Crit Rev Neurosurg (1999) 9: 161 – 166 © Springer-Verlag 1999
Robert A. Zimmerman Zhiyue Wang
R. A. Zimmerman (½) · Z. Wang Department of Radiology, Children’s Hospital of Philadelphia, 34th Street and Civic Center Boulevard, Philadelphia, PA 19104, USA Tel.: +1-215-590-2569 Fax: +1-215-590-4127
161
Proton magnetic resonance spectroscopy
Abstract Proton magnetic resonance spectroscopy (MRS) permits in vivo determination of biochemical parameters within brain tissue, utilizing the same magnetic resonance (MR) scanner and head coil that are utilized for conventional MR imaging. This technology has been evolving and improving over the past decade, with most of the current published work based on the measurement of rather large single voxels, utilizing variable echo times, aimed at characterizing changes in brain tumors, demyelinating diseas-
es, abscesses, and metabolic diseases. Future work will utilize multiplevoxel techniques so that volumes of tissue can be examined with smaller voxels in reasonable acquisition times, providing a greater understanding of the metabolite composition of the brain, especially as more and more metabolites are identified within the spectra.
important brain metabolites could be identified with these techniques. These include choline (Cho), a cell membrane metabolite; phosphocreatine and creatine (Cr) , energy metabolites; N-acetylaspartate (Naa), a metabolite that is thought to be within neurons and perhaps oligodendroglia; and, at these longer TEs, lactate, a by-product of metabolism, primarily of an anaerobic metabolism, that can be seen in very small amounts normally, but increases dramatically in a number of pathological states such as infarction. Alanine is also seen at these longer TEs, produced by bacteria and found within
bacterial abscesses. Recently, shorter TEs, down to 5–10 ms, allow the detection of more spectral peaks, representing a larger number of proton-containing compounds within the brain tissue. Among others, these include inositol, alanine, glutamate, and glutamine. Inositol is of interest as an osmolite related to intracellular sodium content, the belief being that it is increased with increased osmolality in the brain. Glutamate has a role as an excitatory amine that is released with insults of the brain, such as trauma, and leads to further brain injury and brain swelling. Brain tumors, metabolic diseases,
Key words Magnetic resonance spectroscopy · Brain tumors · Demyelinating diseases of the central nervous system · Brain abscess
Introduction Proton magnetic resonance spectroscopy (MRS) did not evolve until the signal intensity of water could be suppressed and localization techniques developed that would allow determination of where the spectral information was coming from within the brain. In the late 1980s, these techniques became available, and initially single-voxel proton MRS has been performed. The techniques that have been required have become more user friendly and are gradually coming into wider clinical use. The echo times (TEs) utilized with proton MRS were initially in the range of 135–270 ms. Several
162
stroke, and head injuries have been among the four most important indications for proton spectroscopy. The usefulness of spectroscopy in cases of brain tumors has been an indication of the degree of malignancy of the tumor and the differentiation of tumor from non-tumorous diseases. Preliminary evidence suggests that proton spectroscopy will play a role in determining the response of tumor
to treatment and in detecting recurrent brain tumor. The role of spectroscopy in metabolic diseases is to identify metabolic lactic acidosis by an increased lactate level, and to identify a variety of compounds that are not normally present in detectable amounts in the brain tissue, such as occur in a variety of inborn errors of metabolism. Such compounds could represent potential
[1] Proton MR spectroscopic characteristics of pediatric pilocytic astrocytomas AJNR Am J Neuroradiol (1998) 19:535–540
Information. The authors examined eight children (aged 1–12 years; mean 6.8) with pilocytic astrocytomas with proton MRS with 4-cm3 single voxels from the solid portion of the tumor. Point-resolved spectroscopy (PRESS) with a long TE of 270 ms and stimulated-echo acquisition mode (STEAM) with a short TE of 20 ms were used to measure the peaks of Cho, Cr, Naa, and lactate. Cho/Naa peak area ratio was significantly elevated at 3.40. Normal reported ratios are 0.75–0.84. Elevated lactate was also observed. The authors were surprised by the elevated lactate since the tumors were not grossly necrotic. Analysis. The authors have recapitulated what has already been reported by a number of other authors, that is, in low-grade astrocytomas (in this case, pilocytic astrocytomas) the Cho level is elevated and the Naa level is decreased. The difference here is one of magnitude. Prior authors found the ratio to be on the order of 1.80 versus these authors’ ratio of 3.40. In all likelihood, the difference is simply the voxel size. Prior authors used 2.5×2.5×2.5 cm3 in the early 1990s, whereas these authors use 1.5×1.5×1.5 cm3 in the late 1990s. The larger voxel size may have enabled some contribution to the spectra of normal surrounding tissue, where the newer smaller voxel would allow less. The differences are more the result of further capability of newer and more accurate methods with MRS. The authors also express great concern over the elevated lactate in pilocytic astrocytomas. This is because of the literature on malignant adult glioblastoma in which lactate elevation
markers for these diseases, and knowledge of what they are could lead to a better understanding of their deleterious effects on brain substance. With those thoughts in mind, we look at recent publications in the neuroradiology literature regarding the potential applications of MRS for a variety of disease states.
