Use of biomarkers and imaging to assess pathophysiology, mechanisms of action and target engagement

Abstract

Multidisciplinary basic research led to an evolving knowledge of the molecular pathogenesis of Alzheimer's disease (AD). These advances have been translated into defined therapeutic concepts and distinct classes of compounds with putative disease-modifying effects that are now being tested in clinical trials. There is a growing consensus that disease-modifying treatments may be most effective when commenced early in the course and progression of AD pathophysiology, before amyloid deposition and neurodegeneration become too widespread. Biological indicators of pathophysiological mechanisms are required to chart and identify AD in the prodromal phase or, preferably, in asymptomatic individuals. Biomarkers are becoming even more important, owing to the challenges in demonstrating efficacy of candidate-drugs that hit pathophysiological targets using clinical and cognitive outcomes in early AD trials with limited duration. Currently, there is emerging consensus that advances in therapeutic strategies for AD that delay predefined milestones or slow the cognitive and disease progression would considerably decrease the expanding global burden of the disease. To effectively test preventive compounds for AD and bring therapy to affected individuals as early as possible there is an urgent need for a concerted collaboration among worldwide academic institutions, industry, and regulatory bodies with the aim of establishing networks for the identification and qualification of multi-modal biological disease markers.

Introduction

There is evolving evidence that distinct pathophysiological mechanisms in genetically complex sporadic Alzheimer’s disease (sAD) – hypothesizing a chronic non-linear-dynamic cascade – somehow commence earlier in life, years to decades before consecutive onset of first clinical symptoms, conveying a diversity of converging and diverging, adaptive to maladaptive, compensatory to non-compensatory neurophysiological, molecular/cellular mechanisms, ultimately leading to complex brain “systems failure” (1, 2). Consistent with the “amyloid cascade hypothesis” (3., 4., 5.), derived from monosystem autosomal dominant genetic amyloid precursor protein (APP) and presenilin mutation models, AD begins with anomalous processing of the App leading to soluble amyloid-beta (Aβ) peptides (6), aggregating into oligomers and fibrillar forms and deposited into amyloid plaques. Microtubule associated tau protein is abnormally hyperphosphorylated and forms paired helical filaments and neurofibrillary tangles (NFTs) (7). Together with associated hypothesized “downstream” mechanisms – inflammation and oxidative stress - these pathophysiological cascades interact somehow and contribute to progressive loss of functional, structural and metabolic synaptic/neuronal/network integrity, progressive neurodegeneration, atrophy and cognitive impairment (8., 9., 10.).

Advances in basic AD research are currently translated into conceptualized treatment strategies with distinct target engagement and disease- or mechanism-modifying potential. Candidate anti-Aβ compounds are marching through matured phases of clinical trials. Despite enormous financial and scientific efforts, however, still no approved disease-modifying therapies exist for AD. Disappointments in therapeutic clinical trials in mild-to-moderate AD, such as the two recent Phase III study failures of bapineuzumab (11), force researchers to re-examine the way of designing trials. Stressing the confidence that phase II clinical trials will translate into successful phase III trials, and that this is a global problem in the neuroscience field requiring multi-national solutions, a European Union/North American Task Force of experts from academia, industry, regulatory agencies, and private foundations was summoned (12). Besides learning from inconclusive trials, the task force investigated the role of biomarkers in prevention trials. A concern was that the timing of therapeutic trials, treating patients with established mild-to-moderate AD, may be too late in the disease process to impact and improve clinical/functional outcomes. Furthermore, new diagnostic criteria for AD – from the International Working Group for New Research Criteria for the Diagnosis of AD (13, 14) as well as from the National Institute on Aging and Alzheimer’s Association initiative (NIA-AA) (15., 16., 17.) – were proposed. They integrate the use of biomarkers as required and defining supportive criteria for preclinical stages of AD (17). Therefore, the traditional definition of the disease is currently extended to prodromal and the very early asymptomatic stages that may be a more feasible target for disease-modifying therapeutic interventions.

Categories of Biomarkers

During the past decades, biomarkers have attained acknowledged importance in pharmaceutical drug development across all therapeutic areas. Supported by recent progress in imaging, molecular diagnostics, and pharmacogenomics as well as in elucidating the pathophysiology of diseases at the molecular level, biomarkers are promising to improve all phases of a drug discovery/development program by allowing validation of mechanism of action (MoA), stratification of patients to enhance clinical response, identification of prognostic signatures, development of diagnostic assays, monitoring the progression of disease and response to therapy. Accordingly, biomarkers are now regularly incorporated in research and development (R & D) strategies for searching new candidate drugs, from drug discovery/development to regulatory approval. Biomarkers are essentially the “tools for escorting the drug candidates from bench to bedside” (18). They can be sorted in different classes: (I) target engagement biomarkers, testing the hypothesis on the interaction of the candidate drug with its molecular target; (II) MoA biomarkers, measuring downstream biological effects; (III) outcome biomarkers, assessing both efficacy and safety, (IV) safety biomarkers, assessing and predicting tolerability and adverse side effects.

