Proton Magnetic Resonance Spectroscopy in Intracranial Gliomas

Chapter 8

Proton Magnetic Resonance Spectroscopy in Intracranial Gliomas Eftychia Z. Kapsalaki, Ioannis Tsougos, Kyriaki Theodorou, and Kostas N. Fountas

Abstract Recent MRI advances have focused on the development and application of molecular and physiological imaging capabilities. One of these relatively new MRI methods, Magnetic Resonance Spectroscopy (MRS) reflects the continuing evolution from purely anatomic to physiological and molecular imaging of the brain. Magnetic Resonance Spectroscopy yields images of the distribution and concentration of naturally occurring molecules such as N-acetyl aspartate (NAA) (one of the most abundant amino acids in the brain), choline (Cho) (a key constituent of cell membranes), lactate (Lac) (a reflection of anaerobic metabolism) and Creatines (Cr). It has been suggested that the sum of Cr and Phosphocreatine is relatively constant in the human brain, and for this reason Cr is often used as a reference signal, and it is a common practice for metabolite ratios to be expressed as a ratio relative to Cr. However, with the development of quantitative analysis techniques, it is clear that total Cr is not constant, both in different brain regions and in pathological processes, so the assumption of Cr as a reference signal should be used with caution. It is well known that gliomas represent the most common type of primary intracranial tumors. The establishment of an accurate diagnosis, the preoperative evaluation of tumor metabolism and the obtaining information regarding tumor histological grade, may increase the efficacy of the currently employed treatments and eventually improve the overall clinical outcome in gliomas cases. Magnetic

K.N. Fountas () Department of Neurosurgery, University Hospital of Larisa, School of Medicine, University of Thessaly, Larisa, Greece email:[email protected]

Resonance Spectroscopy may substantially improve the non-invasive categorization of human brain tumors, especially gliomas. The utilization of MRS (coupled to conventional MRI techniques) in the evaluation of tumors provides greater information concerning tumor activity and metabolism. In addition, it may provide valuable information regarding the tumor response to the employed treatments. Keywords Choline · Creatine · Glioma · Lipids · MR spectroscopy · N-acetyl-aspartate

Introduction Magnetic Resonance Imaging (MRI) is considered the method of choice for the identification and the evaluation of brain tumors. The MR characteristics of brain tumors depend on their consistency and their histopathology. The role of MRI in the evaluation of patients with intracranial masses is to identify the lesion, provide a differential diagnosis and contribute to the selection process of the best therapeutic approach for these patients. Since many brain tumors have similar imaging characteristics, advanced MR techniques have been introduced and may further attribute to the differential diagnosis of these lesions. Such techniques include Diffusion Weighted Imaging (DWI), fractional anisotropy and tractography (DTI), perfusion weighted Imaging (pWI), and Magnetic Resonance Spectroscopy (MRS). Use of these advanced methodologies in the preoperative evaluation of gliomas, may significantly narrow the differential diagnosis and minimize the possibility

M.A. Hayat (ed.), Tumors of the Central Nervous System, Volume 1, DOI 10.1007/978-94-007-0344-5_8, © Springer Science+Business Media B.V. 2011

67

68

of misdiagnosis, particularly in the cases of ring enhancing lesions. In addition, the establishment of an accurate diagnosis may clarify some occasionally perplexing issues, regarding the best surgical planning for these patients. It is well known that gliomas represent the majority of primary intracranial brain tumors. They may be circumscribed (WHO I), or diffuse and infiltrative (WHO II-IV). The degree of malignancy increases as pyrinokinesias and mitotic activity increase, and necrosis and neovascularization develop in these lesions. Preoperative imaging evaluation needs to provide accurate information regarding the extent of the lesion, which is better identified after contrast administration, but also to detect the presence of any neoplastic cells in the peritumoral edema. In a large number of cases, infiltrative gliomas are characterized by a remarkable heterogeneity and significantly varying stages of malignancy in different areas of the same tumor. The employment of advanced MR imaging, using DWI, pWI and MRS, may accurately identify the areas of highest malignancy in a tumor, and thus navigate the surgeon to the most aggressive area of this tumor. Moreover, MRS provides information regarding tumors’ biochemical profile, by measuring specific metabolite concentrations and their ratios, and thus may potentially predict their histological grade. Furthermore, identification of the presence of tumor cells in the peritumoral edema, contributes to a more aggressive surgical resection, and a more efficient post operative radiation planning. In addition, the question of postsurgical radiation necrosis and/or interstitial brachytherapy associated necrosis versus tumor recurrence constitutes a puzzling issue in the management of patients with surgically resected gliomas. Magnetic Resonance Spectroscopy combined with pWI may be useful, particularly in those cases that these two methodologies provide concordant results. In this chapter, we provide a brief overview of the MRS general principles, a brief description of the necessary hardware and software for performing MRS and its technical limitations. We outline the usually recognized metabolites in normal brain and other pathological conditions, we describe the most frequently used metabolic ratios in clinical practice, we analyze the spectral characteristics of gliomas, and we also refer to the future directions of MRS.

E.Z. Kapsalaki et al.

