Emerging Biomarkers in Cognition

Synopsis

The field of aging and dementia is rapidly evolving with the aim of identifying individuals in the earliest stages of disease processes. Biomarkers allow the clinician to demonstrate the presence of an underlying pathologic process and resultant synapse dysfunction and neurodegeneration, even in those earliest stages. For example, PET amyloid imaging and CSF Aβ₄₂ provide direct evidence of amyloid deposition and structural MRI, FDG-PET or SPECT and CSF tau provide indirect evidence of synapse dysfunction and neurodegeneration when the pathologic process is due to Alzheimer's disease (AD). While this review will focus on biomarkers for mild cognitive impairment (MCI) due to AD, structural MRI, FDG-PET or SPECT, and PET with dopamine ligands are also valuable in suggesting non-AD pathologic processes. While these biomarkers are very useful and can even be applied to diagnostic criteria in MCI, several limitations exist. As the field continues to grow, several new biomarkers are emerging and ultimately, a more biological characterization of subjects’ underlying pathophysiologic spectra will be possible.

Introduction

Degenerative cognitive disorders, such as Alzheimer's disease (AD), exist along a spectrum, from a preclinical stage of individuals without symptoms but with biological predisposition,^1-3 to mild cognitive impairment (MCI) being the earliest manifestation of clinical symptoms, and dementia the most advanced stage. The field of aging and dementia is moving toward an earlier identification of those pre- and initially symptomatic subjects⁴ with goal of intervening with disease-modifying treatments. While 40-60% of subjects with MCI will progress to dementia over 5 years,^{5, 6} MCI is a heterogenous group of variable etiologies, some of which are treatable or reversible.^{7, 8} To that end, there has been considerable research in identifying markers of the underlying pathophysiologic process in those earliest stages and, recently, biomarker data has even been incorporated in the updated research diagnostic criteria of MCI due to AD.⁹

Biomarkers are measurable, in vivo indicators of a specific disease related pathological process. The most common underlying pathological process in MCI is AD; thus, most of the research in biomarkers for MCI has focused on that process and will be highlighted in this review. The use of biomarkers in identifying AD has the following goals:⁹

1. Provide direct evidence of an underlying AD related pathologic process;

2. Provide indirect evidence of AD pathology through markers of synapse dysfunction and neurodegeneration;

3. Assist in determining the stage or rate of disease progression; and

4. Provide evidence of alternative non-AD pathologic process.

Examples of biomarkers for each category above as they relate to MCI due to AD are provided in Table 1. Imaging and CSF measurements are the most widely used biomarkers. Amyloid plaques are the hallmark pathologic process in AD and amyloid PET imaging or CSF measurement of Aβ₄₂ provide direct evidence of that process. While tau deposition in the form of neurofibrillary tangles is also critical for the pathological diagnosis of AD,¹⁰ tau levels are less specific than Aβ markers and may better reflect neuronal injury and synapse loss; thus, tau is a better indicator of neurodegeneration. Other measures of neurodegeneration include structural and functional imaging studies, namely MRI brain atrophy, FDG-PET hypometabolism and SPECT hypoperfusion.

Table 1.

Biomarkers in MCI due to AD

Purpose	Biomarker
Marker of amyloid pathology in AD	PET amyloid imaging CSF Aβ42
Marker of synapse dysfunction and neurodegeneration in AD	FDG-PET or SPECT temporoparietal hypometabolism/perfusion CSF tau Structural MRI (e.g. hippocampal atrophy, cortical thinning)
Marker of disease progression from MCI to dementia due to AD	FDG-PET or SPECT temporoparietal hypometabolism/perfusion CSF tau Structural MRI (e.g. hippocampal atrophy, cortical thinning)
Marker of Non-AD process	Structural MRI (e.g. significant infarcts/cerebrovascular disease, subdural hematomas, tumor) FDG-PET or SPECT non-temporoparietal hypometabolism SPECT with dopamine ligands

Biomarkers of Aβ deposition

During the pathologic process of Alzheimer's disease, the amyloid protein forms insoluble fibrils and is deposited extracellularly. There is considerable interest in instruments that can detect and quantify this process via imaging¹¹ and CSF biomarkers of amyloid deposition. The ability of plasma amyloid levels to directly quantify brain amyloid levels, though, has not been firmly established¹² and will be discussed elsewhere.

