) and the effect of magnetic field nonuniformities (extrinsic effects) is referred to as T2* relaxation (34).

RF Coils and Pulses

As noted earlier, special small RF (receiver) coils have been optimized for small animal size and designed to maximize the SNR and imaging resolution by providing greater homogeneity of the RF field (35). In MRI, the SNR is perceived from the quality of the image produced; it is characterized as the ratio of the strength of the MRI signal in the object being imaged to the strength of the unintentional nonspecific background noise in the image. As with most imaging modalities, a higher SNR indicates a better-quality image, with a clearer and more detailed representation of the object/features being imaged. In CMR, the SNR can be influenced by various factors, such as the field strength of the MRI magnet, the RF coil used to detect the signal, the amount of time taken to acquire the image, and the suitability of the sequence/imaging parameters used (36). To obtain high-quality images with a high SNR, it is important to optimize the MRI parameters and minimize sources of noise including physiological noise such as motion artifacts, and electronic noise. Several techniques can be used to improve the SNR and enhance imaging quality in CMR studies of mice and will be discussed further in later sections of this review.

MR imaging requires a series of RF pulses of defined magnitude and timing (MR pulse sequences). Various imaging parameters can be used to form pulse sequences in MRI. These parameters can be adjusted to achieve different contrasts, spatial resolution, and other imaging characteristics. A brief description of common imaging parameters, terminology, and relevance are provided in Table 1. This includes repetition time (T_R), echo time (T_E), flip angle (FA), inversion time (T_I), phase encoding, slice thickness, field of view (FOV), matrix size, echo train length, and spatial acquisition of k-space (Fig. 2C). By adjusting these and other imaging parameters, different pulse sequences can be created to achieve a wide range of imaging characteristics for various clinical and preclinical applications (see Ref. 30 for further details).

Table 1.

Imaging parameters and MR terminology

Imaging Parameters and Terminology	Description
Repetition time (TR):	The time between successive RF pulses. Longer TR values result in greater longitudinal magnetization recovery, whereas shorter TR values can enhance T1-weighted contrast.
Echo time (TE):	The time between the RF pulse and the acquisition of the MRI signal. Longer TE values can enhance T2-weighted contrast whereas shorter TE values are used for T1-weighted contrast.
Flip angle (FA):	The angle of the RF pulse used to excite the tissue magnetization. Greater FA can increase the signal intensity of T1-weighted images, whereas smaller flip angles can increase the signal intensity of T2-weighted images. The Ernst angle, which refers to the specific FA at which the signal intensity is maximized for the tissue of interest with a known T1 relaxation time, is used for optimizing tissue contrast.
Inversion time (TI):	The time between the inversion RF pulse and the excitation RF pulse. TI can be used to suppress the signal from certain tissues, such as fat or blood.
Phase encoding:	The application of additional gradients in the imaging plane to encode spatial information. Phase encoding can be used to form 2-D or 3-D images.
Slice thickness:	The thickness of the imaging slice. Thinner slices provide greater spatial resolution.
Field of view (FOV):	The area of the body that is imaged. A larger FOV can provide greater coverage, whereas a smaller FOV can provide greater spatial resolution for a given matrix size.
Matrix size:	The number of voxels used to form the image. A larger matrix size can provide greater spatial resolution for a given FOV.
Echo train length:	The number of echoes acquired in a single acquisition. Longer echo train lengths can improve the efficiency of the imaging sequence, but can also lead to greater susceptibility artifacts. Susceptibility artifacts are primarily caused by magnetic field distortions arising from variations in magnetic susceptibility between different tissues or interfaces within the imaged object.
Spatial acquisition of k-space:	k-space is a mathematical space where the raw MRI data are collected before being transformed into the final image (Fig. 2C), and the spatial acquisition of k-space determines the spatial resolution and contrast of the final image. A higher spatial resolution can be achieved by acquiring more lines of k-space, while a greater contrast can be achieved by adjusting the timing and amplitude of the RF and gradient pulses used to acquire the data.

MRI, magnetic resonance imaging; RF, radiofrequency.

BASIC CMR PULSE SEQUENCES

An MRI pulse sequence is a programmed set of instructions that controls the timing of RF pulses (as well as their magnitude), gradient switching (and magnitude), and data acquisition. Multiple pulse sequences can be grouped into the MRI scanning protocol and applied to produce images with particular characteristics to visually distinguish between tissues and/or organs. Depending on the MRI scanner, there may be a variety of pulse sequences installed, and each sequence may have different options or parameters that can be adjusted to optimize image quality for a particular application. The following are some of the basic pulse sequences for CMR:

• 1)Spin echo (SE) sequence: A common pulse sequence that involves a 90° RF pulse followed by one or more 180° RF refocusing pulses. This results in the rephasing of the dephased protons, generating an SE signal. SE sequences are commonly used for T1-weighted and T2-weighted imaging, providing different types of tissue contrast. In CMR, SE sequences are used to detect fibrosis by producing T1/T2 images with low signal intensity in fibrotic areas (37).
• 2)Inversion recovery (IR) sequence: the IR sequence is a specialized MRI sequence used to enhance tissue contrast by nulling or suppressing specific signal intensities. It involves applying a 180° RF pulse to invert the magnetization of certain tissues before acquiring the image data. This allows for the visualization and characterization of specific tissues or pathologies that exhibit different signal behaviors. In CMR, the IR sequence is commonly used for assessing myocardial viability and detecting areas of MI (focal fibrosis). By adjusting the inversion time, healthy myocardial tissue can be nulled, whereas infarcted or scarred tissue appears bright in the resulting images due to the T1 shortening properties of a gadolinium (Gd)-based contrast agent.
• 3)Gradient echo (GRE) sequence: The GRE sequence uses gradient magnetic fields to encode spatial information in the MRI signal. It does not involve a 180° RF pulse, which leads to a faster imaging time compared with the SE sequence. GRE sequences generate T2*-weighted images, which are sensitive to variations in magnetic susceptibility. In CMR, GRE sequences are used to detect fibrosis by producing T2*-weighted images with low signal intensity in fibrotic areas because of the altered magnetic properties of these tissues.

It is important to note that the choice of pulse sequence depends on the specific clinical or research question and the goals of the study. In addition, there are many variations and combinations of these basic pulse sequences that can be used to produce more specialized images for assessing fibrosis in CMR. Further to preinstalled pulse sequences, MRI scanners offer the ability to create custom pulse sequences, which can be tailored to optimize image quality for specific research or clinical question. However, creating custom pulse sequences can require specialized knowledge and expertise in MRI physics, pulse programming, and/or engineering.

MRI TECHNIQUES FOR THE DETECTION OF CARDIAC FIBROSIS

In the clinic, although CMR is the most accurate technique for assessing cardiac structure and function (particularly in settings of obesity and restricted imaging windows), it is unique among imaging modalities in its ability to characterize myocardial tissue, and is commonly used as a noninvasive technique for assessing myocardial fibrosis. In mice, where imaging windows are typically less of a limitation, the ability to assess tissue characterization has been a major driver for developing CMR techniques. The detection of cardiac fibrosis in mice has become possible with improvements in technology, the use of high-field strength MR systems (≥4.7 T), dedicated surface coils, and multiple pulse sequences (3, 38, 39).

Weighted Imaging vs. Parametric Mapping

T1/T2-weighted imaging and parametric mapping in CMR can both be used to detect cardiac fibrosis. Weighted imaging in CMR can provide semiquantitative analysis of cardiac fibrosis based on signal intensity ratios or differences. T1-weighted imaging in conjunction with a Gd-based contrast agent is particularly useful for detecting areas of focal fibrosis and defining the borders with healthy myocardium, since areas of focal fibrosis will typically appear clearly as regions of considerably higher signal intensity. However, standard assessment of Gd enhancement requires designation of a reference normal segment, which is not available in the presence of diffuse fibrosis (Fig. 1B). In contrast, parametric mapping provides a pixel-by-pixel map of magnetic relaxation parameters (absolutely denominated numerical T1/T2 properties) expressed in units of time (Fig. 1, C and D). T1 mapping can be used to quantify fibrosis with or without contrast agents and allows a direct quantitative comparison of parametric maps between patients/animals, as well as within patients/animals longitudinally (40). Extracellular volume (ECV) mapping can also be used to quantify the amount of ECM expansion that occurs with fibrosis (Fig. 1D).

