Introduction

Hyperpolarized-¹³C MRI (HP-¹³C MRI) is a minimally invasive technique that enables imaging of ¹³C-pyruvate and labelled downstream products such as ¹³C-lactate and ¹³C-bicarbonate. A prior study using HP-¹³C MR found increased ¹³C-bicarbonate signal in the occipital lobe during functional activation, attributed to increased pyruvate oxidation¹. In this study, whole-brain HP-¹³C MR was performed to compare regional ¹³C-metabolites between visual stimulus conditions.

Methods

Cognitively normal volunteers (N=6) were recruited under a protocol approved by the Sunnybrook Research Institute Research Ethics Board and Health Canada. Participants' age ranged from 24 to 33, and were all screened for cognitive impairment using the Montreal Cognitive Assessment².

'Task' and 'control' HP-¹³C MR scans were acquired for every participant using a previously described 3D-echo-planar sequence^3,4. The task scan involved the participant viewing a 8Hz flashing checkerboard stimulus during the HP-¹³C MR acquisition. During the control scan the participant had their eyes closed. There was a half-hour wait period between the two HP-¹³C scans. During this interval, anatomical and BOLD fMRI images were acquired. The ordering of the task and control HP-¹³C scans was swapped for three participants to control for scan order effects. HP-¹³C MR images were reconstructed in MATLAB (The MathWorks Inc., Natick, MA). A block-design was used for the BOLD scans with the same minute-long block duration as a single HP-¹³C MR task acquisition (Fig. 1). A set of example control and task ¹³C-lactate scans are shown in Fig. 2.

For image analysis, anatomical T1w images were parcellated using an automatic parcellation method⁵, and regions with volume below the 1.5cm³ HP-¹³C voxel resolution were excluded, resulting in 89 brain regions. The parcellation map for each participant was used to compute regional ¹³C-pyruvate, ¹³C-lactate, and ¹³C-bicarbonate. Regional ¹³C-metabolite signals were normalized by the ¹³C-metabolite signal from the brainstem to remove inter-scan ¹³C polarization variability. The brainstem normalized ¹³C-metabolite signal from the task scan was divided by the brainstem normalized control scan ¹³C-metabolite signal, resulting in a single set of regional ratios of ¹³C-metabolite signal for each participant (referred to as task/control signal ratio below). Task/control signal ratio were grouped into occipital and non-occipital lobe regions, as the occipital lobe contains visual processing areas. A one-tailed Datta-Sutten clustered Wilcoxon rank-sum test was run to test for a difference in metabolism between occipital lobe regions relative to non-occipital lobe regions⁶. The percentage change between normalized ¹³C-metabolite task and control scans were calculated for each participant, grouped by non-occipital lobe and occipital lobe regions (Fig. 3). Region-specific task/control signal ratios were plotted for all assessed brain regions, with color coding indicating occipital lobe and non-occipital lobe regions in Fig. 4.

The fMRI scan parameters were as follows: 3.5 mm isotropic resolution, 2000 ms repetition time, 29 ms echo time, 40^o flip angle, 64 by 64 matrix size, 38 slices. BOLD image analysis was performed using the AFNI toolbox^7,8, and involved removal of the first two TRs of each acquisition, registering across time to reduce motion artifacts, and applying a Gaussian blur function. BOLD images were regressed against the task paradigm and voxelwise t-statistics were calculated. BOLD activation was consistently achieved in the occipital lobe in all six participants with the designed task block design. A representative t-statistic map was created by registering the BOLD images of the six volunteers to the MNI152 brain standard⁹, and summing the final results (Fig. 5).

Results and Discussion

A significant difference in the ¹³C-lactate task/control signal ratios between occipital lobe regions and non-visual regions was found (p = 0.04, Z= -1.8). Neither the ¹³C-bicarbonate nor ¹³C-pyruvate signal ratios showed a difference (p = 0.98, Z = 2.0 and p = 0.19, Z = -0.9 respectively). Occipital lobe regions are consistently increased in the task ¹³C-lactate scans for five out of six participants. Small increases in ¹³C-pyruvate signal can be observed in the same set of participants, but no noticeable difference is seen between conditions in ¹³C-bicarbonate. These results are consistent with prior studies that have observed increases in lactate signal during visual stimulus^10,11. Interestingly, the difference in ¹³C-lactate signal between occipital lobe and non-occipital lobe regions appears to result from a bidirectional change: non-occipital lobe regions had slightly reduced ¹³C-lactate production during the visual stimulus, while occipital lobe regions had increased ¹³C-lactate signal.

Conclusion

Increased ¹³C-lactate production in occipital lobe regions relative to non-occipital lobe regions was observed during a visual task using hyperpolarized-¹³C MRI. Future studies will explore the mechanism behind this signal change.

Acknowledgements

Funding support from the Canadian Institutes for Health Research grant PJT152928.