Community Research and Development Information Service - CORDIS

  • European Commission
  • CORDIS
  • Projects and Results
  • Final Report Summary - SCHIZDX (Developing minimally invasive, tools and technologies for high throughput, low cost molecular assays for the early diagnosis of schizophrenia and other psychiatric disorders)
FP7

SCHIZDX Report Summary

Project ID: 223427
Funded under: FP7-HEALTH
Country: United Kingdom

Final Report Summary - SCHIZDX (Developing minimally invasive, tools and technologies for high throughput, low cost molecular assays for the early diagnosis of schizophrenia and other psychiatric disorders)

Executive Summary:
The key goal of the SchizDX project was to develop “Minimally Invasive, High Throughput, Low Cost Molecular Assays for the Early Diagnosis of Schizophrenia and Other Psychiatric Disorders”. The project had 3 key phases: 1) Discovering candidate biomarkers that may be capable of accurately classifying schizophrenia patients based on the expression profiles of these proteins in the blood of individuals; 2) Developing a panel of high throughput assays to measure the biomarker fingerprints of selected proteins in larger numbers of individuals in a minimally invasive, cost effective manner; and 3) Validating the candidate biomarkers and defining their classification performance in the identification of patients and control samples from a prospectively collected cohort of individuals suffering from a range of psychiatric disorders.

The current diagnosis of schizophrenia involves examination and monitoring by psychiatrists, a process which can take between 2 and 5 years and is a rather subjective and sometimes long-lasting uncertain process. Earlier research had indicated that brain “signature” changes in biomolecules should be reflected by changes in the cerebrospinal fluid (CSF) and blood (serum/ plasma) levels of the same or related molecules. It is through this process that peripheral biomarkers for brain disorders, such as schizophrenia, can be identified.

Glycoprotein analysis was carried out on serum and CSF samples from schizophrenia patients and healthy controls. Statistically significant glycoprotein changes were found in serum and CSF samples from first-onset schizophrenia patients compared to control subjects. Comparative proteomic and phosphoproteomic analyses of serum from 20 first-onset antipsychotic-naïve schizophrenia patients and 20 control subjects was also carried out. 24 proteins were identified with altered levels in schizophrenia and 10 of these were phosphoproteins. Samples were collected from first onset drug naive schizophrenia patients before and after treatment with antipsychotic drugs for the identification of drug response protein biomarkers for schizophrenia using LC-MSE. Two proteins, lumican and apolipoprotein C2, were increased by all treatments. Glycosylation-specific changes were also found in serum from schizophrenic patients after treatment with olanzapine.

Multiplex immunoassay profiling of drug treated schizophrenic patients before and after treatment identified 7 candidate treatment response biomarkers that changed in expression levels in two or more of the following treatment cases: olanzapine, risperidone, quetiapine, or a combination. These data suggest that even antipsychotic compounds with overlapping pharmacological profiles are likely to have specific treatment response biomarker signatures.

A novel ex vivo blood culture system (TruCulture) combined with the use of a 33-plex cytokine/chemokine immuno assay panel was implemented to identify further schizophrenia biomarkers and to provide a clinically-relevant system for identification of drug-predictive or response biomarkers and for novel drug discovery.

The project has resulted in the filing of 9 patent applications. In addition, the project has so far resulted in 21 publications in peer reviewed journals and at least 4 others which have been submitted.

The SchizDX project resulted in the launch of the first molecular test for the diagnosis of schizophrenia (VeriPsych®) by Psynova Neurotech and Myriad-RBM in the USA. This is a novel 51-plex immunoassay panel that allows reproducible identification of schizophrenia patients compared to controls with high sensitivity and specificity for presence of schizophrenia in subjects compared to matched controls (806 subjects), with a receiver operating characteristic-area under the curve of 88%. The test is being marketed to assist with the diagnosis of first onset schizophrenia and may result in a quicker and more cost effective diagnosis for some patients. It is well known that early intervention can result in a much better outcome for schizophrenia patients. Therefore, the outputs from the SchizDX project may be of significant benefit to patients.

Project Context and Objectives:
Schizophrenia is a complex psychiatric disorder affecting approximately 1% of the population with equal risk across genders. Schizophrenia and bipolar affective disorder are a major burden to affected individuals and their families and to society at large. These two severe mental illnesses affect at least 2% of the population worldwide, and whilst 50% of sufferers do not receive adequate treatment, they cost hundreds of billions in healthcare provision, treatments and lost earnings.

The current diagnosis of schizophrenia (and bipolar disorder etc.) is rather subjective and psychiatrists currently come to their diagnosis based on observation of both the presence and duration (up to 6 months) of certain signs and symptoms. Many times patients visit their doctor during the prodrome phase due to anxiety, social isolation, difficulty making choices, and problems with concentration and attention, symptoms presenting many psychiatric and medical conditions, making absolute diagnosis difficult and protracted. This often results in long periods (up to 1-3 years) of untreated psychosis during which the disease severity increases before appropriate therapeutics are prescribed to first episode schizophrenics. Schizophrenia diagnosis is difficult not only because of the complex spectrum of symptoms and their similarity to other mental disorders, but also due to the lack of empirical disease markers. In addition, contemporary drug treatments do not effectively treat all aspects of the disease and can often have severe side effects that make it difficult for many patients to continue with medication. Despite intensive efforts by the pharmaceutical industry, therapeutic regimes available to treat these disorders are largely aimed at relieving symptoms and work best at slowing or halting the underlying disease progression in early stage or less severe cases, making early and accurate diagnosis essential. It is worth noting that, where early treatment has been possible, it is associated with greatly improved patient outcomes. Early treatment is associated with greatly improved patient outcomes. There is therefore a major unmet clinical need for empirical diagnostic tests for high throughput screening of biological fluids that would enable early and accurate diagnosis of schizophrenia and related disorders.

The identification of specific biomarkers for mental disorders would revolutionise the clinical management of affected individuals. Biomarkers will help in the identification of disease sub-types, aid in predicting and monitoring treatment response and compliance, and identify novel drug targets. If such biomarkers can be found in readily accessible body fluids they open up the possibility of developing new early or pre-symptomatic diagnostics and/or treatments to improve outcomes or even prevent disease.

The objective of the SchizDX project is to identify biomarkers of disease and develop a diagnostic assay panel for the high throughput screening of biological samples for clinical research. Moreover this platform will be utilised by SMEs for drug design and development, research into new animal models and identifying biomarkers for other mental disorders.

The SchizDX application proposed to identify and validate up to 50 biomarkers relevant to the first onset of schizophrenia. In this project label free nano-LC-MS based proteomic profiling, in combination with glycoprotein profiling and existing analyte assay screening, will be utilised to establish a schizophrenia-specific disease biomarker pattern/signature detectable in blood. The identification of specific biomarkers for mental disorders would revolutionise the clinical management of affected individuals. Biomarkers have the potential to help in the identification of disease sub types, aid in predicting and monitoring treatment responses and compliance, and facilitate novel approaches to drug discovery. This approach would also allow us to develop an understanding of the basic mechanisms that underlie the disease processes of schizophrenia and bipolar disorder. If such biomarkers can be found in readily accessible body fluids, they open up the possibility of developing new early or pre-symptomatic clinical tools to aid diagnosis, speed up treatments and improve patient outcomes or even prevent disease.

Furthermore, biomarker assays that can detect early changes specifically correlated to reversal or progression of clinical symptoms of schizophrenia would have tremendous potential benefit for intervention studies. Used as predictors, these biomarkers can help to identify high risk individuals and disease sub groups who could serve as target populations for chemo-intervention trials, whilst as surrogate end points, biomarkers have the potential for assessing the efficacy and cost effectiveness of novel preventative interventions with an efficiency not possible currently, when the severity of manifest mental disorder is used as the end point.

There is therefore clearly a major unmet clinical need for empirical biomarker tests to allow for the high throughput screening of clinical samples that would enable early and more accurate diagnosis of schizophrenia and related disorders.

This platform will also be utilised by SMEs for;
• Research into drug design and development by assessing the impact of novel drug entities on biomarkers for mental disorders.
• Research into the development of better animal models for mental disorders.
• Research into identifying proteins/peptides as biomarkers for other mental disorders

The rationale for early detection of schizophrenia is based on several observations:

Diagnosis and treatment of schizophrenia are often seriously delayed. Thus, patients on average suffer from productive psychotic symptoms such as delusions or hallucinations for an average of 1–3 years before this disorder is finally diagnosed and treated for the first time (duration of untreated psychosis, DUP). In addition, even before then, patients suffer from an ‘‘unspecific prodromal phase’’ for an average of 2–5 years (duration of untreated illness, DUI) (for reviews, see e.g. Norman and Malla, 2001; Bottlender and Moller, 2003; Riecher- Rossler et al., in press).

Consequences of the disease can be very severe, even in the early preclinical, undiagnosed phase of the disorder. Before first admission, most patients already suffer from serious impairments and losses in various social domains, such as education, work, partnership or independent living (Hafner et al., 1995). Recent studies have also shown that the quality of life is seriously impaired already at first admission and is associated with DUP (Browne et al., 2000).

Early treatment seems to improve the course of the disease. There is a large body of evidence confirming that the early treatment of psychosis can substantially improve the course and outcome of the disease (for reviews, see e.g. Norman and Malla, 2001; McGorry, 2002; Harrigan et al., 2003; Riecher-R?ssler et al., in press). Thus, the majority of studies found a statistically significant correlation between a long duration of untreated psychosis (DUP) and a poor outcome.

Identification of fundamental disease mechanisms using a “multi omics approach”

At present little is known about the basic mechanisms that underlie the disease processes of schizophrenia and bipolar disorder. This lack of knowledge is partly due to the fact that, until recently, global expression profiling studies were technologically impossible. Thus most researchers employed a “candidate gene” approach to look for the “needle in the haystack”. With recent technological advances in the “omics” field, genomics, proteomics and metabolomics (functional genomics) the timing is now right to organize and concerted effort to understand the fundamental disease processes of psychiatric disorders and to translate this improved knowledge into new (pre-symptomatic) diagnostic, therapeutic and preventative regimes.

The combined application of advanced computing and bioscience technologies with functional genomics studies offers unprecedented powerful approaches to explore the molecular “fingerprints” of diseases from early onset through their progressive stages, exploring alteration at the gene, protein, lipid and metabolite level.

Novel technologies offer new investigative avenues

Recently there have been technological advances that allow for the identification of disease specific markers or patterns from patients and controls. Techniques such as DNA microarrays (genomics), proteomics (including cytokine/neurotrophin profiling) and metabolomics, combined with chemometric analysis (e.g. principle component analysis (PCA)), orthogonal signal correction (OSC) and partial least square determinant analysis (PLS-DA) have already been shown to offer an efficient approach to identify individuals with diseases.

Among these novel technologies, label free proteomics, championed by the Bahn lab (U. Cambridge), possesses a great potential for biomarker discovery and the development of antibody-based, high throughput biomarker assays, especially for serum and plasma samples (Levin et al 2007; Huang et al 2007a). Indeed some of these comprehensive approaches have recently shown their high sensitive and specificity to diagnose the presence and severity of coronary heart disease (Brindle et al., 2002) and to identify candidate biomarkers in various cancers (Fujii et al 2005; Hoffman et al, 2007)). In addition, high-throughput glycan analysis and subsequent identification of relevant glycoproteins has demonstrated great potential in identifying secreted biomarkers (Sheridan 2007). The Dublin-Oxford Glycobiology laboratory has developed and will use sensitive methods to fragment glycans and identify disease specific markers (Vickers et al, in press). Lastly, it also makes sense to check whether existing biomarkers identified for a variety of disorders have any utility in schizophrenia. Therefore we will profile >200 existing analytes using a Luminex platform (EDI GmbH) against first episode, drug naïve schizophrenics compared to healthy controls and following antipsychotic treatment to see if any of these existing assays have potential as diagnostic or treatment responsive biomarkers of schizophrenia.

There are six scientific/technical objectives

1. To identify up to 50 candidate biomarkers, specifically up-or down-regulated in drug naive, first onset schizophrenics compared to healthy controls or changing in first onset schizophrenics following 4 weeks of antipsychotic drug treatment, using high throughput proteomic profiling methods. (M18)

2. To develop high throughput, sandwich immunoassays to allow the validation of candidate biomarkers identified in the discovery phase as required. (M36)

3. To validate candidate biomarkers against a larger validation sample set to identify those most useful for further diagnostic assay panel development. (M36)

4. To develop a diagnostic assay panel allowing the high throughput analysis of samples for schizophrenia and to carry out a clinical proof of concept study on at least 800 psychiatry inpatient serum samples. (M42)

5. To commercialize novel biomarker assay panels for clinical profiling to aid diagnosis and drug development. The panel will also be utilized in the development of better animal models. (M42+)

6. To further our understanding of the pathophysiology of schizophrenia. (M1+)

Project Results:
Work Package 3 - Candidate Biomarker Discovery
Objectives

The objective of this work package is to identify up to 50 candidate biomarkers specifically up-or down-regulated in drug naive, first onset schizophrenics CSF and serum samples or changing in first onset schizophrenics following 4 weeks of antipsychotic drug treatment. This was achieved using three complementary approaches:

Label-free nano-LC-MSE-based proteomic profiling
Glycoprotein profiling
Multiplex immunoassay profiling

It was agreed that the CSF profiling objective was removed as an objective to enable greater focus on the serum samples which could be obtained in higher numbers and this was agreed at the mid period review.