PET Amyloid Imaging

Amyloid imaging via a positron emission tomography (PET) tracer binding to fibrillar amyloid was introduced around 2004, using the carbon 11 Pittsburgh compound B (PiB). ¹³ The short half-life of the PiB limits the clinical application of the test and in April 2012, the Food and Drug Administration (FDA) approved a fluorinated tracer, ¹⁸F-AV-45 or florbetapir, while other fluorinated tracers are in development. This tracer has a longer half-life, allowing for regional manufacturing and distribution that is not available with PiB.¹⁴ While an amyloid PET is interpreted as either positive, indicating the presence of tracer uptake of amyloid fibrils, or negative, that is lacking the typical uptake of tracer, regional information may be useful. Compared to controls, amyloid PET scans in patients with AD show widespread uptake of tracer in cortical areas, with the most prominent uptake in the frontal cortex and precuneus followed by the parietal and temporal cortices (Figure 1); minimal uptake is seen in cortical areas known to be relatively unaffected in AD, such as the primary motor, sensory and visual cortices.¹⁵ Temporal amyloid deposition may also be independently related to memory deficits in cognitively normal elderly or patients with MCI.¹⁶

Figure 1. Amyloid PET Imaging.

Axial sections of PET images obtained with the administration of ¹⁸F-AV-45, or florbetapir, as a tracer for fibrillar amyloid in an individual without (B) and with (C) AD. The tracer binds nonspecifically to white matter; using cerebellar white matter as a reference (A, arrow), the lack of tracer uptake in cortical areas creates a distinct gray-white junction in an individual with a negative scan (B, arrow) while the gray-white junction is lost with the uptake of tracer in cortical areas in an individual with a positive scan (C, arrow).

Positive amyloid imaging correlates with AD pathology at autopsy in most, but not all, cases.^{11, 17-19} In MCI, a positive PiB^{20, 21} or florbetapir^{22, 23} amyloid imaging scan correlates with progression from to dementia due to AD, while a positive PiB scan also correlates with time-to-progression.²⁴ While these data are convincing, precise prediction models have not been generated as of yet. There still can be a considerable time lag for individuals with positive amyloid imaging scans with respect to their progression from the MCI to the dementia stage of AD, and more research needs to be done to clarify this progression pattern.

CSF Aβ₄₂

B-amyloid protein ending at amino acid 42 (Aβ₄₂) is the core protein of neuritic plaques. As the protein is deposited in the brain, lower levels remain in the CSF. Low CSF Aβ₄₂ has been shown to correlate with pathologic diagnosis of AD at autopsy.^{25, 26} Across multiple studies, low CSF Aβ₄₂ has a mean sensitivity of greater than 85% with a specificity of 90% in detecting AD versus normal aging across studies.²⁷ In MCI, low CSF Aβ₄₂ is a predictor of progression to dementia^28-30 with similar or lower sensitivity.²⁷

Since both amyloid PET imaging and CSF Aβ₄₂ are biomarkers of brain amyloid deposition, it stands to reason that a positive amyloid scan should correlate with low CSF Aβ₄₂. This has, in fact, been demonstrated,^31-34 further asserting the validity of these tests as biomarkers for amyloid deposition.

[Tags: amyloid; Alzheimer's disease; biomarkers; amyloid imaging; Abeta; CSF]

Biomarkers of Neurodegeneration

Amyloid deposition is insufficient by itself to result in cognitive impairment. The second goal of biomarkers is therefore to provide evidence of neurodegeneration.³⁵ This can be accomplished through a variety of markers, including MRI, FDG-PET, and CSF tau levels.

MRI

There has been more literature generated on the utility of structural MRI in predicting progression to dementia than most of the other biomarkers.^{24, 36, 37} Cerebral atrophy correlates with neuronal loss;³⁸ thus volumetric measurements on MRIs, such as hippocampal atrophy, ventricular volume expansion, whole brain atrophy, and cortical thinning provide can useful information on neurodegeneration (Figure 2). While these measurements are not specific to AD, they do predict progression from MCI to dementia. ^39,40 Hippocampal atrophy can be measured both qualitatively and quantitatively. Several scales are being introduced to allow the clinician to assess medial temporal lobe atrophy in the office and visual rating of hippocampal atrophy can be quite useful.⁴¹ Automated volumetric MRI measurements have also become available through programs like FreeSurfer;⁴² as these are becoming more widely available for use in clinical practice, they could soon be used in a fashion similar to bone density measures used to assess osteoporosis.⁴³

Figure 2. MRI in cognitively normal individuals, MCI and dementia.

Coronal T1 MP RAGE sections through the hippocampus in a cognitively normal individual (A) and individuals with MCI (B) and dementia due to AD (C). Hippocampal atrophy (long white arrows), ventricular dilation (arrowheads), and cortical thinning (black arrows) with widening of the sulci are first subtly noted in the MCI stage (B) and rapidly increase through the dementia stage (C), when they become readily apparent. These structural MRI changes are predictors of conversion from MCI to dementia.

