Abstract

Numerous recent studies have demonstrated the use of genomic data, particularly gene expression signatures, as clinical prognostic factors in cancer and other complex diseases. These studies highlight the opportunity for strategies to achieve truly personalized cancer treatment. In particular, the ability to develop gene expression signatures will likely allow us to guide the use of currently available cancer drugs, develop new targeted therapeutics, and provide an opportunity to better match the most effective drug or drugs with the molecular characteristics of the individual patient.

INTRODUCTION

The advent of genomic technologies has now provided the means to develop data that address the complexity of biologic states, including cancer. Particularly important has been the use of genome-scale gene expression analyses to identify discrete disease classes not previously recognized. Clear examples can be seen for both lymphoma ( 1 ) and breast cancer ( 2 ). The complex data generated in genomics allow us to identify patterns of gene expression, providing a snapshot of gene activity in a cell or tissue sample at a given instant of time which can be used to describe a phenotype. Genomic techniques are transforming biology from an observational molecular science to a data-intensive quantitative genomic science.

Most successful applications of genomic technology have been in the study of human cancer, in which gene expression patterns can be identified that provide phenotypic detail not previously obtainable by traditional methods of analysis: profiles and patterns that identify new subclasses of tumors, such as the distinction between acute myeloid leukemia and acute lymphoblastic leukemia, without prior knowledge of the classes ( 1 , 3 ). Such studies have led to an improved understanding of the underlying disease biology and the opportunity for development of novel therapeutics. Indeed, much of the activity in employing genomic technologies to achieve the goal of personalized cancer therapy has been directed at the identification of targets for new drug development that uniquely attack a given tumor. Certainly, imatinib mesylate (Gleevec) and trastuzumab (Herceptin) are examples of the end goal of this strategy of new drug development ( 4–6 ).

Genomic techniques may also be useful in determining more targeted applications for existing cancer therapeutics, many of which are very effective for subsets of cancer patients. Indeed, many testicular cancer patients are cured by standard adjuvant cytotoxic chemotherapy regimens, as is also the case with many lymphoma and leukemia patients. Likewise, anti-angiogenic drugs have been shown to have significant effects on certain populations of patients ( 7 , 8 ). Nevertheless, a primary limitation of cytotoxic therapeutics, as well as the majority of targeted therapeutic agents, is our current inability to guide their use by identifying the fraction of cancer patients who will benefit from a given regimen based on their specific tumor biology.

The practice of oncology continually faces the challenge of matching the right therapeutic regimen with the right patient, balancing relative benefit with risk to achieve the most successful outcome. Needless to say, this challenge is daunting, and, in present practice, we are often disappointed by suboptimal outcomes. In fact, as an example, 70–80% of patients receiving cytotoxic therapy for lung cancer obtain little-to-no clinical benefit from their treatment ( 9 ). The marginal success rate achieved in many types of cancer is likely a reflection of the enormous complexity of the disease process coupled with an inability to properly guide the use of available therapeutics. To the extent that these cytotoxic agents do have specificity, not necessarily for a biologic target but rather for a particular class of tumors, there is an opportunity to improve cancer care by implementing methods that improve a physician’s ability to choose the appropriate therapy for each cancer patient. Gene expression patterns can help define specific tumor classes in a way that correlates directly with chemotherapeutic options. Even many targeted therapeutics, developed to specifically inhibit a biologic target activity, also lack a basis for selecting those patients who might benefit from the drug. There are exceptions including trastuzumab, imatinib and various hormonal therapies but for the most part, these next-generation drugs are also used in a non-guided manner. The magnitude of the challenge of personalized cancer therapy, and also the opportunity, is further reflected in the fact that over 700 000 cancer therapeutic decisions are made annually in the USA alone, many of which are currently made empirically.

USING GENE EXPRESSION SIGNATURES TO REFINE PROGNOSIS

Several classification systems based on clinical, gross and pathologic criteria exist that attempt to categorize populations of patients into subclasses with different survival patterns. Among these several classification systems, the tumor, node, metastasis (TNM)-staging system has evolved to become the standard for prediction of survival in most solid tumors ( 10 ).