is common and is attributed to anaerobic glycolysis with tumor necrosis. The authors fail to take into account the histology of pilocytic astrocytoma, which shows a loose stroma with clefting. This, as well as the frequent adjacent macrocysts, are sufficient holes for metabolic waste in the form of lactate to accumulate. Our own experience is that lactate elevation means little in the pediatric brain tumor population relative to benign versus malignant.
[2] In vivo single-voxel proton MR spectroscopy in intracranial cystic masses AJNR Am J Neuroradiol (1998) 19:401–405
Information. The authors evaluated the fluid-filled contents of 39 cystic intracranial masses with proton MRS using 2×2×2-cm voxels with PRESS with a TE of 270 ms. Two-thirds of patients were also studied with a TE of 135 ms. The pathology studied included 14 gliomas, the most common finding being only lactate resonance present in the cystic fluid. Similar findings were found in 2 out of 3 metastases. Seven of eight patients with brain abscesses showed different results. Acetate (1.9 ppm); succinate (2.4 ppm); amino acids including valine, alanine and leucine (0.9, 1.5, 3.6 ppm, respectively) and lipid were found along with lactate. Four patients with cysticercosis showed lactate, alanine, and/or leucine. Analysis. Previous authors have observed that lactate, acetate, and succinate are present in the contents of abscesses, originating from enhanced glycolysis and fermentation of the organism. Valine and leucine are known by-products of proteolysis, when enzymes in
163
neutrophils are released in the pus. The authors have confirmed prior reports by demonstrating these same findings in brain abscess cyst fluid. That these products, other than lactate, do not occur in the cystic contents of necrotic gliomas is a step forward in the use of proton MRS in differentiating tumor from abscess. Thus far, it is not clear that cysticercosis can be differentiated from abscess by the analysis of cyst contents.
[3] Treatment of brain tumors in children is associated with abnormal MR spectroscopic ratios in brain tissue remote from the tumor site AJNR Am J Neuroradiol (1998) 19:963–970
Information. The authors used single-voxel proton MRS with voxels of between 5 and 8 cm3 using PRESS with a TE of 270 ms on 70 pediatric patients and 11 normal subjects to study the non-neoplastic regions of the frontal lobes away from the site of the primary tumor. Ratios of Naa to Cho and Cr were analyzed. When chemotherapy was utilized as a treatment component, the Naa was significantly decreased. When both chemotherapy and radiation therapy were carried out, if chemotherapy preceded radiation, lower Naa levels were found. Whole head radiation showed lower Naa levels than in those patients with focal radiation. Analysis. Treatment-related sequelae of pediatric brain tumors are not a new phenomenon. However, demonstrating an effect of treatment on neurons, as demonstrated by a decrease in Naa, a neuronal marker, is new. It should be noted that in this study, patient studies were carried out after administration of a contrast agent, while control data were obtained without contrast material. We have investigated the effects of contrast administration on metabolite ratios in ten patients and have found that the administration of paramagnetic contrast agent increased the mean value of Naa/Cr by 9%. This increase, however, did not reach statistical significance in our study. If the effects of contrast administration were indeed to increase the Naa/Cr ratio, then the effects of tumor therapy of decreasing this metabolite ratio for brain tissue remote from tumor site would even be somewhat more underestimated here. This paper opens avenues of investigation for the future, in meaningful, well-structured studies.
