Plasma biomarkers for Alzheimer's disease - Future Medicine

Review For reprint orders, please contact: [email protected]

Plasma biomarkers for Alzheimer’s disease: much needed but tough to find Alzheimer’s disease is a complex age-dependent neurodegenerative disease where definitive diagnosis is only possible after autopsy and where there is a long prodromal or preclinical phase. Biomarkers for both early diagnosis and prediction of disease progression are needed and extensive efforts to discover them have been undertaken. In this article, we have attempted to summarize the findings of current studies using proteomics and metabolomics approaches. We are also discussing how the use of emerging technologies and better study designs can support the identification of the much-needed Alzheimer's disease plasma biomarkers. KEYWORDS: Alzheimer n biomarkers n diagnosis n endophenotypes n metabolomics n plasma n progression n proteomics

Chantal Bazenet1,2 & Simon Lovestone*1,2 King’s College London, Department of Old Age Psychiatry, Institute of Psychiatry, Box PO70, De Crespigny Park, London, SE5 8AF, UK 2 NIHR Biomedical Research Centre for Mental Health, South London & Maudsley NHS Foundation Trust, Denmark Hill, London, SE5 8AZ, UK *Author for correspondence: Tel.: +44 20 7848 5866 Fax: +44 20 7848 0632 [email protected] 1

Alzheimer’s disease (AD) is one of the most common and most debilitating neurodegenerative disorders of the aging population. It manifests itself by a progressive decline in memory and other cognitive abilities and in functional and behavioral decline [1] . It is believed that those clinical symptoms occur 15–20 years after the underlying pathology initiates. This is therefore the single most important opportunity and challenge in the field. An opportunity as an efficacious treatment in the preclinical phase would in effect be a preventative therapy, albeit secondary prevention. This is a challenge as those elderly people in the population with pathology but no clinical manifestation cannot today be identified and selected for clinical trials of such putative agents. For this reason, biomarkers are needed. Identification of biomarkers for AD is therefore an urgent task and quite possibly an essential pre-requisite for the development of diseasemodifying therapies and potentially reducing costs of clinical trials [2] . Unsurprisingly, regulatory bodies have strongly encouraged the inclusion of biomarkers in clinical trials [3,4] . Most efforts conducted so far have focused on finding a diagnostic biomarker, but we need to refine and expand our search. More specifically, new biomarkers are needed for: early diagnosis, ideally even before cognitive symptoms arise (sensitive biomarkers); accurate diagnosis, distinguishing between the different diseases causing dementia (specific biomarkers); and measurement of disease progression within the time of a clinical trial that could measure pathology directly or indirectly (surrogate biomarker) (Figure 1) [5–8] . In addition, biomarkers will need to be reliable

(robust against main confounding factors such as age, gender and genotype), reproducible, easily obtainable, noninvasive, well-validated, showing little variability over several time-points in normal healthy people and little variation due to unrelated comorbidities. It is improbable that a single biomarker will be discovered that meets all these requirements and most likely a panel of biomarkers and/or a combination of different types of biomarkers (proteomics and metabolomics combined or either combined with imaging) as well as biomarkers identified from longitudinal analyses will be needed [9] . Multiple biomarker panels will serve multiple different utilities. In the clinic, biomarkers meeting even some of these taxing requirements would likely find widespread utility. As AD comes to the foreground as public awareness increases and as we approach a potential disease-modifying treatment, clinicians are encountering patients earlier in their disease journey. Diagnosis at this point is exceedingly difficult and clinicians look to biomarkers to aid diagnostic decision-making. However, it is in the context of clinical trials that biomarkers are most urgently needed, first in the early phases of patient recruitment, as they will help identify patients who are in the preclinical stage of AD pathology [5,10–14] . Second, they may provide ways to measure and follow the effects of the drug treatments (novel or pre-existing in early-phase clinical trials or experimental medicine/proof-of-concept studies). Third, if secondary preventative therapeutics were to be found, then very-long-term trials will be required. The challenges of such trials look to be almost

10.2217/BMM.12.48 © 2012 Future Medicine Ltd

Biomarkers Med. (2012) 6(4), 441–454

part of

ISSN 1752-0363

441

Review

Bazenet & Lovestone

Sensitivity Specificity

Biomarker characteristics

Biomarker Early diagnostic utility Progression

Surrogate

Differential diagnostic

Figure 1. Type of biomarkers needed to support clinical trials. Early diagnosis markers need to be very sensitive to identify a patient population before cognitive symptoms arise. Differential diagnostic markers must be very specific as they need to distinguish between the different diseases causing dementia. Markers of disease progression can be less specific, but should be as sensitive as possible to follow subtle cognitive changes within the time of a clinical trial. Surrogate markers need a good balance between sensitivity and specificity, as they will be reflecting a particular aspect of the pathology.

insurmountable at times partly due to the huge costs of a development program that might be longer than compound patent life. Biomarkers provide the possibility of shortening development timelines if they could achieve the status of surrogates for clinical outcomes. Indeed, using such biomarkers to identify and recruit early in the disease process and to assess the effects of putative therapies – especially in experimental medicine and Phase II trials – may be a necessary step on the path towards successful drug development. Up to now, most efforts have focused on measuring changes either directly in the brain or in cerebrospinal fluid (CSF), the fluid that bathes the brain [11] . The biomarkers that are most replicated and developed include CSF markers of core AD pathology including structural MRI (e.g., for hippocampal volume), amyloid PET imaging and FDG PET to examine brain metabolism. Imaging techniques such as structural MRI and amyloid PET are already widely used in clinical trials [5] . CSF biomarkers have also been explored extensively and are used in clinical contexts in some health services as well as in clinical trials. A panel of three markers has now been established: a reduction in Ab42 levels with a concomitant increase in both CSF tau and phospho-tau has been observed in people with AD and other dementias [15–17] . However, these markers have not yet been shown to adequately predict early diagnosis or disease progression, although the data look promising. Although both PET and CSF analyses are very promising markers of AD pathology and show potential application in clinic, they are respectively costly and sparsely available or invasive 442

Biomarkers Med. (2012) 6(4)

procedures. In addition, neither is amenable to repeated sampling, an important aspect of a biomarker’s role and utility. Peripheral sources of biomarkers (plasma, serum, urine and saliva) present considerable advantages (e.g., easy to obtain, cheap and repeated samplings can be performed for regular monitoring), but have the disadvantage of being further away from the brain, the affected organ in AD. This implies peripheral markers will need to be correlated with accepted markers of AD pathology observed both in the brain and CSF to demonstrate that they reflect AD pathology (Figure 2) . In the past 10 years or so, considerable efforts have been spent on identifying susceptibility factors for AD using genomics. These findings have become singularly large and successful, identifying APOE, CLU, PICALM and CR1 as genetic risk factors, for example. However, by definition a genetic variant is a trait marker, and while risk traits are important for identification they cannot be markers of disease state. For a true state marker, other ‘omics are needed – markers of state that are discoverable by proteomics, transcriptomics or metabolomics. From here on, we will focus on plasma as a source for potential biomarkers using both candidate as well as true ‘omics ‘unbiased’ discovery approaches. This review is not intended to be a systematic review of all data pertaining to the develop ment of peripheral biomarkers, but is instead an attempt to summarize what the technologies and study designs used for the identification of plasma biomarkers have delivered and to consider in what direction this field of research might move forwards.

