Virchows Arch (2006) 448: 744–755
DOI 10.1007/s00428-006-0189-2
REVIEW ARTICLE
Manfred Dietel . Christine Sers
Personalized medicine and development of targeted therapies:
the upcoming challenge for diagnostic molecular pathology.
A review
Received: 4 July 2005 / Accepted: 20 September 2005 / Published online: 22 April 2006
# Springer-Verlag 2006
Abstract Due to continuous technical developments and
new insights into the high complexity of many diseases,
molecular pathology is a rapidly growing field gaining
center stage in the clinical management of tumors as well as
in the pharmaceutical development of new anti-cancer
drugs. The application of novel compounds in clinical trials
has revealed promising results; however, the current
diagnostic procedures available for determining which
patients will primarily benefit from rational tumor therapy
are insufficient. To read a patient’s tissue as “deeply” as
possible, in the future, gaining information on the
morphology and on genetic, proteomic, and epigenetic
alterations will be the upcoming task of surgical pathologists
experienced in molecular diagnostics to provide the
clinicians with information relevant for an individualized
medicine. Among the different high-throughput technologies,
DNA microarrays are now the first array approach
close to enter routine diagnostics. Technically advanced
and well-established microarray platforms can nowadays
be evaluated by distinct bioinformatic tools capable of
identifying both novel genes associated with disease
development and clusters of genes predicting clinical
outcome of an individual tumor. The automatic, highly
parallel analysis of proteins and complex proteins lysates
for early detection of cancers such as breast, prostate and
ovary as proteomic patterns in the serum also appears at the
horizon. In addition, an improved analysis of tumor
samples via antibody or reverse-phase protein arrays is
likely to provide the pathologist in the future with
information about activated oncogenic signaling pathways
and other cell functions, such as drug response or the
potential to metastasize. While expression microarrays and
proteomic analysis rely on relatively unstable material
incompatible with paraffin-embedded tissue samples, an
investigation of DNA methylation using specialized highthroughput
platforms has revealed the potential of being
used in future diagnostics. Each of these approaches on its
own might not suffice to extract all information required for
an efficient individualized diagnostics. Therefore, a “multiplex
approach” combining the different biological levels
DNA, RNA, and protein, may be necessary to functionally
classify malignant tumors. This appears to become a major
challenge for diagnostic pathologists.
Keywords Review . Molecular pathology .
Targeted therapy
Introduction
Standard diagnostic procedures for human tumors are
currently based on a combination of histopathology and
immunohistology closely connected with clinical data.
This strategy, in most cases, provides precise information
on dignity, tissue origin, tumor type, stage, and grade as
well as information on the completeness of surgical tumor
removal. Up to now, these data comprise the most relevant
information on a patient’s prognosis and are a rational basis
for therapy design. As the diversity of alterations in tumor
cells is not completely mirrored by tumor cell morphology,
pathologists and clinicians often observe that two patients
harboring the “same” type of tumor in a seemingly
“identical” stage show different clinical outcome with
respect to survival and, in particular, with respect to therapy
response. To improve this unsatisfactory situation, a
patient-specific disease prediction based on tissue examination
by pathologists would be extremely helpful.
During recent years, many molecular markers and
marker patterns (“signatures”) have been disclosed
which, at least, partially predict prognosis and therapeutic
M. Dietel . C. Sers
Institute of Pathology, Charité,
University Hospital Berlin,
Humboldt-University,
Schumannstr. 20-21,
10117 Berlin, Germany
M. Dietel (*)
Oligene GmbH,
Berlin, Germany
e-mail: manfred.dietel@charite.de
Tel.: +49-30-450536002
Fax: +49-30-450536900
effectiveness. Considering the fact that cell function is
controlled by a complex network of functionally active
signaling pathways, it is unlikely that expression analysis
of a single or a small number of proteins will precisely
predict the clinical outcome of an individual tumor.
Hanahan and Weinberg (2000) have described six essential
alterations in cancer cells that are common to most human
tumors: self-sufficiency in growth signals, insensitivity to
growth-inhibitory signals, evasion of apoptosis, limitless
replicative potential, sustained angiogenesis, and tissue
invasion and metastasis. However, each of these alterations
is achieved by different modifications in signal transduction,
gene or protein expression, protein modification, and
localization. Thus, a different and improved characterization
of malignant tumors as compared to current standard
diagnostics is required to reflect the multitude of genetic,
proteomic, and metabolomic alterations typical for malignant
cells. Several newly developed high-throughput
technologies, such as DNA microarrays, protein and
antibody arrays, proteome technologies such as mass
spectroscopy MALDI-TOF, metabolomic analyses, methylation
arrays, reverse-phase protein microarrays (RPAs),
etc., provide the technical platforms for this approach and
will gain more interest in the future. By integrating these
technologies into standard diagnostic procedures, future
molecular pathology will be able to dramatically increase
the relevance of tissue-based examinations. Pathology will
play a crucial role in therapy design and, by providing
molecular diagnostic approaches, also in therapy adjustment
and selection of effective drugs.
In this review, we will describe genomic, proteomic, and
epigenomic approaches which have been shown recently to
improve diagnostics and prognosis. We will focus on DNA
microarrays as the most advanced technology in this field
and suggest the development of an integrated approach for
future diagnostic molecular pathology.
Genomics—DNA microarrays
Technology
The generation of gene expression profiles from cultured
cells and tissue samples is nowadays a well-established
method. In the majority of experiments, DNA fragments
specific for individual genes on activated glass surfaces are
used [13, 62]. RNA is prepared from the biological material
to be analyzed, reverse transcribed into cDNA, labeled
with fluorescent dyes, and hybridized to the array. The
hybridization signals are detected by a laser scanner, the
images are normalized in various ways, and the relative
expression levels obtained for individual genes are further
clustered into groups of genes with similar or identical
expression patterns.
DNA microarrays consist of thousands of DNA
elements (10,000–80,000 or even more) robotically
tethered to a solid surface, e.g., coated glass slide, silicon,
or nylon. The DNA sequences are either oligonucleotides
or cDNA and represent different genes of interest. Each
gene is usually represented by more than one feature to
increase the specificity of the analysis. The cDNA clones
are often selected from public databases, amplified from
cDNA libraries, and purified before spotting. Oligonucleotides
(20–80 bp) are alternatively chemically synthesized
and spotted or may be directly synthesized on the array
surface. In addition to sequences representing known
genes, sequences representing expressed sequence tags
(EST) coding for genes of unknown function can also be
spotted on the solid surface, offering the potential of
investigating new marker genes.
The target DNA usually is a cDNA derived from tumor
cell mRNA which is amplified and simultaneously labeled
by reverse transcriptase-polymerase chain reaction (PCR).
The RNA is prepared from rapidly processed cell cultures
or fresh frozen tissue. Novel fixation techniques and
preparation methods are currently tested which might also
allow in the future the use of RNA prepared from paraffinembedded
tissues to be utilized in diverse microarray
experiments [28, 30, 54]. Target DNA, of course, can also
be prepared from genomic DNA fragments. For labeling,
fluorescent dyes are used nowadays. The expression
intensity of each gene is determined after hybridization
of the target DNA to the immobilized array DNA.
The hybridization signal generated on each spot reflects
the expression of the corresponding gene. The quantification
of the signals is performed using special software
which allows to correct for spot integrity and for technical
deviations during array production and hybridization and
to interpret the signal intensity. For this purpose, cDNA of
normal tissue or any other reference cDNA has to be
introduced. The final results provide a genetic expression
profile indicating overexpression, underexpression, no
change, or complete absence for each gene in the tissue
samples to be compared to each other (for review, see [52]).
Bioinformatics
The enormous amount of data obtained not only from each
high-throughput experiment but also from data documentation,
data processing, and data interpretation is a unique
challenge requiring close cooperation between pathologists
and bioinformatic specialists. Therefore, methods and
bioinformatic tools have been developed for the extraction
of complex information, the evaluation via statistical
methods, and the translation of molecular information
into clinically relevant data. cDNA microarrays today have
been used for a variety of different objectives. To choose
the method of statistical analysis that is appropriate for each
study, it is helpful to distinguish different types of array
analysis. Simon et al. [72] have distinguished three
different types of study objectives: “class discovery”
(find a new class), “class comparison” (find differences
between two predefined classes), and “class prediction”
(find a predictive gene set for a certain predefined class).
