Mass spectrometry-based methods for robust measurement of Alzheimer's Disease biomarkers in biological fluids

Abstract

Alzheimer’s disease (AD) is the most common form of dementia affecting 60-70% of people afflicted with this disease. Accurate antemortem diagnosis is urgently needed for early detection of AD to enable reliable estimation of prognosis, intervention and monitoring of the disease. The National Institute on Aging/Alzheimer’s Association sponsored the “Research Framework: towards a biological definition of AD”, which recommends using different biomarkers in living persons for a biomarker-based definition of AD regardless of clinical status. Fluid biomarkers represents one of key groups of them. Since cerebrospinal fluid (CSF) is in direct contact with brain and many proteins present in the brain can be detected in CSF, this fluid has been regarded as the best biofluid in which to measure AD biomarkers. Recently technological advancements in protein detection made possible the effective study of plasma AD biomarkers despite their significantly lower concentrations vs to that in CSF. This and other challenges that face plasma based biomarker measurements can be overcome by using mass spectrometry.

In this review we discuss AD biomarkers which can be reliably measured in CSF and plasma using targeted mass spectrometry coupled to liquid chromatography (LC/MS/MS). We describe progress in LC/MS/MS methods development, emphasize the challenges and summarize major findings. We also highlight the role of mass spectrometry and progress made in the process of global standardization of the measurement of Aβ42/Aβ40. Finally, we briefly describe exploratory proteomics which seek to identify new biomarkers that can contribute to detection of co-pathological processes which are common in sporadic AD.

Graphical Abstract:

During the last 30 years, immunoassays and mass spectrometry (MS/MS) have been used for reliable measurement of Alzheimer’s disease (AD) biomarkers in CSF and plasma. However, substantial variations in the results obtained from different immunoassay-platforms was observed and for MS/MS based methods as well. MS/MS including the use of neat CSF-based Certified Reference Materials for calibrator standardization has now been established as reference methodology. This technology can be used for simultaneous analysis of multiple compounds and their different pathological forms that enables more rapid and efficient biomarker research on AD and provides for the first time in the history of Aβ42 measurements the prospect of developing cutoff concentration value that can be applied across analytical platforms and across centers throughout the world.

Alzheimer’s Disease

Alzheimer’s disease (AD) is the most common form of dementia affecting around 50 million people worldwide with 5.5 million AD cases in United States (alzheimersnewstoday.com 2020). AD is an irreversible, progressive brain disorder with common manifestations including loss of short-term memory, impaired reasoning, difficulty handling complex tasks, and poor judgment. Alzheimer's disease is often described as a continuum with three distinct stages: preclinical, prodromal [mild cognitive impairment (MCI)] and dementia. Although, nearly everyone with Alzheimer’s disease will eventually have the same symptoms there are two main types of AD; early-onset Alzheimer's disease in people who are younger than age 65 which accounts for approximately 5% of all people with AD and, late-onset Alzheimer's disease, the most common form of the disease, which affects to people age 65 and older. Rare forms of early-onset AD are strongly linked to mutations in genes for preseline 1/2 (PSEN 1/2) or amyloid precursor protein (APP) (Hardy & Higgins 1992). By contrast, late-onset AD is a multifactorial disorder in which age-related changes, genetic risk factors, such as allelic variation in apolipoprotein E (Apo E) and many other genes, vascular disease, Lewy Bodies, TDP-43 pathology, traumatic brain injury and risk factors associated with diet, the immune system, mitochondrial function, metal exposure, and infection are all implicated (Armstrong 2013).

The AD is pathologically characterized by the progressive accumulation of extracellular deposits of amyloid-β peptide (Aβ) in the form of amyloid plaques as well as intracellular deposits of tau protein as neurofibrillary tangles (Jack et al. 2013; Blennow et al. 2015) together with neuronal degeneration, glial activation and neuroinflammation (Bronzuoli et al. 2016). Amyloid plaques in AD are predominately composed of Aβ peptides, derived from the amyloid precursor protein (Cummings & Cole 2002). Among various Aβ peptides the 42-amino acid form of Aβ (Aβ42), is commonly observed in neuritic plaques (Snyder et al. 1994); and it appears to be central to AD pathogenesis. Tau protein, a microtubule-associated protein, is also intimately linked to the pathologic changes in AD pathogenesis: hyper phosphorylation, truncation, and oligomerization of tau proteins are critical to the formation of neurofibrillary tangles (Cummings & Cole 2002; Alonso et al. 2001; Basurto-Islas et al. 2008; de Calignon et al. 2010). Although plaques and tangles are considered the main pathologic features of Alzheimer’s disease it has been demonstrated that neuritic plaques and tangles in the AD brain contain other protein deposits such as α-synuclein and TDP-43 (James et al. 2016; Kovacs et al. 2013).

The AT(N) Research Framework

Historically AD diagnosis was most definitively confirmed post-mortem by histopathological analysis of brains and detection of the hallmark pathologic findings of amyloid plagues and fibrillary tangles (McKhann et al. 1984). However, accurate antemortem diagnosis is urgently needed for early detection of AD pathology to enable reliable estimation of prognosis and early intervention and monitoring of the disease. In 2011, the National Institute on Aging and Alzheimer’s Association (NIA-AA) described diagnostic recommendations for the preclinical, mild cognitive impairment, and dementia stages of Alzheimer’s disease (Jack et al. 2011; McKhann et al. 2011) which were consequently updated in 2018 and called “research framework” because its intended use is for observational and interventional research (Jack et al. 2018). The NIA-AA research framework grouped biomarkers of Alzheimer’s disease into those that detect amyloid β deposition [A], pathologic tau [T], and neurodegeneration [N]. The AT(N) classification system categorized different biomarkers like imaging using magnetic resonance imaging (MRI) (Wong et al. 2010) or amyloid positron emission tomography (PET) (Klunk et al. 2004) and biofluids by the pathologic process each biomarker detects. The AT(N) system is flexible in that new biomarkers that detect additional pathologic processes, e.g. vascular, Lewy Body, TDP-43 or others, can be added to the three existing AT(N) groups when validated tests for the co-pathologies become available.

The neuroimaging techniques such as MRI and PET are used to observe pathological changes in the brain and have shown promising results in identifying patients with mild cognitive impairment who are at a high risk of developing of AD (Jack et al. 1999; Desikan et al. 2010) while PET imaging of amyloid burden has potentially improved the accuracy of AD diagnosis (Serrano et al. 2014; Boccardi et al. 2016). However, imaging techniques are expensive and with access limited to large medical centers making early diagnostic screening as well as routine imaging monitoring challenging (Cilento et al. 2019). Thus, there is an urgent need for AD fluid biomarkers based on the AT(N) research framework.

Alzheimer’s Disease Fluid Biomarkers

CSF biomarkers

Cerebrospinal fluid (CSF) is an ultrafiltrate of plasma contained within the ventricles of the brain and the subarachnoid spaces of the cranium and spine. Taking into consideration that CSF is in direct contact with brain parenchyma many brain proteins present in the brain extracellular space can be detected in CSF (Blennow et al. 2010). Indeed, in 1992 it was shown that Aβ is present in the CSF (Seubert et al. 1992) and that there is a strong correlation between low postmortem ventricular CSF Aβ42 levels and high brain amyloid plaque load (Strozyk et al. 2003). In addition, reduced Aβ42 concentrations in lumbar CSF antemortem correlated with amyloid plaque load at autopsy (Tapiola et al. 2009). It has been postulated that the reduction in CSF Aβ42 concentration in AD reflects aggregation of Aβ into plaques, i.e., that the retention of the peptide in the brain parenchyma, results in reduced concentration of soluble Aβ in the CSF (Andreasen et al. 1999). In addition to Aβ42, two shorter Aβ forms, Aβ40 and Aβ38, are present in CSF (Janelidze et al. 2016b; Lewczuk et al. 2017). These peptides, similar in size to Aβ42, are produced by processing Aβ precursor protein (APP) by the concerted actions of β-secretase and the γ-secretase protease complexes (Portelius et al. 2011). It has been suggested that the concentration of Aβ42 in CSF is affected by the total amount of Aβ peptides present in the brain parenchyma in addition to the pathophysiological Aβ status (Lewczuk et al. 2017). By normalizing to the concentration of Aβ40, the most abundant Aβ peptide in the CSF, the ratio normalizes the differences in overall Aβ concentration, providing a better index of Aβ-related pathology. Recently, several studies reported that adding the CSF Aβ42/Aβ40 ratio to AD diagnostic tools: 1) improves prediction accuracy for amyloid plaque burden in patients with mild cognitive impairment, 2) improves discrimination of AD from other forms of dementia, and 3) increases the concordance between CSF and PET amyloidosis (Lewczuk et al. 2017; Janelidze et al. 2017; Janelidze et al. 2016b; Korecka et al. 2020).

Unlike Aβ peptides, tau protein and phosphorylated tau increase in the CSF of AD patients (Tapiola et al. 2009; Buerger et al. 2006) reflecting the intensity of neuronal and axonal degeneration and damage in the brain (Blennow et al. 2010).

