Personalized Medicine: Marking a New Epoch in Cancer Patient Management

Published OnlineFirst August 6, 2010; DOI: 10.1158/1541-7786.MCR-10-0264

Review Molecular Cancer Research Personalized Medicine: Marking a New Epoch in Cancer Patient Management

Maria Diamandis1, Nicole M.A. White1,2, and George M. Yousef1,2

Abstract Personalized medicine (PM) is defined as “a form of medicine that uses information about a person's genes, proteins, and environment to prevent, diagnose, and treat disease.” The promise of PM has been on us for years. The suite of clinical applications of PM in cancer is broad, encompassing screening, diagnosis, prognosis, prediction of treatment efficacy, patient follow-up after surgery for early detection of recurrence, and the strat- ification of patients into cancer subgroup categories, allowing for individualized therapy. PM aims to eliminate the “one size fits all” model of medicine, which has centered on reaction to disease based on average responses to care. By dividing patients into unique cancer subgroups, treatment and follow-up can be tailored for each individual according to disease aggressiveness and the ability to respond to a certain treatment. PM is also shifting the emphasis of patient management from primary patient care to prevention and early intervention for high-risk individuals. In addition to classic single molecular markers, high-throughput approaches can be used for PM including whole genome sequencing, single-nucleotide polymorphism analysis, microarray analysis, and mass spectrometry. A common trend among these tools is their ability to analyze many targets simultaneously, thus increasing the sensitivity, specificity, and accuracy of biomarker discovery. Certain challenges need to be addressed in our transition to PM including assessment of cost, test standardization, and ethical issues. It is clear that PM will gradually continue to be incorporated into cancer patient management and will have a significant impact on our health care in the future. Mol Cancer Res; 8(9); 1175–87. ©2010 AACR.

Introduction genes, proteins, and environment to prevent, diagnose, and treat disease.” The promise of PM has been on us for some The field of medicine has seen many new advances in years and is rapidly gaining momentum. The concept of PM the last several decades. With the completion of the human is to use clinical, genomic, transcriptomic, proteomic, and genome project (1), an enormous amount of new informa- other information sources to plot the optimal course for tion has been gained about the human genome and the an individual in terms of disease risk assessment, prevention, genetic variations between individuals. Added to this is treatment, or palliation. Thus, PM aims to eliminate the the emergence of new technologies for global genomic “one size fits all” model of medicine, which has mainly cen- analysis, including high-throughput sequencing, single- tered on reaction to a disease (treating the symptoms) based nucleotide polymorphism (SNP) genotyping, and tran- on average responses to care, by shifting the emphasis of script profiling. The construction of haplotype maps of patient care to cancer prevention and early intervention the genome (2, 3) is now allowing us to view DNA in a for high-risk individuals (4, 5). Moreover, understanding much bigger picture than ever before. This, coupled with the molecular profiles of individuals and how these can cause enormous advances in computer systems and bioinfor- variations in disease susceptibility, symptoms, progression, matics, has resulted in a revolutionary shift in medical care and responses to treatment will lead to tailoring medical care to the era of personalized medicine (PM). to fit each individual patient (6-8). PM is defined by the U.S. National Cancer Institute as The objectives of this review are to provide a simplified “a form of medicine that uses information about a person's yet comprehensive overview of the scope of applications of PM and their effect on cancer patient management. We Authors' Affiliations: 1Department of Laboratory Medicine and also summarize the basic principles of the most promising Pathobiology, University of Toronto and 2Department of Laboratory approaches for PM. Finally, we address the challenges and Medicine and the Keenan Research Centre in the Li Ka Shing controversies that face our transition into an era of PM and Knowledge Institute, St. Michael's Hospital, Toronto, Canada provide a vision of how PM can be incorporated into clin- Corresponding Author: George M. Yousef, Department of Laboratory Medicine, St. Michael's Hospital, 30 Bond Street, Toronto, Ontario, ical practice. We will avoid technical details and sophistica- Canada M5B 1W8. Phone: 416-864-6060, ext. 6129; Fax: 416-864-5648. tions about specific applications when possible, referring E-mail: [email protected] the reader to more specialized articles in this regard. For doi: 10.1158/1541-7786.MCR-10-0264 a number of excellent reviews covering many aspects of ©2010 American Association for Cancer Research. PM, please refer to the April 1, 2010 issue of Nature.

www.aacrjournals.org 1175

Diamandis et al.

