Metabolic profiles of human brain tumors using quantitative in vivo 1H magnetic resonance spectroscopy

Proton spectroscopy can noninvasively provide useful information on brain tumor type and grade. Short- (30 ms) and long- (136 ms) echo time (TE) (1)H spectra were acquired from normal white matter (NWM), meningiomas, grade II astrocytomas, anaplastic astrocytomas, glioblastomas, and metastases. Very low myo-Inositol ([mI]) and creatine ([Cr]) were characteristic of meningiomas, and high [mI] characteristic of grade II astrocytomas. Tumor choline ([Cho]) was greater than NWM and increased with grade for grade II and anaplastic astrocytomas, but was highly variable for glioblastomas. Higher [Cho] and [Cr] correlated with low lipid and lactate (P < 0.05), indicating a dilution of metabolite concentrations due to necrosis in high-grade tumors. Metabolite peak area ratios showed no correlation with lipids and mI/Cho (at TE = 30 ms), and Cr/Cho (at TE = 136 ms) best correlated with tumor grade. The quantified lipid, macromolecule, and lactate levels increased with grade of tumor, consistent with progression from hypoxia to necrosis. Quantification of lipids and macromolecules at short TE provided a good marker for tumor grade, and a scatter plot of the sum of alanine, lactate, and delta 1.3 lipid signals vs. mI/Cho provided a simple way to separate most tumors by type and grade.

White matter damage on diffusion tensor imaging correlates with age-related cognitive decline

BACKGROUND

Damage to white matter tracts, resulting in "cerebral disconnection," may underlie age-related cognitive decline.

METHODS

Using diffusion tensor MRI (DTI) to investigate white matter damage, and magnetic resonance spectroscopy (MRS) to look at its underlying pathologic basis, the authors investigated the relationship between white matter structure and cognition in 106 healthy middle-aged and elderly adults. Fractional anisotropy (FA) and mean diffusivity (MD) values, whole brain white matter histograms, and regions of interest placed in the white matter of the centrum semiovale were analyzed. Correlations with executive function, working memory, and information-processing speed were performed.

RESULTS

There was a progressive reduction in FA and increase in diffusivity with age in both region of interest (r = 0.551, p < 0.001), and whole brain histograms (r = 0.625, p < 0.001). DTI values correlated with performance in all three cognitive domains. After controlling for age, DTI parameters correlated with working memory but not with the other two cognitive domains. MRS studies found a correlation of N-acetyl aspartate, a neuronal marker, with DTI parameters (r = 0.253, p < 0.05).

CONCLUSION

The results are consistent with white matter damage due to axonal loss, causing age- related cognitive decline. Working memory may be particularly dependent on complex networks dependent on white matter connections.

Dynamic asymmetry of phosphocreatine concentration and O(2) uptake between the on- and off-transients of moderate- and high-intensity exercise in humans

The on- and off-transient (i.e. phase II) responses of pulmonary oxygen uptake (V(O(2))) to moderate-intensity exercise (i.e. below the lactate threshold, theta;(L)) in humans has been shown to conform to both mono-exponentiality and 'on-off' symmetry, consistent with a system manifesting linear control dynamics. However above theta;(L) the V(O(2)) kinetics have been shown to be more complex: during high-intensity exercise neither mono-exponentiality nor 'on-off' symmetry have been shown to appropriately characterise the V(O(2)) response. Muscle [phosphocreatine] ([PCr]) responses to exercise, however, have been proposed to be dynamically linear with respect to work rate, and to demonstrate 'on-off' symmetry at all work intenisties. We were therefore interested in examining the kinetic characteristics of the V(O(2)) and [PCr] responses to moderate- and high-intensity knee-extensor exercise in order to improve our understanding of the factors involved in the putative phosphate-linked control of muscle oxygen consumption. We estimated the dynamics of intramuscular [PCr] simultaneously with those of V(O(2)) in nine healthy males who performed repeated bouts of both moderate- and high-intensity square-wave, knee-extension exercise for 6 min, inside a whole-body magnetic resonance spectroscopy (MRS) system. A transmit-receive surface coil placed under the right quadriceps muscle allowed estimation of intramuscular [PCr]; V(O(2)) was measured breath-by-breath using a custom-designed turbine and a mass spectrometer system. For moderate exercise, the kinetics were well described by a simple mono-exponential function (following a short cardiodynamic phase for V(O(2))), with time constants (tau) averaging: tauV(O(2))(,on) 35 +/- 14 s (+/- S.D.), tau[PCr](on) 33 +/- 12 s, tauV(O(2))(,off) 50 +/- 13 s and tau[PCr](off) 51 +/- 13 s. The kinetics for both V(O(2)) and [PCr] were more complex for high-intensity exercise. The fundamental phase expressing average tau values of tauV(O(2))(,on) 39 +/- 4 s, tau[PCr](on) 38 +/- 11 s, tauV(O(2))(,off) 51 +/- 6 s and tau[PCr](off) 47 +/- 11 s. An associated slow component was expressed in the on-transient only for both V(O(2)) and [PCr], and averaged 15.3 +/- 5.4 and 13.9 +/- 9.1 % of the fundamental amplitudes for V(O(2)) and [PCr], respectively. In conclusion, the tau values of the fundamental component of [PCr] and V(O(2)) dynamics cohere to within 10 %, during both the on- and off-transients to a constant-load work rate of both moderate- and high-intensity exercise. On average, approximately 90 % of the magnitude of the V(O(2)) slow component during high-intensity exercise is reflected within the exercising muscle by its [PCr] response.

