Magnetic resonance spectroscopy (MRS) could be used for dozens of metabolites [1]. In this paper we will focus especially on proton (hydrogen) MRS. Generally published literatures consist of reviews and articles about metabolites are long, complicated and emphasise on physical or on technical data. But with this article, we will try to explain basically how MRS could be used in daily practice of radiologists. First of all we tried to search all current literature only referencing the main ones to make this review and we want all of our colleagues to give feedback about this review. With collection of all these trick points and line up to appropriate order, it is possible to make this review a common guide for radiologists and a useful tool while reporting MRS.
After the 3 tesla MRS came into use, we are able to measure main metabolites more accurately, also with current softwares we are able to measure more metabolites. So with all those new developments, differential diagnosis of metabolical diseases, follow up of demyelinating diseases, differentiation of malignant and benign tumours and staging of tumours are much easier. The most important part of MRS is, it has ability to measure chemical component of the area that we study, non invasively (
Generally metabolites are distributed between water and lipid peaks. There are different techniques to obtain metabolite information but this is not a subject for this review. But at least we can say that there are mainly two
methods in use and with PRESS (point resolved spectroscopy) method signal-noise ratio is higher [
During examination, voxel of interest should not include especially scalp, sinuses, air or haemorrhagic foci. Normal CSF may dilute concentration but does not cause significant degradation of spectrum content. Also jewellery and ventriculoperitoneal shunts could disturb optimal settings of shimming and could change spectrum. Gadolinium could decrease the cho peak by % 12 - 15, because of that MRS should be done before administration of contrast material. Because there is no certain normal international reference values for MRS metabolites, measurements could be compared with examined cases’ own normal and abnormal regions. Also, some determined ratios of well known ratios can help in differential diagnosis as defined in
On starting to examination, single or multi voxel protocol must be selected considering the purpose of the study. In searching of brain tumour which has got a heterogeneous or infiltrative pattern we should use multi-voxel protocol [
In grey matter NAA, Cr and glx are higher quantities, but in white matter Cho is relatively high [
NAA is the major neuronal metabolite and Cho, Cr, mIns, glutamate are the other ones. Whereas, the brain normally doesn’t include high quantity of some metabolites (like lactate, hydroxy butyrate, acetone, glycine, succinate, pyruvate, alanine, lipid, phenylalanine, mannitol, ethanol, etc.) to definable on normal resonance line.
Diagnosis could be made by determining such abnormal metabolites or with determining changes of normal metabolites. Also with using a special technique, NAA’s amid component’s changes could be determined by administering histidine and homocarnosine. And so, brain PH levels could be measured by determination of NAA level changes. Also with whole brain spectroscopy, quantitative values of all metabolites could be measured.
Different components of metabolites could be useful for defining the different peak values of the same product. For example: NAA 2.02 and 2.6 ppm, Cr at 3.02 and 3.9 ppm, Glx between 2.05 - 2.4 and 3.65 - 3.8 ppm, glucose at 3.43 and 3.48 ppm, myo-inositol between 3.56 and 4.06. But, generally, only the established dominant ones to be considered for diagnosis.
Main metabolites: Cho (choline, 3.2 ppm), Cr (creatine, 3.0 ppm), NAA (n-acetile aspartic acid, 2.02 ppm), Mi (myo-inositole, 3.56 ppm) Lip (Lipid, 0.8 - 1.5 ppm), Lac ( lactate, 1.3 ppm) . The other metabolites will be explained later by shortly. Water peak visualised at 4.7 ppm and it is used as reference value. All these metabolites that rises before water peak could be identified by spectral editing.
Cho: Marker of membrane proliferation, above normal values usually gives warning about malignity [
Cr: Peaks at 3.2 ppm. Useful as an internal reference [
NAA: Neuronal marker. The only situation causes increased NAA is Canavan disease. Generally all other diseases cause reduced NAA levels [
mI: As a glial proliferation marker, it increases in diabetes, Alzheimer disease, PMLE, Down syndrome; and decreases in stroke, tumours and encephalopathies. While mI is increases in Alzheimer and Down syndrome, it decreases in other cases of dementia [
mI and glycine’s peak values are so close to each other, they both resonate at 3.5 - 3.6 ppms (gly: 3.55-singlet, mIns: 3.52 - 3.61-multiplet). Myo-inositol makes multiplet at short TE’s. It’s strength decreases at long TE values. Contrary to this, glycine resonates as a singlet peak and its visualisation enhances at long TE values. Gly increases in stroke, encephalitis, Huntington, Parkinson, Alzheimer and Van der Knaap diseases [
In infantile periods, mI is the dominant peak at first months and Cho is also high at these times. But then mI’s and Cho’s peak values decreases as NAA, Cr increases. After the end of age 2, MRS patterns of metabolites becomes similar with adult level [
Normal | Neonatal | Infantile | Child | Adult |
---|---|---|---|---|
NAA/Cho | 0.67 ± 0.17 | 1.56 ± 0.67 | 3.43 ± 1.04 | 2.37 ± 0.40 |
Cho/Cr | 1.27 ± 0.40 | 0.84 ± 0.21 | 0.49 ± 0.16 | 1.05 ± 0.23 |
Normal | NAA/Cr = 1.5 ± 0.2 | Cho/Cr = 1.05 ± 0.23 | NAA/Cho = 2.37 ± 0.40 |
---|---|---|---|
Gliom low grade | NAA/Cr = 1.12 ± 0.48 | Cho/Cr = 1.71 ± 0.63 | NAA/Cho = 0.65 ± 0.37 |
Gliom high grade | NAA/Cr = 0.66 ± 0.37 | Cho/Cr = 1.95 ± 0.72 | NAA/Cho = 0.33 ± 0.17 |
Glutamate/glutamine: Some times these couple of metabolites are called as Glx. Glx peaks mainly between 2.1 and 2.45. In MS plaques, hepatic encephalopathy, and Reyes syndrome, increased ammonia level causes increase in Glx peak values [
GABA: It is visualised at 1.9, 2.3, 3.0 ppm’s, but 2.3 ppm should be first choice because at 1.9 and 3.0 ppm values it overlaps by other dominant metabolites. Simply, for GABA 2.31 ppm and for Glutamine 2.62 pm values could be used. As a new method, MegaPRESS sequence could be used for quantitative differentiation of GABA [
Lactate (resonates at 1.3 ppm) is more prominent at longs TE values. In 1.5 tesla 150 ms, in 3 tesla 140 ms is the most suitable values. Inversion of lactate at short TE values could be incomplete at 3 Tesla systems. But generally, in shorts TE’s lactate is inverted but lipid is more prominent.