General Principles of Magnetic Resonance Spectroscopy Proton Magnetic Resonance Spectroscopy (also Hydrogen−1 MRS, or H1 MRS) is the application of nuclear magnetic resonance with respect to Hydrogen-1 nuclei within the molecules of a substance. Since the atomic number of hydrogen is 1, a positive hydrogen ion (H+ ) has no electrons and corresponds to a bare nucleus with one proton. Regarding clinical applications, water is the biggest source of protons in the human body, therefore, allows the application of brain H1 MRS. Magnetic Resonance Spectroscopy depends on the chemical shift theory, which corresponds to a change in the resonance frequency of the nuclei within the molecules, according to their chemical bonds. The presence of an electron cloud constitutes an electronic shield, which slightly lowers the static magnetic field to which the nucleus would normally be subjected. Thus, same nuclei will resonate in different frequencies, according to which molecular group they belong to, as they experience this different “shielding effect”. This resonance frequency difference (chemical shift) is expressed as parts per million or ppm, a value that is independent of the amplitude of the magnetic field. Thus, the value of the chemical shift provides information about the molecular group carrying the hydrogen nuclei, and therefore provides differentiation among several metabolites. Therefore, within a certain region of interest, ideally a voxel of at least 8 cm3 , it is possible to gather information on these molecular groups and present it as a spectrum. In this spectrum, the x-axis (oriented from right to left) represents the precession frequency, which differentiates the identity of a certain compound. On the other hand, the intensity on the y-axis can be used to quantify the amount of a substance, although this is a matter of great dispute and includes serious risks that should be taken into account. To obtain reliable absolute concentrations, one has to consider potential complicating factors, with respect to both the data acquisition method and the data processing method. For example, relaxation effects in data acquisition can be either corrected or eliminated, whereas data fitting is complicated by factors such as the contribution of macromolecules (Jansen et al., 2006).

8

Proton Magnetic Resonance Spectroscopy in Intracranial Gliomas

Peaks of several molecular groups on the spectrum are also called resonances. Some metabolites do not have simple resonances, but may be split into two (called doublet), three (triplet) or even more sub-peaks. As spins can be considered to be small magnets, they interact with the main magnetic field, thus the degree of interaction, known as spin-spin coupling, causes a peak to split into more than one sub peaks. The frequency separation of each peak and the clear representation of its characteristics depends on the field strength, on the magnetic field homogeneity, the degree of chemical shift, the samples’ chemical composition, and on the digital resolution i.e. the precision with which the signal is sampled. At low fields with a poor shim, the peaks tend to overlap and that causes difficulty in interpreting and in measuring peak heights. In proton MRS (H1 MRS), the water signal must be suppressed, as it is much greater than the signal from other H1 containing compounds and it has overlapping spectroscopic peaks. The reference frequency used, set at zero ppm, is that of the standard tetra-methyl silane Si-(CH3 )4 , which has a single proton resonance due to its a completely symmetrical molecule. It should be mentioned here, that above 4 ppm the spectrum becomes unreliable, since the suppression of the water peak at 4.7 ppm tends to destroy the neighbouring portions of the spectrum.

Proton MRS Apparatus In-vivo H1 MRS uses MRI apparatus that is virtually identical to that used in routine imaging, but nevertheless requires: • A sufficiently strong and very homogenous magnetic field (at least 1.5 T), to distinguish resonance peaks, shimming ideally less than 0.5 ppm over the central 20 cm Diameter of a Spherical Volume (DSV). • Specific sequences for spectroscopic signal acquisition. There are two types of H1 MRS: Single Voxel Spectroscopy (SVS), which receives the spectrum from a single voxel only, and Chemical Shift Imaging (CSI), which measures spectra in a single dimension projection (1D), on a two dimension slice (2D), or a three dimensional volume (3D)

69

• Adapted data processing software, and • An adapted radiofrequency system (in the resonance frequency of the studied nucleus).

Single-Voxel (SV) Spectroscopy In SV spectroscopy, the signal is received from a volume limited to a single voxel. This acquisition is fairly fast (1–5 min) and a spectrum is easily obtained. It is performed in three steps. Initially, the suppression of the water signal is performed, followed by the selection of a voxel of interest, and lastly by the acquisition of the spectrum. For the spectra acquisition there are two types of sequences available: (a) PointRESolved Spectroscopy (PRESS) and (b) STimulated Echo Acquisition Mode (STEAM). It is worth mentioning that the analyzed volume is selected by three selective radiofrequency pulses (accompanied by gradients) in the three directions in space (either 90◦ –90◦ – 90◦ in STEAM, or 90◦ –180◦ –180◦ in PRESS). These pulses determine three orthogonal planes, whose intersection corresponds to the studied volume. Only the signal within this voxel will be recorded, by selecting the echo resulting from a series of the three studied radiofrequency pulses.

Chemical Shift Imaging (CSI) Chemical shift imaging (CSI) consists of spectroscopic data of a group of voxels, in slice(s) (2D) or in a specific volume (3D). It is based on a repetition of STEAM or PRESS type sequences to which a spatial phase encoding is added. The number and direction of phase encodings depend on the number of dimensions explored, having as a consequence though, a longer acquisition time. The results appear in the form of a matrix of the obtained spectra from the studied regions, or as parametric images (metabolic maps). However, it has to be emphasized that CSI has several disadvantages. In this technique, voxels of interest (VOIs) are much larger than those in SVS, so it is more difficult to achieve magnetic field homogeneity. Localization of multiple VOIs is not as accurate as localization via a single voxel in SVS,

70

as phase encoding causes much more spin dephasing. In addition, adjacent voxels’ interference can add up to 10% of their signal to the voxel of interest, thus degrading the obtained spectra. As a result, there is a signal loss of approximately 13% per direction of phase encoding, which has to be appropriately multiplied in the case of 3D CSI. Another disadvantage of 3D CSI is the increased acquisition time, which can last up to 15 min.