Inferences from pulmonary O2 uptake with respect to intramuscular [phosphocreatine] kinetics during moderate exercise in humans

1. In the non-steady state of moderate intensity exercise, pulmonary O2 uptake (Vp,O2) is temporally dissociated from muscle O2 consumption (Vm,O2) due to the influence of the intervening venous blood volume and the contribution of body O2 stores to ATP synthesis. A monoexponential model of Vp,O2 without a delay term, therefore, implies an obligatory slowing of Vp,O2 kinetics in comparison to Vm, O2. 2. During moderate exercise, an association of Vm,O2 and [phosphocreatine] ([PCr]) kinetics is a necessary consequence of the control of muscular oxidative phosphorylation mediated by some function of [PCr]. It has also been suggested that the kinetics of Vp,O2 will be expressed with a time constant within 10 % of that of Vm,O2. 3. Vp,O2 and intramuscular [PCr] kinetics were investigated simultaneously during moderate exercise of a large muscle mass in a whole-body NMR spectrometer. Six healthy males performed prone constant-load quadriceps exercise. A transmit-receive coil under the right quadriceps allowed determination of intramuscular [PCr]; Vp,O2 was measured breath-by-breath, in concert with [PCr], using a turbine and a mass spectrometer system. 4. Intramuscular [PCr] decreased monoexponentially with no delay in response to exercise. The mean of the time constants (tauPCr) was 35 s (range, 20-64 s) for the six subjects. 5. Two temporal phases were evident in the Vp, O2 response. When the entire Vp,O2 response was modelled to be exponential with no delay, its time constant (tau'Vp,O2) was longer in all subjects (group mean = 62 s; range, 52-92 s) than that of [PCr], reflecting the energy contribution of the O2 stores. 6. Restricting the Vp,O2 model fit to phase II resulted in matching kinetics for Vp,O2 (group mean tauVp,O2 = 36 s; range, 20-68 s) and [PCr], for all subjects. 7. We conclude that during moderate intensity exercise the phase II tauVp,O2 provides a good estimate of tauPCr and by implication that of Vm,O2 (tauVm,O2).

Blood-brain barrier permeability is increased in normal-appearing white matter in patients with lacunar stroke and leucoaraiosis

BACKGROUND AND AIM

The pathogenesis of cerebral small-vessel disease (SVD) is incompletely understood. Endothelial dysfunction has been implicated and may result in increased blood-brain barrier (BBB) permeability with leakage of blood constituents into the vessel wall and white matter. We used contrast-enhanced MRI to determine whether there was any evidence for BBB permeability in the white matter of patients with SVD, and whether this was present not only in areas of leucoaraiosis (white-matter lesions) but also in normal-appearing white matter (NAWM).

METHODS

Subjects underwent T1 volumetric MRI before and after bolus injection of contrast. Scanning was continued for 30 min postinjection to determine the contrast-enhancement time course. The mean signal intensity change was plotted against time to calculate the area under the curve values, a parameter related to BBB permeability. Automated brain segmentation and regions of interest analysis were performed to determine 'permeability' in different brain compartments.

RESULTS

Compared with controls (n=15), the SVD patient group (n=24) had signal changes consistent with increased BBB permeability in NAWM (p=0.033). Multivariate regression analyses identified leucoaraiosis grade as an independent predictor of these permeability related signal changes in NAWM after adjustment for age, gender, weight, brain volume, area under the curve in the internal carotid arteries and cardiovascular risk factors.

CONCLUSION

This study provides evidence for increased BBB permeability in SVD, and this is particularly seen in SVD with leucoaraiosis. Its presence in NAWM would be consistent with it playing a causal role in disease pathophysiology.

