Current Cardiovascular Imaging Reports

, Volume 5, Issue 2, pp 77–82

Usefulness of Cardiac Magnetic Resonance in Early Assessment of Cardiomyopathies: Myocardial Fibrosis Is a Common Denominator

Authors

  • Ana Pastor
    • King’s College London, Department of Cardiovascular Imaging, Division of Imaging Sciences and Biomedical EngineeringThe Rayne Institute
  • Tobias Voigt
    • King’s College London, Department of Cardiovascular Imaging, Division of Imaging Sciences and Biomedical EngineeringThe Rayne Institute
    • Philips Research, Clinical Research Europe
  • Tobias Schaeffter
    • King’s College London, Department of Cardiovascular Imaging, Division of Imaging Sciences and Biomedical EngineeringThe Rayne Institute
  • Eike Nagel
    • King’s College London, Department of Cardiovascular Imaging, Division of Imaging Sciences and Biomedical EngineeringThe Rayne Institute
    • King’s College London, Department of Cardiovascular Imaging, Division of Imaging Sciences and Biomedical EngineeringThe Rayne Institute
Cardiac Magnetic Resonance (E Nagel, Section Editor)

DOI: 10.1007/s12410-012-9125-9

Cite this article as:
Pastor, A., Voigt, T., Schaeffter, T. et al. Curr Cardiovasc Imaging Rep (2012) 5: 77. doi:10.1007/s12410-012-9125-9

Abstract

Myocardial fibrosis is a common denominator in a wide spectrum of cardiomyopathies; it plays a major role in the pathophysiology of structural remodelling and as a predictor of adverse outcome. Quantification of myocardial fibrosis by invasive biopsy is limited in clinical practice due to the commonly scattered tissue distribution and procedural risks. Late gadolinium enhancement by cardiac magnetic resonance has revolutionized the assessment of ischemic cardiomyopathy by visualization of regional scarring after myocardial infarction. The binary (white-black) principle of contrast development and a requirement for well separated layers of healthy and diseased myocardium is straightforward for clinical decision making in ischemic heart disease, but renders this technique less powerful in conditions where the myocardium is affected diffusely. Recently emerged T1 mapping techniques allow for an individualized quantification of global and regional myocardial signal and are promising tools to separate between healthy and diseased. In this article we review the emerging evidence for T1 mapping techniques and outline the necessary future directions for a successful translational pathway.

Keywords

Cardiac magnetic resonanceTissue characterizationLate gadolinium enhancementT1 mappingDiffuse fibrosis

Introduction

Cardiovascular magnetic resonance (CMR) allows for a comprehensive assessment of myocardial anatomy and function with high levels of accuracy and reproducibility. Myocardial fibrosis is a fundamental process in the development of myocardial dysfunction in various cardiomyopathies [14]. As CMR may uniquely characterize the extent and pattern of myocardial fibrosis in various cardiomyopathies, CMR is increasingly applied as the routine clinical first-line investigation into the cause and follow-up of cardiomyopathies [5•]. One of the most successful examples of tissue characterization is late gadolinium enhancement (LGE), which revolutionized the assessment of ischemic cardiomyopathy by visualization of regional scarring after myocardial infarction [6]. In this technique, visualization of fibrosis by CMR relies on a greater distribution volume and slower washout of gadolinium contrast agents from infarcted tissue and a relative difference in signal recovery between the normal and infarcted myocardium [7, 8]. The use of relative signal intensities and apparent degree of enhancement is dependent on the timing (and the tissue concentration of accumulated contrast agent within the scar) and efficacy of “the nulling” [9], which requires well-separated layers of healthy myocardium from the diseased, underscoring the binary “white-black” principle of LGE. Unlike the regional ischemic fibrosis, the processes, which diffusely affect the myocardium, commonly leave insufficient amounts of normal myocardium to permit optimal nulling and limit differentiation of normal from diffusely scarred areas. Recently, several studies explored novel T1 mapping techniques, as potentially useful in characterization of the non-ischemic, diffuse cardiomyopathic processes. In this review we focus on available evidence and potential applications and streamline the future directions for a successful translational pathway.