The Scope of Clinical Applications of The benefits of PM for the pharmaceutical industry are Personalized Medicine summarized in Fig. 2. Some of the methods currently being used for screening of cancer mutations are described in Cancer is a heterogeneous group of diseases whose causes, the succeeding paragraphs. pathogenesis, metastatic potential, and responses to treatment can be very different among individuals (9). Great Cancer screening variations exist, even between individuals with the same A combination of genetic and environmental factors is type of cancer, suggesting that genetic factors play an thought to contribute to an individual's predisposition to important role in cancer pathogenesis. These differences cancer (11). Knowledge of the precise nature of these con- make cancer an ideal target for the application of PM. tributors is important because it can affect the course of ac- The suite of clinical applications of PM in cancer is broad, tion for disease prevention (modifications to behavior and encompassing a wide variety of fields. Constituting some of lifestyle) and monitoring of high-risk patients (10). In some these clinical applications are screening, diagnosis, progno- cases, the association between genetic factors and cancer is sis, prediction of treatment efficacy, patient follow-up after very clear and has a significant effect on clinical intervention. surgery for early detection of recurrence, and the stratifica- For example, individuals with mutations in the tumor sup- tion of patients into specific smaller subgroups, thus allow- pressor genes breast cancer susceptibility gene 1 (BRCA1) ing for individualization of treatment (ref. 10; Fig. 1). and breast cancer susceptibility gene 2 (BRCA2)havea Incorporation of the concept of PM into our health care significant risk of developing breast cancer (12) and may system will allow patients and doctors to become aware of an choose to take early preventive surgical measures (e.g., pro- underlying disease state, even before clinical signs and symp- phylactic mastectomy; refs. 13, 14), undergo regular screen- toms become apparent. In turn, treatment or preventive ing, or receive adjuvant therapies (15, 16). BRCA1 and measures could be taken, even before the disease manifests, BRCA2 mutations are also associated with other cancers, in- which can delay disease onset and symptom severity. PM cluding ovarian (17), hematologic (18), and early-onset will also guide treatment decisions by stratifying patients prostate cancers (19), and knowledge about these mutations into unique subgroups based on their specific molecular may affect strategies for clinical management of these can- characteristics. Subgroups would preferentially receive tar- cers (e.g., prophylactic salpingo-oophorectomy; refs. 20, geted therapies to which they are more likely to respond, 21). Genetic testing has also been used to identify indivi- with the goal of improving response rates and survival out- duals with inherited mutations in the DNA mismatch repair comes (6). Unnecessary side effects and toxicity of treatment genes, MLH1 and MSH2, who have a higher risk of devel- will be reduced, especially in individuals predicted to be oping colon cancer (22). Knowledge of their predispositions “nonresponders” to a particular targeted therapy. Nonre- can promote early screening colonoscopy for colon cancer, sponders would be stratified into alternate subgroups for with more frequent testing, to enable early cancer detection personalized therapy. Furthermore, an expectation of who and treatment. In addition, screening for RET gene muta- will and will not respond to a particular treatment will have tions can identify individuals who are at a higher risk of de- an effect on how clinical trials are designed and carried out, veloping multiple endocrine neoplasia type 2 (23) and reducing their costs and failures. Ultimately, PM is expected allows for closer follow-up or prophylactic thyroidectomy. to result in an overall reduction in the cost of health care Public databases are constantly being updated with new (10). A summary of the currently available commercial tests mutations and polymorphisms associated with cancer for the different aspects of PM is shown in Table 1. (24). These databases can be useful resources for identifying PM can also benefit pharmaceutical companies looking new biomarkers for screening. to identify unique molecular targets for drug design. Such strategies will reduce the cost of lead discovery, reduce the Tumor classification and subtyping timeline and costs of clinical trials, and potentially revive One of the revolutionary aspects of PM is that it changes drugs that were previously thought to be ineffective (10). the traditional paradigm of classifying cancers. With PM,

FIGURE 1. Clinical applications of PM in cancer patient management.

