Diagnostic and prognostic biomarker discovery strategies for autoimmune disorders

Abstract

Current clinical, laboratory or radiological parameters cannot accurately diagnose or predict disease outcomes in a range of autoimmune disorders. Biomarkers which can diagnose at an earlier time point, predict outcome or help guide therapeutic strategies in autoimmune diseases could improve clinical management of this broad group of debilitating disorders. Additionally, there is a growing need for a deeper understanding of multi-factorial autoimmune disorders. Proteomic platforms offering a multiplex approach are more likely to reflect the complexity of autoimmune disease processes. Findings from proteomic based studies of three distinct autoimmune diseases are presented and strategies compared. It is the authors' view that such approaches are likely to be fruitful in the movement of autoimmune disease treatment away from reactive decisions and towards a preventative stand point.

Full text

Review
Diagnostic and prognostic biomarker discovery strategies for
autoimmune disorders
David S. Gibson
a,
⁎, Joao Banha
b,c,d,e
, Deborah Penque
b
, Luciana Costa
c,e
,
Thomas P. Conrads
d
, Dolores J. Cahill
f
, John K. O'Brien
g
, Madeleine E. Rooney
a
a
Arthritis Research Group, Queen's University Belfast, Belfast, BT9 7BL, UK
b
Laboratório de Proteómica, Departamento de Genética, Instituto Nacional de Saúde Dr Ricardo Jorge, 1649-016 Lisboa, Portugal
c
Grupo de Imunologia Molecular e Celular, Departamento de Promoção da Saúde e Doenças Crónicas,
Instituto Nacional de Saúde Dr Ricardo Jorge, 1649-016 Lisboa, Portugal
d
Department of Pharmacology and Chemical Biology and Mass Spectrometry Platform, University of Pittsburgh Cancer Institute,
Pittsburgh, PA, USA
e
Human Molecular Genetics and Functional Analysis Unit, Center for Biodiversity, Functional & Integrative Genomics (BioFIG) Campo Grande -
1749-016 Lisboa, Portugal
f
Conway Institute of Biomedical and Biomolecular Science, University College Dublin, Belfield, Dublin 4, Ireland
g
Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SA, UK
ARTICLE INFO ABSTRACT
Current clinical, laboratory or radiological parameters cannot accurately diagnose or predict
disease outcomes in a range of autoimmune disorders. Biomarkers which can diagnose at
an earlier time point, predict outcome or help guide therapeutic strategies in autoimmune
diseases could improve clinical management of this broad group of debilitating disorders.
Additionally, there is a growing need for a deeper understanding of multi-factorial
autoimmune disorders.
Proteomic platforms offering a multiplex approach are more likely to reflect the complexity
of autoimmune disease processes. Findings from proteomic based studies of three distinct
autoimmune diseases are presented and strategies compared. It is the authors' view that
such approaches are likely to be fruitful in the movement of autoimmune disease treatment
away from reactive decisions and towards a preventative stand point.
Crown Copyright © 2009 Published by Elsevier B.V. All rights reserved.
Keywords:
Proteomics
Inflammation
Autoimmune disease
Behçets disease
Juvenile idiopathic arthritis
Dilated cardiomyopathy
Contents
1. Introduction .........................................................1046
2. Behçet's disease .......................................................1046
2.1. Epidemiology and immunology of Behçet's disease ................................1046
2.2. Diagnosis and treatment of Behçet's disease. ...................................1047
2.3. Proteomic approaches in Behçet's disease .....................................1047
⁎Corresponding author. Arthritis Research Group, Queen's University Belfast, Room 4.4a Microbiology Building, Royal Victoria Hospital,
Grosvenor Road, Belfast, BT12 6BA, UK. Tel.: + 44 28 90632617; fax: + 44 2890 661112.
E-mail address: d.gibson@qub.ac.uk (D.S. Gibson).
available at www.sciencedirect.com
www.elsevier.com/locate/jprot
JOURNAL OF PROTEOMICS 73 (2010) 1045–1060
1874-3919/$ –see front matter. Crown Copyright © 2009 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.jprot.2009.11.013

3. Juvenile idiopathic arthritis................................................. 1048
3.1. Classification, symptoms and pathology of juvenile idiopathic arthritis .................... 1048
3.2. Current issues in the management and treatment of JIA .............................. 1050
3.3. Proteomic approaches in juvenile arthritides ................................... 1051
4. Dilated cardiomyopathy .................................................. 1053
4.1. Autoantibody detection ............................................... 1053
4.2. Pathology of dilated cardiomyopathy (DCM).................................... 1053
4.3. Proteomic study of dilated cardiomyopathy .................................... 1054
4.4. Protein sequence identification and characterisation............................... 1055
5. Bioinformatics and statistical analysis........................................... 1055
6. Discussion and conclusions ................................................ 1057
Acknowledgements........................................................ 1058
References............................................................. 1058
1. Introduction
Autoimmune diseases are a heterogeneous group of disorders
characterized by a humoral cell mediated immune response
directed against a variety of tissues. A cardinal feature of
autoimmune disease is the presence of autoantibodies in the
systemic circulation and in specific proximal fluids and tissues,
such as the synovial joint in arthritides. The aetiology of
autoimmune disease remains poorly understood and due to the
multi-factorial nature of such a complex disorder, approaches
directed at implicating individual molecules have had limited
impact on early diagnosis and reliable prediction of outcome. The
strongest evidence for the cause of autoimmune disease points to
a deregulation of genes involved in the inspection and tolerance
of self antigens and possible manipulation of these genes by
external environmental factors. Although genetic alterations may
act as a source of uncharacteristic immune system behaviour, the
clinical manifestations of autoimmune disease at the system or
organ level, such as joint inflammation and erosion in arthritis,
are ultimately enacted by proteins.
Unsupervised proteomic platforms offerthe ability to reliably
quantify and identify a large subset of the proteins which
contribute to autoimmune disease. Typically two-dimensional
gel electrophoresis and liquid chromatography columns are
used to separate thousands of proteins by their physical
attributes such as molecular weight, isoelectric point and
hydrophobicity [1]. Importantly, differential comparisons can
be made between autoimmune diseases and their subtypes,
outcomes and therapeutic responses. The unbiased nature of
these techniques is what really makes them stand apart from
the limitations of genetic screening methods such as transcript
arrays [2]. Within protein chips, candidate biomarker identity
can be preordained by the user through a matrix of designated
capture antibodies or nucleotides. These arrays offer a high
throughput systemwhich could highlight, for example, putative
diagnostic proteins expressed at significantly higher or lower
levels at an earlier disease stage than is currently possible. Pools
of candidate biomarker proteins which segregate or identify
patients on the basis of clinically relevant criteria can then be
identified by various modes of mass spectrometry. Supervised
array methods are of particular use when the diagnostic or
predictive power of a predetermined set of proteins requires
validation in much larger patient cohorts, as they are capable of
much higher throughput [1,3]. Both modes of proteome
investigation in clinical samples are described to illustrate
their ability to isolate clinically useful biomarkers.
This review highlights the application of three proteomic
strategies applied against a range of autoimmune diseases
presentedat the 3rd European Proteomics Association: Behçet's
disease, juvenile idiopathic arthritis and dilated cardiomyopa-
thy. As with a number of autoimmune disorders, a common
theme across these three conditions is the cyclical pattern of
remission and relapse insymptoms. These disorders present in
the human population with variable incidence, degrees of
severity and clinical outcome. They collectively present chal-
lenges in reliable diagnosis by conventional clinical means and
so providean opportunity to showcase several recent biomarker
discovery studies utilising various proteomic methods. The
conditions described generally have a similar or more frequent
incidence than better recognized autoimmunediseases such as
rheumatoid arthritis orsystemic lupus erythrematosus and are
just as dehabilitating and require more reliable diagnostic
criteria than those currently available. A final section is also
included to highlight the role of bioinformatics in biomarker
discovery and validation. The development of reliable biomark-
er panels could help in characterising the disease type, outcome
and response to therapy of many autoimmune disorders.
2. Behçet's disease
2.1. Epidemiology and immunology of Behçet's disease
Behçet's disease (BD) is considered a multi-systemic inflam-
matory disease, mainly characterized by oral and genital
ulcers, uveitis (eye inflammation that can lead to blindness)
and skin lesions, the pathogenesis of which remains unclear
[4]. The pathophysiology of BD equally involves the joints,
lungs, kidneys [5], central nervous system and gastrointestinal
system [6,7]. Further, BD is characterized as a vasculitis in both
arteries and veins in any kind of organ [8]. The clinical
characteristics of BD include recurrent episodes of remission
(inactive BD) and the exacerbation (active BD) of the fore
mentioned symptoms and generalized vasculitis [6].
The geographic distribution of BD is more prevalent along the
“Silk Route”, an ancient trade route between the Mediterranean
1046 JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

