MAGMA (2003) 16:29–42 DOI 10.1007/s10334-003-0004-x

Alexander P. Lin Frederick Shic Cathleen Enriquez Brian D. Ross

Received: 15 July 2002 Accepted: 5 November 2002 Published online: 21 February 2003 © ESMRMB 2003 Submitted by Alexander P Lin, Winner, in partial fulfillment of the requirements of the Young Investigator Award Finalists, European Society of Magnetic Resonance in Medicine and Biology (ESMRMB), Cannes, France, 2002

A. P. Lin (✉) · B. D. Ross Clinical Spectroscopy Unit, Huntington Medical Research Institutes, 660 S. Fair Oaks Avenue, Pasadena, CA, 91105, USA e-mail: [email protected] Tel.: +1-626-3075840 Fax: +1-626-3975846 F. Shic · C. Enriquez Rudi Schulte Research Institute, Santa Barbara, CA, USA

R E S E A R C H A RT I C L E

Reduced glutamate neurotransmission in patients with Alzheimer’s disease – an in vivo 13C magnetic resonance spectroscopy study

Abstract Cognitive impairment in Alzheimer’s disease (AD) is not fully explained. PET indicates reduced cerebral metabolic rate for glucose. Since glutamate neurotransmission (GNT) consumes more than 80% of the ATP generated from metabolism, a pilot study was carried out to determine the neuronal tricarboxylic acid cycle (TCA) based on the hypothesis that reduced GNT could contribute to cognitive impairment in AD. Three AD patients with cognitive impairment (mini-mental state exam: 24 vs 30, P<0.05) and significant reduction in both N-acetyl aspartate (NAA)/Creatine (Cr) (P<0.009) and NAA/myo-inositol (mI) ratio (P<0.01), and three age-matched controls each received 0.014–0.016 g/kg/min 99%1–13C glucose IV. Quantitative 1H and proton-decoupled 13C MR brain spectra were acquired from combined posterior-parietal white matter and posterior-cingulate gray matter every 5 min for 140 min. 13C magnetic resonance spectroscopy (MRS) measures of glucose oxidation and neuronal TCA rate, in-

Introduction Underlying biochemical abnormalities have long been sought to explain the symptoms and pathobiology of Alzheimer’s disease (AD) [1–10]. A favored hypothesis identifies the predominantly cholinergic neurons within the hippocampus and limbic systems as the first to be

cluding prolonged time to 13C enrichment of glutamate (Glu2) (P<0.004) and bicarbonate (HCO3) (P<0.03) as well as reduced relative enrichment of Glu2/Glu4 between 60 and 100 min (P<0.04), were significantly different in AD patients vs. controls. 13C measures of GNT, glutamine (Gln)2/Glu2 (P<0.02) and rates of glutamate enrichment (Glu2/glucose: 0.34 vs 0.86, P=ns and Glu4/glucose 0.26 vs 0.83, P=ns), were also reduced. 13C MRS measures of neuronal TCA cycle, glucose oxidation and GNT were significantly correlated with measures of neuronal integrity: NAA/Cr, [NAA] and mI/NAA as determined by 1H MRS (R2=0.73–0.95; P<0.05–0.01), suggesting that impairment of GNT may be a contributing factor in the cognitive impairment characteristic of AD. Keywords Alzheimer’s disease · Glutamate-glutamine cycle · 13C Magnetic resonance spectroscopy · 13C Glucose infusion · Glutamate neurotransmission · N-acetyl aspartate · Mini-mental state exam

lost in AD, thereby contributing to memory loss [11]. Neuronal loss in AD patients is not confined to the hippocampus, however, as confirmed by atrophy on magnetic resonance imaging (MRI) and by proton magnetic resonance spectroscopy (1H MRS) findings [12]. N-acetyl aspartate (NAA) is an amino-acid derivative synthesized in neurons and transported down axons and is

47 60 27 43.5±23.3 0.25 F M M Mean±SD p value

1 2 3 Controls

Mean±SD

lipid peaks and do not affect 13C measurements (unpublished work from laboratory)

n.a. 9.97 10.89 10.42±0.65 0.05 0.45 0.44 0.44 0.44±0.003 0.01 0.58 0.62 0.64 0.61±0.03 0.09 1.31 1.39 1.33 1.38±0.06 0.009 V−, A− V−, A− V−, A− 30 30 30 30.0±0.0 0.05

27 29 17 24.3±6.4 1 <1 2 71 53 56 60.0±9.6 F F F 1 2 3

a Significant gender differences for 1H and 13C glucose MRS are confined to subcutaneous b Patient had a snack after overnight fast and prior to 1-13C glucose infusion

0.0147 0.0132 0.0145 0.014±0.002 0.14

0.0143 0.0145 0.0169 0.015±0.007 Fastedb Fasted Fasted V+, A+ V−, A+ V+, A+

1.10 0.91 1.11 1.04±0.11

0.83 0.66 0.70 0.73±0.09

0.76 0.72 0.63 0.70±0.07

9.17 7.21 9.17 8.51±1.13

Fasted Fasted Fasted

1-13C glucose dose (g/kg/min) mI/NAA mI/Cr

Duration of disease (years)

MMSE

MRI

NAA/Cr

1H

MRS results

[NAA]

1-13C glucose infusion

Fed/fasted

Patients

Three AD patients (age 53–71 years) were examined with quantitative 1H and 13C MRS using a [1-13C] glucose infusion protocol. The results of the combined examination were compared with those from three control subjects of similar mean age (P>0.25). Details of age, gender, clinical diagnoses, and the 13C infusion protocols of all six subjects are summarized in Table 1. Diagnosis and grading of AD according to DSMIII-R criteria and MMSE were performed independently. Patients (with one exception, noted in Table 1) and control subjects fasted for a minimum of 12 h (overnight) prior to [1-13C] glucose administration. [1-13C] glucose (99% enriched; Cambridge Isotope Laboratories, Andover, Mass.) was prepared on a named-patient basis as a pyrogen-free (Celsis Labs, St. Louis, Mo.), sterile, 20% solution (Fairview University Hospital, Minneapolis, Minn.). Residues of glucose infusions were stored for bacteriological, chemical, and NMR analyses. The glucose was administered intravenously as a single dose over 15 min (N=5: one AD patient received a marginally higher