Advances in plasma biomarker identification Plasma is the noncoagulated liquid part of the blood that is devoid of cells and contains mainly proteins, including antibodies, metabolites and circulating RNA. However, the circulating blood, hence plasma, is separated from the brain by the blood–brain barrier (BBB), which restricts the diffusion of microscopic pathogens (e.g., bacteria) and of large or hydrophilic molecules into the CSF. Only, small hydrophobic, lipophilic molecules will more easily transit into the blood, in addition to an active transport of specific molecules and other metabolic products. This implies that the most relevant markers (i.e., those that reflect what is happening in the brain) will be in low abundance in the plasma, at least in normal healthy people (Figur e 2) . However, increasing evidence points towards future science group

Plasma biomarkers for Alzheimer’s disease: much needed but tough to find

a compromised BBB in neurodegeneration and this will impact on the type of molecules leaking out of the brain into the circulation [18] . Nevertheless, plasma represents a rich source of undiscovered information. Changes in plasma proteins, metabolites or even miRNA levels could reflect either already known mechanisms of AD pathology, including neuronal cell death (e.g., cell content has spilled out), inflammation (production and secretion of cytokines and chemokines), oxidative mechanisms and activation of complement cascades [19] or could provide useful insights into novel pathways involved in the disease etiology. Ab is at the center of AD pathology, and as such, variations in its levels have been measured not only in CSF, but also in plasma. Although a plethora of studies have examined Ab (Ab42, 40 and ratios 42:40) as a potential diagnostic or predictive biomarkers, the overall results are contradictory (for reviews, see [20–22]). In some studies, changes in Ab42 but not Ab40 levels were associated with a higher risk of developing late-onset AD [23–25] , whereas others found no particular correlation between Ab levels and late-onset AD risk or implicated Ab40 rather than Ab42 [26,27] . Despite those discrepancies, plasma Ab peptides may still be useful as pharmacodynamic biomarkers rather than a diagnostic test [28] . Recent efforts to discover novel AD plasma biomarkers have been utilizing technologies that offer the possibility of multiplexing (i.e., measuring the concentration of several analytes at once) and thus accelerating the discovery process.

Technologies frequently used in biomarker discovery Proteomics Proteomic technologies have been increasingly used in disease biomarker discovery in general and more recently in the field of neurodegenerative disorders [29] . They are based on various approaches including immunoassays and mass spectrometry (MS).

in one experiment (i.e., multiplex). There are different solid support and detection methods available, some of which are described below. Overall, these technologies are well suited for both targeted (i.e., only measuring putative AD biomarkers in a customized assay) or semitargeted approaches (i.e., measuring levels of multiple proteins for which assays have been developed, albeit not specifically in the context of AD). In the Quantibody ® array (Raybiotech Inc., GA, USA), capture antibodies are arrayed onto a glass support. The sample is then incubated with the array. Alexa Fluor® 555/Streptavidinconjugated antibodies (Molecular Probes, OR, USA) specific to the proteins of interest are added for detection (and thus quantification) by a laser scanner. Using this approach, Ray et al. identified a panel of 18 cytokines that could be used for classification between AD and controls with 90% accuracy [30] . They introduced the concept of a blood signature, arguing that a combination of plasma biomarkers (i.e., a signature) rather than a single marker will be needed to reflect the complexity of AD pathology. Interestingly, these 18 proteins pointed to possible deregulation in immune responses, neuronal support and apoptosis, as well as hematopoiesis. This illustrated another use of biomarker discovery as having potential for identifying new target molecules or pathways involved in AD pathology. The major multiplex proteomics discovery studies (ADNI [Alzheimer’s Disease Neuroimaging Initiative], AIBL [Australian Imaging, Biomarkers Lifestyle Flagship Study of Ageing], TARC [Texas Alzheimer’s Research Consortium] our own AddNeuroMed and associated studies) have used another antibody Compartment Biomarker acquisition

future science group

Brain

Cerebrospinal fluid

Imaging

Lumbar puncture

Blood Blood drawing Tourniquet

L4 Spinal tap L5

Antibody-based detection

Moving beyond assays for single proteins, a number of multiplex, sandwich-based ELISA arrays are now commercially available. Similar to a traditional sandwich-based ELISA, these arrays use a pair of specific antibodies to detect the protein of interest. However, in this case, multiple specific capture antibodies are immobilized onto a solid support, allowing for the detection and quantification of several proteins

Review

Distance to disease Ease of sampling Analyte concentration

Figure 2. Alzheimer’s disease biomarker sources.

www.futuremedicine.com

443

Review

Bazenet & Lovestone

capture multiplex approach – the Luminex xMAP® technology (Luminex, TX, USA) – and thus measured several hundreds of samples. Using this approach, specific capture antibodies are coated onto microspheres that are colorcoded, and up to 500 distinct sets can be combined, with different intensities of red and green dyes. These dyes will confer the microspheres with a unique signature and they can then be identified using a laser. Samples are added to the microspheres and the captured proteins of interest are detected by biotinylated antibodies. Rules Based Medicine Inc. (TX, USA), using the xMAP technology, measured over 141 analytes in the ADNI, TARC and AIBL samples. TARC identified a set of 11 proteins that distinguished AD from controls in both plasma and serum [31–33] , a set that was distinct from the Ray et al. panel [30] . This discrepancy may be partially due to the use of a different platform. However, in both cases, these proteins were all inflammation components that, for most, had already been implicated in AD pathology and may reflect an inflammatory endophenotype. 2D-gel electrophoresis/MS approaches

2D-gel electrophoresis (2D-GE) is a well-established technique where proteins are separated in a first dimension by their isoelectric point (isoelectrofocussing), then by their mass on a SDS-PAGE. This generates a protein fingerprint/map and is particularly useful for examining proteins of medium abundance as they can be somewhat separated from more highly abundant proteins. Differential protein spots are then identified by MS. Although not a very high-throughput method compared with multiplex ELISA arrays, it is closer to the ambitions of proteomics in that it is a so-called ‘unbiased’ approach delivering data on proteins not limited by a priori assay construction. However, as for all proteomics technologies, it is limited in the range of proteins it can assess. In contrast to genomics, where essentially the whole genome can be assayed, 2D-GE, such as arrays and gelfree MS approaches, is only able to assess a small proportion of the proteome and this is limited still further in complex tissues and fluids such as blood. In our hands, 2D-GE ‘sees’ only 300–500 distinct spots from plasma samples, each of which contains one to five proteins. This means that although assessing up to 1000 or so proteins, with an order of magnitude more detailed than the best antibody capture arrays, 2D-GE still only assesses the most abundant fraction of the plasma proteome. 444


In one of the first large-scale studies on plasma in AD, we demonstrated using 2D-GE that differences in the plasma proteome can be observed in AD versus control samples [34] . Changes in the levels of complement factor H (CFH), an innate immunity marker, were the most robust. This work was replicated by several groups [35,36] including a longitudinal study using the ADNI’s proteome dataset [9] . Changes in CFH and CD40 and CFH and adiponectin levels over 12 months showed pair-wise differences longitudinal signatures achieving a sensitivity and specificity close to 90%. Interestingly, a recent study showed that CFH levels could be controlled by miRNA in AD brains [37] . The parallel genomic data also increasingly point to complement biology as being implicated in AD pathogenesis, again suggesting that such biomarker studies might have secondary, but useful, outcomes in identifying potential therapeutic targets in AD. Liquid chromotography-MS/MS approaches: quantitative proteomic profiling using isobaric mass tagging

Multiple variants on MS-based proteomics can be employed in biomarker discovery. Typically, plasma samples are first enzymatically digested using trypsin. The peptides generated are further separated by reverse phase liquid chromatography (LC) and identified by MS/MS. The peptides identified are then matched to unique proteins using appropriate databases. The use of isobaric mass tags in an LC-MS/MS workflow has rendered more robust quantification of peptides/proteins [38,39] . In an example of such an approach, we performed an untargeted discovery using tandem mass tags in a small cohort of AD and control samples, finding that gelsolin levels were significantly decreased and correlated with rates of decline in AD patients [40] . Gelsolin, an actinbinding protein, has antiamyloidogenic, anti oxidant and antiapoptotic properties [41–44] . Brain gelsolin appears to be cleaved in AD brains, in regions of the brains that are pertinent to the disease [42] . This finding further supports the notion that plasma contains markers that are relevant to the disease pathology. Metabolomics Although still a relatively nascent field, metabolomics is being developed and applied as a more global approach to the discovery of biomarkers for neurodegenerative disorders such as motor neurone disease, Parkinson's disease and Huntington's disease (for reviews, see [45–48]). future science group