1176 Mol Cancer Res; 8(9) September 2010 Molecular Cancer Research

Personalized Medicine in Cancer Management

Table 1. A partial list of commercial tests currently available for PM in cancer patients

Test Company Cancer type Test type Technique Application

Oncotype Dx Breast Genomic Breast Expression profile RT-PCR Predicts risk of recurrence Cancer Assay Health of a panel of and guides chemotherapy 21 genes treatment decision Oncotype Dx Colon Genomic Colon Expression profile RT-PCR Predicts recurrence and Cancer Assay Health of a panel of assists treatment decision 12 genes in stage II colon cancer MammaPrint Agendia Breast Expression of a Microarray Predicts risk of recurrence panel of 70 genes HercepTest Dako Breast c-erbB-2 IHC Predicts response to overexpression trastuzumab (Herceptin) Ventana Pathway Ventana Breast c-erbB-2 IHC Predicts response to overexpression trastuzumab (Herceptin) TheraScreen: EGFR29 DxS NSCLC EGFR29 mutation RT-PCR Predicts response to gefitinib (Iressa) and erlotinib (Tarceva) TheraScreen: KRAS DxS mCRC KRAS mutation RT-PCR Predicts response to Mutation Kit panitumumab (Vectibix) and cetuximab (Erbitux) CYP450 Test Amplichip Breast Identify CYP2D6 Microarray Predicts response to tamoxifen and CYP2C19 and determines optimal genotype treatment dose BCR-ABL Mutation Genzyme CML T315I mutation RT-PCR Predicts response to imatinib Analysis Test Genetics (Gleevec) EGFR Amplification Genzyme CRC EGFR amplification FISH Predicts response to cetuximab Test Genetics (Erbitux) and panitumumab (Vectibix) EGFR Amplification Genzyme NSCLC EGFR amplification FISH Predicts response to gefitinib Test Genetics (Iressa) and erlotinib (Tarceva) ALK Gene Genzyme NSCLC ALK gene FISH Predicts response to Rearrangement Test Genetics rearrangement erlotinib (Tarceva) EGFR Mutation Genzyme NSCLC EGFR Mutation RT-PCR Predicts response to tyrosine Genetics kinase inhibitors

Abbreviations: RT-PCR, real-time PCR; IHC, immunohistochemistry; NSCLC, non–small cell lung cancer; mCRC, metastatic colorectal cancer; CML, chronic myelogenous leukemia; CRC, colorectal cancer; FISH, fluorescence in situ hybridization.

pathologic classification can shift from the histologic scale, tion of cancers into unique molecular groups with distinct which often gives little information on prognosis, individu- prognosisortreatmentoptionswillsignificantlyimprove alized treatment options, and chance of recurrence, over- patient management. looking the fact that many patients with similar histologic types might experience markedly different disease courses, Targeted therapy and predictive markers for to the molecular scale, which offers a highly detailed, global treatment efficiency perspective of the disease process and promises superior per- The ultimate goal of PM is to define the disease at the formance over traditional classifications (25-28). Molecular molecular level so that therapy can be directed at the right analyses at the protein, DNA, RNA, or microRNA population of patients (the predicted “responders”). Thus, (miRNA) levels can contribute to the identification of novel elucidating the molecular differences between individuals tumor subclasses, each with unique prognostic outcome or will have an impact on how new treatments are developed response to treatment, which could not be identified by tra- and would be of great benefit to pharmaceutical companies ditional morphologic methods (10). Recently, molecular looking to market new cancer drugs. It is now known that classification has been used to identify unique subclasses even in individuals with similar clinical phenotypes (clini- of cancers, including acute myeloid leukemia (29, 30), cal manifestations and pathologic type), drug therapy is glioblastoma (31), breast cancer (32, 33), and renal cell usually only effective in a fraction of those treated, suggest- carcinoma (34, 35), and to differentiate between Burkitt's ing that other differences account for individual drug lymphoma and diffuse B-cell lymphoma (36). Subclassifica- responses (7). Being able to predict which patients will

www.aacrjournals.org Mol Cancer Res; 8(9) September 2010 1177

Diamandis et al.

FIGURE 2. Benefits of PM for the pharmaceutical industry.