For each type of study, different statistical strategies for
evaluation should be used. According to Simon et al., using
methods incompatible with the goal and the design of the
745
study is one of the most common errors in the analysis of
microarray data.
Several methods for data analysis and interpretation
have been developed, which can be separated into two
classes: supervised and unsupervised methods [11]. A
supervised approach means that some external information
such as tumor grading or patients’ survival define the
relevant classes before the analysis; this method is used for
“class prediction” and “class comparison” studies.
In contrast, for “class discovery” studies, an unsupervised
approach is used, which defines different groups of
tumors exclusively according to their gene expression
profiles without using external information. An unsupervised
classification detects similarities in gene expression
profiles. For this purpose, the hierarchical clustering [24]
combined with a graphic presentation is an excellent tool
(Fig. 1). K-means or self-organizing maps are further
examples of unsupervised methods frequently used in
microarray analyses [79, 82]. It is important to note that
cluster analysis should not be used for class prediction and
class comparison studies.
Especially for “class prediction” studies, supervised
approaches should be applied. The supervised algorithm
establishes a model on the training set of samples, which
assigns gene sets with a defined expression profile to
predefined groups of tumor samples. After the class
prediction algorithm has been trained, it can be used to
predict the class of new samples. To assess the error rate
(e.g., percentage of incorrectly classified tumors), a
crossvalidation approach is often used. In this approach,
the available data sets are split repeatedly into a training set
and a test set. The algorithm is then trained on the training
set and applied to the test set, and correct and false
predictions are counted. A special type of cross-validation
is the leave-one-out strategy, where just one sample is
omitted from the training set (training on n−1). This one
sample is then used as the test set. This procedure is
repeated n times, each time using another sample as a test
set. This approach achieves maximal efficiency of training
yet requires high computational resources.
Such sample classification can be helpful to identify the
set of genes that discriminates between predefined groups
of patients with good or poor survival, e.g., estrogen
receptor (ER) positivity or ER negativity. This approach
has been described to discriminate, e.g., between tumor
entities like different solid tumors [78], subgroups of acute
myeloid leukemia and acute lymphocytic leukaemia [31],
diffuse large B-cell lymphomas, between early- and latestage
ovarian carcinomas [52], and breast carcinomas with
or without the potency to metastasize [85].
In a critical commentary, Simon, Radmacher, Dobbin,
and McShane [72] already worried that “many investigators
are not experienced in the analytical steps needed to
convert thousands of noisy data points into reliable and
Fig. 1 Hierarchical clustering
analysis exemplified for
Topotecan-resistant (R) and
Topotecan-sensitive (S) cell
lines. All cell lines resistant to
Topotecan (left panel) and all
cell lines sensitive to Topotecan
(right panel) express a unique
set of genes. Each row in the
cluster indicates the expression
profile of a specific gene across
all 19 cell lines. Each column
indicates the individual cell line
in which the gene is expressed.
Red, green, and black squares
indicate that expression of the
gene is greater than, less than, or
equal to the median level of
expression across all cell lines,
respectively. The scale bar reflects
the fold increase (red) or
decrease (green) for any given
gene relative to the median
level of expression across
all samples
746
interpretable biological information”. Furthermore, the
molecular profiles are often relatively unstable due to
tumour heterogeneity and platform specificity. The authors
listed a number of prerequisites necessary to avoid
misinterpretation of the data. In particular, the issue of
class prediction is discussed as it includes derivation of
classifiers, which predict prognosis, response to therapy,
potential to metastasize, and many more. In a very recent
publication by Michiels et al. [53], several of the most
prominent classifiers were doubted. The authors concluded
that the list of genes included in a molecular signature
largely depends on the constitution and size of the training
sets used and even claimed that a majority of the published
classifiers are not significant. A comprehensive overview is
also given by Scharpf et al. [67], suggesting several
methodological approaches and illustrating the associated
software to generate interpretable, robust, and less platform-dependent
results.
A relatively new unsupervised approach to analyze array
data is provided by linear independent component analysis
(ICA) [37, 48, 49]. In contrast to cluster analysis methods,
which assume that each gene belongs to just one cluster,
ICA is based on the idea of a combinatorial control which
is more realistic from a biological point of view. Gene
expression levels are described as linear functions of
common hidden variables which are (in the ideal case)
related to distinct biological causes of variation such as
transcriptional regulators or responses to treatments. Thus,
ICA-based results show increased biological relevance and
might describe new biological links previously undetected
by clustering methods. An interesting study was conducted
on endometrial carcinomas and benign endometrial
samples [66]. The authors compared the power of ICA to
other established methods and thereby detected (a) an
improved separation of benign and malignant groups, (b) a
characteristic expression pattern corresponding to the
histological classification, and (c) a novel pattern of
coregulated genes related to fatty acid metabolism which
are clinically associated with endometrial carcinoma. Thus,
a careful examination of histopathologically well-characterized
tumor material via high-throughput methods and
sophisticated bioinformatic approaches might even unravel
novel disease mechanisms during such advanced diagnostic
schemes.
Application of DNA microarrays
The application of high-throughput cDNA and oligonucleotide
array technologies and the establishment of gene
expression profiles has already revolutionized experimental
tumor diagnostics in the last couple of years. With the
use of these tools, otherwise indistinguishable tumor
subgroups have been identified [2]. In addition, the
development of gene signatures enabled prognostic conclusions
which could not be drawn using standard histopathological
and immunohistochemical methods [35, 84,
85], opening the door for a more comprehensive tumor
analysis by diagnostic pathology. In the following, some
examples of microarray applications will be described to
demonstrate the wide range of applications where this
technique could support and even improve the predictive
power of histopathological diagnostics.
Molecular classification and prognosis
Based on hundreds of international studies including
thousands of patients, an international classification system
for malignant tumors based on conventional histopathology
supported by immunohistochemistry has been established,
the UICC–TNM system [73]. In general, the TNM
system helps to estimate the prognosis of groups of tumors.
However, this classical approach with conventional histopathology,
analyzing morphological parameters, is not
completely sufficient when the prognosis of an individual
case shall be predicted. Therefore, the DNA microarray
technique is currently under investigation to improve the
current classification system by correlating array results
with histomorphology and clinical behavior of tumors.
The first step towards a molecular classification was
done by Khan et al. [39] who defined expression signatures
from cancer cell lines which were indicative for the organ
type of origin. A landmark study including gene expression
profiling with DNA microarrays as an extension of
histopathological classification has shown the possibility
to differentiate between AML and ALL and to improve the
precision of prognosis prediction [31].
Tumors with similar histological appearance but with
different clinical behavior, such as small blue round tumors
of childhood, are difficult to classify on a routine
microscopical basis. Genetic profiling proved to correctly
identify the four subgroups neuroblastoma, rhabdomyosarcoma,
Ewings sarcoma, and special types of nonHodgkin’s
lymphoma [40] and, thus, can be a valuable help
in tumor classification.
The WHO grading system of oligodendrogliomas is
somewhat limited due to subjectivity of histopathological
evaluation. Gene expression profiling now unraveled two
molecularly distinct subgroups which correspond to the
morphological grading and, thus, can be applied to provide
a more objective tool for grading and prediction of
prognosis [89]. This approach may be extended towards
other brain tumors where reproducible histological assessment
is often difficult, if not impossible.
In a group of 55 breast cancer patients with balanced
clinicopathological features, three subclasses of tumors
with different 5-year survival could be identified by
analyzing mRNA expression of only ≈1,000 candidate
genes [7, 8]. In a population-based study, Sotiriou et al.
[75] were able to suggest distinct aspects of breast cancer
classification and prognosis based on gene signatures. The
molecular analysis of pre-invasive vs invasive breast
tumors revealed extensive similarities of the genetic
alterations [51], suggesting that, in the early stage, the
potential for invasive growth is already almost fully
developed—an insight which might change the current
strategy of early breast cancer treatment.
747
Further genetic subclassifications of tumor entities with
distinct clinical features have been described for breast
cancer [60, 74], cutaneous malignant melanoma [10, 86],
large B-cell lymphoma [2], and B-cell lymphoma [14],
pediatric ALL [92] and acute leukemia with MLL translocation
[3], and adenocarcinoma of the lung [9, 26], ovary
[57, 69], colon [25], and prostate [21].