MR techniques to characterize fibrosis include: 1) late-Gd enhancement (LGE) imaging, 2) native T1 mapping (without contrast), 3) postcontrast T1 and ECV mapping, and 4) T2 and T2* mapping. More advanced CMR techniques including diffusion-weighted imaging (DWI) and T1-rho (T1ρ) have also been applied and are discussed in a later section (and Table 2; including advantages and disadvantages).

Table 2.

Advantages and disadvantages of CMR techniques in assessing fibrosis in humans and mice

Imaging Technique	Description	Advantages	Disadvantages
LGE	Based on differences in contrast washout between normal and pathological myocardium.	• Reference standard method for focal fibrosis.	• Semiquantitative analysis. • Not suitable for diffuse fibrosis. • Not suitable for patients or mouse models with severe renal impairment.
Native endogenous • T1 mapping	Based on differences in T1 relaxation between normal and pathological myocardium without contrast.	• Assessment of diffuse fibrosis,useful tool for tissue damage assessment. • Provides a quantitative value for fibrosis assessment expressed in units of time. • Exogenous contrast agents not required.	• Low sensitivity for fibrosis and normal myocardial differentiation.
Contrast enhanced • T1 mapping • ECV mapping	T1 mapping: relies on differences in T1 relaxation between normal and pathological myocardium with contrast enhancement. ECV mapping: based on the value calculated from pre- and postcontrast T1 mapping.	• Provides a quantitative value for fibrosis assessment. T1 mapping: • Commonly used for focal fibrosis assessment. ECV mapping: • Shows good potential in assessing diffuse fibrosis for disease progression or post-therapeutic follow-up.	• Not suitable for patients or mouse models with severe renal impairment. • Long acquisition time as both pre- and postcontrast scanning is required for ECV.
T2 and T2* mapping	T2: relies on tissue composition. T2*: depends on structure of tissue.	• Shows potential of quantitative assessment for cardiac fibrosis. • Exogenous contrast not required.	• Has not been clinically recognized for assessing cardiac fibrosis. • T2*: shimming is crucial. • Unclear whether hemorrhage can be distinguished from fibrosis. • T2*: false positive area at the lung-heart interface.
T1ρ mapping	Advanced MRI technique uses spin lock pulse to measure spin lattice relaxation time.	• Provides a quantitative value for fibrosis assessment. • Exogenous contrast not required.	• Has not been clinically recognized for assessing cardiac fibrosis. • Technical development for fast and accurate T1ρ measurement for fibrotic myocardium in mice is needed.
DWI	Noninvasive tracking of myocardial fiber orientations and the quantification of myocardial fiber disarray.	• Shows potential as a quantitative assessment for cardiac fibrosis. • Exogenous contrast not required.	• Has not been clinically recognized for assessing cardiac fibrosis.

CMR, cardiovascular magnetic resonance imaging; DWI, diffusion-weighted imaging; ECV, extracellular volume fraction; LGE, late gadolinium enhancement; T1ρ, T1 rho. Information from the table comes from Refs. 3 and 41.

LGE.

LGE imaging is widely considered to be the reference standard CMR technique for the assessment of myocardial scars and focal fibrosis, but not diffuse fibrosis (42, 43) (Fig. 1, A and B). Briefly, Gd is a biological tracer that freely distributes in the extracellular space after injection but is unable to move across the intact cell membrane. An increase in Gd concentration shortens T1 relaxation time and produces a high-intensity signal on T1-weighted imaging. Continued imaging after the initial uptake and washout of Gd in healthy myocardium allows the detection of diseased myocardium because of the increased volume of distribution and slow clearance rate of Gd in the myocardium containing focal fibrosis. LGE imaging can clearly identify focal fibrosis (bright) from healthy myocardium (dark) by enhancing the signal achievable from these areas and thus the contrast with the surrounding myocardium (44, 45) (Fig. 1B).

Clinically, Gd-based contrast agents are injected intravenously for LGE imaging, whereas in animal models both intravenous (iv) and intraperitoneal (ip) injections can be used (46). A recommended delay stabilizes T1 variation after contrast agent injection, ensuring reliable T1-weighted images; duration depends on contrast agent, protocol, and objectives, allowing agent-tissue interaction for a stable state, minimizing T1 variation, and enhancing imaging accuracy (7, 44). Clinically and in mice, LGE MRI is achieved by using a T1-weighted inverse recovery (IR) sequence (47, 48). The inversion time of the IR sequence is chosen to null the signal of the healthy myocardium after contrast administration to improve the detection of focal fibrosis. The optimal T_I needs to be identified before imaging, which can be achieved by using an IR sequence (Look-Locker) with multiple T_Is (48). Bright blood T1-weighted Cine fast low angle shot (FLASH) sequence can also be used as an alternative to IR LGE CMR (44). However, the signal intensity of the areas of myocardial delayed enhancement might have a similar value to the blood pool, making it difficult to identify subendocardial focal fibrosis. Dark-blood T1-weighted LGE is able to increase the contrast between scars and blood (49). The comparison between bright blood and dark blood LGE is discussed in more detail in a later section on experimental technical considerations for CMR.

Native T1 mapping (without contrast).

Typically, weighted images are generated by MR to provide visual contrast between different tissues or organs (semiquantitative imaging). Mapping of relaxation values (e.g., T1, T2, and T2*) can provide additional quantitative myocardial tissue characterization. In contrast to traditional LGE T1-weighted imaging, cardiac T1 mapping can quantify T1 relaxation on a voxel-by-voxel basis. T1 mapping has proven to be useful in many cardiac pathologies associated with diffuse fibrosis (7, 43, 50) (Fig. 1C).

T1 mapping uses multiple inversion recovery images with different T_Is to assess the whole recovery of the longitudinal magnetization after inversion, and the signal intensities of these multiple images with different T1 weightings can be used to fit the equation for T1 relaxation. This method can be time consuming for cardiac imaging, and has been enhanced by variations; for example, the Look-Locker technique is an efficient method for T1 mapping (51), which includes five well-defined T_Is with multiple segments to allow complete T1 recovery. Its long T_R and long total acquisition time make the classical Look-Locker method sensitive to motion artifacts, especially if the electrocardiogram (ECG) signal is poor, or if there are variations in heart rate (HR). This leads to inconsistencies of k-space acquisition, thus generating blurred images. In the clinic, the single-shot modified Look-Locker imaging (MOLLI) sequence and its variants (52) can acquire images spaced by the R-R interval along the T1 recovery curve that significantly stabilizes the sampling of the T1 recovery curve (53). To manage the requirements of high HR and small heart size in mice, multishot MOLLI can provide reliable measurements of high spatial resolution T1 maps in mice for HRs up to 600 beats/min (54). In addition, a different approach of T1 mapping in the mouse heart using a three-dimensional (3-D) IntraGated (self-gated, discussed in a later section) FLASH sequence with variable flip angles was shown to enable a constant T_R, maintain a steady-state condition, and achieve high SNR by being less sensitive to cardiac motion (55). With the advancements of modern scanners, this method has allowed 3-D T1 mapping of the mouse heart to be obtained with adequate spatial resolution in as little as 20 min using a two-flip angle approach (55).

Clinically, elevated native T1 values can be associated with an increase in tissue water content, such as during the early stages of acute myocardial injury/infarction (53); however, cardiac diseases that involve lipid or iron deposition are associated with low native T1 values (43). In the clinic, native T1 mapping can also be useful when Gd-based contrast agents are not recommended, for example, patients with severe/end-stage renal disease who are not capable of excreting Gd, and for who it has been associated with nephrogenic systemic fibrosis (56, 57)—albeit this risk having been reduced with newer Gd agents. The same is true for mice with renal disease/pathology (58). Importantly, native T1 mapping has also shown potential as an imaging biomarker for diffuse myocardial fibrosis, as a positive correlation was found between native T1 values and collagen volume fraction in mice (Tables 3 and 4). As native T1 varies depending on the vendor, field strength, pulse sequence, and specific pulse sequence parameters, scanner-specific normal ranges are required. Furthermore, sex differences should also be considered; before menopause, women tend to have higher T1 (74).