Biomarkers used to identify and monitor the biochemical effects of drugs are called theragnostic biomarkers (9, 10). These may be valuable in drug development to bridge the gap between studies with animal models and large clinical trials by evaluating whether a drug has a true disease-modifying effect in small-scale clinical studies. Thus, only the most promising drug candidates would be designated for further investigation to improve the success rate of larger Phase II/III trials. Theragnostic biomarker trials should be feasible in AD, as cerebrospinal fluid (CSF) concentrations of total tau (t-tau), phospho-tau (p-tau), and the 42-amino acid fragment of Aβ (Aβ1-42) have shown low intra-individual variability in longitudinal samples (19, 20). However, only preliminary evidence suggests that CSF biomarkers might be useful in detecting/monitoring biochemical effects of novel compounds against AD related mechanisms (10, 21, 22). Data from biomarker studies may raise confidence that a novel compound truly engages with the specific molecular target and that, based on the generated hypothesis, should hopefully alter the specified pathophysiological mechanism, which again should optimistically result in quantifiable efficacy signals at a certain stage of the disease during a certain duration of testing. If efficacy is not observed, then the failure of the study can be attributed to a failure of this approach from a mechanistic perspective (i.e. targeting the specific molecular target is not effective in impacting the pathophysiological mechanisms which again is insufficiently impacting complex downstream surrogate effects, such as disease symptoms, e.g. cognition and function) rather than a failure in the study design (e.g. selection of an inappropriate dose to adequately test the hypothesis). Such studies are particular important when investigating novel, unprecedented, targets (23).

When a candidate drug is selected for development, it seems necessary to get access to literature-based data as well as pre-clinical data by which it is possible to develop a hypothesis for the MoA of the candidate drug. MoA biomarkers are often linked to pharmacodymanic markers that enable the establishment of a chain of molecular events, known in literature as the “pharmacological audit trail” (24, 25) starting from the entry of the drug into the blood stream, to the target tissue, to the impact on cellular/molecular targets and, therefore, on the pathophysiological cascade of the target disease. Thus, these markers are also known as proof-of-principle (POP) markers since they help establish the principle for the potential activity of a drug via an intended pathway (26).

Neurobiochemical Markers in AD

The currently maturing status of multimodal core biomarker development and validation integrating neurobio chemistry/neurogenetic and structural/functional/metabolic neuroimaging studies has been extensively reviewed providing the converging perspectives of industry stakeholders and regulatory bodies on the AD biomarker discovery/development area (1, 8., 9., 10., 27., 28., 29., 30.).

The most important markers under examination and their association with the pathological processes in AD can be clearly divided into two categories: (I) core markers that reflect crucial neuropathological events in AD such as deregulated metabolism of Aβ and APP; (II) downstream markers that reproduce secondary alterations to brain structure/function, i.e. volumetric and metabolic changes to temporo-medial structures (31) (Table 1).

Table 1.

Markers under examination for Alzheimer’s disease

Pathogenic correlates	Category	Associated markers
Amyloidogenic pathway of APP metabolism	Cerebrospinal fluid	Aß1-42 concentration
	In vivo molecular imaging	Intracerebral load of Aß (11C-PiB-PET)
Tau phosphorylation and neuronal/axonal degeneration	Cerebrospinal fluid In vivo molecular imaging	T-tau concentration P-tau181 and p-tau231 concentration Intracerebral aggregates of NFTs (18F-FDDNP-PET)
Secondary alterations to brain structure/ function	Structural neuroimaging	Regional (medial temporal) atrophy Volumetry of hippocampus/entorhinal cortex Rate of brain/regional atrophy Voxel-based morphometry (VBM) Diffusion tensor imaging (DTI)
	Functional neuroimaging	Functional MRI Metabolic alterations (18F-FDG-PET) Regional perfusion (SPECT)

Abbreviations: 18F-FDDNP-PET, 2-(1-{6-[(2-[18F]fluoroethyl)(methyl)amino]-2-naphthyl}ethylidene)malononitrile; 11C-PiB, 11C-labeled Pittsburgh Compound-B, 18F-FDG, 2-deoxy-2-(18F)fluoro-D-glucose; APP, Amyloid precursor protein; Aß, Amyloid-beta, MRI, magnetic resonance imaging; NFTs, neurofibrillary tangles; PET, positron emission tomography; SPECT, single photon emission computed tomography.