FDG-PET

The brain metabolic rate directly correlates with synaptic activity.^{44, 45} Degenerating areas of the brain theoretically show less synaptic activity and therefore less metabolic activity. As glucose is the source of fuel for the brain, providing, a glucose tagged PET tracer (e.g. fluorodeoxyglucose or FDG) provides valuable information about the metabolic activity of the brain and thereby serves as a marker of neurodegeneration. Patients with clinical and pathologically confirmed AD show a characteristic FDG-PET pattern of hypometabolism in the bilateral temporoparietal association regions, including the posterior cingulate and precuneus, sparing the primary motor, sensory and visual cortices,^{46, 47} even when correcting for cerebral atrophy (Figure 3, part C).⁴⁸

Figure 3. FDG-PET in cognitively normal individuals, MCI and dementia.

Axial section of FDG-PET scans in a cognitively normal individual (A) and individuals with MCI (B) and dementia due to AD (C). Less tracer uptake is apparent in hypometabolic areas. FDG-PET hypometabolism is a marker of neuronal injury and dysfunction. Subtle temporoparietal hypometabolism is noted in MCI (B, white arrows), which is a predictor of conversion to dementia. The temporoparietal hypometabolism is more apparent in an individual with dementia due to AD (C, white arrows).

The temporoparietal hypometabolic pattern on FDG-PET characteristic of AD can appear prior to the demonstration of any clinical symptoms. ⁴⁹ This is most prominent in apolipoprotein E4 homozygotes but also has been seen in heterozygotes.⁵⁰ Among patients with MCI, the temporoparietal metabolic pattern is indicative of conversion to dementia (Figure 3, part B).^51,52

SPECT

Single-photon emission computed tomography (SPECT) is an imaging technique using radioligands to measure regional cerebral blood flow (rCBF). As brain perfusion is tightly correlated with brain metabolism, brain SPECT provides a surrogate marker similar to FDG-PET for functional brain activity, and, inversely, neurodegeneration. SPECT scans are the most widely available functional neuroimaging technologies given the relative ease of use and lower cost of the radioligands compared to PET; SPECT radioligands generally have longer half-lives, do not need a cyclotron for manufacturing, and are FDA approved for brain imaging. PET scans, though, have greater spatial resolution.

Similarly to PET, SPECT scans show hypoperfusion in the posterior temporoparietal association regions sparing the primary cortices,⁵³ even correcting for atrophy⁵⁴ in dementia due to AD that correlates with neurofibrillary tangle pathology.⁵⁵ In patients with MCI, hypoperfusion in these areas is predictive of progression to dementia.^{56, 57}

While PET and SPECT yield similar results, PET has greater spatial resolution and is less confounded by coexistent cerebrovascular disease. Direct comparisons of the two imaging modalities have shown greater sensitivity and overall accuracy for PET vs SPECT, even when utilizing high resolution SPECT scanners.^{58, 59}

CSF tau

Tau protein plays a key role in the pathology of AD.¹⁰ In AD, tau accumulates in neurons, disrupting normal activity and causing its release into the extracellular space. CSF tau levels are therefore increased in AD,^{60, 61} and this finding does correlate with the presence of neurofibrillary tangles in the pathological diagnosis of AD.^62,63 Likewise, normal CSF tau levels are found in AD mimickers, such as depression,⁶⁴ alcoholic dementia⁶⁵ and Parkinson's disease,⁶⁶ aiding in the discrimination of these diseases.

However, elevated CSF tau is not specific for AD and can be seen in a variety of neurologic conditions, such as stroke, traumatic brain injury or prion diseases.^67-69 The intensity of elevated CSF tau reflects the intensity of neurodegeneration. In AD, though, phosphorylated tau is the principal component of neurofibrillary tangles.⁷⁰ Since phospho-tau is not elevated in stroke⁷¹ or Creutzfeldt-Jakob disease,⁷² elevated CSF levels of phospho-tau may provide improved diagnostic discrimination from AD. Phospho-tau levels may also aid in discrimination from frontotemporal dementia⁷³ or dementia with Lewy bodies.⁷⁴