T1 Mapping in a Nutshell

Quantification of T1 signal relaxation describes the recovery of longitudinal magnetization [10, 11]. For this, signal intensity per se is unsuitable as it is expressed on an arbitrary scale that differs from one imaging examination to another. On the contrary, T1 mapping permits signal quantification (in milliseconds) on a standardized scale. Alternatively, and probably more exact in describing the amount of contrast agent in a given tissue, relaxivity (R = 1/T1) can be used. In T1 mapping imaging acquisition, a preparation-pulse is first delivered upon which the signal recovery is sampled during multiple measurements of varying T1 weighting (obtained by varying prepulse delays) in order to calculate T1 values. The signal relaxation time is calculated for every pixel by the combination of all acquisitions in a parametric image referred to as a T1 map. Since the T1 map incorporates a continuum of signal intensities throughout the whole myocardium and thus allows for quantification as well as detection of diffuse changes.

A number of approaches have been used to measure the myocardial T1 values. They can be based as sets of separate acquisitions or integrated into a single sequence and utilize saturation or inversion recovery prepulses with different prepulse delays [1215, 16•, 17•], [18]. Early approaches suffered from low temporal and spatial resolution, which limited their clinical applicability. The more successful methodologies followed the principle of the traditional single-shot steady-state free precession Look-Locker sequence which employs an inversion prepulse and subsequent sampling of T1 recovery (inversion recovery). The modified Look-Locker sequence (MOLLI) allows T1 mapping with high spatial resolution [15]. In contrast to Look-Locker where sampling relies on increasing trigger delay with images acquired in the different parts of the cardiac cycle, MOLLI is distinguished by a selective acquisition at a fixed delay time of the cardiac cycle in late diastole (Fig. 1). MOLLI, however, requires 17 heart beats to obtain a single slice map, based on 11 image-phase acquisitions, which leads to a relatively long breath-hold between 15 and 20 ms, prone to undesired breathing motion and errors in the pixel-wise estimation of the T1 map. A recently introduced, shorter 7-phase alternative called “short MOLLI” (ShMOLLI) showed good levels of agreement for post-contrast T1 values and a significant reduction of breath-hold time (mean 9 s) as compared to MOLLI [17•]. It is uncertain whether this shorter version potentially leads to undersampling of the T1 recovery in regions of longer T1 values and whether it affects the values of tissues with long T1, such as the myocardium (in the range between 900 and 1,100 msec), myocardial scar (1,200–1,400 msec), and the blood pool (in the range of 1,800–2,000 msec) in native T1 imaging. ShMOLLI appears robust for postcontrast T1 values and in line with this, shows smaller variation for post-contrast T1 values in comparison to MOLLI. As increasing heart rate reduces time for longitudinal relaxation to occur, all T1 imaging techniques suffer with some heart rate dependency. In systematic comparisons in phantoms, shMOLLI again shows the least heart rate dependency for acquisition of short T1 values [17•], whereas acquisition of longer T1 values at heart rates above 80 bpm significantly diverge and require heart rate correction [15, 17•].
https://static-content.springer.com/image/art%3A10.1007%2Fs12410-012-9125-9/MediaObjects/12410_2012_9125_Fig1_HTML.gif
Fig. 1

Representative images of T1 magnetization recovery curve and 11 sampling phases from a MOLLI measurement of native T1 values

T1 maps are usually reconstructed using nonlinear curve fitting to an inversion recovery signal model
$$ SI = A - B{e^{{ - t/T_1^{*}}}} $$
(1)
where SI is signal intensity, and A and B are coefficients containing non-T1* related effects influencing the signal. T1* is the apparent relaxation time taking into account signal loss due to image acquisition. Relaxation time T1 can be determined according to
$$ {T_1} = T_1^{*}\left( {B/A - 1} \right) $$
(2)
and Relaxivity (R) as
$$ R = 1/T1 $$
(3)

Curve fitting to Eq. 1 can be performed on a pixel-by-pixel basis. Pixel-by-pixel matching of myocardial maps is shown to be very sensitive to respiratory motion artefacts and heart rate variability [15, 17•, 19, 20]. This can especially affect pixels on the border of two different tissue types as motion between source images, ie, different inversion times, can contaminate the magnetization recovery curve and derived values. Delineation of scar in the T1 map may be challenging in the presence of breathing artifacts, emphasizing the need for robust co-registration of the source phases in postprocessing tools [19, 20]. However, for diffuse fibrosis this may pose less of a problem.