Despite minor modifications by disease type, the TNM-staging system has not changed in the past decade and clearly does not account for the prognostic diversity within each stage. The limited information on the characteristics of tumor size, degree of invasion and presence or absence of metastasis simply does not reflect the extreme complexity of disease, seen either in the variability of clinical outcomes or in the plethora of genetic alterations. Substantial data are now available using microarray analysis to identify genes and gene expression profiles that are associated with the different biologic behaviors of tumors and with prognosis prediction that goes far beyond traditional staging methods and prognostic criteria in solid tumors ( 11–14 ).

Although gene-expression studies have great potential for improving cancer management and for increasing our understanding of the disease biology, strong claims about the clinical value of these signatures cannot be made without prospective clinical trials that validate their benefit above and beyond the use of standard clinico-pathologic prognosis variables. Such efforts are currently underway in breast cancer; a 70-gene Amsterdam signature (Mammaprint ® ) ( 15 , 16 ) and the Oncotype DX ® recurrence score (RS) using 21 genes ( 17 ) have reached the final step of prospective clinical trial testing and are currently being validated in prospective phase III trials, the MINDACT ( 18 ) and TAILORx ( 19 ) trials, respectively (Fig.  1 ). Similar validation strategies are planned to evaluate the clinical benefit of genomic prognosis in lung cancer as well ( 11 ).

Figure 1.

The schema for the TAILORx trial. Stratification of patients into prognostic risk scores (RS) determines the individualized treatment arm.

Figure 1.

The schema for the TAILORx trial. Stratification of patients into prognostic risk scores (RS) determines the individualized treatment arm.

DIRECTING THE USE OF STANDARD-OF-CARE CYTOTOXIC CHEMOTHERAPEUTICS

The clinically based treatment guidelines presently used to guide the administration and selection of therapy for cancer patients can be imprecise. Often, multiple potential regimens with roughly equivalent efficacy and side-effect profiles are indicated for a patient population, and little or no guidance is provided on how to select among the choices. The result is an inefficient administration of therapy with average response rates that are less than optimal ( 9 ).

A number of studies over the past several years have sought to develop methods to predict who would respond to various chemotherapeutic drugs, both singly and in combination. Such prediction usually involves the analysis of tumor samples collected prior to treatment to identify gene expression profiles that could be associated with response. Examples of success with this approach have been seen in breast cancer neo-adjuvant treatment studies, using single agents as well as combination therapy ( 20–23 ).

An alternative strategy has made use of drug sensitivity data from cancer cell lines treated with many of the commonly used cytotoxic chemotherapies. The NCI-60 panel of cancer cell lines was developed to create a resource for drug sensitivity studies through the assay of many thousands of compounds in great detail. Recently, these drug sensitivity data have been coupled with baseline Affymetrix gene expression data for these cells, allowing the development of gene expression signatures that could predict sensitivity to various chemotherapeutic agents ( 24 , 25 ) (Fig.  2 A). Additional studies with other cancer cell line panels have yielded similar results ( 26 , 27 ). These signatures have been further employed in translational research to assess the ability to predict response to these drugs in patient studies. In several instances, including patient cohorts treated with docetaxel, paclitaxel, adriamycin and topotecan, the expression signatures generated from in vitro drug sensitivity measurements were able to accurately predict response of patients to these drugs ( 25 ).

Figure 2.

( A ) Patterns of predicted sensitivity to cytotoxic chemotherapy in non-small cell lung cancer. The top panel shows the signatures of drug response for commonly used cytotoxic agents. The bottom panel shows the patterns of predicted chemosensitivity in a cohort of 91 lung tumors (red, sensitive; blue, resistant). ( B ) A strategy to translate genomic-guided cytotoxic therapy into the clinic.

Figure 2.

( A ) Patterns of predicted sensitivity to cytotoxic chemotherapy in non-small cell lung cancer. The top panel shows the signatures of drug response for commonly used cytotoxic agents. The bottom panel shows the patterns of predicted chemosensitivity in a cohort of 91 lung tumors (red, sensitive; blue, resistant). ( B ) A strategy to translate genomic-guided cytotoxic therapy into the clinic.

