Introduction 

Glutamine is a target of new cancer therapies due to the abnormal dependence of cancer cells on glutamine [1,2]. Previous research using ex vivo nuclear magnetic resonance spectroscopy observed increased glutamine concentration was positively correlated with glioma tumor grade [3]. In vivo, glutamine can be quantified in the brain using magnetic resonance spectroscopy (MRS), and prior work using 3 Tesla (3T) MRS reported increased glutamine concentrations in gliomas [4]. Reliable quantification of glutamine at 3T, however, remains challenging due to spectral overlap with glutamate and low signal-to-noise ratio (SNR) [5]. The use of ultra-high field (UHF) MR scanners (e.g., 7T) can facilitate higher spectral resolution and SNR [5]. While glutamine has been quantified using MRS at 7T, the repeatability of these measurements is less characterized. The goal of this study was to assess the repeatability of in vivo glutamine quantification using single-voxel MRS acquired in the brain on a 7T MR scanner and compare to acquisition at 3T, with the future goal of translating these methods to glioma patients.  

Methods 

MR data was collected in ten healthy volunteers (mean ± standard deviation age = 27 ± 11 years; five males, five females) at 7T (Siemens TERRA) using an 8-channel transmit, 32-channel receive head coil (Nova Medical) and 3T (Siemens PrismaFit) using a 32-channel receive only coil (Siemens). T1-weighted images were used to position the MRS voxel using a magnetization-prepared rapid gradient echo (MPRAGE) sequence at 3T and MP2RAGE at 7T. A semi-localized by adiabatic selective refocusing (sLASER) sequence for MRS (TR=3000 (3T)/5000 (7T) ms, TE=35 ms, flip angle=90°, bandwidth=4500 Hz, transients=64, complex data points=2048, voxel size=20x20x20 mm³) was acquired in the posterior cingulate cortex (PCC). Two repeated scans were acquired, followed by repositioning, and two additional repeated scans. In a subset of subjects (n=4), an additional 3T MRS scan was acquired with the same acquisition time as the 7T scan (transients=110; 5 min 58 sec). Scans were conducted in a single session, with the order of field strength (7T or 3T) randomized. 

Spectra were analyzed using LCModel (v6.3-1L). SPM12 was used for voxel segmentation. Absolute glutamine concentrations (mM) were calculated using water-normalized values from LCModel, corrected for cerebrospinal fluid and T1/T2 relaxation times [6]. Metrics for spectral quality include SNR and Cramér-Rao lower bounds (CRLBs) for glutamine fitting. Repeatability between pairs was assessed using a two-way, mixed effects, absolute agreement intraclass correlation coefficient (ICC) (3,2), reported as ICC [95% confidence interval]. ICC values were categorized as excellent (>0.90), moderate (0.75-0.90), good (0.5-0.74), and poor (<0.5). Wilcoxon signed-rank tests were used to compare CRLBs and glutamine concentrations between field strengths. 

Results and Discussion 

MRS voxel positions overlaid on T1-weighted images and corresponding spectra are shown in Figure 1. ICC values for SNR at 7T (0.98 [0.95-0.99]) indicate excellent repeatability, compared to high variability at 3T with values ranging from good to excellent (ICC=0.92 [0.66-0.98]). ICC values for glutamine CRLBs were moderate (>0.8) at both field strengths; however, CRLBs were significantly lower at 7T compared to 3T (Z=-2.5, p=0.012) indicating improved quantification. Glutamine concentrations were also significantly lower at 7T versus 3T (Z=-2.7, p=0.007). With a matched number of transients, longer TR at 7T necessitates longer scan times; however, time-matched scans also demonstrated lower glutamine CRLBs at 7T (9-10%) versus 3T (12-13%). These data suggest 7T MRS results in improved concentration estimates even when controlling for scan time, supporting the use of ultra-high field strength MR scanners for in vivo glutamine quantification. 

Conclusion 

Improved repeatability and glutamine quantification was observed using 7T MRS. Immediate next steps include applying this protocol in glioma patients to aid in the development of glutamine as a prognostic biomarker. 

Image/Figure:

Click to view full size

Image/Figure Caption:

Figure 1. Representative MRS voxel positioning overlaid on T1-weighted images and corresponding MR spectra at 3T and 7T acquired in the same subject. T1-weighted images were acquired using MPRAGE (3T, left) and MP2RAGE (7T, right) sequences. Images are displayed in neurological orientation. Single-voxel MRS data were acquired in the posterior cingulate cortex at 3T (left) and 7T (right) using the sLASER MRS sequence. Raw data (black), as well as the fit obtained using LCModel (red), are shown for both spectra. 

Presentation Poster:
Click to View

Author