Table 3.

Studies using CMR to assess fibrosis for the monitoring of disease progression in mice

Pathological Model	MRI Scanner	Imaging Technique for Fibrosis Assessment	Imaging Sequence	Gating/Trigger	Contrast Delivery Route	Assessment Time Point	Fibrosis Type	Major Outcomes from CMR
MI—Permanent occlusion of LAD in adult male C57BL/6 mice Bohl et al. (46)	9.4 T (Varian)	• LGE • Native T1 • Postcontrast T1	Snapshot-IR-FLASH	Prospective trigger (ECG and respiratory double gated)	ip iv	Prior MI Day 1 post-MI	• Local fibrosis	T1 contrast peaked earlier after intravenous than after intraperitoneal, contrast between viable and nonviable myocardium was comparable for both routes. Excellent correlation for infarct size between LGE and histology.
MI—Permanent occlusion of LAD in adult male Swiss mice Coolen et al. (59)	9.4 T (Bruker)	• LGE	First-pass perfusion FLASH	Retrospective trigger	iv	Day 1 post-MI	• Local fibrosis	Regions of decreased perfusion agreed very well with regions of LGE.
MI—I/R: occlusion of LAD for 1 h in adult male C57BL/6 mice Gao et al. (45)	9.4 T (Agilent)	• LGE	T1W SE sequence	Prospective trigger (ECG and respiratory double gated)	iv	24 h post-I/R 4-wk post-I/R	• Local fibrosis	Left ventricle histological cross sections with picrosirius red staining for fibrotic scars, matched with corresponding CMR images, showed a linear correlation (r = 0.96, P < 0.0001).
DMD (Mdx male mice) Stuckey et al. (60)	11.7 T (Magnex Scientific)	• LGE	GRE cine-FLASH	Prospective trigger (ECG and respiratory double gated)	ip	1, 3, 6, 9, and 12 mo of age	• Diffuse interstitial fibrosis	LGE was detected in 3 out of 9 mdx mice at 6 mo of age, all LGE displayed at 12 mo of age but no LGE presented in control mice. LGE showed strong linear correlation with sirius red collagen volume fraction and impaired cardiac function.
MI—Reperfusion after 30-min occlusion in adult Swiss male mice in comparison to healthy mice Coolen et al. (55)	9.4 T (Bruker)	• Native T1 • Postcontrast T1 (using contrast of paramagnetic liposomes)	Steady-state FLASH in combination with a variable flip angle DESPO T1	Retrospective trigger	iv	Healthy mice: 3 different days with 2-day interval MI mice: 24 and 48 h after reperfusion	• Local fibrosis	High-quality, bright-blood T1 maps showed feasibility to detect regional differences between infarcted and healthy myocardium. The mean T1 value of the healthy myocardium of the black blood mode was significantly lower (905 ± 110 ms) compared with the bright blood mode (1,764 ± 172 ms). 3-D T1 mapping in mouse heart was achieved in 20 min.
T1DM (STZ induced, 60 mg/kg, in 8- wk-old male C57BL/6 mice for consecutive 5 days, BG ≥ 13.9 were considered DM) Zhang et al. (61)	7.0 T (Varian)	• Native T1 • Postcontrast T1 • LGE • ECV	T1: GRE Look-Locker IR LGE: multislice GRE IR	Prospective trigger (ECG and respiratory double gated)	ip	From 8 wk and every 4 wk until 24 wk	• Diffuse interstitial fibrosis	No LGE observed in both control and T1DM groups. Significant difference between T1DM vs. control, native T1: 20 wk onward (increase in T1DM), postcontrast T1: 12 wk onward (decrease in T1DM) and ECV: 12 wk onward (increase in T1DM). Linear correlation with the CVF (from picrosirius red): native T1 (r = 0.56) and postcontrast T1 (r = 0.70) show moderate correlation, ECV show strong correlation (r = 0.86).
Hypertension model (l-NAME 3 mg/mL in water for 7 wk in adult male WT mice compared with placebo tap water treated control group TAC (C57BL/6 adult male mice)Coelho-Filho et al. (62)	4.7 T (Bruker) 9.8 T (Bruker)	• ECV	Look-Locker IR	Prospective trigger (ECG and respiratory double gated)	sc	7-wk post-l-NAME 4 and 7 wk after TAC	• Diffuse interstitial fibrosis	Myocardial ECV was significantly higher in l-NAME treated mice (0.42 ± 0.08) than placebo treated (0.25 ± 0.03) and TAC (0.30 ± 0.04) post-7 wk, and showed a direct linear correlation with Masson’s trichrome connective tissue fraction.
T2DM High-fat diet (60% calories from fat) plus STZ injection (100 mg/kg) in adult 8-wk-old male C57/BL6J mice for 1 wk, BG ≥ 13.9 were considered DM Shi et al. (63)	7.0 T (Varian)	• ECV	GRE Look-Locker IR	Prospective trigger (ECG and respiratory double gated)	ip	From 8 wk and every 4 wk until 24 wk	• Diffuse interstitial fibrosis	Myocardial ECV was significantly elevated in the T2DM mice vs. control from week 12. Impaired myocardial strain and function in T2DM from week 12. ECV showed a linear correlation with myocardial strain and LVEF.
Myocarditis induced by systemic inflammation-inbred mouse line both 10- to 12-wk-old male and female mice (C57BL/6J, FVB/NJ and NOD/ShiLtJ) treated with TLR-7 agonist, Resiquimod (100 µg/30 µL per 30 g) Baxan et al. (51)	9.4 T (Bruker)	• T1 • T2 • T2*	T1: GRE Look-Locker IR T2: multiecho SE fat-suppressed T2*: multiecho 2 D-UTE	Prospective trigger (ECG and respiratory double gated)	N/A	2 wk post-treatment 10–12 wk old	• Diffuse interstitial fibrosis	Myocardial T1 and T2* showed a linear correlation with fibrosis score but not in T2. Interstitial iron in the heart of Resiquimod-treated mice, and the strong paramagnetic characteristics of iron led to a decrease in T1 and T2.
T1DM (STZ induced) Bun et al. (64)	11.75 T (Bruker)	• T2	2 multislice short-axis SE acquisitions	Prospective trigger (ECG and respiratory double gated)	N/A	8-wk post-STZ induced diabetes (imaging perform at age of 16 wk)	• Diffuse interstitial fibrosis	Myocardial T2 was significantly lower in diabetic mice (13.8 ± 2.8 ms) than controls (18.9 ± 2.3 ms), and a good correlation between T2 and fibrosis area (Picrosirius red) obtained by histopathology
MI—Reperfusion after 30-min occlusion in adult male C57BL/6J Coolen et al. (65)	9.4 T (Bruker)	• T2 • LGE	T2 prep CPMG-MLEV with SSFP (FISP)	Prospective trigger (ECG and respiratory double gated)	N/A	Day 1, day 3, and day 7 post-MI	• Local fibrosis	T2 in the infarct region was significantly higher than in remote tissue, whereas remote tissue was not significantly different from baseline and T2 value was highest at day 3 postinfarct. Infarct areas based on T2 mapping and LGE showed a strong correlation.
Aging (15 mo old) male CF-1 mice TAC in adult male C57BL/6J mice Lee et al. (66)	7.0 T (Bruker)	• T2	Fast spin eco TurboRARE sequence	Retrospectively trigger IntraGate	N/A	Aging: 15 mo old TAC: 2-wk postsurgery	• Diffuse interstitial fibrosis	The aged mice and the TAC mice showed a marked decrease in T2 (25.3 ± 0.6 in aged vs. 29.9 ± 0.7 ms in young mice; and 24.3 ± 1.7 in TAC vs. 28.7 ± 0.7 ms in shams). An inverse correlation between myocardial fibrosis percentage (picrosirius red) and T2.
MI—Permanent occlusion of LAD TAC Both in adult male C57BL/6 mice Van Nierop et al. (67)	9.4 T (Bruker)	• LGE • T2*	LGE: FLASH T2*: T2*-weighted with UTE and 3-D center-out radial readout Look-Locker	LGE: retrospective trigger T2*: prospective trigger (ECG and respiratory double gated)	iv	MI: 1- to 2-wk postsurgery TAC: 11-wk postsurgery	• Diffuse interstitial fibrosis • Local fibrosis	Region-of-interest analysis in the in vivo post-MI and TAC hearts revealed significant T2* shortening due to fibrosis. Infarct volumes from histology and subtraction images showed a linear correlation. However, in vivo contrast on subtraction images was suboptimal, hampering a straightforward visual assessment of the spatial distribution of the fibrotic tissue.
MI—Permanent occlusion of LAD in adult female C57BL/6 mice Musthafa et al. (68)	9.4 T (Varian)	• T1ρ	Spin-lock pulse with SE sequence	Prospective trigger	N/A	1, 3, 7, and 20 days after surgically induced MI	• Local fibrosis	A trend toward an increase of T1ρ was observed at day 3 post-MI. Increase of T1ρ in the infarcted area was detected 7 days post-MI and remained elevated at day 20 compared with a reference area.