and Eastern Asia, with Turkey having the highest incidence of
BD with prevalence rates between 110 and 420 per 100,000
habitants. In Portugal, the National Study Group for BD reported
a prevalence of 2.5 per 100,000 habitants [9].
Although the specific aetiology remains unknown, the
present dogma suggests that BD is caused by unknown
environmental triggering factor(s), ranging from infectious
agents to pollution, against a background of genetic suscep-
tibility [4]. HLA-B51 is the strongest genetic predisposition
factor for BD defined so far, while other MHC associated genes,
such as MICA and TNF, might contribute to this disease due to
a linkage disequilibrium with HLA-B51 [10].
Vasculitis is considered very important in BD pathogenesis,
being responsible for most of the symptoms observed during
the course of the disease. In fact, the endothelium seems to be
the primary target in this disease; however, it may just be the
subject of the abnormal behaviour of the immune system [10].
Neutrophil hyperactivity has been described as an important
aspect of the immunological dysfunction observed in BD. In
fact, there is an increase in activation markers, chemotaxis,
phagocytosis, superoxide production and myeloperoxidase
expression in neutrophils from patients with BD [11]. Also,
Oliveira R. et al. elucidated a possible role of leukocyte-
associated ceruloplasmin in BD pathogenesis by flow cyto-
metry with increased ceruloplasmin expression at the surface
of monocytes in BD patients in comparison to controls [12].
The mechanism underlying the neutrophil hyperactivity in
this disease is unknown but a fundamental role for T cells has
been suggested, by priming neutrophils with pro-inflamma-
tory cytokines [13].
Since there is no consensus regarding the autoimmune
origin of DB, this disease has also been referred to as an
autoinflammatory disease [14]. Autoinflammatory describes a
broad group of diseases distinct from the classical organiza-
tion of immune diseases which are autoimmunity, allergies or
immunodeficiencies [15]. Autoinflammatory diseases are
characterized by recurrent episodes of systemic inflamma-
tion, usually manifested by fever and inflammation in specific
tissues such as joints, skin, eyes and gastrointestinal system
which is only characteristic of hereditary periodic fever
syndromes such as Familial Mediterranean Fever [15,16].
Despite the clinical characteristics of these diseases being
similar to infection or common rheumatologic diseases, there
is no evidence of pathogens and no increase in antibody
concentration or T cell specific antigens which are usually
evident in autoimmune diseases [17].
2.2. Diagnosis and treatment of Behçet's disease
Since a specific laboratory test for BD is lacking, its diagnosis is
based exclusively on clinical assessment of patients [18]. This
situation poses a great challenge for physicians since BD
symptoms resemble several other confounding immune
dysfunctions. Several classifications had been proposed but
only in 1990 there was a consensus with the creation of the
International Study Group Classification for BD [8].
Several laboratory tests have indicated leukocytosis,
increases in C-reactive protein, IgG, IgA and IgM levels but
none have revealed to be sensitive or specific enough for BD
diagnosis [5]. Likewise, HLA typing in the clinical context is not
useful due to the low association with HLA-B51 [7]. On the other
hand, it has been reported that BD presents an inflammatory
response against autoantigens such as retinal S-antigen [19],
heat shock protein [20],α-tropomyosin [21,22], selenium-
binding protein [23] and oxidized low-density lipoproteins [24].
It is still unclear whether these autoantigens are truly patho-
genic or if the immune response directed against them results
from an inflammatory reaction attributed to disease activation.
Treatment regimens for BD patients remain empirical and
there are considerable differences in the practical approaches
to treatment despite all being based in anti-inflammatory and
immunosuppressant agents. The main objectives in treating
patients with BD are symptom management, inflammation
suppression at an early stage and damage control to specific
organs.
2.3. Proteomic approaches in Behçet's disease
There are few publications using proteomic tools and mass
spectrometry (MS) technologies to study BD. When using the web
search engine dedicated to life sciences and biomedical topics
(PubMed) to search for “Proteomics AND Behçet”as keywords in
all fields, only three papers are retrieved [23,25,26]. Changing the
keyword from “Proteomics”to “Mass Spectrometry”four addi-
tional papers were obtained [22,27–29]. In 1999, Ames et al. aimed
to evaluate the occurrence and clinical significance of lipid
peroxidation (oxidative stress) in rheumatic diseases character-
ized by vascular involvement, such as systemic lupus erythema-
tosus (SLE), rheumatoid arthritis and BD, using gas
chromatography-MS to measure plasma oxidative stress marker
8-epi-PGF2alpha [29]. They concluded that oxidative stress may
be pathogenically relevant in some autoimmune rheumatic
diseases with vascular involvement. Later, in 2005, Karasawa et
al. investigated whether autoimmunity to the anti-oxidative
peroxiredoxin enzymes exists in patients with systemic autoim-
mune diseases [25]. The results obtained showed autoantibodies
to peroxiredoxin I may be involved in the pathophysiology of
systemic autoimmune diseases such as SLE and primary
vasculitis syndrome, but not so significant in RA and BD patients.
In 2002, Mor et al. identified alpha-tropomyosin as a target
self-antigen in BD. They used patient sera to immunoblot
tissue lysates, and found that some patients manifest anti-
bodies to a 37 kDa band detected in extracts of skin, tongue,
vagina, muscle and heart but not in brain, kidney, lung, liver,
intestine and thymus [22]. In-gel digestion followed by MS
analysis revealed the band to be alpha-tropomyosin. Addi-
tionally, they suggested autoimmunity to alpha-tropomyosin
to be pathogenic since immunized Lewis rats developed
lesions in the uveal tract and skin, with features of BD. In
fact, these findings were sustained by Mahesh et al. when they
reported an increase in lymphoproliferative response to
alpha-tropomyosin as well as observing autoantibodies to
the same protein in BD patients with posterior uveitis [21].
Another self-antigen was found in 2003 by Lee et al., where
they used a 2D-PAGE–immunoblot–MALDI-TOF analysis pipe-
line to identify alpha-enolase as the target protein in anti-
endothelial cell antibody (AECA) reacting BD patients [26].
Using the same pipeline, Okunuki et al. identified selenium-
binding protein as novel autoantigen in BD patients with
uveitis [23]. A different set of patients with the same clinical
1047JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

characteristics (uveitis) was analysed by Mao et al., in 2008,
who found elevated levels of haptoglobin and amyloid A in
serum from these BD patients [27].
Throughout the last decade several proteomic platforms
have been developed and perfected with purposes that go
from studying protein–protein interactions and post-transla-
tional modifications to protein detection and quantification
[31]. Mass spectrometers measure with high precision the
mass-to-charge (m/z) ratios of ionizable compounds. Hence,
mass spectrometry (MS) quickly emerged as the technology of
choice for protein and peptide analysis [31]. Furthermore, the
development of quantitative proteomic approaches has facil-
itated the application of MS to identifying biomarkers for
various diseases, including cancers [33].
The new proteomic platforms of high throughput analysis
provide the tools for the discovery of specific biomarkers for
BD. Improved protein extraction protocols combined with
recently developed MS techniques and fully annotated
genomic databases has allowed the identification of trace
amounts of proteins present in complex samples [34]. Most of
these techniques utilise peptide labelling with heavy isotopes.
The absolute amount of a given protein in a sample can be
measured using as internal control labelled peptides added to
the sample in a known amount [35]. One promising area in
quantitative proteomics is the use of label-free methods in
which the peak area for peptide ions in mass spectra is directly
used as a measure of protein abundance [36].
Mass spectrometers are already in clinical practice as a
diagnostic tool in laboratories. Several intermediary metabolic
compounds in serum or urine are detected using MS to diagnose
metabolic diseases. The biological principle on which this
technique is based is that the composition of ionizable molecules
in biological samples can change in diseases and thus can be
used as the basis for their diagnosis [31]. For example, matrix-
assisted laser desorption/ionisation (MALDI) MS can now be
used directly in tissues when combined with imagiology. The MS
identification of a specific peptide in a tissue section gives
information on its localization, profile and distribution between
different tissues. If these techniques become robust enough it is
most likely that direct profiling of tissues and serum or other
biological fluids such as urine or saliva become common practice
in a diagnostic laboratory. Advancement in instrumentation is
essential for MS-based clinical diagnosis. Miniaturization, ro-
bustness, and simplicity of operation are key factors in making it
feasible for routine diagnostics. The standardization of mass
spectrometers across laboratories is also essential for the
reproducibility of diagnostic tests [32].
Absolute quantification approaches have been used to
measure the actual amount of proteins in a sample. This is
usually done by spiking known amounts of heavy synthetic
peptides and quantifying signature ions in the selected-reaction-
monitoring mode. Then, in the selected-reaction-monitoring
mode, using mass filters in tandem MS, a precursor ion and one
of its fragment ions (a prominent ion in MS/MS spectrum) are
selected for quantification [32]. On the basis of the same principle,
multiple analytes can also be simultaneously monitored by MS
(multiple reaction monitoring), which is analogous to multiplexed
quantification in a single experiment. One of the clinical
applications of this is the routine analysis of peptides derived
from established biomarkers in serum. The advantage of these
methods is that it is possible to monitor unique isoforms of
proteins because analysis is carried out at the peptide level [37].
Sample separation and protein identification methods from key
studies referred to within this review are summarised in Table 1.
Recently, due to the absence of a specific laboratory test for BD
so far, the group at Instituto Nacional de Saúde, Portugal (authors
JB, DP, TPC and LC), has sought to discover new and specific
biomarkers for BD by doing a large scale profiling of proteins/
peptides in serum proteome of BD patients [30]. The workflow is
illustrated in Fig. 1. Briefly, the serum was pooled and depleted
from the top 14 abundant proteins and further separated with
SDS-PAGE followed by reversed-phase liquid chromatography
(RPLC)-tandem MS with a linear ion trap MS. As expected,
numerous proteins involved in inflammatory processes were
identified as differentially abundant between BD and control
individual sera. A different set of proteins identified discrimina-
tory for BD patients relative to healthy individuals is directly
associated to coagulation and adhesion processes. Overall, the
results obtained in this study are in accordance to the vasculitis
processes characteristic of BD. Interestingly, two proteins associ-
ated with eye development were identified as being of decreased
abundance in sera from BD patients as compared to controls. On
the other hand and rather surprisingly, these results suggest that
alterations in lipid metabolism could be involved in BD patho-
physiology. This proteomic approach allowed the unbiased
identification of a large number of proteins involved in different
biological processes mainly in defence mechanisms and stress
response pathways as expected in chronic inflammatory diseases
such as BD and is in accordance to some of the previously
reported findings for this disease.
3. Juvenile idiopathic arthritis
3.1. Classification, symptoms and pathology of juvenile
idiopathic arthritis
Around one in every thousand children in the UK suffers from
juvenile idiopathic arthritis (JIA) [38]. JIA is a heterogeneous group
of inflammatory disorders primarily affecting the musculoskele-
tal system. Unlike adult rheumatoid arthritis, JIA may affect
skeletal growth and development [39].OfthesevensubsetsofJIA
identified according to the ILAR classification [40], three are the
most common: oligoarticular, extended oligoarticular, and poly-
articular. Adverse outcomes can present to varying degrees
regardless of disease subtype, but persistently inflamed joints
are a major risk factor [41]. All three subtypes are characterised by
periods of exacerbation and remission. However the impact of
inflammation flare ups on management and outcome differ
significantly between the three groups. In addition, nearly half the
patients on disease modifying therapies will have relapses after
premature discontinuation of treatment. The rate of relapses may
be influenced by residual synovial inflammation, which is not
clinically apparent [42].
The most common symptoms of JIA are persistent joint
swelling, pain and stiffness. The main adverse outcomes
include limitation of joint function, structural damage due to
erosions and in some cases severe disability or death. Adverse
outcomes are observed to varying degrees regardless of
disease subtype, but persistently inflamed joints are a major
1048 JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

