WO2008112232A2 - Methods for monitoring gamma secretase inhibition in vivo - Google Patents

Abstract

The present invention provides a means for monitoring the inhibition of γ secretase in an Alzheimer's disease (AD) patient or control subject in vivo using a γ secretase inhibitor biomarker. Statistically relevant biomarkers identified from a fluid sample of a patient treated with an AD therapeutic are identified and used to monitor γ secretase inhibition and the efficacy of a γ-secretase inhibitor.

Description

TITLE OF THE INVENTION

METHODS FOR MONITORING GAMMA SECRETASE INHIBITION IN VIVO

FIELD OF THE INVENTION The present invention relates generally to therapeutics and disease progression in the field of Alzheimer's disease. More specifically, it relates to biomarkers that can be used to monitor gamma secretase inhibition or to determine the efficacy of drugs given to treat Alzheimer's disease.

BACKGROUND OF THE INVENTION Alzheimer's disease (AD) is a major neurodegenerative disease of unknown etiology that is characterized by the selective degeneration of basal forebrain cholinergic neurons. The degeneration of these cells leads to a secondary loss of neurons in the limbic system and cortex that control learning and memory. The consequent symptoms of the disease include a progressive loss of memory, the loss of the ability to communicate and the loss of other cognitive functions which occur over a course of approximately eight years. Over the course of this cognitive decline patients often become bedridden and completely unable to care for themselves. Although several symptomatic therapies have been approved to provide some compensation for the cholinergic deficit, for example, Aricept (donepezil HCL) (Eisai, Inc. and Pfizer, Inc.), the clinical effects of these are modest and none are able to significantly alter the course of the disease. Improving upon strategies for the treatment of AD has become a focus for the medical and scientific communities due to increases in the average age of the world population, the consequent increase in incidence and prevalence of age-related disorders such as AD, and the severe socioeconomic impact associated with supporting such cognitively impaired patients over the long term.

Requisite to improving the treatment of AD is improving the ability of clinicians to accurately diagnose the disease early in its course and to accurately monitor the progression of the disease. Currently, a diagnosis of possible or probable AD is typically made based on clinical symptoms. A definitive diagnosis of AD can only be made post-mortem and requires a pathological examination of the affected brain tissue. The key pathological hallmarks of the disease are plaques consisting of deposited amyloid beta (Aβ) protein and tangles consisting of degenerated neuronal cells and their cytoskeletal elements (neurofibrillary tangles). There are currently no tests that, in and among themselves, have been validated to identify AD and differentiate it from other diseases affecting cognition. Compared to the pathological diagnosis, the pre-mortem clinical diagnosis can achieve an accuracy of approximately 80% to 90% at the very best of centers. However, this level of diagnostic accuracy more commonly occurs at well-experienced AD centers and for patients who have been manifesting clinical symptoms for several years (Rasmusson, D. X., et al., Alzheimer Pis. Assoc. Disord.. 10(4): 180-188 (1996); Frank, R.A. et al.. Proceedings of the Biological Markers Working Group: NIA Initative on Neuroimaging in Alzheimer's Disease. Neurobiol. Aging, 24: 521-536 (2003)). Following the clinical diagnosis, the progression of the disease is typically monitored through cognitive testing and assessment of everyday function. The course is often variable across patients and may be influenced by both organic and environmental elements.

Aβ peptide, a 39-43 amino acid peptide derived by proteolytic cleavage of the amyloid precursor protein (APP), is the major component of amyloid plaques (Glenner and Wong, Biochem. Biophvs. Res. Comm,, 120: 885-890 (1984)). APP is actually a family of polypeptides produced by alternative splicing from a single gene. Major forms of APP are known as APP695, APP751, and APP770, reflecting the number of amino acids in each splice variant (Ponte et al., Nature 331 : 525-527 ( 1988); Tanzi et al., Nature 331 : 528-530 ( 1988); Kitaguchi et al., Nature 331 : 530-532(1988)). APP is a ubiquitous membrane-spanning (type 1 ) glycoprotein that undergoes proteolytic cleavage by at least two pathways (Selkoe, Trends Cell Biol. 8: 447-453 (1998)). In one pathway, cleavage by an enzyme known as α-secretase occurs while APP is still in the trans-Golgi secretory compartment (Kuentzel et al., Biochem. J.. 295: 367-378 (1993)). This cleavage by α- secretase occurs within the Aβ peptide portion of APP, thus precluding the formation of Aβ peptide. In an alternative proteolytic pathway, cleavage of the Met596-Asp597 bond (numbered according to the 695 amino acid protein) by an enzyme known as β- secretase occurs. This cleavage by β-secretase generates the N- terminus of Aβ peptide. The C-terminus is formed by cleavage by a second enzyme known as γ-secretase. The C-terminus is actually a heterogeneous collection of cleavage sites rather than a single site since γ-secretase activity occurs over a short stretch of APP amino acids rather than at a single peptide bond. Peptides of 40 or 42 amino acids in length (Aβ40 and Aβ42, respectively) predominate among the C-termini generated by γ-secretase. Aβ42 peptide is more prone to aggregation than Aβ40 peptide, the major secreted species (Jarrett et al., Biochemistry. 32: 4693-4697 91993); Kuo et al., J. Biol. Chem. 271 : 4077-4081 ( 1996)), and its production is closely associated with the development of Alzheimer's disease (Sinha and Lieberburg, Proc. Natl. Acad. Sci. USA. 96: 1 1049-1 1053 (1999)). The bond cleaved by γ-secretase appears to be situated within the transmembrane domain of APP. For a review that discusses APP and its processing, see Selkoe, Trends Cell. Biol.. 8: 447-453 (1998).

Much interest has focused on the possibility of inhibiting the development of amyloid plaques as a means of preventing or ameliorating the symptoms of Alzheimer's disease. To that end, a promising strategy is to inhibit the activity of β- and γ-secretase, the two enzymes that together are responsible for producing Aβ. This strategy is attractive because, if the formation of amyloid plaques is a result of the deposition of Aβ is a cause of Alzheimer's disease, inhibiting the activity of one or both of the two secretases would intervene in the disease process at an early stage, before late- stage events such as inflammation or apoptosis occur. Such early stage intervention is expected to be particularly beneficial (see, for example, Citron, Molecular Medicine Today, 6: 392-397 (2000)).

The last decade has seen an increase in efforts to identify and validate AD-related biomarkers that might increase the sensitivity and specificity of diagnosis and provide a convenient and objective measure of disease progression (Regan Research Institute and National Institute of Ageing (NIA) Consensus Report of the Working Group on: 'Molecular and Biochemical Markers of Alzheimer's Disease,' Neurobiol. Aging. 19(2): 109-1 16 (1998); Frank et al., (2003). Among the techniques that currently hold promise in this regard is the biochemical analysis of cerebrospinal fluid (CSF). The value of CSF analysis is based on the fact that the composition of this fluid may reflect brain biochemistry due to its direct contact with brain tissue and the interstitial fluid.

The CSF proteins that have received the most attention are those thought to reflect key features of the disease pathogenesis, including Aβ deposition and neuronal degeneration. Studies have demonstrated reduced levels of the Aβ42 peptide in the CSF of clinically diagnosed AD patients compared to controls (Andreasen, N., et al., Arch. Neurol., 58: 373-379 (2001 ); NIA Consensus Report, 1998; Frank et al, 2003, Andreasen, N., et al., Clin. Neurol. Neurosurg.. 107: 165-173 (2005)). Aβ42 is a cleavage product of the amyloid precursor protein (APP) and is thought to be a major constituent of the senile plaque. One theory of disease progression is that reduced CSF levels in AD patients may be due to increased deposition of the peptide in the brain. In contrast, many studies have shown that the expression of the Aβ40 peptide, another APP cleavage product that is also a plaque component, may be similar in clinically diagnosed AD and control CSF (Frank et al, 2003).

