In Vivo Proton Exchange Rate (k_ex) MRI for the Characterization of Multiple Sclerosis Lesions in Patients

Abstract

Background:

Currently available radiological methods do not completely capture the diversity of multiple sclerosis (MS) lesion subtypes. This lack of information hampers the understanding of disease progression and potential treatment stratification. For example, inflammation persists in some lesions after gadolinium (Gd) enhancement resolves. Novel metabolic and molecular imaging methods may improve the current assessments of MS pathophysiology.

Purpose:

To compare the in vivo proton exchange rate (k_ex) MRI with Gd-enhanced MRI for characterizing MS lesions.

Study Type:

Retrospective.

Subjects:

Sixteen consecutively diagnosed relapsing-remitting multiple sclerosis (RRMS) patients.

Field Strength/Sequence:

3.0T MRI with T₂-weighted imaging, postcontrast T₁-weighted imaging, and single-slice chemical exchange saturation transfer imaging.

Assessment:

MS lesions in white matter were assessed for Gd enhancement and k_ex elevation compared to normal-appearing white matter (NAWM).

Statistical Tests:

Student’s t-test was used for analyzing the difference of k_ex values between lesions and NAWM, with statistical significance set at 0.05.

Results:

Of all 153 MS lesions, 78 (51%) lesions were Gd-enhancing and 75 (49%) were Gd-negative. Without exception, all 78 Gd-enhancing lesions showed significantly elevated k_ex values compared to NAWM (924 ± 130 s⁻¹ vs. 735 ± 61 s⁻¹, P < 0.05). Of 75 Gd-negative lesions, 18 lesions (24%) showed no k_ex elevation (762 ± 29 s⁻¹ vs. 755 ± 28 s⁻¹, P = 0.47) and 57 (76%) showed significant k_ex elevation (950 ± 124 s⁻¹ vs. 759 ± 48 s⁻¹, P < 0.05) compared to NAWM. MS lesions with k_ex elevation appeared nodular (118, 87.4%), ring-like (15, 11.1%), or irregular-shaped (2, 1.5%).

Data Conclusion:

For Gd-enhancing lesions, k_ex MRI is highly consistent with Gd-enhanced images by showing 100% of elevated k_ex. For all Gd-negative lesions, the discrepancy on k_ex MRI may further differentiate active slowly expanding lesions or chronic inactive lesions, supporting k_ex as an imaging biomarker for tissue oxidative stress and inflammation.

Level of Evidence:

2

Technical Efficacy Stage:

3

MULTIPLE SCLEROSIS (MS) is the most common autoimmune inflammatory disease of the central nervous system, and a leading cause of neurologic disability in young adults.¹ MS is characterized by multifocal demyelination, axonal injury, neuronal loss, and inflammation.² Early therapeutic intervention is commonly considered the key to minimize the neurological disability of MS patients; however, this is dependent on early and accurate diagnosis.³ Patients exhibit substantial heterogeneity in clinical and histopathological presentation and progression over time.⁴ Novel molecular and metabolic imaging methods may be useful to refine the identification of lesion subtype to assist in understanding disease, including the contribution of reactive oxygen species and other direct causes of tissue damage.⁵ Better imaging strategies may enable prospective identification of aggressive disease to facilitate patient stratification into aggressive vs. conservative treatment plans.⁶ Moreover, new noninvasive imaging biomarkers of neuroprotection and repair are needed to guide clinical decisions and allow evaluation of experimental therapeutics, particularly those directed toward regeneration in gadolinium-negative lesions.^7–10