Task 3.1. Label-free LC-MSE based proteomic profiling of 20 CSF samples from first onset, drug naive schizophrenic patients and 20 matched normal controls

The original goal for this deliverable was to identify candidate biomarkers that are significantly up- or down-regulated in serum and/or CSF of first episode schizophrenia patients compared to carefully matched controls. This formed the initial plan for work package 3 of the application [Candidate Biomarker Discovery]

Stage 1. Biomarker discovery study (original plan)

This first phase of the project would focus on the identification of schizophrenia-specific proteomics patterns and candidate biomarkers using state of the art proteomic profiling methodologies on the following sample sets (already collected and available for analysis from CIMH and Muenster):

Drug naive, first onset schizophrenics vs matched controls [n=80 in total 20 schiz; 20 controls; both CSF and serum];

Originally it was intended that we would use three complementary state of the art profiling platforms to enable the deepest interrogation of the relevant proteomes for specific biomarkers.

Label-free nano-LC-MS based proteomic profiling (Psynova/U. Cambridge)
Glycoprotein profiling (The National Institute for Bioprocessing Research and Training/Psynova)
Existing multianalyte profiling using Rules Based Medicine’s (RBM) HuMAP assay panel (89 analytes)

The goal of these studies was to identify up to 50 candidate biomarkers significantly up or down regulated in schizophrenic samples compared to controls. However, during the delay between submitting the grant application and actually initiating the project, three developments resulted in a change to this plan.

The University of Cambridge performed a label-free LC-MSE experiment to profile proteins of 54 CSF samples from first onset schizophrenia patients and matched controls. Strong correlations were found between almost all proteins which were likely to be caused by external variables such as the cerebral flow or the hydration state of the individual subjects in the study. In biomarker discovery studies, these differences between samples have to be normalised in order to detect potentially small, disease relevant molecular changes. It was demonstrated that this normalisation can have a profound effect on the correlation structure between originally uncorrelated analytes and may, therefore, lead to erroneous downstream interpretation of the data. It was concluded that no reliable biomarkers could be identified in CSF given these concerns.

Preliminary pilot studies demonstrated that both LC-MS and multianalyte profiling approaches identified dozens of candidate biomarkers in studies on serum. As the goal of the project is to develop “minimally invasive, high throughput, low cost molecular assays” for the early diagnosis of schizophrenia and other psychiatric disorders it was agreed amongst partners to focus efforts on development of the obviously less invasive and higher throughput blood based assays.

It was originally intended that NIBRT would focus on the identification of novel glycoprotein candidate biomarkers. However, due to the 9 month delay in funding being available from the EC after the project was initiated and the excellent progress made in the meantime in identifying many candidate biomarkers (see later) it was agreed amongst the project participants that NIBRT would focus on investigating potential schizophrenia specific glycobiology changes on the candidate biomarkers already identified by LC-MS and multianalyte profiling methods, rather than identify further marker candidates per se.

A modified plan was proposed in that the first phase of the project focused on the identification of schizophrenia-specific proteomics patterns and candidate biomarkers using state of the art proteomic profiling methodologies on the following sample sets (already collected and available for analysis from CIMH and Muenster):

Serum samples obtained from drug naive, first onset schizophrenia patients vs. matched controls [n= at least 20 schiz; 20 controls]

Two discovery platforms were proposed.

Label-free nano-LC-MS based proteomic profiling (Psynova/U. Cambridge)
Existing multianalyte profiling using Rules Based Medicine’s (RBM) HuMAP assay panel

For the reasons outlined above, this task was removed. After review and discussion with the REA and technical reviewer this was accepted.

Task 3.2. Label-free LC-MSE based proteomic profiling of 20 serum samples from first onset, drug naive schizophrenic patients and 20 matched normal controls

In the first study, a total of 55 individual (non-pooled) clinical serum samples and 12 QC serum samples were compared in this study. These comprised 22 samples taken from first-onset drug-naive patients and 33 samples from demographically matched healthy volunteers from centre 1 (CIMH). The samples were analysed using a label-free nano-LCMSE- based global proteomic profiling approach as described in the work package description.

In the second study, a total of 37 individual (non-pooled) clinical serum samples were compared. These comprised 20 samples taken from first-onset drug-naive patients and 17 samples from demographically matched healthy volunteers from centre 1 (CIMH).

The samples were analysed using a similar label-free nano-LCMSE- based global proteomic profiling approach as above, except that two-dimensional liquid chromatography was performed before the mass spectrometry analysis. The main difference was in application of strong cation exchange chromatography fractionation. Each sample was separated on the SCX column; fractions were collected and analysed by the same LCMSE platform mentioned previously.

A further nine proteins were identified that demonstrated showed a statistically significant change in protein expression in schizophrenia subjects compared with that in controls (P < 0.05)

In summary, 18 candidate biomarkers have been identified from serum samples that are significantly differentially expressed in schizophrenia patients compared to matched controls (Tables 2 and 3). Prior to these studies CCNR and Psynova had already identified 38 candidate biomarkers that are significantly differentially expressed in a variety of biofluids and using a variety of proteomics platforms. A subset of these markers (n=22) have been selected for the immunoassay development work package to enable their further validation in larger sample sets using the RBM multiplex immunoassay platform).

Task 3.3. Glycoprotein profiling of 20 serum samples from first onset, drug naive schizophrenic patients and 20 matched normal controls

This is 100% complete and a paper published (Stanta et al., 2010). In addition, an additional contribution on identification of altered serum phosphoprotein patterns in schizophrenia has been added at the end of this section. This provides additional insight into effects on post-translational modification in schizophrenia.

Note, that although the collection of CSF was discontinued for the large scale profiling and validation studies, samples which were already in hand from Period 1 were used for the glycoprotein analysis.

Glycoprotein analysis was carried out on serum and CSF samples from schizophrenia patients and healthy controls. Statistically significant glycoprotein changes were found in serum and CSF samples from first-onset schizophrenia patients compared to control subjects. Further studies using this approach should lead to identification of a new class of biomarkers for schizophrenia, based on glycosylation-specific changes.

The current diagnosis of schizophrenia involves examination and monitoring by psychiatrists, a process which can take between 2 and 5 years. In order to increase the efficiency of the diagnostic process and to guide treatment approaches, the identification of molecular biomarkers appears to be essential. Most biomarker approaches for schizophrenia and most other medical conditions have thus far focussed on overall changes in protein levels. However, disease or treatment-associated changes in post-translational modifications could be even more informative, as this could lead to information on functional changes in protein networks. For example, previous studies have described glycosylation modifications in schizophrenia. Glycosylation is one of the most common forms of post-translational modification, with over 60% of proteins being N- or O-glycosylated. Numerous glycosylation modifications have been reported in certain diseases, such as in cancer, congenital disorders of glycosylation and rheumatoid arthritis. In addition, correct glycosylation of proteins is crucial in many biological processes, such as in cellular development, cell signalling, protein folding and cellular recognition. Therefore, the project focussed on identification of changes in the glycosylation pattern of glycoproteins to identify biomarkers to aid in the diagnosis of schizophrenia.

Schizophrenia-related candidate proteins
A list of proteins of interest in schizophrenia research was provided by UCAM. From the list of candidate proteins, those displaying N- or O-linked glycosylation include: transferrin, haptoglobin, hemopexin, ?1 acid glycoprotein, glycoprotein HS2?, vitronectin, apolipoprotein D, vitamin K-dependent protein S, thyroxine binding globulin, factor XIIIB, cathepsin D, lumican, dopamine ?-hydroxylase, complement C4A and pregnancy zone protein. Initial studies on the isolation of these candidates from serum at NIBRT concentrated on haptoglobin and ?1 acid glycoprotein (AGP). Haptoglobin consists of ? and ? chains (1-1, 2-1 and 2-2 isoforms), however, only the ?-chain is glycosylated. There are 4 sites of N-glycosylation on haptoglobin. AGP has a molecular weight of 41-43 kDa and has 5 sites of N-glycosylation.

Glycan analysis
Hydrophilic interaction liquid chromatography (HILIC)-based glyco-analytical technology was employed in the characterisation of N-glycans (3). Briefly, the samples were immobilised in gel blocks and glycans released with peptide N-glycosidase F (Prozyme; Hayward, CA, USA). Subsequently, the glycans were fluorescently labelled with 2-aminobenzamide (2-AB). Phy tips (PhyNexus; San Jose, CA, USA) were used to remove excess 2-AB label. Samples were then analysed by HILIC, using a 150 x 4.6 mm TSK-Gel Amide-80 column (3 µm particle size; Apex Scientific; Kildare, Ireland) on a 2695 Alliance separations module with a Waters 2475 fluorescence detector (Waters; Milford, MA, USA). The peaks in the HILIC profiles were integrated and glycan structures assigned using exoglycosidase digestions and GlycoBase as a guide (http://glycobase.nibrt.ie/glycobase.html). The nomenclature used for glycan structures. The relative percentage areas of the peaks were subjected to statistical analyses.

Glycan analysis of serum and CSF samples from controls and schizophrenic patients
Stanta et al (2010) describes a number of studies, including glycan analysis of serum and CSF samples from controls and first-onset schizophrenics, which were performed in the Dublin-Oxford Glycobiology laboratory at NIBRT. Serum samples were separated into two fractions: high abundant serum protein fraction (HAS) and low abundant serum protein fraction (LAS) using the Proteoprep 20 plasma immunodepletion kit (Sigma). Approx 95% of serum protein is accounted for by the high abundant proteins, thereby potentially masking significant disease-related alterations in the low abundant proteins. The relative percentage areas for each peak in the HPLC profiles were calculated and values were compared between controls and schizophrenia samples.

Glycan analysis of HAP fraction from schizophrenic patients and controls
Significant differences were found between the % areas of peak H6 in the HAP serum fraction from schizophrenia patients compared to controls. For males, this peak was 30% greater in samples from schizophrenia patients compared to controls. Conversely, the same peak was 25% lower in samples from female schizophrenia patients compared to samples from female controls. The predominant glycan in peak H6 was identified as A3F1G3S3.

Glycan analysis of the LAP fraction from schizophrenia patients and controls
Analysis of the LAP fraction revealed two peaks with significantly different % areas between schizophrenia and control samples. Peaks U19 and U23 were larger in samples from male patients compared to controls although no difference was found for females. Exoglycosidase digestions showed that peak U23 contained A4G4LacS4 and peak U19 contained A3F1G3S3.

Glycan analysis of CSF from schizophrenia patients and controls
In CSF samples, four peaks were significantly altered in schizophrenia samples compared to controls. Peak C3 was higher in female patients and lower in male patients compared to the respective controls. The relative % areas of peaks C17, C18 and C20 were lower in all schizophrenia samples than in control samples. The predominant glycans in each peak were: C3 = FA2B, M5, A2G1, A3B; C17 = FA2BG2S2, FA2G2S2, A3F1G3; C18 = A3G3S1, FA4G3; C20 = A4BG4 and A3G3S2.

Higher levels of the tri-antennary tri-sialylated glycan, A3F1G3S3, were found in both LAP and HAP serum fractions from male schizophrenia patients. A3F1G3S3 contains the tetrasaccharide epitope, sialyl Lewis X (sLex), which is a ligand for the selectins (E-selectin, P-selectin and L-selectin) and is involved in leukocyte extravasation, and possibly metastasis, making it an interesting target in cancer and inflammation research. There are reports of inflammation, evident by increased C-reactive protein levels, in schizophrenia. Therefore the increased sLex levels found in this study may be linked to inflammation in schizophrenia. The increased levels of A4G4LacS4 in male schizophrenia patients suggests increased activity of glycosyltransferases such as ?-1,4-galactosyltransferase and a ?-1,3-N-acetylglucosaminyltransferase, which are responsible for the generation of polylactosamines.

In addition, disease-related glycan alterations in 4 HILIC peaks were found in CSF samples from schizophrenia patients. The glycans found in each peak were identified by carrying out exoglycosidase digestions and their implications are discussed in Stanta et al. However, considering the observation that serum is a much more accessible fluid than CSF, the glycan alterations in serum samples may prove more useful as biomarkers in the clinical environment than those from CSF. The glycan alterations identified here provide new information on aspects of schizophrenia pathophysiology and therefore further studies are warranted to determine if these can be used in the development of new diagnostic tests for schizophrenia.

Glycan analysis of HAS fraction
Significant differences were found between the relative percentage areas of peak H6 in HAS samples from first-onset schizophrenia patients when compared to control samples. This peak was 30% higher in samples from male schizophrenia patients when compared to samples from male controls. Conversely, the same peak was 25%. The predominant glycan that was contained in peak H6 was identified as A3F1G3S3.