It has been demonstrated that CSF total tau correlates with postmortem neurofibrillary tangle load (Tapiola et al. 2009). This suggests that neurofibrillary tangle-bearing neurons contribute to the CSF level of total tau while CSF levels of phosphorylated tau (p-tau) reflect both the phosphorylation state of tau and the formation of neurofibrillary tangles in the brain (Blennow et al. 2010). Phosphorylation of tau takes place at several different sites including threonine 181, a well-established biomarker of AD, serine 199, threonine 217, threonine 231, and serine 235. All of these phosphorylated forms of tau increase in the CSF of AD patients (Blennow et al. 1995; Ishiguro et al. 1999; Kohnken et al. 2000). Recent studies provide evidence that p-tau217 and p-tau181 closely parallel one another in relationship to PET and CSF measures of neocortical amyloid-β burden and is closely associated with amyloidosis, improving on detection of the latter at the asymptomatic stage (Barthelemy et al. 2020a) and more accurately distinguishes AD dementia from non-AD neurodegenerative disorders (Janelidze et al. 2020b; Thijssen et al. 2020). This is an area in need for direct comparison studies, by different laboratories and analytical platforms, of these findings to help determine which of these proteoforms of p-tau, or others, is diagnostically and prognostically most effective in various cohorts. It has been suggested that combining the levels of Aβ42 and p-tau181 or tau and α-synuclein could aid in the differentiation of AD from dementia with Lewy Bodies (DLB) (Viode et al. 2019). Alpha-synuclein is a neuronal protein, localized predominantly in presynaptic terminals, and involved in vesicle fusion and neurotransmitter release. Although, α-synuclein aggregates are the main component of Lewy bodies in DLB (Kim et al. 2014), the aggregates containing α-synuclein were also found in approximately half of sporadic AD brains (Hamilton 2000) and is a major co-pathology frequently also seen in familial AD (Leverenz et al. 2006). Additionally, the other member of the synuclein family, β-synuclein was proposed as a new candidate marker of synaptic loss. When measured by mass spectrometric assay β-synuclein was highly increased in CSF and serum of patients with AD compared with age matched controls and other neurodegenerative diseases (Oeckl et al. 2016; Halbgebauer et al. 2020; Oeckl et al. 2020).

Although Aβ- and tau-related AD biomarkers remain at the center stage of the AT(N) research framework, there are continuous efforts to develop novel diagnostic and prognostic fluid biomarker for Alzheimer’s disease. For example, several synaptic proteins have been identified in CSF including presynaptic synaptosomal-associated protein 25 (SNAP-25) and presynaptic proteins synaptotagmin-1 (SYT-1) (Davidsson et al. 1996; Davidsson et al. 1999), as well as the dendritic (post-synaptic) protein neurogranin (Thorsell et al. 2010; Kvartsberg et al. 2015) which is present in CSF in several molecular forms like monomeric full-length neurogranin, and N- and C-terminal truncations of neurogranin, as well as larger forms of still unknown composition (Nazir et al. 2020). Neuronal pentraxin 2 (NPTX2) and neuronal secretory protein VGF have been found as promising synaptic biomarkers for predicting progression vs. no progression from mild cognitive impairment to AD (Swanson et al. 2016; Hendrickson et al. 2015; Spellman et al. 2015).

It has been found that endo-lysosomal dysfunction is an another mechanism in the etiology of Alzheimer’s disease. Thus, it is possible that proteins involved in the normal function of endo-lysosomal vesicles can serve as biomarkers of disease. For example, it has been demonstrated that the lysosomal protein LAMP2, involved in chaperone-mediated autophagy, shows increased levels in CSF indicating endo-lysosomal dysfunction in Alzheimer’s disease (Sjodin et al. 2016).

The discovery that heterozygous missense mutations in the gene encoding triggering receptor expressed on myeloid cells 2 (TREM2) are risk factors for Alzheimer’s disease, has led to increased interest in immunobiology in the brain. There are several studies that have reported significantly higher concentrations of the soluble variant of TREM2 (sTREM2) in AD compared to control CSF and significant correlations were observed in CSF between soluble TREM2 and total tau as well as p-tau181 (Heslegrave et al. 2016; Brosseron et al. 2018; Gispert et al. 2016).

Apolipoprotein E (ApoE) is a well-established genetic risk factor associated with AD (Genin et al. 2011). It is a 299 amino acid protein synthetized by many tissue with the highest expression in liver and brain (Mahley & Rall 2000) and since it does not cross the blood-brain barrier, its secretion and function in blood and CSF are independent (Liu et al. 2012). The human ApoE gene possesses three major alleles: ε2, ε3, and ε4 with a worldwide frequency of approximately 8%, 78%, and 14%, respectively (Farrer et al. 1997). There are three polymorphic forms of ApoE: ApoE2, ApoE3 and ApoE4 which differ by two amino acids (cysteine and arginine) interchanges (Mahley & Rall 2000). These isoforms have been associated with the occurrence or progression of several pathological conditions for example traumatic brain injury , coronary atherosclerosis and AD (Verghese et al. 2011). ApoE4 is the most important genetic risk factor for sporadic AD and might be an interesting biomarker candidate. The presence of ApoE4 increases the risk of late-onset AD by a factor of three for heterozygous (ε3/ε4) or of 12 for homozygous (ε4/ε4) compared with individuals with no ApoE ε4 (Roses 1996).

Blood biomarkers

Although significant efforts have been devoted to the development of AD biomarkers in cerebrospinal fluid, CSF has limitations for establishing it in routine clinical practice, as the lumbar puncture is a modestly invasive procedure. Thus, AD biomarkers for pre-clinical diagnosis, screening of large population of patients and monitoring of disease progression based on body fluids other than CSF are urgently needed.

There have been extensive efforts to develop blood-based AD biomarkers using either immunoassay or mass spectrometry methodology. Advantages of blood over CSF include easier access, the sampling is minimally invasive, cost effective and the procedure is suitable for repeated analysis in longitudinal studies. However, from the analytical point of view, blood is a more challenging matrix for measurement of AD biomarkers. As a result of the selectivity of the blood-brain barrier, and the high blood to CSF volume ratio, the concentrations of protein biomarkers delivered from the central nervous system (CNS) to blood are much lower than in the CSF. For example, the concentration of total tau in blood is approximately 100 times lower than in CSF (Budelier & Bateman 2020) making analysis in blood analytically challenging. In addition, if a biomarker is not CNS-specific, but it is also present in peripheral tissues, it will be difficult to determine if its altered concentrations reflects the changes in brain or is also connected with systemic changes (Zetterberg 2019). Yet another important factor that may complicate analysis of biomarkers in blood is the presence of a high number of proteins, many in orders-of-magnitude higher concentrations-for example, albumin or endogenous antibodies, which may interfere in the assay (Zetterberg 2019; Apweiler et al. 2009). Finally, the analyte of interest may undergo proteolytic degradation by different proteases present in plasma (Yoshimura et al. 2008). All these analytical issues must be taken into consideration when developing an assay for the measurement of AD biomarkers in blood based matrix, i.e. plasma or serum.

Initial study of biomarkers in plasma by an ultrasensitive digital enzyme-linked sandwich immunoassay (ELISA) for Aβ42 in plasma were very encouraging and showed correlation of plasma with CSF Aβ42 and a reduced ratio of Aβ42/Aβ40 in plasma of amyloid PET-positive individuals but to a lesser degree of difference as compared to the CSF Aβ42/Aβ40 ratio (22% lower ratio in plasma vs 43% lower in CSF)(Janelidze et al. 2016a). The results of tau analysis in blood using immunoassays indicated increases in plasma total tau in AD patients when compared with mild cognitive impaired and normal control patients (Zetterberg et al. 2013; Mattsson et al. 2016). Additionally to this, several studies demonstrated significant plasma p-tau181 increases in AD and in MCI when compared to cognitive normal patients (Tatebe et al. 2017; Mielke et al. 2018; Janelidze et al. 2020a). Documented plasma p-tau181 was more strongly associated with both amyloid-β-PET and tau PET than plasma total tau (t-tau) (Mielke et al. 2028) and can differentiate AD patients from other tauopathies (Janelidze et al. 2020a; Thijssen et al. 2020).

Technical developments in the field of mass spectrometry currently allow not only routine detection of AD biomarkers in CSF but also to overcome some of the challenges associated with assays that use a blood-based matrix. Mass spectrometry (MS/MS) is characterized by high sensitivity and therefore ability to measure very low concentrations, specificity to distinguish between very closely related compounds, the availability to generate high-throughput methods to be used as a screening tool and its high multiplexing capacity. Over all, MS/MS is a powerful and highly sensitive platform free of many analytical deviations associated with antibody-based approaches (Hale 2013). Mass spectrometric plasma assay development initially focused on the core, well established CSF biomarkers, i.e. amyloid-β peptides, Aβ42, Aβ40 and amyloid-β precursor protein fragment 669-711 (APP669-711), and the ratios APP669-711/Aβ42 and Aβ42/Aβ40, since lowered concentrations reflect early neuropathological events in AD and thus could be used for early detection, and as a screening test. However, recently also mass spectrometry assay for analysis of plasma phosphorylated tau forms (p-tau) was described (Barthelemy et al. 2020b) indicating progress in the field of blood-based biomarkers for AD.

Analytical methods used for determination of AD fluid biomarkers

Measurement of almost all of the known AD biomarkers in CSF and blood are carried out using antibody-based assays, which are often associated with relatively high variability, particularly when different antibodies, sample preparation, reference material for calibrators preparation are used, leading to inconsistent results across laboratories (Zetterberg 2015; Olsson et al. 2016; Cilento et al. 2019). Based on the data for CSF Aβ42 obtained from the Alzheimer’s Association external quality program the inter-laboratory (n=40 laboratories, three different assays) variability (CV%) ranged from 13% to 36% and the intra-laboratory CV% were in the range of 2.3%-26% (this is assay dependent variation, data for Luminex xMAP) (Mattsson et al. 2011). The overall wide inter-laboratory variability for these assays has likely been caused by differences in pre-analytical procedures (e.g., handling/storage of QC samples or commercial kits at individual sites), analytical procedures, and variations related to the commercial assays themselves (Mattsson et al. 2011). Development of fully automated assays has significantly reduced variability across common samples and different laboratories for CSF Aβ42, t-tau and p-tau181 and this development has diminished difficulties in interpretation of results obtained from different centers (Reiman 2017). Nevertheless, the commercially available assays have been standardized using diverse in-house calibrator materials, which can lead to difficulties in the comparison of their results. Recent studies have provided data that strongly supports a recommended procedure that includes the use of certified reference material (CRM) with defined target values (Kuhlmann et al. 2017; Andreasson et al. 2018). Three such CRMs based on human CSF, with low, medium and high level of Aβ42, currently available in Europe and the USA (Boulo et al. 2020) were developed under the auspices of the International Federation of Clinical Chemistry Working Group for CSF proteins. The availability of certified reference material for calibration of different assays and developing common procedures for sample collection and handling, similar to the one established for biomarker studies in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (Shaw et al. 2011; Shaw et al. 2009), could decrease the variability of obtained results, both for batch-to-batch and between assays comparison (Andreasson et al. 2018).