Using genomic predictors of chemosensitivity may dramatically improve response rates to chemotherapy, significantly impacting the risk–benefit ratio for these patients. As an example, previous studies have shown that topotecan, anthracyclines and taxanes as salvage agents in patients with platinum-resistant advanced ovarian cancer have demonstrated a response rate that ranges between 20 and 30% ( 28 ). If one were to then use a predictor of chemosensitivity to a given agent that has a > 80% accuracy in predicting response, one would essentially increase the ‘effective’ response rate by selecting only those patients likely to respond, based on the drug sensitivity predictor. Predictors of chemotherapy response provide an opportunity to guide the selection of a specific drug that is optimal for an individual patient. This is an especially relevant issue in many cancer treatment scenarios given that there are often multiple regimens approved for use but without a clear superiority for any one agent or regimen. Although there is likely to be overlap in the sensitivities to the chemotherapeutic agents, it is also likely that there are distinct groups of patients predicted to be sensitive to a given agent ( 25 , 29 ) (Fig.  2 A). This provides an opportunity to better direct the use of individual salvage agents on the basis of the gene expression profile of the patient. Conversely, examples of overlap in the predicted sensitivity to the various agents may suggest the opportunity for combinations of agents not previously employed, to achieve a more effective therapeutic benefit.

The development of gene expression signatures that can predict response to various standard-of-care cytotoxic chemotherapies presents an opportunity to improve clinical practice. Many treatment guidelines present the oncologist with multiple options for therapy with approved, standard-of-care regimens, but without a basis to guide the decision about which potential regimen to use. Given the availability of genomic signatures that can predict response, one can design a prospective trial that compares outcomes of two regimens, one chosen traditionally versus a regimen selected through a genomics-guided model (Fig.  2 B). The attractiveness of this design is that a patient receives standard-of-care treatment regardless of the arm of the trial. The purpose is simply to evaluate the ability of the genomic tools to improve outcomes through more informed selection of chemotherapeutic regimens. A positive outcome in such a trial will provide strong evidence for the utility of the approach and the signatures to guide selection of the most appropriate therapy.

DIRECTING THE USE OF TARGETED BIOLOGIC THERAPIES

The success seen in the ability to identify patients at high risk for recurrence or patients likely to be resistant to standard cytotoxic therapies highlights the critical need for innovative approaches to find new therapies for these patients. One viable option in the pursuit of optimal targeted therapy is to make use of the wealth of pathway-specific therapeutics that have been developed over the past two decades. The challenge, as the case with the cytotoxic drugs, is developing strategies that can identify and then guide the use of these targeted agents. As an example, it is possible that the relative poor efficacy of farnesyl transferase inhibitor (FTI) in initial lung cancer trials may have been due to the lack of identification of the appropriate ‘susceptible’ population ( 30 ). With the exception of trastuzumab, imatinib and the hormonal therapies, there are few mechanisms that can select patients likely to respond to the various drugs ( 4–6 , 31 , 32 ).

One approach to targeting biologic therapies is to employ gene expression data that predict sensitivity to a new therapeutic agent, similar to the strategy with the chemotherapies. Indeed, at least one study has reported a gene expression signature that predicts sensitivity to the Src inhibitor dasatinib ( 33 ). Such an approach is developed by treating a series of cancer cell lines with the drug, measuring sensitivity on the basis of proliferation assays, and then using baseline expression data to develop the signature.

An alternative strategy has been described that makes use of gene expression patterns reflecting deregulation of oncogenic pathways ( 25 , 34 ). For instance, a variety of studies have detailed the role of alterations in Ras, Myc and the retinoblastoma protein in deregulating normal cellular proliferation and contributing to the development of oncogenic cells. These deregulated pathways are also the targets of the majority of targeted therapeutics developed or in development. An example of signatures reflecting the activation of various signaling pathways is shown in Figure  3 . Distinctions in the patterns of gene expression between the quiescent cells and the oncogene expressing cells are clearly evident.

Figure 3.