Table 1 –Key proteomic strategies reviewed and disease associated proteins identified. IEC = ion exchange chromatography; PAGE = polyacrylamide gel electrophoresis; ESI-
MS = electrospray ionisation mass spectrometry; 2-DE = two-dimensional gel electrophoresis; MALDI-TOF-MS = matrix-assisted laser desorption ionisation time-of-flight
mass spectrometry; SELDI-TOF-MS = surface-enhanced laser desorption ionisation time-of-flight mass spectrometry; IP = immunoprecipitation; CIC = combined immune
complexes; SEC = size exclusion chromatography; LC = liquid chromatography.
Autoimmune disease Separation and identification
methods (sample type)
Proteins identified
(*autoantigen)
Biomarker purpose
(disease mechanisms)
Study reference/author
Behcet's disease IEC, SDS-PAGE and ESI-MS
(sera)
Alpha-tropomyosin* Screening/diagnostic (multisystem inflammation) [22] Mor et al.
2-DE and immunoblot
(sera)
Selenium-binding protein* Screening/outcome prediction (ocular inflammation) [23] Okunuki et al.
2-DE and MALDI-TOF-MS
(sera)
Alpha-enolase* Screening/diagnostic (vascular inflammation) [26] Lee et al.
2-DE and MALDI-TOF/TOF-MS
(sera)
Haptoglobin, serum amyloid A Outcome prediction (disease activity) [27] Mao et al.
Juvenile
idiopathic
arthritis
Chromatographic protein
chips, SELDI-TOF-MS
(sera/urine).
Transferrin, ceruloplasmin Diagnostic (active renal disease) [52] Suzuki et al.
Serum amyloid A Therapeutic response/outcome prediction (disease activity) [54] Miyamae et al.
IP of CIC's, 2DE, ESI-MS/MS (serum) Serotransferin*, GAPDH*, alpha-1
anti-trypsin*
Screening/outcome prediction (disease activity) [57] Low et al.
IP of CIC's, SEC/LC, ESI-MS
(serum)
Citrulinated fibrinogen,
complement 3, complement 1q
Outcome prediction (joint inflammation) [58] Zhao et al.
DIGE, MALDI-TOF/TOF
(plasma, synovial fluid)
Complement 3c, apolipoprotein AII,
vitamin D binding protein
Outcome prediction (joint inflammation spread) [60] Gibson et al.
Dilated
cardiomyopathy
High density (37,000)
autoantigen array (plasma)
Low-density lipoprotein receptor
related protein-associated protein-1*,
phospatidic acid phosphatase-2*,
K+ channel-interacting
proteins*, tubulin alpha*
Screening/diagnostic (cardiac muscle disease) [72] Lueking et al.
[93] Lueking et al.
SDS-PAGE, immunoblot
(sera)
Myosin long chain , tropomysin, heat
shock protein 60, alpha-myosin,
beta-myosin heavy chain
Diagnostic (cardiac muscle disease) [85] Latif et al.
1049JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

risk factor [41,43]. The body's normal repair mechanisms are
unable to compensate for the loss of cartilage tissue, resulting
in radiological changes in the affected joint [44]. JIA targets the
lining of the joint, known as the synovial membrane, causing
inflammation or synovitis [45]. The resultant hyperplastic
tissue is composed of a collection of activated mononuclear
cells, and macrophages and fibroblasts that produce an
inflammatory soup, containing multiple inflammatory cyto-
kines, and matrix degradative enzymes, some or all of whom
contribute to joint damage. Historically, investigators have
examined single or families of inflammatory mediators in an
attempt to understand the mechanism of disease in JIA.
However while there are some insights, it is clear that the
most likely mechanism is a complex interaction between
numerous inflammatory and repair mediators.
3.2. Current issues in the management and treatment of JIA
In persistent oligoarticular disease only four or less joints are ever
involved. Here the disease is managed with intra-articular steroid
injections [46]. Some respond well to treatment but for others the
synovitis recurs requiring repeated joint injections taken under
general anaesthetic. For some, considerable growth deformity
and joint damage can be the outcome in a limited number of
joints. Currently we have no way of reliably identifying children
who will not respond to intra-articular treatments resulting in
numerous general anaesthetics for joint injections.
Oligoarticular JIA accounts for about 60% of cases at time of
diagnosis. However in around 25% of these children, the
disease will spread to multiple joints —extended oligoarti-
cular JIA sometime after 6 months from diagnosis. Often
spread occurs several years after diagnosis. These children
require systemic disease modifying treatment, frequently
with disappointing results, possibly because unlike children
with polyarticular disease the disease is well established by
the time drugs such as methotrexate are instigated [47].Ifwe
could identify, early, those children for whom the disease will
spread we may be able to obtain a better response to
treatment. Children with more than four joints affected
within six months of onset are defined as polyarticular JIA.
These children require disease modifying treatment virtually
from the outset. However for 30–40% of these children,
despite escalating doses of methotrexate the response is
poor and they will ultimately require effective but very
expensive biological therapies [42]. Again there is currently
no reliable way of identifying these poor responders early,
thus years can be wasted and joints irreparably damaged
before effective treatments are instigated.
Fig. 1 –Experimental workflow illustrating large scale serum proteome analysis methods. The serum was pooled from patients
and depleted from the top 14 abundant proteins and further separated with SDS-PAGE followed by reversed-phase liquid
chromatography (RPLC)–tandem mass spectrometry with a linear ion trap. This strategy enabled the isolation of putative
biomarkers which reliably discriminate Behcet's patients. The proteins identified related to the immune system defence
mechanisms and lipid metabolism.
1050 JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

Current clinical, laboratory or radiological parameters can
not accurately predict disease extension. In approximately
25% of children with oligoarticular JIA, over time the disease
will spread to involve many joints, evolving to extended
oligoarticular disease [48]. Extended oligo JIA is much more
difficult to treat due its characteristic resistance to first-line
therapies [49]. Predictive tests that forecast disease extension
could allow subsequent treatment decisions to be made in a
preventative instead of a reactive manner. It is pertinent to
define more sensitive markers for determining the risk of
unremitting inflammatory arthritis in JIA.
3.3. Proteomic approaches in juvenile arthritides
Although proteomic investigations of arthritic conditions focus
mainly on adult rheumatoid arthritis, there is a growing interest
in using this platform to investigate pathology specific to juvenile
forms of the disease. Various gel-based and gel-free separation
platforms have been coupled with modes of mass spectrometry
to investigate drug turnover and response, autoantigens, immune
complexes and inflammatory protein profiles.
A number of studies highlight the significance of acute phase
proteins such as haptoglobin and serum amyloid A and other
relatively high abundance proteins such as transferrin and
ceruloplasmin in plasma, synovial fluid and urine [50–52].Ithas
been suggested that transient infiltration of negative APP plasma
proteins may partially explain the relapse–remission cycles
characteristic of JIA and other forms of inflammatory arthritis
[53]. In addition, serum amyloid A was identified amongst 23
SELDI-TOF spectral peaks, from sera of systemic JIA patients,
which were proposed as potential biomarkers of active disease
and identified responders to biologic therapy directed at the IL-6
receptor [54]. Investigators have used liquid chromatography tied
with tandem mass spectrometry to gauge the response, turnover
and clearance of anti-inflammatory drugs such as anticorticos-
teroid and methotrexate in plasma and urine [55,56].Inthis
regard it is particularly useful to use high sensitivity proteomic
platforms to titre therapeutic doses for juvenile patients,
especially in clinical trials where toxicity concerns need to be
addressed alongside therapeutic efficacy.
Several studies have exploited the avidity of binding
between antibody and corresponding antigens to study
autoantibodies, their targets and immune complexes in
the context of juvenile disease. Complement components,
immunoglobulins and fibrinogen have been shown to co-
localise in swollen joint tissues and blood from juvenile
patients as circulating immune complexes [57,58]. Low et al.
enriched for these immune complexes from sera by
Proceptor
TM
affinity chromatography prior to 2-dimensional
gel electrophoresis and identified 28 active disease associ-
ated proteins including glyceraldehyde-3-phosphate dehy-
drogenase, serotransferrin, and α-1-anti-trypsin. In a similar
approach, Zhao et al used C1q protein to capture immune
complexes from plasma, separated these by liquid chroma-
tography and later ascertained that citrullinated fibrinogen
was absent from juvenile patients. These complexes are
thought to contribute to synovitis, inflammation of the
membranes surrounding the synovial joint, which is one of
the earliest clinical signs in the evolution of arthritic
disease.
An initial 2-DE gel study carried out at Queen's University
Belfast, UK uncovered fragments of extracellular matrix and
T-cell receptor proteins in the joint (synovial) fluid from
juvenile patients, evidence of the degradative proteolytic
environs of the recurrently inflamed knee [50]. The synovial
fluid proteome has a dynamic range of protein concentration
which reflects its derivation from plasma, but specific addi-
tions are made at the site of joint disease [59]. It is this
difference between blood and joint fluid samples which can be
exploited to isolate and identify those proteins most likely to
be involved at the site of joint inflammation.
Two-dimensional gel electrophoresis is an established plat-
form which facilitates the analysis of complex protein mixtures.
O'Farrell was first to introduce the technique by resolving proteins
to individual isoelectric point and molecular weight coordinates
[61]. In theory thousands of proteins can be visualised at once,
giving a unique qualitative ‘map’or ‘fingerprint’of changes
between given samples. Though developments, such as standar-
dised immobilised pH gradients, have led to improvement in
inter-run consistency, deficiencies in sensitivity and spot match-
ing necessitated further adaptation using fluorescent stains.
Comparison between large groups of conventionally (silver or
Coomassie) stained gels is complicated by spot to spot warping,
caused by variations in gel casting, electric and pH fields and
thermal fluctuations during electrophoresis. Gel heterogeneity
makes it difficult to distinguish with confidence between varia-
tions in the technique and those of genuine induced biological
change, such as in disease states [62]. Difference in-gel electro-
phoresis (DIGE) addresses a number of these issues in that 2–3
samples can be subjected to exactly the same running conditions
within a single gel. Unlu et al. developed DIGE to allow a more
direct and reproducible comparison between protein samples,
differentiated by prelabelling with spectrally resolvable fluores-
cent Cyanine dyes [63]. The Cy dyes are charge matched and have
similar molecular weights (0.5 kDa), so result in only slight gel
shifts. The Cy dyes are based on extended organic ring structures
and hence are highly hydrophobic. Concerns with protein pre-
cipitation prior to electrophoresis have been surmounted by using
a‘minimal labelling’strategy, whereby binding is limited to only
1–2% of lysine residues available within a sample [64].
Excitation of each fluor allows the creation of a digital
image of each individually labelled sample. These dyes give
additional validity to the two-dimensional technique in the
form of higher sensitivity, wider dynamic range and linearity
of detection. Detection limits of 0.025 ng are possible, with a
dynamic range around five orders of magnitude. One of the
strongest features of the technique, however, is the ability to
include an internal pooled standard which is loaded on all gels
within an experiment [65]. The internal standard permits the
linking of all gels in an experiment, thus offering more reliable
and intuitive software assisted comparisons. The accuracy of
protein quantification between samples is increased dramat-
ically and much smaller changes in protein expression can be
studied with greater confidence. Evaluations of DIGE alongside
traditional and more recent proteomic methods using isotope-
coded or isobaric tags (cICAT and iTRAQ), reveal that it remains
competitive in sensitivity and can be used with confidence as a
platform for biomarker discovery [64,66].
In a more recent study by the authors (DG, MR), the synovial
fluid (SF) proteome was been analysed by DIGE to reveal SF
1051JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

