Benjamin C Rowland1, Min Zhou1, Huijun Liao1, Nils Arvold2, Raymond Y Huang3, and Alexander P Lin1
1Centre for Clinical Spectroscopy, Brigham and Women's Hospital, Boston, MA, United States, 2Radiation Oncology, Brigham and Women's Hospital, Boston, MA, United States, 3Radiology, Brigham and Women's Hospital, Boston, MA, United States
Synopsis
MR spectroscopy is a powerful technique for understanding tumour metabolism. Chemical shift imaging allows spatial variations in spectra to be observed at the cost of signal to noise and acquisition time. In this abstract we examine the heterogeneity of metabolite concentrations in both tumour and healthy tissue, in particular the onco-metabolite 2-hydroxyglutarate, to ascertain if the benefits of CSI outweigh the penalties.Introduction
In recent years, neuro-oncologists have become increasingly interested in the capability of MR Spectroscopy (MRS) to assess the chemistry of brain tumours. The concentrations of choline and n-acetylaspartate indicate the degree of tumour activity, whilst elevated lactate indicates hypoxia/necrosis. Of particular interest is the ability of MRS to test for the presence of 2-hydroxyglutarate (2HG), an oncometabolite produced only by gliomas carrying mutations to the isocitrate-dehydrogenase (IDH) gene1. Not only is this useful in monitoring patients post-treatment, but determining whether or not the glioma carries the IDH mutation is also important for surgical planning, where IDH mutant tumours show a signifcant survival benefit with more aggressive resection2.
When prescribing MRS exams physicians have two options available: single voxel spectroscopy (SVS) gives a spectrum with a high signal-to-noise ratio (SNR) with a short acquisition time; chemical shift imaging (CSI) provides a grid of spectra, showing the spatial variation of the chemistry at the cost of a lower SNR and longer sequence duration. In addition, it is usually not possible to acquire an unsuppressed water reference for CSI, making absolute quantification impossible. However, in glioma cases there is often a high degree of heterogeneity across the tumour which may be hidden in SVS measurements by partial volume effects. In this study we investigate the level of this heterogeneity in metabolite concentrations and therefore evaluate the importance of using CSI.
Methods
14 patients with pathologically confirmed IDH+ gliomas were prospectively recruited for the study and MRI/MRS were obtained prior to commencing radiochemotherapy. Spectra were acquired on a 3T MRI (Siemens Skyra) with a 32 channel head coil using a single slice semi-LASER chemical shift imaging sequence with the following parameters: TE1/2: 21/76 ms for a total of 97 ms, TR: 1700 ms, 16x16 matrix with 160 mm FOV and 15 mm slice thickness for a 1.5 cm voxel resolution. Voxels were placed on axial localizers including T1w MPRAGE and 2D FLAIR to include both the lesion and contralateral normal tissue. Typical excitation volumes were 80 mm.
For each subject the CSI voxel grid was fused with the FLAIR image and a radiologist delineated the voxels inside the tumour, as well as a set of matched normal appearing voxels on the contralateral side. Metabolite concentrations in each voxel were quantified using LCModel3. All concentrations were normalised by the total creatine value. In addition the FIDs in each group were phase and frequency corrected, then added to give a composite signal representing the entire volume. This data was also quantified using LCModel and considered as a surrogate for an SVS acquisition of the tumour region. Of the 14 subjects recruited, 4 were rejected due to insufficient spectral quality and a further 4 due to issues with registering the MRS data to the FLAIR, leaving six patients with high quality results.
Results
The principal metabolites quantified in the normal appearing brain (TNAA, TCho, Glx, Ins and GSH) were found to have fairly consistent values, with a mean standard deviation of 19.7%, as shown in Figure 1. By contrast the same metabolites in the tumour voxels had on average twice the variation, with a standard deviation of 40.0%. The only exception was the choline concentration, with a very similar SD in both normal and diseased tissue, although the absolute concentration in tumour is on average 2x higher.
2HG was detected in the tumour voxels of five of the six patients. The heterogeneity of 2HG was much higher than for the other metabolites, with an average standard deviation of 85.2%. The degree of heterogeneity also correlated with the size of the lesion as judged by the number of included voxels. In the five patients with detectable 2HG, 17% of voxels were measured to have no 2HG, while the highest measured 2HG concentration in each tumour was between 150% and 300% of the average concentration.
Conclusion
Examining gliomas with CSI identified significant intra-tumor heterogeneity. Mapping this heterogeneity is important for targeted surgical/radiotherapy planning. Averaging over an entire region using SVS gives only mean concentrations, which can significantly underestimate the peak values present in sub-regions. Consequently CSI should become the preferred option for MRS examinations within neuro-oncology.
Acknowledgements
This research was made possible by the BWH Institute for Neurosciences Seed Grant.References
1 Choi C, Ganji S, DeBerardinis R et al. 2-hydroxyglutarate detection by magnetic resonance spectroscopy in IDH-mutated patients with gliomas. Nat Med. 2012; 18(4):624-9
2 Beiko J, Suki D, Hess K et al. IDH1 mutant malignant astrocytomas are more amenable to surgical resection and have a survival benefit associated with maximal surgical resection. Neuro Oncol. 2014; 16(1):81-91
3 Provencher S Estimation of metabolite concentrations from localized in vivo proton NMR spectra. Magn Reson Med. 1993; 30(6):672-9