Assessment of serine quantification reproducibility using advanced 1H-MRS in the human brain at 3T
Homa Javadzadeh1,2 and Jean Théberge1,2,3

1Department of Medical Biophysics, University of Western Ontario, London, ON, Canada, 2Imaging Division, Lawson Health Research Institute, London, ON, Canada, 3Diagnostic Imaging Department, St. Joseph's Health Care, London, ON, Canada

Synopsis

D-serine supplements alleviate some of the most debilitating features of schizophrenia believed to be associated with glutamatergic abnormalities. Assessment of endogenous serine is impossible using standard proton Magnetic Resonance Spectroscopy (1H-MRS). This work employs a novel 1H-MRS sequence called DANTE-PRESS (D-PRESS) and presents test-retest reliability study for serine levels acquired at 3T in phantoms and initial data in two human subjects. We conclude that reproducibility and precision of serine measurements on a 3.0T scanner is sufficient to assess endogenous levels in vivo and is a valuable tool to examine abnormalities in schizophrenia and monitor supplementation.

Purpose

Precise and reliable measurements of endogenous serine in the human brain using advanced 1H-MRS is crucial for examining any abnormalities of its concentration in patients living with schizophrenia (1, 2). The detection of endogenous serine has been unattainable utilizing standard proton Magnetic Resonance Spectroscopy (1H-MRS) because of its relatively low concentration and its intensive spectral overlap with the neighbouring metabolites (3). Previous measurements of serine were obtained using advanced 1H-MRS techniques at 4.0T and 7.0T (3, 4). We introduce a novel pulse sequence called DANTE-PRESS (D-PRESS) using a narrowly frequency-selective refocusing DANTE pulse to select serine signals while suppressing interfering neighbouring resonances on a 3T clinical scanner (Siemens Biograph mMR). In this study, repeated in vivo measurements were obtained in two subjects and the test-retest reliability of the assessment of in vitro measurements was demonstrated using “in vivo” [~0.732] mM and “double in vivo” [~1.464] mM serine concentrations at baseline and a week later.

Methods

The D-PRESS sequence was implemented through inserting a spectrally-selective single-DANTE pulse to a symmetrical PRESS sequence, at time TE1 after the first 180 degree refocusing pulse: 90- TE1/2-180-TE1/2-DANTE-TE2/2-180-TE2-Acq (Figure 1). The 0th and 1st sideband of the DANTE pulse were used to achieve very selective frequency passbands centered on the 3.83 ppm resonance of serine and the 2.01-ppm resonance of NAA (N-Acetylaspartic) (5). The DANTE pulse (duration of 118.54 ms) was amplitude-modulated with a Gaussian envelope truncated at 5% producing a frequency-domain inversion profile with Gaussian passbands repeating every 223Hz and it consisted of a series of 0.086 ms short, high power square RF pulses, interleaved with delays of 4.48 ms. This profile was centered on a frequency determined at run-time by assigning the NAA resonance from a DANTE OFF acquisition the chemical shift valued of 2.01 ppm and adjusting the carrier frequency to the serine resonance at 3.83ppm (TR=2.0s,TE=286ms, and NA=64). The serine data was obtained from a DANTE ON acquisition with TR=2.0s, TE=286ms, BW=1.5 kHz and NA=240. Water unsuppressed signal was used as concentration reference for metabolite quantification.

In vitro experiments: A final quantification template for in vitro data analysis was developed from the spectral quantification templates obtained in two spherical phantoms (inner diameter [i.d.] 10 cm) containing either high concentrations of the metabolites of interest and NAA (8mM) as a chemical shift reference. Metabolites of interest were: serine (30 mM), creatine (25mM), glutamate (30mM) and glutamine (30mM), all to be included in the in vivo template (6). Ten consecutive spectra were obtained within-session and from between-session (phantom scanned a week later). The data was obtained with DANTE pulse with narrow passbands (15Hz = 0.11ppm at 3T), TR=2.0, TE=286ms, NA=240, number of data points in DANTE waveform=2560, FDP (Frequency Domain Period) of 223 Hz.

In vivo experiments: Two healthy volunteers (2 male; age of 27 ± 4) were recruited and scanned using the same parameters as for the phantom work. For these subjects, two consecutive spectra were obtained within sessions at baseline. 1H -MRS data were obtained from a single voxel (20x20x20mm3) positioned in the anterior cingulate (Figure 2). High resolution 3D MP-RAGE images were used for voxel placement. Metabolite concentrations were quantified using Fitman (7).

Results

An example spectral fit is displayed in figure 3, illustrating the data, fitting line, fit components and the fit residuals. Figure 4 displays in vitro demonstration of ten repeated measurements that were obtained from “in vivo” and also acquired from “double in vivo” serine concentrations at two intervals each obtained one a week apart. The resulting ICC were both high (in vivo ICC =0.91, double in vivo ICC =0.93) indicating excellent repeatability of in vitro measurements within and between scans. The two repeated measurements from the first and second in vivo experiments were quantified and the in vivo concentrations were: 1.17 mM, 1.99 mM and 1.11 mM and 2.17, respectively. The reproducibility of MRS measurements greatly depend on the quality of adjustments and more advanced MRS post-processing capabilities; as illustrated in MRS literature (8), thus this sequence is currently being further modified for obtaining final in vivo test-retest data.

Discussion and conclusion

This study demonstrated the first examination to investigate the reliability of method to reliably detect endogenous serine using a metabolite-selective single-voxel 1H-MRS technique on a clinical scanner of 3T. In the proposed work we demonstrated the accuracy and precision (reliability/reproducibility) of the measurements of serine obtained in phantoms. The proposed method will be able to assess metabolic changes in neuropsychological conditions such as schizophrenia and support the investigation of safer and more effective treatment for the affected patients.

Acknowledgements

The salary support of this study was funded by a grant from NSERC.
The chemicals that were used in this study and the serine project were funded by Lawson internal research fund.

References

1. Tsai et al., Biol Psychiatry 44:1081-1089, 1998. 2. Tran et al. Neurol Clin 27: 21-60 (2008). 3. Choi et al., Mag Reson Med 62(4):1042-46, 2009. 4. Theberge et al., ISMRM (Abstract 1373), 2007. 5. Geen et al., J Mag Reson 81:646-52, 1989. 6. Kumashiro et al., Brain Research 681:117-125, 1995. 7. Bartha et al., NMR in Biomedicine 12(4):205-216, 1999. 8. Deelchand 2015, MRM 73 (5):1718-1725.

Figures

Figure 1. A diagram of the DANTE-PRESS sequence.

Figure 2. The illustration of an in vivo voxel positioning

Figure 3. An illustration of an in vivo spectrum. Line (A) data is represented in black, fitting line in red, line (B, C, D, E) fitting components of serine in orange, creatine in purple, glutamate in green and glutamine in pink. Line (F) displays the residuals. LW correction-factor of 2Hz.

Figure 4. The average value of in vivo baseline, one week, plus double in vivo baseline and one week were 1.066 ± 0.078 with CV of 7.33%, 1.002 ± 0.098 with CV of 9.77%, 2.093 ± 0.110 with CV of 5.24% and 2.130 ± 0.140 with CV of 6.48%, respectively.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
2381