Glycan analysis of LAS fraction
In the LAS fraction, two peaks had significantly altered relative percentage areas between the control samples and the schizophrenia samples; U19 and U23 were significantly higher in samples from male schizophrenia patients. The U23 peak was found to be two-fold higher in male schizophrenia patients but not significantly different in female schizophrenia patients versus controls, with A4G4LacS4 identified as the corresponding glycan for this peak. In addition, the U19 peak was significantly higher in male schizophrenia patients with an increase of 16% and A3F1G3S3 was identified as the glycan responsible for this peak.

Glycan analysis of CSF
In the CSF samples, four peaks were significantly altered in schizophrenia samples when compared to the age-matched controls. Peak C3 was higher in female schizophrenia patients compared to female control samples and C3 was lower in male schizophrenia patients compared to male control samples. The relative percentage areas of peaks C17, C18 and C20 were all lower in schizophrenia samples than in control samples.

Optimisation of protocols for isolation of candidate proteins
At present, 2DE and affinity chromatography methods are being optimised at the Dublin-Oxford Glycobiology laboratory at NIBRT as a means of isolating the candidate proteins for glycan analysis. The preliminary experiments are discussed below.

Isolation of Haptoglobin
2DE analysis was utilised to separate the haptoglobin ?-chain from other serum proteins. In addition, a haptoglobin standard was analysed by 2DE and shows the ?-chain isoforms at the bottom of the gel and the ?-chain isoforms in the middle of the gel. The presence and location of control serum haptoglobin ?-chain isoforms were determined using Western Blot on a 2DE gel. HILIC fluorescence profiling of the N-glycans liberated from the excised gel spots was also performed.

The proteins in each gel spot were identified by LC-MS/MS. A number of spots contained haptoglobin ?-chain and one spot did not contain any protein, as would be expected because it was a control taken from an unstained region of the gel. Certain spots contained additional proteins.

Isolation of ?1 acid glycoprotein
AGP was not isolated using antibody attached to Protein G Dynabeads. This is evident from the observation that AGP was not present in either the non-denaturing or the denaturing elution fractions. AGP has a molecular weight of 41-43 kDa and protein of this size was found in the flow-through fractions from the serum and the AGP standard. Similar sized protein was additionally present in which AGP standard, which had not been incubated with the Dynabeads, was loaded. The presence of AGP in the two flow-through fractions was confirmed by Western blot analysis.

These results elucidate the existence of schizophrenia-associated glycan alterations in both serum and CSF, with the sex of the patient playing a significant role. Serum is a much more accessible fluid than CSF and for this reason the glycan alterations in serum samples may prove more useful as biomarkers in the clinical environment. The N-glycans that were altered in both serum fractions were complex-type glycans, containing sialylated N-acetyllactosamines. Higher levels of the triantennary trisialylated glycan, A3F1G3S3, were found in both LAS and HAS fractions from male schizophrenia patients. A3F1G3S3 contains the tetrasaccharide epitope, Sialyl Lewis X (SLex), which has been found elevated in cancer and chronic inflammation (Saldova et al 2007, Abd Hamid et al 2008, Tabares et al 2006, Brinkman-van der Linden et al 1998). There are reports of inflammation, evident by increased C-reactive protein levels, in schizophrenia (Dickerson et al 2007) and therefore increased SLex levels may be linked to inflammation in schizophrenia. Additional schizophrenia-specific glycosylation alterations were found in this study. These alterations provide potential targets that can be used to develop diagnostic tools for schizophrenia and provide valuable knowledge on certain aspects of the disease biochemistry. Glycosylation plays important roles in a number of cellular processes and for this reason, changes in glycan structures and composition, as reported in this study for example, may have damaging downstream effects such as, altering the lifespan and concentrations of proteins in the serum, modification of protein folding methods and affecting cellular recognition processes.

The technique employed in the isolation of haptoglobin in this study proved successful and therefore, can be utilised in the isolation of the protein from the first-onset schizophrenia serum samples and control serum samples. Haptoglobin was confirmed by LC-MS/MS and Western Blot in a number of the excised gel spots. However, it is important to note that two proteins, aside from haptoglobin, were located in specific spots (complement 3 precursor) and (apolipoprotein A IV). Apolipoprotein A IV is not a glycoprotein and therefore its isolation, in conjunction with haptoglobin, will not interfere with the glycan analysis of haptoglobin. However, complement 3 precursor is a glycoprotein and thus, care must be taken in future studies to eliminate this spot from the isolation of haptoglobin so as to prevent interference in the results.

The HPLC chromatograms that were reported for specific spots are comparable with previously published results from the NIBRT laboratory. Saldova et al (2007) determined the glycan profiles for haptoglobin in gel spots from serum and observed that as the analysis moved from the more acidic spots to the more basic, there was a shift in the glycan profile. A similar shift was elucidated in this report. Saldova et al (2007) showed that the more acidic spots had higher levels of complex triantennary structures, whereas the gel spots that were more basic had no triantennary structures but had higher levels of high mannose glycans than the more acidic gels spots. The observation that the results of this report are in agreement with this publication supports this isolation technique as being suitable for obtaining glycosylation information on haptoglobin in the serum samples from first-onset schizophrenic patients and age-matched controls.

AGP proved more difficult to isolate than haptoglobin. The gel-based isolation method was not employed initially for the isolation of AGP, because of its low pI range (2.8-3.8). Commercially available IPG strips all have a lower limit of pH 3 and therefore there is a possibility that some AGP may be lost if isolated with 2DE. Isolation using antibody-attached magnetic beads in this study was not successful and showed that AGP did not attach to the chosen antibody. For this reason, alterations in the conditions of the isolation should be tested, such as, incubating the antibody for longer with the beads, using higher antibody concentrations and investigating alternative antibodies.

Depletion of serum proteins with Proteoprep 20 immunodepletion kit
The 20 most abundant proteins were depleted from serum using the Proteoprep 20 immunodepletion kit (Sigma; Poole, UK). Each sample (8 µL) was diluted to 100 µL with equilibration buffer (10 mM sodium phosphate, 150 mM NaCl, pH 7.4) according to the manufacturer’s instructions, centrifuged at 1,000 × g and the flow-through fraction was collected and washed twice with 100 µL equilibration buffer. Flow-through (depleted) fractions were combined, concentrated and buffer-exchanged into 50 mM ammonium bicarbonate using a spin column. The proteins in this fraction constituted the low abundant proteins (LAPs). The high abundant proteins (HAPs) were eluted from the columns with 2 mL 0.1 M glycine-HCl, 0.1 % octyl ?-D-glucopyranoside, pH 2.5. This step was carried out 3 additional times and the harvested fractions were combined. The resulting LAP and HAP fractions were prepared for glycan analysis.


Identification of changes in protein phosphorylation patterns in serum from schizophrenia patients using LC-MSE profiling
Despite the progress made through molecular profiling studies in schizophrenia research, efforts are required to investigate the potentially crucial role of post-translational modifications of proteins such as phosphorylation, in the context of the disease. Protein phosphorylation acts as a switching mechanism in a range of processes including cell signalling and protein transport. Perturbation of the regulation of kinase and phosphatase activity is one of the underlying mechanisms in cancer and other diseases. One example is the hyperphosphorylation of the microtubule-associated protein tau which has been implicated in the pathophysiology of Alzheimer’s disease. Therefore, differential phosphorylation of proteins can provide a biomarker readout to cell function, especially if these can be detected in an easily accessible peripheral body fluid such as blood. Furthermore, the factors regulating protein phosphorylation may be important targets for future therapeutic interventions, and the study of phosphoproteins could increase our understanding of schizophrenia aetiology and the mechanism of action of current and upcoming antipsychotic medications.

A combination of IMAC and LC-MSE was used to compare the total protein and phosphoprotein profiles in serum from first-onset, antipsychotic-naïve schizophrenia patients and matched controls. The main objective was to determine whether specific changes in phosphorylation patterns could be identified. First onset, antipsychotic naive schizophrenia (DSM-IV 295.30; n=20) and matched control (n=20) subjects were recruited and serum samples prepared as described in previous sections.

IMAC fractionation
Serum samples were depleted of 14 high-abundant proteins using the MARS14 depletion system (Agilent, Wokingham, UK) and then fractionated using an IMAC resin (Sigma; Poole, UK) charged with Fe3+ ions. The eluate (containing mostly phosphoproteins) and the flow-through (consisting mainly of non-phosphorylated proteins) fractions were digested separately with trypsin and diluted with a solution containing 50 mM (NH4)2HPO4 and 25 mM EDTA (pH = 9) for enhanced detection of phosphopeptides.

LC-MSE analysis
The samples were analysed by LC-MSE for identification and quantitation of all proteins. Intensities of peptides corresponding to the same identified protein were summed to produce the total protein intensity in each analysed fraction. Phosphorylation of serine, threonine and tyrosine residues was detected based on the experimentally determined loss of a phosphate (?mass = 80) after high-resolution fragmentation during LC-MSE analysis. A significantly changed phosphorylated peptide had to correspond to a protein with at least 3 detected (phosphorylated and non-phosphorylated) peptides.

Identification of abundance biomarker candidates using label-free LC-MSE
In both fractions combined, 312 proteins were detected using LC-MSE by >3 unique peptides and 24 proteins were present at significantly different levels between the patient and control groups.

Identification of phosphorylation biomarker candidates
In the enriched (IMAC-eluted) fraction, 2566 unique phosphorylated peptides were detected, corresponding to 242 unique proteins. Statistical analyses of all detected phosphorylated peptides resulted in identification of 173 peptides with significantly different phosphorylation levels, corresponding to 75 different proteins. Of the 75 proteins showing phosphorylation changes, 65 showed no significant change in abundance. Eighteen of the proteins showed increased phosphorylation, 25 showed decreased phosphorylation and 22 contained peptides with both increased and decreased phosphorylation profiles. Therefore, these 65 proteins had specific changes in phosphorylation with no overall change in protein levels. By contrast, 10 proteins showed the same directional phosphorylation and protein changes, suggesting that the phosphorylation changes were likely to be a result of a corresponding change in protein abundance.

In silico pathway analysis of proteins showing phosphorylation specific changes
Accession codes of altered phosphoproteins were uploaded into the Ingenuity Pathways Knowledge Base (IPKB; www.ingenuity.com) to determine the most significant canonical pathways affected. The most significant pathways identified were acute phase response signalling, with 25 out of the 172 known proteins in this pathways identified (p=1.16E-33) and the complement system, with 16 out of 33 proteins identified (p=3.09E-31). Significant enrichment was also identified of LXR/RXR activation (p=3.00E-26), coagulation system (p=9.39E-15) and intrinsic prothrombin activation (p= 9.68E-12) pathways.

The most novel aspect of this study was the finding that 65 proteins had significantly different phosphorylation patterns in schizophrenia patients compared to controls. An additional 10 serum phosphoproteins were identified which had parallel changes in phosphorylation and abundance, indicating that these were affected predominantly at the protein level. Most of the findings were novel in the case of serum phosphoproteomic studies and, therefore, there is considerable scope for further studies on the functional significance and potential uses of these findings. For example, the finding of phosphorylation specific changes in fibronectin is interesting as previous studies have shown that actively growing cells secrete higher levels of phosphorylated fibronectin compared to corresponding quiescent cells. We also found both increased and decreased phosphorylation of gelsolin which suggests potential changes in blood viscosity. Our findings of decreased phosphorylation of clusterin may be important as previous studies have shown that circulating levels of this protein may be associated with the severity of Alzheimer’s disease.

Considerable attention has been paid recently on the potential role of serum paraoxonase 1 in lipoprotein metabolism, as an inhibitor of oxidative changes and susceptibility to atherogenesis. We also found decreased phosphorylation of serum paraoxonase 1. This is the first report showing that this molecule is phosphorylated, let alone that this is changed in schizophrenia. It will therefore be important to determine what effect phosphorylation has on the function of this protein. Prothrombin is well known for its role in the coagulation/clotting cascade. We showed that phosphorylation of this protein was increased. Functional effects of prothrombin phosphorylation have been suggested by previous studies showing that protein kinase C phosphorylates prothrombin. It will therefore be important to determine whether protein kinase C phosphorylation is also involved in regulation of blood coagulation in schizophrenia. The effects on coagulation were consistent our finding of phosphospecific effects on angiotensinogen, plasminogen antithrombin III and coagulation factor XIII B chain, which are all components of the clotting cascade. Also, previous studies have shown that markers of thrombogenesis are activated in unmedicated patients with acute psychosis (8).

Several lines of evidence suggest that immunological factors contribute to schizophrenia. Increased activity of the innate immune system C1, C3, C4 complement components has been reported by several groups. Also, complement components such as C3, C5 and C9 are known to undergo phosphorylation which affects cleavage or binding activity. We found phosphorylation-specific changes in 16 out of the 33 known components of the complement system, including C1q subcomponent subunit B, C1r subcomponent, C1s subcomponent, complement factor H, C3, C4A, C4B, C4B-binding protein ?-chain, C6, C7, C8 ?-chain, C9, C10 and component C11. Consequently, the complement system was identified as a significant canonical pathway by in silico analyses using the Ingenuity Pathways Knowledgebase. Thus, future investigation on the effects of phosphorylation on activity of this pathway, and other pathways identified here, are warranted.

