4997

A new EPI multislice evaluation for APT-CEST

Jan Rüdiger Schüre¹, Manoj Shrestha², Ralf Deichmann², Elke Hattingen¹, Marlies Wagner¹, and Ulrich Pilatus³

¹Department of Neuroradiology, University Hospital Frankfurt, Frankfurt am Main, Germany, ²Brain Imaging Center, Frankfurt am Main, Germany, ³University Hospital Frankfurt, Frankfurt am Main, Germany

Synopsis

To investigate APT-CEST feasibility as pH-sensitive contrast, a multislice CEST EPI sequence was evaluated. The proposed sequence allows continuous saturation during EPI measurement, leaving sufficient time to capture robust CEST data by multiple repetitions. The saturation scheme was optimized in a phantom study and with healthy volunteers. With the availability of 3D CEST data a subsequent registration of APT-CEST contrast images to 31P-MRSI can be achieved for evaluating pH-related changes in APT-CEST contrast. This is demonstrated in a study of a patient with glioblastoma.

Introduction

APT-CEST (Amid Proton Transfer - Chemical Exchange Saturation Transfer) is a ¹H-based method that provides information on pH via the chemical exchange of saturated amide protons with water protons. Due to the higher spatial resolution and by not requiring the use of specific hardware, this method offers advantages for intracellular pH imaging compared to ³¹P-MRS, which exploits the pH dependent chemical shift of inorganic phosphate (Pi). However, APT-CEST is challenging since contrasts depend on multiple factors such as sequence parameters and magnetization transfer effects arising from dissolved proteins, peptides, amino acids and metabolites ¹ .

To assess the suitability of APT-CEST for pH evaluation, data were acquired with a multislice CEST-EPI sequence. A phantom and volunteer studies were used to optimize the saturation scheme, while the feasibility of monitoring local pH-changes was confirmed in a glioblastoma patient based on ³¹P-MRSI data.

Methods

APT-CEST data were acquired on a 3T scanner (Prisma, SIEMENS) with a 32 channel phased-array head receive-coil. The acquisition was based on a multislice CEST-EPI sequence (TE=22ms, voxel-size=3x3x3mm³). Frequency selective saturation was achieved via a series of rectangular pulses (pulse duration=250ms, gap between saturation pulses=250ms, B1=1µT). Image data were acquired by recording each slice in the initial phase of a specific gap between saturation pulses. The sequence allows to freely choose the number of pulses between data acquisition and the total number of slices (1-8 slices), resulting in a minimum TR of 4000 ms for a slab with 8 slices. Z-spectra were recorded applying RF irradiations with different offset frequencies ranging from -8 ppm to + 8 ppm (measurement repetitions=2, increment=0.5 ppm) and from +/-3 to +/-4 ppm (measurement repetitions=6, increment=0.1 ppm). 10 dummy scans (8 rectangular saturation pulses pro dummy) were applied to reach the steady state. ³¹P-MRS data were acquired with a 3D CSI sequence, using a double tuned ¹H/³¹P coil. Parameters were TR=2000ms, FA=60°, 8x8x8 matrix, FoV=240x240x240 mm³ (voxel size of 15x15x12.5mm³ after matrix extrapolation). The spectra were analyzed voxel-wise with jMRUI, employing the nonlinear least-square fitting algorithm AMARES in the time domain. For sequence validation a gelatin-based APT-CEST phantom was prepared adding agarose and gadovist® to obtain white matter T1 and T2 values, and NaCl to adjust the coil loading. The pH was calibrated to 7.2 with KOH. For each subject, motion correction was performed before Z-spectra were determined voxel-wise, interpolated and corrected for B0 field inhomogeneity. MTR_asym was calculated at 3.5 ppm distance from the water resonance. Furthermore, the APT signal was evaluated via Lorentzian difference analysis (LDA), according to ². Subsequently, resulting CEST contrasts were co-registered with the ³¹P-MRS data. Spectroscopic data were used to calculate pH maps according to Petroff et al ³ using the spectral distance between Pi and phosphocreatine and to quantify the pH-dependent chemical shift of Pi. The resulting values were compared to APT-CEST contrast.

Results

Based on the phantom data, it could be shown that the number of saturation pulses between acquisitions of a single slice has no influence on the APT-CEST contrast. Figure 1 shows the results for a slab with 8 slices at a TR of 4000/8000/12000/16000ms (1 pulse/2pulse/3pulse/4pulse, 1 slice).

Figure 2 shows the APT-CEST contrast for a healthy subject obtained either from MTR_asym (top) or from the residual signal of the amides after LDA (bottom). To improve the SNR, APT-CEST data were averaged 6 times (duration: 20min for a total of 175 offsets over the whole volume).

Figure 3 shows data from a patient with glioblastoma in the frontal medullary bed: pH map obtained via ³¹P-MRS (top) and the co-registered APT-CEST data (bottom). The spectroscopic data show that pH is increased in the glioblastoma and decreases with increasing distance from the focal point. Similarly, MTR_asym shows an increased contrast in the area of the glioblastoma.

Discussion

The phantom data show that a saturation pulse of 250 ms followed by a 250 ms delay is sufficient to maintain the steady state even if an EPI slice is acquired during the delay. This can be used to record multislice EPI data with a TR of 4000 ms per 8 slices, allowing repeated measurements to increase SNR. Thus, it is possible to calculate parametric maps like MTR_asym or via the Lorentzian difference analysis, which requires robust data for the fitting process.

Conclusion

The presented multislice sequence provides robust data. With the availability of 3D CEST data a subsequent registration to the spectroscopic data allows a direct comparison of pathological pH-changes for both methods over the entire tumor offering the feasibility to study pH-related changes in the APT-CEST contrast.

Acknowledgements

No acknowledgement found.

References

1. van Zijl Peter C.M. ,et al. Magnetization Transfer Contrast and Chemical Exchange Saturation Transfer MRI. Features and analysis of the field-dependent saturation spectrum.Neuroimage 2017

2. Schüre Jan-Rüdiger ,et al. Correlation of tissue pH via 31P-MRSI with MTRasym derived from APT CEST-MRI in glioblastoma and normal appearing white matter.ISMRM 2018

3. Petroff OA,et al. Cerebral Intracellular pH by 31P nuclear magnetic resonance spectroscopy. Neurology 1985;35(6)781-8

Figures

Fig.1: Phantom measurement of the Z spectra (left) and the residual signal after the LDA (right) over 8 slices after irradiation of a) 1, b) 2, c) 3 and d) 4 saturation pulses. The data show that the e Z spectra and thus the APT signal at +3.5 ppm are independent of the number of pulses.

Fig.2: In vivo APT-CEST results for a healthy volunteer over 8 slices, 6 averages over the spectral range from ±3 to ±4 ppm. A) Calculated MTR asymmetry at 3.5 ppm B) Residual signal of the amides at 3.5 ppm after LDA. Both data sets illustrate a homogeneous distribution of amides in the white matter.

Fig.3 Representation of the calculated pH map via phosphorus spectroscopy (above) with the corresponding co-registered MTR asymmetry via APT-CEST measurement (below) in a patient with glioblastoma. Both data sets show a good agreement in the glioblastoma.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)

4997