Proton MRS in Mild Cognitive Impairment

Abstract

Mild cognitive impairment (MCI) is a clinical syndrome operationalized for early diagnosis and treatment of Alzheimer’s disease (AD). Many individuals with MCI are at the prodromal stage of AD or other dementia. Various quantitative MR techniques that measure the anatomic, biochemical, microstructural, functional, and blood-flow changes are being evaluated as possible surrogate measures for early diagnosis and disease progression in AD. The pathology underlying MCI heterogenous, dominated by AD, cerebrovascular disease, Lewy body disease, or a mixture of these pathologies in community-based autopsy cohorts. Proton MRS metabolite markers may help identify and track etiologies that typically underlie MCI in the elderly. The role of proton MRS will especially be critical for pathophysiological processes for which a reliable biomarker does not exist such as glial and microglial activation in neurodegenerative dementia.

The incidence and prevalence of Alzheimer’s disease (AD) increase significantly with aging. With improvements in healthcare and increasing life expectancy, AD has become a significant public health problem of this century (1). There are no proven treatments for AD pathology, however current efforts to arrest or slow disease progression generate the prospect for preventive interventions (2). There is considerable interest in early diagnosis by identifying individuals with cognitive difficulties who eventually progress to dementia, from those who are aging with normal cognitive function (3). Mild cognitive impairment (MCI) was established on clinical grounds in order to identify symptomatic individuals who do not meet the criteria for dementia (4). A majority of these individuals develop dementia in the future. This article reviews the proton magnetic resonance spectroscopy (¹H MRS) findings in MCI and the value of ¹H MRS for early diagnosis and disease progression in dementia. First, ¹H MRS findings in dementia syndromes that are commonly preceded by MCI will be reviewed in order to interpret the role of ¹H MRS in MCI.

¹H MRS Findings in Dementia

¹H MRS is unique among diagnostic imaging modalities, because the signals from several different metabolites that are sensitive to a different aspect of the neurodegenerative disease processes are measured within a single acquisition period. Each metabolite may be a surrogate marker to in-vivo pathologic processes at the molecular or cellular level. The potential utilities of ¹H MRS as a biomarker in MCI and dementia include early diagnosis, differential diagnosis, and monitoring of pathological progression (5).

¹H MRS metabolite markers of AD have been investigated in clinical cohorts for more than two decades and the neuronal metabolite NAA is consistently found to be lower and the glial metabolite mI is found to be higher in the ¹H MR spectra of patients with AD than cognitively normal elderly(6-15). There are conflicting reports on the membrane integrity marker choline (Cho) levels. Some studies identified elevated Cho levels (10,13,16) in people with AD, and some did not (14,17-20). Creatine (Cr) levels are typically stable in AD compared to age matched controls (14,15,18,19,21-23). For this reason, Cr peak is generally used as an internal reference to adjust for atrophy and acquisition related variability. Postmortem MRS analyses of perchloric acid extracts of AD brains show correlations between MRS metabolites and density of NFT and senile plaques in the tissues (9,24). Furthemore, ¹H MRS in studies in transgenic mouse models of AD consistently showed in vivo metabolite alterations with increasing age (25-27). Recently, we validated these findings in a clinical cohort by demonstrating that antemortem posterior cingulate gyrus NAA/Cr, mI/Cr and NAA/mI on MRS correlate with the likelihood of postmortem Alzheimer type pathology at autopsy (28) (Figure 1).

Figure 1. Posterior cingulate gyrus voxel ¹H MRS findings by pathological diagnosis of AD.

The pathological diagnosis of AD is classified as low, intermediate and low likelihood of AD. For each pathological diagnosis, the plot shows individual values, a box plot of the distribution, and the estimated mean and 95% CI for the mean. The strongest association was observed with the NAA/mI ratio (R_N²=0.40) (28). With permission from Radiology.

Although AD is the most common etiology of MCI, Lewy body related pathology, ischemic vascular disease and infarctions are other pathophysiological processes that are common along with early AD pathology in people with MCI (29-32). ¹H MRS findings in vascular dementia is characterized by a reduction in NAA and NAA/Cr levels even in cortical regions remote from the infarction (16). White matter NAA/Cr is lower in patients with vascular dementia than in patients with AD, (33,34). NAA levels were also lower in patients with stroke who had cognitive impairment compared to those who were cognitively normal (35). Because reduction in NAA was observed in regions remote from the infarction, it is thought that the reduction in NAA/Cr is associated with neuronal dysfunction (35). Cortical mI/Cr levels on the other hand are normal in patients with vascular dementia (16,36). Because mI/Cr is elevated in patients with AD, mI/Cr may help identify the presence of AD in a demented patient with cerebrovascular disease or in a patient with MCI.

