Several studies from different laboratories have now reported that greater NAA is associated with better performance on some type of cognitive test, particularly broad measures of cognitive functioning such as those found in measures of intellectual functioning . The details vary, as different brains regions have been studied with diverse tests, but the general finding seems fairly well established: in vivo measures of NAA are positively correlated with cognitive performance in normal individuals. In 1999 we reported that NAA concentrations from a left parieto-occipital white matter voxel predicted overall intellectual functioning in a young cohort of college aged subjects (Mean age = 22 +/-4.6) . We have now conducted two independent replications, with different participants and slightly different imaging acquisition and processing methods, and found similar results [27, 28]. Across the three studies, NAA accounted for 27%, 36%, and 25% of variance in IQ scores. Among the diverse neurobiologic predictors of IQ (e.g., brain volume, evoked potentials), NAA seems to be one of the strongest biological correlate of intellectual functioning . Importantly, this relationship is not observed when NAA is assessed in left frontal white matter. Across the same three data sets, frontal NAA accounted for 1%, 6%, and 6% of variance in IQ scores (each nonsignificant), and in the latter two data sets an inverse relationship was noted [unpublished data].
A core component skill underlying individual variation in intelligence is speed of cognitive processing . As our NAA-intelligence findings emerged from voxels containing mostly myelinated axonal fibers, we hypothesized that NAA might be most related to neuropsychological measures tapping speeded cognitive performance . Our battery of neuropsychological tests was split into those emphasizing rapid processing (i.e., timed performance measures) vs. those that did not (e.g., word finding ability). Left posterior white matter NAA accounted for 42% of the variance on a composite measure of speeded performance, vs. 8% (non-significant) on non-speeded tests. Interestingly, as also found in our TBI studies, correlations between NAA and motor tests were low (Grooved Pegboard Test, r = 0.18; Grip Strength, r = -0.05). Finally, one recent study reports that left frontal gray matter NAA predicts greater verbal intelligence, though only in women . Thus, studies to date suggest that there may be both cognitive, anatomic, and sex specificity in the white matter NAA-cognition relationship. Obviously, we need to survey many more brain regions, especially gray matter voxels, in larger samples of individuals equally divided among males and females to confirm the hypothesis that regional NAA differences underlie cognitive ability in normal human brain.
The mechanism(s) underlying the WM NAA-cognition relationship are of great interest. It may be, however, that rather "global" cognitive measures such as described here (i.e., IQ, total z-score) are the ones that best map on to NAA, which has been linked to myriad metabolic functions [3-6]. As noted above, the correlational nature of human studies limits implications for understanding causal relationships. Human studies can, however, indicate possible covariance of NAA with other variables known to predict general intellectual functioning. The best known of these is overall brain volume, which correlates with IQ at about r = 0.3 or 0.4 . In our sample (N = 28) we find this same relationship; total brain volume correlated with IQ at r = 0.49 (p < .01) . When both NAA and brain volumetric values are regressed against FSIQ, only occipito-parietal NAA is retained. Thus, NAA levels appear to characterize important variance independent of sheer brain volume, and likely begin to capture subtle brain design variables underlying intellectual attainment: indeed, this point becomes manifest merely by noting that at least two brain designs (e.g., male, female) attain similar IQ's independent of brain size.
Recent fMRI  and voxel-based morphometry (VBM) studies  clearly show that some brain regions seem to be much more important for individual variation in intellectual functioning than others. In particular, frontal gray matter activations and volumes appear to be important to intellectual functioning  although both frontal and parietal regions tend to covary with IQ . As we have recently demonstrated specific brain regions associated with intellectual functioning in a cohort of normal subjects  using Voxel Based Morphometry (VBM), we sought to determine the relationship between NAA levels within frontal and posterior white matter and gray matter morphometry in a cohort of normal subjects . VBM is based on making regionally-specific (voxel-wise) inferences on the local relative concentrations of different tissue types after spatial normalization and segmentation of the underlying anatomical images . The results of the procedure are presented in standard stereotactic (i.e., Talairach)
Figure 4 (right panel). Voxel Based Morphometry (VBM) analysis of GM cluster volumes covarying with left hemisphere frontal white matter NAA concentration: (See color insert that appears between pages 364 and 365)
space that facilitates direct comparisons across studies. VBM has been extensively cross-validated with both ROI and functional analyses [e.g., 39].
Eighteen normal subjects (age range: 18 - 37), free of neurological or psychiatric disorders, were included in the analysis. MRIs were obtained with a clinical 1.5-T scanner, head coil, and software (Signa 5.4; General Electric Medical Systems, Waukesha, WI). Metabolite concentrations were determined using time-domain fitting (LC Model) to measure peak areas of NAA . Significant positive correlations between occipito-parietal NAA and gray matter were found in predominantly frontal lobe regions including the bilateral middle and superior frontal gyrus (BAs 9, 10), left middle frontal gyrus (BA 46), and the anterior cingulate gyrus (BAs 24, 32). Positive correlations between frontal NAA and gray matter were limited to the posterior right cingulate gyrus (BA 31), and right parietal lobule (BA 7). This last result is particularly compelling, as the posterior cingulate cortex metabolic integrity has been linked to early detection of dementia in both AD and VCI , and suggests a critical interplay between frontal white matter metabolic integrity and posterior gray matter morphology. These results (Figure 3) demonstrate a compelling double dissociation between regional metabolic integrity and cortical volume, with occipito-parietal NAA related predominantly to frontal lobe gray matter volume, and frontal NAA related to predominantly posterior gray matter volumes. Moreover, the frontal regions in which posterior NAA predicts gray matter volume are well associated with cognitive activation elicited via functional magnetic resonance imaging and positron emission tomography . Although preliminary in nature, this analysis highlights the importance of intact metabolic connectivity underlying frontal lobe morphology, the first time that this has been demonstrated in a cohort of normal subjects.
The studies described above have focused on relationships between white matter NAA and rather global measures of cognitive ability. 1H-MRS may also prove quite useful in the investigation of functional differences in discreet gray matter nuclei. We have recently reported that NAA/Cre in the hippocampus predicts individual variation in hippocampal function . The Morris Water Maze has been extensively employed in studying the relationship between hippocampal function and spatial learning in rodents, and more recently. Hippocampal damage reliably produces deficits on tasks such as this that require learning configural as opposed to elemental stimuli configurations. Use of the Virtual Water Maze Test (VWMT) in humans provides a solid basis for cross-species comparisons . Both histological and volumetric studies of the hippocampus reveal age-related changes  that could potentially contribute to the well documented decline in memory. In this study we compared 16 young adults (mean age = 26.1 years) with a sample of 16 very healthy elderly individuals (mean age = 11.6 years), all of whom were homozygous APOE-3 genotypes, as the APOE-4 allele has been linked with risk of Alzheimers Disease. Magnetic resonance spectroscopic imaging data were obtained at the level of the hippocampus from a 15 mm thick axial slab, as shown in Figure 5. Concentraion was expressed as a ratio of NAA/Cre. As expected, age-related deficits were observed in VWMT measures reflecting configural memory, but not in VWMT measures reflecting visual-motor processing (and motivation). VWMT memory declines were accompanied by decreased hippocampal volumes and decreased NAA/Cre, though Cho/Cre showed no age effect. In a regression analysis of VWMT performance, NAA/Cre and age were retained as significant predictors, while hippocampal volume did not provide an independent contribution. These results suggest that normal aging is associated with hippocampal structural and biochemical changes, and that these changes
may constitute an important component of age-related deficits in hippocampus-dependent learning and memory.
Was this article helpful?