benefit from a drug would promote a shift to identify development of new or combination treatments for other alternative therapies to improve outcomes in the predicted cancers based on an understanding of their pathogenesis “nonresponders,” in turn reducing the administration of (47). Additionally, old drugs that may have been dismissed costly and potentially toxic treatments to them. Further- as ineffective in a broader category of cancer patients, or more, clinical trials of new drugs could focus on enrolling drugs being used to treat other diseases, could be reintro- patient populations in which treatment is predicted to be duced into clinical trials that are focused on a specific most efficacious, thus increasing the likelihood of positive cancer subgroup to investigate novel effects. outcomes (7). The concept of PM gained much positive momentum Pharmacogenomics and treatment safety from the development of highly successful U.S. Food and Genetic variations can also predict treatment dose and Drug Administration–approved targeted therapies, such as safety in different subgroups of cancer patients. Variations imatinib mesylate (Gleevec) for chronic myeloid leukemia in genes that encode drug-metabolizing enzymes, drug and gastrointestinal stromal tumor (37, 38) and trastuzu- transporters, or drug targets can have detrimental effects mab (Herceptin) for breast cancer (39). Using specific mo- on patient outcomes following treatment (48). For exam- lecular characteristics of these cancers as predictive markers ple, polymorphisms in cytochrome P450 (CYP) enzymes for treatment response (abnormal protein tyrosine kinase ac- can result in too slow or very fast metabolism of drugs, tivity in chronic myeloid leukemia and gastrointestinal stro- causing patients to exhibit symptoms of overdose or have mal tumor and overexpression of the HER-2 receptor in no drug response at all (49, 50) by altering the pharmaco- breast cancer), only those individuals who harbor the appro- kinetics of drug metabolism and distribution (48), and priate molecular alterations are eligible for treatment. It is could also account for the common occurrence of adverse this subclassification of cancers, and the consequent custo- drug reactions (51). Thus, these polymorphisms could be mization of therapy, that has resulted in the remarkable used to predict optimal drug dosage, minimize harmful surge in survival rates for some cancers from zero to 70% (7). side effects, reduce the costs of “trial-and-error dosing,” This application of PM has been further shown by the use and ensure more successful outcomes (50). In familial of mutation screening in the treatment of patients with lung breast cancer, for example, genetic variations in CYP2D6 cancer who received the tyrosine kinase inhibitors gefitinib that confer “poor metabolizer” status have been shown to (Iressa) or erlotinib (Tarceva; ref. 40). Non–small cell lung reduce survival in individuals treated with the chemother- cancers with mutations in the kinase domain of epidermal apeutic drug tamoxifen (52). Genetic variability in P- growth factor receptor (EGFR) are much more responsive to glycoprotein,adrugtransporter,affectsthepharmacokinetics treatment with gefitinib/erlotinib, and more specifically, of the antineoplastic drug paclitaxel in ovarian cancer and is these drugs are especially effective in subgroups of Asian also associated with clinical outcomes (53, 54). The transi- female never-smokers with adenocarcinomas (41, 42). In tion to PM already has some pharmaceutical companies contrast, individuals with mutations in the downstream ef- placing information about genetic testing (including testing fector KRAS are resistant to erlotinib (43). Activating muta- for CYP450 polymorphisms) on drug labels (8, 55). tions in KRAS also cause resistance to the drugs cetuximab (Erbitux) and panitumumab (Vectibix) in colon cancer pa- Prognosis tients, whereas these drugs are effective in cancers with wild- Until recently, the clinicopathologic parameters of the type KRAS (44, 45). Specific responses to drugs based on patient (mainly tumor type, grade, and stage), along with one's molecular profile has led to the recent recommenda- biochemical testing for few tumor markers, were the sole tion by the American Society of Clinical Oncology that indicators for prognosis and tumor aggressiveness and were molecular testing of KRAS be done before the administration consequently used for therapeutic decision-making for can- of cetuximab or panitumumab to colon cancer patients (46). cer patients. However, it has become increasingly clear that Insurance companies have even offered to pay for prognostic accuracy may greatly be improved by under- medication for those individuals with wild-type KRAS standing the molecular basis of the different cancer sub- (knowing they will respond well to therapy), adding a layer types. Genotyping or gene expression profiling by of complexity to the concept of PM. Targeted therapy, as microarray (56) and protein analysis by mass spectrometry in the cases of lung and colon cancers, highlights the po- (discussed in the following sections; ref. 57) have already tential for PM to guide patient care and gives hope for the been used to identify numerous prognostic biomarkers.

