Improvement of in vivo glucoCEST imaging in rat brain using inverse z-spectrum analytical scheme at 7.0 T
Ping-Huei Tsai1,2,3, Hua-Shan Liu4,5, Fei-Ting Hsu6, Yu-Chieh Kao3, Chia-Feng Lu3, Li-Chun Hsieh2, Pen-Yuan Liao2, Hsiao-Wen Chung7, and Cheng-Yu Chen1,2,3

1Department of Radiology, School of Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan, 2Department of Medical Imaging, Taipei Medical University Hospital, Taipei Medical University, Taipei, Taiwan, 3Translational Imaging Research Center, Taipei Medical University, Taipei, Taiwan, 4Graduate Institute of Clinical Medicine, Taipei Medical University, Taipei, Taiwan, 5Department of Medical Imaging, Taipei Medical University Hospital, Taipei, Taiwan, 6Taipei Medical University Hospital, Taipei Medical University, Taipei, Taiwan, 7Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan

Synopsis

GlucoCEST have been proposed to assess the discrepant concentrations of glucose in vitro. However, it is still a challenge to obtain accurate signals from glucose in vivo. Our preliminary finding demonstrated that the proposed analytical scheme provides an alternative to extract glucose profile and could be more robust to the field inhomogeneity, which may be helpful in further implementation in disease models.

Introduction

Chemical exchange saturation transfer (CEST) method is an advanced MR imaging approach to detect the signal from metabolites with low concentration, such as protons from amine, amide or hydroxyl. GlucoCEST measurements have been proposed to successfully assess the discrepant concentrations of D-glucose or 2-deoxy-D-glucose (2-DG) in vitro [1]. Using asymmetric analysis, the asymmetric magnetization transfer ratio (MTRasym) and glucoCEST contrast maps can be derived directly from the z-spectrum, which may provide opportunities to quantify altered glucose levels in vivo. Although some previous reports demonstrated the potential role of glucoCEST in detecting the altered glucose assimilation in mouse tumor model, revealing comparable information to 18FDG-PET findings [2,3], poor reproducibility due to its high sensitivity to field inhomogeneity and possible contamination from asymmetric MT and signals from adjacent metabolites, which restrict its further application. More recently, Lorentzian function has been performed to fit the z-spectrum as a combination of multiple components [4], which in corporation with specific mathematical manipulation may provide some possibility for extracting the glucose signal with better accuracy and reliability. As a result, the purpose of this study is to develop the analytical scheme and assess the reliability in simulation as well as phantom study for glucoCEST imaging, and finally validate it in application of the rat brain in vivo.

Theory & Method

In vivo z-spectrum consists of multiple components including direct water proton saturation, asymmetric MT, aliphatic nuclear Overhauser effect(NOE), amide proton transfer(APT) and other CEST phenomena [5], leading to a challenge of deriving glucoCEST contrast by asymmetric analysis. Fortunately, the signals from the individual components could be approximated using Lorentzian function, which provides some room to calculate the individual signal from Lorentzian line fitting on the inverse z-spectrum more efficiently and accurately. Based on this concept, we first simulated the z-spectrum with signals from the five components at 0(Water), -1.5(MT), -3.2(NOE), 1(Glucose), and 3.5 ppm (APT), to mimic the situation. After testing the reliability of the developed fitting models, this approach was performed with a phantom of 50 mM D-glucose and in vivo rat scan. The glucoCEST imaging was performed on 3 Sprague Dawley rats at a 7.0T animal MR scanner (PharmaScan, Bruker, Erlangen, Germany) with a 72 mm transmitting coil and a quadrature surface coil for receiving. After manual shimming, a turbo rare based CEST sequence with a continuous form MT saturation pulse (duration =800 ms, B1=2.5 μT) was performed before and after administration of 0.2 ml 2-DG (6 mM) with TR/TE=1000/20 ms, FOV=20x20 mm2, matrix size=128x128, slice thickness=1 mm, rare factor=8, acquisition time=16min 16sec, and z-spectrum offset=±5 ppm with 0.167 ppm steps. After data acquisition, a 2D median filter was performed first to alleviate the noise effect. Inverse z-spectrum analysis was then applied to fit the 5 individual components. For glucoCEST contrast in rat brain, the signal was derived by subtracting the fitted pre-contrast z-spectrum from the post-contrast z-spectrum.

Results

The demonstration of the fitted inverse z-spectrum and the 5 individual components were in consistent with the simulated data(Figure 1(a)). With decreasing of the sampling rates, similar glucose profiles were obtained with less than 1% differences(Figure 1(b)). Moreover, MTRasym derived from the 50 mM glucose phantom using conventional asymmetric analysis shows a shift of the profile as compared with that from the inverse Lorentzian analytical scheme(Figure 2). The results became even worse when the glucose level decreased. Significant overestimation of the altered glucose contrast was shown in rat brain using asymmetric analysis(Figure 3(a)) as compared to that derived from the proposed method(Figure 3 (b)).

Discussion

This present study demonstrated the ability to extract glucose profile from CEST imaging using the inverse z-spectrum analytical scheme and indicated the feasibility of detecting the altered glucose assimilation in rat brain. Although the acquisition time is relative longer for the clinical application, results from the simulations suggest that there are still some rooms for further acceleration. In addition, greater than 6% increase of the glucoCEST enhancement was observed in vivo after administration of 2DG using the proposed analytical method. The finding also shows more reliable changes of glucose assimilation in both cortical and subcortical regions of the brain due to that the contamination from creatine signal at 2 ppm was eliminated by the subtraction. In conclusion, our preliminary finding demonstrated that the proposed analytical scheme provides an alternative to extract glucose profile and could be more robust to the field inhomogeneity, which may be helpful in further implementation of glucoCEST imaging in disease models.

Acknowledgements

This project is supported by TMU grant (TMU103-AE1-B20). We also thank Dr. Gang Zhu, in Bruker U.S.A. for technical consulting support.

References

1. Nasrallah FA, et al. J Cereb Blood Flow Metab 2013;33:1270-8

2. Walker-Samuel S, et al. Mat Med 2013;19(8): 1067-73

3. Rivlin M, et al. Sci Rep 2013; 3(3045) DOI: 10.1038

4. Cai K, et al. NMR Biomed 2014 ; 28 : 1-8

5. Jin T, et al. Mag Reson Med 2013; 69: 760-70

Figures

Figure 1 (a) Fitted inverse z-spectrum and 5 individual components (b) glucose profiles derived from discrepant sampling rates.

Figure 2 MTRasym and glucose profile derived from the conventional asymmetric analysis (a) and the inverse z-spectrum analytical scheme (b) in phantom study.

Figure 3 GlucoCEST contrast derived from the conventional asymmetric analysis (a) and the inverse z-spectrum analytical scheme (b) in rat brain.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
2910