Fig. 2 –Proteomic strategy to discover biomarkers which stratify patients by clinical outcome. Illustration of an investigation focused on identifying juvenile arthritis patients at
risk from disease extension to joints not involved at diagnosis (T=0 months). A. Importantly samples of synovial (joint) fluid, plasma and synovial membrane are removed at
diagnosis, prior to any treatment. The patients suffering from the adverse outcome (green), defined by preset criteria, are identified as the longitudinal study progresses.
B. Difference in-gel electrophoresis is used to simultaneously separate the proteins within clinical samples and differentiate patients in remission (purple) from those with
extension of inflammation (green). Multivariate bioinformatics methods such as principal component analysis and hierarchical cluster analysis are used to isolate protein
pools which distinguish an outcome subgroup. C. Putative biomarker candidates are digested from gel spot cores and identified by matrix-assisted laser desorption ionisation
(MALDI-TOF) mass spectrometry (i). Further validation of protein expression is provided by specific antibody based methodologies such as immunhistochemistry of synovial
membrane biopsies (ii) and Western immunoblotting of individual patient synovial fluids (iii). Protein samples in lanes 4, 5 and 6 correspond to patients exhibiting disease
extension (green).
1052 JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

protein expression patterns to identify reliable predictors of JIA
subgroup and track disease course in detail [60]. The rationale
and methods employed within this study are illustrated in
Fig. 1. Multivariate analyses were used to isolate a panel of 40
proteins which distinguish patients who suffered spread of
inflammation to previously unaffected joints. Proteins were
identified using MALDI-TOF mass spectrometry and expres-
sion can be verified in SF and synovial membranes by western
immunoblotting and immunohistochemistry. This strategy is
illustrated in Fig. 2.
Juvenile patients in whom inflammation extended, dis-
played a general trend towards reduced transferrin, haptoglo-
bin and complement C3c relative to other patients. These host-
response proteins could be involved in the controlling the
spread of inflammation to previously unaffected joints. Hapto-
globin has been shown to be involved in angiogenesis, tissue
remodeling and cell migration [67]. In addition transferrin has
also been shown to promote angiogenesis, whereas decreased
levels of complement components may protect breast cancer
cells from complement mediated immune surveillance [68,69].
Complement component C3c is increased in SF in adult arthri-
tis patients and correlates to levels of polymorphonuclear cells
in the SF. Therefore C3c in SF may be indicative of inflamma-
tory activity in the SM [70].
Overall, proteomics presents a powerful means to identify
those biomarker profiles which predict evolution into a particular
disease outcome. It is anticipated that by identifying autoimmune
disease patients susceptible to specific outcomes earlier, one
could instigate more effective therapies to prevent joint and peri-
articular damage.
4. Dilated cardiomyopathy
4.1. Autoantibody detection
Autoimmune diseases are characterized by the presence of
low and high-affinity autoantibodies. Although the patho-
genic role for most of the autoantibodies in various autoim-
mune diseases is not clear, the specificity and pathogenicity
of the autoimmune response may present an important tool
for diagnosis, classification and prognosis. Additionally,
profiling the autoantibody repertoire may elucidate the
pathophysiology of autoimmunity, enabling novel treat-
ments such as antigen-tolerising therapy. It is known that a
humoral immune response is generated to disease associated
antigens during the development of autoimmune diseases.
This response seems to appear months or years before the
clinical diagnosis of the disease, rendering serum autoanti-
body detection suitable for early-stage diagnosis of autoim-
mune diseases. Protein arrays have been shown to have
applications in profiling the antibody repertoire in healthy
individuals and subjects with autoimmune diseases [71].We
have previously demonstrated the application of this ap-
proach in autoimmune dilated cardiomyopathy (DCM) [72]
and alopecia areata [73]. A number of reports have highlight-
ed the potential of this approach for the generation of
biomarkers in cancer detection [74]. Protein arrays offer a
number of advantages, primarily that tens of thousands of
proteins can be screened in high throughput and since only
femtomole of proteins are required per spot, the technique is
very sensitive.
4.2. Pathology of dilated cardiomyopathy (DCM)
Dilated cardiomyopathy (DCM) is a myocardial disease
characterized by progressive depression of myocardial
contractile function and by ventricular dilatation. DCM is a
pertinent cause of heart failure and a common indication
for heart transplantation. Thirty percent of DCM patients
belong to the inherited genetic form, and the rest may be
idiopathic, viral, autoimmune or immune-mediated in
association with a viral infection. Despite recent improve-
ments in therapy, the incidence of DCM is 1/20,000 per
annum and mortality remains high [75].Disturbancesin
humoral and cellular immunity have been described in
cases of myocarditis and DCM [75–77]. Inflammatory and
autoimmune mechanisms play an important role in the
pathogenesis of DCM. At least, one-third of DCM patients
reveal IgG cardiac-specific antibodies [78,79]. It has been
shown, that the injection of blood lymphocytes from DCM
patients containing human autoantibodies induces early
stage of heart failure in mice [80]. Before we performed the
profiling of plasma from CM subjects, it was known that the
extraction of circulating autoantibodies by immunoadsorp-
tion from blood of patients with DCM leads to an improve-
ment in cardiac function indicating an active role of these
autoantibodies in the pathogenesis of DCM. Apart from
simply reflecting an inflammatory response to myocyte
necrosis, this indicated an epiphenomenon, such that
antibodies could initiate the disease process or contribute
to the progression of myocardial contractile malfunction. In
particular, autoantibodies against cellular cardiac proteins,
like G-protein coupled receptors e.g. β
1
-adrenergic receptors
and muscarinergic receptors (M2) have been identified
[81,82]. Antibodies against membrane components and
mitochondrial proteins, especially against the ADP/ATP-
carrier protein of the inner mitochondrial membrane had
been associated with DCM [83]. It has been shown that
autoantibodies against the ADP/ATP-carrier protein cause a
reduction in cardiac function by enhancing the calcium
channel current, which leads to calcium overload [84].Also,
autoantibodies against myosin light-chain-1, tropomyosin,
HSP-60 and α-andβ-myosin heavy chain-isoforms have
been observed [85]. Analysis of IgG subclasses against
alpha- and beta-myosin heavy chain protein indicated
increased levels of IgG3 in DCM patients [86].Itwas
demonstrated that immuno adsorption removes functional
active negative ionotropic antibodies from the plasma of
patients with DCM and these antibodies belong to the IgG
subclass 3 [87].
Past experimental studies have demonstrated the direct
involvement of autoimmunity in the pathogenesis of DCM.
Mice deficient in the programmed cell death-1 (PD-1) immu-
noinhibitory coreceptor develop DCM [88]. Sera of these mice
contained high-titer autoantibodies against a heart-specific
30-kD protein, which was shown to be cardiac troponin I. An
experimental animal model with rats sensitized against the
β
1
-receptor, gave direct evidence of β
1
-receptor antibodies as a
cause of dilated cardiomyopathy [89].
1053JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