2-Hydroxyglutarate: gioma cases that have isocitrate dehydrogenase mutations. So, 2-hydroxyglutarate increases and visualised especially with TE: 90 - 100 ms at 2.25 ppm. Glioma cases that have high rate of this metabolite have good response to angiogenic therapy [
Lipid-macromolecular peaks resonates at 0.8, 1.2 1.5, 6.0 ppms. 0.8 - 1.2 ppm values usually called as lipid-macromolecular non-specific peak. If there is too much lipid we should use long TE for decrease lipids that have shorter TE values to get clear resonance graphics (It increases in tumours, metastasis and post-radiotherapy changes) [
Pyruvate: it peaks at 2.37 ppm. Normally couldn't detected but increase and could be defined in pyruvate dehydrogenase enzyme deficiency [
Syllo-inositole: peaks at 3.35 ppm, increases in alcoholism and glioneuronal tumours. It can be rise in early infancy especially at examinations obtained in 3 Tesla systems. It can be superposed by taurine but taurine has simple peak while sylloinositole has a more complex one [
Guanidinoacetate: Increases at 3.78 ppm. It is lower in anaplastic astrocytomas then other gliomas [
Citrate: Increases at 2.6 ppm in some brain tumours like PNET and high grade glial tumours [
Melanin: Peaks at 6.72 ppm. It can be detected in lesions that contain melanin.
Glucose: Peaks at 3.43 and 3.80 ppm. Increases in DM and hyperglycemia crisis.
Glutathione: Detected at 2.95 ppm. Increases in GBM, meningioma, and ischemic oxidative stress.
Additional to searching all these metabolites on resonance line, also there are some useful metabolite ratios to increase the strength of differential diagnosis (
It’s very hard to keep all those values defined on
NORMAL: NAA/Cr > 2.0, NAA/Cho > 1.6, Cho/Cr < 1.2
ABNORMAL: NAA/Cr < 1.6, NAA/Cho < 1.2, Cho/Cr > 1.5
Generally, instead of using MRS alone, we should use other conventional and multi parametric methods together for diagnosis. For example in high grade tumours; Cho and rCBV is high, in low grade tumours myo-in- ositol and glycine is high but rCBV is low. In PNET, Cho is high but NAA is low and taurine can also be high.
MRS establishes high levels of lipid and lactate due to rich cholesterol crystals and lipid deposits whereas normal neuronal metabolites are lower or absent. As cranial fossa base and bones are superposed and volume of hypophysis gland is lower than 2 mm, MRS is not very effective. If craniopharyngioma is large enough, especially at 1.5 ppm, it could be seen that central lipid-cholesterol peaks are high but Cho-NAA peaks are not. However, high Cho/Cr levels and presence of NAA are detected in gliomas that mimic craniopharyngioma which are located at hypothalamic or chiasmatic region [
Cytosolic amino-acides like Alanine (1.48 ppm), Acetate (1.9 ppm), Valine (0.9 ppm), Leucine (3.6 ppm) could increase. Furthermore, increase in Glx peak between 2.1 - 2.4 ppm and Glutathione peak at 2.95 have been notified in some literature. With respect to these characteristics, it is possible to distinguish schwannomas located in the same region or atypically located meningiomas (intraosseous or intraventricular) [
At 1.2 ppm, an increased triplet peak may be defined.
In early phase of acute stroke NAA decreases and this decrease continues during the first week. Creatine acts similar to NAA. But choline levels are not reliable and they are variable. On the other hand, lactate level starts to increase a few minutes after stroke. It decreases again by reperfusion [
Lac and Glu increase and NAA ratio decreases. A severe variability in Cho levels is not expected.