The Tau protein is another CSF protein that has been studied for disease etiology. Tau is an axonal protein that, when hyperphosphorylated, assembles into the paired helical filaments that form neurofibrillary tangles. Whereas the presence of Tau in the CSF is thought to be a general reflection of axonal (i.e., neuronal) degeneration in the brain, the presence of phosphorylated Tau (p-Tau) may be a more specific indicator of AD-related pathology. CSF levels of both Tau and p-Tau in clinically diagnosed AD patients have been shown in many studies to be elevated compared to that in controls (Andreasen, 2001 ; and for review Consensus Report, 1998; Frank et al., 2003 and Andreasen, 2005). A recent review article describes not only the status of biochemical biomarkers but also the active area of imaging biomarkers and their use in longitudinal clinical trials (Thai, L. J., et al., Alzheimer Pis. Assoc. Disord., 20(1 ): 6-15 (2006)). Generally, the field has focused on diagnostic or prognostic biochemical biomarkers and there are few papers related to disease progression markers from fluid samples.

SUMMARY OF THE INVENTION

The present invention relates to biomarkers and methods for monitoring γ-secretase inhibition.

In one embodiment, the invention is a γ secretase inhibition marker selected from the group consisting of apolipoprotein H (ApoH), CD40-Ligand (CD40L), fatty acid binding protein (FABP), ferritin, leptin, macrophage inflammatory protein- 1 beta (MIP-I β), macrophage derived chemokine (MDC), prostatic acid phosphatase (PAP), serum amyloid P (SAP), tumor necrosis factor-α (TNF-α) and tumor necrosis factor receptor 2 (TNF-R-II). In a preferred embodiment of the invention the marker is CD40L, ferritin, MDC or SAP. In an alternate preferred embodiment said marker is FABP or TNFα. In still another preferred embodiment the marker is apolipoprotein H, leptin, MIP-I β or PAP.

In another embodiment the invention is a method for monitoring the inhibition of γ secretase in an AD patient or control subject comprising: a. selecting at least one γ secretase inhibition marker previously identified as being statistically significant and which is differentially expressed; b. obtaining a fluid sample from an AD patient or control subject; c. analyzing the fluid sample of step (b) for the presence of the γ secretase inhibition marker or markers to form a reference level; d. obtaining a second fluid sample from said AD patient or control subject at a later prescribed time interval; e. analyzing the second sample for the presence of the γ secretase inhibition marker or markers; f. comparing the results of step (e) to those of step (c) to obtain an output; g. where the output of step (f) is the change in the marker levels between the two sample measurements; and where a change in the marker levels is indicative of γ secretase inhibition. In still another embodiment the invention is a method for monitoring the efficacy of an AD therapeutic comprising: a. selecting at least one γ secretase inhibition marker previously identified as being statistically significant and which is differentially expressed; b. obtaining a fluid sample from an AD patient or control subject to which the AD therapeutic is to be administered; c. analyzing the fluid sample of step (b) for the presence of the γ secretase inhibition marker or markers to form a reference level; d. administering an AD therapeutic to said AD patient or control subject; e. obtaining a second fluid sample from said AD patient or control subject at a prescribed time interval after therapeutic administration; f. analyzing the second sample for the presence of the γ secretase inhibition marker or markers; g. comparing the results of step (f) to those of step (c) to obtain an output; h. where the output of step (g) ) is the change in the marker level between the two samples; and where a change in the marker level is indicative of γ secretase inhibition.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 shows the PK and PD results from a CMP rhesus study showing peak Aβ42 lowering at four hours post administration of Compound A in conscious CMP rhesus monkeys.

Figures 2 A and 2B show the 90 analytes of the Rules-Based Medicine human antigen MAP v 1.6. Those analytes in the panel for which there was not at least 10 AD subjects and 10 control (CTL) subjects with at least 2 CSF samples available are shown with an asterisk (N=27) and thus only the remaining (N=63) analytes were considered in the statistical analysis.

Figure 3 shows a histogram of the distribution of number of measurable RBM analytes above the lowest calibrator concentration as a function of sample % in the pilot rhesus CSF samples (N=16): 35 of 90 analytes had measurable levels in all 16 samples; 46 of 90 analytes had fewer than six missing values.

Figures 4A and 4B show histograms of the distribution of number of measurable RBM analytes above the lowest calibrator concentration as a function of sample % in the human CSF samples (Fig. 4A: placebo (N=12); Fig. 4B: GSI treated (N=12)) from a single dose γ-secretase inhibitor (GSI) crossover study: 55 of 89 analytes had measurable levels in all 24 samples; up to 23 analytes were not measurable in all samples.

Figures 5 A and 5B show histograms of the distribution of number of measurable RBM analytes above the lowest calibrator concentration as a function of sample % in the human CSF samples (Fig. 5A: N=I 2 placebo; Fig. 5B: N=I 2 GSI treated) from the multiple dose GSI study: 50 of 89 analytes and 54 of 89 analytes in placebo and GSI treated samples, respectively, had measurable levels; 25 of 89 analytes were not measurable in all samples.

Figure 6 shows a plot of CSF Aβ40 (AB 140) expression in placebo and high dose samples from the CMP rhesus study: "•" = placebo (control) group; "+" = GSI treated group; the vertical line represents the mean and standard error for each group; solid line connects the mean expression level for both groups.

Figure 7 shows a plot of CSF Aβ42 (AB142) expression in placebo and high dose samples from the CMP rhesus study: "•" = placebo (control) group; "+" = GSI treated group; the vertical line represents the mean and standard error for each group; solid line connects the mean expression level for both groups.

Figure 8 shows a plot of plasma Aβ40 (AB 140) expression in placebo and high dose samples from the CMP rhesus study: "•" = placebo (control) group; "+" = GSI treated group; the vertical line represents the mean and standard error for each group; solid line connects the mean expression level for both groups Figure 9 shows a plot of CSF CD40L (CD40-Ligand) expression in placebo and high dose samples from the CMP rhesus study: "•" = placebo (control) group; "+" = GSI treated group; the vertical line represents the mean and standard error for each group; solid line connects the mean expression level for both groups.

Figure 10 shows a plot of CSF ferritin expression in placebo and high dose samples from the CMP rhesus study: "•" = placebo (control) group; "+" = GSI treated group; the vertical line represents the mean and standard error for each group; solid line connects the mean expression level for both groups.

Figure 1 1 shows a plot of CSF MDC expression in placebo and high dose samples from the CMP rhesus study: "•" = placebo (control) group; "+" = GSI treated group; the vertical line represents the mean and standard error for each group; solid line connects the mean expression level for both groups.

Figure 12 shows a plot of CSF serum amyloid P (SAP) expression in placebo and high dose samples from the CMP rhesus study: "•" = placebo (control) group; "+" = GSI treated group; the vertical line represents the mean and standard error for each group; solid line connects the mean expression level for both groups. Figure 13 shows a box plot of CSF Aβ40 (AB 140) expression in placebo and single dose GSI samples from a balanced two-period crossover single dose human study: "•" = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line. Figure 14 shows a box plot of CSF Aβ42 (AB 142) expression in placebo and single dose

GSI samples from a balance two-period crossover single dose human study: "•" = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line.