4.3. Proteomic study of dilated cardiomyopathy
In a recent study, carried out at University College Dublin,
Ireland (by authors DC and JO'B) the autoantibody repertoire of
plasma from DCM patients was profiled against a human
protein array consisting of 37,200 redundant, recombinant
human proteins comprising over 10,000 different human
proteins. We performed qualitative and quantitative valida-
tion of these putative autoantigens on protein microarrays to
identify novel putative DCM specific autoantigens. In addition
to analyzing to the whole IgG autoantibody repertoire, we
profiled the binding of the IgG3 subclass antibodies. By
combining the screening of a protein expression library with
protein microarray technology, we have detected twenty-six
proteins identified by the IgG antibody repertoire and six
proteins bound by the IgG3 subclass. Several of these
autoantibodies in plasma of DCM patients, such as the
autoantibody against the Kv channel-interacting protein, are
associated with heart failure.
High density protein arrays are robotically generated from
a human brain cDNA expression library (hEx1) and can be
arrayed onto a variety of surfaces including nitrocellulose or
PVDF membranes [90–92]. The protein chip layout generally
contains disease associated proteins, a panel of autoimmune-
related antigens and also a panel of control antigens for
normalisation purposes. The generation and screening of
protein chips have been described in detail by Lueking et al.
[93]. Briefly, the protein microarrays are blocked with BSA
prior to overnight incubation with diluted serum. A mouse
anti-human antibody followed by a cyanine dye labelled anti-
mouse antibody are used to detect any autoantibodies bound
to the antigen array. The arrays are then scanned using a
Genepix 4000B confocal microarray reader and associated
software.
In earlier studies, the data sets were generated from 10
DCM patient plasma screenings were compared to 10 age- and
sex-matched control plasmas from persons without heart
diseases, and to background incubations with anti-human-IgG
and anti-human-IgG3 [72]. We also identified a large number
of protein antigens in plasma from both DCM patients and
controls, the natural autoantibody (NAA) profile. IgG anti-
bodies against cytoskeletal proteins such as tubulin, its
interacting protein stathmin and myosin were found in 80%
of all samples (disease and controls). The presence of such
natural autoantibodies is well recognized and they have been
found in all individuals. We determined such NAAs to
ascertain the quality of the plasma used. Here, we used
skeletal myosin as a marker to monitor the plasma quality and
identified it in 88% of the plasma samples in this study.
From these data, proteins exclusively detected by anti-
bodies in DCM patient plasma were determined and their
potential role was analysed. For example, the gene coding for
LRPAP1 was identified and has been shown to contain a silent
polymorphism among females with coronary artery disease
and an additional variation in intron 1 of the LRPAP1 gene
could contribute to the risk of developing myocardial infarc-
tion. Another protein identified, PAP2, has been shown to be
inhibited by Mg
2+
and was localised to the sarcolemmal
membrane. PAP2 was considered an important enzyme, as it
has been shown to modulate the sarcolemmal levels of
phosphatidic acid and 1,2 diacylglycerol [94]. The importance
of phosphatidic acid in heart function is evident from its
ability to stimulate sarcolemmal Ca
2+
-related transport sys-
tems, and 1,2 diacylglycerol has been shown to activate PKC
isozymes which phosphorylate several myocardial proteins
including ion channels [95]. This may be important in relation
to DCM, where a change in intracellular Ca
2+
homeostase has
also been observed in other studies [96,97].
Mammalian voltage-gated potassium K+ channels are
assemblies of pore-forming alpha-subunits and modulating
beta-subunits. Kv4 alpha-subunits in the heart and central
nervous system require recently identified beta-subunits of
the neuronal calcium sensing protein family called K+
channel-interacting proteins (KChIPs). KChIP1 belongs to
four Kv channel-interacting proteins and KChIP1, 3 and 4 are
predominantly expressed in brain whereas KChIP2 is predom-
inantly in heart. Sequence comparison between the identified
brain specific protein KChIP1 and the heart-specific KChIP2
shows a homologous region of about 107 aminoacids with a
homology of 71% in this region, suggesting potentially a
common epitope or epitopes were recognized. It has been
shown that KChIP2 (−/−) knock-out mice were highly suscep-
tible to cardiac arrhythmias and were characterized by a
complete absence of transient outward potassium current.
Tubulin alpha was also identified using anti-human-IgG3
secondary antibodies. Tubulin alpha acts as a subunit of the
microtubuli, which are important for cell structure, the
intracellular transporting system and cell division. Tubulin
was already known to be an autoantigen in other autoimmune
diseases, such as systemic lupus erythematosus [98]. Anti-
bodies against tubulin have also previously been identified in
the plasma of healthy individuals. However, in this study we
have identified tubulin as a putative IgG3 specific autoantigen
in DCM patients.
Although several antigens with a closer connection to heart
failure were identified in the DCM study, the biological role of
these antibodies and their corresponding antigens was
unclear. It remains to be investigated if the antibodies
identified in plasma cause or have a strong influence on this
disease. For example, it has been shown that the extraction of
circulating antibodies (subtype IgG3) from the blood of
patients with DCM by immunoadsorption has lead to an
improvement in cardiac function, indicating an active role of
these antibodies in the pathogenesis of DCM [99]. However, it
cannot be ruled out that some of these circulating antibodies
in plasma may be as a result of other factors such as a result of
cross-reactivity to other proteins or as a secondary product of
aberrant pathways or as a result of processes such as
apoptosis without direct influence on the disease.
Hence, we refer to profiling the antibody repertoire in
plasma or serum. The potential of this approach is that it may
identify a disease associated antibody profile which may have
the potential to improve the diagnosis of disease, where
circulating antibodies play a role, or to potentially improve the
sub-typing or perhaps even assist in the analysis of the
progression or prognosis of such disease. The proteins
identified could potentially be candidates for further charac-
terization and validation as potential diagnostic markers.
The advantage of developing an autoantibody based test is
that the proteins these antibodies bind have appealing
1054 JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

features as biomarkers for a number of reasons. Firstly, they
can be detected at 1:100 dilution of serum (or up to 1:500 or in
some cases 1:1000) so the assay is compatible with the ELISA
format. The biochemical properties of antibodies are well
understood and there are many available reagents for their
detection, simplifying assay development. Secondly, when
profiling the autoantibody repertoire, there is no requirement
for sandwich ELISA development as the autoantibody is
detected with the same reporter antibody in all cases (the
human IgG binds the protein and the secondary antibody is
anti-human IgG). Finally, the antibody response is long lasting
and in a number of disorders, their presence predates the
onset of clinical symptoms. An additional asset of autoanti-
body arrays is that the assay format could be used with minor
adaptation for rapid testing, or as point of care devices. One
possible disadvantage of profiling the autoantibody repertoire
against high content protein arrays is that as the proteins are
generally denatured on the array, continuous (linear) epitopes
are detected. Discontinuous epitopes are not detected. For
potential diagnostic applications, this is not such a significant
disadvantage, as the proteins used in diagnostic assays, such
as in ELISA formats are also denatured as they are adsorbed to
the microtiter plate surface and stored dry.
However, for such an application, the proteins identified in
this type of analysis would need to be further characterized
and functionally and clinically validated and would be
required to be sufficiently disease associated to make them
relevant for diagnostic applications. A further application of
the type of study performed here is that the antigens
identified may give insights into the causation of the disease
or the mechanism of the progression of such diseases. To
improve the power of this approach, in common with the
other approaches and studies discussed in this report, larger
numbers of well-characterized diseased and control patients
would need to be performed. Moreover, addition of protein
expressing clones deriving from a heart cDNA expression
library, or by directed subcloning of specific proteins, to the
present expression library may lead to improved results.
4.4. Protein sequence identification and characterisation
The identified protein expressing cDNA clones can be se-
quenced to confirm the identity of the protein and reading
frame. In most cases, this leads to a direct identification of the
human protein in the array by sequence homologyto a database
entry (typically based on a cutoff e-value of 1e
−30
). In other cases,
the best homology found is to an entry derived from some other
organism, though usually the score is only marginally better
than a similar match to an entry derived from a human gene. In
some cases, the sequence has no significant homology to any
database entry, and is therefore likely to represent a novel
protein.After verification, by repeat subcloning and sequencing,
further characterisation would be required, such as obtaining
further sequence data by extended reads past the end of the
initial sequence. Then protein domains in the sequence are
determined using Pfam1 (http://pfam.sanger.ac.uk/) and Inter-
Pro2 (http://www.ebi.ac.uk/interpro/). Both of these tools allow
the identification of known protein domains with particular
functions in the protein sequence which may shed some light
on the role or structure of the protein itself.
In cases where a homology to an animal protein is found, this
can help to at least identify the family of the human protein.
Where the homolog is a primate better inferences can usually be
made, but in some cases, particularly mouse, the matched protein
can often be very well characterised. Finally, where a human
protein is matched it is still sometimes necessary to research the
protein to identify a plausible role for it in the progression of
the disease. We will approach this using OMIM (http://www.ncbi.
nlm.nih.gov/sites/entrez?db=omim) and GeneCards (http://www.
genecards.org/). The proteins can be expressed and purified for
further analysis and validation.
5. Bioinformatics and statistical analysis
The published autoimmune studies of dilated cardiomyopa-
thy [72] and alopecia areata [73] present examples of the types
of bioinformatics analysis required to interpret results from
large proteomic datasets. Custom written perl scripts are used
to assign and collate protein identifiers and to count the
proportion of positive subjects in each group for every protein
on arrays. Fisher's exact test is used to measure the statistical
significance of proportions of positive case and control
subjects. Further analysis involves clustering the protein
profile of each experimental subject according to their
similarity to each other, and the R project is used to perform
both hierarchical and k-means based clustering of the data.
Similarly, the proteins are clustered according to which of the
subjects has shown a positive result for it. The sensitivity and
specificity of the candidate biomarker(s) are calculated, in
addition to the receiver operator characteristic (ROC) curves
and areas under the curves.
Since as with the other autoimmune diseases described in
this review, the purpose of these screening experiments
against high content protein arrays is to generate a diagnostic
marker or a panel of markers is to identify patients with a
condition or a more severe version of that condition and to
distinguish them from those without the condition or to aid in
the early detection of disease. The markers under examina-
tion in this application are continuous, though part of the
analytic technique is to identify appropriate cut-points for
these markers. The development of diagnostic screening
assays and the further validation of these biomarkers in the
clinical setting on independent patient cohorts is a crucial part
of the translation of such assays into the clinic for improved
patient care. Different statistical approaches are used in the
initial discovery phase to identify candidate markers (because
of the large number of markers relative to the number of
patients available). In the phase involving validation of
already identified candidate biomarkers, the calculation of
sample sizes for the validations is based on classical
approaches [100]. The most common metrics used to evaluate
diagnostic ability are sensitivity and specificity which are
applicable to binary tests when a test result is either just
positive (indicating disease) or negative (indicating normal).
Sensitivity is the ability of the test to identify cases with disease
and is defined as the proportion of positive tests in those
persons with disease. Specificity is the ability of the test to
identify normals and is the proportion of negative tests in
persons who are disease-free. When dealing with a continuous
1055JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

