Proton magnetic resonance spectroscopy and apparent diffusion coefficient in evaluation of solid brain lesions

Background/Aim. Advanced magnetic resonance techniques can provide insight in physiological changes within pathological canges and contribute to better distinquishing between different tumor types and their discrimination from non-neoplastic lesions. The aim of this study was to evaluate the role of proton magnetic resonance spectroscopy (1H-MRS) and apparent diffusion coefficients (ADC) in distinguishing intracranial glial tumors from tumor like nonneoplastic lesions, as well as for differentiating highfrom low-grade gliomas. Methods. This retrospective study included 47 patients with solid brain lesions (25 nonneoplastic, 14 low-grade and 8 anaplastic glial tumors). In all patients 1H-MRS (at a TE of 135 ms and 30 ms) and diffusion-weighted imaging (DWI) were performed. The choline to creatine (Cho/Cr), choline to N-acetyl aspartate (Cho/NAA), N-acetyl aspartate to creatine (NAA/Cr) and myoinositol to creatine (mIn/Cr) ratios and the apparent diffusion coefficient (ADC) were determined. Results. The Cho/Cr ratio was significantly higher in glial tumors grade II than in non-neoplastic lesions (p = 0.008) and in glial tumors grade III than in non-neoplastic lesions (p = 0.001). The Cho/NAA ratio was significantly higher in glial tumors grade II than in non-neoplastic lesions (p = 0.037). ADC/ADC between glial tumors grade II and glial tumors grade III showed a statistical significance (p = 0.023). Conclusion. Our study showed that 1H-MRS and apparent diffusion coefficients can help in evaluation and differentiation of solid brain lesions.


Introduction
Conventional magnetic resonance (MR) imaging has become the gold standard for detection and morphological assessment of solid brain lesions [1][2][3][4][5] .However, MR imaging based differentiation of neoplastic from non-neoplastic brain masses and the establishment of tumor grade are often difficult 4,6,7 .Further evaluation and follow-up are often necessary, including histopathological examination of biopsy specimens 8 .When lesions cannot be treated surgically or when they are located at areas of high risk for biopsy, greater accuracy of non-invasive imaging evaluation is desirable 1,2 .Assessment of MR images obtained after administration of a paramagnetic contrast agent must be done with caution, because any pathology associated with disruption of bloodbrain barrier (BBB) results in post-contrast enhancement 9 .Advanced MR techniques, like MR spectroscopy and diffusion-weighted imaging, can provide insight in physiological changes within pathology and contribute to more successful distinguishing between tumor types and their separation from tumor mimicking lesions 1,10 .
The radiological differential diagnosis of solid brain masses varies from tumors (gliomas WHO grades I-III), benign pseudotumoral lesions to demyelinating or ischemic lesions 7 .Therefore, establishment of correct diagnosis is crucial for choosing appropriate therapeutic procedure and patient outcome 11,12 .Gliomas are the most common primary neoplasms of brain, typically heterogeneous, varying histologically from low grade to high grade [11][12][13][14] .Although golden standard in diagnosis of brain glioma, histological evaluation can be misleading, because sampling regions may or may not correspond to increased cellularity and/or neoangiogenesis 11,[15][16][17][18] .Therefore, more accurate information about tumor physiology, such as metabolism, cellularity and microstructure are important in determining tumor grade and cannot be collected only based on conventional MR imaging 11 .Advanced MR imaging techniques, such as proton MR spectroscopy ( 1 H-MRS) and diffusion-weighted MR imaging (DWI) could provide insight in those features and hence increase accuracy of prediction of tumor histological grade 8,9,15 .Tracing of brain metabolites concentrations using 1 H-MRS can provide information about cell proliferation, degradation, energetic metabolism and appearance of ischemia or necrosis 5,14,16 .DWI and apparent diffusion coefficient (ADC) values obtained from DWI, provide complementary information about cellular density and tissue microstructure 16 .
The aim of this study was to assess the role of proton MR spectroscopy and DWI in discrimination of gliomas from non-neoplastic mimics, as well as for differentiation of grade II from grade III of glial neoplasms.All the glioma patients underwent MR imaging examination followed by surgery and histological evaluation of the lesion.Fourteen of them were assigned to be grade II (5 diffuse astrocytomas, 4 oligoastrocytomas and 5 oligodendrogliomas) and 8 as anaplastic astrocytomas grade III (Table 1).