Metastatic potential
Up to now, it is impossible to determine precisely whether a
tumor which has not developed clinically detectable
metastases at the time of diagnosis will metastasize during
the following years. This information, however, would be
of utmost importance to decide, e.g., whether an adjuvant
therapy is necessary or not. The first steps to elucidate the
metastasizing potency of an individual tumor have been
published by van’t Veer et al. [85]. In breast carcinomas
with T1/T2, NO at time of diagnosis, the authors defined 70
“predictor genes” whose signature predicts with approximately
80% sensitivity and specificity the chance to set
lymph node metastases. A recent study by Wang et al. [88]
confirmed these data. This provides a rational to treat
patients with adjuvant therapy or not.
Tissue assignment, primary and occult metastases
In clinical practice, it may happen that a metastasis is found
while the primary tumor is unknown or it is unclear
whether a certain metastasis is derived from a known
primary tumor. Most often, histopathology, in addition to
morphological immunophenotyping, can disclose the tissue
origin. Nonetheless, gene expression profiling may be
of additional help to classify cancer cells according to their
tissue of origin [29, 39, 63, 78]. It was shown that the
molecular signature of the primary tumor is preserved in its
metastases and, thus, allows a clear assignment which
gives the necessary information for an adequate treatment.
Furthermore, it could be shown that DNA arrays are able to
distinguish between lung adenocarcinomas and metastases
of extrapulmonary origin [9]. The origin of cancer of
unknown primary can be derived from characteristic
expression patterns as the tumors often harbor distinct
gene expression profiles characteristic for the organ type of
origin [83].
Therapy response and drug resistance
Drug resistance remains a major problem during therapy
for systemic cancer disease. Due to the high potency to
adapt to therapeutical intervention, malignant tumor cells
frequently develop escape mechanisms in response to
radiation or cytostatic drugs. Although many attempts have
been made [45, 46], up to now, reliable and practically
applicable techniques to predict a tumor’s reaction to drugs
or radiation do not exist. The reason for this are multiple
cellular mechanisms such as increased DNA repair,
elevated levels of drug transporters, overexpression of
detoxifying enzymes, and decreased rates of apoptosis
which are often involved in the development of drug
resistance. To monitor multiple alterations occurring in
tumors being drug insensitive, highly parallel analyses,
such as the DNA microarray technique and also proteomic
analysis, are required. This opportunity opens new
dimensions to predict therapy resistance and sensitivity.
A recent National Institutes of Health study [68]
investigated 60 tumor cell lines (NCI60) which have
been treated independently with more than 70,000 different
agents, one at a time. Among other results, the study linked
bioinformatics and chemoinformatics by correlating cellular
drug response with transcriptomic information derived
from DNA microarrays. The aim was to associate clusters
of related drugs with clusters of gene alterations and,
thereby, to define drug–gene relationships. This approach
may contribute to the establishment of a defined expression
database on which an individualized molecular pharmacology
of tumor drug response can be established.
By comparing wild-type cell lines with derivatives
resistant for thymidilate synthase (TS) inhibitors, Wang et
al. [87] were able to identify certain patterns of genetic
alterations correlating with TS resistance. The associated
gene expression profile was partly tissue dependent, e.g.,
YES1 was overexpressed in the epithelial cell lines while it
was not upregulated in a lymphoblast cell line. Cisplatinum
(cDDP) resistance was shown by microarray
analysis to be accompanied with altered expression of
genes coding for membrane proteins and a glycoprotein
hormone subunit [33] not previously known to play any
role in cDDP resistance.
In an excellent study by Zembutsu et al. [94], 85 human
cancer xenografts were tested with regard to characteristic
expression profiles in response to nine anticancer drugs
often used in clinical therapy. More than 1,500 genes were
identified whose expression profile has correlated in some
way with chemosensitivity. The authors identified sets of
genes which could partly be associated with chemosensitivity
of particular tumor types (colon, breast, non-small
cell lung cancer, etc.) to the different drugs applied. To
predict the efficacy of adjuvant therapy in esophageal
tumors, the DNA microarray technique was applied to 20
cancer specimens with clinically known responses [41].
There were 52 genes identified which were likely to
correlate with patients’ outcome and possibly with
chemosensitivity and chemoresistance. This approach
shows some potential to determine drug response in
advance. In addition, bone marrow samples from 19
patients with acute lymphoblastic leukemia were investigated
with regard to resistance to an ABL tyrosine kinase
inhibitor [34]. On the basis of 95 differentially expressed
genes, it appeared to be possible to distinguish responder
from non-responder. Raponi et al. [65] showed the
possibility to detect the pathways modulated by inhibitory
drugs in AML elucidating the mechanisms of drug action.
Thus, gene expression profiling can be helpful for the pretreatment
assessment of anti-cancer therapy. Although a
748
vast number of experiments still have to be conducted, it
might become possible to predict chemoresistance (Fig. 2)
and to avoid non-effective drugs and unnecessary side
effects for the patients. The discrimination between
responders and non-responders before therapy will further
stimulate the development of an individualized therapeutic
strategy with a personalized combination of drugs.
Pitfalls of DNA microarray analyses
Numerous studies have shown characteristic expression
patterns related to certain aspects of tumor biology. Dozens
of genes were associated with prognosis, drug resistance,
potential to metastasize, etc.; however, the clinical
confirmation of the results is so far only on the way. The
clinical application of DNA microarrays will require a high
level of reproducibility and reliability of technology, of
sample processing, and of analysis. A major obstacle is the
still low inter-laboratory and inter-platform reproducibility
of microarray data. Several of the reasons underlying this
problem were described in 2003 by Tan et al. [80]. A
comparison of the three most widely used commercially
available platforms revealed significant discordance, which
could in part be attributed not only to differences between
the types of array but also to the different algorithms used
for data evaluation. In 2001, the MIAME standard was
already suggested, which includes a predefined set of
information for a certain microarray experiment which
allows for an independent evaluation of the results and for a
conclusion to be drawn [12]. In a more recent publication
by the Toxicogenomics Research Consortium, the impact
of array type, data handling, image analysis, and experimental
protocol onto the reproducibility of microarray
analysis between different laboratories was investigated
[5]. This study revealed that, like in the earlier reports, the
type of microarray used has by far the largest influence
onto the comparability of microarray data. More importantly,
the authors conclude that, upon application of a
common, commercial microarray platform, standardized
experimental image analysis and data processing approaches
correlation coefficients up to 0.9 can be achieved
between experiments performed in different laboratories.
Thus, careful experimental design and a strict standardization
of the whole procedure will be of utmost
importance in the future for the reproducibility of
microarray experiments. Furthermore, one should keep in
mind the fundamental challenges, which have frequently
obstructed cancer research during the last decades. These
include inter- and intratumor heterogeneity, specimens
either extracted, stored, or diagnosed inappropriately, and
unexpected changes of cell line properties. These complications
are still relevant and may be even accelerated by
high-throughput techniques.
Proteomics
The rapid development of proteome technology has lifted
tumor diagnostics to a new level. Despite the high
relevance of DNA- and RNA-based information discussed
above, nucleic acids are several layers of abstraction away
from the physiological events that determine disease
characteristics. Proteins govern metabolic processes, protein
interaction, and posttranslational modifications and
are, therefore, the major targets to be included in future
molecular pathological approaches (see also review by
Hanash [32]). Three fields of application are beginning to
emerge: (a) highly sensitive mass spectroscopy coupled to
an array-based separation of complex protein lysates
(surface enhanced laser desorption time-of-flight, SELDITOF)
or microdissected tumor samples are used for early
Fig. 2 A possible approach to
predict chemosensitivity and/or
chemoresistance of clinically
relevant tumor samples. The
low-density DNA arrays are
specially designed to indicate
the altered expression profile of
sensitivity/resistance associated
genes or EST. Different patterns
appear to be correlated with
sensitivity or resistance of
malignant tumors for certain
cytostatic drugs
749
detection and identification of new biomarkers for human
tumors [59, 61], (b) a direct mass spectroscopic analysis
and the establishment of protein profiles of tumor and
normal tissue [14, 76, 91], and (c) the application of protein
and antibody arrays for the specific identification of certain
proteins and their activated forms in tumors [50, 55, 77].