Several studies demonstrated that CSF concentrations of Aβ1-42, t-tau, and p-tau reflect central elements of the underlying AD process. Together, these validated core feasible biomarkers identify and predict AD pathophysiology in prodromal mild cognitive impairment (MCI) with a sensitivity and specificity of 75% to 95% (10). AD patients exhibit a significant reduction in CSF Aβ1-42 levels (6) and increases in t-tau and p-tau concentrations (32) showing dynamic signals as early as in the asymptomatic and expanding to the prodromal to dementia stages. The combined use of t-tau, p-tau, and Aβ1-42 exhibits an optimized predictive value for identifying subjects of prodromal AD in MCI subjects (33). The high diagnostic performance of these core feasible “gold standard” CSF markers has been confirmed in large multinational multi-center validation studies (1) including the Worldwide Alzheimer’s Disease Neuroimaging Initiative (WW-ADNI) (8), the DESCRIPA study (34), and the Swedish Brain Power initiative (10).

Recently, Buchhave and colleagues suggested that perturbations in Aβ metabolism precede tau-related pathology and neuronal degeneration (35). Further evidence on the order and dynamics of biomarkers in AD pathogenesis was provided by Jack and colleagues (36., 37., 38.). Many families carrying APP, presenilin-1 (PSEN1), or presenilin-2 (PSEN2) mutations – typically characterizing the Mendelian autosomal dominantly inherited form of AD (ADAD) – have been examined to detect the time course of dynamic fluid biomarker changes, neuroimaging signal modifications, and clinical alterations before the presumed onset of AD symptoms. In agreement with earlier hypotheses preliminary studies of a familial AD cohort, performed by the Dominantly Inherited Alzheimer Network (DIAN), suggest that Aβ1-42 levels in CSF begin to decline as early as 25 years before expected symptom onset (39). This is followed by the appearance of fibrillar amyloid deposits in the brain, increased levels of tau in CSF, and progressive brain atrophy roughly 15 years before expected symptom onset (39). Cerebral hypometabolism and subtly impaired episodic verbal memory seem to begin some 10 years or so before expected symptoms (39). These results derived from the first cross-sectional analysis suggest that a clinical diagnosis of dementia will be established at a very late stage in the course of ADAD pathophysiological progression, which fits closely with existing hypothetical models of disease and biomarker progression in sAD (36., 37., 38.). Bateman and colleagues state, however, despite the possibility of pathophysiological commonalities between ADAD and sAD, that the timing and order of biomarker and psychometric changes might differ between these genetically distinct AD subtypes (39). Intriguingly, different ADAD variants and sAD subtypes might represent distinct entities exhibiting individual pathophysiological dissimilarities that result in different multimodal biomarker profiles and progression patterns, thus requiring differently adapted therapy regimens over time to be most effective or effective at all (30).

Compared with established core feasible CSF biomarkers, attempts to disclose reliable AD biomarkers in peripheral blood have been of limited success. There are publications on potential candidate blood biomarkers; however, follow-up studies are either lacking or failed to confirm and validate the diagnostic value (9). Although plasma Aβ1-42 and Aβ1-40 have been measured in peripheral blood, they do not contribute to the identification of AD in a reproducible manner (40) and are unlikely to reflect Aβ processing in the central compartments of the brain (41). In light of this, most research has focused on CSF biomarkers, which more directly reflects brain neurobiochemistry, and can be obtained through lumbar puncture without substantial side effects (42). Nevertheless, the successful discovery, development, validation and qualification of blood-serum-based biomarkers would be a very significant step forward, allowing more generalized, widespread and frequent use worldwide in both trials and clinical practice. The collection of biological material including blood-serum and cells for RNA, should be integrated wherever possible into clinical trial protocols throughout all stages (43).

Major Pathophysiological Mechanisms and Patterns of Neurobiochemical Markers in AD

To date there exists no valid overarching integrative global hypothesis of the complex systems biology of the genetically complex sAD with a valid consideration of its largely unknown, still dark, long and winding asymptomatic disease stages across many years and decades. Various hypotheses for mono-system mechanisms, however, have been proposed to explain aspects of the biology of AD. While the most widely accepted is the amyloid cascade hypothesis, other models have also received independent or even competitive consideration. Some of them are linked to the hypothesized central pathophysiological cascade in AD and promise to mirror major fundamental pathogenic mechanisms (amyloid cascade or abnormal phosphorylation of tau protein), whereas others have been related to seemingly “non-specific” or “downstream” pathological disease processes (e.g. inflammation, oxidative stress, apoptosis, altered lipid signaling) (Table 2). Some of these concepts are currently put to the test with numerous mechanistic compounds which march through the stages of clinical development. Around the individual disease concepts, hypothesis-driven pathophysiological candidate biomarkers are being developed. They are complemented by novel biomarker candidates generated by hypothesis-free exploratory research (e.g. from the proteome-based technologies).