Another limitation related to immunoassays is accessibility to specific antibodies which are not always available, especially for the newly proposed candidate biomarkers for AD, and unfortunately immunoassays’ antibodies performance can be interfered with by other proteins present in biofluid samples (e.g. with blood albumin and immunoglobulin; (Apweiler et al. 2009). Additionally, immunoassays are challenging to multiplex (Hoofnagle & Wener 2009) but besides of this difficulty, a few multiplex immunoassays are available both, for CSF and plasma, like SIMOA (Quanterix, Billerica, MA, USA), Meso Scale Discovery (Rockville, MD, USA) or recently described a densely aligned carbon nanotubes sensor array for AD biomarkers in plasma (Kayoung et al. 2020).

Currently, there are several immunoassays available for the quantitation of Aβ42, tau and p-tau in CSF. The most widely used are: the INNOTEST® enzyme-linked immunosorbent assay (measure Aβ42, p-tau181P, and total-tau); the first fully automated assay platform and reagents is the Elecsys® CSF immunoassay for Aβ42, Aβ40, total tau and p-tau181 (Roche Diagnostics, Rotkreuz, Switzerland), run on the Cobas 6000 instrument (Roche Diagnostics), the fully automated Lumipulse®G1200 instrument (Fujirebio, Malvern, PA, USA) with chemiluminescence assays for Aβ42, Aβ40, p-tau181, and total-tau; the ADx-EUROIMMUN enzyme-linked immunosorbent assay for Aβ42 and Aβ40 (ADx Neurosciences, Gent, Belgium); Meso Scale Discovery (MSD) with the Aβ triplex (Aβ42, Aβ40, and Aβ38) or total-tau kits.

Early studies of plasma Aβ42 and Aβ40 as biomarkers of AD were disappointing with results showing variability and overall not significant changes in Aβ42 or Aβ42/40 between patients and controls (Olsson et al. 2016). Likely factors behind these disappointing early results include biologic, methodologic and pre-analytic variables and these early results pointed to the need for development of more highly standardized assays and pre-analytics (Rissman et al. 2012). The inconsistent clinical and analytical performance results with these “first-generation” assays, e.g. ELISA or Luminex, likely included additional issues such as epitope masking by hydrophobic Aβ peptides binding to plasma proteins (Kuo et al. 1999; Blennow & Zetterberg 2018). In 2011 a new method based on the single-molecule array (SIMOA) technique for measurement of Aβ42 in plasma was described (Zetterberg et al. 2011) and the data collected during evaluation of this assay found weak but significant correlations between both plasma Aβ42 and the Aβ42/Aβ40 ratio and the corresponding CSF measures, as well as to cortical [18F]flutemetamol PET retention (Janelidze et al. 2016a). Analytical approaches to overcome immunoassay limitations were sought after and led to development of mass spectrometry-based methods. As an antibody-free platform with high sensitivity, accuracy and excellent multiplex capability mass spectrometry is an obvious choice as an alternative technique for plasma Aβ42 and Aβ40 measurements, and to serve as reference methodology.

Mass spectrometry

Mass spectrometry has been long recognized as an important and powerful analytical tool. This technique is based on the principle that charged particles moving from the ion source to the detector through an electric field can be separated from one another by their mass-to-charge ratio (m/z) (Willoughby et al. 2002). Mass spectrometry methodologies can be divided into five basic elements: samples preparation, followed by sample introduction, ionization, mass analyses and detection. Sample preparation in itself is a major component of any mass spectrometry method. Generally, sample preparation involves more or less extensive preparation which sometimes is followed by an initial separation by liquid chromatography. The molecules of the sample must be ionized and the most common ionization techniques are electrospray ionization (ESI), atmospheric pressure chemical ionization (APCI) and matrix assisted-, or surface enhanced laser desorption ionization (MALDI and SELDI, respectively). The mixture of ionic species is consequently introduced to the mass analyzer and obtained data are transformed into useful information in the form of mass spectra.

Proteomics, as a large-scale study of proteins in CSF, plasma and brain tissues, can be used for identifying potential biomarker candidates (exploratory proteomics) and for selective and precise quantification of unique peptides or intact proteins (targeted proteomics) (Portelius et al. 2017; Brinkmalm et al. 2015). The term “proteomics” is increasingly being used to designate work in which mass spectrometry is the central technology platform. A defining advantage of MS/MS for discovery or hypothesis generation in clinical proteomics is the capability to confidently identify thousands of proteins in complex biological samples without pre-specification of the analytes to be measured. However, this unbiased coverage comes with the cost of reduced sensitivity and stochastic sampling (Gillette & Carr 2013). To verify biomarkers and to test hypotheses more sensitive quantitative protein approaches must be used. Targeted MS/MS platforms provide an ideal tool for such activities by focusing the resources of the mass spectrometer on one of a few defined analytes (Gillette & Carr 2013; Spellman et al. 2015)

Targeted proteomics in Alzheimer’s disease

Selected reaction monitoring (SRM, sometimes also referred to as multiple reaction monitoring, MRM) and parallel reaction monitoring (PRM) are two targeted MS/MS approaches with powerful applications for biomarker detection and quantification. These two techniques are antibody independent which makes them a valuable and attractive alternative to immunoassay especially for initial discovery and clinical validation studies. Despite methodologic simplicity results produced by immunoassays, are limited by availability, selectivity and cost of antibody reagents especially when developing such assays for new biomarkers. In addition, “de novo” development of immunoassays is time consuming and a significant fraction of commercially available assays requires further validation (Vidova & Spacil 2017).

In SRM using triple quadrupole analysis, a previously found precursor ion of compound of interest is selected by the first quadrupole, then fragmented in the collision cell (second quadrupole). The third quadrupole filters only specific product ion/s according to the described transition/s for each parent biomarker molecule, and the collected data are recorded over time as the compound of interest elutes from a LC, and are used for target compound identification and quantification. SRM provides for highly specific detection and can be applied to complex biological samples with a reasonable degree of sample preparation (Portelius et al. 2017; Picotti & Aebersold 2012; Vidova & Spacil 2017). In comparison to immunoassays the SRM technique is cost effective, features fast assay development and deployment and it is in principle capable of distinguishing highly similar proteoforms such as isoforms and post-translationally modified proteins (Picotti et al. 2013). The SRM assay approach provides an opportunity for simultaneous detection of multiple target compounds and absolute quantification of the proteins/peptides of interest can be achieved by using a stable isotope labelled version of the proteins/peptides as an internal standard which is added to samples prior to sample preparation. Because these molecules have identical physicochemical properties to the endogenous analyte, and are only distinguished from the analyte in the mass analyzer, they can be used to track and compensate for variations in the sample preparation and instrumental analysis procedures (Pannee et al. 2016a).

Parallel reaction monitoring is an alternative method of targeted quantitation and requires much less assay development (Ronsein et al. 2015). In PRM analysis using a quadrupole mass filter in line with a high resolution and accurate mass analyzer (like Orbitrap), precursor ions are selected and fragmented, similar to SRM, but all resulting product ions are detected. The PRM assays require less laborious method development compared to SRM, since product ions are not selected a priori therefore making PRM an ideal targeted approach for qualification and verification of protein profiles (Peterson et al. 2012; Andersson et al. 2019). However, an important PRM disadvantage is that it relies on an analyzer that is slower (Peterson et al. 2012), making this a less efficient approach for targeting large sets of peptides/proteins simultaneously and SRM remains the most common targeted MS/MS approach for studying biomarkers (Cilento et al. 2019) and indeed this approach has been successfully used for quantification analysis of numerous of AD biomarkers both in CSF and blood-based matrices.

Analysis of CSF biomarkers by mass spectrometry

Amyloid beta peptides in CSF

The first SRM method for analysis of 3 amyloid-β peptides, Aβ42, Aβ40 and Aβ38 in human CSF was published by Lame in 2011 (Lame et al. 2011). In this method N15-labelled human Aβ42, Aβ40 and Aβ38 were used as internal standards and a Waters Acquity UPLC coupled with Waters Xevo TQ-S was employed for quantitative analysis. Sample preparation was based on extraction performed on an Oasis MCX μElution plate, after denaturation of aggregated forms of Aβ peptides in human and artificial CSF using high concentration guanidine hydrochloride, without the need for proteolytic digestion. Artificial CSF with bovine serum albumin (BSA; 4mg/mL) was used as a surrogate matrix for calibrator preparation and the equivalent performance of this matrix and human CSF was confirmed by the authors.

This pioneer paper opened the gate for development of mass spectrometric assays for analysis of Aβ peptides in human CSF by several groups and two years later another method was published (Pannee et al. 2013). A unique approach of this assay was using a reverse calibration method. N15-labelled Aβ42, Aβ40 and Aβ38 were used as calibration standards in pooled human CSF with fixed amounts of endogenous Aβ peptides serving as internal standards. However, for human CSF unknown samples the “heavy” peptides (labelled) served as internal standards. SRM analysis of positively charged peptide ions was performed on a triple quadrupole mass spectrometer (TSQ Vantage, Thermo Scientific).