[4] Single-voxel proton MR spectroscopy of nonneoplastic brain lesions suggestive of a neoplasm AJNR Am J Neuroradiol (1998) 19:1695–1703
Information. Five patients out of 241 who had proton spectroscopy for suspected brain neoplasms and spectra consistent with neoplasm were identified by postoperative histology to have non-neoplastic lesions. One additional patient with clinically suspected non-neoplastic mass did not have histology. These patients had spectra showing reduced or absent Naa and elevated Cho. Lipid and/or lactate was elevated in 4 out of 5 patients with histology. These were histologically characterized by significant white blood cell (WBC) infiltrates with interstitial and perivascular accumulations of lymphocytes, macrophages, histiocytes, and in one case, plasma cells. Subsequent diagnoses are two cases of fulminant demyelination, one human herpes virus, type six encephalitis, one arteriovenous malformation (AVM) with organizing hematoma, and one inflammatory pseudotumor. Magnetic resonance (MR) spectra were acquired on a 0.5-T MR system utilizing PRESS with a TE of 40 ms and a voxel size of 1–3 cm3. Analysis. That all non-invasive imaging techniques, including MRS, are not perfect in their ability to differentiate neoplastic tissue from non-neoplastic, should not be an issue at present. The authors point out the pitfall that with current basic techniques such as were used here there is a small likelihood (2.0%) that what appears neoplastic by MRS will not be by histology. The decreased Naa in neoplasms represents neuronal replacement by tumor tissue, whereas in the nonneoplastic cases it is the neuron-destroying nature of each lesion. The elevated Cho levels in tumors are indicative of increased turnover of cell membranes (synthesis and/or degradation), while in non-neoplastic cases they are due to the proliferation of cellular elements of the immune system and astroglia. The elevated lipid/lactate peak in neoplasms reflects cellular necrosis and anaerobic metabolism, respectively, whereas in non-neoplastic cases the mechanisms are the same.
164
[5] Single-voxel proton MR spectroscopy and positron emission tomography for lateralization of refractory temporal lobe epilepsy
ized to the hippocampal structures should hold even more promise in demonstrating abnormalities in mesial sclerosis.
AJNR Am J Neuroradiol (1998) 19:1–8
Information. Twenty-five patients with refractory temporal lobe epilepsy with onset clinically lateralized to one temporal lobe were studied by both positron emission tomography (PET) utilizing fluorodeoxyglucose [18F] and proton MRS. MRS was single voxel 8 cm3 with a TE of 135 ms. Twenty-four (96%) of the 25 patients had either MRS or PET showing ipsilateral lateralization to the side of clinical seizure focus. Concordant MRS and PET lateralization was possible only in 14 (56%). Forty-two temporal lobes were abnormal on MRS, while 25 were abnormal on PET. Bilateral low Naa was present in 10 patients. All hypometabolic temporal lobes on PET also had low Naa on MRS. Five patients with PET-negative studies were correctly localized by MRS. Thus, MRS appeared to be more sensitive than PET. Seventeen out of 25 have had surgery. There is no pathology in 8. All operated patients had sclerosis. Analysis. This study is of interest because it achieves decent results despite the large voxel size used for MRS, which samples much more than hippocampus, e.g., temporal lobe white matter. This suggests that the Naa axons arising from the pyramidal cells of the temporal neocortex and mesial structures have suffered and that the MRS is detecting this more widespread damage. The PET is detecting metabolic abnormalities that are indeed cortical in origin. Smaller MRS voxels local-
[6] Benign versus secondary-progressive multiple sclerosis: the potential role of proton MR spectroscopy in defining the nature of disability AJNR Am J Neuroradiol (1998) 19:223–229
Information. The authors studied 30 patients with multiple sclerosis (MS) and 13 controls with proton MRS with STEAM, single voxel 2×2×2 cm with a TE of 20 ms. All patients with MS showed lower Naa/Cr and Naa/Cho in white matter than the controls. Patients with acute plaques differed from those with chronic lesions in the following: in patients with benign MS, Naa reduction was more evident in acute lesions with recovery in chronic lesions; in patients with secondary progressive MS, Naa reduction was stable in both acute and chronic lesions. Analysis. The authors present interesting data, despite relatively large voxels relative to the size of the MS plaques. This indicates that they are really measuring a more global process within the brain, in which the reduction of Naa in acute lesions is related to several possible etiologies: edema with inflammation, demyelination, and/or neuronal dysfunction, whereas recovery of Naa is related to absence of inflammation and/or remyelination. In chronic secondary progressive MS, the decreased Naa is likely to be due to irreversible loss of axonal integrity.
