Magnetization transfer (MT), first demonstrated in vivo by Wolff and Balaban, is a contrast mechanism based on the exchange of magnetization occurring between groups of spins characterized by different molecular environments. MT produces a source of contrast alternative to T 1 and T 2 , which has become widely used in clinical imaging to improve the suppression of static tissue in MR angiography and to increase lesion visibility on conventional MRI when gadolinium-based contrast agents are used. As the MT effect reflects the relative density of macromolecules such as proteins and lipids, it has been associated with myelin content in white matter (WM) of the brain and of the spinal cord. In the attempt to measure myelination, several approaches to quantify the MT effect have thus been proposed. This chapter will review some of the basic theory of MT, the modeling, and the application to spinal cord imaging. It will also present some of the most recent developments, including chemical exchange saturation transfer (CEST).

The MT Phenomenon and Its Relationship to Myelin

The simplest model of MT identifies two compartments (or “pools”) of protons: those in free water (“liquid” or “free” pool) and those bound to macromolecules (“macromolecular” or “semisolid” pool). The former pool is the main contributor to the MRI signal, while the latter one is assumed to be MRI invisible due to its extremely short T 2 (∼10 −5 s). Nevertheless, macromolecular protons are characterized by a broader absorption line shape than the liquid ones, which makes them sensitive to off-resonance irradiation ( Figure 3.4.1 ).

FIGURE 3.4.1, Schematic representation of the absorption line shapes of the liquid (black) and macromolecular pool (gray). The narrow line width of the liquid protons makes them relatively insensitive to off-resonance irradiation at frequencies larger than 1 kHz.

Selective saturation of the macromolecular pool can thus be achieved by applying radio-frequency (RF) energy several kilohertz off resonance from the Larmor frequency. At the field strength typical of clinical scanners (1–3 T), the natural line width of free water is of the order of few tens of Hz; therefore, the direct saturation of the liquid pool by off-resonance irradiation is minimal.

Nevertheless, as an exchange of magnetization occurs between the two pools via cross-relaxation and chemical exchange, this selective saturation is transferred to the liquid pool, and hence results in an attenuation of the MRI signal.

As a consequence, MT is able to probe indirectly macromolecules such as proteins and lipids. Myelin is a lipid–protein structure (its dry mass is approximately 70–80% lipids and 20–30% proteins) wrapped around axons in both the central (CNS) and the peripheral nervous system (PNS). In the CNS (including the spinal cord), myelin is primarily found in the WM, although it is present in smaller quantities also in the gray matter (GM). The main purpose of myelin is to act as an insulator, thus increasing the speed of action potential transmission. The ability to measure myelin in vivo would have very important consequences, as myelin is the primary target of demyelinating diseases (such as multiple sclerosis, MS), but it is also believed to be involved in degenerative processes secondary to neuronal or axonal damage. Additionally, processes of myelination during neurodevelopment are believed to be responsible for a number of psychiatric disorders. MT imaging, together with diffusion-weighted imaging ( Chapter 3.1 ), q -space imaging ( Chapter 3.2, Chapter 3.3 ), and short T 2 mapping ( Chapter 3.5 ) are among the most promising MR techniques for the assessment of myelin.

MAGNETIZATION TRANSFER MEASURES

  • MT-weighted —Magnetization transfer affects the contrast of the image, but it is not quantitative

  • MTR (MT ratio) —Semiquantitative measure reflecting the amount of bound protons and calculated from two images

  • MTCSF —MT-weighted scan normalized to the CSF signal. Only one image needed, but results dependent on T 1 / T 2

  • qMT —quantitative MT technique that models the signal change with two pools (bound and free). Parameters determined from it include the exchange rate ( RM 0 B ), the macromolecular pool ratio ( F ), and the T 2 of the bound pool ( T 2 B ). Long scan times are associated with this measure because it needs several data points to fit the model to the data

  • CEST —Selective irradiation of labile proton pools resonating at frequencies other than water

MT Contrast and MTR

MT saturation results in a suppression of the MRI signal that is proportional to the amount of macromolecules in a given tissue. For this reason MT contrast has been used from the start to suppress unwanted signal. For example, in MR angiography, the signal contrast between the blood and other tissue can always be enhanced by using MT (which does not affect blood) to further suppress the background tissue signal. With a similar purpose MT imaging can be applied to spinal cord imaging to improve contrast and sharpness between spinal cord and cerebrospinal fluid (CSF) and to increase the conspicuity of focal lesions.