Total tau levels have been more extensively studied than phospho-tau; there is a mean sensitivity greater than 80% to differentiate AD from non-demented aged individuals.²⁷ Elevated levels of CSF total tau are also a predictor of progression from MCI to dementia,^28-30 with similar or lower sensitivity.²⁷ Phospho-tau levels have a similar specificity around 80%, with modest improvement in sensitivity to 92% in discriminating AD from non-demented aged individuals.²⁷ Phospho-tau can likewise predict conversion from MCI to dementia.^{28, 75} However, there is a large variation in these sensitivity and specificity figures across studies, and these studies examine different phosphorylated tau epitopes.²⁷ Likewise, there is large variability in assay results across different centers with no standardization of analytical techniques or clinical procedures.²⁹

Combinations of CSF markers

All 3 CSF markers discussed, Aβ₄₂, total tau and phospho-tau can independently predict conversion from MCI to dementia (Table 2). Several studies have examined if combinations of these markers improve diagnostic accuracy. A meta-analysis did, in fact, find that the best predictor of conversion is an abnormal ratio of Aβ₄₂ to total tau or abnormal levels of both markers; the addition of phospho-tau did not have added benefit.⁷⁶ This combination of markers produces a sensitivity of 87% with a specificity of 70% and positive predictive value of 65%.⁷⁶

Table 2.

CSF biomarkers in MCI due to AD

Biomarker	Level
Aβ42	Low
Total tau	High
Phospho-tau	High
Aβ42/total tau	Low

[Tags: biomarkers; Alzheimer's disease; MRI; FDG-PET; SPECT; tau; phospho-tau; CSF; hypoperfusion; hypometabolism; hippocampal atrophy; cortical thinning; neurodegeneration; synaptic dysfunction]

Biomarkers of Non-AD Processes

The last role of biomarkers is to exclude other pathologic processes that may account for cognitive impairment. Any of the above biomarkers can be helpful in this regard. Structural brain MRI can show, among others, cerebral infarcts, significant cerebrovascular white matter disease, subdural hematomas, hydrocephalus or tumors underlying a cognitive process. Furthermore, the posterior temporoparietal hypometabolism and hypoperfusion pattern via FDG-PET and SPECT, respectively, is in distinction to that seen in other neurodegenerative disorders, such as frontotemporal lobar degeneration (FTLD) or dementia with Lewy bodies (DLB), that can clinically appear similar (Figure 4); hypometabolism in FTLD, for example, is most prominent in the anterior frontal and temporal lobes.⁷⁷ FDG-PET can thereby improve diagnostic accuracy between these two disorders,⁷⁸ but do not define these disorders as frontal variants of AD can appear similar.^{79, 80} Furthermore, while parietal hypometabolism/perfusion can be a feature in DLB as it is in AD, the additional presence of occipital hypometabolism⁸¹ or hypoperfusion⁸² can aid in distinguishing DLB from AD. More recently, several radioligands for use in SPECT have been developed that bind to the dopamine reuptake or transporter site. Low radiotracer uptake in the caudate and putamen is a distinguishing feature of DLB from AD,^82-84 and this finding correlates with autopsy diagnosis;⁸⁴ however, reduced tracer uptake does not distinguish DLB from PD or PD with dementia,⁸² and DLB and AD frequently co-exist.⁸⁵

Figure 4. Comparison of FDG-PET in AD, DLB, and FTD.

Sagittal section of FDG-PET scans in individuals with AD (A), DLB (B), and FTD (C). AD is characterized by temporoparietal and posterior cingulate hypometabolism (A, white arrow) relative to other cortical areas, while occipital hypometabolism is noted in DLB (B, dashed arrow) and frontal hypometabolism is apparent in FTD (C, arrowhead).

[Tags: biomarkers; non-AD dementia; MRI; FDG-PET; SPECT; dopamine; dementia with Lewy bodies; frontotemporal dementia; Parkinson's disease; hypometabolism; hypoperfusion]

Temporal Evolution of Biomarkers

Alzheimer's disease is a dynamic process and the various biomarkers do not reach abnormal levels or maxima at the same time but are dependent on the stage one is in along the pathophysiologic process.⁸⁶ From a hypothetical sequence of events proposed by Jack et al, as is seen in Figure 5, the initiating process in the AD cascade involves the deposition of amyloid in the brain, typically during the preclinical stage.⁸⁷ The precise timing of this relative to the subsequent presentation of clinical symptoms is not known, but the construct of the initiating event being amyloid deposition is well documented.⁸⁸ Thus, in vivo markers of amyloid deposition, including amyloid PET imaging and CSF Aβ₄₂, are likely to be abnormal before presence of clinical symptoms. In fact, 20-40% of cognitively normal elderly have abnormal amyloid imaging or CSF Aβ₄₂ levels, with a similar proportion having a pathological diagnosis of AD despite no clinical symptoms.⁸⁷ Serial PiB imaging scans have suggested that amyloid may begin accumulating as many as 15-20 years before the presence of clinical symptoms.^{89, 90} By the time clinical symptoms appear, PiB retention rates and CSF Aβ₄₂ levels have largely saturated and do not correlate with cognitive decline.^{22, 23, 89, 91, 92}

Figure 5. The theoretical evolution of biomarkers in the Alzheimer's pathological cascade.