Signatures of signaling pathways underlying the oncogenic phenotype. In each instance, expression signatures developed in vitro represent differences between quiescent cells and ‘activated’ cells and clearly depict the status of an individual oncogenic pathway.

Figure 3.

Signatures of signaling pathways underlying the oncogenic phenotype. In each instance, expression signatures developed in vitro represent differences between quiescent cells and ‘activated’ cells and clearly depict the status of an individual oncogenic pathway.

Other studies have demonstrated an ability of the pathway profiles to predict tumors arising from deregulation of these pathways, providing a rationale for using these tools to identify those genetic pathways that have been deregulated during the development of the tumor and to map out the pattern of pathway deregulation associated with tumor development ( 35 , 36 ) (Fig.  4 A). Importantly, gene expression profiles provide a way to measure the consequences of the oncogenic process, irrespective of which specific aspect of the pathway is altered. Even if the known oncogene is not mutated, but rather another component of the pathway is altered, the gene expression profile will still detect the alteration.

Figure 4.

( A ) Patterns of oncogenic pathway deregulation seen in non-small cell lung cancer. A panel of signatures (top) representative of individual pathway activation is applied to lung tumors to reveal distinct patterns of pathway deregulation (bottom). The samples are on the X-axis and the status of each pathway is depicted by row (red, activated pathway; blue, quiescent pathway). ( B ) An approach to identifying appropriate choices for combination therapy. Patient samples (on the X-axis) have been sorted on the basis of the predicted sensitivity to docetaxel and examined for patterns of pathway deregulation (red, activated pathway; blue, quiescent pathway). This example shows that patient subsets that might be resistant to docetaxel are likely to be sensitive to PI3kinase inhibition and thus identifies a combination strategy that can be further studied in clinical trials.

Figure 4.

( A ) Patterns of oncogenic pathway deregulation seen in non-small cell lung cancer. A panel of signatures (top) representative of individual pathway activation is applied to lung tumors to reveal distinct patterns of pathway deregulation (bottom). The samples are on the X-axis and the status of each pathway is depicted by row (red, activated pathway; blue, quiescent pathway). ( B ) An approach to identifying appropriate choices for combination therapy. Patient samples (on the X-axis) have been sorted on the basis of the predicted sensitivity to docetaxel and examined for patterns of pathway deregulation (red, activated pathway; blue, quiescent pathway). This example shows that patient subsets that might be resistant to docetaxel are likely to be sensitive to PI3kinase inhibition and thus identifies a combination strategy that can be further studied in clinical trials.

The analysis of oncogenic pathways through gene expression analysis offers a potential opportunity to identify new therapeutic options for these patients by providing a potential basis for guiding the use of pathway-specific drugs. Although in principle one could generate a predictor of response to each potential therapeutic agents, in reality this would quickly become impractical for the hundreds of targeted therapies. Pathway predictors provide a more efficient mechanism to achieve the same goal. One major value of this approach is the capacity to direct combinations of therapies using multiple drugs that target multiple pathways on the basis of genomic information that describes the state of activity of the pathways.

To demonstrate the value of the use of pathway prediction to guide drug use, pathway deregulation has been assessed in a series of breast cancer cell lines that are used for screening of potential therapeutic drugs. In parallel with mapping the pathway status, the cell lines are employed for assays with a variety of drugs known to target specific activities within given oncogenic pathways. The goal is to directly demonstrate that a cell line is sensitive to a drug on the basis of the knowledge of the pathway deregulation within that cell. This can then be extrapolated to tumors, predicting sensitivity to a given drug on the basis of the knowledge of pathway deregulation in the tumor. Drugs employed include a variety of available reagents that target each of the pathways for which signatures have been developed. Results from testing sensitivity of the cell lines to an FTI (L-744832), as well as Src-inhibitor (SU6656), employing growth inhibition as the assay, have shown a clear relationship between prediction of pathway deregulation and sensitivity to the respective therapy ( 25 , 34 ). These initial positive results have now been further validated in a larger set of 50 cancer cell lines to include lung, ovarian and melanoma tumor types. Currently, systematic studies are underway that include the use of combinations of drugs (chemotherapy and targeted agents) to determine the extent to which enhanced sensitivity can be predicted by pathway analysis. These experiments aim to translate our pathway predictions into clinically relevant identification of individual therapeutic sensitivity as well as identification of the most effective combination of therapeutic interventions.