Another common dementia pathology identified in people with MCI is Lewy body pathology. Lewy body pathology by itself is less common than the mixed (AD and Lewy body) type (37). In our ¹H MRS series, patients clinically diagnosed as dementia with Lewy bodies have normal NAA/Cr levels, whereas patients with AD and vascular dementia have lower NAA/Cr levels than normal, in the posterior cingulate gyri(16). Patients with dementia with Lewy bodies have preserved neuronal numbers at autopsy (37). Likewise, normal NAA/Cr levels suggest integrity of neurons in the posterior cingulate gyri, which may be useful in distinguishing patients with dementia with Lewy bodies from AD or vascular dementia. White matter NAA/Cr however was significantly reduced in patients with DLB compared to the control group (38). It is possible that NAA/Cr is decreased in other cortical regions of people with dementia with Lewy bodies, that have not yet been studied with ¹H MRS. On the other hand, Cho/Cr ratios are elevated in patients with dementia with Lewy bodies compared to normal. Elevation of Cho in dementia with Lewy bodies and AD may be the consequence of increased membrane turnover due to dying back of the neuropil. Another explanation however is the down regulation of choline acetyltransferase activity which may be responsible for this change in both AD and dementia with Lewy bodies. Both AD and dementia with Lewy bodies are characterized by cholinergic dysfunction, although cholinergic dysfunction may be more severe in dementia with Lewy bodies than in AD (39). The finding that Cho/Cr levels decrease with cholinergic agonist treatment in AD (40), raises the possibility that Cho/Cr levels may be a useful marker of cholinergic dysfunction associated with both AD and Lewy body related pathology (Figure 2)

Figure 2. Posterior cingulate gyrus voxel ¹H MRS findings in common dementia syndromes.

Alzheimer’s disease (AD), frontotemporal dementia (FTD), dementia with Lewy bodies (DLB), vascular dementia (VaD).

The Heterogeneity of Mild Cognitive Impairment

The progression of AD pathophysiological processes start decades before the clinical diagnosis of AD and the earliest cognitive impairments occur in the memory domain (41). The syndrome of amnestic MCI represents this prodromal phase in the progression of AD (4). More recently, the construct of MCI has been broadened to include individuals with impairments in non-amnestic cognitive domains such as attention/executive, language or visual-spatial processing domains (42). The clinical presentation of this broadened definition of MCI is heterogeneous.^. Both the amnestic and non-amnestic subtypes of MCI may present with involvement of a single cognitive domain or multiple cognitive domains. It is clear from several independent studies that most people with the amnestic form of MCI who progress to dementia in the future, develop AD(43-49). People with non-amnestic MCI on average have more vascular comorbidity and infarctions as well as a higher prevalence of extra pyramidal features, mood disorders, and behavioral symptoms than people with amnestic MCI (50,51). The ethiology of MCI is also heterogenous. A variety of early stage dementia-associated pathophysiological processes such as AD, cerebrovascular disease and Lewy body pathology have been identified in patients with MCI at autopsy (29,30,31,32). Many of these pathologies co-exist in MCI (29) and require different therapeutic strategies.

Furthermore, all patients with MCI do not develop dementia at a similar rate (52,53). The heterogeneity of MCI warrants development of non-invasive biomarkers that can predict the rate of future progression to different dementias, for early diagnosis and treatment with potential disease-specific preventive interventions.

