Mini-Mental State Examination and proton spectroscopy of the posterior cingulate in Alzheimer disease

To compare metabolite ratios in the posterior cingulate with the Mini-Mental State Examination (MMSE) test scores in patients with mild or moderate Alzheimer disease and in controls. Methods We evaluated 29 patients with mild or moderate Alzheimer disease and 15 controls by proton spectroscopy with the voxel located in the posterior cingulate. The MMSE was applied to all patients and controls. The metabolic ratios: N-acetyl-aspartate/creatine (Naa/Cr), mio-inositol/creatine (mI/Cr) and mio-inositol/N-acetyl-aspartate (mI/Naa) were obtained and then post-processed using the MRUI software (magnetic resonance user interface). Results Correlation between Naa/Cr and mI/Naa ratios in the posterior cingulate with the MMSE was observed, and a positive correlation with Naa/Cr and negative correlation with mI/Naa were seen. The mI/r ratio presented no correlation with MMSE scores. Conclusion The positive correlation with Naa/Cr, and negative correlation with mI/Naa may corroborate that neuronal density/viability is associated to a higher MMSE score.

Alzheimer disease (AD) is the most common cause of dementia with a relative frequency which increases with age. Conventional magnetic resonance imaging (MRI) may not detect abnormalities until late in the course of the disease. 1 Newer techniques such as proton magnetic resonance spectroscopy (1H-MRS), which allows noninvasive assess-ment of some metabolites in vivo, can be used to detect abnormalities earlier in the disease.
In Alzheimer disease (AD) N-acetil aspartate concentration or N-acetyl-aspartate/creatine (Naa/Cr) ratio are decreased while mio-inositol concentration, mio-inositol/ creatine (mI/Cr) and mio-inositol/N-acetyl-aspartate/creatine (mi/Naa) ratios are increased. 2 Both the hippocampus and posterior cingulate are the limbic regions primarily affected in AD.
Cognitive tests are commonly employed to assess such patients, of which the MMSE is the most widely used.
The aim of the present study was to assess correlation between metabolic ratios obtained by proton spectroscopy in the posterior cingulate, with MMSE scores in patients with mild to moderate AD and in cognitively normal controls.

Methods
This was a prospective study based on proton spectroscopy analysis carried out in the Magnetic Resonance section of Hospital This project was approved by the Ethics committee of the Hospital das Clinicas of the University of São Paulo School of Medicine, where patients or their legal guardians signed the free informed consent term after agreeing to participate in the study.
The inclusion criteria of the study were: patients with diagnosis of probable AD according to the NINCDS-ADRDA criteria, and mild to moderate dementia according to the DSM-III R criteria, signing of the informed consent term by the patient or their legally responsible guardian, and a collaborative patient.
All patients underwent tests for levels of Vitamin B12, thyroid hormones, serology for syphilis, hemogram, levels of urea and creatinine, hepatic enzymes, total fraction proteins, and magnetic resonance imaging (MRI) examination to rule out other causes of cognitive defi cits.
Exclusion criteria applied were: patients with psychiatric or neurologic diseases, history of cranial trauma, use of psychotropic medication (except drugs for treatment of AD), diabetes mellitus, evidence of focal or diffuse brain lesions such as tumors, hydrocephalus or cerebral infarcts on MRI. The presence of sparse focus of high signal in the white matter of cerebral hemispheres on T2-weighted sequences did not constitute exclusion criteria given that these are commonly observed in elderly patients (class 1 and 2 on the Fazekas and Schmidt scale); 3,4 spectroscopies were excluded if after homogenization of the magnetic fi eld, the value of the water peak width in frequency units (FWHM -frequency width at half maximum) exceeded 7 Hz.
Based on these criteria, 16 patients were excluded (11 with diabetes mellitus, four due to cerebral infarcts and one due to diabetes mellitus and meningioma), giving a fi nal study sample of 29 patients. Age ranged from 56 to 87 years (mean 74.2±7.6 years and median of 75 years). Of the 29 subjects included, 17 were female (59%) and 12 male (41%). Moreover, 23 patients had mild AD while 6 had moderate AD.
A control group of 15 volunteers was used, constituting individuals without cognitive defi cits, these patients were from the general population. Control group age ranged from 66 to 79 years (mean of 72.5±3.3 years). Nine volunteers were female (60%) and 6 were male (40%).
All volunteers presented normal values on the MMSE, 5 as well as on the delayed recall of 10 fi gures 6 and on the category fl uency test-animals/min, 7 where values were adjusted for schooling when appropriate.
The MRI studies were performed on a 1.5 Tesla Unit (Horizon LX 8.3, GE Medical Systems Milwaukee, WI, USA) for all patients and controls using a brain quadrature coil. The MRI and 1 H-MRS examinations took approximately 60 minutes.
The MRI protocol included: spin echo sagittal T1weighted, fast spin echo axial T2-weighted-images, axial FLAIR images (fl uid attenuated inversion recovery), axial diffusion-weighted images, axial SPGR -spoiled gradient recalled acquisition in steady state, and localizer axial T2weighted images for planning the spectroscopy.
The 1 H-MRS protocol included a single voxel acquisition using the PRESS (point resolved spectroscopy) technique with TR=1500 ms, TE=135 ms, field of view (FOV)=24 cm, 8 NEX (number of excitations), and 96 and 128 excitations, lasting approximately 3 minutes. The voxel was located in the posterior cingulate. Preceding the spectroscopy acquisition, automatic adjusting transmission-reception, water suppression and fi eld homogeneity optimization were performed for the selected voxel.
The voxel measured 2x2x2 cm (8 cm 3 ) and was located in the posterior cingulate in the median image of the saggital plane. It was positioned below the cingulate and above the parietoccipital sulcus, including the posterior cingulate as well as the inferior pre-cuneus, a location previously described in the literature [8][9][10][11][12] (Figures 1 and 2).
We evaluated the N-acetyl-aspartate/creatine (NAA/ CR), Mio-inosytol/creatine (mI/CR) and Mio-inosytol/ N-acetyl-aspartate (mI/NAA)/ratios. Post-processing was carried out using the MRUI software (Magnetic Resonance User Interface), Java version, from Advanced signal processing for medical resonance imaging and spectroscopy, TMR, FMRX-CT97-0160. The NAA/CR, mI/CR, and mI/NAA ratios obtained in the posterior cingulate for patients and controls were compared.
Two illiterate patients (12 and 13), together with a mild AD patient (patient 14), presenting with a marked compromise in language and a very low MMSE score (MMSE=6), disproportionate to the other cognitive functions which remained relatively preserved, were excluded from this analy-sis. This lack of literacy and marked language impairment hampered the MMSE analysis and scoring which could have in turn introduced a bias in the association of the relationship between metabolites and the MMSE.
The Chi-square test was used to compare the gender distribution of the two groups (nominal data). The Student-t test was employed to compare the metabolite ratios in patient and control groups. Correlation between MMSE and metabolic ratios were calculated using Pearson's linear coefficient (13). The descriptive level considered signifi cant was 5% (<0.05).