Figure 15 shows a box plot of CSF apolipoprotein H expression in placebo and single dose GSI samples from a balanced two-period crossover single dose human study: "•" = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line.

Figure 16 shows a box plot of CSF leptin expression in placebo and single dose GSI samples from a balanced two-period crossover single dose human study: " •" = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line.

Figure 17 shows a box plot of the CSF MlP-I β (MIP 1 beta) expression in placebo and single dose GSI samples from a balanced two-period crossover single dose human study: " •" = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line.

Figure 18 shows a box plot of the CSF prostatic acid phosphatase (PAP) expression in placebo and single dose GSI samples from a balanced two-period crossover single dose human study: " • " = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line. Figure 19 shows a box plot of the CSF TNF-RII expression in placebo and single dose

GSI samples from a balanced two-period crossover single dose human study: " •" = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line.

Figure 20 shows a box plot of the CSF Aβ40 (AB140) expression in placebo and multiple dose GSI samples from a balanced two-period crossover multiple dose human study: "•" = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line.

Figure 21 shows a box plot of the CSF Aβ42 (AB 142) expression in placebo and multiple dose GSI samples from a balanced two-period crossover multiple dose human study: " •" = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line.

Figure 22 shows a box plot of the CSF Fatty Acid Binding Protein (FABP) expression in placebo and multiple dose GSI samples from a balanced two-period crossover multiple dose human study: " •" = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line. Figure 23 shows a box plot of the CSF TNFα (TNFa) expression in placebo and multiple dose GSI samples from a balanced two-period crossover multiple dose human study: " •" = placebo (control) group; "+" = GSI treated group. Each subject received both placebo and GSI treatments; samples corresponding to these treatments from each subject are connected by solid line.

Figure 24 shows a screenshot of the output of the analysis of the natural logarithm of CSF Aβ40 from an R program for a one-way ANOVA rhesus study to compare high dose GSI versus placebo treated samples. The number shown in the first column ("Estimate"), second row ("GroupD2"), of the table is an estimate of the difference between the high dose GSI and placebo treated samples of log(Aβ40), i.e. -1.027829. The corresponding p-value is shown in the fourth column ("Pr(>|t|)"), second row, of this table, i.e. 5.863324e-03. The anti-log of the estimated difference in log scale between the GSI and placebo treated samples (i.e., exp(-l .027829)) corresponds to the ratio of Aβ40 expression of high dose GSI to placebo (0.36). Figure 25 shows a screenshot of the output of the analysis of the natural logarithm of

CSF Aβ40 from an R program for a linear mixed effects model for a balanced two-period crossover single dose human study to compare GSI versus placebo treated samples. The number shown in the first column ("Value"), fourth row ("Drugplac"), of the table is an estimate of the difference between the placebo and single dose GSI groups of log(Aβ40), i.e. 0.725876. The corresponding p-value is shown in the fifth column ("p-value"), fourth row, of this table, i.e. 6.01 1699e-09. The anti-log of the estimated difference in log scale between the GSI and placebo treated groups (i.e., exp(-0.725876)) corresponds to the ratio of Aβ40 expression of single dose GSI to placebo (0.48).

Figure 26 shows a screenshot of the output of the analysis of natural logarithm of CSF Aβ40 from R program for a linear mixed effects model for a balanced two-period crossover multiple dose human study to compare GSI versus placebo treated samples. The number shown in the first column

("Value"), fourth row ("Drugplac"), of the table is an estimate of the difference between the placebo and multiple dose GSI groups of log(Aβ40), i.e. 0.43357937. The corresponding p-value is shown in the fifth column ("p-value"), fourth row, of this table, i.e. 1.67343 le-05. The anti-log of the estimated difference in log scale between the GSI and placebo treated groups (i.e., exp(-0.43357937)) corresponds to the ratio of Aβ40 expression of multiple dose GSI to placebo (0.65).

DETAILED DESCRIPTION OF THE INVENTION Definitions

As used herein, the term "analyte" or "marker" or "biomarker" are used interchangeably and refers to a protein or protein complex that are the subject of analysis herein, for example, Aβ42 or CD40L.

As used herein, the term "covariate" refers to variables such as the baseline age and sample storage time that are used as additional independent variables in the univariate analysis of variance (ANOVA). As used herein, the term "disease state" or "disease status" refers to the intrinsic state or status of the underlying pathophysiology of the disease process, including biomolecular, cellular, and system dysfunctions that are not known or understood, but whose affects are measurable in a mammalian host, in particular a human subject. As used herein, the term "disease progression" refers to the change in the disease state or disease status over time on its own or along with another variable over time, such as the cognitive status.

Disease progression consists of a change in a disease state variable, such as a measurable biomarker or panel of markers, as a function of another independent variable, most typically time, and may be either individually or along with another measurement, such as cognition. As used herein, the term "γ secretase inhibition marker" refers to an individual, i.e. univariate, analyte, marker or biomarker that has been identified as being statistically significant and differentially expressed.

As used herein, the term "γ secretase inhibition panel" refers to any combination of two or more analytes, markers or biomarkers used to monitor γ secretase inhibition or to determine efficacy of an AD therapeutic. In one embodiment of the invention it refers to a multi-analyte panel or composite of individual γ-secretase inhibition markers.

® ®

As used herein, the term "RBM Human MAP " or "Human MAP " refers to the collection of 90 analytes, markers or biomarkers comprising the proprietary human Multi-Analyte

® Profiles (Human MAP ) (See Figures 2A and 2B) from Rules-Based Medicine, Inc. (RBM) Austin, TX. For the human single and multiple dose studies, only 89 analytes from this panel were measured in that IFN-gamma was not included in the data analyses.

As used herein, the term "monitoring Alzheimer's disease" means both the ability to evaluate the progression of the disease over time or progression of the disease with a cognitive measure, as well as the ability to evaluate the change of the disease state after AD therapeutic intervention. As used herein, the term "AD therapeutic" or "AD therapeutic intervention" means any compound, molecule or biologic administered to a patient diagnosed with AD to alter or modify the disease state, progression, or its deleterious effects cognitive effects. This includes, but is not limited to, any therapeutic administered in oral, intravenous, intramuscular or subcutaneous form and may be administered alone or with other excipients or adjuvants in a pharmaceutical composition or formulation. As used herein, the term "CSF" refers to cerebrospinal fluid.

As used herein, the term "CSF Aβ42" or "CSF Aβ40" refers to a protein of Aβ42 or Aβ40, respectively, isolated and quantified specifically from a CSF sample as compared to other forms of the proteins present in brain tissue or other fluids.

As used herein, the term "plasma Aβ42" or "plasma Aβ40" refers to a protein of Aβ42 or Aβ40, respectively, isolated and quantified specifically from a plasma sample as compared to other forms of the proteins present in brain tissue or other fluids.

As used herein, the term "sAPP" refers to any or all species of an N-terminal secreted APP fragments created from the proteolytic cleavage of APP, including but not limited to its major forms known as APP695, APP751 and APP770, by enzymes or activities known as α-secretase or β-secretase. Such N-terminal fragments, include but are not limited to, proteins known sAPPα and sAPPβ.