Limitations of MR Spectroscopy Though a promising advance, several technical limitations potentially compromise the efficacy of MRS as a diagnostic modality. Intraparenchymal calcification, contamination from adjacent bone and fat (from the skull), cerebrospinal fluid (CSF) or the presence of intratumoral hemorrhage may alter the MRS signal and, if inadvertently included in an assessed voxel, may confound the obtained results. In order to obtain a good and accurate H1 MRS analysis the following guidelines should be scholastically followed: good magnetic homogeneity, good water suppression, proper localization in respect to the lesion, well optimized pulse sequences, and finally a cooperative patient. To obtain good homogeneity, excellent shimming is a prerequisite, and calibration of water suppression must be exceedingly efficient. This allows a good analysis of the metabolite intensities. Another important issue is to obtain adequate signal-to-noise ratio in order to permit reproducible peak area integration, avoiding artefacts. Finally, the appropriate voxel size should be selected, and has to be large enough to obtain a detectable metabolite signal, thus avoiding distortion from other peaks in the spectrum and should not be placed near the skull, to avoid contamination from bone and fat. However, when the voxel size has to be very small, the obtained spectrum requires a larger acquisition time, in order to obtain good shimming and water suppression. Stronger magnetic fields, as 3 T, can spread out precession frequencies over a wider range and may double spatial and temporal resolution. Moreover, stronger magnetic fields may also allow the detection of compounds that are currently considered to be not clearly detectable with lower magnetic fields. Nevertheless, at

E.Z. Kapsalaki et al.

higher magnetic fields, water is much more difficult to be suppressed, due to the increased magnetic field inhomogeneity, poorer radiofrequency coil efficiency, poorer shimming, and different relaxation times. Kim et al. (2006) have concluded that although there is an increased signal to noise ratio at 3 T compared to 1.5 T, a better spectral resolution at a short TE (35 ms) but not at long TE (144 ms), there was no significant difference in the metabolite ratios at 3 T. Their findings are in agreement with our experience from using 3 T MRI and MRS. Another issue that needs to be taken into consideration is the usage of commercial MRS analysis packages, which may be user-friendly but should be used with extreme caution, since in these packages there are metabolites, which are not detected in a routine brain spectrum (Xu et al., 2008).

Proton Magnetic Resonance Spectroscopy Brain Metabolites Choline (Cho) At 3.2 ppm a prominent resonance arises from the methyl protons of choline (Cho) containing compounds. It constitutes a structural compound of the cell membrane and is routinely detected in a normal brain spectrum, but in small concentrations. Choline signals in brain tumor studies are considered as a surrogate marker of cellular membrane turnover. Ex vivo studies of perchloric acid extracts of intracranial tumor tissue indicate that Cho spectral signal represents mostly free Cho, phosphocholine (PC) and glycerophosocholine (GPC) molecules (Miller et al., 1996). Nevertheless, at routine clinical magnetic field strengths, these three compounds are seen as a single resonance, as they cannot be resolved, hence they are referred as the total Cho signal. Possible causes for the elevated Cho signal seen in brain tumors can be separated into: (a) intracellular and (b) extracellular mechanisms. In the intracellular environment, Cho may be elevated as a result of enhanced transport, phosphocholine may be elevated as a result of accelerated phosphorylation and transport, and all three compounds (Cho, PC and GPC) may be elevated as a result of cellular membrane break down. In the extracellular environment

8

Proton Magnetic Resonance Spectroscopy in Intracranial Gliomas

and in the blood, Cho may be elevated as a result of hyperperfusion and/or systemic metabolic alterations. According to several in vitro MRS studies based on brain tumor biopsy material, increased intracellular PC is the predominant cause for the Cho signal elevation (Tzika et al., 2002). It has been demonstrated, that PC increases with cell proliferation rate, and that glioma cells have enhanced Cho transport and phosphorylation during their growing phase (Podo, 1999). Recently, the technique of 1 H high-resolution magic angle spinning (HR-MAS) has been shown to produce high-quality data, allowing the accurate measurement of many metabolites present in unprocessed biopsy tissue, allowing better correlations between in vivo and in vitro studies (Wilson et al., 2009). It has to be emphasized however, that the exact interpretation of changes in total choline signal is complicated, due to multiple contributions to the observed total choline spectrum resonance. Therefore, more general terms as increased cellular membrane turnover or altered cellular membrane metabolism, are frequently used to explain elevated Cho in gliomas.

Creatine (Cr) At 3.03 ppm a composite peak arises from the methyl and methylene protons of Cr and phosphorylated creatine (PCr), often called total Creatine (tCr). In the brain, Cr and PCr are present in both neuronal and glial cells, and together with ATP play a crucial role in the energy metabolism of neuronal tissues. Total Cr constitutes a metabolite routinely detected in normal brain spectrum, and is considered to be a measure of cellular density, while is especially high in glial cells. The concentration of total creatine is relatively constant, with no changes reported with brain aging. As such, its resonance is frequently used as an internal concentration reference and other metabolites are represented as ratios to Cr. While convenient, the use of any internal concentration reference should be used with extreme caution. There appears to be a substantial variation in Cr concentrations between grey and white matter, as well as within individual tumors of a certain type. Although under most conditions, the use of Cr as a constant is reliable, previous absolute quantification studies have demonstrated that in a number of brain

71

tumours, Cr levels were found to be significantly less than those in normal brain (Howe et al., 2003), and are decreasing with increased grade of malignancy.