In this chapter, however, we focus on the quantitative applications of the technique.

The simplest approach to quantify the degree of signal loss in an MT-saturated experiment is to express it as the MT ratio (MTR). The MTR is computed as the percentage difference of two images, one acquired with off-resonance saturation ( M S ) and one without ( M 0 ):


MTR = M 0 M S M 0 × 100.

This simple operation can be computed on a voxel-by-voxel basis to obtain an MTR map, typically expressed as percentage units (pu), although in some cases is expressed as a fraction. As discussed in the previous section, some evidence suggests that molecules associated with myelin dominate the MT exchange process in WM and that MTR increases with myelin content. MTR was shown to be sensitive to subtle damage in the normal-appearing WM of patients with demyelinating diseases, such as multiple sclerosis.

The MTR became very popular as an approach to characterize WM disease and was soon extended to imaging of the cord. An example of MTR map of the cervical cord is shown in Figure 3.4.2 .

FIGURE 3.4.2, Magnetization transfer ratio (MTR) of the cervical cord in a healthy subject. Two scans, without (A) and with (B) saturation are combined to compute the percentage difference, or MTR (C).

The most widely adopted approach to MT-weighted acquisition is the so-called pulsed MT, in which a spoiled gradient echo (2D or 3D) or a spin echo acquisition is modified to accommodate high power, off-resonance, saturation pulses, which are played out just before each excitation. Similar acquisitions are available on most commercial scanners, and MT weighting is often offered as an option with several sequences. More details on MT imaging acquisition are given in Section 3.4.4 ; examples of acquisition protocols suitable for the spinal cord are also shown in Table 3.4.1 .

TABLE 3.4.1
MT Sequence Selection
Sequence TR (ms) TE (ms) MT Prepulse (Offset Frequency) Scan Time (Matrix) B 0 References
Three-dimensional spoiled gradient echo 110 13 5-lobed sinc (1.5 kHz) 8 min (368 × 326) 3.0 T Smith et al.
Three-dimensional gradient echo 28 3.2 Gaussian (1.2 kHz) 5 min (256 × 256) 3.0 T Cohen-Adad et al.
Three-dimensional gradient echo 50 12 5-lobed sinc (1–63 kHz) 32 min (256 × 228) 1.5 T Smith et al.
Fatemi et al.
Two-dimensional gradient echo 600 25 Gaussian (1.5 kHz) (128 × 128) 1.5 T Agosta et al.
Fast spin echo 1600 17 3-lobed sinc (1 kHz) 17.7 min (256 × 192) 1.5 T Silver et al.
Two-dimensional gradient echo 616 22 Gaussian (1.5 kHz) 5.3 min (128 × 256) 1.0 T Lycklama a Nijeholt et al.
Gradient echo 750 23 Sinc (0.7 kHz) 4.8 min (256 × 180) 0.3 T Yoshioka et al.

One of the limitations of the MTR, however, is that it is the result of the combination of several more fundamental quantities, and it is highly dependent on the acquisition parameters. The characteristics of the MT-saturating pulse, including the shape, the amplitude, the duration, and the offset frequency, for example, have a major effect on the measured MTR. Similarly, imaging parameters such as the repetition time (TR) and the excitation flip angle can affect the result.

