The in vivo neuronal contribution to human cerebral metabolic rate of glucose (CMRglc), measured by 18FDG-PET, is unknown. Since NAA is thought to reflect neuron density, evaluating how CMRglc varies as a function of NAA concentration ([NAA]) should reflect the way in which brain glucose metabolism is affected by neuron density and/or by NAA content per neuron. The CMRglc-to-[NAA] relationship could be derived for an individual subject by plotting local CMRglc against local [NAA] across that subject's brain. The CMGglc-to-[NAA] relationship might be expected to vary from subject-to-subject depending on factors such as subject cognitive status. However, there are limitations to this approach. First, while cortical gray matter is the tissue of primary interest in evaluating brain metabolic activity, PET and 1H MRSI data are often expressed in terms of whole, unsegmented brain tissue, rather than as values for cortical gray matter alone. Second, even if regional data are compared, the spatial resolution of MRSI is lower than that of PET, a source of possible signal infidelity. In a study that included 19 demented, cognitively impaired, and control subjects, who had whole-brain PET data, MRI, and MRSI we aimed to establish a method for the assessment of CMRglc and [NAA] in cortical gray matter, accounting for differences between PET and 1H MRSI image resolution. Furthermore, we looked for the quantitative relationship between gray matter CMRglc and [NAA] in individual cognitively normal, cognitively impaired, and demented subjects and explored whether this CMRglc-to-[NAA] relation varies with cognitive status across subjects. In 18 of 19 subjects, a significant linear regression (P < 0.05) resulted when gray matter PET was plotted against gray matter NAA, whereby gray matter PET was higher for higher gray matter NAA. A representative example of this PET-NAA correlation is shown in figure 4.
To the extent that [NAA] can be taken as a marker of neurons (5-6), the correlation with FDG-PET suggests that the metabolic activity measured by 18FDG-PET in a sample of gray matter increases with the density of neurons present in that gray matter sample and/or with the quantity of NAA within those neurons. Furthermore, the slope of the GMCMRglc-against-GMNAA regression decreased with increasing CDR across subjects, suggesting that CMRglc per neuron is lower in cognitively impaired and demented subjects than in cognitively normal subjects. This explanation is consistent with the idea that diminished cortical metabolism is a physiologic substrate of dementia, regardless of etiology (7, for a review see 18). It also suggests that such hypometabolism may be due not simply to losses in neuron numbers in gray matter, but to an alternative or concomitant decrease in the metabolic activity per neuron of those neurons remaining. This method may be used to investigate the relationship of CMRglc to neurons in various conditions.
CSF-corrected GMNAA, IU
Figure 4. Correlation between FDG-PET (glucose metabolism) and NAA levels.
CSF-corrected GMNAA, IU
Figure 4. Correlation between FDG-PET (glucose metabolism) and NAA levels.
5. DEEP GRAY MATTER STRUCTURES IN HIV: A 1H MRSI STUDY (26)
The goal was to determine the concentrations of the neuronal marker N-acetylaspartate (NAA) and of choline-containing metabolites (Cho) in the subcortical brain of HIV-seropositive patients as a function of their cognitive impairment and clinical symptoms. Pathological studies suggest that subcortical gray matter carries a heavy HIV load, and neuropsychological test results are consistent with involvement of subcortical and fronto-striatal brain systems in HIV disease. Single-volume 1H MRS studies suggested neuronal preservation (i.e., unchanged NAA) and macrophage infiltration (i.e., high Cho) in subcortical brain of cognitively impaired and clinically symptomatic HIV+ individuals. Improved 1H MRS methods may allow the early detection of metabolite alterations in subcortical brain of asymptomatic HIV+ individuals.
Two-dimensional 1H MR spectroscopic imaging with volume preselection was performed in 30 HIV- controls and 70 HIV+ participants with varying severities of systemic disease and neuropsychological impairments.
Subcortical NAA was about 20% lower than control only in HIV+ patients with severe cognitive impairments; asymptomatic patients or those with mild cognitive impairments had normal subcortical NAA. Subcortical Cho was about 11% higher compared to controls in HIV+ patients regardless of the presence or absence of cognitive impairment or clinical symptoms. Subcortical NAA correlated with performance on a variety of neuropsychological tests but not with Center for Disease Control clinical stage.