N-Acetyl Aspartate (NAA) At 2.01 ppm a prominent resonance arises from the methyl group of NAA. In general, NAA is considered to be a highly specific neuronal marker, reflecting the number of intact neurons in the gray matter, and the density of intact axons in the white matter. NAA is exclusively localized in the central and peripheral nervous system. Its concentration varies in different parts of the brain, and undergoes significant changes following any developmental central and peripheral nervous system changes. It has been reported that NAA resonance is a marker of neuronal density, and its decreased concentration in tumors reflects neuronal loss (Majos et al., 2004). The reported NAA concentration reduction in brain tumor spectra is typically attributed to the displacement and destruction of neuronal tissue, which accompanies gliotic infiltration (Birken and Oldendorf, 1989). Indeed, the vast majority of in vivo H1 MRS studies have demonstrated that NAA signal is markedly decreased in brain tumor spectra, although it is not unusual to detect a small residual amount of NAA. When NAA is present in a brain tumor spectrum, it is difficult to exclude the possibility that a small amount of NAA-containing (normal) brain tissue has accidentally been included in the sampled volume. It needs to be pointed out however, that NAA also decreases in other pathological conditions such as in the chronic stages of stroke (Gideon et al., 1992), or in multiple sclerosis (Simone et al., 2001). Moreover, NAA levels may reflect neuronal dysfunction rather than actual neuronal loss. This is substantiated by recovery of NAA levels in cases of incomplete reversible ischemia (Brulatout et al., 1996) or reversible traumatic brain injury (Sinson et al., 2001).

Alanine (Ala) At 1.47 ppm a prominent resonance arises from the three methyl protons coupled to a single methane in the alanine molecule. Alanine is known to be the substrate of several amino acid transporters. The signal from

72

Alanine is usually overlapped by lipids, therefore, the use of spectral editing or the application of longer echo times is necessary, in order to distinguish these two resonances. Alanine is a non essential amino acid, with its main function being in the metabolism of tryptophan and pyridoxine. It also assists in the metabolism of sugars and organic acids, providing energy for muscle tissues and brain, without any significant contribution however, to the actual energy metabolism, but mainly to the carbon recycling process (Broer et al., 2007). Yudkoff et al. (1986) found, in cultured astrocytes, an important flux of nitrogen from glutamate to alanine, corresponding to net alanine synthesis. They proposed that alanine released from astrocytes is utilized by neurons to synthesize glutamate, whereas we would suggest that in higher eukaryotes, alanine may actually be the substrate that feeds the neurons to sustain their energy needs. The total concentration of Ala is very low and is almost not detectable in a normal brain spectrum. It is not usually detected in gliomas, but has been expressed in tumors of meningeal origin, and it may be a distinct metabolite for meningiomas. Manton et al. (1995) suggested that the presence of ala in meningiomas may indicate that their metabolism involves partial oxidation of glutamine rather than glycolysis. Nevertheless increased Ala has also been observed in cases of evolving ischemia.

Lipids (Lip) Lipid signals arise at about 0.9 and 1.4 ppm. Lipids are normally absent from a normal MR brain spectrum. They have short transverse relaxation times thus, are relatively easier to be detected when using short TE SVS (35 ms or less). The appearance of lipid resonances in a tumor usually represents necrosis, thus Lip are present in high grade gliomas, which typically have a larger necrotic fraction than lower grade tumors. Lipid concentration increases with the increased cellularity of the tumor, however, Lip are not always specific and reproducible markers. Since free lipid signals appear to be associated with necrotic areas, they may indicate high degree of malignancy or a tumor of metastatic origin. Therefore, detection of lipids is indicative of a malignant tumor. Gotsis et al. (1996) have reported that lipids are detected in anaplastic

E.Z. Kapsalaki et al.

gliomas and GBM, in metastatic lesions, but also in abscesses of bacterial origin. They are usually not detected in low grade tumors, but may be found in lipid containing meningiomas. It is widely known that lipids remain a perplexing finding in brain H1 MRS.

Lactate (Lac) At 1.33 ppm lactate arises as a doublet resonance from the three equivalent methyl protons, while the single methine proton resonates as a quartet at 4.10 ppm. Lac (or lactic acid) is the end-product, hence the reflection, of anaerobic glycolysis. In normal human brain, lactate is below or at the limit of detectability in most studies. Its absence may conversely correlate with increased neo-vascularization of a highly aggressive malignant tumor. Increased Lac concentrations have been observed in a wide variety of conditions that are associated with restricted blood flow, hence hypoxia, such as ischemic stroke (Barker et al., 1994), mitochondrial myopathy, encephalopathy, and lactacidosis. It is also present in aggressive tumors (Negendank et al., 1996), and abscesses. The increased concentration of Lactate and lipids, are consistent with rapid tumor growth leading to hypoxia, hypoxic stress, and finally to central necrosis, as the tumor outgrows its blood supply. Although Lac concentration is considered to increase with astrocytoma grade (Auer et al., 2001), the statistical significance of this finding is questionable, due to its high variability within tumor groups. Moreover, lactate and lipid levels, as a marker of tumor grade, have shown varied results in long-TE studies (Negendank et al., 1996). Lipids and lactate may coincide since they appear in the same region of the spectrum. These resonances are better distinguished at higher magnetic fields and by using CSI, which produces inversion of the lactate peak but not of the lipid peak.

MyoInositol (mI) MyoInositol is a rather complex sugar alcohol that gives rise to four groups of resonances. The main resonance peak can be seen at 3.56 ppm. MyoInositol is located in astrocytes and is considered to be a marker of glial cells, but its exact function is not known.