If it is tuberculosis abscess, lipid lactate levels increase, NAA and Cho decrease and a small increase is observed in aminoacid levels [
In pyogenic abscess, increase of aminoacid levels are more apparent [
In fungal abscess, lipid lactate levels increase more then aminoacid levels. Unlike Tbc abscess, especially in angioinvasive type, the intricated walls called crenelated outlines [
NAA decreases, Cho, lactate and myoinositol increase [
Due to increase in ammonia level, Glu peak rises. Myoinositol and Cho decrease and lactate increases [
Since the quality of spectrum is more clear and relatively more quantitative, MRS is implemented as single voxel. Multivoxel-CSI needs longer time, and since many of epilepsy patients are pediatric, they move during the process and distortion occurs in images. During acute phase, sodium/myoinositol pump increases myo-inositol levels. But the artefacts due to movements of epileptic ictus make it difficult to detect. In the interictal phase, NAA decreases due to neuronal decrement and myo-inositol shows moderate increases due to reactive astrogliosis in hypocampal sclerosis [
A solvent which is used in several operations, several treatment protocols and in some drugs that are used in preoperative or intraoperative anesthesia, peaks at 1.13 ppm at MRS. This solvent also exists in several drugs and benefical in security and follow-up of the treatment.
At 0 - 2 ppm interval, there is lip-macromolecule peak increase that supress other peaks occur [
During Lithium treatment, Cho levels diminishes [
NAA/Cho ratio decreases, myo-inositol, taurin and Glx peaks arise [
Tumour is identifiable as; myo-inositol is very high [
In MRS, lymphoma shows abnormally high lipid-lactate ratio, ADC levels are low relatively to other lesions. rCBV ratio is generally lower than expected (<2.3). In metastasis and GBM these ratio is very high. In metastasis peritumoural perfusion has not increased versus to GBM. Discrimination of diagnosis is safe based on these indirect indicators. In cerebral lymphoma, very high increase of Cho is useful to distinguish from toxoplasma, tuberculosis and cryptococcus infection [
In MSUD (maple syrupe urine disease): at 0.9 - 1.0 ppm long branched-chained amino acids, in diabetic ketoacidosis at 2.2 ppm acetone and at 3.4 ppm glucose increase.
In galactosemia: galactitol (which mimics lactose) peak between 3.67 - 3.74 ppms.
In phenylketonuria: phenylalanine peak arises at 7.6 ppm.
In lipid storage diseases: high levels of lipid resonance occur generally at 0.7 - 1.6 ppms.
Increased lactate is obvious in mitochondria storage disease as Leigh, MERF (Myoclonus epilepsy with ragged-red fibers) and MELAS (Mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes).
Scyllo-inositol increases at 3.35 ppm in peroxisomal diseases.
For diagnosis and follow-up of glial tumours, especially perfusion MRI, all the imaging parameters should be used in collaboration. For example, low rCBV and high ml/Cr indicates low grade tumour whereas Cr is low and it is at least grade III GBM. These alterations are proportional to Ki-67 which is a indicator of proliferation-mi- tosis and high Cho and low Cr are indicators of poor prognosis [
Tumours after radiotherapy, if rCBV > 1.3 we have to think relapse instead of necrosis. If rCBV > 1.75, Cho/Cr > 1.83 and Cho/NAA > 1.92 it is acceptable as high grade. Peritumoural infiltration is possible if tissue which is surrounding lesions has Cho/Cr ratio more than 1.3 times.
Cho may have misleading elevations in pilocytic astrositoma. This is due to increase of glycerophosphocholine instead of pathologic phosphocoline. Unfortunately discrimination is not possible with MRS. Clinical situations, age and conventional sequences may help. Beside these, in pilocytic astrositoma unassigned peak elevation occurs at 2.8 ppm due to unsaturated fatty acids increase.
For differential diagnosis in such conditions, data from MRS should be used in conjunction with conventional and advanced neuroradiological findings. For example, in T2 perfusion, signal fall to 75% level is important, as line approaches to baseline at normal brain and gliomas. In lymphoma and pilocytic astrocytoma, it may arise to higher than baseline. But in meningiomas it is remote from baseline. Unlike choroid plexus papillomas, intraventricular meningiomas has rich perfusion. In choroid plexus papillomas myo-inositol is very high. As atypia and malignity progress, myo-inositol falls and Cho increases [
As another example, in addition to MRS, ADC (apparent diffusion coefficient) values gain advantage when making diagnosis. At posterior fossa tumours, which display similar conventional findings, ADC measurement can help to differentiation. Because, ADC value is between 1.0 - 1.3 (×10−3 mm2/s) for ependymoma, lower than 0.7 for medulloblastoma, and generally higher than 1.4 for pilocytic astrocytoma.
In conclusion, with this condensed review, despite a high technical-physical burden about MRS which is highly complicated and intense, we provide much practical information. We also believe to ensure knowing basic informations from performing protocol to permutation of metabolites as well as essential data about rare metabolites and differences on spectrum made by several diseases. With development of software and hardware, utilisation of MRS in diagnostic radiology will expand beyond today.