NEXT STEPS: EXPLORING OPPORTUNITIES FOR COMPLEX COMBINATION THERAPY

Beyond the individual signatures assayed to date, an additional valuable resource is the availability of a database of signature information. Each of the existing panels of signatures, along with newly developed signatures, can be used to profile large collections of cancer expression data sets to generate profiles of signature activity across multiple tumor types. An example is shown for pathway activity and chemosensitivity profiles in a series of lung cancer samples (Fig.  4 B). This data repository, summarizing the profiles of signatures across many tumor data sets, provides a further resource of substantial value. In the context of drug development strategies, this information cannot serve as a guide to the identification of populations of patients that might be appropriate for a given therapeutic but can also identify potential combinations of drugs that might be most efficacious given the overlap in predicted sensitivities.

These data also provide a resource for evaluating opportunities for identifying patient subsets that may be appropriate for utilization of a given therapeutic in combination with a signature. This also provides the important additional opportunity of looking for appropriate combinations on the basis of patterns of these profiles. An example of the latter is shown for the case of docetaxel (Fig.  4 B) in which patient samples have been sorted on the basis of predicted sensitivity to docetaxel and examined for predictions of pathway activation. This provides the potential for identifying patient subsets that might be used in combination with docetaxel or to provide alternatives for those patients likely to be resistant to docetaxel. Obviously, this can be extended to any of the available signatures as well as new ones developed in specific projects.

Conflict of Interest statement . This work was supported by grants from the National Institutes of Health/National Cancer Institute (5-U54-CA112952-03 and 1-R01-CA106520-03) and by a gift from the V Foundation.

