3303

Quantitative imaging biomarkers of demyelination and remyelination: reproducibility of MTsat vs. MTR.
Beth York1,2, Michael J. Thrippleton1,2, Rozanna Meijboom1,2,3, and Adam Waldman1,2,4

1Centre for Clinical Brain Sciences, University of Edinburgh, Edinburgh, United Kingdom, 2Edinburgh Imaging, Edinburgh, United Kingdom, 3UK Dementia Research Institute, University of Edinburgh, Edinburgh, United Kingdom, 4Imperial College London, London, United Kingdom

Synopsis

MTsat provides magnetization transfer information less confounded by B1 and T1 inhomogeneities than MTR (largely derived from bound protons within axonal myelin in the brain). The present study measured the reproducibility of MTsat versus MTR, and compared tissue-type contrast for MTsat and MTR in healthy volunteers and patients with relapsing-remitting multiple sclerosis. Quantitative maps were created for histogram analysis, descriptive statistics and sensitivity comparisons. MTsat and MTR showed similar reproducibility but MTsat showed higher tissue contrast. Our data suggest MTsat is a superior biomarker for myelin integrity, with utility for the study of demyelination and remyelinating therapies in multiple sclerosis.

Introduction

Magnetization transfer (MT) imaging allows quantitative imaging of protons bound to macromolecules within axonal myelin, and hence provides an indirect in vivo measure of myelin integrity. MT ratio (MTR) imaging has been applied in a number of studies of multiple sclerosis (MS), to detect white matter myelin destruction, a pathological hallmark of the disease. MTR signal is however influenced by other variables and MTsat, which corrects for B1 inhomogeneities and T1 relaxation, has been proposed as a more reliable measure of myelin content1.

Previous research suggests MTsat is sensitive to MS pathology2 and may thus have potential as a biomarker for demyelination and remyelination. However, no studies have yet investigated the test-retest variance of MTsat in comparison to MTR. The aim of this study was to determine the reproducibility of MTsat compared with MTR in healthy volunteers. We additionally investigated the performance of MTsat and MTR for tissue-type and lesion delineation in patients with MS.

Methods

Healthy volunteers (n=12, mean age = 45±10yrs) were imaged twice with the same protocol (including structural and MT acquisition) one week apart. At both time points, processing steps included: (1) segmentation of the T1-weighted structural scan (whole brain, grey matter, white matter); (2) co-registration of these tissue masks to the MT-off proton density image; (3) creation of MTR and MTsat maps and, (4) calculation of descriptive statistics (mean, standard deviation, coefficient of variation) per tissue type for each metric. MTR and MTsat histograms for individual subjects were similarly compared across the two time-points.

The reproducibility of MTR and MTsat across time-points was visually assessed with Bland-Altman plots3. One-way random effects, absolute agreement intra-class correlation co-efficients (ICCs) per tissue-type across time-points were calculated for MTsat and MTR mean values4.

Patients with relapsing-remitting MS (RRMS, n=18, mean age=39±6yrs) were imaged at a single time-point using a similar protocol for comparison. Processing steps were identical to healthy volunteers with additional manual segmentation of white matter into normal-appearing white matter (NAWM) and lesion masks.

Mean MTsat and MTR values were calculated from patients’ masked quantitative maps for grey matter, NAWM and lesions. The differences between (a) NAWM and lesions and between (b) NAWM and grey matter mean values were then calculated for each individual.The relative NAWM to grey matter difference was calculated as %Δ=(MTNAWM-MTGM)/MTNAWM x 100.

As an indicator of the ability of each technique to detect longitudinal change, the average MTNAWM to MTlesions difference in patients was displayed on the normalised white matter Bland-Altman plot (Figure 1, right). MT values were normalised to account for scaling differences.

Results

Descriptive statistics (Table 1) are reported for healthy volunteers. Bland-Altman plots (Figure 1) across the two time-points show similar test-retest error for MTsat and MTR. All tissue-type mean differences across the two time-points did not significantly differ from zero. Histograms of MTsat values illustrated the greater separation of grey matter and white matter compared to MTR in both patients and healthy volunteers (Figure 2).

Similarly, intra-class correlation co-efficients (Table 2) for MTsat and MTR demonstrate high reproducibility for both techniques. The ICC for white matter was marginally higher for MTsat compared to MTR. However, ICCs in grey matter and white matter minus grey matter were slightly higher for MTR than MTsat.