Measuring MT in the Spinal Cord

The human spinal cord contains segregated sensory and motor pathways that have been difficult to quantify using conventional MRI techniques. Spinal cord lesions are typically more symptomatic than brain lesions and correlate better with the degree of physical disability. MT-MRI provides quantitative values about the degree of damage, while conventional MRI gives a simple volumetric measure of disease burden. Using MT-MRI, a study showed that it is possible to assess demyelination of specific spinal pathways and that signal abnormalities in the dorsal and lateral columns of the spinal cord are correlated with vibration sensation (dorsal) and strength (lateral). Another study also demonstrated that measures of MTR in the dorsal and ventrolateral spinal cord predicted the sensory and the motor disability, respectively.

Clinical Relevance of MT-MRI in the Spinal Cord

The structural changes occurring within and outside T 2 -visible lesions have been quantified using MT-MRI. As one major advantage of MT is its specificity to demyelination and degeneration, as assessed by histopathology, it is especially helpful in identifying microscopic disease activity in normal-appearing brain tissue on conventional MRI scans. MT-MRI of the spinal cord has shown great potential for clinical assessment of patients suffering from demyelinating diseases, such as MS, adrenomyeloneuropathy (AMN), and neuromyelitis optica (NMO).

Multiple Sclerosis

MS is an inflammatory demyelinating disease of the CNS as extensively described in Chapter 1 of this book. MS selectively affects the myelin sheath of the CNS and oligodendrocytes, which causes multifocal demyelinated plaques in the brain, the spinal cord, and the optic tract. MS is characterized by both focal and spatially diffuse lesions of the brain and spinal cord with heterogeneous pathologies. Especially, the spinal cord is a common site of involvement in MS as cord pathology is a major cause of the disability suffered by MS patients. MTR is most commonly used to quantify tissue changes in MS and has been measured in the spinal cord of MS patients. Previous studies showed that significant changes in MTR can be detected between controls and MS patients in the spinal cord WM. In MS patient groups, cord MTR decreased in both normal-appearing tissue and lesions, and furthermore correlated with the degree of demyelination and axonal density in the spinal cord, similarly found in the brain. A correlation between MTR values and disability has also been observed in the spinal cord of MS patients. MT-MRI studies in the upper cervical cord showed that diffuse abnormalities measured by MTR are an important contributor to disability, implying that the assessment of regional cord damage can contribute to the understanding of the factors associated with the development of disability. Such results show that MTR has greater pathological specificity for changes in myelin content than conventional MRI does. Furthermore, several groups reported that the MTR values in the cord are independent or only partially correlated with focal abnormalities in the brain, suggesting that changes seen in the cord are not merely the result of Wallerian degeneration of the motor tract and do not simply reflect brain pathology. This indicates that measuring pathology of the spinal cord using MT-MRI may be a “rewarding exercise in terms of understanding MS pathophysiology”.

Adrenomyeloneuropathy

AMN is a noninflammatory neurodegenerative disease with millimeter-size lesions in the dorsal and lateral columns of the cervical cord. In contrast to MS, conventional MRI has shown no significant changes in patients with AMN other than cord atrophy late in the disease as AMN does not have an overt inflammatory component. Previous studies showed that MTR values are sensitive to cord pathology in the cervical-dorsal columns of AMN patients and correlates well with EDSS and quantitative sensory-motor tests. A subsequent study performed by the same group demonstrated for the first time that qMT-derived metrics can be a potential quantitative biomarker to assess and characterize human spinal cord tissues in disease.