Magnetic resonance neurography

Radiological methods exist for generating tissue-specific images of bone, vessels, lymphatics, abdominal viscera, and the central nervous system, but there has been no reliable means to generate a clinical image of a nerve. We present the first "image neurogram" and report a method for producing such images by use of commercial magnetic resonance imaging systems. The image depicts a human nerve in situ in relation with a nerve graft, wherein the nerve is rendered in isolation much like a vessel appears in isolation in a subtraction angiogram.

Issues in flow and oxygenation dependent contrast (FLOOD) imaging of tumours

The sensitivity of blood oxygenation level dependent (BOLD) contrast techniques to changes to tumour deoxyhaemoglobin concentration is of relevance to many strategies in cancer treatments. In the context of tumour studies, which frequently involve the use of agents to modify blood flow, there are underlying physiological changes different to those of BOLD in the brain. Hence we use the term, flow and oxygenation dependent (FLOOD) contrast, to emphasize this difference and the importance of flow effects. We have measured the R(2)* changes in a prolactinoma tumour model for a variety of vasoactive challenges [carbogen, 100% oxygen and 100% nitrogen as different breathing gases, and administration of tumour blood flow modifiers such as calcitonin gene related peptide (CGRP), hydralazine and nicotinamide]. In addition we have measured other relevant physiological parameters, such as bioenergetic status from (31)P MRS, and blood pH and glucose, that may change during a vasoactive challenge. Here we discuss how they relate to our understanding of FLOOD contrast in tumours. We frequently observe R(2)* changes that match the expected action of the vascular stimulus: R(2)* decreases with agents expected to improve tumour oxygenation and blood flow, and increases with agents designed to increase tumour hypoxia. Unlike most normal tissues, tumours have a chaotic and poorly regulated blood supply, and a mix of glycolytic and oxidative metabolism; thus the response to a vasoactive challenge is not predictable. Changes in blood volume can counteract the effect of blood oxygenation changes, and changes in blood pH and glucose levels can alter oxygen extraction. This can lead to R(2)* changes that are smaller or the reverse of those expected. To properly interpret FLOOD contrast changes these effects must be accounted for.

Application of magnetic resonance neurography in the evaluation of patients with peripheral nerve pathology

Currently, diagnosis and management of disorders involving nerves are generally undertaken without images of the nerves themselves. The authors evaluated whether direct nerve images obtained using the new technique of magnetic resonance (MR) neurography could be used to make clinically important diagnostic distinctions that cannot be readily accomplished using existing methods. The authors obtained T2-weighted fast spin-echo fat-suppressed (chemical shift selection or inversion recovery) and T1-weighted images with planes parallel or transverse to the long axis of nerves using standard or phased-array coils in healthy volunteers and referred patients in 242 sessions. Longitudinal and cross-sectional fascicular images readily distinguished perineural from intraneural masses, thus predicting both resectability and requirement for intraoperative electrophysiological monitoring. Fascicle pattern and longitudinal anatomy firmly identified nerves and thus improved the safety of image-guided procedures. In severe trauma, MR neurography identified nerve discontinuity at the fascicular level preoperatively, thus verifying the need for surgical repair. Direct images readily demonstrated increased diameter in injured nerves and showed the linear extent and time course of image hyperintensity associated with nerve injury. These findings confirm and precisely localize focal nerve compressions, thus avoiding some exploratory surgery and allowing for smaller targeted exposures when surgery is indicated. Direct nerve imaging can demonstrate nerve continuity, distinguish intraneural from perineural masses, and localize nerve compressions prior to surgical exploration. Magnetic resonance neurography can add clinically useful diagnostic information in many situations in which physical examinations, electrodiagnostic tests, and existing image techniques are inconclusive.

Effects of prior exercise on oxygen uptake and phosphocreatine kinetics during high-intensity knee-extension exercise in humans

1. A prior bout of high-intensity square-wave exercise can increase the temporal adaptation of pulmonary oxygen uptake (.V(O2)) to a subsequent bout of high-intensity exercise. The mechanisms controlling this adaptation, however, are poorly understood. 2. We therefore determined the dynamics of intramuscular [phosphocreatine] ([PCr]) simultaneously with those of .V(O2) in seven males who performed two consecutive bouts of high-intensity square-wave, knee-extensor exercise in the prone position for 6 min with a 6 min rest interval. A magnetic resonance spectroscopy (MRS) transmit-receive surface coil under the quadriceps muscle allowed estimation of [PCr]; .V(O2) was measured breath-by-breath using a custom-designed turbine and a mass spectrometer system. 3. The .V(O2) kinetics of the second exercise bout were altered compared with the first such that (a) not only was the instantaneous rate of .V(O2) change (at a given level of .V(O2)) greater but the phase II tau was also reduced - averaging 46.6 +/- 6.0 s (bout 1) and 40.7 +/- 8.4 s (bout 2) (mean +/- S.D.) and (b) the magnitude of the later slow component was reduced. 4. This was associated with a reduction of, on average, 16.1% in the total exercise-induced [PCr] decrement over the 6 min of the exercise, of which 4.0% was due to a reduction in the slow component of [PCr]. There was no discernable alteration in the initial rate of [PCr] change. The prior exercise, therefore, changed the multi-compartment behaviour towards that of functionally first-order dynamics. 5. These observations demonstrate that the .V(O2) responses relative to the work rate input for high-intensity exercise are non-linear, as are, it appears, the putative phosphate-linked controllers for which [PCr] serves as a surrogate.