Most of the metabolomics studies have been performed so far on very small cohort sizes and will need further replication. Several studies have recently demonstrated the involvement of lipids and sterols in AD pathology [49–54] , which is of considerable interest given the proteomics and genomic marker studies demonstrating the critical role of ApoE and ApoJ (also known as clusterin). Metabolomics quantifies small-molecule metabolite changes indicative of alterations in anabolic and catabolic processes in a given person. These changes can be indicative of a disease or a healthy state, or be adaptive responses. Although they can clearly be affected by specific dietary habits, age seemed to be the most confounding factor [55] . Metabolomic research focuses mainly on changes in polar and nonpolar lipids and other small metabolites related to the disease. Metabolites are a very good reflection of genomic and proteomic changes due to a given pathology and constitute the final element of a broader systems biology approach (Figure 2) . An extensive study of the metabolome in AD could lead to the discovery of new pathways implicated in AD. The biochemical pathways behind metabolite changes remain to be identified and correlated with memory loss and cognitive impairment. Multidimensional MS-based shotgun lipidomics

Multidimensional MS-based shotgun lipidomics is a nontargeted, multidimensional, sensitive and quantitative method that has high accuracy, sensitivity and reproducibility [56–58] . Han et al. applied this method in a small prospective study where they examined 26 AD versus 26 controls using this multidimensional MS-based shotgun lipidomics approach [50] . Over 800 metabolites were examined, including choline glycerophospholipid, lysoglycerophospho lipid, ethanolamine glycerophospholipid, phosphatidyli nositol, sphingomyelin (SM), ceramide and triacylglycerol. The authors correlated their findings with diagnosis, APOE genotype and cognitive performance, and identified an AD signature separating cases and controls: a decrease in several SM species (those with particularly long fatty acid chains with 22 or more carbons atoms) concomitant with an increase in two specific ceramide species (N16:0 and N21:0) in AD patients. The ratio of SM/cereamide species, containing the same number of carbon atoms, was also affected in AD patients. These ratios appear to be more discriminatory than the levels of the single species themselves. future science group

Review

This signature also correlated with disease severity and cognitive impairment and was more pronounced in ApoE4 carriers. Although this was a small study that needs repeating, its results corroborate other findings (e.g., [53]). In this recent study, the authors examined changes in plasma lipids with progression of AD. No crosssectional associations were found with changes in SM, ceramide levels and ratios with cognitive measures such as mini-mental state examination, AD assessment scale-cognitive or clinical dementia rating sum. However, baseline levels of SM (high) and ceramide (low) were predictive of slower cognitive decline (as measured by minimental state examination and AD assessment scale-cognitive, but not for clinical dementia rating sum) in longitudinal studies. Global metabolite profiling by ultra performance LC & 2D gas chromatography coupled to time-of-flight MS

Ultra performance LC is another method applied for global lipid profiling, whereas 2D gas chromatography coupled to time-of-flight MS is preferentially used for the measurement of small polar molecules such as amino acids, various organic acids, sterols and sugars [59,60] . Using both platforms, Orešič et al. examined a longitudinal cohort of AD and control samples [61] . They confirmed that AD patients had diminished baseline levels of ether phospholipids, phosphatidylcholines, sphingomyelins and sterols as compared with controls, but did not observe the lipid profile to be a good predictor set for progression to AD. However, they found a distinct progression signature from mild cognitive impairment (MCI) to AD, comprised of three other distinct metabolites, consisting of 2,4-dihydroxybutanoic acid and metabolites on the pentose pathway. As these biomarkers indicated a state of increased hypoxia, this implies the potential involvement of hypoxia in the development of AD pathology. Sterol analysis by LC-MS

A new workflow was established by Sato et al. to measure sterol content in biofluids by LC-MS [54] . The authors found that plasma levels of desmosterol were decreased in AD and MCI patients. In addition, desmosterol and cholesterol ratios significantly differed between controls and MCI or AD cases. Intriguingly, those changes were not affected by ApoE4 genotype and they were more pronounced in female AD patients. Finally, the authors reported that CSF www.futuremedicine.com

445

Review

Bazenet & Lovestone

ratios of desmosterol:cholesterol correlated with the plasma ratios. This study needs repeating and to be expanded into a larger cohort. The above studies have highlighted that: Metabolomic approaches are powerful and present a broad picture of the biochemical events happening in our body; The approaches used (liquid or gas chromatography followed by MS) allow for the measurement of a large number of metabolites at the same time; It is possible to obtain an AD metabolic signature to distinguish cases versus controls. Many other proteins and analytes have been found to show some differences in AD, including but not limited to adiponectin [62] and 3a,5aTHP [63,64] . None of these have yet been sufficiently validated or replicated to be confident that they have biomarker utility, but together, they add considerably to the evidence that there is a peripheral signal worth exploring. We expect many more analytes to be discovered as technologies and study designs improve. The hard task will be to assemble the best combination of analytes into suitably robust assays and validate these in sufficiently large and representative populations.

Emerging technologies in biomarker discovery Plasma is a complex matrix with a very wide dynamic range with over 900 identified proteins to date and almost certainly more to be discovered [65] . In total, 99% of the plasma protein content is composed of very-high-abundance proteins (22 main proteins, among which albumin represents 50% of plasma content), hence 1% of the plasma protein content consists of hundreds of proteins that are present in very low abundance. The presence of high-abundance proteins can thus hinder the measurement of other less abundant proteins. In addition, because AD is a chronic disease, it is reasonable to think that a progressive build-up of changes rather than an acute change in plasma and other fluid biomarkers would be expected. This might translate into subtle changes rather than big changes in biomarker profiles depending on how quickly patients decline. Newly developed technologies should be sought to increase the chances of discovering biomarkers of real utility and might be applied in particular to: go deeper into the plasma proteome; increase multiplexing capabilities; access 446


the full range of analytes; and bridge candidate discovery to validation steps. Diving deeper into the plasma proteome: improved LC-MS workflow The current methods, for the most part, assay only the tip of the proteome iceberg and measure largely high-to-medium-abundance proteins. For more comprehensive exploratory proteomics, future strategies might need to include extensive prefractionation of the plasma followed by high-end proteomic technologies. The recent arrival of, for instance, Orbitrap mass spectrometers provide a next level of performance with increased resolving power and increased mass accuracy. They also support the use of multiple fragmentation techniques, which result in better identification and quantification of peptides, particularly peptides labeled with isobaric tags. One of the most commonly applied and effective strategies is depletion of the high-tomedium-abundance proteins (including albumin and immunoglobulins). Depletion of the most abundant proteins has greatly enhanced the detection of low-abundance proteins in plasma, as shown by Schenk et al. [65] . However, there is concern that such approaches might introduce variability [66,67] , which might limit the application of depletion strategies. In addition, some of the high-abundance proteins have themselves been identified as potential biomarkers (such as gelsolin and CFH). Other possibilities for plasma proteome fractionation include multi dimensional separation first by LC or in isoelectrofocussing solution followed by 1D-gel separation or combined with a second LC dimension. Regardless of the method used, fractionation of plasma yields a higher number of unique proteins to be identified, allowing access to a deeper plasma proteome. However, this will result in an increase in MS/MS time and in the costs of the overall discovery phase. Increased multiplexing capacity Luminex xMAP or other antibody-based technologies are inherently limited in the number of analytes in one single panel. This is because of two main reasons: the problem of antibody crossreactivity and the requirement of different sample dilution factors for different antibody pairs. The development of a new class of DNA-based aptamers that have protein-binding functional groups has tremendously increased the number of proteins that aptamers can now recognize [68] . This new technology offers a wider dynamic range and allows the possibility of measuring hundreds of future science group