Neuromyelitis Optica

Neuromyelitis optica (NMO), also known as Devic's disease, is an inflammatory demyelinating disease that selectively affects the optic nerve and the spinal cord. There has long been debate as to whether NMO is a variant of MS. However, clinical, immunological, and pathological characteristics of NMO are increasingly being used to distinguish it from MS. In patients with NMO, no brain abnormalities are generally found on the T 2 -weighted scans, as one of the supportive criteria for NMO diagnosis is a negative conventional MRI scan of the brain at disease onset. Filippi et al. investigated the occult damage of the normal-appearing brain tissue (NABT) in patients with relapsing NMO using MT-MRI and found no difference between patients and control subjects, whereas MS patients had a significantly lower histogram average MTR and peak height. Rocca et al. reported the abnormal changes in the normal-appearing WM and GM (NAWM/NAGM) in patients with NMO, and found reduced MTR of the NAGM in patients with NMO, which suggested the presence of GM damage in these patients. Despite potential advantages of investigating pathophysiology of NMO, MT-MRI has not been widely used to study the spinal cord in vivo. Recently, a study based on a novel approach to MT-MRI, namely MTCSF, MT-weighted scan normalized to the CSF signal (described further in Section 3.4.3.2 ), demonstrated that MTCSF values of NMO patients were significantly higher than those of control subjects, suggesting that the assessment of NMO cervical cord damage is feasible using the quantitative capability of MT-MRI.

Challenges

MT-MRI of the cervical cord presents technical difficulties, mainly because of the size of its structure and tendency to move during imaging. First of all, around 1 cm in diameter of region of interest requires at least submillimeter spatial resolution to distinguish the small GM and WM structures in the spinal cord. However, acquisition of high-resolution images at 1.5 T or 3 T results in reduced signal-to-noise ratio (SNR) and increased motion sensitivity through longer scan times. Another technical challenge is that the cord is subject to many types of motion such as cardiac pulsation and respiration. Despite availability of various motion correction algorithms, certain types of motion such as out-of-plane motions along the slice direction are more difficult to accurately correct, especially when imaging the cross section of the spinal cord where variations along the cord are slower than in plane. To this end, spinal cord MT-MRI has been mainly limited to assess total disease burden, which allows detection of large inflammatory lesions. To overcome such limitations, the MTCSF approach has been developed and utilized in studies with various clinical applications. In MTCSF, MT effects are quantified in the spine using cerebrospinal fluid as an internal intensity standard, allowing interindividual comparison of MT-weighted data without the need for a reference scan:


MTCSF = M S M mean CSF ( ROI )


M mean CSF ( ROI )
is obtained from an automatically selected region of interest (ROI), within the CSF, where no MT is expected to occur, CSF being substantially free water. A study showed tract-specific signal abnormalities in the dorsal and lateral columns of the spinal cord in MS patients using MTCSF. One limitation of MTCSF is its dependency on T 1 and T 2 contrast, which might be altered in the presence of inflammation. For this reason this method is unlikely to be able to distinguish between demyelination and inflammation, and might be limited to clinical applications in which inflammation is known not to occur. Figure 3.4.3 shows a comparison between standard MTR and MTCSF of the cord. As MTCSF is an MT-weighted image scaled by a single value, the image quality appears to be greater than MTR, which instead is a voxel-by-voxel measure where noise contributes to both the nominator and denominator of the fraction.

FIGURE 3.4.3, A comparison between MTCSF (A) and MTR (B) in the cervical cord. In MTCSF the signal intensity is normalized by the average signal intensity measured in a region of interest containing CSF. Notice that MTCSF is a normalized MT-weighted image while MTR is a semiquantitative measure.

It should be noted that MTR is a semiquantitative measure that not only depends on the size of the macromolecular pool but also on the exchange rate between the bound and mobile proton pools. To overcome the multiparametric dependence of MTR, qMT may provide a more direct surrogate of myelin content. However, estimation of qMT requires assumptions on the number of proton pools in the sample and necessitates multiple MT measurements as well as an independent measurement of T 1 . Alternatively, MT-MRI can be combined with diffusion-weighted MRI (DW-MRI). Studies showed that combining those two biomarkers can provide information more specific to WM pathology in MS, spinal cord injury, and amyotrophic lateral sclerosis. Another study demonstrated the high reproducibility of MT and DW-MRI at 3 T suggesting robust assessment of WM integrity in the human cervical cord.

You're Reading a Preview

Become a Clinical Tree membership for Full access and enjoy Unlimited articles

Become membership

If you are a member. Log in here