¹H MRS Findings in MCI

Early ¹H MRS studies in MCI included individuals who had impairments in memory function (i.e. amnestic MCI) (8,54-56). A majority of patients with amnestic MCI develop AD in the future, and many of these individuals have early AD pathology (46). In keeping with this, the ¹H MRS findings in amnestic MCI are similar to but milder than the findings in AD (8,55,56) (Figure 3). However there are distinct group wise differences in MRI and ¹H MRS findings between amnestic MCI and non-amnestic MCI subtypes. Patients with amnestic MCI tend to have smaller hippocampal volumes and elevated mI/Cr ratios compared to patients with non-amnestic MCI and cognitively normal controls. On the other hand, non-amnestic MCI patients have normal hippocampal volumes and normal mI/Cr ratios, but a greater proportion of these patients had cortical infarctions compared to the amnestic MCI patients (50). Both hippocampal atrophy and elevated mI/Cr are sensitive markers of early AD pathology, and the severity of these abnormalities correlate with the pathologic severity of AD (28,57-62). For this reason, hippocampal atrophy and elevated mI/Cr most likely represent a higher frequency of early AD pathology in patients with amnestic MCI compared to non-amnestic MCI. On the contrary, normal hippocampal volumes and mI/Cr ratios in the non-amnestic MCI subtype suggest that other pathologies in addition to AD underlie non-amnestic MCI. Higher prevalence of cortical infarctions on MRI, history of TIA and stroke in non-amnestic MCI patients suggest that cerebrovascular disease is one of the pathological contributors to non-amnestic MCI.

Figure 3. Posterior cingulate gyrus voxel ¹H MRS findings in patients with mild cognitive impairment (MCI) and Alzheimer’s disease (AD) compared to a cognitively normal subject.

An 8 cm³ (2×2×2cm) voxel includes the right and left posterior cingulate gyri and inferior precuneate gyri. The anterior border of splenium, the superior border of corpus callosum and the cingulate sulcus are used as the anatomical landmarks for voxel placement.

MRS as a Biomarker in for Early Diagnosis of Dementia in MCI

The pathologic and clinical heterogeneity of MCI require multimodality imaging markers that are sensitive to the various dementia related pathophysiological processes for early diagnosis in patients with MCI. The most common dementia-related pathologies observed in MCI include AD, cerebrovascular disease and Lewy body disease (29-31). Lesions that are associated with cerebrovascular disease on MRI include infarctions and white matter hyperintensities on T2 weighted images. These cerebrovascular lesions are more common in patients with MCI compared to cognitively normal older adults (50). An MR marker that is highly sensitive to the pathophysiological processes of AD specifically the neurofibrillary tangle pathology-associated neurodegeneration early in the disease course is hippocampal atrophy (60,62). Both the presence of cortical infarctions (63) and hippocampal atrophy (64) are predictors of dementia risk in MCI.

There is evidence that MRS is sensitive to the pathophysiological processes associated with dementia in patients with MCI (65,66). Decreased NAA/Cr ratio in the posterior cingulate gyrus voxel is associated with an increased risk of dementia in patients with MCI (63) (Figure 4). Furthermore, posterior cingulate gyrus voxel NAA/Cr levels decline over time in patients with MCI who progress to AD diagnosis (67). MRS is complementary in predicting future progression to dementia in MCI when considered with other strong predictors of dementia risk in MCI such as hippocampal volumes and cortical infarctions. Decreased posterior cingulate gyrus NAA/Cr increases the risk of progression to dementia in MCI patients with hippocampal atrophy and the risk of dementia increases even further when cortical infarctions are present in a patient with MCI (63) (Figure 5). This is consistent with cross-sectional studies showing the added value of ¹H MRS and hippocampal volumes for discriminating cognitively impaired but non-demented individuals from cognitively normal subjects (56). Furthermore, hippocampal volumes and NAA/Cr levels are independent and complementary predictors of verbal memory on neuropsychometric testing in non-demented older adults, demonstrating that verbal memory depends on both structural and metabolic integrity of the hippocampus (68).

Figure 4. Lower NAA/Cr increases the risk of dementia in MCI.

Estimates of the probability of remaining free of dementia over time based on age, gender, and education -adjusted Cox models with NAA/Cr as the predictor. Estimates assume a 76 year old patient with 14.6 years of education and represent a weighted average of the curves for men and women where the weights are the proportion of men and women in our sample (63). With permission from Neurology.

Figure 5. Multiple MR markers of underlying dementia pathologies improve the ability to identify patients with prodromal dementia over a single MR marker.

Estimates of the probability of remaining free of dementia for four patient groups with increasingly negative prognoses. Group A has adjusted hippocampal volume and NAA/Cr 1 SD above MCI average and no cortical infarctions. Group B has adjusted hippocampal volume 1 SD below MCI average with NAA/Cr 1 SD above MCI average and no cortical infarctions. Group C has adjusted hippocampal volume and NAA/Cr both 1 SD below MCI average and no cortical infarctions. Group D has adjusted hippocampal volume and NAA/Cr both 1 SD below MCI average and cortical infarctions (63). With permission from Neurology.