Task 3.4. Label-free LC-MSE based proteomic profiling of longitudinal samples from first onset schizophrenic patient longitudinal samples at first onset and after 4 weeks of antipsychotic treatment

The aim of this task was to identify molecular biomarkers which could help to guide treatment selection. A comprehensive molecular analysis of serum from first onset schizophrenia patients before and after treatment with olanzapine, risperidone, quetiapine or a mixture of antipsychotics using LC-MSE analysis was carried out.

Subjects were recruited from the Departments of Psychiatry at the Universities of Cologne (centre 1) and Muenster (centre 2), Germany. Schizophrenia was diagnosed based on the Structured Clinical Interview for Diagnostic (SCID) and Statistical Manual (DSM)-IV and all patients fulfilled criteria for the paranoid subtype (classification 295.30). The medical faculty ethical committees of the respective universities approved the protocols of the study. Informed consent was given in writing by all participants and clinical investigations were conducted according to principles in the Declaration of Helsinki.

All patients were antipsychotic naive or had been free of antipsychotic treatment for at least 6 weeks at the start of the study. Patients were assessed by experienced clinicians for psychopathology on the day of sample collection and after the 4 week treatment period using the Positive and Negative Syndrome Scale (PANSS). Total PANNS scores of patients in centre 1 were 86.1 ± 21.4 before and 70.0 ± 28.4 after treatment and those in centre 2 were 71.2 ± 21.0 before and 53.3 ± 15.6 after treatment.

Blood samples were collected from all subjects between 8:00 and 12:00 am before and after the 4 week treatment period into S-Monovette 7.5mL serum tubes (Sarstedt; Numbrecht, Germany). The blood was clotted at room temperature for 2 hours and centrifuged at 4000 x g for 5 minutes to pellet the clots. The resulting supernatants were stored at -80C in Low Binding Eppendorf tubes (Eppendorf; Hamburg, Germany).

Mass spectrometry
Serum samples (40µL) were immuno-depleted using the Multiple Affinity Removal System (Agilent; Santa Clara, CA, USA) on the ÄKTA™ Purifier UPC 10 liquid chromatography system (GE Healthcare; Little Chalfont, Bucks, UK). The flow through, containing the low abundance proteins, was concentrated using spin columns with a 5kDa molecular weight cut-off (Agilent). Proteins in samples were trypsinized and subjected to LC-MSE analysis as described.

Data analysis
Data were processed using the ProteinLynx Global Server (PLGS) version 2.4 (Waters) and Rosetta Elucidator Biosoftware version 3.3 (Seattle, WA, USA) for alignment of raw MS1 data in time and m/z dimensions as described previously. Aligned peaks (features) were extracted and quantitative measurements obtained by integration of the volumes (time, m/z, intensity) of each feature and database searching was carried out using PLGS version 2.4 with the ion accounting algorithm described by Li et al against the Swiss-Prot protein database (version 57; 20,334 entries). The criteria for database searching and protein identifications were as described previously. Protein abundance was calculated by summing peptide intensities for each protein. A two-tailed paired Student’s T-Test was used to identify differential expression, after logarithmic transformation. Significance was set at p<0.05. Fold changes were calculated as the mean intensity after treatment, first onset schizophrenic patients before (T0) and after (T4) treatment, divided by the mean intensity before treatment (T4/T0).

Validation
Matched serum samples were selected and randomized, blinded by a code numbers and analysed using enzyme-linked immunoadsorbent assays (ELISAs) for lumican and Apolipoprotein C2 (both from Uscn Life Science Inc; http://www.uscnk.net/sitemap/elisa.php, UK) according to the manufacturer’s recommendations. Assay plates were read with a Bio-Rad 680 reader (Carlsbad, CA, USA) at 450nm.

Olanzapine study
LC-MSE profiling of serum from first onset schizophrenic patients (n=23) taken before and after 4 weeks of olanzapine treatment resulted in identification of 20 proteins which showed statistically significant changes in concentration of greater than 1.1 fold (T4/T0; P<0.05).

Risperidone, quetiapine and combination therapy study
Patients were treated with either risperidone (n=6), quetiapine (n=9) or a combination of these antipsychotics (n=8). LC-MSE analysis of serum samples from these patients led to identification of 9 proteins which were altered after the 4 week treatment period. Only 2 of the differentially expressed proteins overlapped between the two studies. These proteins were lumican and apolipoprotein C2.

Validation
To validate the two reproducible findings of the LC-MSE profiling study, we carried out ELISA analyses to measure the levels of apolipoprotein C2 and lumican. These assays were performed using whole serum to ensure that the changes were not artefactually induced by the depletion protocol. This showed that significantly higher levels of both proteins were found in sera of patients after treatment, validating the findings of the profiling study.

Discussion
This is the first LC-MSE study that has been carried out to identify biomarkers which track antipsychotic treatments. The main findings were that two proteins, lumican and apolipoprotein C2 were increased by all treatments. Further studies will be required to determine whether assays for these proteins should be incorporated into the multiplex immunoassay panels.

Task 3.5 Glycoprotein profiling of longitudinal samples from first onset schizophrenic patient longitudinal samples at first onset and after 4 weeks of antipsychotic treatment.

Blood samples were collected and serum prepared as described in previous sections from antipsychotic-naive schizophrenia patients (n=23) at first-onset (T0) and after 4 weeks treatment with olanzapine (T4).

Depletion of serum protein with MARS-14 column
Samples were depleted of the 14 most abundant serum proteins using the Agilent MARS-14 column, as described in previous sections. Glycan analysis was carried out as described in Task 3.3.

Isolation of serum proteins by 2D electrophoresis
Glycoproteins in serum samples were isolated using two-dimensional gel electrophoresis (2DE) as described (4). Four high abundant proteins (AGP, haptoglobin ?-chain, ?1 anti-trypsin and transferrin) were isolated from the gel and digested with PNGase F for glycan analysis, as described above. The resulting HILIC peaks were assigned with the same numbers as those in the whole serum profile allowing easy comparisons to be made.

MALDI-TOF MS identification of isolated proteins
Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) analysis was utilised to identify the proteins that were in the excised gel pieces using a 4800 plus MALDI TOF/TOF Analyser (Applied Biosystems, Foster City, CA, USA), as described. The resulting mass fingerprints were analysed by the Global Protein Server Workstation (Version 3.6, Applied Biosystems) for matching MS and MS/MS data against an in silico digested version of the Swiss-Prot database to confirm protein identifications.

Determination of serum AGP concentration
Serum AGP levels were determined using the AGP AssayMax ELISA kit (AssayPro; St. Charles, MO, USA) using the manufacturer’s instructions. Absorbance values at 450 nm were determined on a Perkin Elmer Victor X3 multi-label plate reader.

Statistical analysis. General Linear Model, Repeated Measures ANOVA tests were used to identify significant differences between the patients pre- and post-treatment with olanzapine. When significant interactions were found, post-hoc paired sample t-tests were carried out to identify peaks that were significantly altered after treatment. For testing the effects of olanzapine on glycosylation patterns in whole serum and the LAP fraction, the significance levels of post-hoc tests were corrected for multiple comparisons (?/n; 0.05/23=0.0022) with only p values ? 0.0022 considered to be significantly different. The significance levels of the paired sample t-tests on glycosylation of the isolated high abundant proteins and on the levels of branching, galactosylation and sialylation in whole serum were not corrected for multiple comparisons, as only 6 patients were considered in these tests. Glycan analysis was carried out on whole serum samples and on LAP fraction proteins from schizophrenia patients before and after 4 weeks treatment with olanzapine.

Effects of olanzapine treatment on whole serum glycosylation
Olanzapine treatment altered the glycosylation profile of whole serum proteins from schizophrenia patients. Peak 16 was increased (p = 0.0018) and peak 20 decreased (p = 0.0016) after treatment. Glycans A2G2S2 was identified in peak 16 and A3G3S2, A2F1G2S2 and A3BG3S2 were present in peak 20.

Effects of olanzapine treatment on glycosylation of low abundant serum proteins
Glycan analysis was carried out on the low abundant serum protein (LAP) fraction from schizophrenia patients. Fifteen peaks were assigned to the HILIC profiles and exoglycosidase digestions carried out to identify the glycans in each peak. Statistical analysis was performed on the relative percentage areas of the peaks and no differences were found after treatment with olanzapine. Peak 9 contained the most abundant glycan (A2G2S2) which was the equivalent of peak 16 in the whole serum glycan profile.

Isolation and glycan analysis of high abundant serum proteins
2DE was carried out in order to isolate AGP, haptoglobin ?-chain, transferrin and ?1 anti-trypsin from serum samples before and after olanzapine treatment. MALDI-TOF MS analyses confirmed the identity of all four proteins. Glycan peak 16 from AGP was significantly increased from 22.4 ± 1.2 % to 28.1 ± 1.2 % (p = 0.027) after olanzapine treatment. Peak 16 contained A2G2S2 and peak 24 (which showed a non-significant decrease) contained tetra-antennary structures, including A4G4S4, A4G4S3, A4F1G4S2, A4F1G4S3, A4F1G4S4 and A4F2G4S4. Olanzapine treatment did not significantly alter glycosylation of haptoglobin, ?1 anti-trypsin and transferrin, although these proteins showed a trend of increased peak 16 and decreased peak 20 as found in the analysis of whole serum proteins above.

Determination of AGP concentration
An ELISA kit was used to examine the concentrations of AGP in the serum samples from schizophrenia patients before and after olanzapine treatment. This showed that the olanzapine treatment did not have a significant effect on the serum concentration of the AGP protein (pre-treatment = 0.50 ± 0.02 mg/mL, post-treatment = 0.44 ± 0.03 mg/mL).

Discussion
The olanzapine treatment led to significant changes in the glycosylation profile of whole serum proteins but not for the low abundant serum protein fraction. The bi-antennary, di-sialylated glycan, A2G2S2, was found to be increased post-treatment with olanzapine and a number of other di-sialylated glycans, A3G3S2, A2F1G2S2, A3BG3S2, were found to be decreased after treatment in whole serum samples. Possible sources of the olanzapine-induced glycan modifications include alterations to the proteins that carry these glycans, altered inflammatory processes and changes in the activity of specific glycosyltransferases. Further experiments in the current study identified AGP as a carrier of some of the glycans that were altered in whole serum. A2G2S2 was increased and the levels of some tetra-antennary glycans were reduced on AGP post-treatment. Also, haptoglobin ?-chain, transferrin and ?1 anti-trypsin showed increased A2G2S2 and decreased A3G3S2, A2F1G2S2 and A3BG3S2, although these effects were not statistically significant. The finding that the concentration of the AGP protein was not altered by olanzapine treatment suggested that the effects on the glycans were not a result of altered protein levels.

AGP is a 41-43 kDa glycoprotein that has 5 sites of N-glycosylation and glycosylation accounts for 40 % of the protein molecular weight. AGP has been shown, along with albumin, to be a primary carrier of olanzapine in the blood. Previous reports have shown that the removal of glycans from AGP did not affect the binding ability of drugs and therefore it is unlikely that the binding of olanzapine to AGP is responsible for the glycan alterations shown here. The finding that olanzapine altered galactosylation and sialylation of serum glycans and did not affect the levels of branching of glycans, suggests an altered activity or expression of galactosyltransferases and sialyltransferases. This is consistent with the finding of previous studies which showed that treatment with olanzapine increased the expression of the B4GALT1 galactosyltransferase in liver of schizophrenia patients. Therefore, the glycan changes identified here provide important new information on the mechanism of action of olanzapine treatment of schizophrenia patients. Further studies are warranted to test the possibility these and other glycan-related changes can be used for prediction or monitoring of drug responses. This is an objective of the next phase of this project.

Task 3.6 - HuMAP multianalyte profiling of 20 serum samples from first onset, drug naive schizophrenic patients and 20 matched normal controls

The Multi-Analyte Profiling (HumanMAP®) platform was used to measure 147 analytes in serum collected from 445 first or recent onset schizophrenia, bipolar disorder (BD), major depressive disorder (MDD) and control subjects, recruited from three psychiatric centres. Biomarkers associated with schizophrenia were identified by testing samples from 71 first onset antipsychotic-naïve schizophrenia and 59 control subjects (CIMH). The resulting biomarkers were validated by comparative testing using samples from 46 first onset antipsychotic-naïve schizophrenia patients and 46 controls (Muenster) and from 46 recent onset minimally-treated schizophrenia patients and 45 controls (Magdeburg). Serum was also tested from 32 BD patients (CIMH) and from 50 MDD patients (Muenster) for preliminary determination of the disease specificity of the biomarker panel.

A set of 45 differentially expressed analytes was initially identified which distinguished first onset antipsychotic-naïve schizophrenia patients from control subjects (centre 1). Correlation and in silico pathway analyses showed that some of these biomarkers were associated with growth factor, hormonal and acute phase response pathways. Application of the complete set of biomarkers across clinical centres 1, 2 and 3 yielded an average sensitivity of 0.90, specificity of 0.83 and a receiver operating characteristic (ROC) area under curve (AUC) of 0.87. Analysis of covariance showed that 36 of the 45 markers remained significant changed in schizophrenia vs. control after adjustment for recorded baseline characteristics including age, gender, body mass index, smoking and cannabis consumption.