Magnetic resonance neurography

We have made cross-sectional image "neurograms" in which peripheral nerve has a greater signal intensity than that of other tissue. Neurographic images of the rabbit forelimb were obtained using a spin-echo magnetic resonance imaging (MRI) technique that combines fat suppression and diffusion weighting. After fat suppression the nerve shows up in relative isolation and is brighter than the surrounding tissue due to its longer T2 relaxation time of approximately 50 ms compared to approximately 27 ms for muscle. The addition of pulsed gradients for diffusion weighting of the MR signal further enhances the intensity of the nerve signal relative to that of surrounding muscle tissue. The greater diffusional anisotropy of nerve tissue (D parallel/D perpendicular = 3.1) compared to that of muscle (D parallel/D perpendicular = 1.9) allows further enhancement of the nerve by a subtraction of two diffusion-weighted images, one with the gradients oriented parallel and one with the gradients oriented perpendicular to the nerve orientation. We show that by manipulation of the MRI parameters, either echo time or pulsed gradient strength, the nerves can be made to show up as the most intense feature. This verifies the feasibility of generating three-dimensional "neurographic" images, analogous to angiograms, but which demonstrate the peripheral nerve tracts in apparent isolation.

The response of human tumors to carbogen breathing, monitored by Gradient-Recalled Echo Magnetic Resonance Imaging

PURPOSE

Gradient-Recalled Echo (GRE) Magnetic Resonance Imaging (MRI), which detects changes in blood vessel deoxyhaemoglobin content, has been investigated as a noninvasive monitor of changes in human tumor oxygenation and blood flow, in response to carbogen (95% O2, 5% CO2) breathing.

METHODS AND MATERIALS

GRE images (TE = 60 ms, TR = 200 ms, alpha = 40 degrees, 256[2] matrix) were acquired from 31 patients with primary and metastatic disease, prior to and during carbogen breathing. Three patients underwent a follow-up examination after radiotherapy.

RESULTS

Seventeen out of 34 tumors showed enhanced image intensity, consistent with an improvement in tumor oxygenation and blood flow, while 11 showed no response; 6 studies were technical failures. In one patient a metastatic node that had eluded orthodox investigation was visualized. A reduction in response was observed in the three patients studied postradiotherapy.

CONCLUSION

This method, which can be performed on a standard clinical MRI instrument, provides a noninvasive measurement of tumor response to oxygenation/blood flow modification. In principle, this should enable the clinician to optimize treatment protocols, such as carbogen breathing, for individual radiotherapy patients.

Classification of brain tumours using short echo time 1H MR spectra

The purpose was to objectively compare the application of several techniques and the use of several input features for brain tumour classification using Magnetic Resonance Spectroscopy (MRS). Short echo time 1H MRS signals from patients with glioblastomas (n = 87), meningiomas (n = 57), metastases (n = 39), and astrocytomas grade II (n = 22) were provided by six centres in the European Union funded INTERPRET project. Linear discriminant analysis, least squares support vector machines (LS-SVM) with a linear kernel and LS-SVM with radial basis function kernel were applied and evaluated over 100 stratified random splittings of the dataset into training and test sets. The area under the receiver operating characteristic curve (AUC) was used to measure the performance of binary classifiers, while the percentage of correct classifications was used to evaluate the multiclass classifiers. The influence of several factors on the classification performance has been tested: L2- vs. water normalization, magnitude vs. real spectra and baseline correction. The effect of input feature reduction was also investigated by using only the selected frequency regions containing the most discriminatory information, and peak integrated values. Using L2-normalized complete spectra the automated binary classifiers reached a mean test AUC of more than 0.95, except for glioblastomas vs. metastases. Similar results were obtained for all classification techniques and input features except for water normalized spectra, where classification performance was lower. This indicates that data acquisition and processing can be simplified for classification purposes, excluding the need for separate water signal acquisition, baseline correction or phasing.