proteins over a broad range of concentrations in the same well. Results of the application of this aptamer-based technology to biomarker discovery will be eagerly awaited. Other lesser explored ‘omics Although methodologies to explore a wider range of analytes or to specifically measure posttranslational modifications (PTMs) in plasma matrices are available, they have not yet been applied to biomarker discovery in a systematic manner. For instance, although circulating miRNA molecules have shown great promise as potential cancer biomarkers, their role as potential biomarkers for neurodegenerative disorders has not yet been studied very extensively. This is despite an increasing interest into the investigations of the biological and pathological functions of miRNAs in CNS development, aging and disease states [69–72] . The existence of new array platforms using quantitative reverse transcription-PCR (e.g., Exiqon, Denmark) to analyze and quantify miRNA will facilitate the profiling of large numbers of miRNA molecules in patient plasma samples. In addition, a disease or predisease state can be measured not only by changes in protein levels, but also by changes in PTMs. Specific PTMs may provide information such as activation of specific signaling pathways and point to potential new drug targets. The analysis of PTMs is to some degree still in its infancy and is more challenging in many ways than measuring protein quantity. Nonetheless, a range of approaches have been developed, including: Specific stains can be used in conjunction with 2D-GE (Ruby stains looking at phospho proteins and glycoproteins); Specific enrichment for glycoproteins can be obtained by using lectin (ConA) columns and the isolated glycoproteins identified by MS; Enrichment of phosphopetides using TiO2 purification or by a strong cation exchange columns fractionation followed by LC-MS/MS has led to the identification of 70 phospho proteins in human plasma [73] ; Redox proteomics, which is of particular interest as oxidative stress has been suggested to play a crucial role in AD pathogenesis [74–79] . Not many plasma proteins are oxidized, but interestingly, increases in the levels of oxidation of g-fibrinogen and a-antitrypsin have been observed in AD patients [80–82] . future science group

Review

Bridging discovery & validation steps Platforms used in discovery are not necessarily the platforms used for either further validation/ replication or, more importantly, the platforms chosen for the clinical assays. Discrepancies in findings have been observed, for instance, between an MS-based method and an immuno assay trying to measure the same protein. Immunoassays rely on the availability of at least two very specific antibodies for the detection of a single protein, and so can incur long development periods. In addition, as explained above, they have a limited multiplexing ability. In addition to the aptamer arrays described above, quantitative and targeted MS assays can be developed as alternatives. Keshishian et al. combined multiple reaction monitoring (also called selective reaction monitoring) with the use of stable isotope dilution-MS to establish a robust assay for low abundance plasma proteins [83] . Briefly, signature peptides specific for a given proteins are used to identify the protein of interest while a synthetic, stable isotopelabeled version of those signature peptides are used as an internal standard and allow proper quantification of the biomarker protein.

Study designs For the most part, current studies have concentrated on class prediction where the class is defined as a clinical outcome. A typical study compares cases with controls or people with AD to other neurodegenerative diseases. Increasingly, however, innovation is entering the domain of study design with approaches seeking to use other independent variables to discover biomarkers reflecting pathology. Putative single markers and panels of biomarkers distinguishing cases from controls have already been identified and more are underway, as reviewed above. However, the utility of such classifiers is limited in various ways. The AD population is a highly heterogeneous one: pure AD pathology is most often accompanied by other neurodegenerative pathologies such as vascular changes or synucleinopathy. Studies examining cases versus controls may thus identify markers that are not due to the primary disease. Another caveat is that the normal population is also composed of individuals who may have biomarkers of AD pathology without the clinical manifestation of AD because of the long prodromal period. Finally, a possibly even more heterogeneous group is the MCI population, which is likely to be composed of at least three subsets: MCI patients who convert into AD, those who www.futuremedicine.com

447

Review

Bazenet & Lovestone

convert into non-AD pathology or those who never convert. Alternative approaches to discover biomarkers with a trial design targeted to the desired outcome are being increasingly used; here we give a few examples in the search for biomarkers of pathology and progression, recognizing that there are others. Markers of AD pathology Recent studies have aimed to identify plasma markers associated with quantitative surrogate markers of AD pathology such as brain atrophy. An early example of such a study using 2D-GE and LC-MS/MS led to the identification of seven proteins associated with hippocampal volume [84] . Replication studies using ELISA-based assays and whole-brain atrophy to overcome issues with manually estimating hippocampal volumes confirmed five of the seven proteins that can predict disease severity as measured by whole-brain atrophy. Indeed, complement C3 and C3a, complement factor-I, a-1 microglobulin and g-fibrinogen, together with age and sex, explained more than 35% of the variance in brain volume [84] . This implies that they should be strong predictors of AD pathology in in vivo situations. These findings corroborate nicely with the fact that complement markers have also been found in CSF that correlate with AD diagnosis and act as predictors of conversion from MCI to AD. Inflammation may increase the speed or rate of disease progression. Markers of progression An alternative approach is to use disease progression as a quantitative trait variable with the assumption that more rapid progression results from more substantial pathology. Using this approach, with rate of cognitive decline as a measure of progression, we found that clusterin levels were indeed associated with severity of AD pathology and progression [85] . We further demonstrated clusterin levels correlated with longitudinal brain atrophy in MCI subjects [86] . These findings suggest that peripheral plasma clusterin is a marker of central brain pathology. Clusterin is of particular interest as genome-wide association studies showed that a single-nucleotide polymorphism in the clusterin genes was significantly associated with an increased risk of developing AD [87,88] . Further hippocampal function was altered in healthy carriers of this risk variant of CLU [89] . Other common and rare nonsynonymous substitutions in CLU have now been uncovered and correlated with increased 448


AD risk [90] . The effects of those variants on clusterin protein level and function remain to be elucidated. A variant of the study design for markers of progression in AD is the search for markers of conversion from predementia syndromes such as MCI to full dementia. The study by Ray et al. was one such study of this design, although the proteins acting as markers in this study have not been replicated widely [30] . Nonetheless, we and others have confirmed an inflammatory signature in plasma in AD, and in our study, a set of cytokines and related molecules accurately predicted conversion from MCI when combined with imaging [91] .

Ongoing challenges Using proteomics and metabolomics in typical or increasingly sophisticated study designs, it is clear that there is a signature of disease in plasma. However, to turn this discovery into a validated biomarker for clinical trial utility remains a very significant task with some considerable challenges. Standardization of sample collection & analyte measurements Preanalytical factors such as specimen collection, processing and storage can affect the analyte (proteins, metabolites and miRNA) integrity and thus their levels. Indeed, levels of the various components (especially metabolites) of the plasma can be easily affected by diet, medication, handling of samples and other environmental factors, and even gender [92] . Therefore, it is important to develop standardized protocols for plasma collection and of assays used for appropriate and meaningful comparison across studies, especially during the replication and validation phases. This is an issue not only for plasma, but also for CSF and imaging biomarkers. Several initiatives have been put in place worldwide to address those challenges [93,94] ; the National Institute of Aging has published best practice guidelines for specimen collection and storage for AD centers [101] . A globalization effort for harmonization and standardization of biomarkers (plasma, CSF or imaging) is underway with the creation of a global consortium by the Alzheimer’s Association. One small caveat to this important and collaborative effort is that standardization is an unambiguously important process but is best applied after and not before biomarker discovery. Before biomarkers are discovered, standardization of future science group


sample collection and processing, for example, runs the risk of standardizing to the ‘wrong’ standard. A better approach, in the very early phases of research, is collection of data pertaining to sample collection and processing (time of day, relation to last meal and time from venepuncture to storage, among others) and then to use post hoc analysis to identify those factors that influence biomarker outcomes. What is important is to be systematic and consistent with standard operating protocols within a study and to publish and share these data between studies so that potential confounds can be identified. After biomarkers are discovered, then standardization becomes critical – as it currently is for CSF markers. Challenges inherent to AD The quest for AD plasma biomarkers is complicated by the fact that the rate of the progressive build-up of the pathology most likely varies from one individual to another and there is no precise pattern of disease onset across the AD population. The problem caused by this lack of synchronicity causes large variability and can be partly addressed by examining large sample cohorts. This will also partly resolve the issue of a specific diagnostic for AD; indeed, the final common pattern observed among a large population with a common primary AD pathology but with different comorbidities should reflect the pathology they all have in common (i.e., AD). Yang et al. have addressed the lack of synchronicity by developing a bioinformatic model that replicated the evolution of the pathological symptoms [95] . They then established a new timeline of disease progression, which they used to synchronize CSF biomarkers and brain imaging measures. These new synchronized measures allowed them to place CSF and imaging biomarkers temporally along this more accurate progression timeline of the disease. Study replications & data analyses A number of potential biomarkers have been reported, but few have been sufficiently validated or replicated: not all studies consist of separate training and test sets of adequate size, while other studies are more descriptive and performed on a single set. Therefore, independent replications are critical for the establishment of biomarkers. Very recently, two groups independently failed to replicate the original findings of Ray et al. [30] . Marksteiner et al. confirmed future science group