MR imaging
MR imaging examinations were performed on a 1.5 T MR imging device (Avanto, Siemens Medical Solutions, Erlangen, Germany) using the standard 8-channel transmit/receive head coil.The conventional MR imaging protocol consisted of a three-plane localizer sequence, axial T1 weighted spin echo (SE), repetition time eho time [(TR/TE) 550/9.4ms, slice thickness 5 mm, gap 1 mm, matrix 512 × 256, NEX 2, FOV 24 cm], axial and sagittal turbo T2 weighted spin echo (TSE), (TR/TE 4820/94 ms, slice thickness 5 mm, gap 1 mm, matrix 512 × 256, NEX 2, FOV 24 cm), coronal fluid-attenuated inversion recovery (FLAIR), (TR/TE/TI 9900/126/2500 ms, slice thickness Proton MR singl evoxel spectroscopy (SVS) or chemical shift imaging (CSI) with a TE of 30 ms and TE of 135 ms, was performed immediately after completion of conventional MR imaging.SVS was used for well-circumscribed lesions and CSI for diffuse infiltrative lesions.Post contrast T1-weighted MPRAGE images were used for positioning of the volume of interest (VOI).Typical VOIs sizes were 100 × 80 mm 2 .VOIs for SVS were placed at image regions to show post-contrast enhancement.For CSI, voxels which showed the greatest departure of Cho/Cr from the values for normal appearing white matter were selected.To suppress the water signal, chemical shift selective saturation was applied.The acquisition time was approximately 7 min for CSI and 4 min for SVS.Spectroscopic data were processed with the Syngo v14 software implemented on MR imaging console.The processing algorithm included the application of a Hanning filter, baseline and phase correction.Metabolite peak areas were obtained after all the observed resonances in the spectra were fitted.

DWI
An echo planar (EPI) SE sequence (TR/TE 3808/89 ms, slice thickness 5 mm, matrix 128 × 128, NEX 2, FOV 24 cm) was used for obtaining diffusion-weighted images; b value of 0 s/mm 2 , as a reference, and b values of 1,000 s/mm 2 were included in DWI sequences.We used DWI to calculate corresponding ADC maps.ADC values were calculated by using the equation: ln(S/S 0 ) = -bADC where b is a diffusion sensitivity factor, S is a signal at b = 1,000, S 0 is a signal at b = 0 and ADC was previously explained.
The positions of regions of interest (ROIs) placed on the ADC maps corresponded as much as possible to the location of spectroscopic VOIs.The ROIs size varied from 20 to 44 pixels.

Statistical analysis
The SPSS 12.0 for Windows was used for statistical analysis.Unpaired two-tailed Student's t-test was used for comparison of 1 H-MRS metabolite ratios (Cho/Cr, Cho/NAA, NAA/Cr, mIn/Cr) and ADC values (ADC, ADC and ADC/ADC) between the groups (non-neoplastic lesions, glial tumors grade II and III) and all seven different pathologies (demyelinating lesions, ischemic lesions, hamartomas, diffuse astrocytomas, oligoastrocytomas, oligodendrogliomas and anaplastic astrocytomas).The significance level was set to be p < 0.05.