Age

Patients and methods

Sexa

therefore a valuable in vivo “marker” of viable neurons, axons and dendrites [13]. The cerebral concentration of NAA can be directly measured using 1H MRS and has been shown to be decreased in AD patients [14–16]. This observation, in combination with the finding of an increase in myo-inositol (mI), a possible marker of astrocytes, readily detected in short echo-time (TE) 1H MRS, provides a non-invasive test for AD [17] that is 100% sensitive and 75% specific when compared to autopsy results [18]. Using 1H MRS, the concentration of glutamate plus glutamine, important neurotransmitters, was shown to be markedly decreased in vivo [14, 19–21], an observation confirmed by post-mortem enzymatic assay of brains of AD patients [22]. Failure of energy metabolism due to “oxidative stress” has been proposed in neurodegenerative diseases, including AD [23]. While direct determination by in vivo 31P MRS failed to confirm ATP depletion in AD [19, 24–26], PET studies strongly suggest that there are important regional reductions in glucose oxidative rate [27]. Glucose hypometabolism correlates with cognitive deterioration as measured by mini-mental state exam (MMSE) scores [28]. PET measures glucose metabolic rate but not actual neurotransmitter function [29]. Glutamate neurotransmission (GNT) is carried out by a glialneuronal process that includes the oxidation of glucose and the glutamine-glutamate cycle [30]. Together with other components of the electrical action potential, GNT consumes more than 80% of the ATP generated from glucose oxidation [31]. Incorporation of 1-13C-labeled glucose into cerebral glutamate (Glu) and glutamine (Gln) pools allows for direct in vivo measurement of GNT using proton-decoupled 13C MRS [32–35]. We undertook a pilot study using 13C MRS after 1-13C glucose infusion in a small number of AD patients and controls. The goals of the study were: (1) to determine glucose oxidation and GNT in AD patients, and (2) to correlate GNT and glucose oxidation with degree of neuronal loss.

Table 1 Patient demographics. Patients with mild-moderate Alzheimer’s disease and age-matched controls were examined. Controls were healthy subjects with no history of neurological or psychological disease. V Ventricular dilatation, A cortical atrophy, ± increased/normal, n.a. not available.

30

31

Fig. 1 Magnetic resonance imaging (MRI) and proton magnetic resonance spectroscopy (1H MRS) of a representative patient with Alzheimer’s disease (AD) based on clinical criteria. AD diagnosis in a female aged 53 [mini-mental state exam (MMSE)=29] with nearly normal MRI. The marked reduction in N-acetyl aspartate (NAA)/Cr and increased myo-inositol (mI)/Cr in single-voxel 1H MRS, carried out by stimulated echo acquisition mode (STEAM) (left) or point-resolved spectroscopy (PRESS) (right), compared to an age-matched normal subject also fitted the diagnostic criteria [17]. The voxel indicates the middle slice of the STEAM acquisition and is centered in the posterior cingulate gyrus (gray matter)

cluded in the MRS studies could be clearly defined. 1H spectra from a voxel containing mostly gray matter within the posterior cingulate gyrus were acquired from each subject (as shown in Fig. 1), with stimulated echo acquisition mode (STEAM) sequence (TR=1.5 s, TM=13.7 ms, TE=30 ms, voxel size=11.34 cc [17]) and point-resolved spectroscopy (PRESS), (TR=1.5 s, TE=35 ms, voxel size=8 cc [15]). The purpose of 1H MRS was to confirm the clinical diagnosis of probable AD by the typical 1H MRS pattern of decreased NAA, increased mI [and reduced glutamate/glutamine (Glx)] [17]. Absolute concentrations of cerebral metabolites (in mmol/kg brain tissue) measured in this 1H MRS examination were used for quantification of 13C spectra (see below) [36–38].

dose over 37 min). The use of somatostatin to prevent insulin secretion was omitted. Serial blood samples (every 10 or 20 min) and final urine samples were collected from the controls. Initial and final blood and urine samples were obtained from the patients. Total plasma glucose was obtained in situ using a Surestep Pro bedside unit (Lifescan, Milpitas, Calif.). Fractional 13C-enriched glucose levels in deproteinized plasma and urine were determined by 1H NMR using a Varian 500 Unity INOVA spectrometer. Studies were carried out under FDA IND 56,510. Proton-decoupled 13C MRS was performed on a General Electric Signa 1.5-Tesla clinical scanner equipped with a second channel for decoupling. Informed patient consent and approval of the Internal Review Board of Huntington Memorial Hospital were obtained.

13C

MRI and 1H MRS Using the quadrature General Electric 1H volume head coil, fast spin-echo T2-weighted and T1-weighted MR images of the brain were obtained, from which the anatomy of the region of brain in-

MRS

The half-head surface coil used in this study minimizes power deposition during proton-decoupling to the anterior third of the head and to the eyes. Its construction and performance are described in detail in [39,40]. The coil was localized by mounting the 13C surface coil on a headrest adjacent to the occiput, such that the MR signal was derived largely from cortical gray matter of the occipital lobe and from the overlying scalp. 13C MRS spectra were obtained by a simple rectangular pulse-and-acquire experiment with an excitation bandwidth of 4 kHz (250 ppm chemical-shift range). A low-excitation flip angle was used to minimize any possible variations in the steady-state magnetization due to the different T1 relaxation times caused by pathology. The flip angle was calibrated by minimizing the lipid signal (CH2)n at 30 ppm from the skull adjacent to the 13C coil by a spatial saturation pulse. Due to the very short T1 time of lipids, this radiofrequency (RF) power corresponds closely to a 90° pulse. For the actual acquisition, which did not include outer volume suppression, the pulse amplitude was scaled to achieve a 45° flip angle at this location. Decoupling was