Review

that the levels of only five of those proteins were increased in MCI and AD patients as compared with controls [96] . A further replication attempt showed that only three of those proteins differed significantly between AD and controls [97] . There are a number of reasons for the lack of replication: use of small sample cohorts, different technology, type of data processing and analyses performed or even possibly overtraining of the dataset. In addition, not all studies include internal validation in an independent test set. The choice of analyses performed on a given dataset may lead to different outcomes. For illustration, de Paula et al. performed a different analysis of the original dataset from Ray et al. [30] and uncovered both common and new biomarkers [98] . One of the challenges in replicating findings of complex multivariate analyses of studies is that data are processed in various ways, including the use of z-scores, normalization by other variables and other parsing of the data. Open access to raw data for ana lysis by others would facilitate mitigating these problems. Furthermore, reporting in detail on standards and procedures used and information on what could be potential cofounders (e.g., time of day and relationship to last meal) should be reported more systematically and could help decrease reproducibility issues. Combination of markers across various ‘omics platforms: systems biology approach The various types of plasma biomarkers (proteins, metabolites and miRNA) can be combined and/or correlated with CSF markers, brain imaging and genomics information (F igur e 3) . Indeed, this will lead to the development of a more complex but more complete fingerprint of prodromal AD and to a better understanding of how it differs from other neurodegenerative diseases. However, incorporation of imaging may prove problematic as often the available sample sets are small, and this will reduce the significance of the findings. Finally, powerful and innovative bioinformatics tools will be needed to unravel multivariate plasma signatures. Sample collections & bioinformatics analyses The use of very large variable approaches – the ‘omics – is both essential but also problematic in the development of biomarkers for AD and other disorders. Increasing numbers of analytes can be measured in one single plasma sample, www.futuremedicine.com

449

Review

Bazenet & Lovestone

Genome

Environment

Health/disease

Neurobiological networks

Integration of multimodal biomarkers Brain imaging, Alzheimer’s disease CSF/blood ‘omics subtypes Biomarkers

Figure 3. Systems biology approach. Neurobiological networks are affected by genetic make-up and environment. Perturbation of those networks can lead to neurodegeneration. Integration of multimodal biomarkers defines Alzheimer’s disease and its subtypes. CSF: Cerebrospinal fluid.

so the number of predictors is thus often far greater than the number of samples. This can result in overfitting the data and may be an important cause of nonreplication of findings. One approach to compensating for this source of error is to simply increase the sample size. To do so, substantially more samples will need to be available for analysis. Shared access to samples collected in research cohorts around the world would be extremely useful and speed up progress in biomarker discovery. Sharing the burden of not only collecting samples, but also collecting serial clinical measures, brain atrophy evaluations and CSF biomarker information, would facilitate this. Other developments that would enhance biomarker discovery would be improved mathematical and computational tools to reduce data overfitting and to better handle cohort hetero geneity (different disease progression and different disease stages) and integration of multimodal analytes, and then to interpret these using bioinformatics as well as developing biomarker algorithms for use in the clinic.

Future perspective The AD field is at a time of reflection: a large number of clinical trials of anti-amyloid therapies have failed and very few new therapies are emerging. The challenge in developing novel therapeutics appears ever more substantial and both academic 450


and pharmaceutical groups are becoming increasingly concerned, with much thought being given to how to carry out disease-modifying trials in this complicated, chronic but critical disease area. Biomarkers may provide one part of the solution to enhance, or perhaps even permit, the development of disease-modifying therapies. The big opportunity and equally the big challenge in AD is the very long preclinical disease period, lasting perhaps one or two decades before the onset of symptoms. The opportunity is to treat people in this phase when therapies might be most effective. The big challenge is to identify people in this preclinical disease phase. Biomarkers are, by definition, the only way to do this. A second major challenge is measuring pathology in an organ that is inaccessible to conventional medical tools of direct access by surgery or the clinical skills of medical examination. It is a huge challenge to embark upon clinical trials without the ability to directly monitor the effect of therapies on the target organ. Again, biomarkers are necessary to do this in AD and related dementias. A third challenge is to select appropriate patients for trials. Trial participants are needed who have specific dementia diseases, for example, and who are predicted to progress as another form of dementia. Yet again, biomarkers are needed to enable such recruitment and stratification in trials. In all these ways, and in others, biomarkers would considerably facilitate the discovery of novel therapeutics for AD. Biomarkers for diagnosis, pathology estimation and prediction are all needed. Great progress is being made in relatively specific markers using CSF and molecular imaging approaches. MRI and other approaches not discussed here, such as EEG and functional MRI, may all be useful as part of the biomarker armamentarium. However, increasing attention is being paid to blood-based biomarkers. When this work started, the a priori justification for blood-based biomarkers was low because of the BBB and the complexity of blood; both factors contributing to the difficulty in finding a bloodbased marker. However, work over the past 6 years or so has convinced many that there is a blood-based biomarker signature waiting to be discovered, be it with proteins, metabolites, RNA or other approaches alone or in combination. In order to translate the increasing optimism that there is a biomarker signature in blood waiting to be discovered, a number of developments are required. These developments include increasing research but, in particular, larger, future science group


well-controlled, well-designed studies with better replication and validation of data. To support this, we recommend efforts are made to: Enhance and incentivize collaboration and sharing of data and samples between academic groups and with industry; Expand and better assess cohorts for biomarker discovery. Ensure cohorts are available that reflect disease in the community, noting the sometimes atypical cohorts that enter clinical trials and some research programs; Continue to innovate and to improve in technologies (the ‘omics) relevant to biomarker discovery; Develop and improve approaches to analyze ever-larger datasets by including collaborative groups with IT and mathematics solutions; Pay attention to the full pathway of biomarker development from discovery, through validation to qualification and regulatory approval. Paying attention includes attention to intellectual property and business issues (who pays? who profits?), recognizing that the requirements of large pharmaceutical companies, small biotechnology and academic groups are different; Consider innovation in biomarker trial design to be as important as technology innovation.

Review

With these developments and no doubt others, there is considerable reason to be optimistic that blood-based biomarkers are plausible and that progress has been made towards discovering such biomarkers. Although it is clear that there is no biomarker yet that fulfills all the requirements of a biomarker in AD, let alone a bloodbased biomarker, the next 5 years should see blood-based markers come to the forefront and find their place alongside other approaches in clinical trials and in the clinical care of patients with AD. Financial & competing interests disclosure King's College London has intellectual property interests in biomarkers related to Alzheimer’s disease. In the last 5 years, S Lovestone has provided consultancy to AstraZeneca, GEHC, Noscira and Pfizer. S Lovestone and C Bazenet have shared research projects including cofunding with AstraZeneca and Proteome Sciences and have an EU IMI grant pending, which includes Alzheimer's disease biomarkers with more than ten pharmaceutical companies and multiple small and medium enterprises. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. No writing assistance was utilized in the production of this manuscript.

