31P Magnetic Resonance Fingerprinting Method for Efficient Measurement of Creatine Kinase Mediated High Energy Phosphate Metabolism
Charlie Yi Wang1, Yuchi Liu1, Shuying Huang1, Mark Alan Griswold1,2, and Xin Yu1,2

1Biomedical Engineering, Case Western Reserve University, Cleveland, OH, United States, 2Radiology, Case Western Reserve University, Cleveland, OH, United States

Synopsis

We propose a novel 31P Creatine Kinase encoding Magnetic Resonance Fingerprinting method (CK-MRF) to efficiently measure high energy phosphate metabolism through creatine kinase (CK). Measurement reproducibility of CK rate constant (kf,CK) using CK-MRF was compared with both conventional 31P saturation transfer method, and Four Angle Saturation Transfer method in rat hind limb. CK-MRF measurement showed comparable or superior reproducibility using 20 s experiment time compared to 160 s experiment time of either comparison method. Changes in kf,CK following Ischemia/Reperfusion (IR) were also measured.

Background/Purpose

31P Magnetization Transfer spectroscopy (MT-MRS) has been used to measure high energy phosphate metabolism via creatine kinase (CK) in vivo. However, current 31P MT-MRS methods require prohibitively long experiment time, which limits their clinical applications. In this study, we propose a novel Magnetic Resonance Fingerprinting1 (MRF) framework based method, the Creatine Kinase encoding MRF method (CK-MRF), to increase measurement efficiency of 31P metabolism through CK. Measurement accuracy and reproducibility of the creatine kinase forward rate constant (kf,CK) using CK-MRF was compared with conventional methods in rat hind limb. Changes in kf,CK following Ischemia/Reperfusion (IR) were also measured.

Methods

CK-MRF acquisition consisted of an inversion preparation pulse followed by 320 acquisitions arranged in blocks, alternating between phosphocreatine (PCr) and γATP excitation (Fig. 1). The acquisition was organized into two sections, composed of blocks of 10 ramped excitations with 12.8 ms TR. Excitation flip angle (FA) in both sections was modulated by a sinusoidal envelope. Following each excitation block, frequency selective saturation was achieved with a 470 ms BISTRO scheme. Saturation frequency at either 2.4 ppm upfield of PCr (γATP saturation, section 2) or 2.4 ppm downfield of PCr (control saturation, section 1) was used. Total acquisition time for 1 fingerprint average was 20 s. RF excitations used Gaussian shaped spectrally selective pulses (4 ms duration) with alternated phase. FID signals were acquired in a 7.8 ms window with 30 ms dwell time. Signal evolutions for both PCr and γATP were extracted by Fourier transform.

A dictionary was constructed using Matlab-based Bloch-McConnell simulator that included four matching parameters: intrinsic PCr longitudinal relaxation time (T1,PCr), pseudo-first order CK exchange rate (kf,CK), PCr to ATP ratio (M0,PCr/M0,γATP), and PCr chemical shift (δPCr). Dictionary resolution for T1,PCr, M0,PCr/M0,γATP, and δPCr was 0.25 s, 0.3, and 6.25 Hz respectively. kf,CK used an adaptive resolution of 0.006 s-1 and 0.03 s-1 for values below and above 0.415 s-1 respectively. The dictionary totaled 132,660 entries.

Animal studies (n=3) were performed on rat hind-limb at 9.4T using a custom-built 31P saddle coil. An inflatable cuff was placed at the thigh of the rat to induce ischemia. During baseline, 3 methods were compared for measurement precision: CK-MRF, conventional MT by 31P saturation transfer2, and Four Angle Saturation Transfer3 (FAST). A total of 1800 s of data was acquired for each method. Acquired data was retrospectively averaged without view-sharing and the results of parameter estimation with different number of averages were compared. Following baseline data acquisition, 20 min ischemia was induced by inflating the cuff to 300 mmHg followed by 10 minutes of reperfusion for stabilization. Afterwards, CK-MRF and conventional 31P saturation transfer were acquired in interleaved fashion for a total of 20 min each.

Results

Figs. 2a&b show representative fingerprints acquired in vivo. Fig. 3 shows the measurement accuracy and reproducibility of baseline kf,CK for all 3 methods using different number of signal averages, corresponding to an acquisition time of 20 s (CK-MRF and FAST methods only), 160 s, 320 s, and 1800 s, respectively. Even with 20 s acquisition (single average), CK-MRF demonstrated similar or better precision compared to 160 s signal averages of either conventional MT or FAST acquisition. Data from 20 s FAST acquisition yielded nonsensical data. Both CK-MRF and conventional MT detected an increase in post-ischemia kf,CK. CK-MRF observed an increase from 0.315 s-1 at baseline to 0.349 s-1 after reperfusion, while conventional 31P saturation transfer observed an increase from 0.338 to 0.374 s-1.

Discussion/Conclusion

In this study, we present the use of a spectroscopic MRF method to increase the sensitivity to creatine kinase activity. The current iteration shows good agreement with conventional 31P MT measurement methods with far superior measurement efficiency leading to the possibility of performing spatial mapping measurements.

Acknowledgements

The authors would like to acknowledge funding from NIH TL1-TR000441, T32-EB007509, F30-HL124894, R01-HL73315, R21-HL126215.

References

1. Ma, D. et al., Nature 2013;495:187–92.

2. Xiong, Q. et al., Circulation Research 2011;108:653–663.

3. Bottomley, P. et al., Mag. Reson. in Med. 2002;47:850–863.

Figures

Figure 1, CK-MRF pulse sequence diagram (a) and excitation and saturation scheme description (b).

Figure 2, Example in vivo CK-MRF signal evolution using 320 s (a) and 20 s (b) duration of signal average, overlayed by corresponding dictionary match.

Figure 3, Comparison of in vivo measurement method accuracy and reproducibility in CK forward rate constant measurement. Each method acquired a cumulative 1800 s of data per animal. Error bars indicate standard deviation of repeated measurements obtained through retrospective time-averaging of acquired data to the given durations.




Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3957