Specificity of Schizophrenia Diagnostic Biomarker Panel by Comparison with other Neuropsychiatric Disorders
One factor which renders diagnosis of schizophrenia difficult is the overlap of symptoms with other neurological and neuropsychiatric disorders. Therefore, the 36 differentially expressed analytes were tested as a combined diagnostic panel using serum samples from clinical centre 1, comprised of schizophrenia (n=71), BD (n=32) and control (n=59) subjects, and clinical centre 2, comprised of schizophrenia (n=46), MDD (n=50) and control (n=96) subjects. BD subjects were euthymic and such patients can experience disruptions in cognitive behaviours similar to those observed in schizophrenia. MDD subjects were acutely ill and were chosen due to the observed overlap of negative symptoms between depression and schizophrenia. The analysis showed that schizophrenia was identified with an average sensitivity of 0.89, specificity of 0.86 and ROC-AUC of 0.87 compared to BD and MDD. It should be noted that samples from BD and MDD subjects were compared to those from schizophrenia patients and controls from the corresponding clinical centres to minimize any potential confounding factors due to geographical or methodological differences.

Data analysis followed a step-wise procedure starting with determination of molecular differences between schizophrenia patients and controls in clinical centre 1. Differences of group means were assessed by unpaired two-tailed t-tests and p < 0.05 was considered significant. As comparison, non-parametric Wilcoxon rank-sum tests were performed on all analytes to account for potential deviations from normal distributions. Although adjustment for multiple hypothesis testing is controversial (1), the raw and adjusted values (2) are indicated as a guide for estimating the validity of results and for prioritizing validation experiments. The difference between groups was also investigated using analysis of covariance (ANCOVA) models including age, sex, body mass index (BMI), smoking and cannabis consumption as covariates. Interactions between diagnosis and the aforementioned covariates were tested. An average 14% of analyte values were missing from each clinical centre due to low analyte concentrations and were assigned the value 0.0.

Analytes identified as differentially expressed in serum samples from clinical centre 1 were then analysed in centres 2 and 3 using the HumanMAP® platform and the same statistical assessment. The boosted classification tree algorithm was applied to determine sensitivity, specificity and the Receiver Operator Characteristic (ROC) Area Under the Curve (AUC) for classifying the degree of separation between schizophrenia patients and controls, and between schizophrenia patients and subjects with non-schizophrenia psychiatric illnesses. Confidence intervals were calculated by the Bootstrap Case Cross-Validation (BCCV) procedure and adjusted according to the Bias Corrected accelerated (BCa) method due to coverage limitations. For every estimation, 1000 bootstrap samples were used which has been shown to be sufficient for determination of 95% confidence intervals. The leave-one-out cross validation values of the original data is also given as point estimates.

Sample sizes were based on availability and those of the schizophrenia patients and controls from clinical centre 1 were sufficient to detect an effect size >49.7 % with a power of 0.8 using a two tailed significance level of p < 0.05. The respective effect sizes for sample cohorts in centres 2 and 3 were 59.0 % and 59.4%, respectively. Analyses were performed using the statistical software package R (http://www.r-project.org).

Task 3.7. Multiplex immunoassay profiling of schizophrenia patients before and after treatment with antipsychotics.

In addition further profiling studies were added to determine if predictive markers for metabolic side effects and relapse could be identified (in kind contributions). The importance and potential impact of these additional studies to future developments in the treatment of schizophrenia are clear.

The aim of this task was to identify molecular biomarkers which could help to guide treatment selection. A comprehensive molecular analysis of serum from first onset schizophrenia patients before and after treatment with olanzapine, risperidone, quetiapine or a mixture of antipsychotics using LC-MSE analysis was carried out.

Subjects were recruited from the Departments of Psychiatry at the Universities of Cologne (centre 1) and Muenster (centre 2), Germany. Schizophrenia was diagnosed based on the Structured Clinical Interview for Diagnostic (SCID) and Statistical Manual (DSM)-IV and all patients fulfilled criteria for the paranoid subtype (classification 295.30). The medical faculty ethical committees of the respective universities approved the protocols of the study. Informed consent was given in writing by all participants and clinical investigations were conducted according to principles in the Declaration of Helsinki.

All patients were antipsychotic naive or had been free of antipsychotic treatment for at least 6 weeks at the start of the study. Patients were assessed by experienced clinicians for psychopathology on the day of sample collection and after the 4 week treatment period using the Positive and Negative Syndrome Scale (PANSS). Total PANNS scores of patients in centre 1 were 86.1 ± 21.4 before and 70.0 ± 28.4 after treatment and those in centre 2 were 71.2 ± 21.0 before and 53.3 ± 15.6 after treatment. Blood samples were collected from all subjects as described in previous sections.

Multiplex immunoassay
All samples were analysed using the HumanMAP® multiplex immunoassay platform as described previously The protocol for the study participants, clinical samples and test methods was carried out in compliance with the Standards for Reporting of Diagnostic Accuracy (STARD) initiative.

Multiplex immunoassay profiling of serum from schizophrenia patients before and after 4 weeks of treatment with olanzapine (n=23), risperidone (n=6), quetiapine (n=9) or a combination of riseridone and quetiapine (n=8) identified 7 molecules that were changed in 2 or more of the treatment groups. One of these molecules was the hormone prolactin which was significantly increased in 3 of the 4 groups. The molecules connective tissue growth factor (CTGF), interleukin 17 (IL17), matrix metalloproteinase 3 (MMP3) and human chemokine (C-C) motif ligand 16 (HCC-4) were changed in the olanzapine and combination treatment groups and tumour necrosis factor receptor II (TNFRII) and creatine kinase-MB (CK-MB) were altered by the olanzapine and quetiapine treatments.

The change in prolactin most likely reflects the dopamine D2 receptor antagonist effects of many antipsychotics and is consistent with numerous previous studies which have shown that prolactin can be used as a marker for treatment with second generation antipsychotics. The changes in the other molecules have not been reported previously but reflect the known effects of antipsychotics on inflammatory pathways.

Task 3.8 - Bioinformatic analysis and identification of up to 50 candidate biomarkers for further validation

Many candidate biomarkers have been identified to be considered for subsequent validation. Those identified using HuMAP technology are already available in high throughput immunoassay format. Candidates identified by mass spectrometry approaches have been selected for the first wave of immunoassay development were prioritised based on the following criteria: statistical significance, reproducibility of the measured protein expression change across all samples, and the availability of suitable antibodies to enable assay development. Based on this analysis, 22 candidate biomarkers were sent to RBM for multiplexed immunoassay development.

A further 7 candidate biomarkers where no suitable commercial antibodies are available and bioinformatic analysis of various profiling studies at CCNR suggested had potential as a disease specific biomarker were sent to Storkbio for antibody generation and subsequent assay development at EDI – see task 4.1.

Work package 4 - Assay Development and biomarker validation

Task 4.1. Raise Antibodies against candidate biomarkers to enable immunoassay development as required.

In addition 9 immunoassays have been developed as extra contributions to the project

For task 4.1, 7 protein targets as biomarkers were selected in WP3 by workgroups of Cambridge University, Psynova Neurotech and The National Institute for Bioprocessing Research and Training. These proteins were:

Q6XYQ8 SYT10_HUMAN Synaptotagmin-10
Q86V15|CASZ1_HUMAN Zinc finger protein castor homolog
Q86W28|NALP8_HUMAN NACHT, LRR and PYD domains-containing protein 8
P29120 NEC1_HUMAN Neuroendocrine convertase 1(PC1)
P16519 NEC2_HUMAN Neuroendocrine convertase 2(PC2)
P09172 DOPO_HUMAN Dopamine beta-hydroxylase
P48147 (PPCE_HUMAN) Prolyl endopeptidase

To develop novel antibodies, Storkbio’s team selected 2 peptides sequences as antigens for each candidate biomarker, totally 14 peptides, each for one polyclonal antibody project. Selection was made using bioinformatical analysis according to antigenicity, surface probability, hydrophilicity and flexibility indexes. From each protein sequence two peptides with best characteristics (maximal index values) were chosen, including other information accessible about the protein to develop antibodies, which possibly recognize native protein targets.

Peptides were synthesized using Fmoc SPPS chemistry and finalized with conjugation to Keyhole Limpet Hemocyanin (KLH), which helps to raise peptide’s immunogenicity. One polyclonal antibody project included 2 female rabbits. Peptide-KLH conjugate was injected 4 times during 2 month period together with Freund’s Adjuvant to stimulate immune response. After third immunization, test-bleeds were taken and enzyme-linked immunosorbent assay (ELISA) was performed to control, if immune response arose and peptide-specific antibodies have developed. All 28 test-serums showed significant signal compared to zero-serums and production was continued. Total final bleeding was made after fourth immunization and antiserums tested on ELISA, each serum against the specific peptide, which was used as antigen. All serums had good to very good antibody titres, concluding that immunization process was successful.

Antiserum of rabbit with better titre was selected from each project for epitope-specific antibody purification. For this, in each project the specific antigenic peptide was coupled to sepharose matrix column and peptide-specific antibodies were collected from antiserum. This step eliminated other antibodies and non-specific agents, to get a pure sample of antibodies to decrease possible non-specific reactions in later applications. Pure antibody samples were tested again on ELISA against specific peptide. Finally, 14 novel anti-peptide polyclonal antibody samples against 7 different candidate biomarkers were prepared for further assay development and biomarker validation.

Incorporation of new assays into the DiscoveryMAP platform using existing reagents
In cases where reagents were commercially available, new assays were constructed for incorporation into the DiscoveryMAP as indicated in. These assays were used in the final validation screening of serum from 748 subjects as described in WP5.

Development of immunoassays which required novel antibody production
In addition to achieving the main objective with the constructed assays described above, we have also attempted to develop additional new assays in cases where antibodies of sufficient quality were not available. To develop novel antibodies, 2-6 peptide sequences were selected as antigens for each candidate biomarker totalling 53 peptides (1-4 peptides per target) (not shown). Multiple peptides were used in cases of larger protein targets to increase the choice of epitopes or if specific sequences of proteins were of interest. Selections were made using bioinformatic analyses according to antigenicity (Jameson-Wolf), surface probability (Emini plot), hydrophilicity (Kyte-Doolittle), flexibility (Karplus-Schulz) and alpha region (Garnier-Robson) indices. In addition, EMBOSS antigenic analysis (Kolaskar and Tongaonkar) was used. From each protein sequence peptides with the best characteristics (maximal index values) where chosen for antibody development, to maximize chances of binding to the native proteins.

Peptide synthesis and immunizations for novel antibody production
Peptides were synthesized using Fmoc SPPS chemistry and conjugated to Keyhole Limpet Hemocyanin (KLH), to increase immunogenicity. Each polyclonal antibody project used 2 female rabbits for immunizations. Peptide-KLH conjugates were injected 4 times over 2 months with Freund’s Adjuvant to stimulate immune responses. After the third immunization, test-bleeds were taken and ELISAs performed to determine if anti-peptide antibodies had developed. All 44 test-sera showed signals compared to the results seen with pre-immune sera and production was continued. Final bleeds were taken after the fourth immunization and all antisera tested positive again. However, antibody titres were lower than expected, but this was thought to be due to use of more than one peptide per immunization (this option was chosen to preserve resources and animals).

For each project, antisera with the highest titres were selected for epitope-specific antibody purification. For this, antigenic peptides were coupled to 6% agarose, which was activated with iodoacetyl groups to bind sulfhydryl groups of cysteines in the peptides. The columns were incubated with the antisera and antibodies binding to the peptide-agarose were eluted and collected. This step eliminated the presence of other antibodies and non-specific agents, which decreased the chances of possible non-specific reactions in later applications. Finally, 22 novel anti-peptide polyclonal antibody samples against 14 different candidate biomarkers were prepared for further assay development.

Development of novel assays beyond achieving the deliverables
Thirteen bead-based two-site immunoassays have already been incorporated into the Myriad-RBM DiscoveryMAP immunoassay platform which has been used for completion of the large scale validation study comprising 748 subjects. In addition we have continued to produce additional assays to support the potential exciting follow up work arising from this highly productive project. Work has continued on 10 new assays for incorporation into future multiplexes. This includes four additional targets (ceruloplasmin, dopamine ?-hydroxylase, leucine rich ?2 glycoprotein and secretagogin) which were not described in the methods section. All four of these molecules have been found to be altered in in-house molecular profiling studies of schizophrenia or in antipsychotic treatment experiments.

The key objectives of developing novel assays for validation screening of approximately 800 subjects has been achieved twice. Firstly, novel assays were incorporated into the DiscoveryMAP multiplex immunoassay platform and testing with panel this showed high performance for distinguishing schizophrenia from control subjects (n=748; see WP5). Secondly, a novel 51-plex assay system was produced and used to screen samples from 806 subjects, and again this resulted in high performance. In addition to this major achievement, the project also developed further two-site bead based assays for incorporation into multiplexes which are relevant to schizophrenia and/or antipsychotic treatment. It is anticipated that these will be applied to support follow up phases of this project.

Task 4.2. Develop sandwich immunoassays for newly defined biomarkers and multiplex where appropriate.