As used herein, the term "α-1 -antitrypsin," also known as "SERPINAl " or "α-1 antiproteinase, antitrypsin," refers to a liver protein that blocks the destructive effects of certain enzymes. SwissProt Accession Number P02743.

As used herein, the term "CD40L" or "CD40 Ligand" refers to the ligand of cluster designation 40 (CD40), a human leukocyte differentiation antigen recognized by monoclonal antibodies, associated with B cell and T cell proliferation, differentiation and isotype switching.

As used herein, the term "MDC," also known as CCL22, refers to macrophage-derived chemokine synthesized specifically by cells of the macrophage lineage and is a potent chemoattractant for neutrophilic granulocytes. SwissProt Accession Number 000626.

As used herein, the term "SAP," also known as "APCS," refers to serum amyloid P, a small glycoprotein found in normal serum and in all amyloid deposits. SwissProt Accession Number P02743. As used herein, the term "ferritin" refers to a serum protein found at low concentrations and that is directly proportional to the body's iron stores. Swiss-Prot Accession Number: Heavy chain P02794; Light chain P02792.

As used herein, the term "apolipoprotein H" (ApoH), also known as "APOH" or "β-2- glycoprotein-1 ," refers to a molecule that is expressed by placental trophoblast cells at high levels and appears to act as a co-factor for the binding of auto-antibodies to phospholipids in trophoblasts. SwissProt Accession Number: P02749.

As used herein, the term "leptin" refers to a protein hormone produced predominantly in adipocytes and that plays a role in the regulation of body weight through the hypothalamic centers of hunger, body temperature and energy expenditure. Swiss-Prot Accession Number: P41 159. As used herein, the term "MIP-I β" or "MIP-I beta" refers to Macrophage inflammatory protein-l beta, one of two MIP proteins which are the major factors produced by macrophages following their stimulation with bacterial endotoxins. Swiss-Prot Accession Number: Pl 3236.

As used herein, the term "prostatic acid phosphatase" (PAP), also known as "ACPP," refers to a prostate-specific fraction of acid phosphatase commonly elevated in men with metastatic disease. Swiss-Prot Accession Number: Pl 5309.

As used herein, the term "TNF-R-II," also known as "TNFRSFl B," refers to tumor necrosis factor receptor 2, a soluble form of the TNF receptor. Swiss-Prot Accession Number: Q92956.

As used herein, the term "fatty acid binding protein" or "FABP" refers to a plasma marker of acute myocardial infarction (AMI) and a biomarker of cardiovascular risk. Swiss-Prot Accession Number: P05413.

As used herein, the term "TNF α" or "TNF alpha" refers to tumor necrosis factor alpha, a protein secreted by macrophages, monocytes, neutrophils, T-cells and NK-cells following many different stimuli including interferon, IL-2 , GM-CSF, bradykinin, immune complexes, inhibitors of cyclooxygenase and PAF (platelet activating factor). Swiss-Prot Accession Number: POl 37 As used herein, the term "sensitivity" refers to the ability of an individual marker or a composite of markers to correctly identify patients with the disease, i.e. Alzheimer's disease, which is the probability that the test is positive for a patient with the disease. The current clinical criterion for patients who are deemed as probable for having AD is about 85% sensitive when compared to autopsy confirmed cases.

As used herein, the term "specificity" refers to the ability of an individual marker or a composite of markers to correctly identify patients that do not have the disease, that is, the probability that the test is negative for a patient without disease. The current clinical criterion is about 75% specific.

As used herein, the term "accuracy" refers to the overall ability of an individual marker or a composite of markers to correctly identify those patients with the disease and those without the disease.

An open area of investigation in drug discovery today is establishment of a model to accurately identify novel biomarkers of compound activity in the central nervous system. To begin to address this question a CMP rhesus study was conducted with a known γ-secretase inhibitor, Compound

A, to prepare CSF samples for proteomic analysis beyond the classic Aβ analytes. Rhesus CSF samples

® were analyzed in the Rules-Based Medicine (RBM) human Multi-Analyte Profile (Human MAP ). This study was designed as a pilot study to evaluate the potential of the RBM human MAP to cross-react to nonhuman primate (NHP) CSF and identify novel CSF biomarkers of γ-secretase inhibition. Human CSF samples from subjects that received an investigational γ-secretase inhibitor (Compound A) were also assessed for biochemical changes indicative of γ-secretase inhibition.

The changes in the CSF that occur upon γ-secretase inhibition seem to fall into interrelated but separate biological categories, including but not limited to, initiation factors (Aβ42), inflammation/repair mechanisms (inflammatory cytokines such as IL-7), immune responsive markers (CD40L and MDC) and reactionary proteins (SAP). Applicants are not aware of any evidence in the literature that changes in these markers can be attributed to γ-secretase inhibition. Measurement of changes in all of these categories in a CSF sample would allow for a global assessment of how a given treatment affects the brain. Such a global assessment of drug treatment is expected to be a more sensitive marker of compound effects than any one change in the CSF taken in isolation. Banked frozen CSF samples were selected from a rhesus AD study based on CSF Aβ42 reduction at four hours post administration and aliquots of pristine CSF samples were submitted for RBM human MAP analysis. PK and PD results from the rhesus study showing peak Aβ42 lowering at four hours post administration of Compound A are shown in Figure 1. Similar samples from a human clinical trial with Compound A were also submitted to extend the results observed in the primate study and to assess the potential for identification of translational biomarkers from rhesus monkeys to humans.

The present invention relates to the identification of univariate γ-secretase inhibition markers, the construction of multi-analyte panels of γ-secretase inhibition markers and the statistical methods to determine individual markers that are statistically significant individually as well as in an optimal panel of makers. In addition to the classical markers of CSF Aβ40 and CSF Aβ42, Applicants have discovered that apolipoprotein H (ApoH), CD40-Ligand (CD40L), fatty acid binding protein (FABP), ferritin, leptin, MIP-I -β, MDC, prostatic acid phosphatase (PAP), serum amyloid P (SAP), TNF- α and TNF-R-II are each significant biomarkers of γ-secretase inhibition which could be useful clinically in the assessment of the degree and duration of γ-secretase inhibition. In an alternate embodiment of the invention, the biomarker responses would be evaluated for correlation to cognitive endpoints.

Univariate analysis performed on each analyte was carried out using a one-way ANOVA on the log transformed data to ensure the distribution of the data was approximately symmetric. The p- values (false positive rate) and q-values (false discovery rates) from this analysis on all the analytes were reported. The number of analytes considered in the rhesus study was 93 (90 RBM analytes plus CSF Aβ40, CSF Aβ42 and plasma Aβ40). The number of analytes considered in the human single dose and human multiple dose studies was 91 (89 RBM analytes plus CSF Aβ40 and CSF Aβ42). All analytes that have been analyzed and those that had p-values less than 0.05 were reported with special reference to the analyte that also had q-value of less 0.05. An important aspect of the invention described herein is that individually, apolipoprotein H (ApoH), CD40-Ligand (CD40L), fatty acid binding protein (FABP), ferritin, leptin, MIP-I -β, MDC, prostatic acid phosphatase (PAP), serum amyloid P (SAP), TNF-α and TNF-RII were each meaningful univariate markers of γ-secretase inhibition.