BG, blood glucose; CMR, cardiovascular magnetic resonance imaging; CPMG, Carr Purcell Meiboom Gill sequence; CVF, collagen volume fraction; DESPOT1, driven equilibrium single-pulse observation T1; DMD, Duchenne muscular dystrophy; ECG, electrocardiogram; ECV, extracellular volume fraction; FLASH, fast low angle shot; GRE, gradient echo; I/R, ischemia-reperfusion; ip, intraperitoneal injection; IR, inversion recovery; iv, intravenous injection; LAD, left anterior descending artery; LGE, late gadolinium enhancement; l-NAME, N^ω-nitro-l-arginine methyl ester; LVEF, left ventricular ejection fraction; MI, myocardial infarction; MLEV, M. Leavitt’s Cpd Sequence; sc, subcutaneous; SE, spin echo; STZ, streptozotocin; TAC, transverse aortic constriction; TRL-7, Toll-like receptor 7; T1DM, type 1 diabetic mellitus; T2DM, type 2 diabetic mellitus; TurboRARE, turbo rapid acquisition with relaxation enhancement; UTE, ultrashort echo time; WT, wild type.

Table 4.

CMR assessment of fibrosis post-therapeutic intervention in mice

Pathological Model	Therapeutic Intervention	Imaging Technique for Assessing Fibrosis	Follow-Up Image	Major Outcomes from CMR
TAC In adult C57/BL6 male mice Stuckey et al. (50)	Losartan—angiotensin II receptor blockade Provided in drinking water 1 wk before surgery	• Native T1 • LGE • Postcontrast T1 • ECV	All imaging techniques were performed at 7 days and 28 days post-TAC surgery	• Native T1 value elevated at 7 days post-TAC surgery vs. sham but not at 28 days. • ECV was elevated at 28 days post-TAC vs. sham but not at 7 days. • The ECV fraction was elevated in TAC vs. sham and showed a direct linear correlation with picrosirius red collagen volume fraction. • Treatment with losartan prevented an increase in ECV fraction, indicating that T1 mapping is sensitive to pharmacological prevention of fibrosis. • Study suggests suitability of ECV as an in vivo measure of diffuse fibrosis.
TAC In adult C57/BL6 male mice Glasenapp et al. (69)	Reverse TAC (rTAC) at 3-wk post-TAC	• Native T1	Baseline, 7–8 days, 3- and 6-wk post-TAC surgery, 7–8 days and 3-wk post-rTAC surgery.	• Native T1—elevated at 7–8 days post-TAC compared with baseline and age-matched sham, and remained elevated at 3- and 6-wk post-TAC but did not further elevate from 8 days post-TAC. • Mechanical ventricular unloading (rTAC) shown to attenuate the elevation in T1 vs. sham rTAC, observed at both 7–8 days and 3-wk postsurgical intervention. • T1 values showed a direct linear correlation with fibrotic area (picrosirius red staining), inflammation signal (68Ga-pentixafor PET signal) and cardiac function (ESV and EF).
Hypertensive mice Adult C57/BL6 male mice treated with a continuous angiotensin II (480 ± 34 ng/kg/min) infusion for 6 wk. Kwiecinski et al. (70)	Discontinuation of angiotensin II infusion after 6-wk infusion	• Native T1 • LGE • Postcontrast T1 • ECV	Baseline and every 2 wk during angiotensin II infusion. At 4 wk after discontinuation of angiotensin II infusion	• Native T1, postcontrast T1, and ECV showed linear correlation with histological fibrosis (Picrosirius red). • During pressure loading, ECV showed a progressive increase with a similar pattern of change as the blood pressure and inverse correlation with the systolic function. However, the change in native T1 did not reach significance. • No observation of LGE focal replacement fibrosis. • In response to reverse pressure overload, LV mass and ECV regressed at 4 wk after removal of angiotensin II infusion vs. before infusion removal. However, the reverse remodeling was only partial as none of these measurements returned to the baseline values.
HFpEF Adult WT mice (l-NAME, 3 mg/mL in water-7 wk) Coelho-Filho et al. (71)	Spironolactone subcutaneous pellets implantation (50 mg/kg/day)	• ECV	Baseline and 7 wk after therapy	• Mice treated with both l-NAME and spironolactone showed a lower myocardial ECV vs. treated l-NAME alone and no significant difference when compared with control mice. • The ECV fraction showed direct linear correlation with Masson’s Trichrome connective tissue fraction. • Study shows treatment with spironolactone prevented increased ECV fraction and suggests suitability of ECV as an in vivo measure of diffuse fibrosis.
MI In adult OF1 male mice (Reperfusion after 30-min ligation) Van Den Boomen et al. (72)	UPy-hydrogel loaded IGF1 and VEGF (UPyGF-hydrogel) intramyocardial injection (10 µL) at 2 min before reperfusion	• LGE • T1	LGE: Day 1 post-MI T1: • 1-day pre-MI • Days 3 and 22 post-MI	• LGE: showed similar infarct size for saline (12.4 ± 1.2 %) and UPyGF-hydrogel (11.1 ± 1.1%) treated mice at day 1 post-MI. • T1 values: At day 3, only T1 value of saline injected ischemic myocardium (2.172 ± 0.008 s) was significantly increased compared with the healthy myocardium (1.541 ± 0.030 s). At day 22, saline-injected infarcts T1 value remained significantly increased (2.077 ± 0.075 s) compared with the healthy myocardium. T1 value of UPyGF-hydrogel treated ischemic myocardium did not show significant difference when compared with healthy myocardium at both days 3 and 22.
Hypertensive Adult male C57BL/6J mice (1 kidney/DOCA/salt model) Li et al. (73)	Serelaxin combined (0.5/mg/kg/day) with intravenous BM-MSC (25 μg of BM-MSC proteinper mouse)	• LGE • Native T1 • Postcontrast T1 • ECV	7 day posttreatment	• Native T1 and postcontrast T1 values were not significantly altered in all groups. • 1 K/DOCA/salt-injured mice showed a trend toward having decreased average postcontrast T1 values and a significant increase in ECV. • Combined treatment group showed no significant difference with control in ECV.

BM-MSCs, bone marrow-derived mesenchymal stromal cells; DOCA, deoxycorticosterone acetate; ECV, extracellular volume fraction; EF, ejection fraction; ESV, end systolic volume; HFpEF, heart failure with preserved ejection fraction; IGF1, insulin growth factor 1; LGE, late gadolinium enhancement; l-NAME, N^ω-nitro-l-arginine methyl ester; LV, left ventricle; MI, myocardial infarction; PET, positron emission tomography; TAC, transverse aortic constriction; UPy, ureido-pyrimidinone; VEGF, vascular endothelial growth factor.

Post-Gd contrast T1/ECV mapping.