Currently, the diagnosis of MS is based on the McDonald criteria, in which magnetic resonance imaging (MRI) is an essential part.¹¹ Besides clinical symptomatic attacks, fulfilling the imaging criteria for dissemination in space (DIS) and dissemination in time (DIT) are the important points for the diagnosis of MS¹¹ according to the McDonald criteria. Dissemination in space means an MS patient has lesions located in at least two of the four distinct anatomical locations: periventricular, infratentorial, spinal cord, and juxtacortical regions. MS lesions can be classified into five pathological stages: early active, late active, smoldering, inactive, and shadow plaques.¹² Dissemination in time means the lesions in an MS patient are in different stages. Gadolinium (Gd) enhancement on postcontrast T₁-weighted images (T₁WI), signaling blood–brain barrier disruption and active inflammation, is the widely used MRI method to define the stage of an MS lesion.¹³ A patient fulfils the criteria of DIT based on the coexistence of Gd-enhancing and nonenhancing lesions or the appearance of a new lesion on postcontrast T₁WI or T₂-weighted images (T₂WI).¹¹ However, the narrow window of Gd enhancement in MS lesions lead to underestimation of their activities. The average duration of Gd enhancement in individual new lesions is ~3 weeks, and the vast majority of Gd enhancement resolves in 6 months.¹⁴ Afterward, the stages of MS lesions cannot be identified when they turn into Gd-negative lesions. However, inflammation persists for a long time after the Gd enhancement is resolved.^15,16 These nonenhanced lesions can be categorized into chronic active lesions (slowly expanding lesions) and chronic inactive lesions according to the presence or absence of ongoing inflammation.^16–18 Therefore, DIT is underestimated with Gd-enhanced MRI alone and new MRI methods will be highly desirable for improving the DIT determination. In addition, some recent studies proposed a correlation between lesions with typical paramagnetic rims on susceptibility-weighted imaging (SWI) and the chronic active MS lesions (slowly expanding lesions), and there is still a significant ongoing need to seek new MRI biomarkers to improve the characterization of MS lesions and the DIT determination.^19,20

As an emerging MR contrast, the in vivo proton exchange rate (k_ex) quantified from direct water saturation (DS) removed omega plots has been recently shown applicable in healthy human brains.^21,22 Through exchangeable protons, proton exchange happens between bulk water and metabolites, such as proteins, peptides, amino acids, and other small molecules. The major proton exchanging sites in endogenous metabolites include amine −NH₂, amide −NH, hydroxyl −OH, and hydrosulfuric −SH groups. As a fundamental biophysical process, proton exchange plays important roles in producing MRI contrasts including T₁- and T₂-relaxations, chemical exchange saturation transfer (CEST), magnetization transfer (MT), and nuclear Overhauser enhancement (NOE).²² We have previously developed CEST techniques for the in vivo assessment of myo-inositol and of glutamate in healthy brain and neurological lesions.^23,24 The aim of the present study was to assess the use of k_ex MRI, in comparison with Gd-enhanced MRI, to characterize and stage MS lesions.

Materials and Methods

The study was approved by the Institutional Review Board (IRB) and written informed consent was obtained from all patients. All patients were previously diagnosed with relapsing–remitting MS according to the 2017 revisions of the McDonald criteria.¹¹

Imaging Acquisition and Analysis

All patients underwent MR exams on a 3.0T GE MR750 scanner (GE Healthcare, Milwaukee, WI) with a 32-channel head coil and imaging sequences including T₂WI, k_ex MRI based on omega plotting,²² and Gd-enhanced MRI. The parameters for T₂WI were time of repetition (TR) = 5300 msec, time of echo (TE) = 92 msec, field of view (FOV) = 24 × 24 cm², matrix = 512 × 512, slice thickness = 5 mm. The k_ex MRI protocol consisted of the acquisition of five CEST Z-spectral data with different saturation powers (B₁) from a single slice covering most MS lesions delineated by T₂WI, which was decided by a neuroradiologist (Y.Z., 8 years’ experience in neuroradiology). The CEST Z-spectra were acquired using a customized single-shot fast low angle shot (FLASH) sequence with imaging parameters: short TR = 3000 msec, TE = 22.6 msec, FOV = 24 × 24 cm², matrix size = 128 × 128, slice thickness = 5 mm, number of average = 1, CEST saturation power = 1, 2, 3, 4, and 5 μT, and saturation duration = 1.5 sec.^24,25

A total of 33 frequency offsets were acquired at each saturation B₁, ranging from −6 to +6 ppm with an increment of 0.25 ppm, +15.6 ppm, and + 39.1 ppm (used as reference image). The total scanning time for collecting each Z-spectral dataset was 3 minutes 18 seconds and the total imaging time for k_ex MRI in this study was ~16.5 minutes. After the k_ex MRI protocol, postcontrast T₁-weighted imaging was performed 5 minutes after administration of 0.1 mmol/kg gadopentetate dimeglumine (Magnevist, Bayer, Berlin, Germany). The scanning parameters for the Gd-enhanced MRI were TR = 500 msec, TE = 8 msec, FOV = 24 × 24 cm², matrix = 320 × 320, slice thickness = 5 mm.