For patients with RRMS, descriptive statistics are reported (Table 3) and the relative NAWM-to-grey matter difference, %Δ, was higher for MTsat (40.28) than MTR (16.98). In the normalised Bland-Altman plot (Figure 1, right), MTsat shows a wider interval for normalised NAWM-to-lesion difference (-10.18, 9.46) compared to MTR (-8.19, 8.6).

Discussion

Test-retest reproducibility in healthy volunteers for MTsat and MTR is similar although MTsat may slightly outperform MTR in white matter. Reproducibility of a technique is a key determinant of sensitivity to biological change in longitudinal studies. However, the improved white-to-grey matter contrast and distribution of individual subject values suggests that MTsat is more sensitive than MTR to myelination in the brain.

Descriptive statistics and Bland-Altman comparison for patients with RRMS suggest that, while individual tissue-type variation is not substantially different between MTsat and MTR, the former delineates healthy versus pathological brain tissue to a greater extent than MTR. Furthermore, relative tissue-type difference calculations show improved sensitivity for MTsat compared to MTR. This is likely due to reduced variation from B1 inhomogeneities and T1 relaxation.

Conclusion

Reproducibility and tissue contrast in healthy subjects and patients with RRMS suggests advantages of MTsat compared with MTR as a biomarker of demyelination; MTsat may therefore provide a more reliable platform for stratifying MS patients, and validating putative remyelinating therapies in future multi-centre clinical trials.

Acknowledgements

Thank you to:

Volunteers and patients who participated in the present study;

Radiographers at the Edinburgh Imaging Facility;

Francesca Chappell for her helpful statistical advice;

SPRINT-MS/MND PhD funding provided by CSO Scotland;

Cambridge University Hospitals NHS Foundation Trust & the University of Cambridge.

References

1. Helms G, Dathe H, Kallenberg K, Dechent P. High-resolution maps of magnetization transfer with inherent correction for RF inhomogeneity and T1 relaxation obtained from 3D FLASH MRI. Magn Reson Med. 2008;60(6):1396-1407. doi:10.1002/mrm.21732.

2. Lema A, Bishop C, Malik O, et al. A Comparison of Magnetization Transfer Methods to Assess Brain and Cervical Cord Microstructure in Multiple Sclerosis. J Neuroimaging. 2017;27(2):221-226. doi:10.1111/jon.12377.

3. Bland MJ, Altman D. Statistical Methods for Assessing Agreement Between Two Methods of Clinical Measurement. Lancet. 1986;327(8476):307-310. doi:10.1016/S0140-6736(86)90837-8.

4. Shrout PE, Fleiss JL. Intraclass correlations: Uses in assessing rater reliability. Psychol Bull. 1979;86(2):420-428. doi:10.1037/0033-2909.86.2.420.

Figures

Figure 1: MTsat and MTR Bland-Altman plots for healthy volunteers (n=12) showing mean grey matter, mean white matter, mean white matter minus mean grey matter, and normalised white matter across two time-points, one week apart3 . Dashed lines show mean difference ±1.96 SDs. Dotted lines show .95 confidence intervals. As an indicator of the ability of each technique to detect longitudinal change, red solid lines (right plots) show (MTNAWM minus MTlesions for patients) / standard deviation of ΔMTsat in white matter for healthy volunteers.

Figure 2: Example MTsat and MTR histograms for a single, healthy subject at one time-point. Values in whole brain, white matter, grey matter and CSF are shown in separate colours.

Table 1: Descriptive statistics derived from quantitative MTsat and MTR maps for healthy volunteers (n=12) scanned at two time-points (TP1, TP2) one week apart. MT: mean (SD) value within each tissue-type averaged across both time-points for all subjects; CV%: inter-subject co-efficient of variation; MTΔ: mean (SD) difference in mean value at TP1 minus TP2 . Standard deviations are reported in brackets.

Table 2: Intra-class correlation co-efficients (one-way, random effects, absolute agreement) for healthy volunteers (n=12) in grey matter, white matter, and mean white matter minus mean grey matter.

Table 3: Descriptive statistics for MTsat and MTR for patients with relapsing-remitting multiple sclerosis (n=18) scanned at one time-point (TP1). These are derived from mean values within tissue-type masked quantitative maps. NAWM: normal-appearing white matter (i.e. white matter excluding lesions); MT: mean value within each tissue-type at TP1 for all subjects; CV%: inter-subject coefficient of variation. Standard deviations are shown in brackets.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
3303