32

carried out with WALTZ-16 [41, 42] and a bandwidth of 640 Hz during the 0.2-s acquisition period. The local specific absorption rate (SAR) with this coil and pulse sequence was well below the FDA limits [40, 43]. The center frequency of the decoupler was set to 2.7 ppm. The receiver bandwidth was 5 kHz, with 1,024 complex data points sampled. The 13C transmitter frequency was set to 90 ppm, and the TR was 1 s. Natural-abundance 13C spectra were acquired for up to 30 min before starting glucose infusion. After the start of glucose infusion, serial spectra with 2.5-min (during the first 10 min after start of infusion) or 5-min time resolution were acquired for up to 140 min. A complete examination, including localizer MRI, 1H MRS, natural-abundance and post-[1-13C] glucose infusion 13C MRS, lasted on average 160 min. Analysis of 13C spectra Manual and automated analysis of 13C spectra In addition to the manual identification [44] and measurement of 13C peaks, natural abundance and 13C-enriched difference spectra were obtained for each time period, and a newly developed observer-independent, fully automated analysis was applied [45]. In this procedure, three alternative methods were employed to maximize the metabolic information obtained from the 25–30 sequential 13C spectra with 1,000–1,250 peaks acquired in each subject. Method A Data were prepared for analysis by first passing all data through normalization algorithms written in IDL 4.0 (Research Systems, Inc. Brulder, Co.). Baseline spectra acquired pre-infusion were averaged together to form a standard reference, and all subsequent spectra were scaled, zero-order-phased, and frequency-shifted via automated correlation procedures in order to minimize the deviation as compared to this baseline. This made it possible, for instance, to separate 4-13C glutamate from the underlying methylenes of lipids in an observer-independent fashion, and provided a metric of regularization for hardware and patient instabilities. The spectra of all subjects were assessed manually for peak heights and the collected values assembled to produce time courses for analysis as described in [44]. The basis for this analysis was the use of linear and exponential modeling in order to capture the dynamics of metabolism. In addition, qualitative measures, such as the time to the first appearance of bicarbonate and glutamate, were assessed. Method B An alternative method of analysis was direct examination of the average of all spectra over a long timespan (60–100 min after the start of infusion). In order to ensure a maximal signal to noise ratio, spectra to be averaged were normalized to one another as described in method A, the sum total correlated to the baseline, and the baseline subtracted. This resulted in a high-resolution snapshot of the behavior of an extensive window of metabolic activity. An automated peak search routine was employed in the measurement of all peaks, incorporating a priori baseline information only for the measurement of Glu4 due to overlap with lipids. Ratios of enrichment were used in order to characterize the differences and similarities in the observable substrate-product relationships of the TCA cycle. Method C In order to examine the time-varying behavior of metabolism, all spectra acquired post-infusion were assembled into 20-min blocks

by averaging as described in Method B. Interpolation was carried out by shifting the averaging window spectra by spectra until the entire examination was fully covered. Once time-course averaging was completed, a spectrum covering the entire exam was computed. Each 20-min data set was renormalized to this global average, and automated peak search routine employed to gather all peak information from all averaged spectra, including the global average. A separate routine, written in PERL, collected the output from the peak gathering step and constructed a database characterizing peak behaviors. The mI signal, as determined from the global average, was used to calculate the signal intensity per unit concentration of mI. Combined with translation constants gathered from in vitro phantom measurements, an estimate of all enriched peak concentrations was determined by referencing against the 13C mI signal intensity and normalizing based on mI concentration as determined by 1H quantitative MRS [35]. Expression of

13C

metabolite results

Metabolite flux determination for principal pathways, TCA cycle, and glutamine synthesis, introduced by Mason et al. [46, 47] is heavily dependent upon a number of assumptions about in vivo activity of compartmentalized enzymes and applies only under very specific conditions of steady-state that were not always achieved in the present study. A simplified approach, which has proven useful in clinical 13C MRS studies [44, 48] was applied to the present data, as follows: Time to appearance of [2-13C] glutamate in acquisitions acquired with a time resolution of 5 min. The time to the appearance of Glu2will be correlated with the net synthesis rate of glutamate and the overall rate of the neuronal and glial TCA cycle. t(HCO3): Time to appearance of bicarbonate (HCO3) as above. The time to appearance of bicarbonate will be directly tied to the metabolic rate of neurons and glia. Glu2/Glu4: The signal intensity of Glu4 was assumed to be directly correlated to its concentration. As α-ketoglutarate (αKG) and Glu are in fast exchange, the fractional enrichment of αKG reflects the fractional enrichment of the larger glutamate pool. Since V_x is equal for both αKG2↔Glu2 and αKG4↔Glu4, Glu2/Glu4 reflects αKG2/αKG4 and thus provides a simplified measure of the efficiency of exactly one turn of the TCA cycle. NAA/Asp: This number tries to categorize the quantity of NAA (as NAA2+NAA3) enriched per quantity of aspartate (Asp2+Asp3) enriched, with the objective of obtaining some quantity proportional to the rate of NAA synthesis [49]. Asp/Glx: If a defect exists in the segment of the TCA cycle between oxaloacetate (OAA) and αKG, it would be apparent in the measured ratio of Asp/Glx as they are both limited through the arrival of label at OAA2, in the second round of the TCA cycle. Gln2/Glu2: This value was computed by a linear model and interpolation to t=100 min and is used in this study as an inverse measure of GNT. Its deviation from control could be indicative of changes in metabolite pool sizes, alterations in metabolic flux rates, or a change in the distribution of the metabolic load of neurons and glia. If the glial TCA cycle is operating faster, then an enrichment pattern with relatively increased glutamine signal versus glutamate signal is expected [49, 50].

t(glu2):

33

Statistics Unpaired t tests were applied to group means ±standard deviation (SD) using two samples of unequal variance and P<0.05 as significant. Correlations between 13C measures of glucose metabolism and GNT with the 1H MRS measures of neuronal loss, [NAA], [mI]/[NAA], mI/NAA and NAA/Cr ratios were made using MS Excel Statistical Analysis Package (Microsoft, Seattle, Wash.) to determine goodness of fit to a linear model (R2) and the Pearson correlation coefficient (r). Significance was established for r [51]. Power calculations (not shown) indicated N=3 for P<0.05 at 80% level for most of the 13C MRS data obtained [44].

