Magnetic resonance spectroscopy in AD

Phosphorous MR spectroscopy.

Early work in this field involved the study of postmortem AD brain tissue. Pettegrew et al.25-27⇓⇓ reported increases in PME and PDE, with one study finding a strong association between PDE and senile plaque counts.27 A recent postmortem study28 at high field strength reported that this abnormality may be greater in AD brains positive for the APOE-ε4 allele.

Clinical MRS studies have produced mixed results, with three studies29-31⇓⇓ suggesting no differences in PME and PDE levels between patients with AD and control subjects. Results in reports from Pettegrew’s laboratory, on the other hand, suggest a complex relationship in vivo.32-34⇓⇓ PME concentration has been found to increase in the mild stage of AD and then to fall to normal levels as the disease advances. The phosphorous metabolism marker phosphocreatine has been found to follow the converse pattern: it is depleted in mild AD and renormalizes in the later stages. On the other hand, neural membrane breakdown products, reflected by the PDE resonance, have been convincingly shown to fall with age in control subjects33 and yet increase linearly with dementia progression.32-34⇓⇓

Interestingly, the same authors reported a normalization of phosphate metabolites after a year-long trial of acetyl-l-Carnitine, in conjunction with attenuation of symptom progression.35 However, a 21-day trial of metrifonate (an acetylcholinesterase inhibitor), which improved clinical variables, failed to demonstrate phosphorous metabolite change.36 Whether this was due to the medication not affecting phosphorous metabolism, or if such changes were masked by the low sensitivity and resolution of ³¹P-MRS, is unresolved. A clinical finding of particular interest, and which requires follow-up investigation, is a case report of phosphorous metabolism changes preceding clinical symptoms altogether.32 ³¹P-MRS studies in AD are summarized in the table (a more comprehensive summary table appears in the online version of this article at www.neurology.org).

There are a number of limitations to these studies. Postmortem studies are dependent upon time until fixation and close attention to this variable has been lacking. The method of absolute quantitation used in the negative reports has the disadvantage of sampling particularly large brain regions and has lacked demonstrated repeatability. On the other hand, all of the positive ³¹P-MRS findings have in effect come from a single laboratory, detracting from confident generalization.

Despite these limitations, the published reports suggest that in early, and possibly presymptomatic AD, a distinct biochemical abnormality is evident using regional ³¹P-MRS. PME, the membrane lipid precursor group, becomes elevated and then falls back to normal concentration with dementia progression. Phosphocreatine levels move in the reverse direction, dropping off in mild dementia and recovering later. PDE, on the other hand, increase in step with senile plaque counts, a pattern quite distinct from the limited reports in the normal elderly of a gradual decline of this metabolite.

These conclusions warrant two important qualifications. First, MRS changes may not be as significant or extensive as the abnormalities in glucose metabolism observed in AD.31 If, however, these biochemical changes can be detected early in the disease process, they may have major diagnostic implications. Second, few data indicate that these changes are specific to AD. Whereas one study32 showed successful discrimination of AD from multiple subcortical infarct dementia, another study showed that an “other dementias” group could not be accurately distinguished.37 Further metabolite information, particularly that resolved by ¹H-MRS, may be useful in this regard.

Proton MR spectroscopy.

Proton MRS has yielded a growing body of interesting and largely replicable evidence of characteristic metabolite changes in AD, summarized in the table (a more comprehensive summary table appears in the online resource at www.neurology.org). A consistent finding has been a reduction in NAA levels in AD brains. This has been demonstrated in the occipital lobe,20,38⇓ temporoparietal region,39 temporal lobe,40-43⇓⇓⇓ parietal lobe,44 and in the frontal lobe.45 Overall, NAA decrease in AD has been shown in at least 18 reports, including in vitro studies showing a correlation with AD pathology.37,46⇓ Cortical NAA levels do not seem to be related to dementia severity and the decrement is indeed remarkable for its consistent magnitude—about 10% to 15%—across studies with different protocols and between different brain regions. Also, NAA depletion seems higher in gray matter compared with white matter. Studies that have attempted to control for CSF in the VOI suggest that the NAA depletion found in AD is independent of the level of atrophy.

Another striking finding in the literature has been the unforeseen elevation of MI levels by about 15% to 20% in the gray matter of patients with AD. Subjects with age-associated memory impairment show no significant increase in MI in the temporoparietal region,39 yet one study demonstrated an increased MI signal in the posterior cingulate of individuals with mild cognitive impairment.41 No significant MI changes have been confirmed in white matter but a moderate inverse association between frontal white matter MI levels and global mental function has been found.40

Whereas the clinical specificity of the NAA decline in AD is poor, the addition of MI information increases the accuracy of the diagnostic function. The combined NAA/MI ratio is robust in discriminating possible AD cases from age-matched control subjects.18,40,47⇓⇓ For the more challenging task of discriminating AD cases from other possible dementia diagnoses, NAA/MI ratio was reported to yield a modest positive predictive value (74%). A negative MRS result in a subject with possible dementia was of more clinical utility; the same study estimated an 80% negative predictive value.47 A smaller study using a 1.0-T magnet replicated the MI elevation seen in AD and found that this measure could be used to discriminate from patients with multi-infarct dementia.38 Automated AD diagnosis using MRS seems eminently possible, but future studies will require validation of the discriminant function on a prospective sample. More than one VOI should be incorporated into any future studies testing MRS diagnostic ability.