Table 2.

Pathophysiological mechanisms and patterns of neurobiochemical markers in Alzheimer’s disease

Pathogenic process	Biomarker	Changes in biomarker level	Major reference	Usefulness as theragnostic marker	Target engagement
Brain amyloid pathology and APP metabolism	CSF Aβ1-42	Marked reduction in levels in AD and prodromal AD	Hampel et al., 2010a	Yes	A(3 immunotherapy; BACE1 inhibitor; y-secretase inhibitor; A(3 aggregation inhibitor; a-secretase potentiation; Anti-fibrillization agents and chelators
	CSF Aβ1-40	No major changes in AT)	Hansson et al., 2007	Yes	A(3 immunotherapy; BACE1 inhibitor; y-secretase inhibitor; A(3 aggregation inhibitor; a-secretase potentiation; Anti-fibrillization agents and chelators
	CSF ratio A(31-42/A(31-40	Marked reduction in levels in AD and MCI. It maygive a slightly higher diagnostic accuracy than A(31-42 alone	Hansson et al., 2007	Yes	A(3 immunotherapy; BACE1 inhibitor; y-secretase inhibitor; A(3 aggregation inhibitor; a-secretase potentiation; Anti-fibrillization agents and chelators
	CSF BACE1	Increase in levels and activity in AD and MCI progressing to AD	Hampel & Shen, 2009	Yes	BACE1 inhibitor
	CSF sAPP and sAPP|3	No change or marginal increase in levels in AD and MCI. Farge overlap with health}’ elderly	Zetterberg et al., 2009	Yes	BACE1 inhibitor
	CSF A(3 oligomers	Increase in levels in AD	Fukumoto et al., 2010	Yes	A(3 immunotherapy; A(3 aggregation inhibitor
Neurofibrillar}’ tangle pathology	CSF p-taul81	Marked increase in levels in AD and prodromal AD	Hampel et al., 2010b	Yes	Inhibitors of the tau kinase GSK3β; Tau aggregation inhibitors
Cortical neuronal and axonal damage and degeneration	CSF t-tau	Marked increase in levels in AD and prodromal AD	Hampel et al., 2010b	Yes	Tau aggregation inhibitors
	CSF VIFIP-1	Increase in levels in AD. VIFIP-l/A(31-42 predicts progression from cognitively normal to MCI/very mild or mild AT)	Tarawneh et al., 2012		Unknown
	CSF NFL proteins	Normal in AD. Increase in disorders with subcortical pathology (e.g.: vascular dementia, normal-pressure hydrocephalus, frontotemporal	Sjogren et al., 2001	dementia)	Unknown
Neuroinflammation and immunological mechanisms	CSF TGF-P	Increase in levels in AD	Zetterberg et al., 2004	−	Unknown
	CSF TNF-a	Increase in levels in AD	Shen et al., 2006	−	Unknown
	CSF TACE	Increase in activity in AD. Higher activity in MCI vs. AD	Jiang et al., 2011	.	Unknown
	CSF sIF-6R	Reduction in levels in AD	Hampel et al., 1999	−	Unknown
	CSF YKL-40	Increase in levels in AD. YKL-40/A(31-42 predicts progression from cognitively normal to MCE very mild or mild AT)	Craig-Schapiro et al., 2010	−	Unknown
	CSF inflammatory proteins (NrCAM, YKL-40, chromogranin A, camosinase I)	They might improve the diagnostic accuracy of AJ31-42 and tau	Perrin et al, 2011		Unknown
	CSF inflammatory proteins signature	They might complement the best current CSF biomarkers to differentiate very mildly/mildly demented from cognitively healthy subjects.	Craig-Schapiro et al., 2011		Unknown
	Plasma inflammatory proteins signature	Eighteen proteins involved in the immune response, Ray et al, 2007 hematopoiesis, and apoptosis that distinguished AD from healthy control and identified MCI progressing to AD		Unknown
	Plasma complement factor H and α2-macroglobulin	Increase in levels in AD	Hye et al.,2006		Unknown
	Serum inflammatory proteins signature	Serum protein-based algorithm to separate AD from healthy controls	O’Bryant et al., 2010; 2011	−	Unknown
Oxidative stress	CSF F2-isoprostanes	Increase in levels in AD, prodromal AD, and asymptomatic carriers of familial AD mutations	Montine et al., 2007	−	Unknown
	Plasma F2-isoprostanes	Conflicting results	Montine et al., 2007	−	Unknown
Integrity of the blood-brain barrier	CSF/serum albumin ratio	Unchanged in pure AD. Increase in CNS infections, inflammatory disorders, brain tumors, and cerebrovascular disease (including vascular dementia)	Blennow et al., 1990		Unknown
Microvascular homeostasis	Plasma MR-proADM Plasma MR-proANP Plasma CT-proET-1	Increase in levels in AD Reduction in levels in AD	Buerger et al., 2011	−	Unknown Unknown Unknown