8

Proton Magnetic Resonance Spectroscopy in Intracranial Gliomas

It has been found that, mI is elevated in low-grade astrocytomas and demyelinating lesions (as multiple sclerosis). Castillo et al. (2000) reported that elevated mI may be a marker of low-grade astrocytomas. Recent magic-angle spinning (MAS) studies of whole-tumor biopsies have demonstrated that mI is elevated in astrocytomas, and its concentration decreases as malignancy grade increases (Cheng et al., 1998). Moreover, in a recent study by Kallenberg et al. (2009) it was shown that in the contralateral normal-appearing white matter, mean myo-inositol levels were significantly increased in patients with GBM compared to the levels of control subjects. Similarly, mI levels were higher in patients with GBM, compared to those of patients with low grade gliomas. Hence, increased concentrations of mI in the contralateral normal-appearing white matter of GBM patients are consistent with mild astrocytosis, and may suggest an early widespread neoplastic infiltration. A normal brain spectrum is characterized by a marked concentration of NAA and presence of Cho

73

and Cr (Fig. 8.1). Other metabolites, as Ala, Lip or Lac, are not detectable in a normal brain spectrum. Pathological brain areas show variable amounts of decreased concentration of NAA. Depending on the histological profile of the lesion, there may be variable amounts of Cho, Cr, Lip, Lac, Ala, mI, or other metabolites. The absolute amount of these metabolites, but more importantly their ratios, may characterize a brain lesion. Benign and low grade lesions present NAA but in lower concentrations than the normal values (Table 8.1). Absolute metabolite values vary significantly in different normal individuals; therefore, evaluating their ratios is a more accurate marker in grading gliomas, than their absolute values. It is always helpful to compare metabolite values in both hemispheres. In establishing an H1 MRS diagnosis, the ratios of NAA/Cho, NAA/Cr and Cho/Cr are being calculated (Table 8.2). All other metabolites are being identified in pathological conditions. The principal metabolites that are most commonly evaluated in Proton Magnetic Resonance Spectroscopy

Fig. 8.1 MRS normal spectrum, demonstrating a dominant peak corresponding to NAA at 2 ppm, a peak at 3.03 ppm corresponding to Cr and a peak identified at 3.2 ppm corresponding to Cho

74

E.Z. Kapsalaki et al.

Table 8.1 The concentrations of the characteristic metabolites in gliomas

NAA Cho Cr Cho/Cr

Table 8.2 Approximate ratios of the normally detected metabolites in gliomas are shown, as reported by Fountas and Castillo

Low grade (II)

Anaplastic (III)

GBM (IV)

Low High Low High

Very low Very high Low Very high

Very low or absent Very very high Very low Very very high

NAA/Cr Fountas et al. NAA/Cr Castillo et al. NAA/Cho Cho/Cr Fountas et al. Cho/Cr Castillo et al. mI/Cr Castillo et al.

Normal brain

Low grade (II)

Anaplastic (III)

GBM (IV)

1.52 1.26 1.74 1.51 0.62 0.49

Increased 1.23 Decreased 2.15 1.06 0.82

Increased 1.48 Decreased 2.78 1.48 0.33

Very increased 3.24 Very decreased 5.40 2.08 0.15

Table 8.3 Summary of the principal metabolites that are most commonly evaluated in proton magnetic resonance spectroscopy (H1 MRS) Frequency/Cerebral Metabolite concentration Physiological role TE Variation impact Cho

3.2 ppm/0.9–2.5 mmol/kg

Cr Creatine/Phosphocreatine

3.0 ppm/5.1–10.6 mmol/kg

NAA N-Acetyl-Aspartate

Marker of cell membrane metabolism.

Long/short

Compounds related to Long/short energy metabolism marker. 2.02 ppm/7.9–16.6 mmol/kg Neuronal cell marker. Long/short Only seen in healthy nervous tissue.

Ala Alanine

1.5 ppm/0.2–1.4 mmol/kg

Lipids Free Lipids

0.9, 1.4 ppm/> 1.0 mmol/kg

Lactate

1.33 ppm/0.4 mmol/kg

Myo-inositol (mI)

3.6 ppm/3.8–8.1 mmol/kg

Is characteristic Long/short of mengingeal tumors Membrane breakdown Long/short product.

A product of anaerobic Long glycolysis. Glial Marker Long/short

(H1 MRS) of the brain and a brief yet comprehensive explanation of their role in determining the obtained spectrum and potentially predicting their histopathological type and grade are summarised in Table 8.3.

↑ Tumors, inflammation, hypoxia, demyelization. Caution: Higher in white matter than Grey matter. Serves as reference peak as it is relatively constant. ↓ Hypoxia, Stroke, Tumors ↓ Neuronal dysfunction. ↓ Ischaemia, trauma, inflammation, infection, tumors, dementia, gliosis. Expressed in tumors of meningeal origin, may be discriminant metabolite. Indication of histological necrosis. High grade tumors, metastatic lesions Nonspecific marker of tumor aggressiveness. A diagnostic modifier in diseases that affect Cho. ↓ as glioma grade increases

Spectroscopic Tumor Profile It is well known that gliomas are infiltrative tumors, which may extend and spread along the white matter tracts to the adjacent brain tissue, and frequently