In the early 2000’s measurements of amyloid-β peptides in CSF by different immunoassays were characterized by high variability that made it very challenging to use Aβ42 as a biomarker in routine clinical practice because it was not possible to establish uniform cutoff values for diagnosis since the results of studies performed in different laboratories were not comparable (Mattsson et al. 2013; Carrillo et al. 2013). So there was a need for standardization of CSF AD biomarker assays and development of reference measurement procedure was an important step to reduce this variability and would permit global standardization of different assay platforms. In 2014 two, independent reference measurement procedures for analysis of Aβ42 in human CSF were developed, validated, published and approved by JCTLM as reference measurement procedures and listed in the JCTLM database under DB ID C11RPM9 and C12RPM1 (Leinenbach et al. 2014; Korecka et al. 2014). The Leinenbach et al. method used N15-labelled Aβ42 as the standard material for preparation of calibrators in pooled human CSF and C13-labelled Aβ42 as an internal standard, and PRM measurements were performed on the high-resolution quadrupole-Orbitrap hybrid instrument. The Korecka et al. method used N15-labelled Aβ42 as an internal standard and unlabeled Aβ42 as a standard for preparation of calibrators in artificial BSA (4 mg/mL)-based CSF. All SRM measurements were initially performed on a triple quadrupole API 5000 (Sciex) coupled with an Acquity UPLC (Waters). However, the assay was subsequently transferred to a more sensitive mass spectrometer, Xevo TQ-S (Waters) when it became available. Both methods utilized sample preparation with high concentration guanidine hydrochloride and MCX μElution columns based on that established by Lame et al. Four methods, by Lame, Leinenbach, Korecka and one developed at PPD, Inc. (Jenkins et al. 2011) were additionally assessed by Round Robin studies (Table 1). The study demonstrated that these 4 assays established in four different laboratories using different instruments, calibration methods and different calibrators preparations, produced similar results. Additionally, use of a common Aβ42 reference standard further decreased inter-laboratory variation that proved the value of standard reference material (Pannee et al. 2016a; Kuhlmann et al. 2017). The certified reference materials were subsequently developed and characterized with the support of the Global Biomarker Standardization Consortium of the Alzheimer’s Association using the two reference measurement procedures and three CRMs based on human CSF, at 3 different levels of Aβ42 are currently available in Europe and USA and can be ordered directly from JCTLM or from Analytical Reference Materials International/LGC Standards, Manchester, NH, USA. Development of Aβ42 CRMs is crucial for method harmonization since they can ensure the equivalence of results between methods and across platforms and enable widespread use of a common cut-off value for the measurement of Aβ42. Additional mass spectrometric assays can also be developed for use in the future for harmonization of different biomarkers, for example, analysis of Aβ42 in plasma or analysis of tau in human CSF.

Table 1.

CSF Aβ42 and Aβ40 mass spectrometry methods

Center	University of PA, USA	University of Gothenburg, Sweden	Waters, USA	PPD Inc., USA
Method	UPLC/MS/MS	LC/MS/MS	UPLC/MS/MS	UPLC/MS/MS
Sample preparation	Oasis MCX μElution plate (Waters)	Oasis MCX μElution plate (Waters)	Oasis MCX μElution plate (Waters)	Oasis MCX μElution plate (Waters)
Sample volume	0.1 mL	0.18 mL	0.2 mL	0.2 mL
Mass spectrometer	Triple Quadrupole Xevo TQ-S (Waters)	Q-Exactive Quadrupole-Orbitrap (Thermo Scientific)	Triple Quadrupole Xevo TQ-S (Waters)	Triple Quadrupole Xevo TQ-S (Waters)
HPLC	Acquity UPLC (Waters)	Acella 1250	Acquity UPLC (Waters)	Acquity UPLC (Waters)
Calibration Aβ42 Aβ40	100-3000 pg/mL 200-20000 pg/mL	150-4000 pg/mL 1500-40000 pg/mL	100-10000 pg/mL 100-10000 pg/mL	50-5000 pg/mL 100-10000 pg/mL
Column	XBridge-trap column BEH 300-analytical (Waters)	Pro-Swift RP-4H (Thermo Scientific)	BEH 300 (Waters)	XBridge-trap column BEH 300-analytical (Waters)
Transition, m/z Aβ42 Aβ40	SRM 1129.6->1079 1086.6->1054	PRM 1129.58->15 fragments 1083.47->17 fragments	SRM 1129.0->1078.5 1083.0->1053.6	SRM 1129.0->1078.5
References	Korecka et al. 2020	Pannee et al. 2016	Lame et al. 2011	Jenkins et al. 2011

Abbreviations:

UPLC/MS/MS – ultra pressure liquid chromatography with tandem mass spectrometric detection

MCX – Mixed-Mode Cation eXchange

SRM – selected reaction monitoring

PRM – parallel reaction monitoring

An important advantage of singleplex mass spectrometric methods is that they can be relatively fast and easily converted to multiplex assays. This simple, cost and time effective conversion wouldn’t be possible for immunoassays. Since emerging evidence suggested that ratio of CSF Aβ42/Aβ40 may be superior to CSF Aβ42 alone, both reference measurement procedures were modified by adding 2 additional Aβ peptides, Aβ40 and Aβ38 as calibrators, and validated to allow quantification of multiple Aβ species within a single MS analysis (Pannee et al. 2016b; Korecka et al. 2020). Data from the studies which utilized the multiplex methods substantiated earlier studies, using immunoassays, that documented significantly improved concordance with amyloid PET (Fluorbetapir PET, or 18F-flutemetamol PET) using ratios of Aβ42/Aβ40 and Aβ42/Aβ38 vs Aβ42 alone.

Analytical methods with mass spectrometric detection are often called “gold standard” methods and are often used in method comparison studies as a reference procedure. As an example there is one study that compared results for Aβ42 in human CSF using the highly automated and validated Elecsys β-Amyloid (1-42) CSF assay versus the Korecka et al. reference measurement procedure. High correlation was found between Elecsys β-Amyloid (1-42) CSF assay and this reference measurement procedure strongly supported the use of the Elecsys β-Amyloid (1-42) CSF assay for reliable AD diagnosis (Shaw et al. 2019; Korecka et al. 2020).

Tau protein in CSF

Tau is a second CSF AD biomarker which has been analyzed not only by immunoassays but also by mass spectrometric methods. Given the molecular diversity of tau in CSF, with its many isoforms, posttranslational modifications and peptide fragments, the selectivity of the immunoassays has often been questioned (Kang et al. 2013). Between 2014 and 2016 the first two methods for CSF tau analysis by mass spectrometry were published (McAvoy et al. 2014; Bros et al. 2015; Barthelemy et al. 2016). For these methods sample preparation was based on enzymatic digestion of proteins with trypsin to obtain tau specific peptides which were consequently used for quantification of tau by measuring the relative signal of the chosen peptide to that of an internal standard which was isotopically labeled tau, 441(2N4R) or 412 (1N4R) amino acid isoform, dependent on the method. In general, additional steps for sample purification may be included, both before digestion at the protein level or after digestion at the peptide level (Chappell et al. 2014). The SRM method developed by McAvoy et al. employed a monoclonal antibody to selectively enrich tau from CSF, and the PRM method by Bros et al. was antibody-free and used partial protein precipitation with perchloric acid and SPE columns for sample clean-up before digestion with trypsin. Different isoforms of the tau protein were used by these two groups as standard material (recombinant tau 441(2N4R) or tau 381(1N3R)) and different peptide/s have been chosen for the purpose of quantification, which could lead to variable results. The peptide/s selected for tau quantification must not be subject to any truncation or post-translation modification (Bros et al. 2015) . There is also the possibility of using wing peptides as standards or internal standards, when labelled which is justified by the observation that both CSF and plasma contains many truncated forms of tau and less commonly full length protein (Barthelemy et al. 2020b; Sjogren et al. 2001; Johnson et al. 1997). Wing peptide includes the sequence of the selected quantitative peptide with additional amino acids on the N- and C-terminals and it should behave as a substrate of the digesting enzyme similarly to the parent biomarker protein, however it is not clear that this is always the case (Chappell et al. 2014). Both MS/MS methods were compared with ELISA and the comparisons demonstrated high correlations with r² values ranging from 0.95 to 0.99. However, the results obtained by mass spectrometry-based assays were about 20 times higher than those from ELISA and the reason/s for this higher concentration was described elsewhere (Barthelemy et al. 2016; Portelius et al. 2017). Both methods when applied in clinical studies documented significant differences in tau concentration between AD patients and healthy controls.

Mass spectrometry methods represent an orthogonal technique that can confirm the analytical specificity of more commonly used immunoassays. They are also powerful platform to measure specific modifications of proteins like, for example tau, and to investigate proteins at a molecular level (McAvoy et al. 2014; Barthelemy et al. 2016). Additional work, which can lead to the development of a reference measurement procedure, reference standard materials and global standardization of tau measurements if needed, is currently under discussion with the support from Global Biomarker Consortium for the Standardization of CSF Biomarkers and the IFCC CSF Biomarker study group.

In 2017 and 2019 two papers were published to report on development of multiplex MS/MS methods for precise quantification of tau and α-synuclein simultaneously (Viode et al. 2019) and tau and Aβ42 in parallel (Pottiez et al. 2017). For the first method tau and α-synuclein were extracted at the same time by protein precipitation, then digested with trypsin and several peptides for each biomarker were analyzed in the PRM positive mode on a Q-Exactive Quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific). Data processing was optimized by summing the signals of up to six major fragment ions to provide one single extracted ion chromatogram for each targeted peptide. The method was used to determine concentrations of tau and α-synuclein in CSF samples obtained from a small group of patients with AD and healthy controls. Significantly higher results of both proteins were found in AD CSF when compared with CSF from healthy controls, and the levels were determined with each targeted peptide. Since tau and α-synuclein concentrations showed opposite trends in AD (both are elevated) and dementia with Lewy Bodies (both are depressed) combining the two biomarkers can be beneficial for differentiation of these two diseases.