¹H MRS in Preclinical AD

Recently, the diagnostic criteria for AD and MCI were revised by two separate workgroups charged by the National Institute on Aging and Alzheimer’s Association (69-71). A third work group was charged to define the preclinical stage of AD in light of the evidence that the pathophysiological process of AD begins decades before the diagnosis of clinical dementia (41). The change in most well validated imaging biomarkers have been modeled for the three clinical stages of AD (i.e. preclinical AD, MCI and AD). According to this hypothecial model, the change in β-amyloid (Aβ) pathology imaged with PET amyloid ligands precede the change in imaging markers of neurodegeneration associated with the neurofibrillary tangle pathology of AD such as hippocampal atrophy on MRI (72). The model that emerged from evidence on well validated imaging biomarkers will be critical for tracking disease progression and for assessment of primary and secondary preventive interventions in individuals at the preclinical and MCI stage of AD.

Several well validated imaging biomarkers exist for various pathological features of the early AD pathology such as increased Aβ load on PET, atrophy on structural MRI or glucose metabolic reductions on PET. However there are other features of AD pathology for which a well valid biomarker does not exist. For example there is no widely accepted biomarker for glial or microglial activation. The glial metabolite mI quantified with ¹H MRS may potentially be useful as a biomarker for glial activation in neurodegenerative diseases including AD.

Cross-sectional studies indicate that mI/Cr is elevated in MCI and mild AD even in the absence of a decrease in NAA/Cr (8,54,73). We recently validiated this finding in a pathology-confirmed sample of older adults with a range of AD pathology showing that the mI/Cr elevation is associated with intermediate likelihood (i.e. an earlier stage) AD pathology whereas the decrease in NAA/Cr is associated with higher likelihood (i.e. a later stage) AD pathology (28) (Figure 1). Furthermore, mI/ Cr levels increase in the predementia phase of Down’s syndrome(6,74) and in presymptomatic individuals with familial dementia (75,76) even in the absence of structural MRI and NAA/Cr changes(76). The mI peak consists of glial metabolites that are responsible for osmoregulation (77,78). MI levels correlate with glial proliferation in inflammatory CNS demyelination (79). Because the dense cored amyloid deposits in AD are surrounded by clusters of microglia and astrocytes (80), it is thought that the elevation of the mI peak is related to glial proliferation and microglial activation in AD (81,82). A significant correlation between mI/Cr levels and amyloid load measured with Pittsburgh Compound-B PET imaging was found in a population-based sample of 311 cognitively normal older adults (83). It is possible that the mI/Cr levels are associated with the microglial and glial activation that surround the senile amyloid plaques. However a ¹H MRS - histology correlation study in a mouse model of spinocerebellar ataxia type 1 found elevated mI/Cr levels even in the absence of gliosis (84). Based on limited evidence, it is not be possible to attribute elevation in mI solely to glial activation in neurodegenerative diseases. Although there is evidence that mI/Cr elevation is an early marker in sporadic AD, familial AD and frontotemporal lobar degeneration even before cognitive impairment, loss of neuronal integrity and atrophy, histological confirmation is needed to better understand the biological basis of mI/Cr elevation in MCI.

Conclusions

MCI is a clinically and pathologically heterogeneous disorder. Current data indicate that a single imaging marker will be insufficient for determining the underlying pathophysiological processes that contribute to cognitive impairment in MCI. ¹H MRS may potentially provide information on the underlying pathologies in patients with MCI that is not available from other imaging biomarkers. Data from cognitively normal older adults and cognitively normal adults at risk for familial dementia suggest that ¹H MRS may be useful as a biomarker for preclinical pathological processes and potentially assessing the response to preventive interventions. Although significant progress has been made on improving the acquisition and analysis techniques in ¹H MRS, translation of these technical developments to clinical practice have not been effective. The main reasons for ineffective translation of technology to clinical practice or patient-oreinted research are two fold: 1) Lack of standardization for multi-site applications and normative data 2) Insufficient understanding the pathological basis of ¹H MRS metabolite changes. Advances on these grounds would further increase the impact of ¹H MRS as biomarker for the early pathological involvement in neurodegenerative diseases and in turn increase the use ¹H MRS in clinical practice.