8

Proton Magnetic Resonance Spectroscopy in Intracranial Gliomas

invade neighboring lobes, and even the contralateral hemisphere (Mikkelsen and Edvardsen, 1995). Gliomas are categorized into four grades: Pilocytic astrocytomas (WHO I), astrocytomas (WHO II), anaplastic astrocytomas, oligodendrogliomas and mixed glial tumors (WHO III) and Glioblastomas Multiforme (WHO IV). Anaplastic astrocytomas and GBM are more heterogeneous, may produce mass effect, and show variable degree of contrast enhancement. Gliomas frequently need to be differentiated from other tumors such as solitary metastases, lymphomas, but also non neoplastic lesions such as an abscess, or an evolving infarct. There has been interest in using proton MR spectroscopic imaging (H1 MRS) in the evaluation of the tumor borders and disease burden (Croteau et al., 2001). Moreover, several studies sought to determine a correlation between different proton H1 MRS metabolic ratios and the degree of tumor infiltration, and examined the role of H1 MRS in suggesting a preoperative histological diagnosis. An important contribution of H1 MRS is to identify the metabolic profile and the histological grade of a brain glioma. Negendank et al. (1996) and Fountas et al. (2004) concluded that glial tumors have significantly elevated Cho signals, and decreased Cr and NAA signal, compared to normal brain. Choline signal intensities were extremely high, while Cr signal intensities were very low, in anaplastic astrocytomas and GBMs in their series. They also described that anaplastic astrocytomas and GBMs demonstrated an increased Cho/Cr ratio (Table 8.2). Lipids may be detected in all grades of gliomas, however they usually present in higher concentrations in anaplastic astrocytomas and GBMs. Significant variability has been observed regarding the presence or absence of Lip peaks in low-grade gliomas. Likewise, the concentration of Lac varies significantly among patients with astrocytomas of the same histologic grade. Similarly, mI shows great variability in astrocytomas. The ratio of MI/Cr is a marker reported by Castillo et al. (2000) to be higher in low-grade astrocytomas, intermediate in control subjects, and lower in patients with anaplastic astrocytomas and GBMs. Despite the accurate measurement of absolute metabolite concentrations, it seems that their ratios may be a more important marker in grading gliomas. Regarding the metabolite ratio findings, Fountas et al. (2004) reported NAA/Cr ratio to be decreased in all

75

gliomas however, they found no correlation between the degree of reduction and the tumor’s histological grade. Likewise, the ratio of NAA/Cho is found to be inconsistent among tumors of the same histologic grade. Contrariwise, the Cho/Cr ratio is found to be a reproducible and consistent marker among astrocytomas of the same histologic grade. This metabolite ratio is lower in low-grade astrocytomas (grades II and III), but almost two-fold increased in GBMs. This finding suggests that the higher the Cho/Cr ratio is, the higher the tumor grade will be (Table 8.3). The ratios of Cho/Cr in grade II and III gliomas may not vary significantly, but in these cases, their conventional MRI characteristics may be helpful in their differentiation, since in grade III astrocytomas contrast enhancement is present in the vast majority of cases (Figs. 8.2 and 8.3). Moreover, the presence of lip in anaplastic astrocytomas may suggest a more aggressive histological type. Nevertheless, gliomas need to be differentiated from other benign or malignant lesions. The most common differential diagnosis of glioma grade II and III is ischemia and infection, particularly when located in the temporal lobe. The concentration of lactate is increased in ischemia, and additionally the presence of lipids suggests necrosis in an infarcted brain area. Moreover, DWI may help differentiate acute/subacute ischemia from a low grade tumor, since ischemia shows markedly decreased diffusion. It cannot be overemphasized the importance of the patient’s clinical history and symptomatology in the differential diagnosis process along with thorough interpretation of all available imaging studies. In cases of lymphomas, H1 MRS alone may not be able to distinguish them from gliomas, since both show increased Cho, reduced NAA and increased Cho/Cr and Cho/NAA ratios. In such cases, H1 MRS combined with DWI may help in their differentiation, since lymphomas demonstrate characteristically decreased diffusivity. Furthermore, lymphomas have a more homogeneous contrast enhancement compared to gliomas, however this is certainly not a pathognomonic finding. The concordance of the above characteristics may help to establish a diagnosis in these cases. GBM very often has conventional imaging characteristics indistinguishable from abscesses and metastases. Magnetic Resonance Spectroscopy may play an important role in the diagnosis and differentiation of these lesions. Abscesses uniquely contain

76

E.Z. Kapsalaki et al.

Fig. 8.2 A spectrum of a low grade astrocytoma (WHO II). Note the slightly decreased concentration of NAA and the relatively increased concentration of Cho. The concentration of Cr remains relatively unchanged. Lipids are almost undetectable

several aminoacids detectable on H1 MRS. Moreover, abscesses have decreased diffusivity due to the decreased mobility of water protons, a finding that is not present in gliomas. Kapsalaki et al. (2008) reported the H1 MRS characteristics of a wide variety of bacterial abscesses and also the usefulness of this methodology in distinguishing abscesses from gliomas and metastatic lesions. They outline in their study the role of H1 MRS in identifying the causative agent of an abscess, and they point out its importance in following up the abscess’s therapeutic response. The use of H1 MRS in combination with DWI can significantly increase the diagnostic accuracy of conventional MR imaging and thus provide valuable preoperative information, regarding the nature of space occupying, ring-enhancing intracranial lesions.

Metastatic lesions, when solitary and ring enhancing, may also have indistinguishable imaging characteristics from necrotic GBMs. Both lesions show decreased, almost absent NAA, elevated Cho and decreased Cr. Lipids may also be present in both GBM and metastatic lesions, with metastatic lesions usually showing higher concentrations of lipids. On the other hand, the overall appearance of the spectrum between GBM and metastasis may occasionally be indistinguishable. In this case, H1 MRS can differentiate between the two tumor types by evaluating the spectra at the lesion’s periphery, as seen in Fig. 8.4. An infiltrating tumor, as a GBM, would yield elevated Cho and Lipids signals at the periphery of the lesion, whereas a metastasis would yield a relatively normal spectrum.