REFERENCES

1
Golub
T.R.
Slonim
D.K.
Tamayo
P.
Huard
C.
Gaasenbeek
M.
Mesirov
J.P.
Coller
H.
Loh
M.L.
Downing
J.R.
Caligiuri
M.A.
, et al.  . 
Molecular classification of cancer: class discovery and class prediction by gene expression monitoring
Science
 , 
1999
, vol. 
286
 (pg. 
531
-
537
)
2
Perou
C.M.
Sorlie
T.
Eisen
M.B.
van de Rijn
M.
Jeffrey
S.S.
Rees
C.A.
Pollack
J.R.
Ross
D.T.
Johnsen
H.
Akslen
L.A.
, et al.  . 
Molecular portraits of human breast tumours
Nature
 , 
2000
, vol. 
406
 (pg. 
747
-
752
)
3
Alizadeh
A.A.
Eisen
M.B.
Davis
R.E.
Ma
C.
Lossos
I.S.
Rosenwald
A.
Boldrick
J.C.
Sabet
H.
Tran
T.
Yu
X.
, et al.  . 
Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling
Nature
 , 
2000
, vol. 
403
 (pg. 
503
-
511
)
4
Slamon
D.J.
Leyland-Jones
B.
Shak
S.
Fuchs
H.
Paton
V.
Bajamonde
A.
Fleming
T.
Eiermann
W.
Wolter
J.
Pegram
M.
, et al.  . 
Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2
N. Engl. J. Med.
 , 
2001
, vol. 
344
 (pg. 
783
-
792
)
5
Cobleigh
M.A.
Vogel
C.L.
Tripathy
D.
Robert
N.J.
Scholl
S.
Fehrenbacher
L.
Wolter
J.M.
Paton
V.
Shak
S.
Lieberman
G.
, et al.  . 
Multinational study of the efficacy and safety of humanized anti-HER2 monoclonal antibody in women who have HER2-overexpressing metastatic breast cancer that has progressed after chemotherapy for metastatic disease
J. Clin. Oncol.
 , 
1999
, vol. 
17
 (pg. 
2639
-
2648
)
6
Druker
B.J.
Guilhot
F.
O’Brien
S.G.
Gathmann
I.
Kantarjian
H.
Gattermann
N.
Deininger
M.W.
Silver
R.T.
Goldman
J.M.
Stone
R.M.
, et al.  . 
Five-year follow-up of patients receiving imatinib for chronic myeloid leukemia
N. Engl. J. Med.
 , 
2006
, vol. 
355
 (pg. 
2408
-
2417
)
7
Hurwitz
H.
Fehrenbacher
L.
Novotny
W.
Cartwright
T.
Hainsworth
J.
Heim
W.
Berlin
J.
Baron
A.
Griffing
S.
Holmgren
E.
, et al.  . 
Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer
N. Engl. J. Med.
 , 
2004
, vol. 
350
 (pg. 
2335
-
2342
)
8
Sandler
A.
Gray
R.
Perry
M.C.
Brahmer
J.
Schiller
J.H.
Dowlati
A.
Lilenbaum
R.
Johnson
D.H.
Paclitaxel–carboplatin alone or with bevacizumab for non-small-cell lung cancer
N. Engl. J. Med.
 , 
2006
, vol. 
355
 (pg. 
2542
-
2550
)
9
Ettinger
D.S.
Bepler
G.
Bueno
R.
Chang
A.
Chang
T.Y.
Chirieac
L.R.
D'Amico
T.A.
Demmy
T.L.
Feigenberg
S.J.
Grannis
F.W.
Jr
, et al.  . 
National Comprehensive Cancer Network (NCCN). Non-small lung cancer clinical practice guidelines in oncloogy.
J. Natl. Compr. Canc. Natw.
 , 
2006
, vol. 
6
 (pg. 
548
-
582
)
10
Greene
E.L.
Page
D.L.
Fleming
I.D.
, et al.  . 
AJCC Cancer Staging Manual
 , 
2002
6
New York
Springer
11
Potti
A.
Mukherjee
S.
Petersen
R.
Dressman
H.K.
Bild
A.
Koontz
J.
Kratzke
R.
Watson
M.A.
Kelley
M.
Ginsburg
G.S.
, et al.  . 
A genomic strategy to refine prognosis in early-stage non-small-cell lung cancer
N. Engl. J. Med.
 , 
2006
, vol. 
355
 (pg. 
570
-
580
)
12
Ramaswamy
S.
Ross
K.N.
Lander
E.S.
Golub
T.R.
A molecular signature of metastasis in primary solid tumors
Nat. Genet.
 , 
2003
, vol. 
33
 (pg. 
49
-
54
)
13
Arango
D.
Laiho
P.
Kokko
A.
Alhopuro
P.
Sammalkorpi
H.
Salovaara
R.
Nicorici
D.
Hautaniemi
S.
Alazzouzi
H.
Mecklin
J.P.
, et al.  . 
Gene-expression profiling predicts recurrence in Dukes’ C colorectal cancer
Gastroenterology
 , 
2005
, vol. 
129
 (pg. 
874
-
884
)
14
Lyman
G.H.
Kuderer
N.M.
Gene expression profile assays as predictors of recurrence-free survival in early-stage breast cancer: a metaanalysis
Clin. Breast Cancer
 , 
2006
, vol. 
7
 (pg. 
372
-
379
)
15
van de Vijver
M.J.
He
Y.D.
van’t Veer
L.J.
Dai
H.
Hart
A.A.
Voskuil
D.W.
Schreiber
G.J.
Peterse
J.L.
Roberts
C.
Marton
M.J.
, et al.  . 
A gene-expression signature as a predictor of survival in breast cancer
N. Engl. J. Med.
 , 
2002
, vol. 
347
 (pg. 
1999
-
2009
)
16
Glas
A.M.
Floore
A.
Delahaye
L.J.
Witteveen
A.T.
Pover
R.C.
Bakx
N.
Lahti-Domenici
J.S.
Bruinsma
T.J.
Warmoes
M.O.
Bernards
R.
, et al.  . 
Converting a breast cancer microarray signature into a high-throughput diagnostic test
BMC Genomics
 , 
2006
, vol. 
7
 pg. 
278
 