The development of methodology for the simultaneous analysis of tau and Aβ42 was even more challenging because of the compounds different solubility and other properties leading the investigators to establish an SRM method to quantify both biomarkers in parallel. For parallel analysis the sample processing is similar for both analytes so this can be done in parallel however, using two separate aliquots of CSF (Pottiez et al. 2017). Protein digestion was performed for both, tau and Aβ42, using separate protocols. SRM quantification analysis was performed in the positive mode on a TSQ Vantage, triple quadrupole mass spectrometer (Thermo Scientific) coupled to a nano-Acquity UPLC (Waters). Application of this assay in a pilot clinical study showed significant differences in tau concentration between MCI, AD and healthy control patients but no significant decrease of Aβ42 in MCI and AD compared to controls. According to the authors, this lack of significance in Aβ42 concentrations was probably connected with the choice of targeted peptide corresponding to amino acids in positions 17 to 28 of Aβ42 for its quantification.

Recently, beginning in 2018 increased interest in p-tau and its different phosphorylated forms emerged using an immunoassay (Mielke et al. 2018). However, several groups initiated mass spectrometry-based approaches. For example, one group reported development of a highly sensitive and specific mass spectrometry method using PRM for identification of tau phosphorylation sites (Barthelemy et al. 2019). Using this method, they were able to compare the abundance of phosphorylation sites in tau protein in brain tissue and CSF in people with and without AD. They detected 29 distinct phosphorylation tau sites in full-length tau from brain and 12 sites on truncated tau in CSF. They also identified phosphorylation at the threonine 153 and 175 site that were exclusively found in AD CSF. Targeted mass spectrometry multiplexing ability and high-throughput capacity of this method opened a new research avenue for understanding the physiology of tau phosphorylation and its alterations in AD. The same group also used their mass spectrometry assay for precise detection of different, low-abundance forms of phosphorylated tau and showed that increased levels of CSF p-tau217 more sensitively detect both the preclinical and advanced forms of AD (Barthelemy et al. 2020a) although the same group have reported closely parallel performance of p-tau217 and increases of p-tau181 for detection of both the preclinical and advanced forms of AD (Barthelemy et al. 2020a). At the time of publication of their methodology the precise detection on different forms of phosphorylated tau remains challenging but a valuable tool for understanding the time courses of change in degree of phosphorylation of different p-tau forms. Further assessments of the comparative clinical performance of these and other proteoforms of p-tau are warranted and planned using both validated immunoassays and mass spectrometry methods.

Mass spectrometry also enabled studying the kinetics of multiple isoforms and fragments of tau in human CSF using stable isotope labeling kinetics (SILK) and MS/MS approach (Sato et al. 2018) proving again how powerful mass spectrometry is.

Other CSF biomarkers

Alpha- and beta-synuclein

In 2016 MRM mass spectrometric method for simultaneous quantification of unmodified alpha-, beta- and gamma-synuclein in human CSF using a stable-labeled protein standard (α-synuclein) and stable-labeled peptides or recently, labeled-full length β-synuclein, as internal standards was described (Oeckl et al. 2016; Oeckl et al. 2020). Sample preparation included tryptic digestion and peptides condensation/purification by solid phase cation extraction. Sciex Q-Trap 6500 mass spectrometer was used for MRM analysis. This method was characterized by high sensitivity and good inter-assay precision (below 11.6% for peptides from both, α- and β-synuclein, recently improved to 5.1% by using labeled full-length β-synuclein as an internal standard). Synuclein concentrations obtained by this method were highly correlated with each other in CSF (r>0.8) and the concentrations of α-and β-synuclein were increased in AD patients.

SNAP-25, SYT-1 and LAMP2

A mass spectrometric method for quantification of SNAP-25 in human CSF was published in 2014 (Brinkmalm et al. 2014). This method used immunoprecipitation using antibodies on magnetic beads and digestion with trypsin, as the two main steps for sample preparation, followed by quantitative analysis using high resolution selected ion monitoring (HR-SIM) on a Q Exactive Quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific)). Using this method, the authors were able to demonstrate significantly higher levels of SNAP-25 in CSF samples from patients with prodromal Alzheimer’s disease and Alzheimer’s disease compared with controls.

In 2016, after a minor modification, the method for SNAP-25 was used to study SYT-1 and LAMP-2 as candidates for new AD biomarkers (Ohrfelt et al. 2016; Sjodin et al. 2016). The group reported significantly increased levels of synaptotagmin-1 in patients with MCI due to Alzheimer’s disease and patients with dementia due to Alzheimer’s disease compared with controls and increased levels of LAMP2 peptides in CSF from individuals having an AD core biomarker profile compared to subjects with a control biomarker profile. These two findings support the hypothesis that synaptotagmin-1 and LAMP2 could be used as early markers for Alzheimer’s disease.

Neurogranin

Besides ELISA, concentrations of neurogranin in human CSF and brain tissue can be measured by mass spectrometric assays, exemplified by that described by Kvartsberg (Kvartsberg et al. 2015). In this method magnetic beads coated with two in-house antibodies were used for immunoprecipitation of brain extracts or CSF followed by proteolytic digestion and analysis using a MALDI/TOF/TOF mass spectrometer (Bruker Daltonics) for peptide identification or nanoflow liquid chromatography coupled to a Q Exactive electrospray ionization hybrid quadrupole–orbitrap mass spectrometer (Thermo Fisher Scientific) for peptide verification and quantification. Application of this technology to clinical studies revealed a few important observations. The authors of this study concluded that the levels of CSF neurogranin are markedly increased both in AD and MCI-AD, and concentrations predict conversion from MCI to AD dementia and a faster rate of cognitive decline within amyloid-positive prodromal AD dementia cases.

NPTX2 and VGF

Neuronal pentraxin-2 was established as a candidate novel AD biomarker in an FNIH Biomarkers Consortium CSF Proteomics Project whose goal was to evaluate a multiplexed mass spectrometry-based approach for the qualification of candidate AD biomarkers using CSF samples from ADNI. The study used unique peptides derived from candidate protein biomarkers by a trypsinization procedure to determine the ability of a panel of selected peptides measured with mass spectrometry to detect AD dementia. A robust targeted MS/MS assay was developed and run on the system containing a nano-Acquity UPLC (Waters) and 5500 Q-Trap mass spectrometer (Sciex). The study identified several candidate proteins with potential diagnostic or predictive utility including neuronal pentraxin-2 and neurosecretory protein VGF (Spellman et al. 2015). A follow up study used 750 CSF samples from ADNI participants and employed quantitative targeted LC/MS/MS (Sciex 6500 Q-Trap, nano-Acuity UPLC) to study the rates of change over an average of 4 years’ time of the candidate biomarkers. The investigators found that the rates of decrease in NPTX2 concentrations differed significantly-the decreased concentrations were significantly greater in progressors vs non progressors, in MCI vs cognitively normal and in biomarker based AD vs biomarker based non-AD in ADNI participants (Libiger et al. 2020). VGF showed more modest decreases over time in this study.

TREM2

A single reaction monitoring assay for the analysis of the soluble form of TREM2 in human CSF was published in 2016. The assay was established on a Waters Acquity UPLC coupled to a Xevo TQ-S triple quadrupole mass spectrometer which operated in the positive ion mode (Heslegrave et al. 2016). A significantly higher sTREM2 concentration in AD compared to control CSF was found. There were also significant correlations between CSF sTREM2 and T-tau as well as p-tau181.

Apolipoprotein E

Apolipoprotein E can be analyzed in both CSF and blood-based matrices, i.e. plasma or serum, using immunoassay or mass spectrometric methods. It has been reported that the results from the studies using immunoassays are inconclusive, carried a significant risk for interferences and biases, they can detect truncations or fragments, and additionally this technique does not have the capability to independently measure ApoE isoforms (Hesse et al. 2000; Merched et al. 1997; Baker-Nigh et al. 2016). So there was a need for the different technique that overcome the disadvantages of immunoassays but simultaneously measure ApoE isoforms with good precision and with high specificity for the isoforms. According to our search on PubMed three mass spectrometric methods for analysis of ApoE in CSF only or in CSF and plasma were published between 2014 and 2020 (Martinez-Morillo et al. 2014b; Baker-Nigh et al. 2016; Minta et al. 2020). For these methods sample preparation was based on tryptic digestion and the obtained 4 peptides for the individual isoforms (LAVYQAGAR for ApoE3 and E4, LGADMEDVR for ApoE4, LGADMEDVCGR for ApoE2 and E3 and CLAVYQAGAR for ApoE2) and one common peptide (LGPLVEQGR) for total ApoE, were used for quantitation of the three isoforms, total ApoE and phenotype identification, based on the presence of different combinations of the four tryptic peptides (Martinez-Morillo et al. 2014b) . Corresponding peptides labeled with 13C and 15N at C-terminal arginine were used as internal standards. Mass spectrometric analysis was performed on triple quadrupole instruments for SRM/MRM methods (Martinez-Morillo et al. 2014b; Baker-Nigh et al. 2016) or on Q Exactive hybrid quadrupole-orbit high-resolution mass spectrometer for PRM (Minta et al. 2020).

Method comparison is a crucial part of each method validation process and indeed these mass spectrometric assays were compared with ELISA showing strong (R²=0.8) (Baker-Nigh et al. 2016) or moderate correlation (R²=0.6) (Martinez-Morillo et al. 2014b). Various ELISAs for total ApoE quantification have been previously used and described (Wahrle et al. 2007(Cruchaga et al. 2012), showing significantly different results and highlighting a lack of reliability and, therefore, the need for a reference method.

In 2020 one group presented results from analysis of ApoE in CSF by a mass spectrometric assay (Minta et al. 2020). They reported that CSF total ApoE concentrations did not differ between amyloid β negative cognitively unimpaired controls and clinically diagnosed AD patients and even though there was a significant increase of total ApoE in the amyloid β positive group compared with amyloid β negative individuals (p < 0.001), the magnitude of the effect was very small (AUC = 0.55). There was a difference in concentration between isoforms in heterozygous individuals in an isoform-dependent manner (E2 < E3 < E4) (p < 0.001, AUC = 0.64–0.69). They concluded that the results indicate that neither the concentrations of total ApoE nor the different ApoE isoforms in CSF are associated with ApoE-ε4 carrier status, Aβ status, or clinical dementia diagnoses.