1178 Mol Cancer Res; 8(9) September 2010 Molecular Cancer Research

Personalized Medicine in Cancer Management

Such markers can help, either alone or in combination High-throughput whole genome sequencing with classic parameters, to substratify patients into smaller The completion of the human genome project has moti- distinct prognostic risk groups and guide therapy decisions. vated the development of “next-generation sequencing” tech- Using tissue microarray analysis, Kim et al. (58) con- nologies (61). High-throughput analyzers from Roche (454 structed a combined molecular and clinical prognostic Genome Analyzer), Life Technologies (SOLiD analyzer), model for survival that was significantly more accurate and Illumina (Genome Analyzer IIe and HiSeq 2000) now than standard clinical parameters for patients with renal make it possible to analyze millions of nucleotides at once, cell carcinoma. The Oncotype Dx Breast Cancer Assay, at a lower cost compared with the traditional Sanger method. which gives a gene expression profile of 21 genes known These analyzers use pyrosequencing (62) or sequencing by to be involved in breast cancer, has also been shown to ligation (61), rather than the traditional Sanger end-chain be a good predictor of the risk of tumor recurrence (59). termination method. Table 2 summarizes the commonly used high-throughput sequencers with their features and applications for PM. The advantages of high-throughput sequencing Approaches and Tools for Personalized include the capacity to perform multiplex reactions, reduced Medicine operating costs, and very fast acquisition of data. Next-generation sequencing is now being used for global Figure 3 summarizes the different approaches and tools analysis of the genome. Previously, some cancer alleles could available for molecular PM testing. A large array of techni- not be detected by Sanger sequencing because they were ques can be used for PM. Commonly used techniques present in extremely low levels in cells. Now, the use of “deep include PCR, fluorescence in situ hybridization, immuno- sequencing” (extensive repeated coverage of the sequence of histochemistry, and sequencing. More recently, the com- interest; ref. 63) and paired-end sequencing (64) has made pletion of the human genome project opened a new their identification possible, thus expanding our under- horizon for PM analysis by using high-throughput analysis standing of the cancer genome. Laser capture microdissec- (60). These include microarray, mass spectrometry, second- tion (65) has also been useful for isolating and enriching generation sequencing, array comparative genomic hybrid- cancer cell DNA and RNA obtained from a tissue sample, ization, and high-throughput SNP analysis, among others. which can then be used for targeted sequencing. These A common trend among these tools is their ability to si- methods enabled the identification of novel mutations, re- multaneously analyze hundreds or thousands of targets. arrangements, and other genomic alterations (66-68) that This multiparametric approach is likely to improve the lead to tumorigenesis in acute myeloid leukemia (69, 70), sensitivity, specificity, and accuracy of new biomarkers chronic myeloid leukemia (67), and other cancers (71-74). (35). In addition to acceleration of biomarker discovery, Methods for high-throughput sequencing continue high-throughput analysis allows a better understanding of to evolve. Biotechnology companies such as Complete the “cross-talk” or interaction between different molecules Genomics, Helicos, and Pacific Biosciences are now work- in the pathogenesis of cancer. In the following discussions, ing on “third-generation sequencing” methods, which are we will highlight some of the commonly used techniques expected to further reduce cost, increase accuracy and length for global analysis. of reads, and provide even higher-throughput analysis of

FIGURE 3. A schematic representing different approaches available for PM molecular testing. A number of available biological materials including DNA, RNA, miRNA, and protein can be isolated from either fresh-frozen or formalin-fixed paraffin-embedded (FFPE) tissues, blood, or other bodily fluids and used for molecular testing. Common techniques include tissue microarray (TMA), SNP analysis, array comparative genomic hybridization (CGH), immunohistochemistry (IHC), fluorescence in situ hybridization (FISH), and quantitative real-time PCR (qRT-PCR).

www.aacrjournals.org Mol Cancer Res; 8(9) September 2010 1179

Diamandis et al.

Table 2. Commercial analyzers for second-generation sequencing

Analyzer Company Sequencing method Throughput Applications

454 Genome Roche Pyrosequencing >1 million high-quality De novo sequencing of Sequencer reads per run and read whole genomes and lengths of 400 bases transcriptomes of any size per 10-h instrument run Genome Illumina Reversible terminator- Combination of 2 × 100 bp Genome Analyzer IIe based sequencing read length and up to 500 DNA sequencing million reads per flow cell Gene regulation analysis Epigenome ChIP-Seq Methylation analysis Transcriptome Transcriptome analysis SNP discovery and structural variation analysis Small RNA discovery and analysis Cytogenetic analysis HiSeq 2000 Illumina Reversible terminator- ∼30× coverage of two Genome based sequencing human genomes in a DNA sequencing single run Gene regulation analysis Epigenome ChIP-Seq Methylation analysis Transcriptome Transcriptome analysis SNP discovery and structural variation analysis Small RNA discovery and analysis Cytogenetic analysis SOLiD Applied Biosystems Sequential ligation 100 Gb and 1.4 billion Genome tags per run (extendable De novo sequencing to 300 Gb and Targeted resequencing 2.4 billion tags per run) Whole genome sequencing Epigenome ChIP-Seq Methylation analysis Transcriptome Gene expression profiling Small RNA analysis Whole transcriptome profiling

Abbreviation: ChIP-Seq, chromatin immunoprecipitation sequencing.

genomes (61, 75). As the reality of faster and cheaper high- There is no doubt that whole genome sequencing is a throughput analyses is becoming closer, several biotechnol- powerful tool for digging deep into the genetic code to ogy companies are racing to commercialize full genome look for influences on disease; however, with such a tool sequencing, which is expected within the next few years comes the challenge of determining what information is (68, 75). Although currently too expensive (∼$48,000 most important for analysis and interpretation, and this USD for a full genome by Illumina), it is estimated that is one challenge that will have to be addressed as we move the era of the “$1,000” genome is fast approaching (76). forward into the era of PM, as discussed later.