Results
The MMSE data was distributed as depicted in Table  1, which shows a lower MMSE and greater variance in the patient group.
Comparative analysis between the MMSE and metabolic ratios was carried out by calculating the indexes of Pearson's linear correlation for each metabolite, considering the entire sample and also patient and control groups separately. Separate group analysis revealed no signifi cant association. However, analysis of the pooled data (patients and controls) revealed a signifi cant correlation for some measures, being positive for the Naa/Cr ratio in the cingulate (0.61), and negative for the mI/Naa ratio (-0.55). These  values indicated some degree of association, but evidenced large spread between the MMSE and the relationships among the metabolites. This data can be seen in Table 2.
The level of association is clearly illustrated in Figures 3 to 5, where each individual is represented by a point, and a line of correlation is presented for the global data.

Discussion
A signifi cant association was observed between Naa/ Cr and ml/Naa ratios in the posterior cingulate and the MMSE. The Naa/Cr ratio correlated positively while the ml/Naa ratio showed negative correlation. Naa and Naa/ Cr ratio values have correlated positively with the MMSE in earlier studies in the occipital and parietal gray matter, 14 parietoccipital gray matter 15 occipital gray matter, 16 , temporal white matter 17 and para-hippocampal gyrus (in a post-mortem study). 18 There is also evidence of an association between decreased Naa concentration and cognitive decline. 19,20 Since NAA is considered a marker for neuronal density/viability, these associations indicate that the NAA/ Cr ratio obtained by proton spectroscopy could be considered an indicator of cognitive decline in such patients.
The Naa/ml ratio also correlated positively with the MMSE. 15 However, confl icting results have been found for this association regarding separate ml or ml/Cr. Several studies have demonstrated the presence of negative cor-     relation for the ml/Cr ratio in the posterior cingulate, 21 or for ml concentration in frontal white matter. 22 Nevertheless, other authors 14,15 observed no signifi cant correlation of the ml/Cr ratio or the ml concentration, with the MMSE. It is possible that the correlation of the Naa/ml or ml/Naa is due to the Naa component. Indeed, the present study showed a more signifi cant correlation for the Naa/Cr ratio than the ml/Naa ratio. In view of the fact that the increase in ml or the ml/Cr ratio found in this disease is early and associated with the accumulation of neurofi brillar tangles and astrocytic and glial proliferation, the absence of correlation between ml/Cr ratios and the MMSE test remains unclear and warrants future investigations involving a larger patient series. Concluding, the analysis of proton spectroscopy studies on a single voxel within the posterior cingulate carried out in mild or moderate AD patients and controls evidenced positive correlation of the Naa/Cr ratio with the MMSE yet negative correlation of the ml/Naa ratio. These fi ndings corroborate a correlation between neuronal density/viability and the MMSE test.