The assays and methods described herein used to determine either univariate markers or multi-analyte panels of markers for γ-secretase inhibition, including proteomic and ELISA/traditional amyloid based markers, can be used for multiple purposes. They can be used in the development of therapeutics for the treatment of AD and have the potential to decrease the time and cost associated with lengthy cognition clinical trials. As such, they can be used to verify target engagement of the γ-secretase complex and serve as markers of pharmacodynamic endpoints. Without wishing to be bound by any theory, the putative markers described herein may be useful in providing an early readout for a γ-secretase inhibitor clinical trial. By providing early information on the target engagement and possible correlation with cognitive endpoints, this data could serve as one of the early filters a the go/no-go decision making process.

Rhesus Monkey GSI study A γ-secretase inhibitor, Compound A, shown in Example 1 below, was evaluated in conscious CMP rhesus monkeys in a four-way crossover design (vehicle, 0.5%methylcellulose 5 mL/kg versus 10, 30 and 60 mpk PO by nasogastric lavage) to identify CSF biomarkers. Compound A or vehicle was administered at T= 0 hour (following T=O sampling) in six monkeys. CSF (2 mL) and plasma (4 mL EDTA) were collected. The CSF was separated into aliquots for analysis of APP metabolites (neat), novel markers, and compound concentration. The plasma was split evenly into five tubes for PK and PD analyses. The selected doses showed no significant AB effect with the lowest dose (10 mpk) and dose-related effects with the intermediate and high doses (30 and 60 mpk, 34 and 54 % reductions in CSF AB42 levels at four hours post-dose). Rhesus Monkey Sample summary

A pilot subset of rhesus CSF samples (N= 16) was selected from a balanced crossover CMP rhesus study: vehicle only (N=5), 30 mpk Compound A (N=5) and 60 mpk Compound A (N=6), all measured at four hours post-dose to capture peak activity (Table 1 ). Table 1 is a summary of the archived CSF samples from the CMP rhesus study selected for REM human MAP analysis. The total number of samples was limited in this pilot study to sixteen due to logistical considerations getting samples shipped to RBM. CSF samples were selected from five of seven animals which received two or more treatments (placebo, intermediate dose (30 mpk) or high dose (60 mpk)) from the original full crossover design study, while the remaining two samples came from animals representing one of the three treatments. These samples represented all levels of γ-secretase inhibitory activity versus a control (no effect). Aliquots of 200 μL were sent to RBM for human MAP analysis.

Treatment:

30 = 30 mpk Compound A single oral dose 60 = 60 mpk Compund A single oral dose V= Vehicle - 5 mL/kg 0.5% methylcellulose

Rules-Based Medicine human Multi-Analyte Profiles (RJBM Human MAP ) Multi-Analyte Profiles, based on Luminex's xMAP bead based technology, consist of at least three pools of beads representing 90 antigens. As a conservative approach the lowest calibrator concentration of the respective marker was used as the value for samples that fell below the lowest calibrator concentration in the standard curve. Designed originally for human plasma analysis of 100 μL

® samples, the Human MAP has been analytically validated according to NACLES criteria and successfully used on human CSF when using 200 μL samples in manual mode and 250 μL samples in automated mode. Statistical analysis

Due to the small pilot sample size (N= 16) of the rhesus CMP study, the univariate data analysis was performed by assuming independence between samples. This caveat should be noted when interpreting the results from these analyses. As a conservative approach, levels of analytes reported as simply "below the lowest calibrator concentration" were replaced by the concentration of the lowest calibrator used in the calibration curve.

Univariate analysis performed on each analyte was a one-way ANOVA on the log transformed data to ensure the distribution of the data was approximately symmetric. The p-values (false positive rate) and q-values (false discovery rates) from this analysis on all of the analytes was reported. The FDR is the "proportion of significant changes that are false positives." The p-value estimates the False Positive Rate (FPR) which is the proportion of false positives among all the proteins that in reality did not change. The number of analytes considered in the rhesus study was 93 (90 RBM analytes plus CSF Aβ40, CSF Aβ42 and plasma Aβ40). Those that had a p-value of less than 0.05 were included herein, with preference given to the analyte from the RBM panel that also had q-value of less than 0.05.

® Response of rhesus CSF in the RBM Human MAP

The first aspect to be addressed as part of this analysis was the determination of the

® extent to which the RBM Human MAP would interact with antigens in the rhesus CSF. Thirty five of

90 analytes had measurable levels above the lowest calibrator concentration in all sixteen samples, while 46 of these 90 analytes had fewer than six missing values. The histogram of measurable analytical responses in the pilot rhesus CSF study is shown in Figure 3. One can see a similar distribution of analytical responses for human control CSF samples in Figures 4A and 5A and human GSI treated samples in Figures 4B and 5B. Although there appears to be additional analytes measurable in the human CSF group in the highest range (i.e. in the 91 -100% of samples measurable), the response to rhesus CSF in the Human MAP is equivalent. As such, one skilled in the art would acknowledge the feasibility of identifying novel CSF biomarkers of compound activity from the rhesus CSF samples using a focused

® antigen panel such as RMB Human MAP .

Analytes that had less than 5% false positive rate (p-value) between the high dose (60 mpk) and placebo from the study are summarized in Table 2. Analytes that were either marginally statistically significant (p < 0.1 ) or statistically significant (p < 0.05) between the high dose and placebo groups are shown in this table. The q-value (estimate of the false discovery rate) for each analyte is also shown. Among these analytes, SAP was measurable in only 1 1 out of 16 samples and MDC was measurable in 8 out of 16 samples. Samples that were not measurable were imputed using the lowest concentration of the standard curve as a conservative approach. Other analytes in this table are measurable in all samples.

Table 2

All markers were marginally significantly different between high dose GSI and placebo

(p<0.10). CSF Aβ40, CSF Aβ42, CD40-Ligand and serum amyloid P were statistically significant with

® p<0.05. Among the 90 analytes in the RBM Human MAP , CD40L (26 % elevated in high dose group), MDC (4.17 fold reduced in high dose group) and SAP (3.33 fold reduced in high dose group) appear significant with p-values <0.05. Among these three, SAP is considered to have the most robust effect with a q-value <0.05.

Human single dose study of a GSI versus placebo A double-blind, randomized, placebo-controlled, two period crossover study in healthy young men in a single study center was conducted. On the first day of each period, subjects received a single oral dose of either Compound A (500 mg) or a matching placebo according to a randomized schedule. CSF was drawn via lumbar puncture (LP) at 12 hours post dosing. For Compound A plasma concentrations and determination of plasma Aβ40 and sAPP levels, blood was collected prior to dosing and at several time points post dosing (see Example 2 for details). The duration of the study for each subject was approximately three weeks and the study was completed within five weeks.

CSF samples from each of the 12 subjects on both placebo and Compound A (total N=24 samples) were sent to for RBM human MAP analysis, in a similar manner to that described for the rhesus monkey study. The response of the human MAP to the human CSF is shown in Figure 4. Similar to the rhesus study, univariate analysis was applied to each marker in the analysis, including CSF Aβ40 and CSF Aβ42.

Table 3 is a summary of the analytes determined from a human single dose study of a GSI versus placebo. Analytes that were statistically significant (p < 0.05) between the single dose GSI and placebo groups are shown in this table. The q-value (estimate of the false discovery rate) for each analyte is also shown. The analytes listed were measurable in all samples. One sample from a subject receiving GSI in period 1 was not available in sufficient volume for the RBM analysis.

Table 3

The output for each marker's analysis included the data shown in Table 3.

All markers were statistically significantly different between single dose GSI and placebo (p<0.05). CSF Aβ40 and CSF Aβ42 had the most robust effect at less than 5% false discovery rate (FDR), i.e., q<0.05.