Dynamics of intramuscular 31P-MRS P(i) peak splitting and the slow components of PCr and O2 uptake during exercise

The dynamics of pulmonary O(2) uptake (Vo(2)) during the on-transient of high-intensity exercise depart from monoexponentiality as a result of a "slow component" whose mechanisms remain conjectural. Progressive recruitment of glycolytic muscle fibers, with slow O(2) utilization kinetics and low efficiency, has, however, been suggested as a mechanism. The demonstration of high- and low-pH components of the exercising skeletal muscle (31)P magnetic resonance (MR) spectrum [inorganic phosphate (P(i)) peak] at high work rates (thought to be reflective of differences between oxidative and glycolytic muscle fibers) is also consistent with this conjecture. We therefore investigated the dynamics of Vo(2) (using a turbine and mass spectrometry) and intramuscular ATP, phosphocreatine (PCr), and P(i) concentrations and pH, estimated from the (31)P MR spectrum. Eleven healthy men performed prone square-wave high-intensity knee extensor exercise in the bore of a whole body MR spectrometer. A Vo(2) slow component of magnitude 15.9 +/- 6.9% of the phase II amplitude was accompanied by a similar response (11.9 +/- 7.1%) in PCr concentration. Only five subjects demonstrated a discernable splitting of the P(i) peak, however, which began from between 35 and 235 s after exercise onset and continued until cessation. As such, the dynamics of the pH distribution in intramuscular compartments did not consistently reflect the temporal features of the Vo(2) slow component, suggesting that P(i) splitting does not uniquely reflect the activity of oxidative or glycolytic muscle fibers per se.

Continuing ischemic damage after acute middle cerebral artery infarction in humans demonstrated by short-echo proton spectroscopy

BACKGROUND AND PURPOSE

Proton MR spectroscopy is a noninvasive method of monitoring in vivo metabolite concentration changes over time. The aim of this work was to study the ischemic penumbra in humans by measuring the metabolic changes that occur after a middle cerebral artery territory infarction.

METHODS

Diagnostic MRI and short-echo time MR spectroscopy were performed on a 1.5-T system. Localized proton MR spectroscopy was performed within the area of cerebral infarction and in a homologous area of the contralateral hemisphere. The residual water resonance in the spectra was removed with the use of the Hankel Lanczos singular value decomposition method, after which peak area estimates were obtained by means of the variable projection time domain fitting analysis. The unsuppressed water signal was used as an internal concentration standard. Ten patients with acute middle cerebral artery infarction were studied within 28 hours of stroke onset and followed up for a period of up to 3 months.

RESULTS

Significant changes were seen in the initial spectra from the infarct compared with the contralateral spectra. Lactate, a marker of anaerobic metabolism, was present within the infarct but not detected in the contralateral hemisphere. N-Acetyl aspartate, a neuronal marker, and total creatine were significantly reduced. The initial choline signal, arising from choline-containing compounds within the cell and cell membrane, remained unchanged in the infarct core compared with the contralateral hemisphere. Further reductions in N-acetyl aspartate and total creatine concentrations occurred within the first week. A fall in the lactate concentration was seen within the infarct core during the first 7 to 10 days. Similar reductions in the choline concentration were observed during this period.

CONCLUSIONS

The demonstration of the continuing loss of cerebral metabolites within an infarct region suggests that further cell loss occurs up to 10 days after infarction. The continuing loss of neurons may represent continued ischemic damage after middle cerebral artery infarction.

Aging of the adult human brain: in vivo quantitation of metabolite content with proton magnetic resonance spectroscopy

The purpose of this study was to examine the effect of aging on brain metabolite concentrations, including N-acetyl aspartate (NAA), the major marker of neurones, using short echo proton spectroscopy. Single-voxel proton spectra (TE 30 msec, TR 2 seconds) were obtained from white and gray matter using automated software (PROBE, G.E., Milwaukee, WI). Spectra were analyzed using the variable projection technique. Thirty healthy volunteers were studied within the age range 24-89 years. No significant trend in change of concentrations of NAA, total creatine, total choline or myo-inositol were seen with age. The total creatine concentration of parietal white matter in the over 60 age group (6.5 +/- 0.3 mmol/l) was significantly higher than the under 60 age group (6.0 +/- 0.4 mmol/l:; P < 0.05). No other significant difference between the two age groups was seen. The tissue concentration of the major neuronal marker, NAA, does not decline with age. No age-related changes in the concentrations of choline and myo-inositol and occipital gray matter total creatine were observed. These results provide a normal range of values for spectroscopically detectable metabolites within the regions studied, against which neurological diseases such as Alzheimer's disease can be compared in vivo.