Synthesis A limited number of metabolites are now measurable with proton MRS. To some extent, what metabolites are seen is dependent on the TE chosen. Lactate is best seen with TEs of 135–270 ms. Naa, Cho, Cr are well seen with both short and long TEs. Voxel size is important in getting localization of the spectra from the particular area of interest. This is especially true in the case of brain tumors, where contamination from
surrounding normal tissue is likely to explain discrepancies between various authors’ results [1]. Thus, smaller is usually better. There are trade-offs, of course; smaller voxels require longer acquisition times to get the same signal-to-noise ratio that is more easily achieved with larger voxels. Of interest is evidence that even with larger voxel sizes where more than the area of interest is included, e.g., mesial temporal
sclerosis [5] and MS [6], by showing significant abnormality, suggests that the disease process extends well beyond the radiographically defined anatomical abnormality. Further work with two-dimensional or three-dimensional chemical shift imaging and small voxels offers the potential of defining the distribution of these abnormalities in both radiographically normal and abnormal tissue [3, 5, 6].
165
That proton MRS is not foolproof is to be expected [4]. Many disease processes are likely to have metabolic derangements that have similar components, at least in part. Lactate can be found in necrotic or cystic tumors [2], ones that are not radiographically necrotic [1], and in acute inflammatory disease [4], as well as in infarction and metabolic diseases that produce anaerobic glycolysis. That metabolic waste, which is what lactate is, accumulates in holes in the brain, should come as no surprise [2]. Not surprisingly, examining the hole or necrotic cavity of a tumor turned out to be rather non-specific
regarding the diagnosis of tumor [2], whereas examining the solid tumor was a great deal more specific [1], if not totally accurate [4]. Examining the cystic contents of abscesses does pay off because they are different from those of brain tumors [2]. This is because of bacterial fermentation and enzymatic degradation of neutrophils [2]. Cho, the elevation of which, and Naa, the decrease of which, are hallmarks of brain tumors [1]; that they are only so reliable (98%) should not surprise us [4]. Inflammatory lesions can and do present acutely with choline elevation due to proliferation of cellular
elements of the immune system and astroglia, and Naa can be decreased with inflammation when neurons are destroyed. Finally, post-therapeutic changes are detectable in tissue away from the tumor [3]. These findings are in the form of decreased Naa. Reversibility or not of these Naa changes has not been studied yet, but in another disease, benign MS, reversibility has been shown in chronic lesions, indicating that, to some extent, metabolic processes have a potential, given the circumstances, of being reversible. Proton MRS is still young. The future is bright.
