Introduction

Hepatic encephalopathy (HE) is a common complication in patients with liver cirrhosis. It comprises a variety of neuropsychiatric and motor symptoms¹ and leads to an increase of ammonia within the brain². Therefore, it is desirable to create 3D metabolic images of the whole brain to reflect the content of ammonia non-invasively. The present study tests the feasibility of chemical exchange saturation transfer imaging (CEST) to display changes in protein CEST signal depending on the content of ammonia.

Methods

In vivo acquisition was performed on a clinical 3 T whole-body MRI scanner (Siemens MAGNETOM Trio A TIM System, Siemens Healthcare GmbH, Erlangen, Germany) using a 12-channel head matrix coil. A prototype 3D gradient echo sequence (echo time TE=3.06ms; repetition time TR=1390ms; FA=10°; FoV=(230x230)mm²; 24 slices; 5mm slice thickness; 1mm gap) with a pre-saturation block (5 gaussian shaped pulses; pulse duration PD=100ms; inter pulse delay IPD=100ms; 22 equidistant frequency offsets between ±5ppm; B₁=1.5µT; 3s break) was used for in vivo CEST imaging and B₀ correction via WASSR³ (changed parameters: 1 gaussian shaped pulse; PD=5ms; IPD=6ms; equidistant offset ±1ppm; B₁=0.3µT). B₁ correction was performed using a one-point correction method⁴ with a relative B₁ map⁵. In vitro measurements were performed within the steady state, using 48 gaussian shaped pulses at 44 equidistant frequency offsets between ±5ppm. In addition, nine different B₁ amplitudes (B₁ = 0.4, 0.7, 1, 1.3, 1.4, 1.5, 1.6, 1.7, 2 µT) were used for B₁ correction⁴. We evaluated our approach in a two-step procedure: First, bovine serum albumin (BSA) solutions (2.5%) and brain samples consisting of one part of diluted and homogenized pig brain tissue (tissue:water ≡ 1:2) and two parts of buffered ammonia solutions at fixed pH=7 were tested with varying ammonia concentrations ([0, 0.5, 5, 50, 500] µM, [5, 50, 500] mM). Fluorescence spectroscopy was used to determine possible changes in protein folding in the BSA phantom. In a second step, 32 participants (11 female; mean age ± standard deviation: 58 ± 6.6 years) were examined after written informed consent, including 14 healthy control, 7 minimal HE (mHE), and 9 HE I patients, graded according to the West-Haven criteria. Anatomical imaging for segmentation purpose and relative B₁ correction⁴ were performed. Post processing was realised with an in-house written MATLAB script (MathWorks Inc, Natrick, MA, USA). All images were smoothed by Gaussian kernel and normalized to S₀. Interpolation of CEST and WASSR z-spectra was done with MATLAB interp1 algorithm (step size of h=0.001). CEST z-spectra were shifted by the calculated WASSR offset to reduce B₀ inhomogeneity effects. The magnetization transfer ratio asymmetry (MTR_asym) was determined for amide protons around (3.5 ± 0.5) ppm. The in vivo measurements were coregistered onto the first offset image (S₀) with SPM12 (Wellcome Trust Centre for Neuroimaging, University College London, UK) to reduce movement artefacts. In vivo 3D MTRasym maps were coregistered onto the anatomical images and normalized to MNI-152-space to create group atlases. MATLAB students’ t-test was used for statistics.

Results

In the phantom measurements the mean MTR_asym value decreases significantly with increasing ammonia concentrations (Figure 1 & Figure 2). No protein unfolding was visible for concentrations <50 mM (data not shown). Representative slices of the group MTR_asym atlases are depicted in Figure 3, showing that the MTR_asym values within the representative slice of the control, mHE and HE patient decrease with severity of HE. The mean MTR_asym values within the whole cerebellum decreased significantly from 0.9 ± 1.1 % (control), 0.8 ± 1 % (mHE) to 0.6 ± 0.7 % (HE I).

Discussion

The measurements revealed a significant decrease of the CEST signal with increasing concentration of ammonia within the phantom. Possible confounders to this effect are reduction of water T2 values⁶ due to ammonia as it is known to suppress MR water signal or small changes in protein structure⁷. In the in vivo measurements in patients with HE a decrease in the MTR_asym signal occurred in comparison to healthy controls, which may be due to ammonia uptake or protein aggregation in astrocytes⁸.

Conclusion

The Chemical Saturation Transfer (CEST) signal shows an ammonia concentration dependent contrast, which allows for investigations of patients with hepatic encephalopathy. Future investigations with larger patient groups should confirm the current findings and may allow analysing regional alterations within the HE patient brains.

Acknowledgements

This study was supported by the Sonderforschungsbereich (SFB) 974 (TP B07) of the Deutsche Forschungsgemeinschaft (DFG).