Results

Fig. 2 MRI of AD patient acquired in 13C coil. T2-weighted MRI of a representative AD patient acquired with dual-tuned, half-volume 13C surface coil (as described in Materials and methods). The volume of brain examined was approximately 100 cm3 and included gray matter of the posterior cingulate gyrus as well as white matter from both posterior parietal hemispheres

Fig. 3 A Representative spectra acquired after 100 min of 1-13C glucose infusion in a patient with AD and an agematched control. B 13C naturalabundance, 13C-enriched, and difference spectra of AD patients and age-matched controls. Broad-band 13C spectra were acquired without additional localization. The 13C signal from overlying scalp lipid was minimized by appropriate flip angle and repetition time (see Materials and methods). Upper panel: Summed spectra of AD patients 1 and 2. Segments of 13C natural-abundance baseline spectra acquired before (center), and after (top) ~10 g 1-13C glucose infusion (13C-enriched 60–100 min). The difference spectrum showing 13C metabolites enriched in AD patients is below. Lower panel: Summed spectra of normal control subjects, naturalabundance (center), 13C-enriched 60–100 min (top), and difference spectrum (below) showing 13C metabolites enriched in normal controls

Figure 1 illustrates the axial T1-weighted MRI of a representative patient with AD and the proton spectra acquired from the posterior cingulate gyrus, a gray matter location within the limbic system, which best demonstrates the metabolic changes [17] associated with the disease. Striking metabolic abnormalities of decreased NAA/Cr and increased mI/Cr were reflected in both STEAM and PRESS acquisitions. MMSE (24±4 vs 30.0±0.0; Wilcoxon rank P<0.05) and 1H MRS confirmed the diagnosis of mild-moderate AD for the three patients enrolled in this pilot study (Table 1). NAA/Cr, [NAA], mI/NAA, [mI]/[NAA] and αβGlx reached statis-

34

Fig. 3B

tical significance and the appropriate trends were observed in mI/Cr and [mI] as well as the “atrophy index” [37] provided by quantifying the proportion of cerebral spinal fluid (CSF) in the VOI (voxel of interest), which was increased in AD (not shown). 13C MR spectra acquired with the dual-tuned surface coil included not only gray matter of the posterior cingulate gyrus, but also a significant proportion of white mat-

ter (Fig. 2). Natural-abundance 13C spectra acquired at 1.5 Tesla, when scaled to the mI signals, did not show marked differences between AD patients and controls (Fig. 3B, middle spectrum in each group). After 1-13C glucose infusion, α and β glucose resonances appeared with a similar time course in brain spectra of AD patients and controls (Fig. 4A). Time constants (TC) of glucose entry (TC Glcin: 13.1±0.4 vs 10.6±0.5), of glu-

35

Fig. 4 Glutamate (Glu2) (51–58 ppm), HCO3 (156–161 ppm), and lactate (16–22 ppm) regions of 13C difference spectra in AD patients and age-matched controls. Three regions of the 13C difference spectra of Fig. 3 are enlarged to illustrate: A The marked reduction in enrichment of Glu2 and aspartate (Asp2) in AD patients (above) compared to controls (below); B The reduced enrichment of HCO3 in AD patients (above) compared to controls (below); C In the region around 21 ppm (not shown in Fig. 3), there is readily visible lactate enrichment in AD patients (above) compared to controls (below)

cose disposal (TC Glcout: 67.5±22.8 vs 78.3±36.3) and time to peak cerebral glucose [t(Glcmax): 27.8±2.8 vs 25.3±4.5] did not reach statistical significance between AD patients and controls. The metabolic fate of glucose was markedly different between AD patients and controls; in the former, 13C brain spectra enriched with 1-13C glucose were strikingly abnormal (Fig. 3A). In spectra acquired after 60–100 min HCO3 and Glu2 enrichment was markedly less in AD patients than in controls. Summed naturalabundance,13C-enriched, and difference spectra permited ready identification of the principal metabolic differences observed between AD and control (Fig. 3B). Glutamate C4 enrichment was clearly observed after subtraction of overlying lipid resonances, but was not obviously different in AD patients compared to controls. The principal differences between AD patients and controls are shown in Fig. 4: reductions in Glu2, Gln2, and

Asp2 (Fig. 4A), a reduction in HCO3 (Fig. 4B) and an increase in 13C lactate (Fig. 4C). Time courses for some of the principal 13C metabolites are illustrated in Fig. 5. While the time course for Glu4 appeared to be essentially unchanged in AD patients, those for Glu2 and Asp3 were markedly slower. Table 2 summarizes results from 13C MRS studies. Statistically significant abnormalities of 13C glucose metabolism in AD patients included the prolonged time to appearance of bicarbonate (tHCO3) and glutamate (tGlu2) and reduced Glu2/Glu4 at t=100 min, all of which are consistent with reduced flux through the neuronal TCA cycle and a lower rate of glucose oxidation in AD (see Discussion). The ratio Gln2/Glu2 at t=100 min, a possible inverse measure of GNT (see Discussion and Fig. 7), was increased in AD patients compared to controls (P<0.02). Despite rather large reductions in AD patients compared to controls in the ratios Glu4 and Glu2/Glc C1, these did not achieve statistical significance in this small patient population. 13C NAA/13C Asp, a possible indicator of de novo NAA synthesis [52] and 13C Asp/Glx were unchanged. No significant difference could be detected between the natural-abundance 13C concentration of glutamate [Glu] or glutamine [Gln] (or their sum, [Glx]) between AD patients and controls (Table 2). When 13C metabolite alterations were compared with 1H MRS determinants, several of the 13C measures cor-

36

Fig. 5 Effect of AD on the time course of appearance of 13C glucose and its principal cerebral metabolites after 1-13C glucose administration. A 1-13C glucose, B 4-13C glutamate, C 2-13C glutamate, D 3-13C aspartate. Filled symbols: Control 2 (male, age 60 years), open symbols: Summed results of AD patients 1 and 2 (females, age 53 and 56 years). 13C spectra acquired during and after 1-13C glucose administration were processed as described in Materials and methods using automated peak fitting, difference spectroscopy, scaling to intrinsic [mI] to obtain µmol/kg brain of the enriched metabolite

related well with NAA/Cr (Fig. 6). Significant correlations were observed between NAA/Cr, the 1H MRS measure of neuronal number, [NAA], mI/NAA, or [mI]/ [NAA], and 13C measures of glucose oxidation, tHCO3, tGlu2, Glu3/Glu4, as well as with Gln2/Glu2 and other measures of GNT (Table 3).