When compared with native T1 mapping, contrast-enhanced T1 values are more variable and rely on the dose of Gd, acquisition time, injection methods, and clearance by the kidneys (43). These factors cause great variance when applying a single postcontrast T1 map to distinguish healthy and pathological myocardium. However, CMR using varied flip angle postcontrast T1 maps offers a solution to this issue. By adjusting the flip angles, the signal from fibrotic tissue can be enhanced or suppressed to different extents. This allows for improved visualization and assessment of cardiac fibrosis in CMR (55).

In addition, the coefficient of contrast concentration in the myocardium normalized to that of blood can estimate the ECV. By combining changes in the native and post-Gd T1 values of the myocardium and applying a correction factor for the hematocrit, ECV is a reliable marker of myocardial tissue remodeling in clinical settings (43) (Eq. 1). ECV has been proposed as a marker of fibrosis when other pathologies such as myocardial edema or infiltration are absent, as they increase the extracellular space (75). In mice, ECV measurements have shown promise in assessing both focal and diffuse fibrosis for disease progression studies (Table 3), and posttherapeutic intervention assessments (Table 4).

$$\text{ECV}=\left(1-\text{Hct}\right)\frac{\left(\frac{1}{\text{T}1\text{\hspace{0.17em}myo\hspace{0.17em}post}}\hspace{0.17em}-\hspace{0.17em}\frac{1}{\text{T}1\text{\hspace{0.17em}myo\hspace{0.17em}pre}}\right)}{\left(\frac{1}{\text{T}1\text{\hspace{0.17em}blood\hspace{0.17em}post}}\hspace{0.17em}-\hspace{0.17em}\frac{1}{\text{T}1\text{\hspace{0.17em}blood\hspace{0.17em}pre}}\right)}$$ (1)

However, contamination of the myocardial signal by blood pool or epicardial fat pixels straddling the subendocardial or subepicardial border is a technical limitation of all myocardial contrast evaluation, both in humans (76) and mice (63). Thus, although ECV measures are most accurate in the midwall, it is important not to neglect myocardial fibrosis in the subendocardial and subepicardial layers (Fig. 1D).

T2 and T2* mapping.

T2 and T2* mapping have also been investigated as potential tools for the detection of both local and diffuse fibrosis in mice (Table 4). T2-mapping measures the relaxation time of protons in tissue water molecules, which is influenced by tissue water content and collagen deposition. T2*-mapping measures the decay of transverse magnetization, which is influenced by the presence of iron in the tissue, such as with intramyocardial hemorrhage that is common in acute MI (77).

Although clinically, T2 relaxation is most commonly used to determine myocardial edema (78), several studies have investigated the potential of T2 mapping for the assessment of cardiac fibrosis in mice. A study using an ultrahigh field MR scanner (11.75 T) demonstrated that T2-weighted measurements were significantly lower in the hearts of diabetic mice, and showed a strong correlation with cardiac fibrosis (64). T2 mapping could also detect fibrosis in a mouse model with ischemia-reperfusion injury (65). Recently, the potential of applying T2 mapping with a fast SE (TurboRARE) sequence was demonstrated as a tool to distinguish diffuse fibrosis by showing that T2 relaxation time decreased significantly, and correlated with increased fibrosis identified histologically in both aging and pressure overload mouse models with cardiac hypertrophy (66). However, some factors need to be considered when interpreting T2 values, including: T2 values tend to depend on the field strength of the scanner (the higher the field strength, the shorter the T2 relaxation time), and T2-mapping post-MI with low resolution could merge and obscure polar T2 variation within the same voxel including both T2 increasing with myocardial edema and decreasing with myocardial fibrosis.

As described earlier, T2* refers to the decay of the magnetic resonance signal due to inhomogeneities in the magnetic field. T2* can be influenced by many factors, such as local magnetic field gradients, susceptibility differences between tissues, and the presence of metallic implants or air pockets. T2*-weighted images are used in MRI to detect and characterize a range of pathological conditions, including hemorrhage, iron deposition, and calcification. In the clinic, T2* mapping is mainly used for the assessment of cardiac iron overload (41). However, given that T2* relaxation times become shorter in collagen-rich tissue (79), there may be potential for applying T2* mapping for assessing cardiac fibrosis. An ex vivo study provided evidence that T2* shortening was highly correlated with the fibrotic area in the rat heart post-MI by using the ultrashort echo time sequence that enabled the measurement of T2* in the tissue with fast T2* signal decay (80). In addition, another study demonstrated the feasibility of using a T2*-weighted 3-D radial MRI sequence for the assessment of cardiac fibrosis in mice. This study showed that the T2* values were significantly lower in fibrotic hearts than in healthy hearts (67). The major drawback for T2* mapping is the influence of field inhomogeneities that induces a false positive area for fibrosis at the lung-heart interface, this ultimately increases the difficulty of analysis (81).

Overall, T2 and T2* mapping have shown some potential for assessing cardiac fibrosis in mice. Although T2 mapping is more sensitive to edema and inflammation, T2* mapping is particularly sensitive to the presence of intramyocardial iron, as seen in cardiac siderosis and acute MI. Combining these techniques with others, such as LGE, T1, and ECV mapping, may provide comprehensive myocardial tissue phenotyping in mice.

ADVANCED CMR TECHNIQUES FOR CARDIAC FIBROSIS ASSESSMENT

Other advanced CMR techniques such as diffusion-weighted imaging (DWI) and T1 rho (T1ρ) have also been used to evaluate fibrosis in the mouse heart. However, these advanced animal imaging sequences may not be readily available in all preclinical systems and will differ in their implementation.

Diffusion-Weighted Imaging

Diffuse fibrosis typically coincides with myocyte hypertrophy in response to a pathological stimulus, for example, in response to chronic reduced oxygen delivery and/or increased pressure load (7). Diffusion-weighted imaging (DWI) allows the noninvasive tracking of myocardial fiber orientation and the quantification of myocardial fiber disarray induced by cardiac diseases (82). Traditionally, DWI is performed by using SE echo planar imaging (EPI) sequences with multiple diffusion directions. In the clinic, ECG and respiration gating/breath holds are used to reduce cardiac and respiratory motion effects. Normally, more averages and relatively low diffusion weighting (B values) can be used to increase SNR and to reduce the artifacts from heart and lung motions after using ECG and respiration triggers. Although rapid acquisition of the EPI sequence is possible, the high HR in mice can still reduce the quality of DWI. Hence, DWI is mainly applied in ex vivo studies in rodents (83). However, methodical preparation and the susceptibility changes caused by the tissue-fluid interfaces can greatly affect the image quality in ex vivo DWI (84, 85). Ex vivo DWI imaging requires careful preparation of the tissue to minimize changes that can affect its structure and other properties, such as tissue shrinkage caused by fixation. Tissue immersion in a solution during imaging can create susceptibility changes that cause imaging artifacts, so optimization of imaging parameters and sequences may be necessary to reduce distortions and improve accuracy (84, 85). A study demonstrated the potential of using the tractographic propagation angle, a metric of myofiber curvature (degrees/unit distance) derived from DWI to determine fibrosis scar post-MI in both humans and mice (86). Moreover, the helix angle that is calculated based on DWI data provides an opportunity of quantifying diffuse fibrosis in pressure overload rodent models (83, 87).

T1-Rho

T1-rho (T1ρ) uses a continuous low-power radiofrequency pulse (spin-lock pulse) parallel to the magnetization in the transverse plane to prevent and decrease the loss of transverse magnetization (T2 decay: magnetic field inhomogeneity) and further enable the measurement of longitudinal relaxation time in the rotating frame (88). T1ρ as an endogenous advanced MR contrast technique has shown the ability to detect myocardial focal and diffuse fibrosis in patients and large animals (89–91). This technique was used in a mouse study of acute MI and provided evidence that T1ρ maps can be used to determine cardiac fibrosis accurately (68). The sequences used to generate this high-resolution T1ρ mapping ensured high SNR with only four different T1ρ-weighted images but was time consuming as it took ∼20 min of scanning time for a single slice of T1ρ (68). More recent work introduced a Bloch simulation-optimized imaging sequence using high flip angles and a radial view-sharing method that ensured high SNR and efficient data sampling for generating faster (<20 min) T1ρ maps with higher reproducibility in healthy mice (92).