Commercial antibodies were purchased for each analyte and both conjugated to beads and biotinylated. They were screened by Luminex in a checkerboard assay with a purified transthyretinin standard and plasma samples titrated out, as expected levels are in the µg/ml range. Upon completion of the primary screens for all 22 biomarkers, pairs were put into multiplexes based on dilution factor and lack of cross-reactivity and interference.

Each multiplex was validated on the RBM automated platform for precision, spike-recovery, linearity, matrix interference and stability. Where possible, assays were correlated to ELISA and reference samples.

Task 4.3. Carry out biomarker validation screen of 750 case and control samples with the newly developed assays to select best candidate biomarkers for final diagnostic panels.

For testing the performance of the antibodies generated by Storkbio on this platform, each antigenic peptide was coupled to microspheres as BSA-conjugate. The purified antibodies as well as the preimmune, preliminary and final sera were tested for performance and analysed using the FlexMap 3D system. Similar to the ELISA results all purified antibodies were also able to recognize the peptide with good to very good signal intensity.

Cross-reactivity was tested by incubation of each antibody individually in combination with microspheres carrying all the different peptides used for immunization. Results indicate that several antibodies show low or very low cross-reactivity, but some exhibited considerable cross-reactivity.

For development of sandwich immunoassays, biotinylation of the antibodies is required. This was done with an aliquot of all purified antibodies using different excess levels of biotin. After removal of uncoupled biotin by size exclusion chromatography, a biotinylation control was performed showing that all antibodies were successfully biotinylated.

Testing of the antibodies for neuroendocrine convertase 1 (PC1) and prolyl oligopeptidase (PPCE) in a sandwich immunoassay was performed using standard proteins PC1 and PPCE purchased from commercial suppliers. At first, these proteins were used to test if the antibodies generated against peptides sequences of the proteins were also able to recognize the native proteins. Therefore the proteins were coupled to Luminex beads. These beads were then incubated with the different biotinylated anti-PC1 and anti-PPCE antibodies. Results showed that both PC1 polyclonal antibodies were able to bind to the native protein, while for PPCE only the biotinylated PPCE I538-G556 antibody showed signals, suggesting that antibody PPCE I316-L338 is able to bind to the synthetic peptide used for immunization but does not recognize the native PPCE amino acid sequence.

The development of sandwich immunoassays was started with PC1 and PPCE where standard proteins were already available. For assay development both antibodies (not biotinylated) for each protein were coupled to different beads. The second antibody was used in its biotinylated form as detection antibody to identify the best capture and detection pair.

As expected from the results described above, it was not possible to develop a sandwich immunoassay for PPCE, most probably due to antibody PPCE I316-L338 which does not recognize native PPCE protein. To solve this problem, commercially available antibodies against human PPCE will be tested in combination with antibody PPCE I538-G556.

For PC1 the different combinations of biotinylated detection antibody and capture antibody were tested. In contrast to PPCE, it was possible to develop a sandwich immunoassay for PC1 on the Luminex platform, though sensitivity was rather low.

For the competitive catecholamine assay, a proof of principle assay was performed. The company LDN (Labor Diagnostika Nord GmbH & Co. KG) produces ELISA/RIA for the detection of catecholamines. These enzyme immunoassays for the quantitative determination of Adrenaline, Noradrenaline and Dopamine are competitive ELISAs, where derivatized samples and solid phase bound analytes compete for a fixed number of antiserum binding sites. For the derivatisation of the Catecholamines, LDN uses the “one well” sample pre-treatment concept, where the Catecholamines are first extracted by using a cis-diol-specific affinity gel. Afterwards they are biotinylated and then enzymatically O-methylated by the enzyme catechol-O-methyltransferase. LDN provided EDI with material to transfer the assay onto the bead based Luminex system.

To prove the possible transfer of the competitive ELISA to the bead-based technology, we started to establish bead-based assays using immobilized Ovalbumin-Dopamine conjugates or avidin-beads. Using the beads with immobilized Ovalbumin-Dopamine conjugates, a competitive assay could be developed, according to the microtitre plate ELISA of LDN.

For the development of this competitive assay for Dopamine, three different Ovalbumin-Dopamine conjugates (no.1, 2, 3) were immobilized on micropheres. Free Ovalbumin-Dopamine conjugate no. 3 was added. The immobilized and free conjugates compete for the antiserum binding sites.

Short protocol for the competitive assay for Dopamine
The Ovalbumin-Dopamine conjugates no. 1, 2 and 3 were covalently coupled to three different Carboxy-bead sets. Then, the Ovalbumin-Dopamine-coated beads were incubated with different concentrations of free Ovalbumin-Dopamine conjugate no. 3 and an antibody directed against the biotinylated and O-methylated form of Dopamine (antibody no. 14). Thereby the antibody was tested in two different concentrations. After a washing step, a PE-conjugated anti-rabbit-IgG was used for detection. Following a second washing step, the assay was read out using a Luminex 100.

At both antibody dilutions, the signal decreases by adding more of the free Ovalbumin-Dopamine conjugate no. 3. Figure 4 reveals, that the antibody binds the immobilized Ovalbumin-Dopamine conjugate no. 1 and that the free Ovalbumin-Dopamine conjugate no. 3 competes for the binding sites of the antibody, leading to a decreased signal at higher concentrations of Ovalbumin-Dopamine conjugate no. 3.

A protocol for the bead based competitive assay was established. In addition, at EDI, general issues of the bead based immunoassays were addressed, including optimization of blocking and detection steps.

Task 4.4. Develop a schizophrenia multianalyte diagnostic panel to 5-10 biomarkers suitable for high throughput screening of clinical samples.

The TruCuture system was incorporated for use with a 33-plex immunoassay platform as a non-funded activity. This task evolved into two studies.

In the first study, the ability of the 36 biomarkers identified in the discovery phase to accurately discriminate two further independent cohorts of schizophrenia patients (including 87 first onset schizophrenia patients and 80 matched controls) and distinguish schizophrenia from other CNS disorders, including 32 euthymic BD patients, 33 dementia patients and 50 MS patients (CIMH) and 50 acutely ill MDD patients alongside the appropriate matched controls from these centres was assessed. Excellent sensitivity and specificity was observed.

In the second study, 22 of the most consistent biomarkers identified in the HuMAP platform across the 5 clinical centres now tested were identified. These 22 biomarkers, plus a further 29 analytes have been recombined into customised multiplexes (51 analytes) for high throughput validation studies. This new custom multiplex combination was initially validated on 838 subjects (comprised of 593 schizophrenia and 245 control subjects) and used to generate a classification decision rule was constructed using support vector machine methodology. This decision rule yielded a good separation between patients and controls with a cross-validation accuracy of 83% (sensitivity 83%, specificity 82%).

A novel ex vivo blood culture system (TruCulture) combined with the use of a 33-plex cytokine/chemokine immunoassay panel was implemented to identify further schizophrenia biomarkers and to provide a clinically-relevant system for identification of drug-predictive or response biomarkers and for novel drug discovery.

Validation study 1: Does the biomarker signature identified in the discovery phase of the project allow similar classification of schizophrenia patients vs. controls in further independently collected clinical cohorts?

Sample collection for this task in Mannheim and Muenster was delayed due to EC funding delays and the centres being legally obliged to have funds in place before recruiting individuals to carry out tasks. In the meantime, further validation of the 36 biomarkers identified using HuMAP methods has been carried out on two clinical cohorts collected outside the scope of the SchizDX project, from psychiatric centres in Rotterdam and Magdeburg.

The major advantage of whole-blood culture assays is that all components of the blood, including cellular and extracellular components, are present and can therefore reflect the function of the whole immune repertoire compared to purified peripheral blood cell preparations. Furthermore, cells in whole blood cultures have not been deprived of their normal environmental constituents, no xenogenic constituents such as serum supplements have been added and they have not been subjected to the usual stress of preparation by several stages of centrifugation. Finally, leukocytes in these whole-blood cultures do not become plastic-adherent, therefore eliminating this artificial type of background activation.

Validation cohorts

Two independent cohorts of largely drug-naïve, first onset schizophrenia patients and matched controls were obtained from collaborators outside the SchizDX project. Schizophrenia was diagnosed based on the Structured Clinical Interview for Diagnostic and Statistical Manual (DSM)-IV. Patients fulfilled the criteria of the paranoid subtype (DSM-IV 295.30). All diagnoses and clinical tests were performed by psychiatrists following Good Clinical Practice guidelines. Patients whose clinical diagnosis required revision at a later stage were excluded from the study. Control subjects used in this study were matched to the schizophrenia patients for age, gender and social demographics and were recruited from the same economic and geographical area of the university districts. Controls with a family history of mental disease or with other medical conditions such as type II diabetes, hypertension, cardiovascular or autoimmune diseases were excluded from the study.

The cohort consists of schizophrenia patients and matched controls. Schizophrenia was diagnosed based on the Structured Clinical Interview for Diagnostic and Statistical Manual (DSM)-IV and fulfilled criteria for the paranoid subtype (classification 295.30).

HuMAP profiling of these new samples was carried out as before. Discrimination of the 36 biomarker schizophrenia panel against healthy controls was calculated using the Adaboost algorithm and the Bootstrap Case Cross-validation procedure. The algorithm was trained and tested on each dataset using this bootstrapping procedure and generated good levels of sensitivity and specificity. Therefore; the satisfying conclusion is that we have now demonstrated that the 36 marker panel can discriminate schizophrenia patients from controls in five independent clinical cohorts.

Disease specificity of the 36 candidate biomarker was also assessed against other CNS disorders, including included 32 euthymic BD patients, 33 dementia patients and 50 MS patients (CIMH) and 50 acutely ill MDD patients alongside the appropriate matched controls from these centres. The algorithm showed almost perfect accuracy vs dementia and MS cohorts (data not shown) and excellent sensitivity and specificity against the major depression and bipolar disorder cohorts.

Biomarker selection for custom multiplex
The initial identification and validation of 36 candidate biomarkers was carried out using RBM’s HuMAP collection of >140 immunoassays. Therefore, most of the assays tested are not providing data useful for the classification of patients. As a first step to developing the candidate biomarker panel into a tool capable of efficiently testing large number of samples, we recombined the selected assays into new custom multiplexes. Rather than simply build on the 36 biomarkers originally identified in clinical centre 1, we were able to use the HuMAP profiling data from all 5 clinical centres to select the most consistent biomarkers for the custom panel. This resulted in the identification of 22 markers which were altered in the schizophrenia population in three or more of the five clinical centres. Technical reproducibility of the various immunoassays was assessed by repeating the HuMAP measurements on fresh aliquots of a subset (n=63) of the same samples originally tested 3 months after the initial testing. This showed an average correlation of 0.83, an average measurement shift of 29% and an average log distance to the LLD of 1.29. In contrast, analytes that were not selected featured a correlation of 0.65, a measurement shift of 54% and a log distance to the LLD of 0.70.

Nine additional biomarkers were incorporated into the resulting 51-plex due to their known association with schizophrenia or due to the fact that we identified significant changes in these analytes in studies of schizophrenia patients using orthogonal platforms. In addition, we incorporated a further 20 significant analytes identified from a separate study to identify bipolar disorder biomarkers. The thinking behind this strategy is that ultimately, we want to develop one or more tests that can aid the differential diagnosis of psychiatry patients presenting early in their disease progression, when distinguishing schizophrenia and bipolar patients is most challenging.

Re-multiplexing selected assays
This procedure was guided by optimum dilution of serum and mixing of antibodies to give the most sensitive assays. The required dilutions of serum were 1:5, 1:50, 1:200, 1:10,000, and 1:200,000. The 1:5 dilution group consisted of 31 analytes which were divided into 4 multiplexes. For each higher dilution, only one multiplex was used, yielding a total of 8 new multiplexes for the 51 analytes. Once the multiplexes were created, large batches of reagents were manufactured to allow consistent testing of approximately 7000 samples. The reagents were validated using the following parameters: sensitivity, linearity, spike recovery, common serum matrix interferences, cross-reactivity, precision, correlation, freeze-thaw stability and short-term room temperature antigen stability.

Classification Decision Rule - development and performance
To discriminate schizophrenia patients from controls using the markers selected in phase I of this study, we implemented a linear support vector machine (SVM) algorithm. New multiplexed immunoassays were developed incorporating the 51 analytes. These were used to analyze a cohort of 838 subjects comprised of 593 schizophrenia and 245 control subjects. For technical validation, the set also contained 227 samples which had been used previously during the marker selection phase of the study.

A classification decision rule was constructed using SMV (SMV-B).This decision rule yielded a good separation between patients and controls with a cross-validation accuracy of 83% (sensitivity 83%, specificity 82%, ROC-AUC 88%). We also determined the classification accuracy of the SVM-B decision rule for four regions of conditional probabilities. This resulted in an increase in accuracy of up to 93% for patients and controls in the highest probability regions. It is important to note that only 14% of the total samples were excluded when Region III was designated as indeterminate.

TruCulture tubes were manufactured at EDI GmbH. These are essentially blood collection tubes that can be used by anyone trained in drawing blood. Once blood is drawn into the tube and mixed by gentle inversion, the plunger handle is removed and the tube is inverted and placed in a dry heat block at 37oC for up to 48 hours. After incubation, the valve separator component is inserted in the tube to separate cells from the culture supernatant. The tubes are then recapped, and stored at -20oC until needed for downstream analysis.