Discussion The causes of cognitive impairment in AD are incompletely understood. Although current research strongly

supports the concept of defective cholinergic neurotransmission within the hippocampus and limbic system [11], by far the largest number of neurons in the normal brain employ glutamate as a neurotransmitter [31]. Thus, any interference with GNT may be expected to interrupt cognitive functions. It has long been known from PET studies that the cerebral metabolic rate for glucose is reduced on a regional basis in patients with AD. Recent calculations have indicated that ATP required for electrical activity of glutamatergic neurons, including the operation of the glutamate neurotransmitter cycle, could consume as much as 80% of that generated by complete oxidation of glucose through the TCA cycle [31]. Magestretti et al. offered theoretical and experimental evidence for the concept of 1:1 stoichiometry between GNT and glucose turnover in the neuronal-glial glutamine-glutamate cycle itself [30,53]. Even if this stoichiometry is not precisely one-to-one [54], it is tempting to suggest that any impairment of glucose oxidation could prejudice the supply of ATP to the glutamatergic neurons and result in cognitive disturbance. Conversely, a disease that reduced GNT would be re-

37

Fig. 6 Correlation between 13C measures of glutamate neurotransmission (GNT) and neuronal glucose oxidation and 1H MRS measure of NAA/Cr in AD. All data from six subjects (3 AD patients and 3 controls) examined by quantitative 1H and 13C MRS were evaluated. Plots of four 13C measures that were significantly different in AD are illustrated and compared with NAA/Cr, a measure of neuronal integrity, obtained in 1H MRS. For explanation of the significance of each of the 13C parameters, see Materials and methods. Excellent linear fits were obtained for all of the parameters displayed (for further statistical analysis see Table 3)

flected in a global and local reduction in the neuronal TCA cycle and the overall rate of glucose oxidation. Finally, simple loss of neurons would result in a reduction of the associated energy-dependent processes. Preliminary results of experiments designed to examine this hypothesis are described in this report.

The results of this pilot study using 1-13C glucose and in vivo 13C MRS are consistent with what is known of neuronal function in AD from PET. Glucose is the principal metabolic fuel of the neuron. In AD, when neurons are injured or lost, the overall metabolic rate for glucose is expected to fall. 13C MRS in the present study demonstrates this in much more exact detail, because H13CO3 (and 13CO2), which is produced from the complete oxidation of glucose in the third round of the neuronal TCA cycle (Fig. 7), was significantly slowed in AD patients compared to controls. Furthermore, the reduction in H13CO3 was in direct proportion to NAA/Cr, which is believed to be proportional to the number of intact neurons. The appearance of significant quantities of 13C in the glycolytic end-product lactate in AD patients but not in normal controls supports the view that the complete oxidation of glucose is limited. Although it is possible

6.5 4.6 8.3 6.5±1.9 0.30 10.3 9.2 10.2 9.9±0.6 0.44 1.33 0.63 0.54 0.83±0.43 0.16 0.42 0.41 0.66 0.50±0.14 0.40 Mean±SD t test (p value)

Controls

Mean±SD

1 2 3

17 21 23 20.3±3.1 0.04*

50 40 30 40.0±10.0 0.03*

0.75 0.89 0.628 0.76±0.13 0.04*

0.38 0.38 0.27 0.34±0.06 0.02*

0.50 0.29 0.19 0.32±0.16 0.43

1.43 0.64 0.51 0.86±0.50 0.20

4 2.88 6.82 4.6±2.0 7.57 7.2 11.16 8.6±2.2 0.11 0.35 0.32 0.26±0.13 0.41 0.58 0.59 0.53±0.10 Patients

1 2 3

30 29.5 24.5 28.0±3.0

90.5 105.5 70 88.7±17.8

0.649 0.436 0.466 0.53±0.11

0.57 0.51 0.48 0.52±0.05

0.65 0.44 0.47 0.31±0.01

0.61 0.24 0.16 0.34±0.24

[Gln] (mM) t(HCO3) (min)

Glu2/Glu4B

Gln2/Glu2 t=100 min

Asp/GlxB

Glucose oxidation or GNT measure

NAA/AspB

Glu4/Glc slope (min−1)

Glucose oxidation or GNT measure

Glu2/Glc slope (min−1)

[Glu] (mM)

Table 3 Correlation between measures of glucose oxidation or glutamate neurotransmission and cerebral NAA/Cr, [NAA] concentration, mI/NAA, and [mI]/[NAA] ratios

t(Glu2) minutes t(HCO3) minutes Glu2/Glu4 Gln2/Glu2 t=100 min Glu2/Glc slope (min−1) Glu4/Glc slope (min−1) [Glu] mM [Gln] mM

t(Glu2) (min)

Table 2 Impact of Alzheimer’s disease on 13C cerebral metabolites of glucose. Patients and controls received equivalent 1-13C glucose dosages and 13C brain spectra were acquired for 100–140 min. Methods of data analysis are described in the text

and correspond to these previously employed [44] unless otherwise noted. BProcessing method B was utilized to obtain these data; *significant difference as determined by Student’st tests

38

r values NAA/Cr

mI/NAA

[NAA]

[mI]/ [NAA]

−0.81 −0.72 0.75 −0.55 0.46 0.73 0.56 0.68

0.95 0.93 −0.66 0.94 −0.36 −0.93 −0.60 −0.62

−0.67 −0.55 0.85 −0.55 0.50 0.21 0.66 0.84

0.94 0.93 −0.55 0.98 0.03 −0.91 −0.73 −0.90

p values for r [NAA]

[mI]/ [NAA]

NAA/Cr t(Glu2) minutes t(HCO3) minutes Glu2/Glu4 Gln2/Glu2 t=100 min Glu2/Glc slope (min−1) Glu4/Glc slope (min−1) [Glu] mM [Gln] mM