Analysis of blood-based biomarkers by mass spectrometry

Amyloid beta in plasma

Early attempts to use ELISA-based assays to analyze plasma Aβ were not very successful (Rissman et al 2012) so the application of mass spectrometry-based methodology was tested in attempts to achieve this goal. In 2014 the first two mass spectrometry methods for analysis of Aβ peptides in plasma, including immunoprecipitation as a part of sample preparation, were published. One of them required a high volume of 5 mL plasma using MALDI/TOF/TOF but demonstrated feasibility of using 1 mL plasma volume and successful quantitation using SRM LC/MS/MS. Using the latter this group succeeded in measuring plasma Aβ42, Aβ40 and Aβ38 in a small group of patient samples (Pannee et al. 2014). In the second method the investigators measured Aβ42 and another peptide derived from APP, i.e. APP669-711 (Aβ−3-40) and documented sensitivity of 82.5% and specificity of 77.3% for Aβ42 and sensitivity of 92.5 % and specificity of 95.5% for the APP669-711/Aβ42 ratio using amyloid PET as the objective brain amyloid measurement for these assessments (Kaneko et al. 2014).

In 2017 and 2018 two papers, reported promising results for assessing brain amyloid status based on the ratio of Aβ42/Aβ40 in plasma (Ovod et al. 2017; Nakamura et al. 2018). Ovod et al. used for their studies a novel liquid chromatography-tandem mass spectrometry assay, Nakamura et al. employed a MALDI/TOF method based on that described in 2014 (Kaneko et al. 2014). Both methods are characterized by extensive sample preparation with immunoprecipitation using anti Aβ peptide antibody coupled to magnetic beads, several steps of washing and drying, and for the Ovod method – a proteolytic digestion step and further purification on SPE columns, or a second immunoprecipitation step (Nakamura et al.) as the additional purification steps. Stable isotope labelled Aβ peptides were used as internal standards, three separate for Aβ42, Aβ40 and Aβ38 (Ovod et al.) or one common internal standard, stable isotope labelled Aβ38, for all measured Aβ peptides (Nakamura et al.). Ovod et al. measured concentrations of three Aβ peptides (Aβ42, Aβ40 and Aβ38) in plasma samples using a Thermo Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific) coupled to a nano-Acquity LC (Waters) and Nakamura et al. used a MALDI/TOF mass spectrometer (AXIMA, Shimadzu/KRATOS) (Table 2).

Table 2.

Plasma Aβ42/Aβ40 mass spectrometry methods

	Advance Med for Dementia, Shimadzu, Japan	Washington University, USA	UGOT, Univ. College, London, UK	Araclon Biotech, Spain
Method	MALDI/TOF	LC/MS/MS	LC/MS/MS	LC/MS/MS
Sample preparation	IP	IP/protease	IP	direct
Sample volume	0.25 mL	1.0 mL/1.6 mL	0.25 mL	0.2 mL
Mass spectrometer	AXIMA Shimadzu/KRATOS	Thermo Orbitrap/ Xevo TQ-S (Triple Quad)(Waters)	Q-Exactive Quadrupole-Orbitrap (Thermo Scientific)	Q-Trap 6500 (Sciex)
HPLC	N/A	Nano-Acquity (Waters)	Dionex UltiMate LC	Eksigent M3 Micro-HPLC
Calibration Aβ42 Aβ40	11-180 pg/mL 40-640 pg/mL	5-80 pg/mL 50-800 pg/mL	10-200 pg/mL 50-800 pg/mL	10-200 pg/mL 50-1000 pg/mL
Column	N/A	Acquity M-class HSS T3	Pro-Swift RP-4H	Zorbax 300SB-C18 and HALO ES-C18 (analytical)
Transition, m/z Aβ42 Aβ40	N/A	PRM C12,N14 peptide: 699.9-> 4 fragm. C12,N14 peptide: 607.8-> 5 fragm.	PRM 1129.58-> 15 fragm. 1083.47->17 fragm.	Not reported
Sens/Specificity	90.0%/70.6%	88%/76%	86.6%/71.9%	Studies underway
AUC	0.889	0.88/0.94	0.817/0.841	0.83/0.88*
Cutpoint	<0.039	<0.1218	<0.095	Studies underway
References	Nakamura et al. 2018	Ovod et al. 2017 Schindler et al. 2019	Pannee et al. 2014 Keshavan et al. 2021	Janelidze et al. 2021

Note: The AUC values are from ROC analyses of Aβ42/Aβ40 vs amyloid PET status and following the / is the AUC that results from including age + sex + ApoE ε4 status.

^*These AUC values are AUC values for Aβ42/Aβ40 alone and following the / is AUC for a logistic regression model that includes ptau217 + the Aβ42/Aβ40 ratio.

Abbreviations:

UGOT – University of Gothenburg, Sweden

MALDI/TOF - matrix assisted laser desorption ionization with time of flight

LC/MS/MS- liquid chromatography with tandem mass spectrometric detection

IP – immunoprecipitation

PRM- parallel reaction monitoring

The Ovod group reported that the Aβ42 concentrations and Aβ42/Aβ40 ratios were significantly lower in the amyloid-positive patients when compared with the amyloid-negative patients. The authors reported that while Aβ42/Aβ40 ratios in the CSF are decreased by approximately 50% in the presence of amyloidosis, in plasma Aβ42/Aβ40 ratios are decreased by 14.3% on average in amyloid positive relative to amyloid negative individuals. Even if the difference was smaller than that detected in CSF for amyloid-positive vs amyloid-negative groups, it was still significant and the authors were convinced that the ratio of plasma Aβ42/Aβ40 provides a sensitive and reliable measure of amyloid status not only on the date plasma samples were obtained but also that predicts future conversion to positive amyloid PET in cases where plasma positivity was detected in some individuals who were amyloid PET-negative at the time of plasma testing. It has also been shown that plasma Aβ42/Aβ40 correlates moderately well with CSF Aβ42/Aβ40 with correlation coefficients of about 0.7 (Ovod et al. 2017; Schindler et al. 2019; Bateman et al. 2019) and inversely correlates with amyloid PET with a Spearman ρ of −0.55 (Schindler et al. 2019). Thus these two mass spectrometry-based methods were able to differentiate amyloid positive from amyloid negative individuals. The Ovod group besides reporting absolute concentrations of three Aβ peptides in plasma also focused on studying production, transport and clearance of Aβ using the stable isotope label kinetics (SILK) technique and reported for the first time the kinetics of Aβ42 turnover in blood in both amyloid-positive and amyloid-negative patients. As an outcome of this kinetics study they found that the half-life of Aβ isoforms in plasma is approximately 3 times shorter when compared with CSF (3 hours in plasma vs 9 hours in CSF). The authors concluded that their findings suggest that plasma Aβ reflects central nervous system pathology of amyloidosis in a similar fashion as CSF and can be used with a high degree of sensitivity and specificity to detect AD amyloid plaques before symptom onset, as well as in symptomatic patients with an unclear clinical diagnosis (Bateman et al. 2019). The receiver operating characteristic (ROC) curve analysis results for plasma Aβ42/Aβ40 were an AUC of 0.88 (Ovod et al. 2017; Schindler et al. 2019) while for CSF Aβ42/Aβ40 ratio the reported AUC values were 0.95 or 0.93, depending on the study (Pannee et al. 2016b; Korecka et al. 2020).

In addition to Aβ42 and Aβ40, Nakamura et al. also measured the APP669-711 peptide and proposed a composite biomarker, which was generated by combining the plasma ratio of APP669-711/Aβ42 and Aβ40/Aβ42 into a composite of the two. Two studies demonstrated that this composite biomarker yielded a high area under the ROC curve ranging from 0.883-0.97 and has potential clinical utility for predicting Aβ pathology (Nakamura et al. 2018; Lim et al. 2020). The group also found that Aβ42 alone afforded moderately high AUC values ranging from 0.718 to 0.872, and two biomarker ratios, Aβ40/Aβ42 and APP669-711/Aβ42, had significantly better predictive ability compared to Aβ42 alone. Further studies will be required to confirm these findings.

Recently one study described an evaluation of the diagnostic utility of plasma Aβ biomarkers, measured by LC/MS/MS and also by SIMOA immunoassay platform (single molecule array assay, Quanterix), for prediction of brain amyloidosis (Keshavan et al. 2021). Their mass spectrometric assay included immunoprecipitation by antibody coupled to magnetic beads, 15N-labelled Aβ42, Aβ40 and Aβ38 were used as internal standards, and calibration curves were prepared by spiking 8% BSA in PBS with Aβ species. In addition to Aβ42, Aβ40 and Aβ38, Aβ−3-40 was also included in this assay to assess the utility of the previously described composite plasma biomarker (Nakamura et al. 2018). PRM-based analysis was conducted on a quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific) (see Table 2 for more details). These investigators quantified areas under the receiver operating characteristic curves to compare the concordance of the different blood tests with amyloid PET, determined blood test cut-off points, then estimated numbers needed to screen to obtain 100 amyloid PET-positive individuals as an evaluation of this plasma biomarker methodology as a screening test.

The main finding was that mass spectrometry plasma measures performed significantly better than SIMOA measures (Aβ42/Aβ40, and p-tau181). At a cut-off point of 0.095, mass spectrometry results for Aβ42/Aβ40 detected amyloid PET positivity with 86.6% sensitivity and 71.9% specificity. They also reported that screening patients with this LC/MS/MS test could reduce the number PET scans by about 50%, providing evidence for the potential value of this test as a screening tool.

Thus there is increased evidence that using mass spectrometry methodology and an immunoprecipitation step, followed with or without a protein digestion step, as the basis for preparing plasma samples for analysis can provide plasma Aβ42/Aβ40 values that are predictive of brain amyloidosis although mean differences in the ratio values are of the order of 15% when comparing plasma samples from amyloid-negative vs amyloid-positive study participants.