Human multiple dose study of a GSI versus placebo

A double-blind, randomized, placebo-controlled, two period crossover study in healthy young men in a single study center was conducted. The second part of this study consisted of subjects receiving daily witnessed doses on ten consecutive days of each period of either Compound A or a matching placebo according to a randomized schedule. On the last day, CSF was drawn via lumbar puncture at 12 hours post dosing (see Example 3 for details). For Compound A plasma concentrations and determination of plasma Aβ40 and sAPP levels, blood was collected at various time points. The duration of the study for each subject was approximately four weeks and the study took about six weeks to complete. CSF samples from each of the 12 subjects on both placebo and Compound A (total N=24 samples) were sent to for RBM human MAP analysis, in a similar manner to that described for the rhesus monkey study. The response of the human MAP to the human CSF is shown in Figure 5.

Similar to the rhesus study, univariate analysis was applied to each marker in the analysis, including CSF Aβ40 and CSF Aβ42 Table 4 is a summary of the analytes determined from the human multiple dose study of a

GSI versus placebo. Analytes that were statistically significant (p < 0.05) between the multiple dose GSI and placebo groups are shown. The q-value (estimate of the false discovery rate) for each analyte is also shown. The analytes listed were measurable in all samples. One sample from one subject receiving placebo in period 1 was not available in sufficient volume for the RBM analysis.

Table 4

All markers were statistically significantly different between multiple dose GSI and placebo (p<0.05). CSF Aβ40 and CSF Aβ42 had the most robust effect at less than 10% false discovery rate (FDR), i.e., q<0.10.

Serum myloid P as a potential biomarker of γ-secretase inhibition

A literature search for serum amyloid P (SAP) and γ-secretase did not identify any published reports indicating a correlation between SAP and γ-secretase. SAP binds gram negative bacteria and inhibits lipopolysaccharide-mediated activation of the classical complement pathway. It is known to be a precursor of tissue amyloid P component associated with primary idiopathic amyloidosis and Alzheimer Disease deposits. It is believed to be produced not only in the liver, but also locally produced and up regulated in the AD brain, as it associates well with amyloid plaques (McGeer et al., Neurobiol. Aging, 22: 843 (2001)). CSF SAP was previously excluded from a list of potential γ-secretase inhibition biomarkers as it was not significantly different (p=0.40) between well characterized AD (60 +/- 41 ) and control samples (50 +/- 34) (Mudler et al., J. Neural Transmission, 109: 1491 (2002)), which was in contrast to earlier reports (Hawkins et al., Biochem. Biophys. Res. Comm., 201 : 722 (1994); Kimuar et al., Neurosci. Lett., 273: 137 (1999)) that found such a correlation. As such, it was unexpected from the studies herein to determine SAP to be a putative γ-secretase inhibition marker.

The univariate analysis of each analyte separately between high dose and placebo samples in rhesus monkeys and human single and multiple dose studies identified several putative γ- secretase inhibition markers, including the analytes serum amyloid P (SAP), CD40 Ligand, ferritin, macrophages-derived chemokine (MDC), apolipoprotein H (ApoH), leptin, MIP-I -β, prostatic acid phosphatase (PAP), TNF-R-II, fatty acid binding protein (FABP) and TNF-α]. These analytes in the

® RBM Human MAP were statistically significant with a false positive rate of (p) < 0.05. Among these analytes, SAP was the most preferred in that it also had a false discovery rate of (q) < 0.05 in addition to the most robust effect. The standard amyloid markers, CSF Aβ40 and CSF Aβ42, are also statistically significant (p<0.05), while plasma Aβ40 trended towards significance in the selected rhesus sample set.

EXAMPLE 1 CMP rhesus in vivo study and samples

A. CMP rhesus monkey model Applicants developed a unique research model herein referred to as the cisterna magna ported (CMP) rhesus monkey model. The initial requirement in development of this non-human primate (NHP) model was to establish surgical procedures and materials for implantation of catheters to provide a sterile system for accessibility to cerebrospinal fluid for noninvasive sampling. A review of the literature and a search throughout existing regional primate centers and laboratories indicated that such a model did not exist. Through trial and error, optimal surgical conditions were established for the CMP rhesus monkey model (Gilberto et al., Contemp. Topics Lab Anim. ScL 42: 53-59 (2003)). It was also determined that permanent/chronic implantation had clear advantages over acute puncture and sub-acute lumbar catheterization. That is, the CSF collected from the chronic implant was routinely pristine while other methods typically had some level of blood contamination as seen visually and determined by hemoglobin content. In addition to the advantages of uncontaminated CSF samples, the chronically implanted cisterna magna catheter and port system (CMP) provides the capability for repeat sampling over the course of a daily/ weekly experiment and the chronic nature of the system allows for longitudinal studies and comparison to historical data within the same monkeys. Once a CMP is surgically implanted and CSF flow has been established, the ports have remained patent for up 46 months. For this model, a flexible catheter with multiple openings was freely suspended in the cisterna magna and anchored firmly on both sides of the atlanto-occipital membrane. An extension catheter was tunnelled subcutaneous Iy from the outside of the membrane to the midscapular region where it fed into an implanted port body. CSF was accessed by inserting a needle through the skin and membrane covering the reservoir in the port body under sterile conditions. In the upright NHP, CSF flows by gravity through the cannulation system without the need for active withdrawal. This provides the critical benefit of protection of delicate CNS tissue from negative pressure. To maintain port patency, the CSF is accessed at least twice weekly under aseptic conditions in all CMP rhesus monkeys.

B. CMP rhesus study of GSI versus placebo A γ-secretase inhibitor, Compound A,

was evaluated in conscious CMP rhesus monkeys in a four-way crossover design (vehicle, 0.5%methylcellulose 5 mL/kg versus 10, 30 and 60 mpk PO by nasogastric lavage) to identify CSF biomarkers. Compound A or vehicle was administered at T= 0 hour (following T=O sampling) in six monkeys. CSF (2 mL) and plasma (4 mL EDTA) were collected at the following time points relative to dose administration: T= -20, 0, 4, 24, 144 and 268 hours. One mL of CSF was centrifuged at 4000 X g for 4 minutes at 4^°C for CSF analyses. The remaining I mL of CSF was separated into aliquots for analysis of APP metabolites (neat) and compound concentration (CSF (250 μL CSF + 2.5 μL 10% tween 20). The 2 mL of plasma was split evenly into five tubes for PK and PD analyses. Sufficient material was obtained to bank aliquots of CSF and plasma for further use. The selected doses showed no significant effect with the lowest dose (10 mpk) and dose-related effects with the intermediate and high doses (30 and 60 mpk, 34 and 54 %reductions in CSF AB42 levels at 4 hours post-dose). Subsequently, selected banked samples of CSF were submitted for analytical characterization according to methods described in Example 4 below: 16 samples were selected from vehicle (N=5), 30 mg/kg (N=5) and 60 mg/kg (N=6) treatments based on Aβ42 effects and sent to Rules- Based Medicine, Inc., Austin, TX (RBM) for pilot analysis in their human multi-analyte profile (MAP).

The putative markers of γ-secretatse inhibition determined from the analysis of this study are shown in Table 2.

EXAMPLE 2 Compound A single dose human study and samples

A double-blind, randomized, placebo-controlled, two period crossover study in healthy young men in a single study center was conducted.