Although technically advanced screening methods have
improved early cancer detection, in some cancer types,
such as ovarian carcinoma, new methods which allow easy
and routine screening are urgently needed to ameliorate the
currently frustrating clinical outcomes. For this purpose,
the highly sensitive SELDI-TOF approach appeared as the
most promising technological development during recent
years. Petricoin et al. [61] reported the identification of
serum proteomic patterns using SELDI-TOF and developed
a bioinformatic tool to distinguish neoplastic ovarian
disease from the normal status. By applying the serum
spectra from 50 healthy women and 50 ovarian carcinoma
patients as training sets, they proposed a proteomic
signature which segregated cancer from non-cancer with
a sensitivity of 100% and a specificity of 95%. Yet, these
results were severely criticized and the expectations were
smoothened in recent publications [4, 17, 96]. Furthermore,
this provoked an intense debate about future
processes of establishing and proving the reliability of
novel technologies [22, 23, 64]. Thus, despite being an
exciting new approach, K. Coombes et al. [17] recently
stated that the “current state of the art in serum proteome
profiling allows considerable room for improvement”.
While serum proteomic analysis is currently tested in the
identification of new and early cancer markers, SELDITOF
has been successfully applied to the characterization
of distinct tumor areas carefully dissected by the help of
laser capture microdissection. This approach was used to
identify markers differentially expressed in prostate cancer
cells as compared to prostatic prostate intraepithelial
neoplasms (PIN) lesions and normal prostate epithelium
[16, 97]. Similar experimental schemes were also developed
for a proteomic evaluation of bladder carcinoma [43],
colon carcinoma [47], and breast carcinoma [19].
In addition to the direct identification of proteins using
mass spectroscopy, antibody arrays and so-called reverse
phase protein array approaches have also been developed
and applied for the identification of tumor-associated
proteins and also their activated phosphorylated derivatives.
Nowadays still in a developmental stage, both types
Table 1 Examples of targeted therapy approaches
Drug Tumor type Target Known action Detection method Remarks
Trastuzumab Metastatic breast cancer HER-2/neu Rc blocking IH, FISH, ISH 1
Cetuximab + Irinotecan Metastatic colorectal cancer EGFR Rc blocking/immunologic
response
IH, FISH 1
Imatinib Mesylate
(Gleevec)
CML, GIST with activated
c-kit receptor tyrosine kinase,
other sarcomas
Bcr/abl,
c-Kit,
PDGF-R
Tyrosine kinase inhibitor IH 1
Bevacizumab (Avastin)
(+5FU)
Colorectal cancer VEGF Rc blocking IH 2
G3139 (Genta, Berkley) Hematologic malignancies and
malignant melanoma
Antiapoptotic
gene bcl-2 in
Bcl-2 antisense
oligonucleotidecreasing
Bcl-2 mRNA
Immunophenotyping
by IH
2
Bortezomib, Epoxomicin Multiple myeloma Proteasome Proteasome inhibitor
Gefitinib (Iressa) Non-small cell lung cancer Mutated EGFR Kinase inhibitor Mutational analyses,
immunoblotting
4
Erlotinib (Tarceva) Non-small cell lung cancer Mutated EGFR Kinase inhibitor Mutational analyses,
immunoblotting
4
Rituximab (+ CHOP),
Y90-Ibritumomab,
I131-Tositumomab
Non-Hodgkin lymphoma CD20 Lympholytic Immunophenotyping
IH
3
Gemtuzumab-Ozogamicin
(calicheamycin)
AML (>60 years) CD33 Anti-CD33 guided cytotoxic
antibiotic, reduction of
P-glycoprotein
Immunophenotyping
by IH
3
Alemtuzumab (Campath) B-CLL, T-NHL, osteogenic
tumorsa
CD52 Lympholytic Immunophenotyping
by IH
3
Rapamycin RAD001 Breast, prostate, renal cancer TOR Kinase inhibitor 4
BMS 354825 GIST Kit Tyrosine kinase inhibitor IH 4
BAY43-9006 Melanoma RAF kinase Kinase inibitor 4
1 FDA-approved drugs requiring a pretherapeutic diagnostic eligibility test, 2 FDA-approved drugs targeting a specific pathway—no tests
available, 3 FDA-approved antibody-targeted therapies for hematologic maliognancies guided by immunophenotyping, 4 under
development, Rc receptor, IH immunohistochemistry, ISH in situ hybridization, FISH fluorescence in situ hybridization
a
Patent no. EP:03029464.9, V. Krenn, Institute of Pathology (submitted)
750
of arrays might become valuable alternatives to cDNA or
oligoarrays and also for immunohistochemical analyses
during daily pathological routine. A major obstacle, among
others, for these applications is the broad range of antibody
specificity and affinity which complicates detection of
defined antibody–antigen interactions [44]. Several studies
which used antibody arrays to profile protein patterns in
tumor samples derived from heptocellular carcinomas [81]
or breast carcinomas [36, 42] applied verification strategies
such as standard immunohistochemistry on tissue microarrays
to demonstrate specificity of their platform. An
interesting derivative of antibody microarrays are the
RPAs. In this case, instead of antibodies, whole protein
lysates are spotted onto glass surfaces and distinct proteins
are detected via antibodies [58]. This procedure can be
combined with laser capture microdissection and each
patient set can be applied in miniature dilution curves to
improve quantification accuracy and to enlarge the
dynamic range. RPAs are generated from frozen tissue
samples and, therefore, allow the use of detection
antibodies specific for phospho-proteins. This technology
was consequently described to compare survival pathway
activation as measured by phosporylated Akt and Erk in
normal prostate epithelium, in PIN, and in invasive prostate
cancer [58]. In addition, it has been used in pathology for
the investigation of the relative expression levels of Bcl-2
and Akt family members in follicular lymphomas [95]. One
can anticipate the future implication of these methods in the
elucidation of intracellular signaling networks downstream
of tyrosine kinase receptors such as HER2 and also
emerging from intracellular oncogenes such as RAS during
cancer therapy. Many of the substances in the new
generation of cancer drugs are designed to interfere with
specific molecular targets, which are believed to have a
critical role in tumor growth by regulating key signaling
pathways (Table 1; Fig. 3). Yet, this requires the development
of optimized protocols and superior detection
techniques [27] and the use of reference standards as
described recently by Sheehan et al. [70].
Epigenomics
Epigenetic modification of DNA by the addition of a
methyl group to cytosine residues in CpG dinucleotides is
one of the most important mechanisms of tumor suppressor
gene inactivation known today [6, 38]. A DNA-based
methylation analysis has several advantages when compared
to the RNA-based gene expression analysis. The
preparation of DNA suitable for such experiments can be
successfully obtained from both frozen and paraffinembedded
tissues. Furthermore, according to current
knowledge, methylation of genes is a rather stable
modification not prone to produce as many artifacts due
to RNA decay as in mRNA expression analysis. Methylation
arrays are now consequently designed for the highthroughput
analysis of DNA methylation at distinct CpG
positions. Furthermore, a real-time PCR-based assay for
gene- and methylation-specific detection has recently been
developed [18]. For methylation arrays, genomic DNA
from the tissue under investigation is treated with sodium
bisulphite, resulting in the conversion of unmethylated
cytosine residues into uracil [56]. Following this procedure,
specific fragments from the regulatory, CpG-containing
regions of genes are PCR-amplified using fluorescentlabeled
primers. The uracil residues (UpG) are replaced by
tyrosine residues (TpG) during PCR amplification. These
fragments are then hybridized to their complementary
oligonucleotides spotted onto a microarray, which are
specific for the originally non-methylated CpG dinucleotide.
Fig. 3 Novel cancer drugs targeting signaling pathways. Several
growth-stimulatory signaling pathways are activated in tumors due
to tyrosine kinase receptor overexpression (HER2 and EGFR) and
mutation (EGFR and KIT). Inhibition of tumor cell proliferation is
expected to occur via inhibition of receptor tyrosine kinases on
tumor cells or on endothelial cells (VEGFR). Furthermore, intracellular
protein kinases such as bcr/abl and RAF and phosphatidylinositol-related
kinases such as mTOR can be targeted by specific
inhibitors. Proteins like the antiapoptotic Bcl-2 can also be inhibited
to support chemotherapy
751
The hybridization conditions are designed in a such way that
a single mismatch between TpG and CpG variants yields a
specific detection signal. Such applications have been used
for the identification of CpG island methylation correlated
with tumor progression [1] and to evaluate a potential
significance for disease-free survival in ovarian carcinoma
patients after chemotherapy [90]. In addition, sophisticated
technologies (methylation target array) have enabled the
detection of CpG island methylation of known tumor
suppressor genes such as WT1, BRCA1, p16INK4A, and
others in hundreds of tumor samples at the same time [15].