Results
The results obtained for metabolite ratios and diffusion parameters are summarized in Table 2.
Data analysis showed that the Cho/Cr ratio was significantly higher in glial tumors grade II compared to nonneoplastic lesions (p = 0.008) and that the Cho/Cr ratio was significantly higher in glial tumors grade III than in nonneoplastic lesions (p = 0.001) (Figure 1).The Cho/NAA ratio was significantly higher in glial tumors grade II than in nonneoplastic lesions (p = 0.37).
NAA/Cr, and mIn/Cr ratios could not differentiate between non-neoplastic lesions, glial tumors grade II and III.
In two cases of anaplastic astrocytoma ratios NAA/Cr were increased compared to the normal values, which can be assigned to contribution of Glx resonances.Therefore, they were excluded from evaluation of NAA/Cr and Cho/NAA ratios in statistical analysis.
Comparison of ADC, ADC and ADC/ADC values between tumors and non-neoplastic lesions showed a statistical significance of ADC/ADC between glial tumors grade II and glial tumors grade III (p = 0.023).
Axal T 2 weighted MR imaging of anaplastic astrocytoma, demyelinating lesion, ischemic lesion and hamartoma respectively shown in Figure 4a-d while metabolic ratios of

Discussion
Differentiation of non-neoplastic lesions, which look similar to neoplastic, on conventional MR images presents a particular challenge regarding establishing the correct diagnosis and following treatment.Advanced MR imaging techniques, like MR spectroscopy and DWI, which show physiological status of tissue, may contribute to better characterization of those pathologies.When brain lesion is solid, without necrosis, the main diagnosis include beside glial tumors (grade I-III), pseudotumoral demyelinating disease and some ischemic lesions with atypical presentation 7 .
Our research showed that the Cho/Cr ratio is higher in glial tumor grade III than in demyelinating lesions and NAA/Cr ratio is lower in oligoastrocytoma (grade II) than in demyelinating lesions (Figure 5c, d, g, h).These results can be explaned by the higher loss of funcional neuronal cells and the larger membrane turnover in glial tumors compared to demyelinating lesions 7 .Cho is a component of the phospholipid metabolism of cell membranes and its increase is related to cell membrane turnover and higher cell density from tumor proliferation 14 .NAA is a neuronal marker and a decrease of NAA levels is caused by replacement of healthy brain tissue by tumor cells 14 .Brain tumor 1 H-MR spectroscopy typically shows elevated Cho levels and reduced NAA levels 19 .MR imaging findings of acute demyelinating lesions can mimic glial neoplasms especialy tumefactive demyelinating lesions 2,20 .They present as T1 hypointense and T2 and FLAIR hyperintense lesions similar to tumors and can show enhancement after administration of contrast agent because of inflammatory BBB breakdown 21 .Acute demyelinating lesions are also caracterized by the increase of Cho levels and decrease of NAA levels 2,19 .This is due to inflammation, demyelination and intense reactive astrogliosis 21 .Bitsch et al. 22 found that elevated Cho levels correlate with glial proliferation, since Cho is a component of glial cell membranes and that there is a connection between the Cho level, neuronal dysfunction and patient's disability.The decrease of NAA is also common finding in acute demyelinating lesions.Also, Bitsch et al. 22 showed that axonal degeneration and decreased axonal density, characteristic for demyelinating process, are associated with decreased NAA.Majos et al. 7 found that elevated Cho levels and reduced NAA levels are more pronounced in brain tumors than in pseudotumoral demyelinating disease.Therefore, analysis of these metabolites values can help in differentiation between glial neoplasms and acute demyelinating lesions 7 .Our findings are in correlation with former published data.