In the first part of the study, on day 1 of each period subjects received a single oral dose of either Compound A (500 mg) or a matching placebo according to a randomized schedule. CSF was drawn via lumbar puncture (LP) at 12 hours post dosing. For Compound A plasma concentrations and determination of plasma Aβ40 and sAPP levels, blood was collected prior to dosing and at 1 , 2, 3, 4, 6, 8, 12, and 24 hours post dosing. The washout period between the two periods was ten days. A period of nine days or eleven days was allowed if it was determined to be necessary. A washout period is defined as the period time when no additional drug is administered and the organism is given time to metabolize or secrete any of the compound out of its system, in essence, allowing the compound to "wash-out". The duration of the study for each subject was approximately three weeks and the study was completed within five weeks.

The putative markers of γ-secretase inhibition determined from the analysis of this study are shown in Table 3.

EXAMPLE 3 Human Compound A multiple dose study and samples

A double-blind, randomized, placebo-controlled, two period crossover study in healthy young men in a single study center was conducted.

In the second part of this study, subjects received daily witnessed doses on days 1 through 10 of each period of either Compound A or a matching placebo according to a randomized schedule. On day 10 CSF was drawn via lumbar puncture at 12 hours post dosing. For Compound A plasma concentrations and determination of plasma Aβ40 and sAPP levels, blood was collected prior to dosing and at 1, 2, 3, 4, 6, 8, 12, and 24 hours post dosing on Day 10 and prior to dosing on days 1 through 9. The washout period between the two periods was ten days. A washout period of nine days or eleven days was allowed if it was determined to be necessary. The duration of the study for each subject was approximately four weeks and the study took about six weeks to complete.

The putative markers of γ-secretase inhibition determined from the analysis of this study are shown in Table 4.

EXAMPLE 4

® Measuring analytes in the RBM human Multi-Analyte Profile (Human MAP )

CSF samples of placebo and Compound A treatment were sent to Rules-Based Medicine,

Inc. (RBM) (Austin, TX) for analysis in their proprietary human Multi-Analyte Profiles (Human MAP

® Version 1.6) under a fee for service agreement. The Human MAP is a Luminex bead based multi- antigen profile consisting of at least three pools of beads representing some 90 total antigens (Figures 2A

® and 2B). Designed originally for human plasma analysis of 100 μL samples, the Human MAP has been analytically validated according to NACLES criteria and successfully used on human CSF when using 200 μL samples in manual mode and 250 μL samples in automated mode. As seen in Figures 3, 4A/B and 5 A/B, the rhesus and human CSF samples responded well in the antigen panel originally designed for plasma, producing at least 50 analytes with good measurements in the majority of the CSF samples.

EXAMPLE 5 Method to identify biomarkers of GSI from the univariate analysis of the CMP rhesus study

Expression of each of the three amyloid markers (CSF Aβ42, CSF Aβ40 and plasma Aβ40) and the 89 markers from the RBM panel was compared between the high dose GSI (60 mpk) and placebo group rhesus CSF samples using analysis of variance (ANOVA). The analysis was carried out after applying logarithmic transformation on the markers to ensure approximate symmetry. The p-values (false positive rate) and q-values (false discovery rates) from this analysis on the 92 (89+3) markers were determined. False Positive Rate (FPR) or p-value estimates the proportion of false positives among all the proteins that in reality did not change. False discovery rate (FDR) or q-value estimates the proportion of significant changes that are false positives.

Summary statistics obtained from these analyses for each of the markers included the ratio of high dose GSI to placebo using the least squares means from the ANOVA, the false positive rate (p-value) and the false discovery rate (q value) of the high dose GSI versus placebo comparison. Relevant figures that characterize the relationship between these markers and treatment are included as Figures 6 through 12. The analysis was carried out using R version 2.4 (R: A Language and Environment for Statistical Computing, R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2005, ISBN 3-900051 -07-0). Figures 6 through 12 were generated using JMP v5.0.1 from the SAS Institute (Cary, NC). The false discovery rate (q values) for each marker was derived using the method proposed by Benjamini & Hockberg (Benjamini & Hockberg, The adaptive control of the false discovery rate in multiple hypotheses testing with independent statistics, J. Behav. Educ. Statist.. 25: 60- 83 (2000)) and determined using a contributed library within R (Strimmer, Estimation and Control of (Local) False Discovery Rates, fdrtool package, version 1 , August 8, 2006). The analysis of each marker individually was automated in R's scripting language, to produce a csv (comma separated variable) output file, readable in programs like MS Excel. A summary of the statistically relevant analytes from this study is shown in Table 2.

EXAMPLE 6 Method to identify GSI biomarkers from univariate analysis of human single dose study

Expression of each of the two amyloid markers (CSF Aβ42 and CSF Aβ40) and the 89 markers from the RBM panel was compared between the single dose GSI and placebo groups using a linear mixed effects model. Since this study was a balanced two-period crossover design, the linear mixed effects model included sequence, period and treatment as fixed effects and subjects as random effect. The analysis was carried out after applying logarithmic transformation on the markers to ensure approximate symmetry. The p-values (false positive rate) and q-values (false discovery rates) from this analysis on the 91 (89+2) markers were determined. False Positive Rate (FPR) or p-value estimates the proportion of false positives among all the proteins that in reality did not change. False discovery rate (FDR) or q-value estimates the proportion of significant changes that are false positives.

Summary statistics obtained from these analyses for each of the markers included the ratio of GSI to placebo using the least squares means from the linear mixed effects model, the false positive rate (p-value) and the false discovery rate (q value) of the GSI versus placebo comparison.

Relevant figures that characterize the relationship between these markers and treatment are included as Figures 13 through 19. Analysis was carried out using the "lme" (i.e. linear mixed effects) function in R version 2.4 (R: A Language and Environment for Statistical Computing, R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2005, ISBN 3-900051 -07-0). Figures 13 through 19 were generated using JMP v5.0.1 from the SAS Institute (Cary, NC). The false discovery rate (q values) for each marker was derived using the method proposed by Benjamini & Hockberg (Benjamini & Hockberg, The adaptive control of the false discovery rate in multiple hypotheses testing with independent statistics, J. Behav. Educ. Statist. 25: 60-83 (2000)) and determined using a contributed library within R ( Strimmer, Estimation and Control of (Local) False Discovery Rates, fdrtool package, version 1 , August 8, 2006). All the analysis of each marker individually were automated in R's scripting language, to produce a csv (comma separated variable) output file, readable in programs like MS Excel. A summary of the statistically relevant analytes from this study is shown in Table 3.

EXAMPLE 7 Method to identify markers of GSI from univariate analysis of the human multiple dose study

The expression of each of the two amyloid markers (CSF Aβ42 and CSF Aβ40) and the 89 markers from the RBM panel was compared between the multiple dose GSI and placebo groups using a linear mixed effects model. Since this study was a balanced two-period crossover design, the linear mixed effects model included sequence, period and treatment as fixed effects and subjects as random effect. The analysis was carried out after applying logarithmic transformation on the markers to ensure approximate symmetry. The p-values (false positive rate) and q-values (false discovery rates) from this analysis on the 91 (89+2) markers were determined. False Positive Rate (FPR) or p-value estimates the proportion of false positives among all the proteins that in reality did not change. False discovery rate (FDR) or q-value estimates the proportion of significant changes that are false positives.