High thalamic Cho was associated with low CD4 lymphocyte counts. The NAA findings suggest functionally significant neuronal subcortical injury only in severely cognitively impaired HIV+ patients. High subcortical Cho throughout all stages of HIV disease is consistent with early and persistent macrophage infiltration. The findings are consistent with the lack of significant subcortical neuron loss in neuropathological studies. Quantitative 1H MRSI may play a role in the objective assessment of the presence, magnitude and progression of brain involvement in HIV infection.
6. BRAIN DAMAGE IN TREATED HIV-INFECTED INDIVIDUALS (27)
This was the first clinical study of the effects of HIV infection and its progression on brain metabolites using short-TE multi-slice 1H MRSI. Figure 5 shows a representative slice of a multislice 1H MRSI experiment at TE=25ms from a control. Metabolite images of mI, Cho, Cr, and NAA were generated using the automatic fitting program developed in this lab. The raw (solid line) and fitted (dashed) 1H MR spectrum was selected from a region in white matter. We co-registered MRSI to segmented MRI data, determined atrophy-corrected absolute metabolite concentrations in major brain regions, and analyzed by region and tissue type, using linear regression in a mixed effects model. All statistical analyses used a 2x2 ANOVA (with age as a covariate when appropriate), yielding main effects of HIV infection and heavy drinking, HIV-by-alcohol interactions and group contrasts. Statistical analyses were also done for HIV+ individuals on or off highly active antiretroviral treatment (HAART) and for CDC stages, to evaluate the contrast in HIV symptomatology. We detected effects of heavy drinking [see below or Meyerhoff et al. ACER 2004] and HIV infection, the latter reported here.
Main HIV effects in patients on HAART were observed for 7% higher mI in thalami, 6% higher Cr in temporal WM and trends to lower parietal GM NAA (-5%) and Cho. Additional data analyses including the 30% of subjects who were not on HAART, suggested that stable HAART appears to ameliorate metabolite abnormalities in HIV samples.
When 21 neurologically asymptomatic HIV+ participants in CDC A were compared to 35 symptomatic HIV individuals in CDC B and C, we found main effects of CDC status: symptomatic patients had higher mI in parietal and temporal WM, frontal WM, higher Cho in parietal WM and trends to higher frontal GM Cho. NAA was largely normal in these patients. Thus, the primary impact of HIV symptomatology was on WM, suggesting that inflammatory changes and perhaps myelin damage are the primary pathological events in mostly treated HIV+ samples, whereas neuronal/axonal deterioration is largely absent.
Nevertheless, lower NAA in left parietal WM and in frontal WM NAA were associated with higher viral load. CD4 did not correlate with NAA measures in this heavily treated group. Thus, these correlations support our hypotheses that lower NAA levels are associated with greater viral load.
In conclusion, the magnitude and anatomical distribution of the cross-sectional HIV effects differ from previous reports (see our previous studies above), which showed dramatic metabolic abnormalities in WM, GM and subcortical brain in pre-HAART era patients with cognitive impairments of greater severity than today. In this heavily treated population, some NAA loss was only observed in parietal GM, while WM Cho and mI increases are still significant but are not as ubiquitous a finding as in previous HIV-infected samples.
7. 1H MRSI REVEALS BRAIN METABOLITE INJURY IN TREATED HIV+ PATIENTS AND IN CHRONIC HEAVY DRINKERS (28)
The above study showed that cross-sectional HIV effects were relatively subtle and it is unclear if they are premorbid or a function of HIV infection. Therefore, we studied HIV+ individuals about 2 years apart by repeated 1H MRSI to test the hypothesis of ongoing brain damage in HIV patients, who are not treated or have high viral loads despite antiretroviral therapy (HAART). Twenty-four HIV+ patients on HAART showed lower rates of frontal GM and parietal WM NAA loss than 6 HIV+ off HAART. However, even the 13 virally suppressed HIV+ (i.e., those on "successful" HAART) had NAA decreases in temporal WM at a rate of 4±6% compared to 9 viremic patients on HAART at 0±3% (p=0.09). Temporal WM Cho and lenticular Cho further decreased over time in virally suppressed patients on HAART.