Abbreviations: AD, Alzheimer’s disease; APP, Amyloid precursor protein; A(3, Amyloid-beta; BACE1, β-site amyloid precursor protein-cleaving enzyme 1; CSF, cerebrospinal fluid; CT-proET-1, C-terminal endothelin-1 precursor fragment; GSK3β, Glycogen synthase kinase 3; MCI, mild cognitive impairment; MR-proADM, Midregional pro-adrenomedullin; MR-proANP, Midregional pro-atrial natriuretic peptide; NEL, Neurofilament light protein; sAPPα, α-cleaved soluble APP; sAPPβ, β-cleaved soluble APP; sIL-6R, Soluble interleukin-6 receptor; TACE, TNF-α converting enzyme; TGF-β, Transforming growth factor-beta; TNF-α, Tumor necrosis factor-α; VILIP-1, Visinin-like protein-1.

Brain amyloid pathology and APP metabolism. Aβl-42 is a marker of brain amyloid pathology (44., 45., 46.). Longitudinal studies in humans suggest that certain forms of Aβ may act as central initiators of the core disease process with toxic effects at the synaptic level (47). Based on this knowledge, novel treatments aimed at inhibiting Aβ toxicity have been developed and are being tested in patients (34). These encompass: (I) secretase inhibitors and modulators, affecting the production of Aβ from APP; (II) immunotherapy, aimed at increasing the clearance of soluble and/or insoluble Aβ from the brain; (III) Aβ aggregation inhibitors that should prevent pathological build-up of the peptide in the brain (22, 48); (IV) potentiation of the neurotrophic a-secretase (non-amyloidogenic) pathway (49).

Given the existence of many factors affecting CSF Aβ (including the many sub-species and aggregating forms) concentrations (production, aggregation, enzymatic clearance), it is difficult to predict what different amyloid-targeting treatment paradigms might do to CSF Aβ levels at particular stages of the disease. To date, data from animal experiments show that y-secretase inhibitor treatment results in a decrease of cortical, CSF, and plasma levels of Aβ (50, 51). Likewise, treatment of monkeys with a BACE1 inhibitor reduced Aβ CSF levels (52).

It is not yet clearified how and to what extend CSF Aβ1-42 concentrations react to treatment regimen with anti-Aβ drugs in AD patients at different disease stages. Aβ is quickly precipitated by Zn2+; Cu2+ and Fe3+ also promote manifest Aβ aggregation, but only under mildly acidic conditions (53), such as those supposed to occur in AD brain. The precipitation of Aβ by these ions is reversible with chelation, in contrast to fibrillization, which is irreversible. In this regard, the compound named PBT2 (Prana Biotechnology) was designed to alter the course of AD by preventing metal-dependent aggregation, deposition and toxicity of Aβ. PBT2 acts at three levels of the “amyloid cascade”: inhibiting the development of toxic soluble oligomers, preventing deposition of Aβ as amyloid plaques, and inducing clearance by mobilizing and neutralizing Aβ from existing deposits (22, 54). A recent Phase IIa study of PBT2 revealed a substantial dose-dependent decrease in CSF Aβ1-42 levels during treatment (55). Data from a clinical study on the amyloid-targeting compound phenserine also displayed changes in CSF Aβ levels in response to treatment (56). However, no significant treatment result on CSF Aβ1-42 was detected in the interrupted Phase IIa AN1792 trial of active Aβ immunization (57). Additionally, a clinical study on a y-secretase inhibitor failed to identify any effect on CSF Aβ1-42 levels (58). Nevertheless, when the impact of this drug on Aβ production rate was assessed through a stable isotope-labeling kinetic (SILK) technique, an inhibitory effect of y-secretase inhibition on Aβ production was shown (59).

The most abundant CSF Aβ isoform is Aβ1-40. Although there is no major change in Aβ1-40 CSF levels in AD, there is a marked decrease in the Aβ1-42/Aβ1-40 ratio in AD and MCI, more prominent than the reduction in Aβ1-42 alone (60). Shorter carboxy-terminally truncated Aβ isoforms in CSF have been identified/quantified via immunoproteomic approaches (61). Intriguingly, shorter Aβ peptides (Aβ1-14, Aβ1-15, Aβ1-16) should characterize an original APP-processing pathway that is upregulated in a dose-dependent manner in response to y-secretase inhibition (62).