Summary statistics obtained from these analyses for each of the markers included the ratio of GSI to placebo using the least squares means from the linear mixed effects model, the false positive rate (p-value) and the false discovery rate (q value) of the GSI vs. Placebo comparison. Relevant figures that characterize the relationship between these markers and treatment are included as Figures 20 through 23. Analysis was carried out using the "lme" (i.e. linear mixed effects) function in R version 2.4 (R: A Language and Environment for Statistical Computing, R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2005, ISBN 3-900051 -07-0). Figures 20 through 23 were generated using JMP v5.0.1 from the SAS Institute (Cary, NC). The false discovery rate (q values) for each marker was derived using the method proposed by Benjamini & Hockberg (Benjamini & Hockberg, The adaptive control of the false discovery rate in multiple hypotheses testing with independent statistics, J. Behav. Educ. Statist. 25: 60-83 (2000)) and determined using a contributed library within R ( Strimmer, Estimation and Control of (Local) False Discovery Rates, fdrtool package, version 1 , August 8, 2006). The analysis of each marker individually was automated in R's scripting language, to produce a csv (comma separated variable) output file, readable in programs like MS Excel. A summary of the statistically relevant analytes from this study is shown in Table 4.

EXAMPLE 8 Multi-variate analysis of GSI markers for monitoring γ-secretase inhibition or efficacy

One skilled in the art would appreciate that the putative univariate markers of the present invention can be utilized to carry out a multi-variate analysis. In particular, the combined ability of several CSF markers identified in Examples 5, 6 and 7 above to monitor γ-secretase inhibition could be assessed using Linear Discriminant Analysis (LDA). This analysis measures the distance from each point in the data set to each group's multivariate mean (called a centroid) and classifies the point to the closest group. The distance measure is the Mahalanobis distance, which takes into account the variances and covariances between the variables.

The assessment of relative importance of the biomarkers with respect to their ability to discriminate AD from control subjects could be done within the framework of the random forest analysis (Breiman, 2001 ) as well. In each tree of the forest (i.e., large collection of trees derived from many simulated samples from the original data), data are permuted in one biomarker at a time and predictions on the permuted data from the random forest method are obtained. These are then compared to the predictions from the unpermuted data and the loss in accuracy is assessed. A large loss in accuracy, represented as the percent mean decrease in accuracy ("Mean Decrease Accuracy"), indicates the relative importance of the corresponding biomarker. In this way, one identifies the most important markers among a panel of markers. With respect to LDA, the predictive ability of these markers to separate AD from control for hold-out datasets is investigated using fifty replicates of 10-fold cross validation. This analysis entails dividing up the data randomly into ten subgroups, using the fitted model from nine of these subgroups to predict the efficacy of the γ-secretase inhibitor in the tenth subgroup, repeating this for all ten groups and then averaging the results across all ten repetitions. This analysis is repeated fifty times to generate a reliable estimate of the overall accuracy along with the sensitivity and specificity of the biomarker composites for monitoring efficacious from non-effactious GSI activity.

Analyses is based on more than five samples in each group of placebo and GSI treated at multiple doses. The LDA graph is generated using JMP software, v5.0.1 from SAS Institute (Cary, NC). All other analyses are performed using R (R: A Language and Environment for Statistical Computing, R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2005, ISBN 3- 900051 -07-0).

LDA for assessing the performance of the various composites of markers is performed using a contributed library within R (Venables, W. N. and Ripley, B. D., Modern Applied Statistics with S₁, Fourth Edition, Springer, New York (2002) ISBN 0-387-95457-0). The random forest analysis is performed using a contributed library within R (Andy Liaw and Matthew Wiener, Classification and Regression by Random Forest, R News. 2 (3): 18-22 (2002)). The 10-fold cross-validation for obtaining reliable estimates of the performance metrics of the biomarker composites is performed using a contributed library within R (Andrea Peters and Torsten Hothorn, Improved Predictors. R package version 0.8-3(2004)).

One skilled in the art would understand that the addition or subtraction of a marker to a composite may not improve the performance.

EXAMPLE 9 Monitoring of γ-secretease inhibition and/or GSI efficacy

According to the present invention, one skilled in the art would appreciate that the putative γ-secretase inhibition markers can be utilized to monitor γ-secretase inhibition or the efficacy of a γ-secretase inhibitor in an individual or cohort of individuals clinically by selecting at least one γ- secretase inhibitor marker previously identified as being statistically significant and which is differentially expressed with successful compound treatment. In particular, at least one marker is measured according to the methods described in Examples 5, 6 and 7 above in the CSF of an individual at an initial time point and then the same marker(s) is measured in a second CSF sample taken at a subsequent time, such as after treatment with a γ-secretase inhibitor. One utilizes the biochemical marker as a response variable (on the y-axis) to plot the change in the marker as a function of treatment group. Using a putative γ-secretase inhibition marker previously identified according to the methods of Example 5, 6 and 7, an assessment is made regarding the efficacy of the treatment in the individual. If the selected γ-secretase inhibition marker does not change after treatment compared to the reference level, this is an indication that the compound was not efficacious in inhibiting γ-secretase. If the γ-secretase inhibition marker changes at later time points as specified above in Examples 5, 6 and 7, then this is indicative of γ- secretase inhibition and allowing one skilled in the art to monitor γ-secretase inhibition and/or efficacy.

Claims

WHAT IS CLAIMED:

1. A γ secretase inhibition marker selected from the group consisting of apolipoprotein H (ApoH), CD40-Ligand (CD40L), fatty acid binding protein (FABP), ferritin, leptin, MIP-I -β, MDC, prostatic acid phosphatase (PAP), serum amyloid P (SAP), TNF-α and TNF-R-II.

2. A γ secretase inhibition marker of claim 1 wherein said marker is selected from the group consisting of CD40L, ferritin, MDC and SAP.

3. A γ secretase inhibition marker of claim 1 wherein said marker is selected from the group consisting of FABP and TNFα.

4. A γ secretase inhibition marker of claim 1 wherein said marker is selected from the group consisting of ApoH, leptin, MIP-I β and PAP.

5. A method for monitoring the inhibition of γ secretase in an AD patient or control subject comprising: a. selecting at least one γ secretase inhibition marker previously identified as being statistically significant and which is differentially expressed; b. obtaining a fluid sample from an AD patient or control subject; c. analyzing the fluid sample of step (b) for the presence of the γ secretase inhibition marker or markers to form a reference level; d. obtaining a second fluid sample from said AD patient or control subject at a later prescribed time interval; e. analyzing the second sample for the presence of the γ secretase inhibition marker or markers; f. comparing the results of step (e) to those of step (c) to obtain an output; g. where the output of step (f) is the change in the marker levels between the two sample measurements; and where a change in the marker levels is indicative of γ secretase inhibition.

6. A method for monitoring the efficacy of an AD therapeutic comprising: a. selecting at least one γ secretase inhibition marker previously identified as being statistically significant and which is differentially expressed; b. obtaining a fluid sample from an AD patient or control subject to which the AD therapeutic is to be administered; c. analyzing the fluid sample of step (b) for the presence of the γ secretase inhibition marker or markers to form a reference level; d. administering an AD therapeutic to said AD patient or control subject; e. obtaining a second fluid sample from said AD patient or control subject at a prescribed time interval after therapeutic administration; f. analyzing the second sample for the presence of the γ secretase inhibition marker or markers; g. comparing the results of step (f) to those of step (c) to obtain an output; h. where the output of step (g) ) is the change in the marker level between the two samples; and where a change in the marker level is indicative of γ secretase inhibition.