Using 2x2 ANOVA in HIV+ participants on HAART (viral status x alcohol status), we found greater rates of NAA loss in suppressed patients than viremic patients in frontal GM and temporal WM, and greater increases of Cho in viremic than suppressed HIV+ patients in frontal and parietal GM, temporal WM, and thalami. This suggests that in virally suppressed subjects on HAART, neuronal damage continues while inflammatory processes are arrested.
In summary, untreated HIV infection presents with ongoing neuronal injury, in particular in parietal WM. HIV+ patients on HAART show less severe longitudinal brain metabolite damage than those off HAART, but even suppressed HIV+ patients tend to have ongoing regional NAA loss. These results demonstrate the ability of 1H MRSI to detect longitudinal metabolite changes due to HIV infection.
8. *H MRSI SEPARATES NEURONAL FROM GLIAL CHANGES IN ALCOHOL-RELATED BRAIN ATROPHY (29)
Eleven elderly alcoholics (61 ± 7 years), abstinent for approx. 9 months at the time of MRSI study, were compared to 9 age-matched light-drinking controls. 1H MRSI with PRESS volume preselection was used in a single slice above the ventricles. Cortical spectra were extracted from frontal and parietal gray matter regions, corrected for possible atrophy using tissue-segmented MR images and cortical metabolite ratios were compared. We found a location-by-group effect, which indicated significantly lower NAA/Cr in frontal than parietal cortex of recovering elderly alcoholics. Alcoholics with the most ventricular CSF showed the lowest NAA/Cr in frontal cortex. Together with findings of no significant tissue atrophy in these elderly alcoholics, the MRS results suggest neuronal damage or loss in frontal cortex associated with glial hyperplasia (or gliosis). This report demonstrates that: 1) that absolute metabolite measures are needed, 2) that by studying recovering alcoholics several months after cessation of drinking, tissue recovery may mask the full impact of chronic alcohol consumption on neuronal injury, and 3) that longitudinal studies in recovering alcoholics are needed to better understand the reasons for deficits in NAA and its potential recovery: Neuronal loss would not be associated with NAA recovery, whereas loss of dendritic arborization and neuronal cell body shrinkage would be reversible. Thus, this study pointed the way to further studies being performed in this lab.
9. SEPARATE AND INTERACTIVE EFFECTS OF COCAINE AND ALCOHOL (30)
In a 1H MRSI study of the effects of concurrent cocaine and alcohol dependence on the brain, we compared non-dependent controls to subjects dependent on crack cocaine alone and to subjects dependent on both cocaine and alcohol. Our findings included lower NAA concentrations in cocaine-dependent and in cocaine and alcohol-dependent subjects, especially in dorsolateral prefrontal gray matter and in posterior parietal white matter, suggesting damage to neurons and axons in these brain regions of substance abusers. In that study, it was not possible to determine whether effects observed in subjects dependent on both cocaine and alcohol were due to cocaine abuse, to alcohol abuse, or to both. To answer that question, we examined in a follow-up study a fourth group of individuals dependent on alcohol alone and augmented the original sample by additional subjects, allowing to address the question of separate and interactive effects of chronic cocaine dependence and alcohol dependence on regional brain metabolites. We found main effects of alcohol dependence on brain atrophy and NAA concentrations. Alcohol-dependent individuals abstinent for 1-2 years had brain atrophy and about 8% lower NAA concentrations in both cortical and subcortical gray matter, but no NAA loss in white matter. NAA loss was most significant in frontal gray matter regions. There was no significant main effect of cocaine dependence on NAA concentrations, despite some regional atrophy in these cocaine-dependent individuals abstinent from cocaine for about 4 months. Cocaine-dependent individuals showed higher posterior parietal white-matter creatine concentrations than controls. In addition, no significant cocaine dependence-by-alcohol dependence interactions were found for any metabolite in any tissue or brain region. Thus, alcohol dependence, but not cocaine dependence, was associated with long-lasting NAA loss in cortical and subcortical gray matter throughout the brain, suggesting widespread neuronal injury. However, past chronic alcohol use did not aggravate any chronic cocaine-induced metabolite deficits.
Was this article helpful?