An interesting further development is the combined use of
expression and methylation arrays, even in conjunction with
an analysis of histone acetylation [71] using the same tissue
samples. Such an approach allows the correlation of
methylation data with expression data and the identification
of functionally relevant genes harboring hypermethylated
CpG islands with a higher accuracy [93]. Although a
multitude of publications on high-throughput analysis of
DNA methylation in tumor tissues already exist, it is evident
that this technology has to be further developed to enter
routine clinical diagnostic procedures. Nevertheless, it
becomes clear that, in the future, a combination of
transcriptomic, proteomic, and epigenomic analysis is
required to identify the tumor-specific alterations with the
highest relevance for prognosis and therapy.
Outlook—an integrated approach in future diagnostics
In recent years, the efforts devoted to unraveling the
connection between activated receptor tyrosine kinases,
intracellular oncogenes, and distinct intracellular signaling
pathways and their role in growth control have uncovered a
large number of potential therapeutic target structures.
Herceptin and Gleevec targeting the HER2/neu, and BCR/
ABL, c-KIT, and PDGF-R oncoproteins, respectively, are
the first designer drugs which are already used with
considerable success in the clinic. In the meantime, other
drugs which act more or less specifically against a broad
range of receptors and signaling components have entered
clinical trials (Table 1) [20]. This development is intimately
connected with the expectation that tumor therapy can be
dramatically improved in the near future. To date, the
percentage of patients responding to the new inhibitors is
often below 30%. This indicates that the current portfolio
of diagnostic methods based on immunohistochemical or
DNA analysis (e.g., fluorescence insitu hybridization) of
single target molecules is still insufficient. For the identification
of those patients who will benefit from novel
therapies, special methods capable of detecting the entire
spectrum of rate-limiting oncogenic pathways in tumors
before and during therapy have to be developed and
adapted to routine diagnostic pathology. This will clearly
play an increasing role in the future tasks of pathological
institutes.
Before transmission and establishment of such an
integrated approach into daily routine can occur, extensive
research has to be performed to allow a firm prediction on
the activation of a certain pathway in clinical material. In
this case, a strategy consisting of a parallel analysis of gene
expression and methylation combined with a proteomic
analysis of activated signaling pathways and overexpressed
proteins will be the method of choice.
The implementation of bioinformatic procedures, which
will undergo further standardization together with highly
parallel tissue analyses, opens the door for a broad profiling
of human cancer tissue, thus improving the possibility to
predict the biological behavior of single tumors. Such a
“multiplex approach” may help to functionally classify
malignant tumors, provided that a “diagnostic algorithm”,
beginning with conventional histopathology in combination
with immunohistochemistry, provides the basis. This
basic evaluation will then be supplemented by specialized
disease-specific analyses such as different kinds of
micorarrays to predict tumor-associated features including
tumor progression, drug response, and metastatic potential,
which will have an important impact on tumor prognosis
and adequate individual therapy.
The recent technical developments also indicate changing
demands for routine pathological diagnostics within the
next years. Only a consequent integration of the methods
described within this review into pathological diagnostics
will guarantee that, in the future, pathology is also the
discipline holding the main body of knowledge about
disease etiology and disease mechanisms.
However, the demonstration that there exists a real
benefit for the patients with, e.g., longer survival times or
better response to therapy is still open. Only when welldesigned
clinical studies prove the progress an implementation
of molecular tumor profiling with individualized
therapeutic strategies will become accepted to be routinely
integrated in diagnosis and treatment of cancer. One major
prerequisite for an advantageous way is the interdisciplinary
cooperation of basis scientists, clinically oriented
physicians, diagnostic pathologists, and clinicians which
all must speak the same “language”. They must have an
understanding of their partners. The “translation” between
different groups of researchers will be one of the most
challenging tasks in modern experimental and diagnostic
pathology.
Acknowledgements The comments on the manuscript by Nils
Bluethgen, Institute for Theoretical Biology, Humboldt University,
Berlin are gratefully acknowledged. We appreciate the help of Balacz
Gyorffy and we are especially grateful for the secretarial work of
Mrs. von Bogen. This work was supported by Oligene GmbH,
Berlin.
References
1. Adorjan P, Distler J, Lipscher E, Model F, Muller J, Pelet C,
Braun A, Florl AR, Gutig D, Grabs G, Howe A, Kursar M,
Lesche R, Leu E, Lewin A, Maier S, Muller V, Otto T, Scholz
C, Schulz WA, Seifert HH, Schwope I, Ziebarth H, Berlin K,
Piepenbrock C, Olek A (2002) Tumour class prediction and
discovery by microarray-based DNA methylation analysis.
Nucleic Acids Res 30:e21
752
2. Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS,
Rosenwald A, Boldrick JC, Sabet H, Tran T, Yu X, Powell
JI, Yang L, Marti GE, Moore T, Hudson J Jr, Lu L, Lewis DB,
Tibshirani R, Sherlock G, Chan WC, Greiner TC, Weisenburger
DD, Armitage JO, Warnke R, Levy R, Wilson W, Grever MR,
Byrd JC, Botstein D, Brown PO, Staudt LM (2000) Distinct
types of diffuse large B-cell lymphoma identified by gene
expression profiling. Nature 403:503–511
3. Armstrong SA, Staunton JE, Silverman LB, Pieters R, den Boer
ML, Minden MD, Sallan SE, Lander ES, Golub TR, Korsmeyer
SJ (2002) MLL translocations specify a distinct gene expression
profile that distinguishes a unique leukemia. Nat Genet
30:41–47
4. Baggerly KA, Morris JS, Edmonson SR, Coombes KR (2005)
Signal in noise: evaluating reported reproducibility of serum
proteomic tests for ovarian cancer. J Natl Cancer Inst 97:
307–309
5. Bammler T, Beyer RP, Bhattacharya S, Boorman GA, Boyles
A, Bradford BU, Bumgarner RE, Bushel PR, Chaturvedi K,
Choi D, Cunningham ML, Deng S, Dressman HK, Fannin RD,
Farin FM, Freedman JH, Fry RC, Harper A, Humble MC,
Hurban P, Kavanagh TJ, Kaufmann WK, Kerr KF, Jing L,
Lapidus JA, Lasarev MR, Li J, Li YJ, Lobenhofer EK, Lu X,
Malek RL, Milton S, Nagalla SR, O’malley JP, Palmer VS,
Pattee P, Paules RS, Perou CM, Phillips K, Qin LX, Qiu Y,
Quigley SD, Rodland M, Rusyn I, Samson LD, Schwartz DA,
Shi Y, Shin JL, Sieber SO, Slifer S, Speer MC, Spencer PS,
Sproles DI, Swenberg JA, Suk WA, Sullivan RC, Tian R,
Tennant RW, Todd SA, Tucker CJ, Van Houten B, Weis BK,
Xuan S, Zarbl H (2005) Standardizing global gene expression
analysis between laboratories and across platforms. Nature
Methods 2:351–356
6. Baylin SB, Esteller M, Rountree MR, Bachman KE, Schuebel
K, Herman JG (2001) Aberrant patterns of DNA methylation,
chromatin formation and gene expression in cancer. Hum Mol
Genet 10:687–692
7. Bertucci F, Nasser V, Granjeaud S, Eisinger F, Adelaide J,
Tagett R, Loriod B, Giaconia A, Benziane A, Devilard E,
Jacquemier J, Viens P, Nguyen C, Birnbaum D, Houlgatte R
(2002) Gene expression profiles of poor-prognosis primary
breast cancer correlate with survival. Hum Mol Genet 11:
863–872
8. Bertucci F, Viens P, Hingamp P, Nasser V, Houlgatte R,
Birnbaum D (2003) Breast cancer revisited using DNA arraybased
gene expression profiling. Int J Cancer 103:565–571
9. Bhattacharjee A, Richards WG, Staunton J, Li C, Monti S, Vasa
P, Ladd C, Beheshti J, Bueno R, Gillette M, Loda M, Weber G,
Mark EJ, Lander ES, Wong W, Johnson BE, Golub TR,
Sugarbaker DJ, Meyerson M (2001) Classification of human
lung carcinomas by mRNA expression profiling reveals distinct
adenocarcinoma subclasses. Proc Natl Acad Sci USA
98:13790–13795
10. Bittner M, Meltzer P, Chen Y, Jiang Y, Seftor E, Hendrix M,
Radmacher M, Simon R, Yakhini Z, Ben Dor A, Sampas N,
Dougherty E, Wang E, Marincola F, Gooden C, Lueders J,
Glatfelter A, Pollock P, Carpten J, Gillanders E, Leja D,
Dietrich K, Beaudry C, Berens M, Alberts D, Sondak V (2000)
Molecular classification of cutaneous malignant melanoma by
gene expression profiling. Nature 406:536–540
11. Brazma A, Vilo J (2000) Gene expression data analysis. FEBS
Lett 480:17–24
12. Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman
P, Stoeckert C, Aach J, Ansorge W, Ball CA, Causton HC,
Gaasterland T, Glenisson P, Holstege FC, Kim IF, Markowitz V,
Matese JC, Parkinson H, Robinson A, Sarkans U, SchulzeKremer
S, Stewart J, Taylor R, Vilo J, Vingron M (2001)
Minimum information about a microarray experiment
(MIAME)—toward standards for microarray data. Nat Genet
29:365–371
13. Brown PO, Botstein D (1999) Exploring the new world of the
genome with DNA microarrays. Nat Genet 21:33–37
14. Celis JE, Kruhoffer M, Gromova I, Frederiksen C, Ostergaard
M, Thykjaer T, Gromov P, Yu J, Palsdottir H, Magnusson N,
Orntoft TF (2000) Gene expression profiling: monitoring
transcription and translation products using DNA microarrays
and proteomics. FEBS Lett 480:2–16
15. Chen CM, Chen HL, Hsiau TH, Hsiau AH, Shi H, Brock GJ,
Wei SH, Caldwell CW, Yan PS, Huang TH (2003) Methylation
target array for rapid analysis of CpG island hypermethylation
in multiple tissue genomes. Am J Pathol 163:37–45
16. Cheung PK, Woolcock B, Adomat H, Sutcliffe M, Bainbridge
TC, Jones EC, Webber D, Kinahan T, Sadar M, Gleave ME,
Vielkind J (2004) Protein profiling of microdissected prostate
tissue links growth differentiation factor 15 to prostate carcinogenesis.