Fig. 3 –Impact on autoimmune disease management of reliable predictive biomarkers versus traditional, reactive clinical decisions. Juvenile arthritis is used to illustrate how
autoimmune disease activity could be minimized by: (i) proactive screening to identify autoantibodies in the systemic circulation which preempt the arrival of clinical
symptoms, (ii) an earlier reliable diagnosis, (iii) logical stratification of patients such that subgroups at risk of adverse outcomes are clearly identified and (iv) guidance on
therapeutic benefit. In this idealized scenario, molecular biomarkers could diminish the extent of joint inflammation, maintain remission from symptoms which can impact
mobility and ultimately reduce the long term damage to the joint.
1056 JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

marker, different threshold cut-points can be used to define a
‘positive’test result, giving rise to different values forsensitivity
or specificity.
The receiver operating characteristic curve (ROC) is used to
generalise the concepts of sensitivity and specificity and is
essentially a plot of sensitivity versus 1 —specificity as the cut-
point ranges over the possible numerical values of the marker.
The ROC characterises the ability of the biomarker as a whole to
distinguish between those who have and do not have a disease, or
between severity groups. There are a number of measures of a
marker's diagnostic ability based on the ROC. The area under the
curve (AUC) is the most commonly used [101,102]. The AUC
ranges from 50% meaning no diagnostic ability to 100% which
implies perfect discrimination between diseased and non-
diseased. For a single marker the analysis is fairly straightforward
though specialised software is required to perform formal
statistical analyses of ROC data [103]. Logistic regression is the
method of choice to build up a model based on a number of
markers, though CART (classification and regression tress) and
neural networks have also been used as alternatives. In SAS, the
logistic regression procedure provides the AUC for the linear
combination of markers included in the model [104]. Marker
combinations can also be investigated using multivariate techni-
ques such as principal component analysis. This would identify
optimum combinations through a principal component score
which would be subsequently analysed with logistic regression.
This approach can reduce the biases that arise when dealing with
highly correlated markers in a single logistic model [105].
Though diagnostic biomarkers and biomarker combinations
will be evaluated using the AUC, there may be different sets that
have different independent diagnostic abilities. If there is no
statistically significant difference between two sets of biomarkers,
the set with the fewest basic marker components will be chosen
as preferable from the clinical point of view even if it has a (non-
significantly) lower diagnostic ability.
6. Discussion and conclusions
From the different diseases presented, it is apparent that a
number of challenges still confront the clinician charged with
the care of autoimmune patients: (i) to provide a clear diagnosis
as early as possible, (ii) to preventdisease progression by reliably
stratifying at-risk patients with outcome predictors, and (iii) to
instigate timely drug intervention, monitor therapeutic re-
sponse and titre dosage over the disease course. The likely
role that the biomarkers highlighted in this review could play
have been summarised in Table 1, whereas the theoretical
impact on disease management by informing clinical decisions
via reliable biomarkers is emphasized in Fig. 3.Withregardto
the first of these issues, a widevariety of overlappingsymptoms
have been attributed to analogous autoimmune conditions,
confounding timely diagnoses. The ability to screen a patient
with multiple biomarkers associated with several autoimmune
diseases could confer a more robust diagnosis to clinicians. As
previously stated it is most likely that the complex nature of
autoimmune disease is more likely reflected by biomarker
panels which indicate multiple, possibly overlapping, pathways
of molecular pathology. It is also therefore likely that non-
disease specific tests like those illustrated for autoantibodies in
DCM screening could be used to diagnose or prognosticate a
wide variety of disorders where autoimmune pathology is
suspected.
When considering the second of the core issues in
clinical management, there is considerable heterogeneity
in the disease course and clinical outcome among patients
with the same autoimmune condition. The ability to predict
disease course would be a powerful tool in the limitation of
adverse outcome such as disease extension in JIA, as it may
flag such patients for more effective/aggressive treatment
strategies at an earlier stage in the disease process than is
currently possible. The authors consider the collection of
samples at the earliest stage in the disease course an
important feature of study design and vital in the discovery
of biomarkers which not only ‘describe’the initial disease
processes, but may also act as sentinels to predict subse-
quent disease outcome. This is of particular importance
when the opportunity arises to analyse collated samples
from a longitudinal study following patient outcomes over a
number years. Furthermore joint proteome records could
help predict disease spread and organ damage. Within such
a study it is also important that patient subgroups are well
defined and that samples are in pristine condition so that
degradation effects do not confound the discovery reliable
biomarkers.
Of relevance to the final issue, pressure is also mounting on
clinicians to rationalise the use of biologic drugs on a cost–
benefit basis. Even though these disease modifying anti-
rheumatic drugs (DMARDs) can be extremely effective in
alleviating the symptoms of autoimmune disease, they
present an immense financial burden to health trusts and
insurance companies. By cataloguing the protein profiles and
identities which best describe the joint status as disease
progresses, it may also be possible to monitor therapeutic
response over time. Proteomic technology could also provide
more complete evidence of treatment effects and safety in
clinical trials than is currently possible. Considering for
example the heterogeneity of patient outcome in JIA, thera-
peutic suppression of synovial joint inflammation, while
neglecting the systemic components of chronic disease, may
not represent the best possible way to manage non-responsive
disease. Longer lasting remission could be possible by taking
systemic sentinels into account.
It is envisioned that a serum, plasma or proximal fluid
sample will be used to predict the subsequent course of
disease for that patient, thus enabling early appropriate
intervention. The analysis of biomarker profiles early in the
course of the disease could identify these patients and thus
avoid months of fruitless treatment, allow instigation of
more effective regimens earlier, and thus reduce adverse
outcomes. As with all translational research, a final properly
powered prospective study to assess the robustness of such
biomarkers to predict outcome is required. Cross validation
of putative biomarkers is a vital feature of the study design,
in agreement with published discovery strategies [106].
Independent cohorts of patients are required to validate
putative biomarkers. Furthermore other sample types may
be necessary along with orthogonal approaches to validate
findings and provide functional relevance to biomarker
candidates.
1057JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

Acknowledgements
This work has been supported by the Research and Development
Office Northern Ireland Grant RRG 8.42 and an Arthritis Research
Campaign UK Project Grant MP/18748 (to D.G. and M.R.). This work
was also partly funded by Project 126/2007 from the Portuguese
Ministry of Health (to JB, DP, LC, TC).
REFERENCES
[1] Hueber W, Robinson WH. Proteomic biomarkers for
autoimmune disease. Proteomics 2006;6:4100–5.
[2] Hershko AY, Naparstek Y. Autoimmunity in the era of
genomics and proteomics. Autoimmun Rev 2006;5:230–3.
[3] Villalta D, Tozzoli R, Tonutti E, Bizzaro N. The laboratory
approach to the diagnosis of autoimmune diseases: is it time
to change? Autoimmun Rev 2007;6:359–65.
[4] Kalayciyan A, Zouboulis C. An update on Behcet's disease.
J Eur Acad Dermatol Venereol 2007;21:1–10.
[5] Evereklioglu C. Managing the symptoms of Behcet's disease.
Expert Opin Pharmacother 2004;5:317–28.
[6] Hirohata S, Kikuchi H. Behcet's disease. Arthritis Res Ther
2003;5:139–46.
[7] Marshall SE. Behcet's disease. Best Pract Res Clin Rheumatol
2004;18:291–311.
[8] Onder M, Gurer MA. The multiple faces of Behcet's disease
and its aetiological factors. J Eur Acad Dermatol Venereol
2001;15:126–36.
[9] Crespo J, Vaz Patto J, Proenca R. Epidemiology of Behcet's
syndrome in Portugal. BD News 2003;4:1.
[10] Pay S, Simsek I, Erdem H, Dinc A. Immunopathogenesis of
Behcet's disease with special emphasize on the possible role
of antigen presenting cells. Rheumatol Int 2007;27:417–24.
[11] Zierhut M, Mizuki N, Ohno S, Inoko H, Gül A, Onoé K, et al.
Immunology and functional genomics of Behcet's disease.
Cell Mol Life Sci 2003;60:1903–22.
[12] Oliveira R, Banha J, Martins F, Paixao E, Pereira D, Barcelos F,
et al. Lymphocyte ceruloplasmin and Behcet's disease. Acta
Reumatol Port 2006;31:323–9.
[13] Niwa Y, Mizushima Y. Neutrophil-potentiating factors
released from stimulated lymphocytes; special reference to
the increase in neutrophil-potentiating factors from
streptococcus-stimulated lymphocytes of patients with
Behcet's disease. Clin Exp Immunol 1990;79:353–60.
[14] Yazici H, Fresko I. Behcet's disease and other
autoinflammatory conditions: what's in a name? Clin Exp
Rheumatol 2005;23:S1–2.
[15] Galeazzi M, Gasbarrini G, Ghirardello A, Grandemange S,
Hoffman HM, Manna R, et al. Autoinflammatory syn-
dromes. Clin Exp Rheumatol 2006;24:S79–85.
[16] Direskeneli H. Autoimmunity vs autoinflammation in
Behcet's disease: do we oversimplify a complex disorder?
Rheumatology (Oxford) 2006;45:1461–5.
[17] Gul A. Behcet's disease as an autoinflammatory disorder.
Curr Drug Targets Inflamm Allergy 2005;4:81–3.
[18] Verity DH, Wallace GR, Vaughan RW, Stanford MR. Behcet's
disease: from Hippocrates to the third millennium. Br J
Ophthalmol 2003;87:1175–83.
[19] Yamamoto JH, Fujino Y, Lin C, Nieda M, Juji T, Masuda K.
S-antigen specific T cell clones from a patient with
Behcet's disease. Br J Ophthalmol 1994;78:927–32.
[20] Tanaka T, Yamakawa N, Koike N, Suzuki J, Mizuno F, Usui M.
Behcet's disease and antibody titers to various heat-shock
protein 60s. Ocul Immunol Inflamm 1999;7:69–74.
[21] Mahesh SP, Li Z, Buggage R, Mor F, Cohen IR, Chew EY, et al.
Alpha tropomyosin as a self-antigen in patients with
Behcet's disease. Clin Exp Immunol 2005;140:368–75.
[22] Mor F, Weinberger A, Cohen IR. Identification of
alpha-tropomyosin as a target self-antigen in Behcet's
syndrome. Eur J Immunol 2002;32:356–65.
[23] Okunuki Y, Usui Y, Takeuchi M, Kezuka T, Hattori T, Masuko
K, et al. Proteomic surveillance of autoimmunity in Behcet's
disease with uveitis: selenium binding protein is a novel
autoantigen in Behcet's disease. Exp Eye Res 2007;84:823–31.
[24] Orem A, Cimsit G, Deger O, Vanizor B, Karahan SC.
Autoantibodies against oxidatively modified low-density
lipoprotein in patients with Behcet's disease. Dermatology
1999;198:243–6.
[25] Karasawa R, Ozaki S, Nishioka K, Kato T. Autoantibodies to
peroxiredoxin I and IV in patients with systemic
autoimmune diseases. Microbiol Immunol 2005;49:57–65.
[26] Lee KH, Chung HS, Kim HS, Oh SH, Ha MK, Baik JH, et al.
Human alpha-enolase from endothelial cells as a target
antigen of anti-endothelial cell antibody in Behcet's disease.
Arthritis Rheum 2003;48:2025–35.
[27] Mao L, Dong H, Yang P, Zhou H, Huang X, Lin X, et al. MALDI-
TOF/TOF-MS reveals elevated serum haptoglobin and
amyloid A in Behcet's disease. J Proteome Res 2008;7:4500–7.
[28] Gebreselassie D, Spiegel H, Vukmanovic S. Sampling of
major histocompatibility complex class I-associated
peptidome suggests relatively looser global association of
HLA-B*5101 with peptides. Hum Immunol 2006;67:894–906.
[29] Ames PR, Alves J, Murat I, Isenberg DA, Nourooz-Zadeh J.
Oxidative stress in systemic lupus erythematosus and allied
conditions with vascular involvement. Rheumatology
(Oxford) 1999;38:529–34.
[30] Banha J, Conrads TP, Hood BL, Sun M. Biomarkers for
Behçet's disease: in a quest for the holy grail using
proteomics tools; 2009. p. 40–2.
[31] Chaerkady R, Pandey A. Applications of proteomics to lab
diagnosis. Annu Rev Pathol Mech Dis 2008;3:485–98.
[32] Addona TA, Abbatiello SE, Schilling B, Skates SJ, Mani DR,
Bunk DM, et al. Multi-site assessment of the precision and
reproducibility of multiple reaction monitoring-based
measurements of proteins in plasma. Nat Biotechnol
2009;27:633–41.
[33] Conrads TP, Hood BL, Petricoin III EF, Liotta LA, Veenstra TD.
Cancer proteomics: many technologies, one goal. Expert Rev
Proteomics 2005;2:693–703.
[34] Patel V, Hood BL, Molinolo AA, Lee NH,Conrads TP, Braisted JC,
et al. Proteomic analysis of laser-captured
paraffin-embedded tissues: a molecular portrait of
head and neck cancer progression. Clin Cancer Res
2008;14:1002–14.
[35] Kirkpatrick DS, Gerber SA, Gygi SP. The absolute
quantification strategy: a general procedure for the
quantification of proteins and post-translational
modifications. Methods 2005;35:265–73.
[36] Ono M, Shitashige M, Honda K, Isobe T, Kuwabara H.
Label-free quantitative proteomics using large peptide
data sets generated by nanoflow liquid chromatography
and mass spectrometry. Mol Cell Proteomics 2006;5:1338–47.
[37] Barnidge DR, Goodmanson MK, Klee GG, Muddiman DC.
Absolute quantification of the model biomarker
prostate-specific antigen in serum by LC-MS/MS using
protein cleavage and isotope dilution mass spectrometry.
J Proteome Res 2004;3:644–52.
[38] Symmons DP, Jones M, Osborne J, Sills J, Southwood TR, Woo
P. Pediatric rheumatology in the United Kingdom: data from
the British Pediatric Rheumatology Group National
Diagnostic Register. J Rheumatol 1996;23:1975–80.
[39] Liem JJ, Rosenberg AM. Growth patterns in juvenile
rheumatoid arthritis. Clin Exp Rheumatol 2003;21:663–8.
1058 JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