Myocardial T1 Mapping in Clinical Setting

Several studies suggest that T1 mapping could support the characterization of myocardial tissue on a global or regional level by direct myocardial signal quantification. As myocardial fibrosis is a common denominator of pathologic remodelling and characterized by expansion of the extracellular matrix and accumulation of interstitial collagen, a quantifiable readout of myocardial remodelling could facilitate the follow-up on clinical progression and guide treatment [14, 5•]. A summary of reported T1 values for healthy myocardium by previous studies is shown in Table 1. Native T1 values in myocardium are in the range of 900–1,100 msec and are significantly shortened by the effect of gadolinium agents. In addition, studies have shown that in acute or chronic myocardial infarction the infarcted area can be separated from healthy myocardium by significantly greater native T1 values and these values were even higher in the acute setting (Table 2) [17•, 21]. Whereas infarct-related edema could account for the higher values in acute infarction, high native values in chronic ischemic scar with replacement fibrosis and increased extracellular space remain difficult to explain. Post-contrast T1 values within the infarct zone were found considerably shortened compared to remote areas and showed prolonged washout after contrast administration [17•, 21].
Table 1

Summary of reported values in healthy myocardium by T1 mapping techniquesa

Author, year

T1 mapping sequences

Number of subjects

Field strength (Tesla)

T1 native (msec)

T1 post-contrast (msec)

Acquisition time

Contrast and dose

Messroghli et al. [15], 2004

MOLLI

8

1.5

980 ± 53

494 ± 23

15 min

Magnevist® 0.15 mmol/kg

Sparrow et al. [24], 2005

MOLLI

8

1.5

~1000

~476

10 min

Magnevist® 0.15 mmol/kg

Maceira et al. [26], 2005

IR-GRE

15

1.5

Not reported

579 ± 75

4 min

Magnevist® 0.1 mmol/kg

Iles et al. [14], 2008

ShMOLLI

18

3.0

1169 + 45

Not reported

Not reported

Magnevist® 0.2 mmol/kg

MOLLI

1166 + 60

Piechnik et al. [17•], 2010

ShMOLLI

18

1.5

976 + 48

Not reported

Not reported

Omniscan®

MOLLI

966 + 48

0.1 mmol/kg

ShMOLLI

3.0

1169 + 45

MOLLI

1166 + 60

Ng et al. [23], 2011

MOLLI

19

1.5

Not reported

504 ± 34

10 min

Magnevist® 0.1 mmol/kg

aMean ± standard deviation

IR inversion recovery; MOLLI modified Look-Locker imaging; shMOLLI short MOLLI

Table 2

Summary of reported values in various cardiac conditions by T1 mapping techniquesa

Author, year

Condition

T1 mapping sequences

Number of subjects

T1 native (msec)

T1 post-contrast (msec)

Acquisition time

Contrast and dose

Messroghli et al. [15], 2004

Acute MI

IR-GRE

8

Remote: 721 ± 37

Remote: 362 ± 23

10 min

Magnevist® 0.2 mmol/kg

Infarct: 849 ± 60

Infarct 262 ± 19

Sparrow et al. [24], 2005

Aortic regurgitation

MOLLI

8

Myocardium ~1000

Myocardium ~510

10 min

Magnevist® 0.15 mmol/kg

Maceira et al. [26], 2005

Amyloidosis

 

30

Not reported

Subendocardially: 427 ± 73

4 min

Magnevist® 0.1 mmol/kg

Messroghli et al. [21], 2007

Acute MI

MOLLI

24

Remote: 1011 ± 61

Remote ~460

10 min

Magnevist® 0.15 mmol/kg

Infarct: 1197 ± 76

Infract ~350

Chronic MI

MOLLI

Remote: 987 ± 34

Remote ~440

Infract: 1060 ± 61

Infract ~310

Iles et al. [14], 2008

Heart failure

IR-GRE

25

Myocardium: 874 ± 21

Myocardium: 383 ± 17

15 min

Magnevist® 0.2 mmol/kg

Piechnik et al. [17•], 2010

Recent MI

1.5 T

18

  