Cancer Res 64:5929–5933
17. Coombes KR, Morris JS, Hu J, Edmonson SR, Baggerly KA
(2005) Serum proteomics profiling—a young technology
begins to mature. Nat Biotechnol 23:291–292
18. Cottrell SE, Distler J, Goodman NS, Mooney SH, Kluth A,
Olek A, Schwope I, Tetzner R, Ziebarth H, Berlin K (2004) A
real-time PCR assay for DNA-methylation using methylationspecific
blockers. Nucleic Acids Res 32:e10
19. Cowherd SM, Espina VA, Petricoin EF III, Liotta LA (2004)
Proteomic analysis of human breast cancer tissue with lasercapture
microdissection and reverse-phase protein microarrays.
Clin Breast Cancer 5:385–392
20. Dancey J, Sausville EA (2003) Issues and progress with protein
kinase inhibitors for cancer treatment. Nat Rev Drug Discov
2:296–313
21. Dhanasekaran SM, Barrette TR, Ghosh D, Shah R, Varambally
S, Kurachi K, Pienta KJ, Rubin MA, Chinnaiyan AM (2001)
Delineation of prognostic biomarkers in prostate cancer. Nature
412:822–826
22. Diamandis EP (2004) Analysis of serum proteomic patterns for
early cancer diagnosis: drawing attention to potential problems.
J Natl Cancer Inst 96:353–356
23. Diamandis EP, van der Merwe DE (2005) Plasma protein
profiling by mass spectrometry for cancer diagnosis: opportunities
and limitations. Clin Cancer Res 11:963–965
24. Eisen MB, Spellman PT, Brown PO, Botstein D (1998) Cluster
analysis and display of genome-wide expression patterns. Proc
Natl Acad Sci USA 95:14863–14868
25. Eschrich S, Yang I, Bloom G, Kwong KY, Boulware D, Cantor
A, Coppola D, Kruhoffer M, Aaltonen L, Orntoft TF,
Quackenbush J, Yeatman TJ (2005) Molecular staging for
survival prediction of colorectal cancer patients. J Clin Oncol
23:3526–3535
26. Garber ME, Troyanskaya OG, Schluens K, Petersen S, Thaesler
Z, Pacyna-Gengelbach M, van de Rijn RM, Rosen GD, Perou
CM, Whyte RI, Altman RB, Brown PO, Botstein D, Petersen I
(2001) Diversity of gene expression in adenocarcinoma of the
lung. Proc Natl Acad Sci USA 98:13784–13789
27. Geho D, Lahar N, Gurnani P, Huebschman M, Herrmann P,
Espina V, Shi A, Wulfkuhle J, Garner H, Petricoin E III, Liotta
LA, Rosenblatt KP (2005) Pegylated, steptavidin-conjugated
quantum dots are effective detection elements for reverse-phase
protein microarrays. Bioconjug Chem 16:559–566
28. Gianni L, Zambetti M, Clark K, Baker J, Cronin M, Wu J,
Mariani G, Rodriguez J, Carcangiu M, Watson D, Valagussa P,
Rouzier R, Symmans WF, Ross JS, Hortobagyi GN, Pusztai L,
Shak S (2005) Gene expression profiles in paraffin-embedded
core biopsy tissue predict response to chemotherapy in women
with locally advanced breast cancer. J Clin Oncol 23(29):
7265–7277
29. Giordano TJ, Shedden KA, Schwartz DR, Kuick R, Taylor JM,
Lee N, Misek DE, Greenson JK, Kardia SL, Beer DG, Rennert
G, Cho KR, Gruber SB, Fearon ER, Hanash S (2001) Organspecific
molecular classification of primary lung, colon, and
ovarian adenocarcinomas using gene expression profiles. Am J
Pathol 159:1231–1238
753
30. Goldmann T, Flohr AM, Murua EH, Gerstmayer B, Janssen U,
Bosio A, Loeschke S, Vollmer E, Bullerdiek J (2004) The
HOPE-technique permits Northern blot and microarray analyses
in paraffin-embedded tissues. Pathol Res Pract 200:
511–515
31. Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M,
Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri MA,
Bloomfield CD, Lander ES (1999) Molecular classification of
cancer: class discovery and class prediction by gene expression
monitoring. Science 286:531–537
32. Hanash S (2003) Disease proteomics. Nature 422:226–232
33. Higuchi E, Oridate N, Furuta Y, Suzuki S, Hatakeyama H,
Sawa H, Sunayashiki-Kusuzaki K, Yamazaki K, Inuyama Y,
Fukuda S (2003) Differentially expressed genes associated with
CIS-diamminedichloroplatinum (II) resistance in head and neck
cancer using differential display and CDNA microarray. Head
Neck 25:187–193
34. Hofmann WK, de Vos S, Elashoff D, Gschaidmeier H, Hoelzer
D, Koeffler HP, Ottmann OG (2002) Relation between
resistance of Philadelphia-chromosome-positive acute lymphoblastic
leukaemia to the tyrosine kinase inhibitor STI571 and
geneexpression profiles: a gene-expression study. Lancet
359:481–486
35. Huang E, Cheng SH, Dressman H, Pittman J, Tsou MH, Horng
CF, Bild A, Iversen ES, Liao M, Chen CM, West M, Nevins JR,
Huang AT (2003) Gene expression predictors of breast cancer
outcomes. Lancet 361:1590–1596
36. Hudelist G, Pacher-Zavisin M, Singer CF, Holper T, Kubista E,
Schreiber M, Manavi M, Bilban M, Czerwenka K (2004) Use
of high-throughput protein array for profiling of differentially
expressed proteins in normal and malignant breast tissue.