8

Proton Magnetic Resonance Spectroscopy in Intracranial Gliomas

77

a)

b)

Fig. 8.3 A spectrum of an anaplastic astrocytoma (WHO III) (a) and (b): Note the decreased concentration of NAA and the markedly increased concentration of Cho in both spectra. In (a), the concentration of Cr remains relatively stable and lipids are

almost undetectable. In (b), the concentration of Cr is slightly diminished and the lipids are increased, findings that possibly indicate a more aggressive type of glioma

a)

Fig. 8.4 (a) Spectrum obtained from the center of a tumor. Note the absence of NAA and Cr, and the markedly increased concentration of Cho. Also, there is a large amount of lipids in the tumor. These findings are characteristic of a necrotic GBM or

b)

a metastatic lesion. The presence of a normal spectrum identified in the periphery of the lesion (b) is more suggestive of a metastatic lesion

78

The major difficulty presents when we need to differentiate radiation necrosis from progressing or recurring glioma. Despite aggressive surgical resection along with chemotherapy and radiation, gliomas have the tendency to locally recur. An important question that is frequently raised is to differentiate in these cases between tumor recurrence versus radiation necrosis. In radiation necrosis spectrum, lipids are detected but Cho is usually not increased. However, increased levels of Cho may be seen in radiation necrosis due to the presence of inflammation. In such cases, differentiation with H1 MRS alone is almost impossible. Other imaging modalities need to be combined, as DWI and perfusion MRI. A low Cerebral Blood Volume (CBV) through the area of contrast enhancement usually suggests radiation necrosis. On the contrary, high CBV along with increased levels of Cho and markedly increased Cho/Cr, Cho/NAA ratios is more suggestive of tumor recurrence. Despite all the recent MR advances, it is not unusual the differentiation between tumor recurrence and radiation necrosis to be still impossible. In these cases, thorough comparison with preoperative studies and close clinical and imaging follow up is of paramount importance.

Future of MRS Aided Diagnosis It seems that we may have reached the limit of visual interpretation of spectra and as a result the field of neurospectroscopy should extend beyond qualitative analysis. Xu et al. (2008), propose at least two methods, which will take clinical MRS beyond the purely visual interpretation of a few metabolite peaks, to a richer, more automated neurochemical diagnostic concept. One is the wavelet analysis which extracts more information after data acquisition, than can currently be included in Fourier transform (FT). Another, is the widely disseminated LC Model (Provencher, 1993) which breaks the visible spectrum up into its presumed constituents, on the basis that any chemical already known to exist in the FT spectrum (so-called prior knowledge) can be outlined and quantified separately from the overlapping series of peaks we see on the screen of the MR scanner. This should only include patients in which the clinical/radiological diagnosis is uncertain.

E.Z. Kapsalaki et al.

Proton MRS is a promising diagnostic modality, which may be used to offer a better preoperative evaluation of intracranial gliomas. It may provide non invasive information regarding the histological grade of the examined tumor. It needs to be emphasized that H1 MRS is not aiming to replace surgical biopsy in establishing a histological diagnosis. It is a complimentary method to the existent imaging modalities, which may attribute to a better preoperative evaluation of gliomas, and may also help differentiate between post-surgical radiation necrosis and tumor recurrence. H1 MRS should be combined with conventional MR imaging and possibly all other available advanced MR imaging techniques, in order to suggest an accurate preoperative diagnosis, including tumor’s histopathological type and grade.

References Auer DP, Gossl C, Schirmer T, Czisch M (2001) Improved lipid analysis of 1 H-MR spectra in the presence of mobile lipids. Magn Reson Med 46:615–618 Barker PB, Gillard JH, van Zijl PCM, Soher BJ, Hanley DF, Agildere AM, Oppenheimer SM, Bryan RN (1994) Acute stroke: evaluation with serial proton MR spectroscopic imaging. Radiology 192:723–732 Birken DL, Oldendorf WH (1989) N-acetyl-L-aspartic acid: a literature review of a compound prominent in 1H-NMR spectroscopic studies of brain. Neurosci Biobehav Rev 13:23–31 Broer S, Broer A, Hansen J, Bubb WA, Balcar VJ, Nasrallah FA, Garner B, Rae C (2007) Alanine metabolism, transport, and cycling in the brain. J Neurochem 102:1758–1770 Brulatout S, Meric P, Loubinoux I, Borredon J, Correze JL, Roucher P, Gillet B, Berenger G, Beloeil JC, Tiffon B, Mispelter J, Seylaz J (1996) A one-dimensional (proton and phosphorus) and two-dimensional (proton) in vivo NMR spectroscopic study of reversible global cerebral ischemia. J Neurochem 66:2491–2499 Castillo M, Smith JK, Kwock L (2000) Correlation of myoinositol levels and grading of cerebral astrocytomas. Am J Neuroradiol 21:1645–1649 Cheng LL, Chang IW, Louis DN, Gonzalez RG (1998) Correlation of high resolution magic angle spinning proton magnetic resonance spectroscopy with histopathology of intact human brain tumor specimens. Cancer Res 58: 1825–1832 Croteau D, Scarpace L, Hearshen D, Gutiérrez J, Rock J, Rosenblum M, Fisher J, Mikkelsen T (2001) Correlation between magnetic resonance spectroscopy imaging and image-guided biopsies: semi-quantitative and qualitative histo-pathologic analysis of patients with untreated glioma. Neurosurgery 49:823–829 Fountas KN, Kapsalaki E, Vogel R, Fezoulidis I, Robinson JS, Gotsis ED (2004) Noninvasive histologic grading of solid