Amyloid deposition is the initiating event in the pathological cascade and corresponding biomarkers (low CSF Aβ declines and positive amyloid imaging) are detected before the development of clinical symptoms. As neurodegeneration occurs, evidence of brain dysfunction can be measured through CSF tau levels or FDG-PET or SPECT. As memory and clinical function decline, structural brain MRI changes are noted.

Modified from Jack CR Jr, Knopman DS, Jagust WJ, et al. Hypothetical model of dynamic biomarkers of the Alzheimer's pathological cascade. Lancet Neurol 2010; 9: 119-128.

Following the deposition of amyloid, there is a sequence of events characterizing neurodegeneration that leads to the appearance of abnormal biomarkers of neurodegeneration, including MRI, FDG-PET, SPECT and CSF tau.⁸⁷ However, the appearance of abnormalities of these markers is likely also temporally ordered. FDG-PET hypometabolism and elevated CSF tau may precede MRI changes as it is these MRI changes that correlate best with clinical impairment in MCI and AD.^{93, 94} As is depicted in Figure 5, it is only after these events take place that clinical symptoms evolve. Memory impairment is typically the first symptom to appear, characterizing the typical MCI stage. As the disease progresses, other cognitive domains become impaired and daily function is compromised, leading to dementia.

As is also depicted in Figure 5, none of the biomarkers are static, but the rate of change of each biomarker varies over the course of disease progression.^{87, 95} The dynamic changes are easiest to appreciate in imaging modalities, providing an advantage for these biomarkers over CSF (Figure 2 and Figure 3). Cerebral atrophy, for example, begins in the medial temporal lobes and spreads outward to the limbic and paralimbic areas and later the isocortical association areas.⁹⁶ Thus, as depicted in Figure 5 from Jack's hypothetical model of AD, FDG-PET abnormalities in the precuneus and posterior cingulate precede changes in the lateral temporal and frontal areas, with a similar pattern occurring with changes in structural MRI.⁸⁷

Given the dynamic nature of the biomarkers, the presence, absence or intensity of biomarkers can aid in disease staging and prognostication. A cognitively normal elderly individual, for example, is expected to have an abnormal amyloid imaging and low CSF Aβ₄₂ if that individual is destined to develop AD. An individual with early MCI due to AD will likewise have abnormal amyloid imaging and low CSF Aβ42, while CSF tau may be modestly elevated and brain structure on MRI appear relatively normal. As clinical symptoms become more apparent in later stages of MCI and early dementia, levels of CSF tau become definitively elevated and rates of brain atrophy begin to rapidly increase.

This model has proved to be extremely useful for the field of aging and dementia at characterizing putative pathologic processes and their sequence. However, the precise temporal ordering of these events, the shapes of the curves and the thresholds for normal and abnormal on each of these markers are largely unknown.

[Tags: biomarkers; temporal evolution; Alzheimer's disease; pathology; pathological cascade; amyloid; Abeta; tau; synaptic dysfunction; neurodegeneration; CSF; MRI; FDG-PET; SPECT; amyloid imaging]

Application

The combination of a clinical syndrome and biomarker information can result in varying levels of certainty about the underlying process. In a patient with MCI, evidence of both Aβ deposition and neuronal injury confers the highest likelihood that the pathophysiological process is due to AD. Under that hypothesis, new research criteria for the diagnosis of MCI, utilizing biomarker data, have been proposed:⁹

MCI – Core Clinical Criteria

An individual meets core clinical criteria for MCI but

1. No biomarker information is available.

2. Biomarkers conflict with one another.

3. Biomarker results are indeterminate, neither clearly positive nor negative.

MCI due to AD – Intermediate Likelihood

An individual meets core clinical criteria for MCI with either:

1. Positive biomarkers reflecting Aβ deposition but biomarkers of neuronal injury have not or cannot be tested. or

2. Positive biomarkers reflecting neuronal injury but biomarkers of Aβ deposition have not or cannot be tested.

MCI due to AD – High Likelihood

An individual meets core clinical criteria for MCI and also has positive biomarkers reflecting Aβ deposition (positive PET amyloid imaging or low CSF Aβ₄₂) and neuronal injury (structural MRI changes, elevated CSF tau, or hypometabolism/perfusion in the posterior temporoparietal regions).