Breast Cancer Res Treat 86:281–291
37. Hyvärinen A, Karhunen J, Oja E (2001) Independent component
analysis. Wiley, New York
38. Jones PA, Baylin SB (2002) The fundamental role of epigenetic
events in cancer. Nat Rev Genet 3:415–428
39. Khan J, Simon R, Bittner M, Chen Y, Leighton SB, Pohida T,
Smith PD, Jiang Y, Gooden GC, Trent JM, Meltzer PS (1998)
Gene expression profiling of alveolar rhabdomyosarcoma with
cDNA microarrays. Cancer Res 58:5009–5013
40. Khan J, Wei JS, Ringner M, Saal LH, Ladanyi M, Westermann
F, Berthold F, Schwab M, Antonescu CR, Peterson C, Meltzer
PS (2001) Classification and diagnostic prediction of cancers
using gene expression profiling and artificial neural networks.
Nat Med 7:673–679
41. Kihara C, Tsunoda T, Tanaka T, Yamana H, Furukawa Y, Ono
K, Kitahara O, Zembutsu H, Yanagawa R, Hirata K, Takagi T,
Nakamura Y (2001) Prediction of sensitivity of esophageal
tumors to adjuvant chemotherapy by cDNA microarray analysis
of gene-expression profiles. Cancer Res 61:6474–6479
42. Knezevic V, Leethanakul C, Bichsel VE, Worth JM, Prabhu
VV, Gutkind JS, Liotta LA, Munson PJ, Petricoin EF III,
Krizman DB (2001) Proteomic profiling of the cancer microenvironment
by antibody arrays. Proteomics 1:1271–1278
43. Krieg RC, Gaisa NT, Paweletz CP, Knuechel R (2005)
Proteomic analysis of human bladder tissue using SELDI
approach following microdissection techniques. Methods Mol
Biol 293:255–267
44. Kusnezow W, Jacob A, Walijew A, Diehl F, Hoheisel JD
(2003) Antibody microarrays: an evaluation of production
parameters. Proteomics 3:254–264
45. Lage H, Dietel M (2000) Effect of the breast-cancer resistance
protein on atypical multidrug resistance. Lancet Oncol 1:
169–175
46. Lage H, Dietel M (2002) Multiple mechanisms confer different
drug-resistant phenotypes in pancreatic carcinoma cells. J
Cancer Res Clin Oncol 128:349–357
47. Lawrie LC, Curran S (2005) Laser capture microdissection and
colorectal cancer proteomics. Methods Mol Biol 293:245–253
48. Lee SI, Batzoglou S (2003) Application of independent
component analysis to microarrays. Genome Biol 4:R76
49. Liebermeister W (2002) Linear modes of gene expression
determined by independent component analysis. Bioinformatics
18:51–60
50. Liotta LA, Espina V, Mehta AI, Calvert V, Rosenblatt K, Geho
D, Munson PJ, Young L, Wulfkuhle J, Petricoin EF III (2003)
Protein microarrays: meeting analytical challenges for clinical
applications. Cancer Cell 3:317–325
51. Ma XJ, Salunga R, Tuggle JT, Gaudet J, Enright E, McQuary P,
Payette T, Pistone M, Stecker K, Zhang BM, Zhou YX,
Varnholt H, Smith B, Gadd M, Chatfield E, Kessler J, Baer TM,
Erlander MG, Sgroi DC (2003) Gene expression profiles of
human breast cancer progression. Proc Natl Acad Sci USA
100:5974–5979
52. Macgregor PF, Squire JA (2002) Application of microarrays
to the analysis of gene expression in cancer. Clin Chem 48:
1170–1177
53. Michiels S, Koscielny S, Hill C (2005) Prediction of cancer
outcome with microarrays: a multiple random validation
strategy. Lancet 365:488–492
54. Nessling M, Richter K, Schwaenen C, Roerig P, Wrobel G,
Wessendorf S, Fritz B, Bentz M, Sinn HP, Radlwimmer B,
Lichter P (2005) Candidate genes in breast cancer revealed by
microarray-based comparative genomic hybridization of archived
tissue. Cancer Res 65:439–447
55. Nielsen UB, Cardone MH, Sinskey AJ, MacBeath G, Sorger
PK (2003) Profiling receptor tyrosine kinase activation by
using Ab microarrays. Proc Natl Acad Sci USA 100:9330–9335
56. Olek A, Oswald J, Walter J (1996) A modified and improved
method for bisulphite based cytosine methylation analysis.
Nucleic Acids Res 24:5064–5066
57. Ono K, Tanaka T, Tsunoda T, Kitahara O, Kihara C, Okamoto
A, Ochiai K, Takagi T, Nakamura Y (2000) Identification by
cDNA microarray of genes involved in ovarian carcinogenesis.
Cancer Res 60:5007–5011
58. Paweletz CP, Charboneau L, Bichsel VE, Simone NL, Chen T,
Gillespie JW, Emmert-Buck MR, Roth MJ, Petricoin EF III,
Liotta LA (2001) Reverse phase protein microarrays which
capture disease progression show activation of pro-survival
pathways at the cancer invasion front. Oncogene 20:1981–1989
59. Paweletz CP, Trock B, Pennanen M, Tsangaris T, Magnant C,
Liotta LA, Petricoin EF III (2001) Proteomic patterns of nipple
aspirate fluids obtained by SELDI-TOF: potential for new
biomarkers to aid in the diagnosis of breast cancer. Dis Markers
17:301–307
60. Perou CM, Sorlie T, Eisen MB, van de Rijn RM, Jeffrey SS,
Rees CA, Pollack JR, Ross DT, Johnsen H, Akslen LA, Fluge
O, Pergamenschikov A, Williams C, Zhu SX, Lonning PE,
Borresen-Dale AL, Brown PO, Botstein D (2000) Molecular
portraits of human breast tumours. Nature 406:747–752
61. Petricoin EF, Ardekani AM, Hitt BA, Levine PJ, Fusaro VA,
Steinberg SM, Mills GB, Simone C, Fishman DA, Kohn EC,
Liotta LA (2002) Use of proteomic patterns in serum to identify
ovarian cancer. Lancet 359:572–577
62. Pollack JR, van de Rijn RM, Botstein D (2002) Challenges in
developing a molecular characterization of cancer. Semin Oncol
29:280–285
63. Ramaswamy S, Tamayo P, Rifkin R, Mukherjee S, Yeang CH,
Angelo M, Ladd C, Reich M, Latulippe E, Mesirov JP, Poggio
T, Gerald W, Loda M, Lander ES, Golub TR (2001) Multiclass
cancer diagnosis using tumor gene expression signatures. Proc
Natl Acad Sci USA 98:15149–15154
64. Ransohoff DF (2005) Lessons from controversy: ovarian
cancer screening and serum proteomics. J Natl Cancer Inst
97:315–319
65. Raponi M, Belly RT, Karp JE, Lancet JE, Atkins D, Wang Y
(2004) Microarray analysis reveals genetic pathways modulated
by tipifarnib in acute myeloid leukemia. BMC Cancer 4:56
66. Saidi SA, Holland CM, Kreil DP, MacKay DJ, Charnock-Jones
DS, Print CG, Smith SK (2004) Independent component
analysis of microarray data in the study of endometrial cancer.
Oncogene 23:6677–6683
754
67. Scharpf R, Garrett ES, Hu J, Parmigiani G (2003) Statistical
modeling and visualization of molecular profiles in cancer.
Biotechniques 34:S22–S29
68. Scherf U, Ross DT, Waltham M, Smith LH, Lee JK, Tanabe L,
Kohn KW, Reinhold WC, Myers TG, Andrews DT, Scudiero
DA, Eisen MB, Sausville EA, Pommier Y, Botstein D, Brown
PO, Weinstein JN (2000) A gene expression database for the
molecular pharmacology of cancer. Nat Genet 24:236–244
69. Schwartz DR, Kardia SL, Shedden KA, Kuick R, Michailidis
G, Taylor JM, Misek DE, Wu R, Zhai Y, Darrah DM, Reed H,
Ellenson LH, Giordano TJ, Fearon ER, Hanash SM, Cho KR
(2002) Gene expression in ovarian cancer reflects both
morphology and biological behavior, distinguishing clear cell
from other poor-prognosis ovarian carcinomas. Cancer Res
62:4722–4729
70. Sheehan KM, Calvert VS, Kay EW, Lu Y, Fishman D, Espina
V, Aquino J, Speer R, Araujo R, Mills GB, Liotta LA, Petricoin
EF III, Wulfkuhle JD (2005) Use of reverse phase protein
microarrays and reference standard development for molecular
network analysis of metastatic ovarian carcinoma. Mol Cell
Proteomics 4:346–355
71. Shi H, Wei SH, Leu YW, Rahmatpanah F, Liu JC, Yan PS,
Nephew KP, Huang TH (2003) Triple analysis of the cancer
epigenome: an integrated microarray system for assessing gene
expression, DNA methylation, and histone acetylation. Cancer
Res 63:2164–2171
72. Simon R, Radmacher MD, Dobbin K, McShane LM (2003)
Pitfalls in the use of DNA microarray data for diagnostic and
prognostic classification. J Natl Cancer Inst 95:14–18
73. Sobin LH, Wittekind Ch (eds) (2002) TNM classification of
malignant tumors, UICC, 6th edn. Wiley, New York
74. Sorlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H,
Hastie T, Eisen MB, van de Rijn RM, Jeffrey SS, Thorsen T,
Quist H, Matese JC, Brown PO, Botstein D, Eystein LP,
Borresen-Dale AL (2001) Gene expression patterns of breast
carcinomas distinguish tumor subclasses with clinical implications.