All image processing and data analyses were performed using custom-written scripts in MATLAB (MathWorks, Natick, MA). The acquired Z-spectra were used to construct k_ex maps based on the direct-saturation-removed omega plot method.²² In brief, each Z-spectrum was modeled as a linear combination of two Lorentzian components corresponding to the water DS spectrum and the residual spectrum. The residual spectral signal was used to construct the omega plot of M_z/(M₀ − M_z) as a linear function of $1/{\omega }_1^2$, where M₀ and M_z are the residual spectral signals far from water resonance (+39.1 ppm) and at the metabolite resonance, respectively, and ω₁ is the amplitude of the saturation RF pulse. The X-axis intercept of the linear fit to the omega plot provided the k_ex value based on Eq. 1 as follows:

$$k_{ex}^2=\frac{-1}{X_{\mathit{\text{intercept}}}}$$ (1)

A generalized omega plot method (Eq. 2) was generated in the case of multiple proton exchange mechanisms contributing to the saturation transfer signal,²² including MT, CEST, and NOE:

$$\frac{M_z}{M_0-M_z}=\frac{1}{T_1^w}\left(\frac{{\sum }_{i=1}^n\frac{k_{s_i}}{f_{s_i}}}{{\omega }_1^2}+{\sum }_{i=1}^n\frac{1}{f_{s_i}{k}_{s_i}}\right)$$ (2)

in which $T_1^w$ is the longitudinal relaxation rate of the water, $k_{s_i}$ are the exchange rates between the metabolites’ pool and the water pool, and $f_{s_i}$ are the fraction of protons on the metabolites relative to water.

Since Eq. 2 indicates a linear relationship between $\frac{M_z}{M_0-M_z}$ and the inverse of the square of saturation pulse power $\frac{1}{{\omega }_1^2}$, the extraction of the X-axis intercept provided an average readout (Eq. 3) of the convoluted exchanging rate of multiple exchange mechanisms:

$$\frac{-1}{{\mathcal{X}}_{\mathit{\text{intercept}}}}=\frac{{\sum }_1^n\frac{k_{s_i}}{f_i}}{{\sum }_1^n\frac{1}{f_i{k}_{s_i}}}=M_1\left(\frac{k_{s_i}}{f_i}\right)\times M^{-1}\left(f_i{k}_{s_i}\right)=p^2\left(f_i{k}_{s_i}\right)=p^2\left({\overline{k}}_{ex}\right)$$ (3)

where M₁ and M⁻¹ are the arithmetic and the harmonic mean operator, respectively. Following the k_ex quantification for each voxel, k_ex maps were produced.

Lesions with T₂ hyperintensities were assumed to be MS lesions. White matter regions without an abnormal signal were considered to be normal-appearing white matter (NAWM). Regions of interest (ROIs) of MS lesions were drawn manually by a neuroradiologist (H.Y., 3 years of experience) by outlining the edge of the lesions on T₂WI. ROIs of NAWM (circle, 55 mm² in size, at least six ROIs for each MS patient) were also drawn on T₂WI as an internal reference. All masks of ROIs were overlaid on k_ex images to calculate the k_ex values.

Three radiologists (H.Y., Q.C., W.C., with 3, 2, and 19 years of experience in neuroradiology, respectively) reviewed all MR images independently to assess MS lesions based on visual impression of three characteristics: the presence of Gd enhancement, k_ex signal elevation, and lesion shape (nodular, ring-like, or irregular-shaped), with the interobserver variation assessed. Any discrepancy for each characterization by these three neuroradiologists was resolved by majority votes.

