Quantitative Imaging of ATP Production Rates and their Functional Changes in Healthy Human Brain
Xiao-Hong Zhu1, Byeong-Yeul Lee1, and Wei Chen1

1CMRR, Radiology Department, University of Minnesota, Minneapolis, MN, United States


We established a practical protocol for quantitatively imaging the cerebral metabolic rates of ATP production via ATPase and creatine kinase (CK) reactions in human brain at 7T using three dimensional (3D) chemical shift imaging (CSI) and in vivo 31P MR spectroscopy (MRS) in combine with magnetization transfer (MT) approach. Subsequently, we applied this 3D 31P-MT imaging protocol to quantify the regional phosphorous metabolites concentrations, ATPase and CK reaction rate constants and fluxes and the intracellular pH in human brain at rest and during visual stimulation. The results of this study provide the values of key parameters relevant to ATP metabolism in absolute scale, which allow quantitative evaluation of regional cerebral energetics in resting human brain and its functional changes.


To establish an imaging protocol for quantifying phosphorous metabolites and ATP production rates in human brain at 7T; and to determine regional ATP metabolic rates and functional neuroenergetic changes in healthy human.


Cerebral energy metabolism of ATP provides chemical energy for neuronal activities and brain functions. In vivo 31P MRS with magnetization transfer (MT) is the only method capable of directly and non-invasively assessing the cerebral metabolic rate of ATP production (CMRATP) via ATPase and creatine kinase (CK) reactions1-4. However, the applicability of this technique for quantifying regional CMRATP has not been fully demonstrated in human brain. We report here a study protocol for imaging key phosphorous metabolites and ATP metabolic rates in human brain at 7T. It has been applied to determine energetic parameters in resting and stimulated brain and to evaluate their regional changes in response to visual stimulation.


Healthy young subjects were recruited. All measurements were conducted at 7Tesla/90cm Siemens scanner using a 1H/31P surface coil probe placed over the occipital lobe. A small phosphorus reference sphere was fixed at center of the 31P coil for power calibration. T1-weighted anatomic image, fMRI data (stimulation paradigm: 8Hz flashing checkerboard with a central fixation point) and 3D-CSI (FOV=12×12×9cm, matrix=7×7×5, TR=1.2s, total NT=896 and flip angle (FA)=48°) with and without γ-ATP saturation5 at resting and stimulated conditions were acquired. After each human scan, 3D 31P-CSI data was obtained on an ATP phantom ([ATP]=10mM) with identical imaging resolution, coil loading and sample position. The experimental setup is shown in Fig. 1. The 31P RF B1 field and FA map were also obtained. The fMRI and the 31P-CSI data were analyzed with SPM and jMRUI software, respectively. The AMARES algorithm6 was used in spectra fitting; the integrals of phosphorous metabolites were corrected for the saturation effects based on relevant T1 and FA information. The ATP content of each subject was determined by comparing the brain signals in selected brain region to that of ATP phantom (see Fig.1); which was then used as an internal standard for determining the concentration of other phosphorous metabolites including PCr and Pi. The forward reaction rate constant of ATPase (kf_ATP) and CK (kf_CK) were determined according to: Mc/Ms ≈1+kf×T1nom where Mc and Ms are control and g-ATP saturated magnetization and T1nom is the nominal T1 for Pi or PCr7. The metabolic rate of ATP production via ATPase (Flux_ATP) or CK (Flux_CK) reactions were determined by multiplying the kf with [Pi] or [PCr], respectively.


Fig. 2 and 3 display typical 31P-MT spectra and fMRI maps of the human brain obtained in this study, showing excellent spectral quality and robust functional activation. The results of the neuroenergetic measurement in multiple subjects/ROIs are summarized in Table 1. The brain ATP levels in resting human brain were similar with [ATP] = 2.96 ± 0.03 mM between all measurements. The regional reaction rate constants and fluxes within the primary visual cortex are similar to previous literature reports1-4,8. We observed stimulation induced increases of [Pi] (10±8%) and ATPase reaction rate constant (24±11%) and flux (37±8%); a small reduction of [PCr] (-0.5%) and moderate increases (~5%) of CK reaction rate constant and flux were detected in activated brain region. However, including surrounding CSI voxels into the ROIs led to substantially less functional changes in [Pi] and ATPase reaction rate and flux, which were similar to our previous dynamic functional study without using the 3D-CSI approach8. Interestingly, a higher intracellular pH in activated brain region during visual stimulation was reliably detected in all subjects at 7T, despite the fact of elevated cellular lactate in the visual cortex9, suggesting a potential role of pH in brain function and activation.


We established a quantitatively imaging protocol for key phosphorous metabolites and ATP production rates in human brain. The energetic parameters were determined and reported in absolute scale, which is essential for investigating the baseline brain energetics under various pathophysiological conditions. The functional study performed in this work provides a good example that demonstrated the applicability of the 3D 31P-MT imaging technique for evaluating regional energetic changes evoked by functional stimulation. A significant amount of ATP energy consumed during visual stimulation in supporting evoked neuronal activity in the human visual cortex. A 37% increase of ATPase flux detected during visual stimulation presents an increment of ATP expenditure (~1.7 billion more ATPs/s/neuron) from the resting state10. The methodology described herein should provide a practical tool that can be applied in various studies for understanding the neuroenergetics in human health and diseases.


NIH grants of R24 MH106049, R01 NS070839, S10 RR029672, P41 EB015894 and P30 NS076408; and Keck Foundation.


[1] Lei et al., PNAS; 100:14409-14414 (2003); [2] Du et al., PNAS; 105:6409-6414 (2008); [3] Du et al., MRM; 57:103-14 (2007); [4] Chen et al., MRM; 38:551-7 (1997); [5] de Graaf et al., NMR Biomed; 9:185-194 (1996); [6] Vanhamme et al, JMR; 129:35-43 (1997); [7] Xiong et al., Circ. Res; 108: 653-663 (2011); [8] Lee et al., Proc. ISMRM; 22:12 (2014); [9] Prichard et al., PNAS; 88:5829-5831 (1991); [10] Zhu et al., NeuroImage; 20: 2017-2027 (2012).


Experimental setup for quantitative imaging of ATP contents and metabolic rates in human brain using a phantom containing known concentration of ATP.

Localized in vivo 31P MR spectra (ROI shown in Fig.1) with (bottom) and without (top) MT obtained in human brain under resting (left) and visual stimulated (right) conditions.

Representative fMRI map of a subject with marked ROIs displayed on the image.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)