EXPERIMENTAL AND TECHNICAL CONSIDERATIONS FOR CMR

Anesthetic Agents

Movement is one of the major difficulties while performing CMR. In the clinical situation, patients can be instructed to lie still on the scanning bed and can hold their breath to control respiratory movement. In an animal study, body movement can be limited by using an anesthetic agent that can also slow the breathing rate. For animal CMR imaging, the anesthetic agents used should allow for easy administration and maintenance (injection or inhalation) with MR-safe instruments and equipment, and provide adequate and reproducible immobilization, preferably without changes in cardiac function and HR. The choice of anesthetic agent should be carefully considered, maintained consistently throughout a single study, and clearly reported since different agents can impact cardiovascular physiology and respiratory function in different ways (93).

The major advantage of injectable anesthetic agents is their convenient single administration. In contrast to volatile anesthetics, injectable agents may be more achievable for smaller/underresourced laboratories since they do not require expensive equipment such as vaporizers and extended ventilation circuits for administration from beyond the 5 Gauss line. Ketamine has been commonly used in animal research because of its large safety threshold (94). The two major disadvantages of injectable agents in CMR imaging are their limited anesthetic duration and lack of adjustability during scanning, an issue particularly for longer scanning sequences/protocols. Therefore, the preferred approach is isoflurane inhalation for animal imaging protocols for long duration protocols (95). Major advantages of isoflurane include its high tolerance in mouse models of disease, rapid induction, fast recovery times, and convenient adjustability during scanning in response to changes in HR; key components in achieving reproducible CMR assessments of cardiac function (96).

Triggering/Gating

CMR is more challenging than MRI of other organs since cardiac and respiratory motion can cause artifacts in reconstructed images. Performing CMR in mice is particularly challenging compared with humans, largely because of the small heart size (adult: ∼90–150 mg), and high HR (conscious mice: 500–600 beats/min, under 2.5% isoflurane: ∼400–500 beats/min) and respiratory rate (conscious mice: ∼100–230 breaths/min, under 2.5% isoflurane: 60–70 breaths/min, without breath holding as an option; 3). Hence, cardiac and respiratory gating have been developed to reduce artifacts caused by motion and increase image resolution. ECG signals are commonly used to gate CMR acquisitions at specific time points to achieve cardiac synchronization both in the clinic and in mice.

There are two types of triggering/gating: prospective and retrospective. CMR acquisitions with prospective gating, that is, using ECG electrode signals, are set to perform MRI and collect data at a specific phase within a heartbeat/breath or multiple consecutive cardiac phases in the cardiac cycle between breaths. Prospective gating captures images in real time with a triggering mechanism, allowing for accurate synchronization of image acquisition with cardiac motion. However, this process increases the scan time as the gating window restricts the start of the MRI acquisition to a specific time interval, typically during a specific phase of the cardiac cycle, to minimize motion artifacts and obtain clearer images of the heart. Critically, for prospective gating to be practicable, a good quality ECG trace must be achieved in each mouse, which can be compromised by electromagnetic interference (including the gradients), and is particularly challenging in models of disease where the ECG waveform is characteristically perturbed such as ST-elevation with MI.

CMR with retrospective gating (also called self-gating) acquires data continuously throughout the cardiac cycle. Segmental data are then reordered and grouped within specific cardiac phases based on the ECG signal and respiratory rate recorded simultaneously with CMR data (96). The approach was originally developed for human applications, based on acquiring extra echoes or projection images incorporated with imaging readouts. Over the last few years, four primary strategies have been developed and adapted for animal cardiac imaging at high magnetic field. These strategies mainly rely on the acquisition and processing of 1) an additional MR signal (navigator based), 2) obtaining images with high temporal resolution (image based), 3) directly extracting the motion synchronization signal from the image data (k-space based), or 4) nonuniform gating; these gating strategies have been previously discussed in depth (97). The major advantage of retrospective gating is that continuous data acquisition with the flexible reconstruction of a variety of cardiac phases and temporal resolutions can be achieved. Nevertheless, small animals with high respiratory frequency and fast heartbeat usually require sophisticated data processing and signal filtering to extract synchronization signals for retrospective gating.

Pre-CMR Scans and Imaging Planes

The localizer scan is the initial step that provides an overview of the mouse thorax and upper abdomen in axial, sagittal, and coronal views (Fig. 4A). It helps in positioning and aligning the mouse within the scanner, ensuring the FOV is in the isocenter of the magnet for accurate and consistent image acquisition. This scan allows the operator to locate the heart and identify the desired imaging planes for subsequent scans. Following the localizer scan, the planning scan is performed to obtain detailed images of the mouse heart in specific orientations/views (Fig. 4B). The planning scan typically includes the vertical long-axis or two-chamber view (2 Ch) and the horizontal long-axis or four-chamber view (4 Ch) (32). These views help define the long axis of the left ventricle (LV) and serve as a reference for planning the position and orientation of the subsequent short-axis imaging stack. It ensures that the desired regions of interest, such as the ventricles or specific cardiac structures, are adequately captured in the imaging plane. By performing the localizer scan and planning scans, the operator positions the mouse heart optimally for subsequent CMR scans with imaging parameters based on the research question (32). These initial steps are crucial for acquiring high-quality and reproducible images of the mouse heart.

Figure 4.

Schematic representation of cardiovascular magnetic resonance (CMR) scans. Visualizing mouse hearts in healthy and pathological conditions. A: localizer scans illustrating the mouse thorax and upper abdomen in axial, sagittal, and coronal views. B: planning scan images of a representative mouse heart in the vertical long-axis or 2-chamber view (2 Ch) and horizontal long-axis or 4-chamber (4 Ch) view. C: comparison of the short-axis (SAX) and 4 Ch views of a schematic of a healthy mouse heart and a DCM fibrotic mouse heart with pathological remodeling. D: schematic of CMR with varied flip angle (FA): 2°, 5°, 8°, 11°, and 14° postgadolinium contrast (left) and derived T1 maps (right) from both a healthy heart and DCM fibrotic heart. Flip angles can affect the image contrast and visibility of fibrotic regions, as different flip angles may enhance or suppress the signal from fibrotic tissue to varying degrees. Therefore, the choice of flip angles can be adjusted to optimize the visualization and assessment of cardiac fibrosis in magnetic resonance imaging (MRI). For postgadolinium FA of 11° and 14° best highlight the differences in contrast between the healthy and fibrotic heart. Postcontrast T1 maps provide better differentiation than postcontrast T1-weighted images when distinguishing between a fibrotic mouse heart and a healthy control heart, as depicted in the T1 maps on the right. Schematics were generated based on information from Ref. 55.

Different cardiac planes can be acquired depending on the target cardiac structure and chambers to obtain the maximal coverage when performing CMR. The standard cardiac planes are short-axis, four-chamber (horizontal long axis), and two-chamber (vertical long axis) views (Fig. 4C). Ventricular-mapping measurements in mice are commonly performed in the short-axis orientation as it is more reproducible (3).

Clinically, atrial assessments generally use long-axis views (2 Ch and 4 Ch) (98, 99). However, the atrium is a thin-walled structure, and tissue characterization with mapping techniques is difficult; to date, we are unaware of studies of these atrial measurements in mice.