– – – – – – – –

mI/NAA 0.01 0.01 – 0.01 – 0.01 – –

– – – – – – – –

0.05 0.05 – 0.01 – 0.01 – 0.01

that the labeling may have been due to an abnormal ratio of glia to neuronal cells [30], where lactate is thought to be produced [30], two additional 13C measures confirmed the inhibition of oxidation in AD: Glu2/Glu4, which is a simplified marker of the first complete turn of the neuronal TCA (Fig. 7), is significantly reduced. Time to appearance of Glu2 tGlu2, which is an additional measure of neuronal TCA cycle rate, was significantly increased and linearly correlated with 1H MRS measures of neuronal loss, [NAA], mI/NAA, and [mI]/[NAA]. We believe t(Glu2) and t(HCO3) to be reliable indications of the second and third rounds, respectively, of the TCA cycle in neurons and astrocytes and that, with the appropriate caveats [44], C2/C4 directly reflects TCA cycle rate in the human brain, as Lewandowski has demonstrated for intact heart muscle [55]. In an earlier 13C MRS study of chronic hepatic encephalopathy [44], we interpreted the prolonged time to appearance of 13C in Glu2 and H13CO3, and the relatively slower rate of Glu2/Glu4 as indicating reduced carbon flux through the TCA cycle. These aspects of the abnormal brain metabolism in AD that are consistent with PET results could only have been directly observed in vivo with 13C MRS. Is there a link with glutamate-glutamine cycle and GNT rate in AD? The present findings also implicate GNT more directly in the cognitive decline characteristic

39

Fig. 7 Fate of 1-13C atom of glucose through neuronal tricarboxylic acid cycle (TCA) and part of the neuronal-glial glutamate-glutamine cycle involved in GNT. Two rounds of the TCA cycle are represented. Carbon atoms enriched are shown in black. Carbons with 50% chance of enrichment are shaded. For clarity, the third round of neuronal TCA is represented only by 13CO2 release. In the text, 13CO2 is referred to as bicarbonate (H13CO3). Glc Glucose, Pyr pyruvate, acCoA acetyl co-enzyme A, Cit citrate, VTCA tricarboxylic acid cycle rate, αKG 2-oxoglutarate, Vx αKG Glu transaminase rate, VXA transaminase or malate aspartate shuttle rate, Glu glutamate, Glug glial glutamate, Gln glutamine, Glng glial glutamine, αKGg, glial 2-oxoglutarate, Suc succinylCoA, OAA oxaloacetate, Asp aspartate, NAA N-acetyl aspartate, VNAA_S NAA synthesis rate

of AD. Thus, the significant change in glutamine metabolism, reflected in a significantly higher Gln2/Glu2, must take place in the glial compartment (Fig. 7), where glutamine synthetase and the bulk of cerebral glutamine are located. If the glial TCA cycle is operating relatively faster than the neuronal TCA cycle, then we would expect an enrichment pattern with an increased glutamine signal versus glutamate signal as compared to controls. (Note that this does not imply that glial cells are increase or decreased, only that the distribution of neurons versus glia in a particular volume of the brain may be different).

These preliminary, in vivo observations have been combined to suggest a model of the brain in Alzheimer’s disease (Fig. 8): The glutamate-glutamine cycle responsible for neurotransmission in the intact human brain reflects activity specific to both neurons and astrocytes. Recent in vitro studies raising the possibility of involvement of glutamate neurotransmission in AD used animal models (knock-out mice, etc.), which may not fully reflect the human disease [56, 57] or cells isolated from the AD brain and other tissues [58]. While such studies are very informative, there is no way of knowing whether the isolated cells reflect in vivo neurochemistry. Thus the present study is, we believe, the first to directly determine the glutamate-glutamine cycle in intact humans with AD. [NAA], a metabolite found in neurons, dendrites, and axons [13], may reflect neuronal number or possibly some aspect of neuronal membrane integrity (e.g. permeability, mitochondrial electron transport). In the present study, the anticipated reduction in GNT was observed. The direct proportionality of GNT to NAA lends support to the first possibility (loss of neurons) and is therefore in accordance with the histopathological findings in AD patients at autopsy. Among other possible explanations for the parallel reduction in GNT, [NAA]

40

Fig. 8 Hypothetical scheme for interruption of GNT by neuronal loss in AD. The uptake and metabolism of glucose by intact glia is interposed between capillaries and the neuronal TCA cycle (originally proposed by Magistretti et al. [30]), but our hypothesis is not dependent upon this concept. In normal brain, two neurons and two glia are shown. Glucose carbons are transported to the neuron where they enter the TCA cycle, form Glu4 (round 1) and Glu2 (round 2), and are oxidized to CO2 (round 3). Glutamate returns carbon to the glia after neurotransmission (indicated as oo) for conversion to glutamine (Gln4 and Gln2). In AD, when one neuron is “lost” (NAA/Cr reduced) all neuronal processes (including neurotransmission) are systemically reduced to half (tHCO3, tGlu2 prolonged; Glu2/Glu4 reduced). Glucose in the “superfluous” glia enters the glial TCA cycle but enriches glutamine before glutamate because of local availability of glutamine synthase (Gln2/Glu4 increased in AD). Excess glucose in AD is metabolized only as far as lactate. Arrows: Steps verified experimentally in the brains of AD patients in vivo

and NAA/Cr in AD are metabolic disorders of mitochondria that equally impact the TCA and the synthesis and degradation of NAA [59]. An apparent reduction of NAA synthesis has been recently described in children with Canavan’s disease [52]. No such effect on NAA synthesis was observed (NAA/Asp was unchanged; see Table 2) over the 100 min of data acquisition in the present study. In reality, the poor signal to noise ratio makes it difficult to quantify this value directly. Using modifications of the technique described in this report, in particular devising an accelerated 13C glucose infusion and detection protocol that is well tolerated by the majority of Alzheimer’s sufferers, it will be possible to address this question directly in patients with various stages of AD.

While the exact stoichiometry of the two processes [54] and the details of the glutamine-glutamate cycle between neurons and glia remain to be clarified [60], the correlation between NAA/Cr, an in vivo measure of neuron number or activity, and these new measures of glucose flux through the TCA and glutamine-glutamate cycles suggest, for the first time, that in AD fundamental metabolic impairments contribute to cognitive impairment. For the moment, the best candidate is glutamate neurotransmitter rate. 13C MRS studies in larger patient populations are required to explore this hypothesis. Acknowledgements A.P. Lin is grateful for a MR Research Fellowship at HMRI. The authors also thank Drs. Keiko Kanamori and Stefan Bluml for their expert discussion.

