Development of Myelin Growth Charts of the White Matter Using T1 Relaxometry

DISCUSSION

MP2RAGE is a clinically feasible scan with accelerated imaging and can be acquired in as little as 6 minutes.⁷ Using the quantitative T1 mapping function of MP2RAGE and converting to R1 values, we were able to demonstrate the expected and significant increase in R1 values with normal myelin maturation and aging. Summarized white matter and gyral data as well as specific white matter anatomy show high correlation and reproducibility.

When using an R1 threshold, we were able to recreate the well-recognized pattern and timing of increased T1 signal with myelin maturation in the first year of life. Increasing T1 signal generally follows an inferior-to-superior, central-to-peripheral, and posterior-to-anterior pattern.^3,7 Furthermore, there is agreement with the histologic study by Yakovlev and Lecours,⁸ demonstrating earlier myelin maturation of the motor fibers of the cortical spinal tract and the associated anatomy: posterior limb of the internal capsule, cerebral peduncle, and pons, as well as the inferior cerebellar peduncles before the more superior brainstem and corpus callosal components.

In a recently published study, Kühne et al⁷ also used MP2RAGE to generate R1 values in white and gray matter regions of the brain in infants and young children (3 months to 6 years of age) demonstrating an expected increase in R1 values with myelin maturation. However, our work has important differences in that we included 9 patients younger than 3 months of age, in whom we can assess the earliest changes in myelination, as well as our ROIs were generated through segmentation from an atlas rather than a subjective approach of manually selecting ovoid ROIs in limited anatomy. This advantage allows a larger selection of brain anatomic structures to be assessed in a less subjective fashion. Furthermore, in our chart presentation for selected white matter structures, we followed the conventional head growth charts in graphing cohort percentile lines to compare more easily with peers and increase sensitivity to changes with time.

There are multiple MR imaging parameters that correlate with myelination.⁹ The lipid and protein macromolecules in the myelin sheath affect T1 and T2 relaxometry^10,11 and the magnetization transfer ratio.¹² Increasing myelination decreases the diffusion of water as measured by diffusion imaging, with resulting parameters of ADC and with diffusion tractography, fractional anisotropy, mean diffusivity, and radial diffusivity.^13,14 Furthermore, myelin also has diamagnetic properties that can be measured by quantitative susceptibility imaging.¹⁵ The T1/T2-weighted ratio has also shown correlation with myelin maturation,¹⁶ and while less direct than the previously mentioned quantitative parameters, it is more easily obtained from standard clinical MR imaging sequences. However, a recent histologic validation study cautions against using T1/T2-weighted ratios to map myelin in the white matter due to discordance compared with the histologic staining of myelin.¹⁷ T1 relaxometry is recognized as having the best subjective visualization for myelin maturation in the first year of life and may be more sensitive in this age range. Compared with diffusion and quantitative susceptibility mapping, MP2RAGE is less susceptible to T2* artifacts and has higher resolution. Previous studies using T1 mapping for myelin quantification do not focus on infants between term and 2 years of age^4,18 and, compared with Kühne et al,⁷ the first 2 months of life.

Indirect measurements of head circumference using cross-sectional imaging have shown high concordance with direct head circumference measurements.¹⁹ Similar to the detection of macrocephaly or microcephaly with a direct head circumference growth chart, a quantitiatve myelination growth chart segmented for specific white matter anatomy may be useful to detect disease involving myelination, particularly for nonlesional hereditary disorders of primary myelination such as Pelizaeus-Merzbacher disease. MR imaging findings in this category of disease are typically symmetric and diffuse and may be difficult to visually detect without sufficient experience in pediatric neuroimaging. Further direction for clinical validation would be to test these data against patients with known primary demyelinating disorders for diagnostic accuracy, using the MP2RAGE technique.

Additionally, using the t test with correction for multiple comparisons, we found significant differences only between sexes for the posterior thalamic radiation, sagittal stratum, and middle temporal pole, with girls demonstrating significantly higher mean R1 values than boys. This is a novel finding because there are no other studies comparing R1 values in white matter between sexes, and indicates faster maturity of girls compared with boys. This finding is in agreement with a study by Wang et al,²⁰ who used diffusion tractography to evaluate adolescent white matter and found that boys had greater longitudinal fractional anisotropy changes than girls, indicating that girls had reached maturity earlier than boys.

Limitations include the retrospective and single-center nature of our study with MR images from subjects who may not be completely healthy. Scanning a similar number of healthy asymptomatic subjects in this age range would be challenging from an ethical and resource use perspective. To ameliorate this limitation, we included only MR imaging scans identified as having normal findings by a pediatric neuroradiologist and excluded known white matter pathology from available electronic medical record data. Similarly, longitudinal data with the same subjects would avoid intersubject variability but would also have limited feasibility due to repeat MR imaging scans that may not be clinically necessary in a retrospective study.

Furthermore, MR images with clinically normal findings may include a variable amount of dilated perivascular spaces, which can affect both gray and white matter. The main disease that presents with enlarged and a greater number of dilated perivascular spaces is mucopolysaccharidosis, which we attempted to exclude from available data in the electronic medical record. Without an objective exclusion criterion, we did not attempt to subjectively exclude presumably normal perivascular spaces, and the possible effect on myelin quantification should be noted.

Finally, most of our subjects required the use of anesthesia for MR imaging acquisition, a standard practice in pediatric imaging. Similar to other institutions, MR imaging for neonates at our institution is generally attempted with timing of the patient’s sleep and gentle immobilization techniques. In neonatal rodent models, anesthesia disrupts oligodendrocyte maturation. Limited human imaging studies demonstrate decreases in white matter volume and fractional anisotropy in infants with previous exposure to surgery and anesthesia.²¹ While there are confounding factors from pre-existing disease, the possibility of white matter alteration from these rodent models may be a limitation in our study. However, the timing of white matter alterations from anesthesia effects may be more delayed than the data obtained during concurrent imaging.