Other CSF candidate biomarkers reproduce pathogenic processes related to Aβ and APP metabolism such as the concentration and activity of the β-site amyloid precursor protein-cleaving enzyme 1 (BACE1), the a- and β-cleaved soluble APP (sAPPa, and sAPPβ), and Aβ oligomers (63., 64., 65.). These biomarkers appear to provide information of limited diagnostic usefulness. However, they might be important for identifying treatment effects of drugs that are meant to inhibit β-secretase or break up amyloid aggregates.

Hyperphosphorylation of tau protein and neurofibrillary tangle pathology. Microtubule-associated tau protein expression is high in non-myelinated cortical axons where it acts as a microtubule-stabilizing protein. Hyperphosphorylation of tau causes the protein to detach from the microtubules. This process promotes axonal and synaptic plasticity in the developing brain, but is pathological in the adult and associated to several tauopathies, including AD. Elevated CSF levels of p-tau are the most specific finding suggesting an on-going AD process and reflect the phosphorylation state of tau and the formation of tangles in the brain (32). Inhibiting tau phosphorylation or aggregation is a strategy to slow down the neurodegeneration process in AD. Drug candidates intervening in tau-related disease mechanisms (e.g., inhibitors of the tau kinase enzymes such as glycogen synthase kinase 3 (GSK3p) and tau aggregation inhibitors) are present but still in earlier phases of developments (48). In this regard, the role of lithium, a mood stabilizers widely used in the chronic treatment of bipolar disorders, has been investigated. Lithium is thought to inhibit GSK3p thus reducing tau phosphorylation. Nevertheless, a randomized clinical trial with lithium (10 weeks, including a 6-week titration phase) in patients with mild AD did not show significant cognitive benefit or change in CSF biomarkers, including p-tau, t-tau, and Ap1-42 (66). However, an effect of lithium may depend upon how the treatment is operationalized in AD. Future studies may assess a potential therapeutic effect of lithium for the treatment of AD using extended treatment and observation periods with higher dose levels. A combination of lithium with other potential GSK3p inhibiting drugs, (e.g., valproic acid), may be a fruitful approach to increase the effect of lithium (66).

Cortical neuronal and axonal damage and degeneration. Pathogenic processes that damage axons in the cortex result in increased CSF levels of t-tau. This is a dynamic marker of the intensity of neuronal/axonal damage and degeneration irrespective of pathognomonic aetiology (67). Given longitudinal studies of conditions involving acute neuronal injury and brain trauma (68) and data from the interrupted Phase IIa AN1792 trial (57), abnormally increased t-tau would be expected to dynamically “normalize” (decrease) if a treatment is successful in substantially slowing or inhibiting the neurodegenerative process in AD.

Other neuronal/synaptic proteins may represent valuable CSF biomarkers since they provide information on cognitive function and disease progression. A recent study of visinin-like protein-1 (VILIP-1) demonstrated that VILIP-1 CSF levels increased significantly in AD. The VILIP-1/Aβ1-42 ratio might help in predicting progression from cognitively normal to very mild dementia and in guiding prognostic assessment in therapeutic trials (69).

Effects on axonal degeneration by disease-modifying compounds could be monitored by using the neurofilament light (NFL) proteins that are structural components of large myelinated axons. NFL proteins are established CSF biomarkers for sub-cortical axonal degeneration/damage and may help differentiate among AD, frontotemporal dementia, and sub-cortical dementia disorders (70).

Neuroinflammation and immunological mechanisms. Accumulated evidence supports the concept that neuroinflammation and immunological mechanisms are involved in AD neuropathogenesis (1). Pilot studies showed increased CSF levels of transforming growth factor-beta (TGF-β) in AD as compared with controls (71, 72).

Tumor necrosis factor-α (TNF-α) elicits inflammatory responses by recruiting microglia or astrocytes to lesion sites, leading to glial cell activation. The TNF-α receptor complex and its functional proteins are involved in AD pathophysiology, linking inflammation pathways with the core amyloid deposition cycle in a chronically destructive and self-propagating dynamic (73).

Significantly higher levels of TNF-α converting enzyme (TACE) activity and soluble TNF-α receptors were shown in MCI subjects vs. AD patients. They represent an early event in AD pathophysiology, with declining mechanistic disease activity from the asymptomatic to the prodromal and to the established AD dementia stage (1). They might be complementary diagnostic markers during the prodromal MCI and AD dementia stages (74) and act as promising markers of target engagement, mechanism of action, and outcome in therapeutic trials (1).

The combined analysis of CSF tau with the soluble interleukin-6 receptor complex (sIL-6RC) added more accuracy to AD dementia diagnosis and should provide another promising candidate for target engagement, mechanism of action, and outcome in therapeutic trials affecting AD related neuroinflammation (75).