3. Waldrop SM, Davis PC, Padgett CA, Shapiro MB, Morris R (1998) Treatment of brain tumors in children is associated with abnormal MR spectroscopic ratios in brain tissue remote from the tumor site. AJNR Am J Neuroradiol 19: 963–970 4. Krouwer HGJ, Kim TA, Rand SD, Prost RW, Haughton VM, Ho K-C, Jaradeh SS, Meyer GA, Blindauer KA, Dusick JF, Morris GL, Walsh PR (1998) Singlevoxel proton MR spectroscopy of nonneoplastic brain lesions suggestive of a neoplasm. AJNR Am J Neuroradiol 19: 1695–1703
5. Achten E, Santens P, Boon P, DeCoo D, Van DeKerckhove T, DeReuck J, Caemaert J, Kunnen M (1998) Single-voxel proton MR spectroscopy and positron emission tomography for lateralization of refractory temporal lobe epilepsy. AJNR Am J Neuroradiol 19:1–8 6. Falini A, Calabrese G, Filippi M, Origgi D, Lipari S, Colombo B, Comi G, Scott G (1998) Benign versus secondary-progressive multiple sclerosis: the potential role of proton MR spectroscopy in defining the nature of disability. AJNR Am J Neuroradiol 19:223–229
Demaerel P, Hecke PV, Oostende SV, et al (1994) Bacterial metabolism shown by magnetic resonance spectroscopy. Lancet 344:1234 Harada M, Tanouchi M, Miyoshi H, Michitani H, Kannuki S (1994) Brain abscess observed by localized proton magnetic resonance spectroscopy. Magn Reson Imaging 12:1269–1274 Kinoshita K, Tada E, Matsumoto K, Asari S, Ohmoto T, Isoh T (1997) MR spectroscopy of delayed cerebral radiation in monkeys and humans after brachytherapy. AJNR Am J Neuroradiol 18: 1753–1761
Negendank W, Sauter R (1996) Intratumoral lipids in 1H MRS in vivo in brain tumors: experience of the Siemens cooperative clinical trial. Anticancer Res 16: 1533–1538 Peeling J, Sutherland G (1993) High-resolution 1H NMR spectroscopy studies of extracts of human cerebral neoplasms. Magn Reson Med 24:123–126 Peeling J, Sutherland G (1993) 1H magnetic resonance spectroscopy of extracts of human epileptic neocortex and hippocampus. Neurology 43:589–594
Papers reviewed 1. Hwang J-H, Egnaczyk GF, Ballard E, Dunn RS, Holland SK, Ball WS (1998) Proton MR spectroscopic characteristics of pediatric pilocytic astrocytomas. AJNR Am J Neuroradiol 19:535–540 2. Chang K-H, Song IC, Kim SH, Han MH, Kim HD, Seong SO, Jung HW, Han MC (1998) In vivo single-voxel proton MR spectroscopy in intracranial cystic masses. AJNR Am J Neuroradiol 19:401–405
Further reading Connelly A, Jackson GD, Duncan JS, King MD, Gadian DG (1994) Magnetic resonance spectroscopy in temporal lobe epilepsy. Neurology 44:1411–1417 Davie CA, Hawkins CP, Barker CJ, et al (1994) Serial proton magnetic resonance spectroscopy in acute multiple sclerosis lesions. Brain 117:49–58 Davis PC, Hoffman JC, Pearl GS, Braun IF (1986) CT evaluation of effects of cranial radiation therapy in children. AJNR Am J Neuroradiol 7:639–644
166
Rand SD, Prost R, Haughton V, et al (1997) Accuracy of single-voxel proton MR spectroscopy distinguishing neoplastic from non-neoplastic brain lesions. AJNR Am J Neuroradiol 18:1695–1704 Remy C, Grand S, Lai ES, et al (1995) 1H MRS of human brain abscesses in vivo and in vitro. Magn Res Med 34:508–514
Roser W, Hagberg G, Mader I, et al (1995) Proton MRS of gadolinium-enhancing MS plaques and metabolic changes in normal-appearing white matter. Magn Reson Med 33:811–817 Sutton LN, Wang Z, Gusnard D, et al (1992) Proton magnetic resonance spectroscopy of pediatric brain tumors. Neurosurgery 31:195–202 Sutton LN, Wehrli SL, Gennarelli L, et al (1994) High-resolution 1H-magnetic resonance spectroscopy of pediatric posterior fossa tumors in vitro. J Neurosurg 81:443–448
Valk PE, Dillon WP (1991) Radiation injury of the brain. AJNR Am J Neuroradiol 12: 45–62 Wang Z, Zimmerman RA, Sauter R (1996) Proton MR spectroscopy of the brain: clinically useful information obtained in assessing CNS diseases in children. AJR Am J Roentgenol 167:191–199