In our research, we found that Cho/Cr ratio is higher in glial tumor grade III than in ischemic lesions.This is because of more intense cell destruction and glial proliferation in glial tumors grade III than in ischemic lesions.Based only on MR imaging examination, ischemic lesions can less frequently be a diagnostic problem for differentiation from glial neoplasms 2,9,18 .On 1 H-MR spectroscopy, infarcts typically display with the reduction of NAA level and the elevation of lactate level and a slight increase in choline 23 .In this research we did not observe lactate peaks.Loss of neuronal cells leades to decreas of NAA and increase of Cho level is due to reactive gliosis 23,24 .The intensity of these metabolite changes reflects the severity of an ischemic process and it is related to the prognosis 24 .Moller-Hartmann et al. 18 found that Cho is a metabolite which can be used for differentiation between ischemic lesions and glial tumors since Cho level observed in glial neoplasm is significantly higher than in an ischemic process.Our findings are in accordance with the previously reported.
By analyzing the spectra obtained in our study we found higher Cho/Cr ratios in anaplastic astrocytomas than in hamartomas and lower NAA/Cr ratio in diffuse astrocytomas than in hamartomas.A lower NAA/Cr ratio in diffuse astrocytomas than in hamartomas could be due to neuronal loss that is more pronounced in glial tumors and higher Cho/Cr ratios in anaplastic astrocytomas than in hamartomas could be explaned by intense tumor glial proliferation in high grade gliomas compared to a glial component within the hamartomas as benign lesions 25 .Hamartomas, on MR imaging, appear as isointense to gray matter on T1 and T2-weighted images, but in more recent studies they have been described as T2 hyperintense lesions 26,27 .On 1 H-MRS, hamartomas present with decrease in NAA/Cr and increase in Cho/Cr and mIn/Cr ratios 26,27 .Since, NAA is a neuronal marker, its decrease is connected with neuronal loss.Reflecting gliosis is related to the increase in mIn 27 .Most commonly, elevated Cho is associated with high -grade glomas, but this also can be found in benign cerebral pathologies like hamartomas 26 .Cho level increase could be due to increasing glial component within the tumor 26 .Majos et al. 7 and Moller-Hartmann et al. 18 found that addition of spectroscopy to routine MR imaging exam helps in characterization of focal intracranial disease and improves decision making in cases suggestive of brain tumors.Our results are in accordance with data in the literature and suggest that 1 H-MRS is useful in evaluation of solid brain masses.
Our research showed a significant difference in ADC/ADC ratio between glial tumors grade II and glial tumors grade III.These findings can be due to the fact that high-grade tumors are characterised by the increased cellularity, microvascular proliferation and/or necrosis, that diffusivity of glial tumors is inversely related to the cellularity and that ADC is inversely proportional to the cellular density 28,29 .Diffusion of a free water molecule in high grade tumors is reduced because of reduction in extracellular space by increased cellularity 30,31 .The areas with the lowest ADC value suggest the areas with the highest cellular density and the highest tumor malignant potential 12 .Previous studies showed that DWI can be useful in differentiating benign and malignant tumors from normal parenchyma and in grading gliomas 4,12,30,32 .The results of our research correspond with previous findings from the literature.
This study is limited by the small sample size (47 patients).Different tumor types in a group of glial tumors grade II (diffuse astrocytoma, oligoastrocytoma, oligodendroglioma) is another limitation of this research.Pure astrocytic tumor differs from oligoastrocytoma and oligodendroglioma in its therapeutic response to chemotherapy, so their distinction is of great importance 15

Conclusion
Our study showed the potential use of 1 H-MRS and DWI in evaluation of solid brain masses.These noninvasive diagnostic techniques have the advantage over histopathologic assessment of focal brain lesions since they allow in vivo examination.ADC/ADC, Cho/Cr and NAA/Cr ratio provided additional valuable information on lesion metabolic structure that can help distinguishing brain tumors from nonneoplastic lesions and tumor grading.

Fig. 1 -Fig. 2 -
Fig. 1 -Box plot of the choline-to-creatine ratio (Cho/Cr) shows a significant difference between non-neoplastic lesions and glial tumors grade II and between non-neoplastic lesions and glial tumors grade III.NNL -non-neoplastic lesions; GTII -glial tumors grade II; GTIII -glial tumors grade III.