The NAA/MI ratio in patients with AD has also been shown to significantly correlate with Mini-Mental State Examination (MMSE) scores,18 and even to significantly predict MMSE change 12 months later.43 Cognitive change has also been related to MI and NAA alteration in the hippocampi of individuals with AD.42,48⇓ Further research is necessary, but there are intriguing suggestions that ¹H-MRS may have a useful role in prognosis of mental function and tracking of disease progression.

A noteworthy finding has been the equivalence of in vivo choline estimates between pathologic groups and control subjects. Furthermore, although it is known that choline levels in the hippocampus of normal individuals are higher than when compared with other brain regions,49 reports have so far failed to show a difference in hippocampal choline levels between patients with AD and control subjects.42,48⇓ Given that the choline peak seems to demonstrate higher variability when assayed with MRS methods,18 strict correction for age, disease severity, medication status, and dietary intake is necessary to make accurate conclusions.

Membrane structure.

The phospholipid and high-energy phosphate pathways have together been implicated in the early and possibly preclinical stages of AD.34 An explanation for early PME elevation may be blockage in the conversion of PME to membrane phospholipid. Abnormal phosphorylation of protein,50 and the known kinase and protease activations that contribute to β-amyloid deposition,51 may contribute directly to structural weakness in neurons via disruption of PME-phospholipid enzymes. Such structural vulnerability can make neurons more susceptible to lesion-induced toxicity and death. The biochemical con-sequence of this—decreased membrane phosphatidylcholine—has been found in AD brains when compared with controls.25 Pharmaceutical interventions based on this rationale may be worth further investigation, given the ease of in vivo assay of PME.

As AD increases in severity, PME levels have been shown to normalize. This may reflect the influence of biochemical factors that link amyloid deposition and cell membrane synthesis; however, it is clear that a better understanding of this discontinuity is required.

PDE levels, reflecting cell membrane breakdown products, are highly correlated with senile plaque numbers (r = 0.89).27 The progression of AD, with the increase in density and number of senile plaques, can therefore be quantified in vivo by the increase of the PDE groups. Researchers and clinicians appear to have hitherto ignored this important finding. Because in normal subjects PDE levels decline with age, possibly reflecting a loss of membranous vesicles, this relationship may aid the differential diagnosis of dementia when serial phosphorous spectroscopy is possible.

High-energy phosphate metabolism.

Pettegrew et al.32-34⇓⇓ showed phosphocreatine depression in the early clinical stages followed by normalization. Depletion of high-energy buffering groups not only impairs neural function but may also leave neurons vulnerable to oxidant toxicity, stress-induced apoptosis, DNA mutation, and other harmful effects. Given that the main calorific requirement in the neuron is for neurotransmitter release and cycling,52 drug therapy aimed at reducing the neuroenergetic deficit early in the disease may be indicated and has the advantage of reliable in vivo measurement.

Metabolic disturbances in AD are well documented even after controlling for atrophy in the ROI.31,53⇓ It is therefore possible that phosphocreatine depletion may provide an additional preclinical marker. Cortical glucose metabolic rates decrease, however, with dementia progression while phosphocreatine levels normalize. Longitudinal studies with these two markers, when combined with neuropsychiatric assessments, may yield useful data.

Neural viability.

Methods based on absolute quantitative techniques in vivo and on postmortem analysis have concluded that an approximate 10% to 15% reduction in NAA concentration occurs in the gray matter of individuals with AD. This change appears to be independent of the degree of atrophy as seen on MRI. These findings have been inconsistently understood in terms of neuronal loss or decreased neural density.

NAA depletion may well represent two separable processes: actual neural cell death in the brain region under examination, and another process reflecting a decreased level of functionality or metabolic integrity in the neurons that are still viable. Abnormalities seen on MRS may anticipate anatomic degeneration discerned on MRI. Reports of the reversal of NAA deficits in certain neurologic conditions support the latter “cellular fitness” interpretation. The functional significance of the NAA variance is underlined by studies showing that NAA levels covary with cognitive performance54; one investigation using a healthy elderly sample demonstrated that this relationship is independent of other neurocognitive predictors and is specific in terms of topographic locale and neuropsychological domain.7

NAA variation in structurally sound tissue may therefore reflect processes related to altered mitochondrial activity16 or neuroinflammation.55 This is not inconsistent with the understanding of AD as an inflammatory reaction,56 or investigation of the mitochondrial basis of neurologic disorders,57 so that the quantitation of NAA may be of value when attempting to monitor pharmacologic efficacy. Whether NAA depletion in AD can be partially reversed by treatment, pharmacologic or cognitive–behavioral, is a question worth examining.