MCI – Unlikely due to AD

An individual meets core clinical criteria for MCI but has negative biomarkers reflecting Aβ deposition and neuronal injury. An alternate cause of MCI should be sought.

[Tags: Mild Cognitive Impairment; diagnostic criteria; biomarkers; Alzheimer's disease; CSF; neurodegeneration; amyloid]

Limitations

The research evaluating the use of biomarkers and their clinical application are not without pitfalls (Table 3). Many studies are in select research populations and not routine clinical populations.²⁷ Likewise, most study participants are clinically diagnosed, which means that the performance of the biomarkers in those studies cannot be higher than the clinical criteria used.²⁷ Clinical criteria, for example, often do not account for co-existing AD pathology in cognitively impaired subjects^{97, 98} or control subjects with asymptomatic AD pathology;⁹⁹ it is unclear, therefore, if suboptimal diagnostic performance of biomarkers is related to the biomarkers themselves or the populations under which they were studied.²⁷

Table 3.

Limitations of Biomarkers for AD

1. Majority of studies performed in select research populations

2. Lack of validation in autopsy studies

3. Lack of data for follow up past 1-2 years

4. Difficulty in collecting, processing and interpreting CSF biomarker data

5. Non-specificity of biomarkers

6. Precise temporal evolution of biomarkers not known

Furthermore, many studies assessing the utility of biomarkers as indicator of progression from MCI to dementia have been completed over a follow up period of 1-2 years. The conversion rate of MCI to dementia, though, is approximately 10-15% per year.^100-102 A longer follow up period is therefore needed to determine utility of biomarkers in slow or non-progressive individuals.²⁷

There is also a great deal of difficulty for the practicing physician in collecting, processing and interpreting CSF biomarkers. A lumbar puncture is invasive, and may be avoided due to fear patient discomfort or complications. When collecting CSF, non-absorbent polypropylene test tubes are needed to prevent the Aβ₄₂ or tau proteins from binding to the tube walls and falsely lowering the results.²⁷ Additionally, while there is an international standardization exercise underway sponsored by the Alzheimer's Association to bring multiple laboratories together to standardize CSF collection and assay methodology, there is currently not a standard cutoff threshold for CSF measures.

Lastly, despite significant gains in research, a great deal of uncertainty remains in the test results of the biomarkers themselves. First, the biomarkers are not specific for AD. Positive amyloid imaging can be seen in cerebral amyloid angiopathy;^{103, 104} low CSF Aβ₄₂ can be seen in DLB,¹⁰⁵ frontotemporal dementia, vascular dementia,⁶⁶ Creutzfeldt-Jakob disease,¹⁰⁶ amyotrophic lateral sclerosis¹⁰⁷ or multiple system atrophy;¹⁰⁸ elevated CSF tau can be found in stroke, traumatic brain injury or prion diseases.^67-69 Second, despite the great usefulness of Jack's model of biomarkers in the AD cascade, the precise temporal ordering of events, shapes of the curves and thresholds for normal and abnormal is unknown.⁸⁷ The lag phase, for instance, between amyloid deposition and appearance of neuronal degeneration remains to be elucidated. Finally, it is not uncommon for biomarker results to conflict, and the exact interpretation of the data in this circumstance is unknown.

[Tags: biomarkers; limitations; CSF; amyloid; tau; Alzheimer's disease; non-AD dementia; dementia with Lewy bodies; frontotemporal dementia; temporal evolution; pathological cascade]

Emerging Biomarkers

Several other measures, including other CSF and imaging studies, as well as blood assays are being considered as potential biomarkers but have yet to be translated into clinical practice (Table 4). For example, CSF levels of visinin-like protein-1 (VILIP-1), a neuronal calcium-sensor protein, has demonstrated utility as a marker of neuronal injury in brain injury models,¹⁰⁹ and CSF VILIP-1 and VILIP-1/Aβ42 predict rates of global cognitive decline similarly to tau and tau/Aβ42 in individuals with very mild or mild AD.¹¹⁰

Table 4.