Testing the effects of antidepressants and antipsychotics on functions of whole blood cells
We tested the capability of the antidepressant drug imipramine and the antipsychotic drug olanzapine to modulate the immune system of stimulated cells in TruCuture tubes. For optimal dose determination, these drugs were tested at 5 different concentrations in a whole blood stimulation approach. The secretion of IL-1beta was used as read out and a stimulation index was performed. In future studies, we can also test functional effects using as readouts analysis with a new 33-plex cytokine/chemokine immunoassay panel.

The TruCulture procedure was incorporated as an extra non-funded activity to follow up future phases of this project which will target more the drug discovery aspects of schizophrenia. Furthermore, this can be combined with use of the 33-plex cytokine/chemokine panel as a readout for the effect of various psychiatric medications in clinical studies. By application of this model it is hoped to provide this as a novel preclinical screening tool for the characterization of existing psychiatric medications and the identification of potential novel treatment strategies.

Work Package No. 5 - Carry out clinical proof of concept screen on 800 psychiatric in patient samples.

Task 5.1. Carry out validation of 800 samples using the novel schizophrenia biomarker panels

This section describes the ground-breaking development of a serum-based test to help confirm the diagnosis of schizophrenia. A multiplex panel of 51 immunoassays was developed that allowed reproducible identification of schizophrenia patients compared to controls with high sensitivity and specificity. Validation of this test consisted of developing a linear support vector machine (SVM) decision rule and testing its performance using cross-validation. The resulting decision rule delivered a sensitive and specific prediction for presence of schizophrenia in subjects compared to matched controls (806 subjects), with a receiver operating characteristic-area under the curve of 88%.

The subjects were recruited from the Departments of Psychiatry at the Universities of Cologne (cohort 1), Münster (cohort 2), Magdeburg (cohorts 3 and 4), Rotterdam (cohort 5) and the USA military (n=110 Bipolar Disorder patients and n=110 healthy controls). Cohorts used for the molecular assay selection phase were comprised of 250 first- and recent-onset schizophrenia patients and 230 healthy control subjects. Schizophrenia patients of cohort 1 (n=71), 2 (n=46), 4 (n=47) and 5 (n=40) were first onset and antipsychotic-naïve, and 32 out of 46 subjects from cohort 3 had not been treated with antipsychotic medication for more than 6 weeks prior to sample collection. We also obtained samples from the USA military sample bank. These were obtained from subjects within 30 days before their first contact with USA military psychiatric services and who later received a confirmed diagnosis of bipolar disorder (BD) (n=110). The cohort used to validate and implement the decision rule was comprised of samples from a mixture of first onset and chronic antipsychotic-treated schizophrenia (n=577) and healthy control (n=229) subjects recruited at the Universities of Cologne, Münster and Magdeburg. Schizophrenia was diagnosed, control samples matched and serum prepared, as described in previous sections.

Multiplex immunoassay profiling was carried out as described in previous sections. Biomarker assays were selected from the DiscoveryMAP platform to reduce this to a smaller set of assays streamlined for classification. The molecules were ranked based on the number of cohorts in which significant differences were observed using unpaired, two-tailed t-tests (p<0.05). Molecule selection was guided by the following criteria: 1) reproducibility (including directionality of change) in three or more cohorts; 2) >80% correlation and low average measurement shift (<40%) in repeat measurements; and 3) mean experimental values separated from the least detectable dose by more than 20-fold.

Multiplex immunoassay construction
This procedure was guided by optimum dilution of serum (1:5, 1:50, 1:200, 1:10,000, and 1:200,000) and mixing of antibodies to give the most sensitive immunoassays. The 1:5 dilution group consisted of 31 assays which were divided into 4 multiplexes. For the higher dilutions, only one multiplex was used, yielding a total of 8 new multiplexes for the 51 assays. Once the multiplexes were created, large reagent batches were produced to allow testing of >7000 samples (important for consistency of the assays in repeat measurements). All reagents were validated according to sensitivity, linearity, spike recovery, common serum matrix interferences, cross-reactivity, precision, correlation and freeze-thaw stability.

Decision rule construction
The new multiplexed immunoassays were used to measure all 51 molecules on a combined data set of 806 subjects comprised of 577 schizophrenia and 229 control subjects and the resulting data were used to develop a decision rule for separating schizophrenia patients from healthy controls employing the linear Support Vector Machines (SVM) approach. This method uses a linear kernel function and the decision rule creates a separating hyperplane in 51-dimensional space (in this case). The result is a classification algorithm defined by 52 parameters (one offset value and coefficients corresponding to the 51 assays in the multiplex).

Schizophrenia biomarker selection
Twenty-two molecules in DiscoveryMAP assay platform were identified which showed significant differences in schizophrenia compared to control subjects in >3 centres. Technical reproducibility was shown by repeating the measurements approximately 3 months later. This showed an average correlation of 0.83, a measurement shift of 29% and log-distance to the least detectable dose of 1.29. Nine additional assays were selected due to published associations of the targeted molecules with schizophrenia, and/or we identified significant changes in these molecules in studies of schizophrenia patients using other molecular profiling platforms, and assays for 20 other analytes were incorporated showed significant differences between bipolar disorder patients and controls to facilitate the future development of a test with differential diagnosis capability.

Decision rule optimization
Briefly, the new 51-plex immunoassay platform was used to analyse sera from 806 subjects (577 schizophrenia, 229 controls). The resulting 806 vectors of 51 dimensions were then used to develop a linear SVM decision rule for separating schizophrenia patients from healthy controls. The hyperplane was described by 51 multipliers (W1-W51) and one bias term (B) for classification using the expression “S = X1W1 + ... + X51W51 + B” and comparing this with zero. If the result is positive, the vector is classified as an element of the first class (C1). If the result is negative, it is classified as an element of the second class (C2). In addition, we used penalty parameter C and ratio F of penalties C1 and C2, such that C1 = C and C2 = C1F. We then applied the holdout method in which approximately 2/3 of the dataset is used for training, and the remaining part is used for validation. The search of optimal performance was carried out among 20,100 pairs of parameters (C,F) organized as a grid covering the following ranges:

Log2C: -10 to 0 with 0.1 steps (100 values)
Log2F: -1.0 to 1.0 with 0.01 steps (201 values)

The probability C that a subject with score S is a schizophrenia patient was calculated for each sample tested. This results in an output of the probability C that a subject with score S has schizophrenia and the complementary probability 1-C that the subject does not have schizophrenia. These can be classified as: -1 or +1 for “schizophrenia” and “not schizophrenia,” respectively, with a confidence C of the classification.

Decision rule performance
The first decision rule (SVM-A) was optimized to discriminate schizophrenia patients from healthy controls in the combined dataset of 806 subjects. This yielded a cross-validation sensitivity of 0.83, specificity of 0.83 and ROC-AUC of 0.89. Since this cohort contained multiple samples from 99 of the antipsychotic-treated patients, a second decision rule (SVM-B) was constructed using the 707 unique samples. This yielded similar results with a sensitivity of 0.83, specificity of 0.82 and ROC-AUC of 0.88.The conditional probability curves for both SVM-A and SVM-B show that probability C “schizophrenia” y=1, corresponding to the score S = 0, is not 50% since the decision rules were balanced to achieve approximately equal levels of sensitivity and specificity (achieved by tilting the relative weights of the corresponding false-positive and false-negative errors).

For an unbiased estimate of classification performance and validation of the test, we determined performance of the decision rules by analysing 480 samples which had not been used for biomarker selection in phase 1. Application of SVM-B yielded an overall classification accuracy of 0.84. We also calculated classification accuracy for regions I, II, II* and I* of the conditional probabilities (leaving out the indeterminate region III comprising 18.5% of the subjects) which resulted in an increase in sensitivity to 0.96 and specificity to 0.97.

Out of the 478 total patients used in Phase 2, 367 were diagnosed with paranoid schizophrenia and 111 were diagnosed with non-paranoid schizophrenia. Application of SVM-A led to identification of 95 (86%) of these patients correctly, indicating that the biomarker signature was present regardless of the schizophrenia subtype.

We also tested 80 schizophrenia patients before and after 4-6 weeks of antipsychotic treatment, which led to symptom improvements (reduction in PANSS scores). Interestingly, application of SVM-B led to correct identification of 85% of these patients at the first time-point and after the treatment period. There was an average correlation of 0.49 across all 51 molecular assays, supporting the stable identification capability of the decision rule. This suggested that schizophrenia patients in remission still feature schizophrenia-like serum profiles even after 4-6 weeks of treatment.

Conclusions
In this multi-centre study, a biomarker panel for schizophrenia based on biological and technical reproducibility of the molecular signature was discovered and validated. All stages of the process, including conduction of the assays, assay selection, assay panel refinement, development and recalibration of the decision rule, were carried out in a CLIA-certified laboratory at Myriad-RBM. Assay selection was based on a large number of samples collected from antipsychotic naïve, acutely psychotic patients to facilitate relatively uniform conditions. Subjects were recruited from four independent clinical centres and samples collected according to strict standard operating procedures to maximize reliability and accuracy of the results. As the assay progresses from beta site testing into exposure to different subpopulations, the performance against present clinical classification and observed prevalence and incidence must be monitored and differences will need examination.

The implementation of the 51-plex molecular assay decision rule was based on a cohort comprised of both untreated and treated schizophrenia patients who were either experiencing a first episode of illness or who were chronically ill (54% of patients were on current antipsychotic treatment). This collection is likely to represent more closely the patient population encountered in clinical practice. High classification performance demonstrated that the decision rule could identify schizophrenia patients with high accuracy irrespective of the disease duration or treatment state. Interestingly, the biomarker signal was still apparent in subjects even after 4-6 weeks of successful treatment with antipsychotic medication. Further work is required for the development of a biomarker panel aiding in the monitoring of patient responses to treatment, which will be a key objective in the next phase of this project.

Further Information

A more detailed report including figures and graphs can be found as a PDF in both the midterm and period 2 reports and also in the Deliverable reports

Potential Impact:
Socio Economic Impact
People with mental disorders experience significant difficulties in living their ordinary lives, participating in their communities and occupational pursuits, and integrating into society in general. This leads to a significant economic impact. Schizophrenia is a severe mental disorder characterized by fundamental disturbances in thinking, perception and emotions and there is currently no “objective” biological test for diagnosis.
The current diagnosis of schizophrenia (and bipolar disorder) is rather subjective, with psychiatrists usually coming to their conclusions based on several patient interviews and clinical observations and evidence of prolonged manifestation of specific clinical symptoms. By any reckoning this is a rather subjective and sometimes long-lasting uncertain process. Moreover, numerous mental disorders have similar characteristics, making absolute diagnosis very difficult. Many times patients visit their doctor during the prodrome phase due to symptoms such as anxiety, social isolation, difficulty making choices, and problems with concentration and attention. As symptoms often present during adolescence, distinguishing schizophrenia from “normal” teenage angst is made more difficult. The current diagnosis of schizophrenia involves examination and monitoring by psychiatrists, a process which can take between 2 and 5 years.
Although schizophrenia is a severe mental disease affecting primarily the brain, it is becoming more apparent that the whole body is involved. Studies over the last two decades have shown that many patients with schizophrenia or other psychiatric disorders have abnormalities in insulin signalling, similar to those seen in type II diabetes mellitus. There are also reports that schizophrenia is associated with a mild pro-inflammatory state resulting in increased permeability of the brain-blood barrier. Therefore, brain “signature” changes in biomolecules should be reflected by changes in the cerebrospinal fluid (CSF) and blood (serum/ plasma) levels of the same or related molecules. It is through this process that peripheral biomarkers for brain disorders can be identified. In order to increase the efficiency of the diagnostic process and to guide treatment approaches, the identification of molecular biomarkers appears to be essential. Most biomarker approaches for schizophrenia and most other medical conditions have thus far focussed on overall changes in protein levels.
The early detection of schizophrenia is a widely accepted goal in psychiatry. The rationale for early detection is based on several observations and as a consequence:
• Diagnosis and treatment of schizophrenia are often seriously delayed.
• Consequences of the disease are already very severe in the early preclinical, undiagnosed phase of the disorder.
• Early treatment seems to improve the course of the disease.
Schizophrenia has serious financial consequences for patients, their relatives and the national economy. The economic impact of schizophrenia is caused by specific symptoms and characteristics of the disease such as early onset, a frequent chronic course, early retirement, excess mortality, regular readmissions to hospital treatment, and high rate of disability.