Statistical Analysis

Statistical analyses were performed using SPSS software (v. 22, Chicago, IL). Descriptive results are presented as mean ± standard deviation (SD). Interobserver agreement between three radiologists was assessed using Fleiss’ kappa statistic. The normality of data was assessed with the Kolmogorov–Smirnov test. Two-tailed Student’s t-test was used for analyzing the difference of k_ex values between lesions and NAWM, and between Gd-enhancing MS lesions and Gd-negative MS lesions, with statistical significance set at 0.05.

Results

Sixteen consecutive MS patients (seven men and nine women) were recruited for this retrospective study between December 2018 and August 2019. Their age ranged from 18–50 years old (32.9 ± 8.1 years), expanded disability status scale scores (EDSS) ranged from 0–6.5 (median, 2.5), and the disease durations ranged from 1–12 years (7.6 ± 3.6 years). Of 16 patients, eight patients had a clinical relapse with corticosteroid therapy following the MRI examinations. Figure 1a demonstrates Z-spectra data from a representative MS lesion at different saturation B₁ powers. Figure 1b shows the fitting of the Z-spectrum for direct saturation and DS-removed residual signals. The quantification of k_ex of lesions or NAWM is demonstrated in Fig. 1c. Most MS lesions showed elevated k_ex values.

FIGURE 1:

Demonstration of k_ex quantification in MS brain. (a) Representative Z-spectra from an MS lesion for five different saturation powers, B₁ = 1–5 μT. (b) Demonstration of inversed Z-spectrum at B₁ = 1 μT with Lorentzian fittings to remove the water direct saturation (DS) contribution. DS was subtracted from the entire Z-spectrum and the residual spectrum was used to construct the omega plot. (c) Typical omega plots constructed using signals from normal-appearing white matter (NAWM) and an MS lesion. The X-intercept of the linear fit provides a readout of the exchange rate (k_ex).

Three neuroradiologists achieved great interobserver agreement on characterizing MS lesions. Specifically, the kappa values for interobserver agreement between the three neuroradiologists were 0.922 for the presence of Gd enhancement, 0.825 for k_ex signal elevation, and 0.825 for lesion shape.

Table 1 summarizes the clinical and imaging information of all patients. A total of 153 MS lesions with k_ex images were identified in 16 MS patients. Of the 153 MS lesions, 78 lesions (51%) were Gd-enhancing and 75 (49%) were Gd-negative. Furthermore, 135 MS lesions showed k_ex elevation relative to NAWM, with the majority (118/135, 87.4%) showing nodular k_ex elevation, the others showing ring-like (15/135, 11.1%) and irregular-shaped (2/135, 1.5%) k_ex elevation.

TABLE 1.

Demographic Data, EDSS Score, and Number of Lesions of Each MS Patient

Patient number	Sex and age	EDSS Score	Total lesion number on T2WI	Number of lesions on kex maps	Number of lesions with Gd enhancement	Number of lesions with kex elevation	Number of lesions without kex elevation
1	Female, 18y	2	6	4	0	4	0
2	Male, 44y	5	43	27	27	27	0
3	Male, 37y	3	17	9	0	9	0
4	Male, 32y	1	3	1	1	1	0
5	Female, 33y	2.5	13	6	0	4	2
6	Female, 41y	1.5	8	4	0	4	0
7	Male, 34y	6.5	38	24	24	24	0
8	Male, 50y	3	19	8	0	6	2
9	Female, 32y	2.5	17	11	3	11	0
10	Female, 23y	2	13	7	2	3	4
11	Female, 31y	1	5	2	0	2	0
12	Male, 27y	3	36	19	1	13	6
13	Female, 33y	3	25	13	13	13	0
14	Female, 31y	2.5	16	7	6	6	1
15	Female, 37y	2	23	10	0	7	3
16	Male, 23y	0	2	1	1	1	0

EDSS = expanded disability status scale.

In all Gd-enhancing lesions, there were significantly elevated k_ex values in comparison to NAWM (924 ± 130 s⁻¹ vs. 735 ± 61 s⁻¹, P < 0.05). MR images from a representative patient with Gd-enhancing lesions are shown in Fig. 2. A representative patient without Gd-enhancing lesions is shown in Figs. 3 and 4, with some lesions that do not have elevated k_ex signal. Overall, in lesions without Gd enhancement, 18 lesions (24%) showed no k_ex elevation (762 ± 29 s⁻¹ vs. 755 ± 28 s⁻¹, P = 0.474) and 57 (76%) showed significant k_ex elevation (950 ± 124 s⁻¹ vs. 759 ± 48 s⁻¹, P < 0.05) compared to NAWM (Figs. 5, 6).