17
Paik
S.
Shak
S.
Tang
G.
Kim
C.
Baker
J.
Cronin
M.
Baehner
F.L.
Walker
M.G.
Watson
D.
Park
T.
, et al.  . 
A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer
N. Engl. J. Med.
 , 
2004
, vol. 
351
 (pg. 
2817
-
2826
)
18
Bogaerts
J.
Cardoso
F.
Buyse
M.
Braga
S.
Loi
S.
Harrison
J.A.
Bines
J.
Mook
S.
Decker
N.
Ravdin
P.
, et al.  . 
Gene signature evaluation as a prognostic tool: challenges in the design of the MINDACT trial
Nat. Clin. Pract. Oncol.
 , 
2006
, vol. 
3
 (pg. 
540
-
551
)
19
Sparano
J.A.
TAILORx: trial assigning individualized options for treatment (Rx)
Clin. Breast Cancer
 , 
2006
, vol. 
7
 (pg. 
347
-
350
)
20
Chang
J.C.
Wooten
E.C.
Tsimelzon
A.
Hilsenbeck
S.G.
Gutierrez
M.C.
Elledge
R.
Mohsin
S.
Osborne
C.K.
Chamness
G.C.
Allred
D.C.
, et al.  . 
Gene expression profiling for the prediction of therapeutic response to docetaxel in patients with breast cancer
Lancet
 , 
2003
, vol. 
362
 (pg. 
362
-
369
)
21
Cleator
S.
Tsimelzon
A.
Ashworth
A.
Dowsett
M.
Dexter
T.
Powles
T.
Hilsenbeck
S.
Wong
H.
Osborne
C.K.
O’Connell
P.
, et al.  . 
Gene expression patterns for doxorubicin (adriamycin) and cyclophosphamide (cytoxan) (AC) response and resistance
Breast Cancer Res. Treat.
 , 
2006
, vol. 
95
 (pg. 
229
-
233
)
22
Holleman
A.
Cheok
M.H.
den Boer
M.L.
Yang
W.
Veerman
A.J.
Kazemier
K.M.
Pei
D.
Cheng
C.
Pui
C.H.
Relling
M.V.
, et al.  . 
Gene-expression patterns in drug-resistant acute lymphoblastic leukemia cells and response to treatment
N. Engl. J. Med.
 , 
2004
, vol. 
351
 (pg. 
533
-
542
)
23
Hess
K.R.
Anderson
K.
Symmans
W.F.
Valero
V.
Ibrahim
N.
Mejia
J.A.
Booser
D.
Theriault
R.L.
Buzdar
A.U.
Dempsey
P.J.
, et al.  . 
Pharmacogenomic predictor of sensitivity to preoperative chemotherapy with paclitaxel and fluorouracil, doxorubicin, and cyclophosphamide in breast cancer
J. Clin. Oncol.
 , 
2006
, vol. 
24
 (pg. 
4236
-
4244
)
24
Staunton
J.E.
Slonim
D.K.
Coller
H.A.
Tamayo
P.
Angelo
M.J.
Park
J.
Scherf
U.
Lee
J.K.
Reinhold
W.O.
Weinstein
J.N.
, et al.  . 
Chemosensitivity prediction by transcriptional profiling
Proc. Natl Acad. Sci. USA
 , 
2001
, vol. 
98
 (pg. 
10787
-
10792
)
25
Potti
A.
Dressman
H.K.
Bild
A.
Riedel
R.F.
Chan
G.
Sayer
R.
Cragun
J.
Cottrill
H.
Kelley
M.J.
Petersen
R.
, et al.  . 
Genomic signatures to guide the use of chemotherapeutics
Nat. Med.
 , 
2006
, vol. 
12
 (pg. 
1294
-
1300
)
26
Lugthart
S.
Cheok
M.H.
den Boer
M.L.
Yang
W.
Holleman
A.
Cheng
C.
Pui
C.H.
Relling
M.V.
Janka-Schaub
G.E.
Pieters
R.
, et al.  . 
Identification of genes associated with chemotherapy crossresistance and treatment response in childhood acute lymphoblastic leukemia
Cancer Cell
 , 
2005
, vol. 
7
 (pg. 
375
-
386