Executive summary Plasma biomarkers for AD Alzheimer’s disease (AD) is the most common form of dementia, affecting nearly 50% of the population over 80 years old. Robust and reliable biomarkers are needed to support decision-making at various stages of clinical assessment and during clinical trials; – Sensitive biomarkers for early diagnostic and specific biomarkers to distinguish between AD and other forms of dementia – biomarkers reflecting pathology; – Biomarkers to predict and to follow disease progression during the relatively short time frame of AD clinical trials; – Surrogate markers in clinical trials. Genetic variants are important risk traits but cannot be markers of disease state. Other markers of disease state discoverable by proteomics, transcriptomics or metabolomics are needed for a complete disease state marker. Advances in plasma biomarker identification Plasma is a rich source of analytes. It is easy to obtain, cheap and repeated samplings can be performed for regular monitoring. However, it is further away from the brain, the affected organ in Alzheimer’s disease. Emerging technologies in biomarker discovery Novel applications of recently improved technologies address the need for increased multiplexing and for exploring plasma proteomics and metabolomics in more depth. Study design Moving away from classical case versus control studies and using other endophenotypes will identify novel sets of biomarkers that will map out the various stages and aspects of AD. Conclusion Although considerable progress has been achieved and a number of panels of protein or metabolite biomarkers have been discovered, their validation will be an important next step. The concept of a panel of biomarkers has now evolved into a panel of multimodal biomarkers that will better reflect the complexity of AD. However, this will bring an additional level of difficulty, as the complex algorithms needed to handle the various datasets have yet to be developed.



451

Review

Bazenet & Lovestone

Alzheimer clinical trials: an overview. Neurobiol. Aging 32(Suppl. 1), S1–S3 (2011).

References Papers of special note have been highlighted as: n of interest nn of considerable interest 1

2

3

4

5

6

15

Thies W, Bleiler L. 2012 Alzheimer’s disease facts and figures. Alzheimer's Demen. 8, 131–168 (2012). Van Rossum IA, Vos S, Handels R, Visser PJ. Biomarkers as predictors for conversion from mild cognitive impairment to Alzheimer-type dementia: implications for trial design. J. Alzheimers Dis. 20(3), 881–891 (2010). Hampel H, Frank R, Broich K et al. Biomarkers for Alzheimer’s disease: academic, industry and regulatory perspectives. Nat. Rev. Drug Discov. 9(7), 560–574 (2010). Broich K, Weiergraber M, Hampel H. Biomarkers in clinical trials for neurodegenerative diseases: regulatory perspectives and requirements. Prog. Neurobiol. 95(4), 498–500 (2011). Cedazo-Minguez A, Winblad B. Biomarkers for Alzheimer’s disease and other forms of dementia: clinical needs, limitations and future aspects. Exp. Gerontol. 45(1), 5–14 (2010). Gomez Ravetti M, Rosso OA, Berretta R, Moscato P. Uncovering molecular biomarkers that correlate cognitive decline with the changes of hippocampus’ gene expression profiles in Alzheimer’s disease. PLoS One 5(4), e10153 (2010).

16

17

19

Querfurth HW, LaFerla FM. Alzheimer’s disease. N. Engl. J. Med. 362(4), 329–344 (2010).

8

Ward M. Biomarkers for Alzheimer’s disease. Expert Rev. Mol. Diagn. 7(5), 635–646 (2007).

21

9

Johnstone D, Milward EA, Berretta R, Moscato P. Multivariate protein signatures of pre-clinical Alzheimer’s disease in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) plasma proteome dataset. PLoS One 7(4), e34341 (2012).

Morris JC, Selkoe DJ. Recommendations for the incorporation of biomarkers into

452

Proteomic research in psychiatry. J. Psychopharmacol. 25(2), 151–196 (2011). nn

Comprehensive review of proteomics approaches and their utility in psychiatric research.

30 Ray S, Britschgi M, Herbert C et al.

Classification and prediction of clinical Alzheimer’s diagnosis based on plasma signaling proteins. Nat. Med. 13(11), 1359–1362 (2007). 31

O’Bryant SE, Xiao G, Barber R et al. A serum protein-based algorithm for the detection of Alzheimer disease. Arch. Neurol. 67(9), 1077–1081 (2010).

33 O’Bryant SE, Xiao G, Barber R et al.

Rissman RA, Trojanowski JQ, Shaw LM, Aisen PS. Longitudinal plasma amyloid beta as a biomarker of Alzheimer’s disease. J. Neural. Transm. 119(7), 843–850 (2012).

A blood-based screening tool for Alzheimer’s disease that spans serum and plasma: findings from TARC and ADNI. PLoS One 6(12), e28092 (2011). A blood-based algorithm for the detection of Alzheimer’s disease. Dement. Geriatr. Cogn. Disord. 32(1), 55–62 (2011). 34 Hye A, Lynham S, Thambisetty M et al.

Proteome-based plasma biomarkers for Alzheimer’s disease. Brain 129(Pt 11), 3042–3050 (2006).

22 Mayeux R, Schupf N. Blood-based

biomarkers for Alzheimer’s disease: plasma Abeta40 and Abeta42, and genetic variants. Neurobiol. Aging 32(Suppl. 1), S10–S19 (2011).

24 Schupf N, Tang MX, Fukuyama H et al.

markers of amyloid beta-related mechanisms in Alzheimer’s disease. Exp. Neurol. 223(2), 334–346 (2010).

29 Taurines R, Dudley E, Grassl J et al.

Very good systematic review and analysis of the role and utility of plasma b-amyloid as a biomarker.

Blennow K, Hampel H, Weiner M, Zetterberg H. Cerebrospinal fluid and plasma biomarkers in Alzheimer disease. Nat. Rev. Neurol. 6(3), 131–144 (2010).

13 Hampel H, Shen Y, Walsh DM et al. Biological

enzyme-linked immunosorbent assay for measuring beta-amyloid protein oligomers in human plasma and brain tissue of patients with Alzheimer disease. Arch. Neurol. 66(2), 190–199 (2009).

32 O’Bryant SE, Xiao G, Barber R et al.

23 Ertekin-Taner N, Younkin LH, Yager DM

harmful to recruitment and retention in Alzheimer’s disease clinical trials? An international perspective. J. Nutr. Health Aging 16(4), 346–348 (2012).

28 Xia W, Yang T, Shankar G et al. A specific

Selkoe DJ, Grodstein F. Plasma amyloid-beta as a predictor of dementia and cognitive decline: a systematic review and metaanalysis. Arch. Neurol. 7(4 Suppl.), S318 (2012).

Aisen PS. Clinical trial methodologies for disease-modifying therapeutic approaches. Neurobiol. Aging 32(Suppl. 1), S64–S66 (2011).

12 Hampel H, Prvulovic D. Are biomarkers

Plasma beta amyloid and the risk of Alzheimer disease and dementia in elderly men: a prospective, population-based cohort study. Arch. Neurol. 65(2), 256–263 (2008).

20 Koyama A, Okereke OI, Yang T, Blacker D,

n

14

Mulder C, Verwey NA, Van Der Flier WM et al. Amyloid-beta(1–42), total tau, and phosphorylated tau as cerebrospinal fluid biomarkers for the diagnosis of Alzheimer disease. Clin. Chem. 56(2), 248–253 (2010). Bell RD, Zlokovic BV. Neurovascular mechanisms and blood–brain barrier disorder in Alzheimer’s disease. Acta Neuropathol. 118(1), 103–113 (2009).

Hampel H, Wilcock G, Andrieu S et al. Biomarkers for Alzheimer’s disease therapeutic trials. Prog. Neurobiol. 95(4), 579–593 (2011).

11

Hansson O, Zetterberg H, Buchhave P, Londos E, Blennow K, Minthon L. Association between CSF biomarkers and incipient Alzheimer’s disease in patients with mild cognitive impairment: a follow-up study. Lancet Neurol. 5(3), 228–234 (2006).