41

References 1. Alzheimer’s Association: general statistics and demographics. http://www.alz.org/hc/overview/ stats.htm 2. Alzheimer A, Stelzmann RA, Schnitzlein HN and Murtagh F (1995) An English translation of Alzheimer’s 1907 paper, „Uber eine eigenartige Erkrankung der Hirnrinde“. Clin Anat 8:429–431 3. Morris RG (1991) The nature of memory impairment in Alzheimer-type dementia. In: Weinman J, Hunter J (eds) Memory: neurochemical and abnormal perspectives. Harwood, London, pp 56–59 4. Knopman DS, DeKosky ST, Cummings JL, Chui H, Corey-Bloom J, Relkin N, Small GW, Miller B, Stevens JC (2001) Practice parameter: diagnosis of dementia (an evidencebased review). Report of the Quality Standards Subcommittee of the American Academy of Neurology Neurology 56:1143–1153 5. de Jager CA, Milwain E, Budge M (2002) Early detection of isolated memory deficits in the elderly: the need for more sensitive neuropsychological tests. Psychol Med 32:483–491 6. Cummings JL, Cole G (2002) Alzheimer disease. JAMA 287:2335–2338 7. Braak H, Braak E (1991) Neuropathological staging of Alzheimer related changes. Acta Neuropathol 82:239–259 8. Gosche KM, Mortimer JA, Smith CD, Markesbery WR, Snowdon DA (2002) Hippocampal volume as an index of Alzheimer neuropathology: findings from the Nun Study. Neurology 58:1476–1482 9. Jack Jr. CR, Petersen RC, Xu YC (1999) Prediction of AD with MRIbased hippocampal volume in mild cognitive impairment. Neurology 52:1397–1403 10. Holmes C (2002) Genotype and phenotype in Alzheimer’s disease. Br J Psychiatry 180:131–134 11. Isacson O, Seo H, Lin L, Albeck D, Granhom AC (2002) Alzheimer’s disease and Down’s syndrome: Roles of APP, trophic factors and ACh. Trends Neurosci 25:79–84 12. Schuff N, Amend D, Ezekiel F, Steinman SK, Tanabe J, Norman D, Jagust W, Kramer JH, Mastrianni JA, Fein G, Weiner MW (1997) Changes of hippocampal N-acetyl aspartate and volume in Alzheimer’s disease. A proton MR spectroscopic imaging and MRI study. Neurology 49:1513–1521

13. Moffett JR, Namboodiri MA, Neale JH(1993) Enhanced carbodiimide fixation for immunohistochemistry: application to the comparative distributions of N-acetylaspartylglutamate and N-acetylaspartate immunoreactivities in rat brain. J Histochem Cytochem 41:559–570 14. Moats RA, Ernst T, Shonk TK, Ross BD (1994) Abnormal cerebral metabolite concentrations in patients with probable Alzheimer disease. Magn Reson Med 32:110–115 15. Kantarci K, Jack Jr. CR, Xu YC, Campeau NG, O’Brien PC, Smith GE, Ivnik RJ, Boeve BF, Kokmen E, Tangalos EG, Petersen RC (2000) Regional metabolic patterns in mild cognitive impairment and Alzheimer’s disease: A 1H MRS study. Neurology 55:210–217 16. Huang W, Alexander GE, Chang L, Shetty HU, Krasuski JS, Rapoport SI, Schapiro MB (2001) Brain metabolite concentration and dementia severity in Alzheimer’s disease: a (1)H MRS study. Neurology 57:626–632 17. Shonk TK, Moats RA, Gifford P, Michaelis T, Mandigo JC, Izumi J, Ross BD (1995) Probable Alzheimer disease: diagnosis with proton MR spectroscopy. Radiology 195:65–72 18. Lin A, Sheih D, Shic F, Chien TY, Ross BD (2002) Validation of MRS in early diagnosis of Alzheimer’s disease. J Geriatric Psychiatry Neurol p.176 19. Ross BD, Bluml S, Cowan R, Danielsen E, Farrow N, Gruetter R (1997) In vivo magnetic resonance spectroscopy of human brain: The biophysical basis of dementia. Biophys Chem 68:161–172 20. Hattori N, Abe K, Sakoda S, Sawada T 2002 Proton MR spectroscopic study at 3 Tesla on glutamate/glutamine in Alzheimer’s disease. Neuroreport 13:183–186 21. Antuono PG, Jones JL, Wang Y, Li SJ (2001) Decreased glutamate + glutamine in Alzheimer’s disease detected in vivo with (1)H-MRS at 0.5 T. Neurology 56:737–742 22. Reinikainen KJ, Paljarvi L, Huuskonen M, Soininen H, Laakso M and Riekkinen PJ (1988) A post-mortem study of noradrenergic, serotonergic and GABAergic neurons in Alzheimer’s disease. J Neurol Sci 84:101–116 23. Beal MF (1995) Aging, energy, and oxidative stress in neurodegenerative diseases. Ann Neurol 38:357–366 24. Smith CD, Pettigrew LC, Avison MJ, Kirsch JE, Tinkhtman AJ, Schmitt FA, Wermeling DP, Wekstein DR, Markesberry WR (1995) Frontal lobe phosphorus metabolism and neuropsychological function in aging and in Alzheimer’s disease. Ann Neurol 38:194–201