[40] Petty RE, Southwood TR, Manners P, Baum J, Glass DN,
Goldenberg J, et al. International League of Associations
for Rheumatology classification of juvenile idiopathic
arthritis: second revision, Edmonton, 2001. J Rheumatol
2004;31:390–2.
[41] Flato B, Lien G, Smerdel A, Vinje O, Dale K, Johnston V, et al.
Prognostic factors in juvenile rheumatoid arthritis: a
case–control study revealing early predictors and outcome after
14.9 years. J Rheumatol 2003;30:386–93.
[42] Hashkes PJ, Laxer RM. Medical treatment of juvenile idiopathic
arthritis. JAMA 2005;294:1671–84.
[43] Flato B, Aasland A, Vinje O, Forre O. Outcome and
predictive factors in juvenile rheumatoid arthritis and
juvenile spondyloarthropathy. J Rheumatol 1998;25:366–75.
[44] Oen K, Reed M, Malleson PN, Cabral DA, Petty RE, Rosenberg
AM, et al. Radiologic outcome and its relationship to
functional disability in juvenile rheumatoid arthritis.
J Rheumatol 2003;30:832–40.
[45] de Kleer IM, Wedderburn LR, Taams LS, Patel A, Varsani H,
Klein M, et al. CD4+CD25bright regulatory T cells actively
regulate inflammation in the joints of patients with the
remitting form of juvenile idiopathic arthritis. J Immunol
2004;172:6435–43.
[46] Khalkhali-Ellis Z, Seftor EA, Nieva DR, Seftor RE, Samaha HA,
Bultman L, et al. Induction of invasive and degradative
phenotype in normal synovial fibroblasts exposed to
synovial fluid from patients with juvenile rheumatoid
arthritis: role of mononuclear cell population. J Rheumatol
1997;24:2451–60.
[47] Ravelli A. Toward an understanding of the long-term
outcome of juvenile idiopathic arthritis. Clin Exp Rheumatol
2004;22:271–5.
[48] Huemer C, Malleson PN, Cabral DA, Huemer M, Falger J,
Zidek T, et al. Patterns of joint involvement at onset
differentiate oligoarticular juvenile psoriatic arthritis from
pauciarticular juvenile rheumatoid arthritis. J Rheumatol
2002;29:1531–5.
[49] Woo P, Southwood TR, Prieur AM, Dore CJ, Grainger J, David J,
et al. Randomized, placebo-controlled, crossover trial of
low-dose oral methotrexate in children with extended
oligoarticular or systemic arthritis. Arthritis Rheum
2000;43:1849–57.
[50] Gibson DS, Blelock S, Brockbank S, Curry J, Healy A,
McAllister C, et al. Proteomic analysis of recurrent joint
inflammation in juvenile idiopathic arthritis. J Proteome Res
2006;5:1988–95.
[51] Suzuki M, Wiers K, Brooks EB, Greis KD, Haines K, Klein-
Gitelman MS, et al. Initial validation of a novel protein
biomarker panel for active pediatric lupus nephritis. Pediatr
Res 2009;65:530–6.
[52] Suzuki M, Ross GF, Wiers K, Nelson S, Bennett M, Passo MH,
et al. Identification of a urinary proteomic signature for
lupus nephritis in children. Pediatr Nephrol 2007;22:2047–57.
[53] Bresnihan B, Gogarty M, FitzGerald O, Dayer JM, Burger D.
Apolipoprotein A-I infiltration in rheumatoid arthritis
synovial tissue: a control mechanism of cytokine
production? Arthritis Res Ther 2004;6:R563–6.
[54] Miyamae T, Malehorn DE, Lemster B, Mori M, Imagawa T,
Yokota S, et al. Serum protein profile in systemic-onset
juvenile idiopathic arthritis differentiates response versus
nonresponse to therapy. Arthritis Res Ther 2005;7:R746–55.
[55] Kumar S, Singh RJ, Reed AM, Lteif AN. Cushing's syndrome
after intra-articular and intradermal administration of
triamcinolone acetonide in three pediatric patients.
Pediatrics 2004;113:1820–4.
[56] Dolezalova P, Krijt J, Chladek J, Nemcova D, Hoza J.
Adenosine and methotrexate polyglutamate concentrations
in patients with juvenile arthritis. Rheumatology (Oxford)
2005;44:74–9.
[57] Low JM, Chauhan AK, Gibson DS, Zhu M, Chen S, Rooney ME,
et al. Proteomic analysis of circulating immune complexes in
juvenile idiopathic arthritis reveals disease-associated
proteins. Proteomics —Clin Applic 2009;3:829–40.
[58] Zhao X, Okeke NL, Sharpe O, Batliwalla FM, Lee AT, Ho PP, et
al. Circulating immune complexes contain citrullinated
fibrinogen in rheumatoid arthritis. Arthritis Res Ther
2008;10:R94.
[59] Gibson DS, Rooney ME. The human synovial fluid proteome:
akey factor in the pathology of joint disease.
Proteomics —Clin Applic 2007;1:889–99.
[60] Gibson DS, Blelock S, Curry J, Finnegan S, Healy A, Scaife C, et
al. Comparative analysis of synovial fluid and plasma
proteomes in juvenile arthritis—proteomic patterns of joint
inflammation in early stage disease. J Proteomics
2009;72:656–76.
[61] O'Farrell PH. High resolution two-dimensional
electrophoresis of proteins. J Biol Chem 1975;
250(10):4007–21.
[62] Marouga R, David S, Hawkins E. The development of the
DIGE system: 2D fluorescence difference gel analysis
technology. Anal Bioanal Chem 2005;382(3):669–78.
[63] Unlu M, Morgan ME, Minden JS. Difference gel
electrophoresis: a single gel method for detecting changes
in protein extracts. Electrophoresis 1997;18(11):2071–7.
[64] Tonge R, Shaw J, Middleton B, Rowlinson R, Rayner S, Young
J, et al. Validation and development of fluorescence two-
dimensional differential gel electrophoresis proteomics
technology. Proteomics 2001;1(3):377–96.
[65] Alban A, David SO, Bjorkesten L, Andersson C, Sloge E, Lewis
S, et al. A novel experimental design for comparative two-
dimensional gel analysis: two-dimensional difference gel
electrophoresis incorporating a pooled internal standard.
Proteomics 2003;3(1):36–44.
[66] Wu WW, Wang G, Baek SJ, Shen RF. Comparative study of
three proteomic quantitative methods, DIGE, cICAT, and iTRAQ,
using 2D gel- or LC-MALDI TOF/TOF. J Proteome Res
2006;5(3):651–8.
[67] de Kleijn DP, Smeets MB, Kemmeren PP, Lim SK, Van
Middelaar BJ, Velema E, et al. Acute-phase protein hapto-
globin is a cell migration factor involved in arterial
restructuring. FASEB J 2002;16:1123–5.
[68] Carlevaro MF, Albini A, Ribatti D, Gentili C, Benelli R, Cermelli S,
et al. Transferrin promotes endothelial cell migration and
invasion: implication in cartilage neovascularization. J Cell Biol
1997;136:1375–84.
[69] Rozanov DV, Savinov AY, Golubkov VS, Postnova TI, Remacle
A, Tomlinson S, et al. Cellular membrane type-1 matrix
metalloproteinase (MT1-MMP) cleaves C3b, an essential
component of the complement system. J Biol Chem
2004;279:46551–7.
[70] Berkowicz A, Kappelgaard E, Petersen J, Nielsen H, Ingemann-
Hansen T, Halkjaer-Kristensen J, et al. Complement C3c and C3d
in plasma and synovial fluid in rheumatoid arthritis. Acta Pathol
Microbiol Immunol Scand C 1983;91:397–402.
[71] Cahill DJ. Protein and antibody arrays and their medical
applications. J Immunol Methods 2001;250:81–91.
[72] Horn S, Lueking A, Murphy D, Staudt A, Gutjahr C, Schulte
K, et al. Profiling humoral autoimmune repertoire of
dilated
cardiomyopathy (DCM) patients and development of a
disease-associated protein chip. Proteomics 2006;6:605–13.
[73] Lueking A, Huber O, Wirths C, Schulte K, Stieler KM,
Blume-Peytavi U, et al. Profiling of alopecia areata
autoantigens based on protein microarray technology.
Mol Cell Proteomics 2005;4:1382–90.
[74] Casiano CA, Mediavilla-Varela M, Tan EM. Tumor-associated
antigen arrays for the serological diagnosis of cancer. Mol
Cell Proteomics 2006;5:1745–59.
1059JOURNAL OF PROTEOMICS 73 (2010) 1045–1060