The clinical value of proton magnetic resonance spectroscopy in adult brain tumours

Proton magnetic resonance spectroscopy (1H MRS) non-invasively provides information on the biochemical profile (typically including up to nine metabolites and mobile lipids) of brain tissue, which varies according to the underlying disease process. A number of studies have assessed its accuracy in the diagnosis of adult brain tumours. This article describes the basic principles of 1H MRS, the metabolic profiles of different brain tumours, and practical points to aid interpretation of spectra. The literature is reviewed regarding the role of 1H MRS in the diagnosis of brain tumours and more specifically where it has proven to be of additional benefit over conventional magnetic resonance imaging.

Tumour dose response to the antivascular agent ZD6126 assessed by magnetic resonance imaging

ZD6126 is a vascular targeting agent that disrupts the tubulin cytoskeleton of proliferating neo-endothelial cells. This leads to the selective destruction and congestion of tumour blood vessels in experimental tumours, resulting in extensive haemorrhagic necrosis. In this study, the dose-dependent activity of ZD6126 in rat GH3 prolactinomas and murine RIF-1 fibrosarcomas was assessed using two magnetic resonance imaging (MRI) methods. Dynamic contrast-enhanced (DCE) MRI, quantified by an initial area under the time-concentration product curve (IAUC) method, gives values related to tumour perfusion and vascular permeability. Multigradient recalled echo MRI measures the transverse relaxation rate T(2)*, which is sensitive to tissue (deoxyhaemoglobin). Tumour IAUC and R(2)* (=1/T(2)*) decreased post-treatment with ZD6126 in a dose-dependent manner. In the rat model, lower doses of ZD6126 reduced the IAUC close to zero within restricted areas of the tumour, typically in the centre, while the highest dose reduced the IAUC to zero over the majority of the tumour. A decrease in both MRI end points was associated with the induction of massive central tumour necrosis measured histologically, which increased in a dose-dependent manner. Magnetic resonance imaging may be of value in evaluation of the acute clinical effects of ZD6126 in solid tumours. In particular, measurement of IAUC by DCE MRI should provide an unambiguous measure of biological activity of antivascular therapies for clinical trial.

Brain tumor classification based on long echo proton MRS signals

There has been a growing research interest in brain tumor classification based on proton magnetic resonance spectroscopy (1H MRS) signals. Four research centers within the EU funded INTERPRET project have acquired a significant number of long echo 1H MRS signals for brain tumor classification. In this paper, we present an objective comparison of several classification techniques applied to the discrimination of four types of brain tumors: meningiomas, glioblastomas, astrocytomas grade II and metastases. Linear and non-linear classifiers are compared: linear discriminant analysis (LDA), support vector machines (SVM) and least squares SVM (LS-SVM) with a linear kernel as linear techniques and LS-SVM with a radial basis function (RBF) kernel as a non-linear technique. Kernel-based methods can perform well in processing high dimensional data. This motivates the inclusion of SVM and LS-SVM in this study. The analysis includes optimal input variable selection, (hyper-) parameter estimation, followed by performance evaluation. The classification performance is evaluated over 200 stratified random samplings of the dataset into training and test sets. Receiver operating characteristic (ROC) curve analysis measures the performance of binary classification, while for multiclass classification, we consider the accuracy as performance measure. Based on the complete magnitude spectra, automated binary classifiers are able to reach an area under the ROC curve (AUC) of more than 0.9 except for the hard case glioblastomas versus metastases. Although, based on the available long echo 1H MRS data, we did not find any statistically significant difference between the performances of LDA and the kernel-based methods, the latter have the strength that no dimensionality reduction is required to obtain such a high performance.

Noninvasive monitoring of carbogen-induced changes in tumor blood flow and oxygenation by functional magnetic resonance imaging

PURPOSE

The response of tumors to radiotherapy can be enhanced if carbogen (95% O2, 5% CO2) is breathed. The timing of carbogen administration is critical, and a noninvasive method of monitoring the response of individual tumors would have obvious utility. Functional gradient recalled echo (GRE) magnetic resonance imaging (MRI) techniques are sensitive to changes in the concentrations of deoxyhemoglobin, which, thus, acts as an endogenous contrast agent for oxygenation status and blood flow.

METHODS AND MATERIALS

Subcutaneous GH3 prolactinomas in three rats were imaged at 4.7 Tesla with a GRE 1H sequence [echo time (TE) = 20 ms, repetition time (TR) = 80 ms, flip angle = 45 degrees, 1 mm slice, 256 phase encode steps, 4 cm field of view, in-plane resolution 0.08 x 0.08 mm, acquisition time = 4 min]. The rats breathed air or carbogen for four periods of 20 min; three control rats breathed only air.

