Purpose

Brain glycogen is an important source of energy, not only in cases of pathophysiological glucose insufficiency such as during transient reduced perfusion, but also in the course of normal brain activity. Abnormalities in glycogen metabolism and its storage have been implicated in various diseases such as Lafora disease and Adult Polyglucosan Body Disease (APBD). It is hypothesized that large quantities of abnormally-branched brain glycogen that cannot be metabolized are accumulating in these conditions¹. However, quantitation of glycogen in brains of patients with this disorder has not been reported. Quantitative measurements of brain glycogen are thus desirable for diagnosis and treatment follow up. However, non-invasive tools for brain glycogen detection in vivo are lacking. Normal brain glycogen in mice is 100% ¹³C NMR visible², suggesting that excess brain glycogen could be detected. Glycogen concentration in the normal brain is low and natural abundance ¹³C glycogen has not been detected in the human brain. Recent ¹³C infusion studies^3,4 have produced estimates of brain glycogen concentration in the range of 3-10 mmol/g. However, the prolonged infusion protocols (2 days) required for human studies would be difficult for patients to tolerate. In this work we have optimized natural abundance ¹³C MRS, capitalizing on the increased sensitivity of high magnetic fields (7T) and NOE signal enhancement, to allow glycogen detection in normal controls. We have also applied the method in a case of an APBD patient. To our knowledge this is the first direct measurement of human cerebral glycogen via natural abundance ¹³C MRS.

Methods

All procedures were performed with local IRB approval and after obtaining a written consent. In adult volunteers (controls=2; APBD patient=1), proton-coupled ¹³C NMR spectra were acquired on a whole-body 7T scanner (Achieva, Phillips Healthcare) with the subject laying comfortably supine, with a quadrature ¹H/¹³C transmit/receive partial volume coil positioned below the occipital region of the head. Saturation slabs were utilized to minimize signal contribution from the occipital belly of the epicranial and trapezius muscles (Figure 1). The ¹³C spectra were acquired for approximately one hour (NSA = 8k), with TR = 400 ms, BW 16 kHz and 4k data points, frequency offset centered on the glycogen C1 signal at 100 ppm. To increase ¹³C SNR, NOE enhancement was implemented with B₁ = 10 μT, 5% duty cycle, 100 ms mix time centered at 180 Hz downfield of water. The NOE enhancements were initially confirmed in a glycogen phantom containing 100 mM glycogen, 40 mM creatine, and 25 mM 100% ¹³C enriched bicarbonate (a reference).

Results and Discussion

In the phantom experiments, a clear NOE enhancement of glycogen C1 SNR was observed at about 50%, Figure 2. The in vivo acquisition protocol, while comparatively long, was well tolerated. In the normal control brain, no glycogen C1 signal was detected when NOE was not utilized (data not shown). With NOE, glycogen C1 doublet (at approximately 100 ppm) was observed in both the normal controls and the APBD patient (Figure 3). Qualitative comparison of the respective SNR in these two cases indicates no dramatic increase of the detectable glycogen concentration in APBD. Given the reported several-fold increase of glycogen in these patients and the poor solubility of the abnormal glycogen, it is possible that the NMR visibility is compromised in this disease, in contrast to previous findings on normal brain glycogen². Lastly, in the control case, if we assume a bicarbonate concentration of 25mM and bicarbonate T₁=10s, a rough estimate of glycogen concentration of about 5mM can be obtained, consistent with previous ¹³C infusion findings.

Conclusion

We demonstrated the feasibility of detection of brain glycogen via natural abundance 7T ¹³C MRS with NOE. While the SNR of the detected signals is low and does not currently allow for reliable quantification of the observed spectra, the sensitivity of this method could be improved in the future by implementing closely-fitting ¹³C receive arrays, as well as with the availability of homogeneous volume excitation transmit coils. Assuming that the glycogen content of the brain of a patient with APBD is increased, these results suggest that ¹³C NMR visibility of abnormal glycogen or similar branched-chain structures is reduced in APBD.

Acknowledgements

This work was supported in part by grants from the National Institutes of Health (EB-015908 and HL-034557) and Cancer Prevention Research Institute of Texas (RP150456), and an internal pilot award (FY19-IA0001).