Schizophrenia and related disorders are a major burden to affected individuals, to their families and to society at large. These severe mental illnesses affect at least 2% of the population worldwide and, while 50% of sufferers do not receive adequate treatment, they cost hundreds of billions in healthcare provision, treatments and lost earnings.
Negative attitudes towards the mentally ill, especially towards persons with schizophrenia are widespread. Affected individuals are looked at as being dangerous and unpredictable. Many media reports reflect this fear even if in reality a potential risk is mainly directed to the closest relatives. This and other stigma attached to schizophrenia creates a vicious cycle of discrimination leading to social isolation, unemployment, drug abuse, long lasting institutionalisation or even homelessness. These are all factors that further decrease the chances of recovery and reintegration into normal life, in addition to the deleterious consequences of the illness itself.
The SchizDX project has resulted in new discoveries about the pathophysiology underlying these disorders, which has culminated in the launch of the first molecular assay for the diagnosis of schizophrenia (VeriPsych®) by Psynova and Rules Based Medicine in the USA. The test is being marketed to assist with the diagnosis of first onset schizophrenia and may result in a quicker and more cost effective diagnosis for some patients. It is well known that early intervention can result in a much better outcome for schizophrenia patients. Therefore, the outputs from the SchizDX project may be of significant benefit to patients.
The suite of biomarkers identified by the project is also being evaluated for diagnosis of other mental disorders including bipolar disorder, various forms of depression and autism etc. It is often challenging for clinicians to distinguish between these disorders due to overlapping symptoms so a molecular test capable of differentiating between such disorders would be of major benefit to patients.
The societal benefit of improved diagnosis of mental disorders will be manifested through more effective treatment of patients leading to a reduced financial burden on healthcare systems and better integration of patients into the community.
The SchizDX project is contributing to a better understanding of mental disorders such as schizophrenia and, in time, this improved understanding may help to combat the widespread negative attitudes towards the mentally ill.
Dissemination activities
Dissemination of knowledge to the wider scientific community as well as the public, which will also include a non-specialist audience, was achieved by setting up of a website that is password controlled. The website is located at http://schizdx.pera.com .The website has been regularly updated during the project with all project documents i.e. meeting minutes, presentations and deliverable reports. The public section of the website contains a publishable summary of the project.
The member site has a document store with folders for:
• Deliverable reports
• Project documents
• Meeting minutes and presentations
• Official reports
Additionally there is contact information for all consortium members to allow communication between partners to be easier.
A public symposium describing the overall impact of the project was held at Cambridge University and a CD-ROM containing the presentations is available on request.
The SchizDX project was promoted at a variety of worldwide conferences/symposiums by a number of the consortium partners:
Conferences Attended
2011
Sabine Bahn - DGPPN Psychiatry Meeting, Berlin, Germany (Nov 23 – 26); American Psychiatric Association, Honolulu, Hawaii (keynote speaker) (May 14 – May 18); 10TH World Congress of Biological Psychiatry, Prague, CZ (May 29 – June 2); Winter workshop on Psychosis; Innsbruck, Austria (Jan 30 – Feb 2); ACES European Entrepreneur Meeting, Zurich (Feb 3)
Emanuel Schwarz - 10TH World Congress of Biological Psychiatry, Prague, CZ (May 29 – June 2); 4th Annual Biomarkers Summit, London (Jan 1 – 2)
Paul Guest - Invited Speaker for “Biomarkers: Their Application in Clinical Trials” (Healthcare Education Services) (May 17 – 18)
Hassan Rahmoune - Next Generation Pharmaceuticals Drug Development Summit Europe. Bremen Germany (April 5 – 7); Next Generation Pharmaceuticals Drug Development Summit Europe. Vilamoura, Portugal (Nov 29 – Dec 1)
Jayne Telford - Glycoscience Ireland Conference, Teagasc, Fermoy, Cork, Ireland, October 2011; Neuroscience Ireland Conference, National University of Maynooth, Kildare, Ireland, September 2011; Wiring the Brain Conference, Powerscourt, Wicklow, Ireland, April 2011

2010
Sabine Bahn - Weißenauer Schizophrenie-Symposium, Cologne, Germany (keynote speaker); SMRI annual meeting, Bethesda, USA; DGPPN Psychiatry meeting, Berlin, Germany (Nov 24 – 27); HUPO Meeting – from bench to bedside, Denver, USA (Mar 7-10); Proteomics in Neuroscience, organised by the Foundation for the National Institutes of Health (FNIH), Bethesda, USA (keynote speaker)
Emanuel Schwarz - Clinical Proteomics, Madrid, Spain

Paul Guest - Invited Speaker for “Biomarkers: Their Application in Clinical Trials” (Healthcare Education Services) (May 18 – 19)
Jayne Telford - Neuroscience Ireland Conference, Health Science Building, UCD, Dublin, Ireland, September 2010; Glycoscience Ireland Conference, Health Science Building, UCD, Dublin, Ireland, August 2010; EuroGlycoforum Glycosylation and Disease meeting, Conway Building, UCD, Dublin, Ireland, August 2010

2009
Sabine Bahn - Proteomics Symposium, Max-Planck Institute Goettingen, Germany (Nov 19); World Congress of Schizophrenia Research, Paris, France (June 28 – July 2); NYAS Biomarkers in Brain Disease Meeting, Oxford, UK (keynote speaker) (Jan 27 – 28)
Paul Guest - Invited Speaker for “Biomarkers: Their Application in Clinical Trials” (Healthcare Education Services) (May 12 – 13)
Hassan Rahmoune - Invited Speaker for “Biomarkers: Their Application in Clinical Trials” (Healthcare Education Services) (May 12 – 13)

Furthermore, 21 scientific papers reflecting several high impact and ground-breaking results were published in peer reviewed international journals, numerous oral and poster presentations were carried out at relevant conferences.
1) Martins-de-Souza D, Guest PC, Vanattou-Saifoudine N, Rahmoune H, Bahn S. Phosphoproteomic differences in major depressive disorder post-mortem brains indicates effects on synaptic function. Eur Arch Psychiatry Clin Neurosci. 2012 Feb 21. [Epub ahead of print]
2) Guest FL, Rahmoune H, Bahn S, Guest PC. The effects of stress on hypothalamic-pituitary-adrenal (HPA) axis function in subjects with psychiatric disorders. Revista de Psiquaitria Clinica (in press).
3) Rahmoune H, Harris LW, Guest PC, Bahn S. Targeting the inflammatory component of schizophrenia. Revista de Psiquaitria Clinica (in press).

4) Bahn S, Harris LW, Rahmoune H, Martins-de-Souza D, Guest PC. Biomarker blood tests for diagnosis and management of mental disorders. Revista de Psiquaitria Clinica (in press).
5) Martins-de-Souza D, Guest PC, Rahmoune H, Bahn S. Proteomic approaches to unravel the complexity of schizophrenia, Expert Rev Proteomics. 2012 Feb;9(1):97-108.
6) Martins-de-Souza D, Guest PC, Harris LW, Vanattou-Saifoudine N, Webster M, Rahmoune H, Bahn S. Identification of proteomic signatures associated with depression and psychotic depression in post-mortem brains from major depression patients. Translational Psychiatry (in press).
7) Schwarz E, Guest PC, Steiner J, Bogerts B, Bahn S. Identification of blood-based molecular signatures for prediction of response and relapse in schizophrenia patients. Translational Psychiatry (in press).
8) Herberth M, Rahmoune H, Schwarz E, Koethe D, Harris LW, Kranaster L, Witt SH, Spain M, Barnes A, Schmolz M, Leweke FM, Guest PC, Bahn S. Identification of a molecular profile associated with immune status in first onset schizophrenia patients. Clinical Schizophrenia and Related Psychoses (in press).
9) Guest PC, Bahn S. Biomarker Discovery: Psychiatric Health. Diagnosing Schizophrenia. European Biopharmaceutical Review, Jan 2012.
10) Krishnamurthy D, Harris LW, Levin Y, Koutroukides TA, Rahmoune H, Pietsch S, Vanattou-Saifoudine N, Leweke FM, Guest PC, Bahn S. Metabolic, hormonal and stress-related molecular changes in post-mortem pituitary glands from schizophrenia subjects. World J Biol Psychiatry. 2012 Jan 17. [Epub ahead of print]
11) Jaros JA, Martins-de-Souza D, Rahmoune H, Schwarz E, Leweke FM, Guest PC, Bahn S. Differential phosphorylation of serum proteins reflecting inflammatory changes in schizophrenia patients. Eur Arch Psychiatry Clin Neurosci. 2011 Dec 15. [Epub ahead of print]
12) Bahn S, Noll R, Barnes A, Schwarz E, Guest PC. Challenges of introducing new biomarker products for neuropsychiatric disorders into the market. Int Rev Neurobiol. 2011;101:299-327.
13) Izmailov R, Guest PC, Bahn S, Schwarz E. Algorithm development for diagnostic biomarker assays. Int Rev Neurobiol. 2011;101:279-98.
14) Schwarz E, VanBeveren NJ, Guest PC, Izmailov R, Bahn S. The application of multiplexed assay systems for molecular diagnostics. Int Rev Neurobiol. 2011;101:259-78.
15) Guest PC, Martins-de-Souza D, Vanattou-Saifoudine N, Harris LW, Bahn S. Abnormalities in metabolism and hypothalamic-pituitary-adrenal axis function in schizophrenia. Int Rev Neurobiol. 2011;101:145-68.
16) Chan MK, Guest PC, Levin Y, Umrania Y, Schwarz E, Bahn S, Rahmoune H. Converging evidence of blood-based biomarkers for schizophrenia: An update. Int Rev Neurobiol. 2011;101:95-144.
17) Schwarz E, Guest PC, Rahmoune H, Martins-de-Souza D, Niebuhr DW, Weber NS, Cowan DN, Yolken RH, Spain M, Barnes A, Bahn S. Identification of a blood-based biological signature in subjects with psychiatric disorders prior to clinical manifestation. World J Biol Psychiatry. 2011 Sep 22. [Epub ahead of print]
18) Schwarz E, Guest PC, Rahmoune H, Harris LW, Wang L, Leweke FM, Rothermundt M, Bogerts B, Koethe D, Kranaster L, Ohrmann P, Suslow T, McAllister G, Spain M, Barnes A, van Beveren NJ, Baron-Cohen S, Steiner J, Torrey FE, Yolken RH, Bahn S. Identification of a biological signature for schizophrenia in serum. Mol Psychiatry. 2011 Apr 12. [Epub ahead of print].
19) Guest PC, Schwarz E, Krishnamurthy D, Harris LW, Leweke FM, Rothermundt M, van Beveren NJ, Spain M, Barnes A, Steiner J, Rahmoune H, Bahn S. Altered levels of circulating insulin and other neuroendocrine hormones associated with the onset of schizophrenia. Psychoneuroendocrinology. 2011 Aug;36(7):1092-6.
20) Stanta JL, Saldova R, Struwe WB, Byrne JC, Leweke FM, Rothermund M, Rahmoune H, Levin Y, Guest PC, Bahn S, Rudd PM. Identification of N-glycosylation changes in the CSF and serum in patients with schizophrenia. J Proteome Res. 2010 Sep 3;9(9):4476-89. Erratum in: J Proteome Res. 2010 Oct 1;9(10):5510.
21) Schwarz E, Izmailov R, Spain M, Barnes T, Mapes JP, Guest PC, Rahmoune H, Leweke FM, Rothermundt M, Steiner J, Koethe D, Kranaster L, Ohrmann P, Suslow T, Levin Y, Bogerts B, van Beveren JM, McAllister G, Weber N, Niebuhr D, Cowan D, Yolken RH, Bahn S (2010). Validation of a Blood-Based Laboratory Test to Aid in the Confirmation of a Diagnosis of Schizophrenia. Biomark Insights 5, 39-47.

Exploitation activities
The major exploitation activity for the SchizDX project has been the launch of the first molecular test for the diagnosis of schizophrenia (VeriPsych®) by Psynova Neurotech and Myriad-RBM in the USA. The test is being marketed to assist with the diagnosis of first onset schizophrenia and may result in a quicker and more cost effective diagnosis for some patients. It is well known that early intervention can result in a much better outcome for schizophrenia patients. Therefore, the outputs from the SchizDX project may be of significant benefit to patients.
The suite of biomarkers identified by the project is also being evaluated for diagnosis of other mental disorders including bipolar disorder, various forms of depression and autism etc. It is often challenging for clinicians to distinguish between these disorders due to overlapping symptoms so a molecular test capable of differentiating between such disorders would be of major benefit to patients. If successful, this could lead to the launch of additional molecular tests for diagnosis of, or discrimination between, other mental disorders.

9 patents have been applied for by Cambridge University
1) Schizophrenia Drug Response
2) 51 schizophrenia markers
3) Specific panel of 13 schizophrenia markers
4) PBMC markers for schizophrenia
5) Whole blood markers for schizophrenia
6) Pituitary Derived Schizophrenia markers
7) Data mined schizophrenia markers
8) Serum schizophrenia markers
9) 34 differential schizophrenia markers
Two markets surveys reports were prepared (Market landscape for biomarkers, antibodies and antibody based research-devices and Report on market penetration) to enable the consortium members to exploit the project deliverables at the end of the project.

List of Websites:
http://schizdx.pera.com
Contact details of the coordinator are as above: -
Dr Paul Rodgers,
Psynova Neurotech Ltd
St John's Innovation Centre, Cambridge, CB4 0WS, UK
Tel: +44 1223 703146
paul.rodgers@psynova.com.
The contact details for the project administrator are: -
Ms Barbara Baron,
Psynova Neurotech Ltd
St John's Innovation Centre, Cambridge, CB4 0WS, UK
Tel: +44 1223 334161
barbara.baron@psynova.com

Related information

Reported by

PSYNOVA NEUROTECH LTD
United Kingdom

Subjects

Business aspects
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top