Development and Application of Quantitative Gradient-Echo MRI Techniques for Assessment of Iron and Myelin in the Multiple Sclerosis Brain

Ahmed Elkady
2018
Multiple Sclerosis (MS) is a progressive autoimmune and neurodegenerative disease of the Central Nervous System that results in demyelination and neuronal/axonal loss. Along with clinical evaluation, Magnetic Resonance Imaging (MRI) is the gold standard modality for diagnosing and monitoring MS, which has been conventionally carried out using T1 and T2weighted pulse sequences. Gradient-echo T2*-weighted has been recently demonstrated as a sensitive predictor of MS disability and cognitive
more » ... e; however, more research is needed to implement gradient-echo based techniques into clinical practice and drug trials. The purpose of this thesis was to develop technical MRI innovations and study research applications to further the field of quantitative gradient-echo MRI of MS. Although gradient-echo acquisitions are most sensitive to iron, myelin also contributes to image contrast. The main clinical hypothesis was that quantitative gradient-echo MRI methods may serve as a biomarker of disease progression by assessing iron and myelin in the MS brain. This hypothesis was tested in one cross-sectional study and two longitudinal studies of MS subjects compared to age-matched controls, where we related MRI findings to clinical outcomes. Since lesions are the most commonly indicator in MS clinical monitoring, we have explored their shape and conspicuity in Quantitative Susceptibility (QS) and Local Field Shift (LFS) maps that have been quantified from multi-echo gradient echo acuistions in Chapter 2 using postmortem and in vivo MS subjects. We have demonstrated that the use of LFS and QS contrast for lesion iron detection is sensitive, but not sufficiently specific, while the use of lesion dipole signature is specific but not sufficiently sensitive. iii Although QS mapping is widely used in MS studies, its adoption is still lagging in clinical practice and trials. We investigated the sensitivity of QS mapping to field of view (FOV) reduction in Chapter 3, with the aim of determining whether restricted-FOV acquisition would affect QS accuracy. We have demonstrated through theoretical simulations and in vivo data that QSM is sensitive to FOV reduction, which was most sensitive in the direction of the static magnetic field. Increased R2* and QS values in MS patients compared to controls in cross-sectional studies has been previously interpreted as increased iron accumulation in MS DGM. However, previous histochemical evaluations have indicated that the MS DGM also suffers from demyelinating lesions. We have developed in Chapter 4 an analysis technique using combined R2* and QS and sparse logistic regression to discriminate iron accumulation from demyelination MS crosssectional studies. We have demonstrated the increased statistical power of the developed technique compared to conventional singular R2*/QS analysis. We have also demonstrated that iron accumulation occurs progressively with advanced MS phenotypes. In Chapter 5 we present a novel longitudinal analysis technique for Discriminative Analysis of Regional Evolution (DARE) of brain iron/myelin changes in MS. Application of DARE to 2year Relapsing-Remitting MS (RRMS) and Progressive (PMS) data compared to age-matched controls revealed that iron changes are prominent over 2 years in RRMS, whereas myelin changes are prominent in PMS compared to age-matched controls. Similar to the developed sparse logistic regression technique, I have demonstrated the increased statistical power of DARE compared to singular use of R2* and QS. Iron increase in identified DARE regions demonstrated the highest correlation with disease severity (r=0.68; Q=0.0005) from amongst all studied parameters. iv No longitudinal evaluation of the MS DGM over 5-years has been previously conducted. In Chapter 6, we analyze 5-year longitudinal changes in RRMS DGM compared to controls using DARE and conventional bulk structure analysis, and correlate the identified significant results to clinical measures. We demonstrate that iron decrease and myelin increase is the most DGM prominent change in RRMS compared to controls, which is somewhat similar to the behavior of 2-year PMS DGM data. Using Pearson's regression, we have also demonstrated that iron decrease in the caudate nucleus is linearly related with disease severity (r = 0.64; Q = 0.03), and myelin increase in the substantia nigra is linearly related with disease duration (r = 0.49; Q = 0.008). In conclusion, this thesis has developed and applied novel analysis techniques using gradientecho acquisitions, which has provided important insight about pathological changes of iron and myelin in the MS brain. v Preface Written informed consent was obtained from all patient and control subjects included in this thesis after the internal Institutional Review Board approved the study design. For all the published and unpublished chapters in this thesis, Dr. Alan Wilman (supervisor) shared unique ideas, helpful discussion and careful editing of the manuscripts. In all cases first drafts were written by the author. The second chapter has been presented as an oral presentation: Elkady AM, Sun H, Walsh AJ, Blevins G, Zhuozi D and Wilman AH. Quantitative Susceptibility Mapping of Lesions in Multiple Sclerosis. Abstract #0281. In: ISMRM 23rd Annual Meeting & Exhibition, Toronto, Canada, 2015. All analyses and compilation of the abstract and presentation were implemented by the author. Dai Zhuozi verified lesions identified by the author. The postmortem patients studied came from the practice of Dr. Ken Warren with Ingrid Catz. The photographs and their processing were done by Andrew Walsh. The preparation of brain slices were performed by the neuropathologists Drs. Johnson and Lu.
doi:10.7939/r30k26t2p fatcat:ehgz2xj4vbbshmymt6oia5i2vq