FIGURE 2:

Images from the number 2 MS patient. A total of 27 lesions were identified on the selected slice of T₂WI, including 12 lesions labeled by arrows. All Gd-enhancing lesions showed k_ex elevation (only 12 lesions labeled by black and white arrows), including 23 lesions with ring-like k_ex elevation (only eight lesions labeled by white arrows) and four with nodular k_ex elevation (black arrows).

FIGURE 3:

Images from number 3 MS patient. Nine lesions were identified on the selected slice of T₂WI (arrows and arrowheads). All lesions showed k_ex elevation without Gd enhancement, including one lesion with ring-like k_ex elevation (white arrow), six with nodular k_ex elevation (black arrows), and two with irregular-shaped k_ex elevation (black arrowheads).

FIGURE 4:

Images from number 5 MS patient. Six lesions were identified on the selected T₂WI slice (white and black arrows). No lesions showed Gd enhancement. While on the k_ex map, two of them (white arrows) showed similar k_ex to the normal-appearing white matter, and the other four lesions (black arrows) showed nodular elevated k_ex.

FIGURE 5:

MS lesions can be categorized into three patterns by Gd enhancement and k_ex mapping.

FIGURE 6:

All Gd-enhancing MS lesions and 76% non-Gd-enhancing MS lesions showed elevated k_ex in comparison to normal-appearing white matter.

Discussion

This study demonstrated the feasibility of k_ex imaging in MS patients. Interestingly, k_ex MRI was found to be highly consistent with Gd-enhanced contrast maps in Gd-enhanced enhancing lesions because all of them showed elevated k_ex. Nonenhancing MS lesions can be further classified into two patterns by k_ex MRI, showing elevated or unchanged k_ex compared to NAWM, suggesting that k_ex mapping may serve as a potential approach to further characterize Gd-negative MS lesions.

Under different B₁s, varied proton species of CEST or MT-expressing metabolites are optimally saturated. For example, the amide proton transfer (APT) effect is optimally saturated under low B₁ around 1 to 2 μT due to the relatively slow amide proton exchange rate.²⁶ The glutamate CEST effect, on the other hand, is optimal under high saturation powers, for instance B₁≥ 3 μT, given its relatively fast amine proton exchange rate.²⁴ Therefore, k_ex as a general physical parameter is the weighted average of all exchanging proton species in tissues.

As a physical parameter, tissue k_ex may be altered by pH, temperature, reactive oxygen species (ROS), or the profile of tissue metabolites containing slow or fast exchangeable protons.²⁷ A previous study using DS-removed omega plots has shown that exchange rate maps can linearly reflect the pH changes in the physiological range (pH 6.2 to 7.4).²¹ PH in MS lesions in vivo with the established ³¹P-MR spectroscopy has been consistently reported at neutral to slightly alkaline (7.03 ± 0.02).²⁸ Based on our research with protein phantoms at different pH, 0.03 unit of pH will lead to k_ex change of only 4 s⁻¹ which is <1%.²² There is literature reporting acidic pH in MS lesions as well.²⁹ However, pH reduction leads to a k_ex decrease, while ROS leads to k_ex enhancement. The fact that we observed positive k_ex contrast in MS lesions indicates that the pH influence is not comparable. As another factor that could affect k_ex, the temperature is well controlled in the brain. There are prior MR spectroscopy studies on metabolite changes in MS lesions.^30,31 While glutamate concentration has been found to be elevated in acute lesions and NAWM, N-acetyl-aspartate was significantly reduced, likely due to neuronal/axonal dysfunction or loss.^30,31 Elevated choline was observed that is likely linked to heightened cell-membrane turnover, as seen in demyelination, remyelination, inflammation, or gliosis.³¹ Evidence was also found for increased glial activity in MS, with a significantly higher myo-inositol concentration.³⁰ However, the aforementioned metabolite changes are unlikely to influence the observed tissue k_ex in MS lesions, given that these changes are at the mM level, a small fraction of the total exchangeable protons (including amide, amine, hydroxyl, and hydrosulfuric protons) whose concentration is estimated to be hundreds to thousands of mM.³⁰