18

7

10

Hampel H, Blennow K, Shaw LM, Hoessler YC, Zetterberg H, Trojanowski JQ. Total and phosphorylated tau protein as biological markers of Alzheimer’s disease. Exp. Gerontol. 45(1), 30–40 (2010).

27 Sundelof J, Giedraitis V, Irizarry MC et al.

et al. Plasma amyloid beta protein is elevated in late-onset Alzheimer disease families. Neurology 70(8), 596–606 (2008). Peripheral Abeta subspecies as risk biomarkers of Alzheimer’s disease. Proc. Natl Acad. Sci. USA 105(37), 14052–14057 (2008). 25 Assini A, Cammarata S, Vitali A et al. Plasma

levels of amyloid beta-protein 42 are increased in women with mild cognitive impairment. Neurology 63(5), 828–831 (2004). 26 Fukumoto H, Tennis M, Locascio JJ, Hyman

BT, Growdon JH, Irizarry MC. Age but not diagnosis is the main predictor of plasma amyloid beta-protein levels. Arch. Neurol. 60(7), 958–964 (2003).


nn

35

First large plasma biomarker study. Akuffo EL, Davis JB, Fox SM et al. The discovery and early validation of novel plasma biomarkers in mild-to-moderate Alzheimer’s disease patients responding to treatment with rosiglitazone. Biomarkers 13(6), 618–636 (2008).

36 Cutler P, Akuffo EL, Bodnar WM et al.

Proteomic identification and early validation of complement 1 inhibitor and pigment epithelium-derived factor: two novel biomarkers of Alzheimer’s disease in human plasma. Proteomics Clin. Appl. 2(4), 467–477 (2008). 37 Lukiw WJ, Alexandrov PN. Regulation of

complement factor H (CFH) by multiple miRNAs in Alzheimer’s disease (AD) brain. Mol. Neurobiol. doi:10.1007/s12035-0128234-4 (2012) (Epub ahead of print). 38 Shiio Y, Aebersold R. Quantitative proteome

analysis using isotope-coded affinity tags and



mass spectrometry. Nat. Protoc. 1(1), 139–145 (2006).

52

39 Thompson A, Schafer J, Kuhn K et al. Tandem

mass tags: a novel quantification strategy for comparative analysis of complex protein mixtures by MS/MS. Anal. Chem. 75(8), 1895–1904 (2003).

Plasma sphingomyelins are associated with cognitive progression in Alzheimer’s disease. J. Alzheimers Dis. 27(2), 259–269 (2011). Aoshima K, Oda Y. Identification of a new plasma biomarker of Alzheimer’s disease using metabolomics technology. J. Lipid Res. 53(3), 567–576 (2012). 55

42 Ji L, Chauhan A, Wegiel J, Essa MM, Chauhan

V. Gelsolin is proteolytically cleaved in the brains of individuals with Alzheimer’s disease. J. Alzheimers Dis. 18(1), 105–111 (2009).

45

Barba I, Fernandez-Montesinos R, GarciaDorado D, Pozo D. Alzheimer’s disease beyond the genomic era: nuclear magnetic resonance (NMR) spectroscopy-based metabolomics. J. Cell. Mol. Med. 12(5A), 1477–1485 (2008).

electrospray ionization mass spectrometric analysis and quantitation of cellular lipidomes directly from crude extracts of biological samples. Mass Spectrom. Rev. 24(3), 367–412 (2005). 58 Yang K, Cheng H, Gross RW, Han X.

Automated lipid identification and quantification by multidimensional mass spectrometry-based shotgun lipidomics. Anal. Chem. 81(11), 4356–4368 (2009). Hyotylainen T. Data analysis tool for comprehensive two-dimensional gas chromatography/time-of-flight mass spectrometry. Anal. Chem. 83(8), 3058–3067 (2011).

Metabolomics: a global biochemical approach to the study of central nervous system diseases. Neuropsychopharmacology 34(1), 173–186 (2009).

Hyotylainen T, Oresic M. Liquid chromatography-mass spectrometry (LC-MS)-based lipidomics for studies of body fluids and tissues. Methods Mol. Biol. 708, 247–257 (2011).

48 Whiley L, Legido-Quigley C. Current

49 Greenberg N, Grassano A, Thambisetty M,

Lovestone S, Legido-Quigley C. A proposed metabolic strategy for monitoring disease progression in Alzheimer’s disease. Electrophoresis 30(7), 1235–1239 (2009). 50 Han X, Rozen S, Boyle SH et al. Metabolomics

51

61

based multiplexed proteomic technology for biomarker discovery. PLoS One 5(12), e15004 (2010). 69 Hebert SS, De Strooper B. Alterations of the

microRNA network cause neurodegenerative disease. Trends Neurosci. 32(4), 199–206 (2009). 70 Hebert SS, Horre K, Nicolai L et al.

MicroRNA regulation of Alzheimer’s Amyloid precursor protein expression. Neurobiol. Dis. 33(3), 422–428 (2009). 71 Lukiw WJ. Micro-RNA speciation in fetal,

adult and Alzheimer’s disease hippocampus. Neuroreport 18(3), 297–300 (2007). 72 Lukiw WJ, Dua P, Pogue AI, Eicken C, Hill

JM. Upregulation of micro RNA-146a (miRNA-146a), a marker for inflammatory neurodegeneration, in sporadic Creutzfeldt– Jakob disease (sCJD) and Gerstmann– Straussler–Scheinker (GSS) syndrome. J. Toxicol. Environ. Health A 74(22–24), 1460–1468 (2011). 73 Carrascal M, Gay M, Ovelleiro D, Casas V,

Gelpi E, Abian J. Characterization of the human plasma phosphoproteome using linear ion trap mass spectrometry and multiple search engines. J. Proteome Res. 9(2), 876–884 (2010).

Orešič M, Hyötyläinen T, Herukka SK et al. Metabolome in progression to Alzheimer’s disease. Transl. Psych. 1(12), e57 (2011).

62 Van Himbergen TM, Beiser AS, Ai M et al.

Biomarkers for insulin resistance and inflammation and the risk for all-cause dementia and Alzheimer disease: results from the Framingham Heart Study. Arch. Neurol. 69(5), 594–600 (2012).

in early Alzheimer’s disease: identification of altered plasma sphingolipidome using shotgun lipidomics. PLoS One 6(7), e21643 (2011).

63 Marx CE, Trost WT, Shampine LJ et al.

Li NJ, Liu WT, Li W et al. Plasma metabolic profiling of Alzheimer’s disease by liquid chromatography/mass spectrometry. Clin. Biochem. 43(12), 992–997 (2010).

64 Smith CD, Wekstein DR, Markesbery WR,


68 Gold L, Ayers D, Bertino J et al. Aptamer-

60 Nygren H, Seppanen-Laakso T, Castillo S,

Metabolomics tools for identifying biomarkers for neuropsychiatric diseases. Neurobiol. Dis. 35(2), 165–176 (2009). strategies in the discovery of small-molecule biomarkers for Alzheimer’s disease. Bioanalysis 3(10), 1121–1142 (2011).

A systematic analysis of the effects of increasing degrees of serum immunodepletion in terms of depth of coverage and other key aspects in top-down and bottom-up proteomic analyses. Proteomics 11(11), 2222–2235 (2011).

59 Castillo S, Mattila I, Miettinen J, Oresic M,

46 Kaddurah-Daouk R, Krishnan KR.

47 Quinones MP, Kaddurah-Daouk R.

67 Smith MP, Wood SL, Zougman A et al.

57 Han X, Gross RW. Shotgun lipidomics:

44 Chauhan V, Ji L, Chauhan A. Anti-

amyloidogenic, anti-oxidant and anti-apoptotic role of gelsolin in Alzheimer’s disease. Biogerontology 9(6), 381–389 (2008).