Proc Natl Acad Sci USA 98:10869–10874
75. Sotiriou C, Neo SY, McShane LM, Korn EL, Long PM, Jazaeri
A, Martiat P, Fox SB, Harris AL, Liu ET (2003) Breast cancer
classification and prognosis based on gene expression profiles
from a population-based study. Proc Natl Acad Sci USA
100:10393–10398
76. Souchelnytskyi S (2002) Proteomics in studies of signal
transduction in epithelial cells. J Mammary Gland Biol
Neoplasia 7:359–371
77. Sreekumar A, Nyati MK, Varambally S, Barrette TR, Ghosh D,
Lawrence TS, Chinnaiyan AM (2001) Profiling of cancer cells
using protein microarrays: discovery of novel radiationregulated
proteins. Cancer Res 61:7585–7593
78. Su AI, Welsh JB, Sapinoso LM, Kern SG, Dimitrov P, Lapp H,
Schultz PG, Powell SM, Moskaluk CA, Frierson HF Jr,
Hampton GM (2001) Molecular classification of human
carcinomas by use of gene expression signatures. Cancer Res
61:7388–7393
79. Tamayo P, Slonim D, Mesirov J, Zhu Q, Kitareewan S,
Dmitrovsky E, Lander ES, Golub TR (1999) Interpreting
patterns of gene expression with self-organizing maps: methods
and application to hematopoietic differentiation. Proc Natl
Acad Sci USA 96:2907–2912
80. Tan PK, Downey TJ, Spitznagel EL Jr, Xu P, Fu D, Dimitrov
DS, Lempicki RA, Raaka BM, Cam MC (2003) Evaluation of
gene expression measurements from commercial microarray
platforms. Nucleic Acids Res 31:5676–5684
81. Tannapfel A, Anhalt K, Hausermann P, Sommerer F, Benicke
M, Uhlmann D, Witzigmann H, Hauss J, Wittekind C (2003)
Identification of novel proteins associated with hepatocellular
carcinomas using protein microarrays. J Pathol 201:238–249
82. Tavazoie S, Hughes JD, Campbell MJ, Cho RJ, Church GM
(1999) Systematic determination of genetic network architecture.
Nat Genet 22:281–285
83. Tothill RW, Kowalczyk A, Rischin D, Bousioutas A, Haviv I,
van Laar RK, Waring PM, Zalcberg J, Ward R, Biankin AV,
Sutherland RL, Henshall SM, Fong K, Pollack JR, Bowtell DD,
Holloway AJ (2005) An expression-based site of origin
diagnostic method designed for clinical application to cancer
of unknown origin. Cancer Res 65:4031–4040
84. van de Vijver MJ, He YD, van’t Veer LJ, Dai H, Hart AA,
Voskuil DW, Schreiber GJ, Peterse JL, Roberts C, Marton MJ,
Parrish M, Atsma D, Witteveen A, Glas A, Delahaye L, van der
Velde T, Bartelink H, Rodenhuis S, Rutgers ET, Friend SH,
Bernards R (2002) A gene-expression signature as a predictor
of survival in breast cancer. N Engl J Med 347:1999–2009
85. van’t Veer LJ, Dai H, van de Vijver MJ, He YD, Hart AA,
Bernards R, Friend SH (2003) Expression profiling predicts
outcome in breast cancer. Breast Cancer Res 5:57–58
86. Vihinen PP, Pyrhonen SO, Kahari VM (2003) New prognostic
factors and developing therapy of cutaneous melanoma. Ann
Med 35:66–78
87. Wang W, Marsh S, Cassidy J, McLeod HL (2001) Pharmacogenomic
dissection of resistance to thymidylate synthase
inhibitors. Cancer Res 61:5505–5510
88. Wang Y, Klijn JG, Zhang Y, Sieuwerts AM, Look MP, Yang F,
Talantov D, Timmermans M, Meijer-van Gelder ME, Yu J,
Jatkoe T, Berns EM, Atkins D, Foekens JA (2005) Geneexpression
profiles to predict distant metastasis of lymph-nodenegative
primary breast cancer. Lancet 365:671–679
89. Watson MA, Perry A, Budhjara V, Hicks C, Shannon WD, Rich
KM (2001) Gene expression profiling with oligonucleotide
microarrays distinguishes World Health Organization grade of
oligodendrogliomas. Cancer Res 61:1825–1829
90. Wei SH, Chen CM, Strathdee G, Harnsomburana J, Shyu CR,
Rahmatpanah F, Shi H, Ng SW, Yan PS, Nephew KP, Brown R,
Huang TH (2002) Methylation microarray analysis of late-stage
ovarian carcinomas distinguishes progression-free survival in
patients and identifies candidate epigenetic markers. Clin
Cancer Res 8:2246–2252
91. Wulfkuhle JD, Sgroi DC, Krutzsch H, McLean K, McGarvey
K, Knowlton M, Chen S, Shu H, Sahin A, Kurek R, Wallwiener
D, Merino MJ, Petricoin EF III, Zhao Y, Steeg PS (2002)
Proteomics of human breast ductal carcinoma in situ. Cancer
Res 62:6740–6749
92. Yeoh EJ, Ross ME, Shurtleff SA, Williams WK, Patel D,
Mahfouz R, Behm FG, Raimondi SC, Relling MV, Patel A,
Cheng C, Campana D, Wilkins D, Zhou X, Li J, Liu H, Pui CH,
Evans WE, Naeve C, Wong L, Downing JR (2002) Classification,
subtype discovery, and prediction of outcome in
pediatric acute lymphoblastic leukemia by gene expression
profiling. Cancer Cell 1:133–143
93. Yu YP, Paranjpe S, Nelson J, Finkelstein S, Ren B, Kokkinakis
D, Michalopoulos G, Luo JH (2005) High throughput screening
of methylation status of genes in prostate cancer using an
oligonucleotide methylation array. Carcinogenesis 26:471–479
94. Zembutsu H, Ohnishi Y, Tsunoda T, Furukawa Y, Katagiri T,
Ueyama Y, Tamaoki N, Nomura T, Kitahara O, Yanagawa R,
Hirata K, Nakamura Y (2002) Genome-wide cDNA microarray
screening to correlate gene expression profiles with sensitivity
of 85 human cancer xenografts to anticancer drugs. Cancer Res
62:518–527
95. Zha H, Raffeld M, Charboneau L, Pittaluga S, Kwak LW,
Petricoin E III, Liotta LA, Jaffe ES (2004) Similarities of
prosurvival signals in Bcl-2-positive and Bcl-2-negative follicular
lymphomas identified by reverse phase protein microarray.
Lab Invest 84:235–244
96. Zhang Z, Bast RC Jr, Yu Y, Li J, Sokoll LJ, Rai AJ,
Rosenzweig JM, Cameron B, Wang YY, Meng XY, Berchuck
A, Haaften-Day C, Hacker NF, de Bruijn HW, van der Zee AG,
Jacobs IJ, Fung ET, Chan DW (2004) Three biomarkers
identified from serum proteomic analysis for the detection of
early stage ovarian cancer. Cancer Res 64:5882–5890
97. Zheng Y, Xu Y, Ye B, Lei J, Weinstein MH, O’Leary MP,
Richie JP, Mok SC, Liu BC (2003) Prostate carcinoma tissue
proteomics for biomarker discovery. Cancer 98:2576–2582
755