Emerging Biomarkers

Purpose	Biomarker
Marker of amyloid pathology in AD	Plasma Aβ42
Marker of synapse dysfunction and neurodegeneration in AD	CSF VILIP-1 Hydrogen 1 MR spectroscopy Functional MRI Diffusion tensor imaging Magnetic resonance perfusion
Marker of oxidative stress and inflammation	CSF Isoprostane CSF interleukins and growth factors CSF S-100B and GFAP CSF YKL-40
Lipidomics	Serum ceramides
Marker of Non-AD process	CSF α-synuclein CSF neurofilament CSF TDP-43

Plasma Aβ

Given the invasive nature of obtaining CSF, a blood marker for amyloid deposition is ideal to identify and treat pre- or early symptomatic individuals. Several studies have investigated plasma measures of Aβ₄₂ and Aβ₄₀ for early biomarkers of AD pathology. However, results thus far have been inconsistent with a recent systematic review and meta-analysis not showing a significant correlation with plasma levels of Aβ₄₂ and Aβ₄₀ alone and risk of AD or dementia, though Aβ₄₂: Aβ₄₀ ratios were suggestive.¹² Lack of standardized assays and longitudinal studies across all stages of cognitive function have also limited the utility of this biomarker.

Markers of oxidative stress and inflammation

AD is associated with an inflammatory reaction and oxidative stress. Several studies have assessed CSF measures of these processes. Isoprostane, a marker of oxidative stress, is the most widely studied and increases early in the course of AD pathology and may increase diagnostic accuracy in MCI.^{111, 112} S-100B and GFAP, as markers of astrocytic activity and gliosis, and several interleukins and growth factors, as markers of the inflammatory response, have been studied, but data on their utility has been inconclusive.¹¹³ YKL-40, a glycoprotein with sequence homology to bacterial and fungal chitinases and chitin binding ability, is involved in inflammation and tissue remodeling and is increased in AD; a combination of CSF levels of YKL-40 and Aβ₄₂ predicts risk of developing cognitive impairment in preclinical AD.¹¹⁴ Likewise, there continues to be interest in developing peripheral blood markers for these processes, but this has been hampered by the short half-life, lack of correlation between CSF and peripheral blood levels, and association of many of these markers with normal aging.¹¹⁵

Lipidomics

Lipids seem to play a crucial role in AD pathology as the brain is composed of 20% lipids, more than any other organ and the ε4 allele of apolipoprotein E (ApoE), which plays a role in lipid processing and transport, increased susceptibility to late onset AD. Two classes of lipids, glycerophosoplipids and sphingolipids, have been most studied. ¹¹⁶ Certain serum ceramides, a metabolite of sphingomyelin, may predict development of AD and less so all-cause dementia in cognitively normal women¹¹⁷ and cognitive loss and right hippocampal volume loss in MCI.¹¹⁸

Hydrogen 1 MR Spectroscopy (¹H MRS)

Dementing illnesses can affect levels of different brain metabolites, including N-acetylaspartate (NAA), choline (cho), myoinositol (mI), and creatine (Cr). N-acetylaspartate (NAA), for example, is a measure of neuronal integrity and is reduced in the cortical grey and white matter in patients with AD, corresponding to areas of significant neurofibrillary tangle deposition. The NAA/Cr ratio is therefore decreased in the posterior cingulate gyrus, an area involved early in AD pathology, but normal in the occipital lobe, an area without significant tangles until later in the disease course.¹¹⁹ On the other hand, choline and myoinositol peaks are elevated in AD, perhaps related to increased phosphatidylcholine catabolism or downregulation of choline acetyltransferase and activated glial cells around amyloid plaques in AD, respectively.¹²⁰ In single domain amnestic MCI, mI/Cr ratios are elevated similarly to AD, whereas mI/Cr are more likely to be normal in non-amnestic MCI, suggesting the contribution of cerebrovascular disease or other neurodegenerative diseases to the cognitive impairment.¹²¹

Disease specific spectroscopic patterns have been noted and thus MRS can aid in diagnosis. While frontotemporal dementia has a similar pattern to AD (decreased NAA/Cr and increased mI/Cr),¹²² vascular dementia is associated with a low NAA/Cr ratio in the posterior cingulate, but normal mI/Cr ratio. Elevated mI/Cr in a demented individual with significant vascular burden may therefore suggest co-existing AD pathology.^{123, 124} Conversely, DLB is associated with normal NAA/Cr in the posterior cingulate, but elevated Cho/Cr.¹²² However, acquisition and interpretation of MRS data is technically challenging and MRS has not yet been validated in autopsy series to permit its use in the clinical setting.