Enhanced detection of low abundance human plasma proteins using a tandem IgY12SuperMix immunoaffinity separation strategy. Mol. Cell. Proteomics 7(10), 1963–1973 (2008).

Lawton KA, Berger A, Mitchell M et al. Analysis of the adult human plasma metabolome. Pharmacogenomics 9(4), 383–397 (2008). multidimensional MS analysis of cellular lipidomes. Expert Rev. Proteomics 2(2), 253–264 (2005).

The neurosteroid allopregnanolone is reduced in prefrontal cortex in Alzheimer’s disease. Biol. Psychiatry 60(12), 1287–1294 (2006). Frye CA. 3alpha,5alpha-THP: a potential plasma neurosteroid biomarker in Alzheimer’s


Schenk S, Schoenhals GJ, De Souza G, Mann M. A high confidence, manually validated human blood plasma protein reference set. BMC Med. Genomics 1, 41 (2008).

66 Qian WJ, Kaleta DT, Petritis BO et al.

56 Han X, Gross RW. Shotgun lipidomics:

43 Hirko AC, Meyer EM, King MA, Hughes JA.

Peripheral transgene expression of plasma gelsolin reduces amyloid in transgenic mouse models of Alzheimer’s disease. Mol. Ther. 15(9), 1623–1629 (2007).

65

54 Sato Y, Suzuki I, Nakamura T, Bernier F,

41 Antequera D, Vargas T, Ugalde C et al.

Cytoplasmic gelsolin increases mitochondrial activity and reduces Abeta burden in a mouse model of Alzheimer’s disease. Neurobiol. Dis. 36(1), 42–50 (2009).

disease and perhaps non-Alzheimer’s dementia. Psychopharmacology (Berl.) 186(3), 481–485 (2006).

53 Mielke MM, Haughey NJ, Bandaru VV et al.

40 Guntert A, Campbell J, Saleem M et al. Plasma

gelsolin is decreased and correlates with rate of decline in Alzheimer’s disease. J. Alzheimers Dis. 21(2), 585–596 (2010).

Mielke MM, Bandaru VV, Haughey NJ, Rabins PV, Lyketsos CG, Carlson MC. Serum sphingomyelins and ceramides are early predictors of memory impairment. Neurobiol. Aging 31(1), 17–24 (2010).

Review

74

Butterfield DA. Oxidative stress in neurodegenerative disorders. Antioxid. Redox Signal. 8(11–12), 1971–1973 (2006).

75 Butterfield DA, Gnjec A, Poon HF et al.

Redox proteomics identification of oxidatively modified brain proteins in inherited Alzheimer’s disease: an initial assessment. J. Alzheimers Dis. 10(4), 391–397 (2006). 76 Butterfield DA, Perluigi M, Sultana R.

Oxidative stress in Alzheimer’s disease brain: new insights from redox proteomics. Eur. J. Pharmacol. 545(1), 39–50 (2006).

453

Review

Bazenet & Lovestone

77 Pratico D. Evidence of oxidative stress in

86 Thambisetty M, An Y, Kinsey A et al. Plasma

Alzheimer’s disease brain and antioxidant therapy: lights and shadows. Ann. NY Acad. Sci. 1147, 70–78 (2008).

clusterin concentration is associated with longitudinal brain atrophy in mild cognitive impairment. Neuroimage 59(1), 212–217 (2012).

78 Pratico D. Oxidative stress hypothesis in

Alzheimer’s disease: a reappraisal. Trends Pharmacol. Sci. 29(12), 609–615 (2008).

Genome-wide association study identifies variants at CLU and PICALM associated with Alzheimer’s disease. Nat. Genet. 41(10), 1088–1093 (2009).

in Alzheimer’s disease. Int. J. Biochem. Cell Biol. 37(2), 289–305 (2005).

Di Domenico F, Coccia R, Butterfield DA, Perluigi M. Circulating biomarkers of protein oxidation for Alzheimer disease: expectations within limits. Biochim. Biophys. Acta 1814(12), 1785–1795 (2011).

82 Yu HL, Chertkow HM, Bergman H, Schipper

Genome-wide association study identifies variants at CLU and CR1 associated with Alzheimer’s disease. Nat. Genet. 41(10), 1094–1099 (2009). 89 Erk S, Meyer-Lindenberg A, Opitz Von

Boberfeld C et al. Hippocampal function in healthy carriers of the CLU Alzheimer’s disease risk variant. J. Neurosci. 31(49), 18180–18184 (2011). 90 Bettens K, Brouwers N, Engelborghs S et al.

Both common variations and rare nonsynonymous substitutions and small insertion/deletions in CLU are associated with increased Alzheimer risk. Mol. Neurodegener. 7, 3 (2012).

HM. Aberrant profiles of native and oxidized glycoproteins in Alzheimer plasma. Proteomics 3(11), 2240–2248 (2003). 83 Keshishian H, Addona T, Burgess M, Kuhn E,

Carr SA. Quantitative, multiplexed assays for low abundance proteins in plasma by targeted mass spectrometry and stable isotope dilution. Mol. Cell. Proteomics 6(12), 2212–2229 (2007). 84 Thambisetty M, Simmons A, Hye A et al.

Plasma biomarkers of brain atrophy in Alzheimer’s disease. PLoS One 6(12), e28527 (2011). 85 Thambisetty M, Simmons A, Velayudhan L

et al. Association of plasma clusterin concentration with severity, pathology, and progression in Alzheimer disease. Arch. Gen. Psychiatry 67(7), 739–748 (2010).

454

preparation for serum/plasma profiling and biomarker identification by mass spectrometry. J. Chromatogr. A 1153(1–2), 259–276 (2007). 95 Yang E, Farnum M, Lobanov V et al.

Quantifying the pathophysiological timeline of Alzheimer’s disease. J. Alzheimers Dis. 26(4), 745–753 (2011).

88 Lambert JC, Heath S, Even G et al.

CC, Gracy RW. Identification of oxidized plasma proteins in Alzheimer’s disease. Biochem. Biophys. Res. Commun. 293(5), 1566–1570 (2002). 81

94 Luque-Garcia JL, Neubert TA. Sample

87 Harold D, Abraham R, Hollingworth P et al.

79 Tuppo EE, Arias HR. The role of inflammation

80 Choi J, Malakowsky CA, Talent JM, Conrad

of cerebrospinal fluid collection and biobanking. Neurology 73(22), 1914–1922 (2009).

91

Furney SJ, Kronenberg D, Simmons A et al. Combinatorial markers of mild cognitive impairment conversion to Alzheimer’s disease – cytokines and MRI measures together predict disease progression. J. Alzheimers Dis. 26(Suppl. 3), 395–405 (2011).

92 Silliman CC, Dzieciatkowska M, Moore EE

et al. Proteomic analyses of human plasma: Venus versus Mars. Transfusion 52(2), 417–424 (2012). 93 Teunissen CE, Petzold A, Bennett JL et al.

n

Describes a synchronization model to predict disease stages.

96 Marksteiner J, Kemmler G, Weiss EM et al.

Five out of 16 plasma signaling proteins are enhanced in plasma of patients with mild cognitive impairment and Alzheimer’s disease. Neurobiol. Aging 32(3), 539–540 (2011). 97 Bjorkqvist M, Ohlsson M, Minthon L,

Hansson O. Evaluation of a previously suggested plasma biomarker panel to identify Alzheimer’s disease. PLoS One 7(1), e29868 (2012). 98 de Paula MR, Ravetti MG, Berretta R,

Moscato P. Differences in abundances of cell-signalling proteins in blood reveal novel biomarkers for early detection of clinical Alzheimer’s disease. PLoS One 6(3), e17481 (2011).

Website 101 National Institue on Aging. Biospecimen best

practice guidelines for the Alzheimer's disease centers. NIA Version 1.0-12/03/07. www.alz.washington.edu/BIO/bio1.pdf

A consensus protocol for the standardization