Craig-Schapiro and colleagues reported that YKL-40, an indicator of neuroinflammation, is elevated in the CSF of AD patients and, together with Aβ1-42, has potential prognostic utility as a biomarker for asymptomatic AD (76). Unbiased proteome based analysis led to discover an array of CSF neuroinflammatory proteins as candidate biomarkers for staging early AD (77) and improving AD diagnosis and prognosis (78). These promising exploratory markers deserve further systematic validation studies.

For a long time efforts to discover and develop robust and reliable biomarkers for AD in peripheral blood-serum had relatively minor success, however, emerging systematic clinical studies begin to show promising novel blood candidates biomarkers and demonstrate considerable performance in AD. Multivariate investigation of 18 plasma signaling and inflammatory proteins was reported to identify AD patients and predict AD, with high accuracy, in MCI subjects (79). Another study using explorative proteomics technology identified AD-associated changes in plasma levels of complement factor H and a2-macroglobulin (80). O’Bryant and colleagues developed a serum protein-based algorithm to differentiate AD patients from controls (81). The original model, utilizing over 100 proteins was next refined by reducing the number of serum proteins to only 30, still retaining an excellent diagnostic accuracy (82).

Taken together, recent genetic research in sAD and the existing peripheral and central biomarker data offer substantial support for a considerable inflammatory component in AD pathophysiology. However, the available results of anti-inflammatory therapies in AD have been contradictory (83); therefore, the link between inflammation and other core disease processes in AD remains elusive.

Oxidative stress. Free radical-mediated injury to neurons characterizes AD. F2-isoprostanes, derived from lipid peroxidation, may be used as biomarkers for this pathogenic process. Increase in CSF F2-isoprostane levels was described in AD (84). In contrast, plasma studies reported conflicting results, probably because the contribution of brain-derived F2-isoprostanes to plasma is masked by the much higher contribution of peripherally derived F2-isoprostanes (84).

Integrity of the blood-brain barrier. The best-established biomarker for the integrity of the blood-brain barrier is the CSF/serum albumin concentration ratio: it is normal in patients with pure AD whereas it is elevated in CNS infections, inflammatory disorders, brain tumors, and cerebrovascular disease (85)

Microvascular homeostasis. Significant changes are observed in plasma of AD patients vs. healthy control subjects, showing a consistent pattern of elevated levels of vasodilators – midregional pro-adrenomedullin (MR-proADM) and midregional pro-atrial natriuretic peptide (MR-proANP) – and decreased levels of vasoconstrictor C-terminal endothelin-1 precursor fragment (CT-proET-1). These markers showed promising predictive accuracy to detect prodromal AD in MCI subjects (86, 87). Such altered expression of microcirculation parameters supports the theory of a characteristic perturbed microvascular homeostasis in AD.

Phases of Development of AD Biomarkers

It is essential to include an appropriate biomarker package in all clinical trials, even though only in a subset of subjects, since clinical outcomes alone may not provide sufficient evidence of the benefit or lack thereof of potential new AD compounds (43). Neurobiochemical candidate markers (as well as imaging candidate markers) usually have to go through time-consuming systematic steps of development (from establishing technical characteristics to method harmonization-standardization, to validation of performance and qualification for use) before they can be designated as markers that can be utilized in clinical trials (43). At present, manual hippocampus volumetry, amyloid-targeting PET (positron emission tomography) ligands, 18F-FDG-PET (18F-2-fluoro-2-deoxy-D-glucose – positron emission tomography), and core AD CSF biomarkers (Aβ1-42, Aβ1-40, t-tau, and p-tau) are those markers that achieved the most advanced stage of development (standardization and validation) (phase III) and are therefore currently being employed in large-scale international multicenter controlled clinical trials (43) (Table 3).

Table 3.

Current development steps of neurobiochemical and imaging markers in Alzheimer’s disease

	Phase I	Phase II	Phase III
Requirements	Establishing technical/methodological features Hypothesis-free search Hypothesis-driven validation studies	Selected patients Determining normal values, sensitivity, and specificity	Validating normal values Determining prevalence, positive/negative predictive values Multicenter controlled clinical trials
Neurobiochemical markers	Exploratory markers based on analysis of proteome (use of proteomics technologies) Blood (plasma/serum)-based biomarkers CSF A3 oligomers	CSF A3 peptides concentration CSF BACE1 concentration and activity	CSF Aß1-42 concentration CSF t-tau concentration CSF p-tau181 and p-tau231 concentration
Imaging markers	Basal forebrain volume Functional activation Functional connectivity (fMRI, EEG, MEG)	Diffusion tensor imaging (DTI) Voxel-based morphometry (VBM) Cortical thickness Automated hippocampus volume Entorhinal cortex volumetry	Manual hippocampal volumetry 18F-FDG-PET Amyloid-PET ligands Combination of imaging markers

Abbreviations: 18F-FDG-PET, 2-deoxy-2-(18F)fluoro-D-glucose-positron emission tomography; A3, Amyloid-beta; BACE1, (3-site amyloid precursor protein-cleaving enzyme 1; CSF, cerebrospinal fluid; EEG, electroencephalography; fMRI, functional magnetic resonance imaging; MEG, magnetoencephalography; PET, positron emission tomography.