)
27
Gyorffy
B.
Surowiak
P.
Kiesslich
O.
Denkert
C.
Schafer
R.
Dietel
M.
Lage
H.
Gene expression profiling of 30 cancer cell lines predicts resistance towards 11 anticancer drugs at clinically achieved concentrations
Int. J. Cancer.
 , 
2006
, vol. 
118
 (pg. 
1699
-
1712
)
28
ten Bokkel Huinink
W.
Lane
S.R.
Ross
G.A.
Long-term survival in a phase III, randomised study of topotecan versus paclitaxel in advanced epithelial ovarian carcinoma
Ann. Oncol.
 , 
2004
, vol. 
15
 (pg. 
100
-
103
)
29
Dressman
H.K.
Berchuck
A.
Chan
G.
Zhai
J.
Bild
A.
Sayer
R.
Cragun
J.
Clarke
J.
Whitaker
R.S.
Li
L.
, et al.  . 
An integrated genomic-based approach to individualized treatment of patients with advanced-stage ovarian cancer
J. Clin. Oncol.
 , 
2007
, vol. 
25
 (pg. 
517
-
525
)
30
Adjei
A.A.
Mauer
A.
Bruzek
L.
Marks
R.S.
Hillman
S.
Geyer
S.
Hanson
L.J.
Wright
J.J.
Erlichman
C.
Kaufmann
S.H.
, et al.  . 
Phase II study of the farnesyl transferase inhibitor R115777 in patients with advanced non-small-cell lung cancer
J. Clin. Oncol.
 , 
2003
, vol. 
21
 (pg. 
1760
-
1766
)
31
Vogel
C.L.
Cobleigh
M.A.
Tripathy
D.
Gutheil
J.C.
Harris
L.N.
Fehrenbacher
L.
Slamon
D.J.
Murphy
M.
Novotny
W.F.
Burchmore
M.
, et al.  . 
Efficacy and safety of trastuzumab as a single agent in first-line treatment of HER2-overexpressing metastatic breast cancer
J. Clin. Oncol.
 , 
2002
, vol. 
20
 (pg. 
719
-
726
)
32
Graff
J.N.
Mori
M.
Li
H.
Garzotto
M.
Penson
D.
Potosky
A.L.
Beer
T.M.
Predictors of overall and cancer-free survival of patients with localized prostate cancer treated with primary androgen suppression therapy: results from the prostate cancer outcomes study
J. Urol.
 , 
2007
, vol. 
177
 (pg. 
1307
-
1312
)
33
Huang
F.
Reeves
K.
Han
X.
Fairchild
C.
Platero
S.
Wong
T.W.
Lee
F.
Shaw
P.
Clark
E.
Identification of candidate molecular markers predicting sensitivity in solid tumors to dasatinib: rationale for patient selection
Cancer Res.
 , 
2007
, vol. 
67
 (pg. 
2226
-
2238
)
34
Bild
A.H.
Yao
G.
Chang
J.T.
Wang
Q.
Potti
A.
Chasse
D.
Joshi
M.B.
Harpole
D.
Lancaster
J.M.
Berchuck
A.
, et al.  . 
Oncogenic pathway signatures in human cancers as a guide to targeted therapies
Nature
 , 
2006
, vol. 
439
 (pg. 
353
-
357
)
35
Bild
A.H.
Potti
A.
Nevins
J.R.
Linking oncogenic pathways with therapeutic opportunities
Nat. Rev. Cancer
 , 
2006
, vol. 
6
 (pg. 
735
-
741
)
36
Rhodes
D.R.
Yu
J.
Shanker
K.
Deshpande
N.
Varambally
R.
Ghosh
D.
Barrette
T.
Pandey
A.
Chinnaiyan
A.M.
Large-scale meta-analysis of cancer microarray data identifies common transcriptional profiles of neoplastic transformation and progression
Proc. Natl Acad. Sci. USA
 , 
2004
, vol. 
101
 (pg. 
9309
-
9314
)