Application of Neuromelanin MR Imaging in Parkinson Disease

Abstract

MRI has been used to develop biomarkers for movement disorders such as Parkinson disease (PD) and other neurodegenerative disorders with parkinsonism such as progressive supranuclear palsy and multiple system atrophy. One of these imaging biomarkers is neuromelanin (NM), whose integrity can be assessed from its contrast and volume. NM is found mainly in certain brain stem structures, namely, the substantia nigra pars compacta (SNpc), the ventral tegmental area, and the locus coeruleus. Another major biomarker is brain iron, which often increases in concert with NM degeneration. These biomarkers have the potential to improve diagnostic certainty in differentiating between PD and other neurodegenerative disorders similar to PD, as well as provide a better understanding of pathophysiology. Mapping NM in vivo has clinical importance for gauging the premotor phase of PD when there is a greater than 50% loss of dopaminergic SNpc melanized neurons. As a metal ion chelator, NM can absorb iron. When NM is released from neurons, it deposits iron into the intracellular tissues of the SNpc; the result is iron that can be imaged and measured using quantitative susceptibility mapping. An increase of iron also leads to the disappearance of the nigrosome-1 sign, another neuroimage biomarker for PD. Therefore, mapping NM and iron changes in the SNpc are a practical means for improving early diagnosis of PD and in monitoring disease progression. In this review, we discuss the functions and location of NM, how NM-MRI is performed, the automatic mapping of NM and iron content, how NM-related imaging biomarkers can be used to enhance PD diagnosis and differentiate it from other neurodegenerative disorders, and potential advances in NM imaging methods. With major advances currently evolving for rapid imaging and artificial intelligence, NM-related biomarkers are likely to have increasingly important roles for enhancing diagnostic capabilities in PD. EVIDENCE LEVEL: 1 TECHNICAL EFFICACY: Stage 2.

Keywords: Parkinson disease; magnetic resonance imaging; magnetization transfer contrast; neuromelanin; nigrosome 1.

FIGURE 1

The anatomy of the SN and its five nigrosome territories shown on serial axial sections based on figure 1 from the article by Massey et al.

FIGURE 2

Cartoon illustration of the nigrosome 1 variants. Source: Figure reprinted with permission from reference 2020 John Wiley & Sons, Inc.

FIGURE 3

(a–f) The N1 sign on QSM‐hpf data in a pseudo transverse plane with different degrees of rotation about the y‐axis for (−20°, −10°, 0°, 10°, 20°, and 30°, respectively) from a 63‐year‐old healthy control. The 0° angle represents the original scanning orientation along the anterior commissure–posterior commissure line. The yellow arrows show that as the scanning plane changes from −20° to 30°, the shape of the N1 becomes more elongated as the angle starts to match the angle to which the SN is tilted away from the transverse plane.

FIGURE 4

The presence of the N1 sign in different movement disorders. The five columns represent, in the following order: the original 30° first echo NM image with a resolution of 0.67 mm × 1 mm × 1.34 mm; the tSWI image as applied to the QSM data of the HPF phase from the first echo; the tSWI image as applied to the QSM data of the HPF phase of the third echo; the QSM data over three echoes using MEiSWIM; and the tSWI of the first echo using the original QSM data (with no HPF). Since the regions of NM can sometimes have non‐zero iron content, using the original QSM can obscure the presence of the N1 sign depending on how the tSWI filter is designed. Hence, we tend to use the tSWI of the QSM data of the HPF phase images for interpreting the presence of the N1 sign. Each row represents a different movement disorder except for the first row, which is for a HC. The ET, RBD, MSA‐C, MSA‐P and the PD‐TD either show bilateral or unilateral N1 signs. The PSP and PIGD cases show bilateral loss of the N1 sign and an atrophied SN. The N4 sign appears to be present in the PIGD case although no N1 sign appears to be present. The red boundaries are from the NM images and are superimposed on all the images to make it clear where the N1 sign must lie if it is to be considered part of the NM territory.

FIGURE 5

The role of different image types and different processing. Best visualizing the NM and iron content depend on the MT preparation used, the sequence type used and the choice of imaging parameters. In these data, we used a 3D MT prepared gradient echo SWI sequence with a resolution of 0.67 mm×1 mm×1.34 mm. These data were collected on a Philips Ingenia scanner. The 30° first echo MTC image in the upper left shows high SNR and visibility. The boundary of the NM is drawn here and copied to the first echo tSWI on the bottom left image. The N1 sign should appear within the NM boundary as is in fact the case. Yellow arrows are then used to highlight the N1 sign in all the lower row tSWI images. This individual has low iron content and so the N1 sign cannot be clearly seen until the fifth echo magnitude image (TE = 37.5 ms). Note the bright region of the N1 is clearly seen even in the early echoes of the tSWI data.

FIGURE 6

A direct comparison of 2D TSE and 3D GRE NM‐MRI sequences for a 41‐year‐old man. Images (a)–(c) in the left panel were from the 2D TSE sequence without an MT pulse. Images (d) and (e) in the middle panel were from the 2D TSE sequence with an MT pulse. Images (g)–(o) in the right panel were from the 3D multiecho GRE sequence. All three sequences show the NM reasonably well although the data are still somewhat noisy (a, d, g) and LC (b, e, h) territories. Of note, images (g) and (h) were averaged over two slices to match the slice thickness in the TSE sequences. The reformatted coronal view (c, f, i) showed that the 3D GRE sequence had a sharper depiction of the LC structure given by thin slice 3D acquisition. In addition, the 3D GRE sequence was also used to generate SWI (j and l), QSM (k), and MRA (m–o) images. Data were acquired using a Siemens VERIO 3 T scanner with a 32‐channel head coil. Detailed imaging parameters are listed in Table S2.

FIGURE 7

Visualizing the NM without an MTC pulse. Top row, NM in the LC: (a) From the 6° STAGE data, the spin density‐weighted short TE (7.5 msec) is good enough to visualize the NM because the surrounding CSF is sufficiently suppressed and the high water content of the LC makes it visible. (b) Coronal reformat of the data showing the LC. Both (a) and (b) are from a resolution of 0.67 mm × 1 mm (interpolated to 0.67 mm × 0.67 mm) × 1.34 mm. (c) High‐resolution MTC coronal view with a resolution of 0.67 mm × 0.67 mm. Bottom row, NM in the SN: (d) an average over two 3 minutes scans with a 3° flip angle; (e) the usual MTC NM images as a reference (5 minutes scan); and (f) the original 3° flip angle (3 minutes scan). The resolution for figures (d–f) is the same as the resolution in figures (a,b).

FIGURE 8

Rapid long‐echo imaging using waveSWI. (a) tSWI of the original third echo (22.5 msec) data, (b) the original third echo magnitude data, (c) the average of the waveSWI 15 msec and 30 msec echoes, and (d) tSWI of this averaged image. The conventional acquisition with 0.67 mm × 1 mm × 1.34 mm takes 4.5 minutes per scan (with TR = 29 msec) while the waveSWI (even with its longer repeat time of 45 msec) takes only 3 minutes for whole brain coverage. The long echo times help to visualize the N1 sign.