Establishing baseline MRI biomarkers for normal brain aging is significant and valuable for separating normal aging effects from neuro-pathological effects on the brain structure and functions. In this study, we use quantitative measurements of tissue specific transverse relaxation properties of gradient recalled echo (GRE) MRI signal to obtain quantitative in vivo information on the evolution of grey matter (GM) tissue structural and functional properties throughout adult human lifespan. We use previously developed approach^1,2,3 to separate tissue-specific R2*_t and BOLD contributions to GRE signal, thus providing information on tissue cellular and functional hemodynamic properties, such as, relative oxygen extraction fraction (OEF_rel), deoxygenated cerebral blood volume (dCBV) and tissue concentration of deoxyhemoglobin (C_deoxy). Our results establish age-related baselines of these parameters that could serve as a basis to study neurodegenerative diseases. In addition, the biological interpretation of R2*_t is discussed compared with the literature data. As R2*_t is not only related to the GM cellular content, but also the cellular arrangements⁴, we unite both of them in the unit of neuron and provide support for a new hypothesis that R2*_t could serve as a biomarker of the cortical “cellular packing density” (CPD), which mostly reflects the neuronal density in this study.

The study was approved by local IRB. Data were obtained from 20 healthy subjects ages 22 to 74 using a 3T Trio MRI scanner (Siemens, Erlangen, Germany) with a 12-channel coil and 3D GRE sequence with 11 echoes, TE1=4ms, ΔTE=4ms, and a navigator echo to correct for physiological fluctuations⁵. Other parameters were: TR=50ms, FA=30°, resolution 1×1×2mm³. Image processing and statistical analysis were done in MATLAB (The MathWorks, Inc.). Effects of background gradients were corrected using VSF approach⁶. T1 weighted images and maps of R2*_t and hemodynamic parameters were generated using algorithms described in previous publications^2,3,7. MPRAGE images with resolution of 1×1×1mm³ were used to generate segmentations and determine cortical thickness (Th) by FreeSurfer⁸. 26 regions of interest (ROIs) from FreeSurfer were registered onto the naturally co-registered T1w images, R2*_t and hemodynamic parameter maps. The median value of each parameter in each ROI across 20 subjects was linearly correlated with age using Eq.[1]:

$$parameter = a + k × (age - 40)$$

Age 40 years was selected as the adult reference for convenience.

The data show that with increasing age, R2*_t significantly increases in most cortical ROIs whereas cortical thickness significantly decreases. However, the product of R2*_t and cortical thickness (SR2*_t), representing the CPD underneath a unit surface, remains constant over the adult life (Figure 1). Since the iron in the cortex remains practically constant after age 30,⁹ the increased R2*_t can be attributed to the increase of the cortical CPD. Indeed, studies have shown that the number of neurons in the neocortex remains unchanged over the adult life.^10,11,12,13 Hence, with shrinking cortical thickness, the CPD should increase which is consistent with our measurements of increased R2*_t and the conservation of SR2*_t.

The distribution of R2*_t corresponding to an averaged 40 years old subject (parameter a in Eq.[1]) on the human brain surface is shown in Figure 2, where the highest R2*_t was found in visual cortex (cuneus, lateral occipital and lingual) and the lowest was found in the anterior cingulate and prefrontal cortex (e.g. superior, rostral-middle frontal). Relatively higher R2*_t was found in primary sensory and motor areas (paracentral, precentral and postcentral). This pattern is similar to the distribution of neuronal density measured in non-human primates by Collins et al.¹⁴ who found highest and second highest neuronal density in visual cortex among all of the examined primates, lowest in prefrontal and premotor cortex in different primates and higher than average in primary sensory cortex. This similarity supports the hypothesis that R2*_t is related to the cortical CPD and mostly reflects the neuronal density.

Additionally, the data show that the tissue hemodynamic parameters, i.e. OEF_rel, dCBV and C_deoxy remain practically constant with age in most cortical regions (Figure 1), consistent with the PET measurements^15,16,17. We also found interesting correlations characterizing relationships between brain structural and hemodynamic properties in different brain regions. Specifically, thicker cortical regions have lower R2*_t, reflecting less CPD, and these regions extract less oxygen from the blood (Figure 3). The possible explanation would be that regions with lower R2*_t (lower neuronal density) show elevated aerobic glycolysis and require more CBF than CMRO₂ resulting in lower OEF¹⁸. This is consistent with the knowledge that regions with lower neuronal density have more complex dendritic and synaptic structure^19,20 and the high aerobic glycolysis is likely needed to support high synaptic activities¹⁸.