Upcoming Prevention Trials: Legitimizing AD Biomarkers

Currently, the emerging international expert efforts are beginning to focus on the very earliest stages of the disease, for primary prevention even before core pathophysiologic mechanisms commence and advance to a late phase corresponding to prodromal symptoms and to syndromal dementia as well as for secondary prevention throughout the complex molecular-cellular mechanistic stages. This paradigm shift towards an earlier AD characterization/definition /detection/diagnosis is crucial to redefine and establish successful interventional trials. Such a paradigm clearly encompasses both primary prevention (preventing the initiation of molecular and cellular mechanisms) and secondary prevention (preventing and progression of pathological mechanisms and subsequent symptoms and syndromes). This goal may be accomplished by incorporating the clinical trial approach to disease into a public health model, using long-term longitudinal databases that include large populations (88). Important initiatives with an international and global perspective are ongoing: the OECD Task Force on AD, the EU-US Task Force on Clinical Trial Development in AD and, most excitingly, the non-profit corporation “Prevent Alzheimer’s Disease 2020” (PAD2020, http://www.pad2020.org) which highlight that a world-wide database should be established by integrating/expanding existing cohorts and registries (88).

The appropriately intensive effort to test potential AD therapeutics earlier and in more rationally (based on data and established facts) designed trials is likely to lead to accurate scientific evidence of disease modification before long. This indication may originate primarily from pre-specified biomarker data, although it will probably be accompanied by positive trends on cognitive tests. Despite being discouraging to patients, carers, physicians, to investors in the pharmaceutical and biotech industry and to the public, recent clinical trial failures will provide ever growing substantial information and experience on what may work partially, what does not, and where to go next (89). According to Selkoe, “as a society, we must invest much more and invest more wisely” (89). In this regard, US president Obama has recently announced a growth in funding for AD next year of $106m (£67m; €80m), a 24% rise over the current $450m a year in the US. The announcement came shortly after scientists gathered at a meeting of the New York Academy of Sciences, in New York, in order to determine how best to spot AD in the “silent years” before the disease is manifest (90).

Generally, it seems recommendable to widen and expand the therapeutic vision beyond the mono-system based Aβ paradigm and focus on trials in very early symptomatic as well as presymptomatic participants (89). The exciting debate on the new AD diagnostic criteria encouraged the development of novel study designs for prevention trials in at risk and prodromal stages of AD. As a next important step forward three complementary prevention trials will start within few months to assess whether anti-amyloid treatments are effective in asymptomatic AD populations. The Alzheimer’s Prevention Initiative (API) trial will test the preventive effects of a specific monoclonal anti-Aβ antibody – crenezumab, developed by the company Genentech – in asymptomatic subjects from an extended family in Colombia with a PSEN1 E280A mutation predisposing carriers to early-onset AD. The DIAN consortium will assess the Eli Lilly and Company anti-Aβ antibody compound Solanezumab (based on recently revealed data from two large-scale phase 3 trials) in a larger cohort of genetically defined early-onset AD patients. A third trial – Anti-Amyloid Treatment of Asymptomatic Alzheimer’s (A4) – will analyze a single unnamed drug in elderly patients who are at high-risk of developing disease, as assessed by amyloid imaging (91, 92). In case of positive outcomes, these prevention trials may help not only to test drugs, but also to corroborate the validity of the “amyloid hypothesis”, thus legitimizing AD biomarkers and accelerating the future clinical development of drugs for larger patient populations (11).

The Long and Winding Biomarker Road Ahead

Given the overall dynamic and as of yet relatively untapped future potential of the multi-modal biomarker development the current status of the field regarding value and integration of biomarkers in clinical trials seems only the beginning of the “systems biology and neural network paradigm” area of AD (1). We can learn much from current research in early asymptomatic populations as well as in ADAD subjects, however, we have yet to carefully chart the full spectrum biomarker map in sAD (30) to advance and optimize effective treatment perspectives for the battle against the devastating worldwide AD epidemic.

Conflict of Interests

HH was supported by the Katharina-Hardt-Foundation, Bad Homburg, Germany. HH declares associations with the following companies: Astra-Zeneca, Avid, Bayer, BMS, Boehringer-Ingelheim, Eisai, Elan, Eli Lilly and Company, GE Healthcare, Genentech, GlaxoSmithKline Bio, Janssen-Cilag, Merz Pharmaceuticals, Novartis, Pfizer, Sanofi-Aventis, Schwabe, Roche, Wyeth. SL declares no competing interests.