Reductions in NAA have been found to be widespread in the cortex and to a lesser degree in the white matter of individuals with AD, suggesting that not only are neural cell bodies damaged, but axonal injury may also be involved. The use of MRS to investigate the temporal primacy of cortical metabolic decline versus axonal decline in AD is of great interest. In contradistinction to the progression of structural changes in AD,58 this pattern of neural peptide depletion, in addition to the early membrane changes seen with ³¹P-MRS, indicates that a diffuse insult to neurocellular viability early in the disease process may be in effect.

Inositol disturbance.

In studies using short echo times, MI levels were found elevated. Similar elevations have been found in the frontal lobes of patients with frontotemporal dementia59 and the basal ganglia of those with Huntington’s dementia.60 Understanding the significance of the MI increase is difficult given the complexity of its biochemical pathways, a summary of which is provided by Ross et al.61 A simple explanation is based upon the enzymatic conversion of MI to inositol triphosphate (IP₃).62 IP₃ functions as an important intracellular second messenger that bonds to plasma membrane. Alterations in this compound can have various neurophysiologic consequences, such as impairment of cholinergic activity and inhibition of Ca²⁺ release. Consistent with this explanation is the finding that IP_n levels are depleted in the hippocampi of patients with AD.63

There are other possible explanations of MI alterations in AD. Inhibition of MI conversion to neural membrane phosphatidylinositol could lead to MI accumulation and phosphatidylinositol depletion, a result also found in postmortem AD brains.64 MI may also act as a marker of neuropathologic lesions, a suggestion supported by the report of a strong correlation between the MI signal and neurofibrillary tangles.37 Glial cells in general have a high MI content,65 and there is related evidence that MI and choline levels increase in areas of high cellularity.17

Another line of speculation is that MI disturbance has its origins in the kidney rather than the brain. Chronic renal failure induces MI inflation and NAA depletion in a pattern similar to that found in AD.47 One study using proton MRS found that patients undergoing hemodialysis and undialyzed patients with chronic renal failure had a 14% increase in neural MI levels.66 Transplantation reversed these metabolic anomalies. The possible role of aluminum neurotoxicity in patients undergoing dialysis and those with AD67 should also be considered, although the evidence remains inconclusive.

MI is known to break down into glucuronate, which plays a detoxifying role in the liver.19 Patients with diabetes, carbon monoxide poisoning, and AD all demonstrate cerebral MI elevation,47 and these entities may be thought to be related through neurotoxicity. In a wider sense, a decline in detoxification efficacy has been implicated in aging and dementia.68 Although the hypothesis is untested, MI may play a similar detoxification role in the brain, and its elevation in AD may reflect increased neurotoxic load rather than pathology per se.

Although the meaning of MI elevation remains unclear, it seems that such a finding may have a prognostic value as well. Investigations of Down’s syndrome suggest that MI alteration may precede and predict subsequent cognitive decline.69 Similar associations between MI levels and changes in MMSE scores have been reported in AD.40,43⇓

Cholinergic imbalance.

Perhaps the most perplexing finding to come from ¹H-MRS study of AD has been the lack of a detectable abnormality in choline levels. Loss of cholinergic neurons in AD, particularly in the hippocampus, predicts an increase in soluble free choline and glycerophosphorylcholine. In vitro studies have confirmed higher free choline and glycerophosphorylcholine levels in AD brains.61 Of the 15 in vivo MRS studies reviewed, however, only two reported a significant choline elevation in AD.70,71⇓ Interestingly, both studies used chemical shift imaging (CSI) methods, and one used increased magnet strength (2 T), which can increase the sensitivity of signal detection. Evaluation of these findings is difficult, however, given CSI interpretation is still developing. Single-voxel MRS in regions of AD cholinergic alteration as prescribed by CSI may help resolve these contradictions. The poor repeatability of choline quantitation17 may otherwise be due to the rapid metabolism of dietary choline.9

Measurement of intraindividual change may be another way to obtain more accurate choline information. One group examined the effect of xanomeline, a muscarinic agonist, in the midparietal lobe of subjects with AD at baseline and 2 months after drug therapy.72 The prediction was that choline therapy should promote decreased cholinergic membrane breakdown and so lead to relative decreases in MRS-visible choline. Choline was found significantly decreased in the drug takers versus placebo control subjects. An interesting follow-up report found a greater choline decrement in pharmacologic responders versus nonresponders.73 The combination of therapeutic intervention, repeat spectroscopy, and cognitive assessment is likely to be a profitable and interesting research design in the investigation of AD.