25. Bluml S, Danielsen ER, Ross BD (1996) Proton decoupled 31P MRS in Alzheimer disease. Proceedings, 4th International Society of Magnetic Resonance in Medicine, vol 1. New York, p. 304 26. Ross BD, Bluml S, Cowan R, Danielsen E, Farrow N, Gruetter R, Tan J (1998) In vivo MRS spectroscopy of human dementia. Neuroimaging Clinics of North America. 8:1–14 27. Silverman DH, Small GW, Chang CY, Lu CS, Kung De Aburto MA, Chen W, Czernin J, Rapoport SI, Pietrini P, Alexander GE, Schapiro MB, Jagust WJ, Hoffman JM, Welsh-Bohmer KA, Alavi A, Clark CM, Salmon E, de Leon MJ, Mielke R, Cummings JL, Kowell AP, Gambhir SS, Hoh CK, Phelps ME (2001) Positron emission tomography in evaluation of dementia: regional brain metabolism and long-term outcome. JAMA 286:2120–2127 28. Desgranges B, Baron JC, Lalevee C, Giffard B, Viader F, de La Sayette V, Eustache F (2002) The neural substrates of episodic memory impairment in Alzheimer’s disease as revealed by FDG-PET: relationship to degree of deterioration. Brain 125:1116–1124 29. Schulman RJ (2001) Functional imaging studies: Linking mind and basic neuroscience. Am J Psychiatry 158:11–20 30. Magistretti PJ, Pellerin L, Rothman DL, Shulman RG (1999) Energy on demand. Science 283:496–497 31. Attwell D, Laughlin SB (2001) An energy budget for signaling in the grey matter of the brain. J Cereb Blood Flow Metab 21:1135–1145 32. Beckmann N, Seelig J, Wick H (1990) Analysis of glycogen storage disease by in vivo 13C NMR: comparison of normal volunteers with a patient. Magn Reson Med 16:150–160 33. Mason GF, Rothman DL, Behar KL, Shulman RG (1992) NMR determination of the TCA cycle rate and α-ketoglutarate/glutamate exchange rate in rat brain. J Cereb Blood Flow Metab 12:434–447 34. Gruetter R, Seaquist ER, Kim S, Ugurbil K(1998) Localized in vivo 13C-NMR of glutamate metabolism in the human brain: Initial results at 4 Tesla. Dev Neurosci 20:380–388 35. Moreno A, Bluml S, Hwang J-H, Ross BD (2001) Alternative 1–13C glucose infusion protocols for clinical 13C MRS examinations of the brain. Magn Reson Med 46:36–48

42

36. Kreis R, Ernst T, Ross BD (1993) Absolute quantitation of water and metabolites in the human brain. Part II. Metabolite concentrations. J Magn Reson 102:9–19 37. Ernst T, Kreis R, Ross BD (1993) Absolute quantitation of water and metabolites in the human brain. I. Compartments and water. J Magn Reson 102:1–8 38. Bluml S, Hwang J-H, Moreno A, Ross BD (2000) Novel peak assignments of in vivo 13C MRS in human brain at 1.5 T. J Magn Reson 143:292–298 39. Adriany G, Gruetter R (1997) A halfvolume coil for efficient proton decoupling in humans at 4 Tesla. J Magn Reson 125:178–184 40. Bluml S (1999) In vivo quantitation of cerebral metabolites concentration using natural abundance 13C MRS at 1.5 T. J Magn Reson 136:219–225 41. Shaka AJ, Keeler J, Frenkiel T and Freeman R. An improved sequence for broadband decoupling: WALTZ-16. J Magn Reson 198352:335–338 42. Shaka AJ, Keeler J and Freeman R (1983) Evaluation of a new broadband decoupling sequence: WALTZ-16. J Magn Reson 53:313–340 43. Gruetter R, Adriany G, Merkle H and Anderson PM (1996) Broadband decoupled 1H-localized 13C MRS of the human brain at 4 Tesla. Magn Reson Med 36:659 44. Bluml S, Moreno-Torres A, Ross BD (2001) [1–13C] glucose MRS in chronic hepatic encephalopathy in man. Magn Reson Med 45:981–993

45. Shic F, Lin A, Ross BD (2002) Automated methods for analysis of large quantitites of13C MRS data in clinical studies. J Magn Reson (in press) 46. Mason GF, Behar KL, Rothman DL, Shulman RG (1992) NMR determination of intracerebral glucose concentration and transport kinetics in rat brain. J Cereb Blood Flow Metab 12:448–455 47. Mason GF, Gruetter R, Rothman DL, Behar KL, Shulman RG, Novotny EJ (1995) Simultaneous determination of the rates of the TCA cycle, glucose utilization, a-ketoglutarate/glutamate exchange, and glutamine synthesis in human brain by NMR. J Cereb Blood Flow Metab 15:12–25 48. Bluml S, Moreno A, Hwang J-H, Ross BD (2001) 1–13C glucose magnetic resonance spectroscopy of pediatric and adult brain disorders. NMR Biomed 14:19–32 49. Bluml S, Moreno-Torres A, Shic F, Nguy C-H, Ross BD 2002 Tricarboxylic acid cycle of glia in the in vivo human brain. NMR Biomed 15:1–5 50. Lebon V, Peterson K, Cline GW, et. al (2002) Astroglial contribution to brain energy metabolism in humans revealed by13C nuclear magnetic resonance spectroscopy: elucidation of the dominant pathway for neurotransmitter glutamate repletion and measurement of astrocytic oxidative metabolism. J Neurosci 22:1523–1531 51. Lentner C (1982) In: Lentner C (ed) Introduction to statistics. Ciba-Geigy, Basel, p 49 52. Moreno A, Ross BD, Bluml S (2001) Direct determination of the N-acetyl-Laspartate synthesis rate in the human 13C MRS and [1–13C] glucose infusion. J Neurochem 77:347–350

53. Sibson NR, Dhankhar A, Mason GF, Rothman DL, Behar KL, Shulman RG (1998) Stoichiometric coupling between brain glucose metabolism and glutamatergic neuronal activity. Proc Natl Acad Sci USA 95:316–321 54. Gruetter R, Seaquist ER, Ugurbil K (2001) A mathematical model of compartmentalized neurotransmitter metabolism in the human brain. Am J Physiol Endocrinol Metab 281:E100–E112 55. Lewandowski ED, Hulbert C (1991) Dynamic changes in 13C NMR spectra of intact hearts under conditions of varied metabolite enrichment. Magn Reson Med 19:186–190 56. Janus C, Westaway D (2001) Transgenic mouse models of Alzheimer’s disease. Physiol Behav 73:873–886 57. Anger WK (1991) Animal test systems to study behavioral dysfunctions of neurodegenerative disorders. Neurotoxicology 12:403–413 58. Robinson SR (2001) Changes in the cellular distribution of glutamine synthetase in Alzheimer’s disease. J Neurosci Res 66:972–980 59. Baslow M (2000) Functions of N-acetyl-l-aspartate and N-acetyl-l-aspartylglutamate in the vertebrate brain: role in glial cell-specific signaling. J Neurochem 75:453–459 60. Kanamori K, Ross BD (2001) The first in vivo observation of 13C-15 N coupling in mammalian brain. J Magn Reson 153:193–202