Delivery Routes for Contrast-Enhanced CMR

As described earlier, contrast-enhanced MRI (e.g., with Gd) is a common technique used to better distinguish tissues/organs within the body. Contrast is typically delivered to mice via intravenous injection. Although intravenous injection is considered the optimal delivery route as it is more clinically relevant, performing intravenous cannulation in the tail vein of mice can be a challenging technique to master, and its success is unable to be visually confirmed (since the cannula is typically inside the scanner bore during infusion). Other delivery routes for contrast agents have been explored, for example, intraperitoneal injection, subcutaneous (sc) injection, and retro-orbital injection (46, 100, 101). Bohl et al. (46) compared the myocardial T1 values in both healthy mice and MI mice after an intraperitoneal or intravenous injection of Gd-based contrast. They showed that myocardial T1 values with intravenous injection dropped faster (in the first minute) and were lower than with intraperitoneal injection in healthy mice (46). Both injection routes provided comparable results and were relatively stable for up to 30–60 min (46). Similar observations were made in infarcted mice, with maximum T1 contrasts (difference between remote and infarcted myocardium) reached earlier and at greater values in the intravenously injected (∼13.5–22 min) compared with the intraperitoneally injected MI mice (∼30–40 min) (46). Both delivery techniques showed clear enhancement of infarcted myocardium and were stable for at least 60 min. This makes intraperitoneal injection a popular alternative route for contrast delivery due to its ease of administration (46), despite its impact on scan time. Delivery of contrast by intraperitoneal injection typically has a higher success rate and is more reproducible between mice and groups compared with intravenous injection, though this can also fail if the contrast is inadvertently injected into the gut. Other delivery routes for contrast agents via subcutaneous or retro-orbital injections show comparable results with intravenous injection (100, 101). The major drawback of subcutaneous injection is the time taken to reach peak values (43–45 min for subcutaneous injection compared with 4–5 min for intravenous injection in healthy myocardium) (100); the ratio of contrast to peak contrast/plateau was similar between retro-orbital and intravenous injection (101). CMR with variable flip angle T1-mapping postgadolinium can be used to aid in the assessment of diffuse cardiac fibrosis (Fig. 4D).

Bright Blood and Dark Blood

Bright/white blood imaging is performed using the high signal intensity of fast-flowing blood and is conventionally used in the clinic with GRE sequences or steady-state free precession MRI (SSFP, a type of GRE sequence). As blood moving into the slice would not have received any previous stimulation, the full volume can be magnetized, and much higher signals can be extracted from the blood. The main advantage of bright blood imaging is the fast acquisition and better visualization of the myocardium even with low SNR because of strong signals produced by the flow enhancement effect. However, the major disadvantage of bright blood techniques is that the blood pool signal can mimic scar tissue and this may lead to false positive observations using conventional bright blood LGE, which can potentially lead to an overestimation of the infarct region. Bright blood techniques are therefore not advisable for the assessment of infarcted myocardium with LGE (Fig. 5A).

Figure 5.

Schematic of white blood vs. dark blood cardiac magnetic resonance imaging in different cardiac planes in the heart with myocardial infarction. Left to right: short-axis, vertical long axis, and horizontal long axis. A: white blood cardiac magnetic resonance imaging (CMR), the blood pool signal can mimic scar tissue and lead to false positive observations using conventional bright blood late gadolinium enhanced (LGE) CMR, which can potentially lead to an overestimation of the infarct region in mice with myocardial infarction. B: for black blood sequences, flowing blood in vessels or the heart is nulled (black). Thereby improving the contrast between the cardiac tissue and blood pool and further benefiting tissue characterization of the myocardium while determining fibrosis with LGE. The red arrows and red dashed circles represent the potential infarcted region. Schematics were generated based on information from Ref. 37.

In black/dark blood imaging, sequences are designed to null the signal from flowing blood and highlight in-plane tissue, this is useful as it improves the delineation of the myocardium and reveals pathological signal changes that may be confounded by the bright blood signal. Blood flowing in coronary vessels and the heart chambers is nulled and appears dark. Double or triple IR can be used to further null signals from the blood for black blood imaging, thereby improving the contrast between the cardiac tissues and the blood pool, which further benefits tissue characterization of the myocardium (Fig. 5B). The major artifact in dark blood imaging is the residual signal caused by insufficient blood suppression. However, even if blood is adequately suppressed, the short T2 of the cardiac tissue results in limited time for data collection before the signal decays, whereas the relatively long T1 imposes SNR penalties on short repetition times, resulting in a trade-off between image quality and acquisition time. The optimization of black blood sequences to suit the goals of each study is crucial for advanced tissue characterization in mice (including assessment of fibrosis) because of the susceptibility of black blood techniques to artifacts (motion and flow) and the need to trade off blood suppression for other imaging parameters (SNR, spatial coverage, and T1- or T2-weighting, total scan time). Overall, when designing experiments involving black-blood imaging techniques, it is important to consider the downstream analysis workflow and potential challenges associated with the limited availability of automated image analysis of black-blood images and the possible need for manual segmentation or specialized analysis software.

Acceleration with Reconstruction Techniques

Constructing a single MR image commonly involves collecting a series of frames of data, called acquisitions. In each acquisition, an RF excitation produces new transverse magnetization, which then collects data along a particular trajectory in k-space. Conventionally the k-space sampling pattern is designed to meet the Nyquist criterion, which depends on the resolution and FOV and requires as many k-space samples as voxels to be reconstructed (102). Fast acquisition is needed for high-temporal-resolution imaging. Scans can be accelerated by applying reconstruction techniques that undersample the k-space data, for example, compressed sensing that is based on the subsampling of the k-space rather than an entire k-space grid (102, 103). However, some major technical issues associated with compressed sensing reconstruction for CMR remain, including high complexity of the algorithms and long reconstruction times, image degradation at high accelerations, and the need for parameter tuning. Parallel imaging is another method used to accelerate imaging by using multiple receiver coils to reduce the number of acquisitions required for MR imaging reconstruction. SENSitive Encoding (SENSE) and GeneRalized Auto-calibrating Partially Parallel Acquisitions (GRAPPA) are two common algorithms used with parallel imaging to resolve the problem of aliasing artifacts as a result of k-space undersampling with signals from different receiver coils. These reconstruction techniques reduce acquisition time significantly by acquiring less k-space data and estimating the nonacquired data on the basis of prior information about the images. Fast acquisitions can minimize movement effects and generate effective and comprehensive images of the heart.

USING CMR FOR SERIAL ASSESSMENTS OF CARDIAC FIBROSIS: DISEASE PROGRESSION AND INTERVENTION STUDIES

Historically, preclinical research relied heavily on histological assessment of fibrosis, a terminally invasive outcome measure requiring independent cohorts of mice to characterize tissue at each relevant time point throughout disease progression or following interventions (3). Echocardiography and CMR have both been used for the serial assessment of cardiac function in mice, each with advantages and disadvantages (Table 5). Although an indirect assessment of cardiac fibrosis can be inferred by echocardiography, for example, using measures of strain (105, 106), improvements in CMR techniques in mice over the last decade make CMR the noninvasive technique of choice to assess cardiac fibrosis over time. However, there are also important considerations for assessing cardiac fibrosis using CMR in mice. A key factor is having validation data to support the accuracy of CMR for detecting cardiac fibrosis. A brief summary or pros and cons is provided in Table 6. CMR with LGE can be used to assess focal fibrosis in the diseased heart. Of note, the contrast agent does not bind to fibrosis itself; the signal enhancement is due to alterations in the volume of distribution as well as wash-in and wash-out kinetics of the contrast agent into the interstitial space or ECM. In other words, a threshold of accumulated myocardial cell death is required for the LGE-CMR technique to detect fibrosis, as well as a reference normal segment. Accordingly, studies that aim to detect significant serial changes of focal fibrosis mass throughout disease progression could require prohibitively long follow-up (107) with the additional risk of underestimating pathology progression at an early stage with diffuse fibrosis (2, 108). The development of LGE-like tracers for fibrosis may improve the utility of LGE CMR imaging in the quantification of interstitial fibrosis, particularly in studies of novel treatments for fibrosis prevention/reversal. However, selective tracers have an intrinsic limitation in that the generated signal is expected to be lower compared with LGE imaging, primarily because of reduced retention of nonselective contrast (108).

Table 5.