A recent study has shown that ROS can enhance tissue k_ex with high sensitivity even at the pM level.²⁷ Due to the extremely short ROS half-lives and very low in vivo concentrations, in vivo ROS imaging is challenging.³² Several techniques have been developed for studying tissue ROS, including biochemical analyses, optical redox scanning, and methods such as electron paramagnetic resonance (EPR) and Overhauser-enhanced MRI (OMRI).^32,33 However, these techniques are either invasive, limited to ex vivo applications, or require exogenous contrast agents. The administration of exogenous contrast agents prevents repeated or longitudinal studies, as the agents react with radicals in the tissue, thereby changing the tissue ROS concentration under investigation. On the other hand, k_ex MRI has recently been developed to image in vivo ROS with advantages due to its endogenous contrast and high sensitivity and specificity, as confirmed by in vivo and ex vivo experiments.^27,33,34 Given the high reactivity of ROS (particularly hydroxyl radicals, ˙OH), ROS promotes proton exchange or elevated k_ex, presumably through an oxidation-catalyzed mechanism, similar to, but likely more effective than, the base(OH⁻)-catalyzed mechanism due to pH.²⁷

Oxidative stress is implicated as one of the key factors in the pathogenesis of MS, given that evidence for oxidative stress is found in the experimental autoimmune encephalomyelitis, pathologic specimens, and autopsied tissues from MS patients, and antioxidant therapy is reported to be a promising approach to treat MS.³⁵ Activated macrophages and microglia and mitochondrial dysfunction are the main sources for oxidative stress in MS lesions.¹⁵ Elevated intracellular ROS due to oxidative stress causes the inflammatory destruction of myelin, as myelin sheaths are highly vulnerable to oxidative stress.³⁶

Differing from Gd enhancement of an MS lesion, which represents the disruption of the blood–brain barrier, elevated ROS production and inflammation is reported to be present in active and slowly expanding MS lesions, but may be absent in inactive MS lesions.^35,37 Therefore, detection and visualization of ROS in vivo is essential for studying oxidative stress in MS pathogenesis. The results of this study have shown that nonenhanced MS lesions could be further classified into two patterns by k_ex mapping, suggesting that k_ex may be an alternative endogenous MRI contrast for separating slowly expanding MS lesions (chronic active lesions) from inactive ones. Recently, other imaging tracers and approaches based on specific pathology were also proposed to separate the nonenhanced MS lesions, such as positron emission tomography (PET) with 11C-(R)-PK11195, an isoquinoline carboxamide that binds selectively to the peripheral benzodiazepine receptor (targeting activated microglia or macrophages), and SWI (targeting abnormal iron deposit).^15,38,39 With k_ex MRI (potentially targeting ROS), the stage of Gd-negative lesions may be further characterized for the better determination of the DIT of MS lesions, encouraging further validation by comparison with other imaging approaches and studies in preclinical MS models.

Limitations

This study focused on the implementation of the novel k_ex MRI for MS patients; however, its clinical implication requires further validation with pathological studies in animal models. Second, this preliminary study of novel k_ex MRI was performed on a small number of MS patients from a single hospital, which needs to be further validated by multicenter studies in a large cohort. Third, the long scan time and only one slice scanning for k_ex mapping may be of concern. The former can be significantly shortened for future clinical applications by reducing the number of Z-spectra to two, given that the omega plot is a linear plot, and by reducing the number of frequency offsets in each Z-spectrum. Fourth, choosing NAWM as an internal reference might underestimate the k_ex elevation in MS lesions. Because MS is considered to be a disease involving the whole brain, NAWM may also have the risk of k_ex elevation.

Conclusion

This study demonstrated the implementation of k_ex MRI in the MS brain. K_ex MRI showed consistent enhancement in Gd-enhancing MS lesions. In addition, k_ex MRI may further differentiate Gd-negative lesions based on their inflammatory status. Once validated, k_ex MRI may serve as a complementary imaging tool for the characterization of MS lesions in addition to Gd-enhanced MRI.