Not reported

Omniscan® 0.1 mmol/kg

MOLLI

shMOLLI

3 T

MOLLI

shMOLLI

Ng et al. [23], 2011

Diabetic cardiomyopathy

MOLLI-SR

25

Not reported

425 ± 72

10 min

Magnevist® 0.1 mmol/kg

Jellis et al. [22], 2011

Diabetic cardiomyopathy

MOLLI-SR

67

Myocardium: 841 ± 185

Myocardium: 432 ± 20

12 min

Magnevist® 0.1 mmol/kg

aMean ± standard deviation

IR inversion recovery; MI myocardial infarction; MOLLI modified Look-Locker imaging; shMOLLI short MOLLI; SR saturation recovery

The current evidence—albeit scarce—for T1 mapping in diffuse fibrosis indicates that this methodology could facilitate characterization of the processes that diffusely affect the myocardium. Iles et al. [14] showed that in patients with heart failure, post-contrast T1 times were significantly shorter than in matched healthy volunteers, even when excluding areas of ischemic regional fibrosis. They further report shorter post-contrast T1 values in those patients with severe impairment of diastolic function. Histologically determined amount of fibrosis significantly correlated with T1 values in hearts of transplanted patients. In diabetic population, Jellis et al. [22] reconfirmed the association between shorter post-contrast T1 values and echocardiographic indices of diastolic dysfunction, and more recently, Ng et al. [23] also showed a strong inverse correlation with longitudinal strain. Contrasting these, Sparrow et al. [24] showed no associations between T1 values and global systolic function, left ventricular mass, or regurgitant fraction, though using a sample of only eight patients. The observed discrepancy may be also explained by the findings in animal models of volume overload and chronic regurgitation, which have shown that extracellular matrix in areas of fibrosis does not contain excess collagen compared with normal tissue but contains increased levels of glucosamine and fibronectin [25]. Finally, in patients with amyloid, the highly affected subendocardial layer showed shorter post-contrast T1 value which correlated with markers of increased myocardial amyloid load, such as left ventricular mass and diastolic function [26].

Limitations and Future Directions

Many limitations apply to interpretation of these values. The use of different imaging methodologies and quantification approaches with inconsistent use of heart rate and breathing motion correction may explain some of the observed variability between the studies and also limit cross-referenced comparisons. Quantification of longitudinal relaxation in tissues with high T1 values, such as native myocardium, remains challenging, as it requires long scanning times and consequently long breath-holds, which are not always feasible in the clinical setting. Whether shorter sequences such as shMOLLI provide a viable option for tissue characterization by post-contrast T1 values is unclear, as many other confounding factors come in play. The overall T1 time of the tissue will depend on the relaxivity, the dose, and the tissue distribution and concentration of the contrast agent as well as on the intrinsic T1 values of the tissue which all affect comparability of observed values [27]. As shortened T1 values after gadolinium administration increase exponentially with the wash-out of gadolinium contrast from the interstitial space, an optimal timing of the acquisition with highest sensitivity and specificity to separate healthy myocardium from the diseased needs to be established. Some investigators proposed that a steady-state of contrast agent by continuous infusion and measurement of extracellular volume fraction may be helpful in controlling for all these influences [28], whereas others suggest that given the relatively slow clearance of gadolinium the post-contrast T1 values after 12 min can accurately facilitate such measurement also after a simple bolus [29].

The majority of evidence so far has been derived with 1.5 Tesla field strength magnets; however, with increasing clinical use of 3 Tesla scanners reconfirmation of the expected values at the higher field strength is required as the increased amount of signal requires longer time for longitudinal relaxation to occur [17•, 27]. Post-processing curve fitting tools used in these previous studies have been custom-made and correction for heart rate dependency and respiratory motion has been limited to fully verify the accuracy of pixel-to-pixel quantification and reported values [19, 20]. Lastly, as most studies have been performed as single-center studies focusing on small groups of highly selected patients, efforts are needed to standardize the imaging and post-processing approaches and consolidate the clinical viability of technique preferably in a multicenter trial.

Conclusions

T1 mapping remains a promising and evolving tool for characterization of myocardial fibrosis. Inconsistent imaging and postprocessing strategies and evidence from single-center studies with small sample sizes have not yet allowed for the current methodology to mature into a clinically robust protocol. Efforts are needed to standardize the imaging and post-processing approaches and consolidate the clinical viability of technique preferably in a multicenter trial.

Disclosure

No potential conflicts of interest relevant to this article were reported.

Copyright information

© Springer Science+Business Media, LLC 2012