RESULTS

Carbogen breathing caused increases of up to 100% in the GRE image intensity of the tumors. Reversion of air breathing caused the image intensity to fall; essentially the same response was observed with the second cycle of carbogen and air breathing. Control rat tumors showed no significant change.

CONCLUSIONS

The response of tumors to carbogen can be monitored noninvasively by GRE MRI. In principle, this could be due to an increase in oxygen content of the blood, a decrease in tumor cell oxygen consumption, or an increase in tumor blood flow. The very large changes in signal intensity suggest that a blood flow increase is the most probable explanation. If this technique can be successfully applied in man, it should be possible to optimize carbogen treatment for individual radiotherapy patients, and perhaps also to enhance tumor uptake of chemotherapeutic agents.

Detection of elevated glutathione in meningiomas by quantitative in vivo 1H MRS

Glutathione has major roles in removing free radicals and toxins from normal tissues, but its presence in tumor cells hinders the effectiveness of many anticancer therapies. Analysis of short echo time brain tumor (1)H spectra at 1.5 T using a linear combination of metabolite spectra (LCModel) suggested a significant contribution of glutathione to meningioma spectra. By in vivo MRS (TE = 30 ms, TR = 2020 ms), reduced glutathione was found to be significantly elevated in meningiomas (3.3 +/- 1.5 mM, Mann Whitney, P < 0.005) compared to normal white matter (1.2 +/- 0.15 mM) and low-grade gliomas (1.0 +/- 0.26 mM), in agreement with published histofluorescence studies of tumor biopsies. Glx concentrations were also found to be elevated in meningiomas compared to astrocytomas or normal white matter, indicative of metabolic differences. The ability to noninvasively quantify reduced glutathione in vivo may aid selection of treatment therapies and also provide an indication of tumor aggressiveness.

Flow and oxygenation dependent (FLOOD) contrast MR imaging to monitor the response of rat tumors to carbogen breathing

Gradient recalled echo (GRE) images are sensitive to both paramagnetic deoxyhaemoglobin concentration (via T2*) and flow (via T1*). Large GRE signal intensity increases have been observed in subcutaneous tumors during carbogen (5% carbon dioxide, 95% oxygen) breathing. We term this combined effect flow and oxygenation-dependent (FLOOD) contrast. We have now used both spin echo (SE) and GRE images to evaluate how changes in relaxation times and flow contribute to image intensity contrast changes. T1-weighted images, with and without outer slice suppression, and calculated T2, T2* and "flow" maps, were obtained for subcutaneous GH3 prolactinomas in rats during air and carbogen breathing. T1-weighted images showed bright features that increased in size, intensity and number with carbogen breathing. H&E stained histological sections confirmed them to be large blood vessels. Apparent T1 and T2 images were fairly homogeneous with average relaxation times of 850 ms and 37 ms, respectively, during air breathing, with increases of 2% for T1 and 11% for T2 during carbogen breathing. The apparent T2* over all tumors was very heterogeneous, with values between 9 and 23 ms and localized increases of up to 75% during carbogen breathing. Synthesised "flow" maps also showed heterogeneity, and regions of maximum increase in flow did not always coincide with maximum increases in T2*. Carbogen breathing caused a threefold increase in arterial rat blood PaO2, and typically a 50% increase in tumor blood volume as measured by 51Cr-labelled RBC uptake. The T2* increase is therefore due to a decrease in blood deoxyhaemoglobin concentration with the magnitude of the FLOOD response being determined by the vascular density and responsiveness to blood flow modifiers. FLOOD contrast may therefore be of value in assessing the magnitude and heterogeneity of response of individual tumors to blood flow modifiers for both chemotherapy, antiangiogenesis therapy in particular, and radiotherapy.

Effects of dichloroacetate on VO2 and intramuscular 31P metabolite kinetics during high-intensity exercise in humans