Comparison of the practicalities to apply echocardiography and CMR in mice for the assessment of cardiac function and fibrosis

	Echocardiography	CMR
Imaging resolution	• Specialized small animal echocardiography systems have very high temporal resolution, and adequate spatial resolution for most study requirements. Excellent for assessing cardiac function but inferior for assessing cardiac fibrosis in comparison to CMR.	• Specialized small animal CMR systems have lower temporal resolution, and exceptional spatial resolution, with additional capabilities including advanced tissue characterization, e.g., fibrosis.
Cost	• Lower	• Higher
Availability	• In small animal research, echocardiography is a relatively high-throughput, widely accessible, safe, and noninvasive imaging technique.	• CMR imaging is a relatively low-throughput, noninvasive imaging technique, becoming available to a growing number of well-resourced laboratories.
Training and expertise	• Acquisition and analysis of small animal echocardiography is practical, achievable, and relatively high throughput, with standardized training available to maintain essential rigor and reproducibility.	• Acquisition of CMR imaging, particularly with less well-characterized techniques, requires specialized expertise. Similarly, CMR analysis requires at least moderate understanding of MR physics, generally requiring multidisciplinary teams (biologists/physiologists and physicists/radiologists).
Time	• Due to its low cost and rapid acquisition time, echocardiography is suitable for imaging of all mice within each cohort of a given study (∼15–50 scans/day), depending on the parameters required.	• Due to its longer setup/scan duration, it is typically only suitable for scanning small study cohorts (∼4–10 scans/day), depending on the parameters/protocols/sequences required. Sample size calculations should be performed for the parameter of interest (104).

CMR, cardiac magnetic resonance; MR, magnetic resonance.

Table 6.

Pros and Cons of using CMR for assessing myocardial fibrosis in mice

Pros	Cons
• Noninvasive: CMR is a noninvasive imaging technique. It doesn’t require complex surgical procedures or tissue sampling, and therefore, is not a terminal procedure.	• Indirect measure of fibrosis: CMR provides an indirect measure of cardiac fibrosis, and not all methods/sequences have been validated against cardiac biopsy samples with direct measures of collagen fibers. Depending on the imaging methods and mouse models used, this may lead to an over or under representation of cardiac fibrosis. CMR results should be validated with histological analysis incorporating stains for collagen including Masson's trichrome or Picrosirius red. Molecular assessment of collagen gene and protein abundance by qPCR or Western blotting can also be used as a complementary approach.
• Longitudinal Studies: Since CMR is noninvasive, it allows for longitudinal studies, enabling researchers to monitor changes in fibrosis over time within the same animal. This is critical for understanding disease progression with or without interventions or treatments.	• Cost and Infrastructure: CMR is a relatively expensive imaging modality, and it requires specialized equipment and facilities. This can be a limiting factor for some research groups.
• Multiparameter Imaging: Multiple imaging sequences or modalities can be used to obtain comprehensive data on cardiac structure and function, in addition to cardiac fibrosis in the same mice.	• Limitations detecting diffuse fibrosis: CMR techniques require further optimization to accurately and reproducibly detect diffuse fibrosis in mice. This limits the ability to detect subtle changes in diffuse fibrosis with interventions or treatments.

CMR, cardiac magnetic resonance; qPCR, quantitative polymerase chain reaction.

The ability to detect subclinical diffuse fibrosis in the early stages of cardiac disease may be important to prevent chronic, irreversible HF. It would also provide the opportunity to monitor the impact of new therapies on fibrosis (9). Optimizing CMR techniques to assess diffuse fibrosis is a work in progress and includes improvements in accuracy, reduction of acquisition time, while ensuring reproducibility. T1 mapping (native or with Gd-based contrast) and ECV have been used more commonly to follow disease progression and evaluate the efficacy of several antifibrotic therapies both in humans and mice, than the more recently explored techniques such as T2/T2* mapping, T1ρ, and DWI (Tables 3 and 4).

Both T1-mapping and ECV measurements have been associated with cardiovascular outcomes and have outperformed LV ejection fraction and other standard parameters in the risk stratification of patients with HF (109–111). ECV also increases with age and shows a strong correlation with the extent of diffuse myocardial fibrosis in mice (Tables 3 and 4). Combined, these support the feasibility of using CMR measurement as biomarkers for cardiac diffuse fibrosis during disease progression in both humans and mice. A comprehensive study by Stuckey et al. (50) examined myocardial fibrosis with LGE, T1 mapping (native and with Gd contrast), and ECV during disease progression and therapeutic intervention in mice with pressure overload (transverse aortic constriction, TAC). They found that native T1 was elevated at 7 days post-TAC but ECV was not. At 28 days post-TAC, native T1 was no longer elevated but ECV was increased (50). This highlights the importance of applying the optimal CMR technique at different disease stages as each technique in CMR has strengths and weaknesses. Alternatively, combining CMR techniques can also be important for a comprehensive assessment of fibrosis in the heart, as different techniques may provide complementary information that can enhance diagnostic accuracy and guide appropriate treatment. The CMR technique applied should be based on the disease type and the estimated outcomes post-therapeutic intervention when designing the study to assess cardiac fibrosis as a diagnostic and prognostic tool in disease progression and therapeutic application follow-up. With continual improvements in conventional CMR and further developments of the advanced CMR techniques in mice such as DTI and T1ρ (Tables 3 and 4), more possibilities to accurately assess cardiac fibrosis in response to therapies will become available.

REMAINING CHALLENGES AND FUTURE OPPORTUNITIES IN DETECTING CARDIAC FIBROSIS WITH CMR IN MICE

CMR techniques in small animals have improved significantly in the last decade, and CMR has become a reliable noninvasive imaging technique for the assessment of cardiac function and tissue characterization in preclinical research. Although mapping of various relaxation parameters in the examination of both focal and diffuse fibrosis in mice has been investigated, the development of these sequences can be technically challenging and requires well-trained operators, especially to ensure accuracy and reproducibility. In addition, advanced tissue characterization by CMR for early diagnosis of diffuse fibrosis (e.g., the assessment of cardiac metabolism or the structural organization of the myocardial fibers) is still limited by long scan times and low sensitivities. Further technical development is required to achieve faster imaging sequences with appropriate sensitivity to contrast agents; this will enable consistent routine quantification of cardiac fibrosis in mice. As noted earlier, validation data to support the accuracy of new CMR techniques will be critical.

Clinically, locating and quantifying atrial fibrosis and structural remodeling may become a valuable tool for the treatment stratification of patients with atrial fibrillation, as well as for preclinical studies. However, there remain challenges associated with the detection of fibrosis in the atria of humans and mice. The earlier-described difficulties inherent to CMR assessment of fibrosis in the mouse ventricle are made more challenging in atria because of their small size (weight of 3–10 mg), irregular structure, and location. Furthermore, it is possible that adequate imaging of the mouse atrial wall exceeds the spatial resolution of currently available machines. Therefore, further development of CMR techniques specific to atrial fibrosis requires optimization of not only the imaging sequences but also a consideration for appropriate scan orientation and analysis.

CONCLUSIONS

The assessment of cardiac fibrosis in preclinical studies is essential to better understand the mechanisms underlying cardiac fibrosis and to accurately assess the therapeutic potential of new antifibrotic agents over time. In the past, studies have largely assessed the ability of drugs to prevent cardiac fibrosis but have been limited by a lack of noninvasive modalities to investigate potential effectors of fibrosis reversal. CMR is a powerful imaging tool with the ability to serially examine cardiac fibrosis in vivo in mice. Following significant improvements in CMR techniques in small animals, protocols for high-fidelity quantification of focal fibrosis in mice have been well described, with LGE prevailing as the current reference standard. Although various MR techniques have been developed to detect diffuse cardiac fibrosis in mice with encouraging results, reproducibility remains challenging. With a growing number of researchers dedicating their efforts to refining CMR techniques to quantify cardiac fibrosis with increasing reproducibility and sensitivity, the ability to detect and measure diffuse fibrosis for the more precise evaluation of novel therapies is a realistic possibility in the near future.

GRANTS

G. Z. is supported by an Australia National Imaging Facility fellowship. D.G.D. is supported by a Shine On Foundation fellowship. D.K.W. is supported by National Health and Medical Research Council (NHMRC) Investigator Grant 1174040. T.H.M. is supported by NHMRC of Australia Investigator Grant 2008129. J.R.M. was supported by the NHMRC Grant 1078985, Baker Fellowship (Baker Heart and Diabetes Institute, Australia), and National Heart Foundation Vanguard Grant 105720.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Y.C.C. and J.R.M. prepared figures; Y.C.C., G.Z., D.G.D., D.K.W., J.P.G., T.H.M., and J.R.M. edited and revised manuscript; Y.C.C., G.Z., D.G.D., D.K.W., J.P.G., T.H.M., and J.R.M. approved final version of manuscript.