Further studies are underway in other laboratories building on this experience and seeking to further streamline and simplify sample preparation. Recently investigators at Araclon Biotech from Spain described an antibody-free liquid chromatography-differential mobility spectrometry-triple quadrupole mass spectrometric method for analysis of Aβ42 and Aβ40 in human plasma (Janelidze et al. 2021). In this method neither immunoprecipitation nor digestion is needed for sample preparation. This group used a proprietary step for extraction of intact Aβ peptides directly from plasma. The described method uses an M3 Micro-HPLC (Exigent) coupled to an electrospray Q-Trap 6500 mass spectrometer equipped with a SelexION and differential ion mobility spectrometry (DMS) interface and an IonDrive Turbo V Ions Source (AB Sciex). This new SelexION technology increases signal to noise ratio by reduction of background noise. Calibrators and quality control samples are prepared in human plasma using 15N-labeled Aβ42 and Aβ40, and deuterated Aβ42 (²H-Aβ42) and Aβ40 (²H-Aβ40) are used as internal standards. Response ratios corresponding to endogenous species in study samples (¹⁴N-Aβ40/²H-Aβ40 and ¹⁴N-Aβ42/²H-Aβ42) were interpolated in the calibration curve made with ¹⁵N analogues. The reported validation data are very promising, with very good linearity, accuracy and inter-assay precision (CV%) below 10% for Aβ40 and up to 11% for Aβ42. The predictive performance of Aβ42/Aβ40 for the state of brain amyloidosis using this method is under study in several different cohorts.

Tau in plasma

Tau in CSF is a second well established biomarker for AD that becomes elevated increasingly as the disease progresses and reflects tau tangle pathology. Based on the data obtained from immunoassays, several groups reported increased concentrations of plasma tau in MCI and AD patients (Mielke et al. 2017; Mielke et al. 2018), or increased concentrations of plasma p-tau181 in the same groups of patients (Tatebe et al. 2017; Mielke et al. 2018). In addition, recent studies have reported that plasma p-tau181 measurement can be used for differentiation of AD patients from other tauopaties (Janelidze et al. 2020a; Thijssen et al. 2020) and to predict the presence of tau and amyloid β pathologies (Karikari et al. 2020). Additionally plasma p-tau217, which increases during early Alzheimer's disease can be used to monitor disease progression (Mattsson-Carlgren et al. 2020; Teunissen et al. 2020) and p-tau231 distinguish AD patients from patients with non-AD neurodegenerative disorders with significantly higher accuracy than established plasma- and MRI-based biomarkers (Palmqvist et al. 2020). The latter differentiates individuals across the entire Braak stage spectrum which was not observed for plasma p-tau181 (Ashton et al. 2021). Given the growing interest in plasma tau and p-tau, driven by the upsurge of studies using sensitive immunoassay methods, one expected consequence has been the development of high sensitivity mass spectrometry methods to study tau and p-tau proteoforms in relationship to disease detection over the disease continuum. Recently one paper described a mass spectrometric method for studying plasma p-tau proteoforms, including p-tau217, p-tau181, p-tau 205 and t-tau, in connection with the progression of AD pathology (Barthelemy et al. 2020b).

A major challenge to overcome when developing an assay for tau and p-tau proteoforms in plasma is their very low concentrations (Mielke et al. 2017; Geyer et al. 2017). For example, the Barthelemy assay includes a specially designed enrichment protocol to purify and concentrate plasma tau from 20mL of plasma to 25μL of final extract. Nano-flow capillary liquid chromatography with nano-electrospray MS/MS was employed in this analytical system. No correlation was found between CSF and plasma total tau, however significant correlation was found between CSF and plasma for p-tau217 (Spearman rho 0.78) and p-tau181 (Spearman rho 0.68) suggesting that plasma p-tau217 and p-tau181 can reflect the changes in CNS tau pathology and therefore serve as a useful biomarker. This group also found that p-tau217 and p-tau181 were highly predictive of amyloid plaque pathology with ROC AUC values of 0.99-0.92 for p-tau217 and 0.98-0.75 for p-tau181, depending on the cohort. These promising results can serve to establish which of the proteoforms affords the detection of AD pathology earliest in the disease continuum and provides for the most accurate and earliest prediction of disease progression.

Apolipoprotein E

During last seven years a numerous of LC/MS/MS method for analysis of ApoE in plasma and serum was published (Simon et al. 2012; Martinez-Morillo et al. 2014b; Han et al. 2014; Hirtz et al. 2016; Blanchard et al. 2018). This approach involves a trypsin proteolysis before analysis of signature peptides carefully selected to maximize sensitivity, specificity, and stability. This methodology is suitable for quantification of total ApoE and the isoforms and also for phenotyping. One group reported that LC/MS/MS has the advantage of being fully specific, with identification of the different isoforms in blood-based matrices, can represent a valuable alternative to genotyping and can be considered as a reference method (Hirtz et al. 2016; Blanchard et al. 2018). However, reported results of total ApoE in plasma by LC/MS/MS for AD subjects are not consistent and does not qualify ApoE as a suitable diagnostic marker for AD. One group reported lower ApoE in serum of AD patients (Han et al. 2014) while two other groups did not find any differences in plasma between AD vs normal controls (Simon et al. 2012; Martinez-Morillo et al. 2014a). The lack of consistency can be connected with the presence of an oxidation-prone methionine residue in the ApoE4 specific peptide (LGADMEDVR) since the sulfur atom is prone to oxidation (Gallien et al. 2011) both in vivo (Shacter 2000) and during sample handling or storage. Additionally the presence of different alleles influences the total concentration of ApoE (Simon et al. 2012; Martinez-Morillo et al. 2014a). All these findings lead to a conclusion that plasma ApoE levels had no obvious clinical significance.

Beta-synuclein

The mass spectrometric assay for analysis of members of synuclein family in CSF described earlier in this paper, was antibody independent, however for analysis of β-synuclein in serum immunoprecipitation with anti-β-synuclein covalently coupled to epoxy-coated magnetic beads was added as an enrichment step of sample preparation (Oeckl et al. 2020). This method for absolute quantification of β-synuclein in serum by IP/MS/MS was characterized by a very good sensitivity (2pg/mL) and intra- and inter-assay precision below 8.6% and 11.7%, respectively. Results of a clinical study utilizing this assay showed increased β-synuclein levels in serum of AD patients in two independent cohorts and highlighted that β-synuclein is already increased in serum in the early phase of AD (Oeckl et al. 2020). The other observation from this study was that β-synuclein shows similar patterns of changes in blood and CSF. Both results suggest β-synuclein as a possible synaptic blood biomarker candidate for AD.

Exploratory proteomics in Alzheimer’s disease

“A particularly attractive aspect of the explorative proteomics approach is that it may lead to the discovery of biomarkers that are not based on existing hypotheses, thereby stimulating the formulation of new hypothesis on disease mechanisms”- Erik Portelius (Portelius et al. 2017). The goal of this technique is to identify exceptional biomarkers which can help to understand and monitor different processes involved in the development and progression of AD.

In the early 2000s using a classic proteomics platform- two-dimensional (2D) gel electrophoresis most often combined with MALDI/TOF mass spectrometry, a panel of new candidate biomarkers was identified in AD patient CSF and plasma. They included biomarkers related to Aβ-peptide processing and transport, neuronal cell adhesion and the inflammatory response (Sancesario & Bernardini 2018). During the past 10 years, rapid development of electrospray ionization tandem mass spectrometry has occurred, and the previously used analytical technique, 2D gel electrophoresis with MALDI/TOF was replaced by liquid chromatography with electrospray MS/MS. Relative quantification can be also performed in explorative proteomics using stable-isotope labeling or a label-free method, both extensively described elsewhere (Portelius et al. 2017; Cilento et al. 2019; Johnson et al. 2020). Quantitative MS/MS-based proteomics studies can provide the high quality data that is necessary for discovery of differences between pathological and non-pathological samples. There were a few quantitative studies reporting proteome changes in AD CSF and blood (Zhang et al. 2005; Abdi et al. 2006; Choe et al. 2007). Several AD biomarkers were studied using SELDI/TOF mass spectrometry. For example, this technique was used to assess the diagnostic utility of a few new candidates as AD CSF biomarkers like cystatin C, VGF and β−2-mocroglobulin (Carrette et al. 2003; Simonsen et al. 2008) just to mention a few of them (Korolainen et al. 2010).

In early 2016 one group reported on using label-free “shotgun” mass spectrometry to analyze the cerebrospinal fluid proteome of Alzheimer’s disease patients and non-demented controls to identify potential biomarkers for Alzheimer’s disease (Khoonsari et al. 2016). They successfully verified significantly lower levels (p<0.05) of eight proteins (A2GL, APOM, C1QB, C1QC, C1S, FBLN3, PTPRZ, SEZ6) in Alzheimer’s disease compared to controls. These proteins are hypothesized to be involved in different biological roles ranging from cell adhesion and migration, to regulation of the synapse and the immune system.

To discover candidate protein biomarkers for early diagnosis of AD in the preclinical stage, label-free based discovery proteomics was employed for measurement of protein changes between healthy and preclinical AD individuals. As a result of this study 732 and 704 proteins with more than one unique peptide were identified in healthy controls and preclinical AD patients, respectively (Zhong et al. 2018).