8

Proton Magnetic Resonance Spectroscopy in Intracranial Gliomas

astrocytomas using proton magnetic resonance spectroscopy. Stereotact Funct Neurosurg 82:90–97 Gideon P, Henriksen O, Sperling B, Christiansen P, Olsen TS, Jorgensen HS, Arlien-Soborg P (1992) Early time course of N-acetylaspartate, creatine and phosphocreatine, and compounds containing choline in the brain after acute stroke. A proton magnetic resonance spectroscopy study. Stroke 23:1566–1572 Gotsis ED, Fountas K, Kapsalaki E, Toulas P, Peristeris G, Papadakis N (1996) In vivo proton MR spectroscopy: the diagnostic possibilities of lipid resonances in brain tumors. Anticancer Res 16:1565–1567 Howe FA, Barton SJ, Cudlip SA, Stubbs M, Saunders DE, Murphy M, Wilkins P, Opstad KS, Doyle VL, McLean MA, Bell BA, Griffiths JR (2003) Metabolic profiles of human brain tumors using quantitative in vivo 1H magnetic resonance spectroscopy. Magn Reson Med 49: 223–232 Jansen JF, Backes WH, Nicolay K, Kooi ME (2006) 1H MR spectroscopy of the brain: absolute quantification of metabolites. Radiology 240:318–332 Kallenberg K, Bock HC, Helms G, Jung K, Wrede A, Buhk JH, Giese A, Frahm J, Strik H, Dechent P, Knauth M (2009) Untreated glioblastoma multiforme: increased myo-inositol and glutamine levels in the contralateral cerebral hemisphere at proton MR spectroscopy. Radiology 253:805–812 Kapsalaki E, Gotsis ED, Fountas KN (2008) The role of proton magnetic resonance spectroscopy in the diagnosis and categorization of cerebral abscesses. Neurosurg Focus 24:E7 Kim JH, Chang KH, Na DG, Song IC, Kim SJ, Kwon BJ, Han MH (2006) Comparison of 1.5T and 3T 1H MR spectroscopy for human brain tumors. Korean J Radiol 7:156–161 Majos C, Julia-Sape M, Alonso J, Serrallonga M, Aguilera C, Acebes JJ, Arus C, Gili J (2004) Brain tumor classification by proton MR spectroscopy: comparison of diagnostic accuracy at short and long TE. Am J Neuroradiol 25:1696–1704 Manton DJ, Lowry M, Blackband SJ, Horsman A (1995) Determination of proton metabolite concentrations and relaxation parameters in normal human brain and intracranial tumours. NMR Biomed 8:104–112 Mikkelsen T, Edvardsen K (1995) Invasiveness in nervous system tumors. In Black P, Loeffler JS (eds) Cancer of the nervous system. Blackwell, Cambridge, MA

79

Miller BL, Chang L, Booth R, Ernst T, Cornford M, Nikas D, McBride D, Jenden DJ (1996) In vivo 1H MRS choline: correlation with in vitro chemistry/histology. Life Sci 58: 1929–1935 Negendank WG, Sauter R, Brown TR, Evelhoch JL, Falini A, Gotsis ED, Heerschap A, Kamada K, Lee BC, Mengeot MM, Moser E, Padavic-Shaller KA, Sanders JA, Spraggins TA, Stillman AE, Terwey B, Vogl TJ, Wicklow K, Zimmerman RA (1996) Proton magnetic resonance spectroscopy in patients with glial tumors: a multicenter study. J Neurosurg 84:449–458 Podo F (1999) Tumour phospholipid metabolism. NMR Biomed 7:413–439 Provencher SW (1993) Estimation of metabolite concentrations from localized in vivo proton NMR spectra. Magn Reson Med 30:672–679 Simone IL, Tortorella C, Federico F, Liguori M, Lucivero V, Giannini P, Carrara D, Bellacosa A, Livrea P (2001) Axonal damage in multiple sclerosis plaques: a combined magnetic resonance imaging and 1H-magnetic resonance spectroscopy study. J Neurol Sci 182:143–150 Sinson G, Bagley LJ, Cecil KM, Torchia M, McGowan JC, Lenkinski RE, McIntosh TK, Grossman RI (2001) Magnetization transfer imaging and proton MR spectroscopy in the evaluation of axonal injury: correlation with clinical outcome after traumatic brain injury. Am J Neuroradiol 22:143–151 Tzika AA, Cheng LL, Goumnerova L, Madsen JR, Zurakowski D, Astrakas LG, Zarifi MK, Scott RM, Anthony DC, Gonzalez RG, Black PM (2002) Biochemical characterization of pediatric brain tumors by using in vivo and ex vivo magnetic resonance spectroscopy. J Neurosurg 96: 1023–1031 Wilson M, Davies NP, Grundy RG, Peet AC (2009) A quantitative comparison of metabolite signals as detected by in vivo MRS with ex vivo 1H HR-MAS for childhood brain tumours. NMR Biomed 22:213–219 Xu V, Chan H, Lin AP, Sailasuta N, Valencerina S, Tran T, Hovener J, Ross BD (2008) MR spectroscopy in diagnosis and neurological decision-making. Semin Neurol 28: 407–422 Yudkoff M, Nissim I, Hummeler K, Medow M, Pleasure D (1986) Utilization of [15 N]glutamate by cultured astrocytes. Biochem J 234:185–192