Traditional control theories of muscle O2 consumption are based on an "inertial" feedback system operating through features of the ATP splitting (e.g., [ADP] feedback, where brackets denote concentration). More recently, however, it has been suggested that feedforward mechanisms (with respect to ATP utilization) may play an important role by controlling the rate of substrate provision to the electron transport chain. This has been achieved by activation of the pyruvate dehydrogenase complex via dichloroacetate (DCA) infusion before exercise. To investigate these suggestions, six men performed repeated, high-intensity, constant-load quadriceps exercise in the bore of an magnetic resonance spectrometer with each of prior DCA or saline control intravenous infusions. O2 uptake (Vo2) was measured breath by breath (by use of a turbine and mass spectrometer) simultaneously with intramuscular phosphocreatine (PCr) concentration ([PCr]), [Pi], [ATP], and pH (by 31P-MRS) and arterialized-venous blood sampling. DCA had no effect on the time constant (tau) of either Vo2 increase or PCr breakdown [tauVo2 45.5 +/- 7.9 vs. 44.3 +/- 8.2 s (means +/- SD; control vs. DCA); tauPCr 44.8 +/- 6.6 vs. 46.4 +/- 7.5 s; with 95% confidence intervals averaging < +/-2 s]. DCA, however, resulted in significant (P < 0.05) reductions in 1). end-exercise [lactate] (-1.0 +/- 0.9 mM), intramuscular acidification (pH, +0.08 +/- 0.06 units), and [Pi] (-1.7 +/- 2.1 mM); 2). the amplitude of the fundamental components for [PCr] (-1.9 +/- 1.6 mM) and Vo2 (-0.1 +/- 0.07 l/min, or 8%); and 3). the amplitude of the Vo2 slow component. Thus, although the DCA infusion lessened the buildup of potential fatigue metabolites and reduced both the aerobic and anaerobic components of the energy transfer during exercise, it did not enhance either tauVo2 or tau[PCr], suggesting that feedback, rather than feedforward, control mechanisms dominate during high-intensity exercise.

The response to carbogen breathing in experimental tumour models monitored by gradient-recalled echo magnetic resonance imaging

Gradient-recalled echo magnetic resonance imaging (GRE MRI), which gives information on blood flow and oxygenation changes (Robinson SP, Howe FA, Griffiths JR 1995, Int J Radiat Oncol Biol Phys 33: 855), was used to observe the responses of six rodent tumour models to carbogen breathing. In one transplanted rat tumour, the Morris hepatoma 9618a, and a chemically induced rat tumour, the MNU-induced mammary adenocarcinoma, there were marked image intensity increases, similar to those previously observed in the rat GH3 prolactinoma. In contrast, the rat Walker carcinosarcoma showed no response. In two mouse tumours, the RIF-1 fibrosarcoma and the human xenograft HT29, carbogen breathing induced a transient fall in signal intensity that reversed spontaneously within a few minutes. The rat GH3 prolactinoma was xenografted into nude mice, and an increase in image intensity was found in response to carbogen, suggesting that any effects that carbogen may have had on the host were not significant determinants of the tumour response. The increases in GRE image intensity of the MNU, H9618a and GH3 tumours during carbogen breathing are consistent with increases in tumour oxygenation and blood flow, whereas the responses of the RIF-1 and HT29 tumours may be the result of a transient steal effect followed by homeostatic correction.

Magnetic resonance imaging techniques for monitoring changes in tumor oxygenation and blood flow

The application of functional magnetic resonance (MR) imaging techniques to the measurement of oxygenation and blood flow in tumors is described. Gradient recalled echo MR imaging (GRE-MRI) offers a real-time noninvasive method for monitoring tumor response to vasomodulators such as carbogen (95% O2/5% CO2) breathing in attempts to overcome tumor hypoxia and improve treatment efficacy. Although the response is tumor-type dependent, increases in signal intensity of up to 100% have been observed in several animal tumor types. Responses are also seen in human tumors. The observed increases in GRE-MRI signal intensity are due to a combination of a reduction of deoxyhemoglobin in the blood causing changes in the MR imaging relaxation time T2* and changes in blood flow and may also reflect the capillary density. Thus, the magnitude of the GRE image intensity change gives an indication of the potential response of an individual tumor to treatments that aim to improve tissue oxygenation and therefore how the tumor may respond to therapy. In addition, carbogen breathing by the host has been shown to increase the uptake and efficacy of chemotherapeutic agents in animal tumors.

The contribution of proton magnetic resonance spectroscopy (1HMRS) to clinical brain tumour diagnosis

Proton magnetic resonance spectroscopy (1HMRS) provides biochemical information from tissue non-invasively, and has an evolving role in brain tumour diagnosis and management. We present 100 consecutive patients with brain tumours who had single voxel 1HMRS as part of their preoperative investigations. We report the histopathological findings and the diagnostic contribution of spectroscopy in an adjunctive role. On the basis of clinical and radiological information the preoperative diagnosis was unclear or inaccurate in 26 out of 100 cases. The discrepancy was of lesion grade in 17 cases and lesion type in 9 cases. In 6 of 100 patients with brain tumours 1HMRS could have made a significant contribution to the preoperative diagnosis if used as part of the routine assessment. There is therefore a useful role for 1HMRS in the evaluation of intracranial mass lesions.