[75] Codd MB, Sugrue DD, Gersh BJ, Melton III LJ. Epidemiology of
idiopathic dilated and hypertrophic cardiomyopathy. A
population-based study in Olmsted County, Minnesota,
1975–1984. Circulation 1989;80:564–72.
[76] Staudt A, Staudt Y, Dorr M, Bohm M, Knebel F, Hummel A, et
al. Potential role of humoral immunity in cardiac
dysfunction of patients suffering from dilated
cardiomyopathy. J Am Coll Cardiol 2004;44:829–36.
[77] Barry WH. Mechanisms of immune-mediated myocyte
injury. Circulation 1994;89:2421–32.
[78] Warraich RS, Dunn MJ, Yacoub MH. Subclass specificity of
autoantibodies against myosin in patients with idiopathic
dilated cardiomyopathy: pro-inflammatory antibodies in
DCM patients. Biochem Biophys Res Commun
1999;259:255–61.
[79] Warraich RS, Noutsias M, Kazak I, Seeberg B, Dunn MJ,
Schultheiss HP, et al. Immunoglobulin G3 cardiac
myosin autoantibodies correlate with left ventricular
dysfunction in patients with dilated cardiomyopathy:
immunoglobulin G3 and clinical correlates. Am Heart J
2002;143:1076–84.
[80] Omerovic E, Bollano E, Andersson B, Kujacic V, Schulze W,
Hjalmarson A, et al. Induction of cardiomyopathy in severe
combined immunodeficiency mice by transfer of
lymphocytes from patients with idiopathic dilated
cardiomyopathy. Autoimmunity 2000;32:271–80.
[81] Magnusson Y, Wallukat G, Waagstein F, Hjalmarson A,
Hoebeke J. Autoimmunity in idiopathic dilated
cardiomyopathy. Characterization of antibodies against the
beta 1-adrenoceptor with positive chronotropic effect.
Circulation 1994;89:2760–7.
[82] Fu LX, Magnusson Y, Bergh CH, Liljeqvist JA, Waagstein F,
Hjalmarson A, et al. Localization of a functional
autoimmune epitope on the muscarinic acetylcholine
receptor-2 in patients with idiopathic dilated
cardiomyopathy. J Clin Invest 1993;91:1964–8.
[83] Schulze K, Becker BF, Schauer R, Schultheiss HP. Antibodies
to ADP-ATP carrier–an autoantigen in myocarditis and
dilated cardiomyopathy–impair cardiac function.
Circulation 1990;81:959–69.
[84] Schulze K, Becker BF, Schultheiss HP. Antibodies to the
ADP/ATP carrier, an autoantigen in myocarditis and
dilated cardiomyopathy, penetrate into myocardial cells
and disturb energy metabolism in vivo. Circ Res
1989;64:179–92.
[85] Latif N, Baker CS, Dunn MJ, Rose ML, Brady P, Yacoub MH.
Frequency and specificity of antiheart antibodies in patients
with dilated cardiomyopathy detected using SDS-PAGE and
western blotting. J Am Coll Cardiol 1993;22:1378–84.
[86] Caforio AL, Grazzini M, Mann JM, Keeling PJ, Bottazzo GF,
McKenna WJ, et al. Identification of alpha- and beta-cardiac
myosin heavy chain isoforms as major autoantigens in
dilated cardiomyopathy. Circulation 1992;85:1734–42.
[87] Staudt A, Bohm M, Knebel F, Grosse Y, Bischoff C, Hummel A,
et al. Potential role of autoantibodies belonging to the
immunoglobulin G-3 subclass in cardiac dysfunction among
patients with dilated cardiomyopathy. Circulation
2002;106:2448–53.
[88] Nishimura H, Okazaki T, Tanaka Y, Nakatani K, Hara M,
Matsumori A, et al. Autoimmune dilated cardiomyopathy
in PD-1 receptor-deficient mice. Science 2001;291:319–22.
[89] Limas CJ, Goldenberg IF, Limas C. Effect of antireceptor
antibodies in dilated cardiomyopathy on the cycling
of cardiac beta receptors. Am Heart J 1991;122:108–14.
[90] Bussow K, Cahill D, Nietfeld W, Bancroft D, Scherzinger E,
Lehrach H, et al. A method for global protein expression and
antibody screening on high-density filters of an arrayed
cDNA library. Nucleic Acids Res 1998;26:5007–8.
[91] Angenendt P, Glokler J, Konthur Z, Lehrach H, Cahill DJ. 3D
protein microarrays: performing multiplex immunoassays
on a single chip. Anal Chem 2003;75:4368–72.
[92] Angenendt P, Glokler J, Murphy D, Lehrach H, Cahill DJ.
Toward optimized antibody microarrays: a comparison of
current microarray support materials. Anal Biochem
2002;309:253–60.
[93] Lueking A, Possling A, Huber O, Beveridge A, Horn M,
Eickhoff H, et al. A nonredundant human protein chip for
antibody screening and serum profiling. Mol Cell Proteomics
2003;2:1342–9.
[94] Yu CH, Panagia V, Tappia PS, Liu SY, Takeda N, Dhalla NS.
Alterations of sarcolemmal phospholipase D and
phosphatidate phosphohydrolase in congestive heart
failure. Biochim Biophys Acta 2002;1584:65–72.
[95] Deng XF, Mulay S, Varma DR. Role of Ca(2+)-independent
PKC in alpha 1-adrenoceptor-mediated inotropic responses
of neonatal rat hearts. Am J Physiol 1997;273:H1113–8.
[96] O'Brien PJ, Gwathmey JK. Myocardial Ca(2+)- and
ATP-cycling imbalances in end-stage dilated and ischemic
cardiomyopathies. Cardiovasc Res 1995;30:394–404.
[97] Studer R, Reinecke H, Bilger J, Eschenhagen T, Böhm M,
Hasenfuss G, et al. Gene expression of the cardiac
Na(+)–Ca2+ exchanger in end-stage human heart failure.
Circ Res 1994;75:443–53.
[98] Pateraki E, Kaklamani E, Kaklamanis P, Portocalas R,
Aessopos A. Autoantibodies in systemic lupus
erythematosus and normal subjects. Clin Rheumatol
1986;5:338–45.
[99] Felix SB, Staudt A, DorffelWV, Stangl V, Merkel K, Pohl M, et al.
Hemodynamic effects of immunoadsorption and subsequent
immunoglobulin substitution in dilated cardiomyopathy:
three-month results from a randomized study. J Am Coll
Cardiol 2000;35:1590–8.
[100] Dignam J. Statistical issues in investigating prognostic
and predictive markers for DCIS. Paper presented at NCI
workshop on Ductal Carcinoma in Situ, San Francisco;
2007. Available at http://appliedresearch.cancer.gov/dcis/
workshop/dcis_dignam.html.
[101] Hanley JA, McNeil BJ. The meaning and use of the area under
a receiver operating characteristic (ROC) curve. Radiology
1982;143:29–36.
[102] Hanley JA, McNeil BJ. A method of comparing the areas
under receiver operating characteristic curves derived from
the same cases. Radiology 1983;148:839–43.
[103] Obuchowski NA. Sample size calculations in studies of test
accuracy. Stat Methods Med Res 1998;7:371–92.
[104] Mandraekar JN, Mandraekar SJ. Statistical methods in
diagnostic medicine using SAS software. Paper 211 30th SAS
User Group International Conference; 2005. Available at
http://www2.sas.com/proceedings/sugi30/211-30.pdf.
[105] Aguilera AM, Escabias M, Valderrama M. Using principal
components for estimating logistic regression with
high-dimensional multicollinear data. Comput Stat Data
Anal 2006;50:1905–24.
[106] Feng Z, Prentice R, Srivastava S. Research issues and
strategies for genomic and proteomic biomarker discovery
and validation: a statistical perspective. Pharmacogenomics
2004;5:709–19.
1060 JOURNAL OF PROTEOMICS 73 (2010) 1045–1060