Functional MRI

Functional MRI (fMRI) modalities highlight the functional connectivity of brain networks, either at rest or task based. At rest, blood oxygen level dependent (BOLD) signal intrinsically fluctuates; brain networks that are functionally connected theoretically demonstrate synchronous temporal fluctuations.¹²⁵ The brain networks active at rest, the default mode network, include the hippocampus, posterior cingulate, precuneus, parietal cortex and medial prefrontal cortex. As these areas are involved early in AD pathology, resting fMRI in AD and MCI shows corresponding decrease in the default mode.¹²⁶ In task based fMRI, demented individuals show decreased activation while preclinical and MCI subjects show increased activation, perhaps representing a compensatory state to mask functional impairment, or, alternatively, a pathologic increase contributing to cognitive decline.¹²⁶

Diffusion tensor imaging

Diffusion tensor imaging (DTI) is a measure of structural connectivity under the principle that water molecules diffuse preferentially along the axon. As AD pathology disrupts cell membranes and thus diffusion of water molecules, the mean diffusivity (MD) is increased and the directional water diffusivity, or fractional anisotropy (FA), is decreased along white matter tracts connecting affected areas of brain, including the medial temporal lobes and posterior cingulate.¹²⁶ Fractional anisotropy measures also can predict conversion of MCI to dementia.¹²⁶

Magnetic resonance perfusion

Employing the same hypothesis as SPECT, MR perfusion imaging aims to measure disease specific regional blood flow decreases via arterial spin labeling techniques.^{127, 128} This modality is attractive as it allows the clinician to obtain similar information as SPECT or FDGPET noninvasively in the same examination with other MR modalities, but sensitivity and specificity measures are yet to be elucidated.^{126, 129, 130}

CSF Markers of Non-AD Dementias

α-Synuclein is the major component of Lewy bodies, the pathologic hallmark of Parkinson's Disease (PD) and Dementia with Lewy Bodies (DLB). α-Synuclein appears to be reduced in CSF of patients with PD and DLB,¹³¹, but α-synuclein measurements have not reliably distinguished DLB from AD.^132-134 Likewise, CSF neurofilament levels appear to be elevated in frontotemporal lobar degeneration (FTLD) compared to early AD, but these measurements have not reliably distinguished between different dementia syndromes.¹³⁵ CSF measurements of TDP-43 have also been of interest given that TDP-43 inclusions are a common abnormality in FTLD; while CSF TDP-43 levels are elevated in FTLD, there is significant overlap with controls.¹³⁶

[Tags: biomarkers; plasma; amyloid; oxidative stress; inflammation; isoprostane; interleukins; growth factors; GFAP; S-100B; lipidomics; ceramides; functional MRI; spectroscopy; diffusion tensor imaging; magnetic resonance perfusion; α-synuclein; dementia with Lewy bodies; neurofilament; frontotemporal dementia; TDP-43; sphingolipids; glycerophospholipids; non-AD dementia; CSF; serum]

Summary

The field of aging and dementia is rapidly evolving with the aim of identifying individuals in the earliest stages. Biomarkers allow the clinician to demonstrate the presence of an underlying pathologic process and resultant synapse dysfunction and neurodegeneration, even in those earliest stages. PET amyloid imaging and CSF Aβ₄₂ provide direct evidence of amyloid deposition and structural MRI, FDG-PET or SPECT and CSF tau provide indirect evidence of synapse dysfunction and neurodegeneration when the pathologic process is due to AD. Structural MRI, FDG-PET or SPECT and PET with dopamine ligands are also valuable in establishing non-AD pathologic processes. While these biomarkers are very useful and can even be applied to diagnostic criteria in MCI, several limitations exist. As the field continues to grow, several new biomarkers are emerging and ultimately, a more biological characterization of subjects’ underlying pathophysiologic spectra will be possible that clinically could translate into earlier and more definitive diagnosis and treatment.

Key Points.

• Biomarkers are measurable, in vivo indicators of a specific disease related pathological process.

• The goal of biomarkers is to provide direct evidence of an underlying pathologic process, provide indirect evidence of a pathologic process through evidence of synapse dysfunction and neurodegeneration, or provide evidence of an alternative pathologic process.

• Amyloid imaging and CSF Aβ₄₂ provide direct evidence of amyloid pathology in Alzheimer's disease (AD).

• Structural MRI, FDG-PET, SPECT and CSF tau can provide indirect evidence of AD as markers of synapse dysfunction and neurodegeneration while certain patterns may suggest pathologic processes other than AD.

• The best CSF marker to predict conversion from mild cognitive impairment (MCI) to dementia is an abnormal ratio of Aβ₄₂ to total tau or abnormal levels of both markers.

• Biomarkers evolve over time and do not reach abnormal or maxima at the same time but are dependent on the stage one is in along the pathophysiologic process.

• Biomarker data can be applied to MCI criteria to improve diagnostic certainty the underlying pathologic process is due to AD.

• Several new biomarkers are in development but have not yet been translated into clinical practice, including plasma Aβ, markers of oxidative stress and inflammation, lipidomics, spectroscopy, functional MRI, diffusion tensor imaging, and magnetic resonance perfusion.