In 2018 a very interesting paper described an exploratory proteomics study to identify markers of cerebral amyloid angiopathy (CAA) (Hondius et al. 2018). Cerebral amyloid angiopathy is a type of vascular disease present in more than 50% of demented elderly and more than 80% of Alzheimer's disease patients. Both CAA and AD are characterized by extracellular Aβ deposits with the distinction that CAA has vascular deposits while AD has only amyloid plaques. The group performed an exploratory laser dissection-assisted LC/MS/MS analysis of AD brain tissue exhibiting severe CAA type-1 pathology (which refers to amyloid deposition in both capillaries and larger vessels), AD brain tissue without apparent involvement of CAA, and control brain tissue without AD related pathology. They showed that the proteome of CAA type-1 is different from that of parenchymal plaque pathology in AD, which led to the identification of proteins selectively associated with CAA. Next to identification of new CAA selective proteins, this study also confirmed the presence proteins already known to be involved in CAA pathology, e.g., CLU, ApoE and APCS. The findings in this study can contribute to studies investigating the role of CAA in AD pathology. One year later, in 2019 a different group used mass spectrometry with immunoprecipitation and showed that the Aβ peptide pattern differed greatly between subjects with no CAA compared to subjects with CAA (Gkanatsiou et al. 2019).

Traditional discovery (unbiased) proteomic approaches for biomarker discovery have struggled to detect low-abundance markers due to the high dynamic range of proteins, the predominance of hyper-abundant proteins, and the use of data-dependent acquisition mass spectrometry. Tandem-mass-tag (TMT) as a novel quantification strategy overcomes this limitation. (Russell et al. 2017). This new technology is designed to ensure that identical peptides labeled with different TMTs exactly co-migrate in all separations. The basic version of TMT marker is comprised of a pair of tags which binds to identical peptides and this pair of TMT-labelled peptide co-elutes as a single peak, which increases the method’s sensitivity when compared with conventional deuterated-based isotope labeling (Thompson et al. 2003). Using this new labeling technique connected with an Orbitrap Fusion Tribrid mass spectrometer, one group was able to report peptides from 41 proteins that had not previously been detected in CSF (Russell et al. 2017). Another group, using a newer generation mass spectrometer, Orbitrap Fusion Lumos Tribrid MS, identified 2,327 proteins in the CSF of which 139 (for example NPTX2 and VGF) were observed to be significantly different in their abundance in the CSF of AD patients versus controls (Sathe et al. 2019).

One of the papers published in 2020 described a bottom-up approach to monitor how proteins from human cerebrospinal fluid associate with Aβ amyloid fibrils form plaque particles (Chaudhary et al. 2020). The study by Chaudhary et al. identified and quantitated 128 components of the captured multiprotein aggregates. The results provide insights into the functional characteristics of the sequestered proteins and revealed distinct interactive responses for the two investigated Aβ peptides, Aβ42 and Aβ40.

A recent study provided results for one of the largest, thus far, proteomics studies on AD (Johnson et al. 2020). In this study more than 2,000 brain tissue samples and nearly 400 cerebrospinal fluid samples were analyzed by quantitative proteomics. For exploratory proteomics purposes they used a few techniques, including a label-free MS/MS approach on a Q-Exactive Plus mass spectrometer (Thermo Fisher) and also TMT-MS/MS (Orbitrap Fusion mass spectrometer, Thermo Fisher) methodology coupled to a nano-liquid chromatograph. The group identified 5,688 proteins in AD brains some of which can serve as therapeutic targets and fluid biomarkers for detecting and monitoring the disease. A major outcome of this study was this group’s finding that biomarkers of astroglial/microglial metabolism were the most significantly associated with AD compared to the other biological processes and functions defined in this study. The biomarker members of this “module” all increased in asymptomatic AD and AD suggesting that the proteins in this module can serve as useful biomarkers for staging AD, detect disease early in the continuum and for development of novel therapeutic approaches to this disease.

Endopeptidomics is an another way of identification of biomarkers for diagnosis, monitoring of disease progression and therapeutic intervention. It focuses on studying endogenous polypeptides present in CSF, without need for a sample trypsinization step (Holtta et al. 2015) (Hansson et al. 2017).

The above-mentioned proteomics studies represent only a small part of extended work on the proteome in AD and normal individuals. As of January 2021 there are 78 papers available on PubMed under CSF, AD, proteomics, mass spectrometry search criteria and published during the last 5 years.

In summary, exploratory proteomics and endopeptidomics studies have been used for identification of new AD biomarkers, however further validation of biomarkers, identified as significant candidates for early detection and disease progression, is a key requirement. This can be achieved by assessing these candidates using targeted proteomics technologies.

Closing remarks

During the last 30 years two analytical platforms, immunoassays and mass spectrometry, have been effectively used in the research on fluid (CSF and plasma) biomarkers for AD. Currently, routine clinical measurement procedures for core CSF biomarkers Aβ42/Aβ40, total tau and phosphorylated tau are based on immunoassays. However, the data collected by the quality control program of the Alzheimer’s Association has documented large variations in the results of Aβ42 obtained from different immuno-platforms. There are several factors that can impact final results including pre-analytical factors and lack of the standardization of the immunoassays. Two reference measurement procedures and three certified reference materials for CSF Aβ42 were developed using mass spectrometry methodology, without antibody enrichment, highlighting the importance of this analytical technique. The developed CRMs were used to re-calibrate commercial immunoassays for CSF Aβ42 from Roche Diagnostics, Fujirebio and EUROIMMUN. Promising results were obtained from this preliminary study. Following re-calibration of each system with CRMs, high consistency was observed among the obtained results with significantly reduced bias across platforms at the median concentration of 700pg/mL (Boulo et al. 2020). As a follow-up, the Global Biomarkers Standardization Consortium has organized an additional study to confirm the findings. If the preliminary results are confirmed, harmonization of different assays and establishment of common cut-off value for CSF Aβ42 could be achieved finally. Otherwise, mass spectrometry can be used as an alternative technique for routine analysis of CSF Aβ42, if needed. A Round Robin study performed in 2016 (Pannee et al. 2016a) showed that the average inter-laboratory CV for 4 participating laboratories was 12.2%, so definitely lower than for the earlier manually performed immunoassays. Since at that time CRMs were not available another approach has been used to harmonize the results and after this correction the average inter-laboratory CV was lowered to 8.3%. Of course this result requires confirmation.

As is the case for CSF, currently available platforms for analysis of AD core biomarkers in plasma also includes immunoassays and mass spectrometry. Immunoassays struggle with high complexity of the matrix and low concentration of compounds of interest, but results for ptau proteoforms and NfL for example using the SIMOA platform are promising as is the highly automated Roche Elecsys platform (Palmqvist et al. 2019). Recent developments in LC/MS/MS analytical instruments with increased sensitivity of detection made it possible to use this technique for analysis of low abundant compounds so in this field mass spectrometry can serve both as reference method for standardization of immunoassays and for studies of dpost-translation modifications such as the degree of phosphorylation over time. There are reports that currently obtained concentrations in plasma using different analytical methods are not translatable across platforms (Budelier & Bateman 2020) and there is a need for reference method procedure and certified reference materials for measurements of AD biomarkers in plasma, as it was true case for CSF, and for this purpose again mass spectrometry could be used. MS/MS is also an excellent technique for efficient, time and cost effective verification of new biomarker candidates and a leading platform for simultaneous detection of multiple biomarkers. The novel biomarker candidates identified by exploratory proteomics must be validated before they can move to clinical testing. Development of quantitative immunoassays for the purpose of this validation is limited by the availability of specific antibodies and mass spectrometry as an antibody independent technique overcomes this concern.

In summary, mass spectrometry plays a very important role in a field of AD fluid biomarkers. It is very efficient tool for the search for novel biomarkers and provides for the possibility of time and cost-efficient validation. Mass spectrometry is a unique candidate to serve as a reference method procedure and can help to develop certified reference materials to harmonize assays. This technology can be easily used for simultaneous analysis of multiple compounds and different pathological forms what could enable more rapid and efficient research on AD and improve the differentiation between AD and other dementias (Cilento et al. 2019). And in case of any doubts in the science on AD biomarkers researchers usually reach out to mass spectrometry for help. This technique can also serve as a platform for routine analysis of AD fluid biomarkers. However, for this purpose more effective sample preparations procedures, preferably antibody-independent, and more automated MS/MS platforms with improved sensitivity, LC separation and high throughput are needed.

Abbreviations:

AD

Alzheimer’s disease

CSF

cerebrospinal fluid

LC-MS/MS

liquid chromatography with mass spectrometric detection

MCI

mild cognitive impairment

PSEN1/2

preseline ½

APP

amyloid precursor protein

ApoE

apolipoprotein E

Aβ

amyloid beta

AT(N)

amyloid beta, tau, neurodegeneration research framework

NIA-AA

National Institute on Aging and Alzheimer’s Association

MRI

magnetic resonance imaging

PET

positron emission tomography

DLB

dementia with Lewy Body

SNAP 25

synaptosomal-associated protein 25

SYT-1

synaptotagmin-1

NPTX2

neuronal pentraxin 2

VGF

neuronal secretory protein

LAMP2

lysosomal protein 2

TREM2

triggering receptor expressed on myeloid cells 2

t-tau

total tau

MS/MS

mass spectrometry

p-tau

phosphorylated tau

CV

coefficient of variation

QC

quality control

CRM

certified reference material

ADNI

Alzheimer’s Disease Neuroimaging Initiative

ELISA

enzyme-linked sandwich immunoassay

SIMOA

single-molecule array technique

ESI

electrospray

APCI

atmospheric pressure chemical ionization

MALDI

matrix assisted laser desorption ionization

SELDI

surface enhanced laser desorption ionization

SRM

selected reaction monitoring

MRM

multiple reaction monitoring

PRM

parallel reaction monitoring

LC

liquid chromatography

JCTLM

The Join Committee for Traceability in Laboratory Medicine

BSA

bovine serum albumin

API

atmospheric pressure ionization

MCX

mixed-mode cation exchange

SPE

solid phase extraction

SILK

stable isotope label kinetics

TOF

time of flight

FNIH

The Foundation for the National Institutes of Health

UPLC

ultra pressure liquid chromatography

ROC

receiver operating characteristics

AUC

area under a curve

CAA

cerebral amyloid angiopathy

TMT

tandem-mass-tag