1180 Mol Cancer Res; 8(9) September 2010 Molecular Cancer Research

Personalized Medicine in Cancer Management

SNP analysis and haplotype mapping (115). Proteomic analysis can be advantageous over mRNA The human genome is estimated to have more than 30 measurements because proteins are the final effector mole- million SNPs, which are essentially the “fingerprints” of cules and their levels do not always correlate with mRNA our genetic code (77, 78). Some of these have been thor- levels due to posttranscriptional modifications (116). Fur- oughly characterized by the International Haplotype Map- thermore, protein-protein interactions are important contri- ping Project (2, 3) in various populations and made publicly butors to cellular pathways and mechanisms that contribute available (79). These databases have provided the necessary to carcinogenesis. In mass spectrometry, proteins are ionized tools for researchers to discover associations between disease into smaller molecules with unique mass-to-charge ratios risk and common SNPs (80). The advent of commercially that can be used to quantify proteins (117). This has led available microarrays (SNP chips; ref. 81) has resulted in a to the identification of several new cancer biomarkers for shift from studying disease by linkage analysis to using ge- different malignancies, including breast (118), ovarian nome-wide association studies (82). SNP arrays use allele- (119), prostate (120), and kidney (121, 122) cancers. Pro- specific oligonucleotide probes that produce a fluorescent teomics can contribute to tumor classification (123), treat- signal when a specific allele of a SNP is present and are ca- ment selection, pharmacoproteomics, and identification of pable of analyzing up to 1 million SNPs in a single sample new drug targets and may also be used for therapeutic drug (83, 84). Alternatively, SNP haplotypes, rather than single monitoring (124-126). Details of the principles of mass SNPs, can be analyzed. SNP arrays can also be used for spectrometry are beyond the scope of this review and have screening for common features of cancer genomes, such as recently been covered in a number of excellent reviews. allelic imbalance, copy number variation, or loss of hetero- zygosity. SNP arrays have been used extensively to assess var- Genome-wide association studies ious aspects of cancer, including risk assessment (85-87), Recent technological advances have allowed the exam- prognosis (88), survival (89), response to therapy (90), ination of genetic variations on a much larger scale than and progression and metastasis (91-93). ever before. Genome-wide association studies (GWAS) allow for a complete picture of the genome rather than Microarray analysis the smaller sections to which we were previously limited. Microarrays are widely used for global analysis of gene ex- High-throughput approaches allow for an unbiased pression in cancer because they are cost-effective and have examination of genetic variations in many different high throughput (94). Microarrays are chips with immobi- tumors. There are a number of ongoing GWAS and lized capture molecules (such as oligonucleotides or cDNAs) some of them are listed in Table 3. The Cancer Genetic that serve as probes for binding fluorescently labeled targets Markers of Susceptibility project, an initiative of the Na- (e.g., cDNA) prepared from the two specimens to be tional Cancer Institute, aims to identify genes involved compared (e.g., normal versus cancer). Using microarrays, in breast and prostate cancers by SNP analysis. The Hu- it is possible to assess the expression levels of thousands of man Cancer Genome Project will examine not only mu- genes in a single experiment. There are several platforms tations but also other genetic abnormalities, such as gene for microarray analysis, including the most popular mRNA silencing, through methylation and other epigenetic me- microarray, miRNA microarrays, DNA arrays (array com- chanisms, as well as gene translocation, amplifications, parative genomic hybridization), and protein arrays. and deletions (127). Microarrays have been extensively used to study the differ- GWAS are all driven by the quest for future PM. ent aspects of malignancy, especially at the discovery phase. Recent results of GWAS have identified 6q25.1 as a Gene expression profiling has been successfully used for the susceptibility locus for breast cancer (128) and two inde- detection of cancer to categorize different subtypes of a can- pendent loci within 8q24 that contribute to prostate cer (95-98), identify invasive versus noninvasive phenotypes cancer in men of European ancestry (129, 130). GWAS (99), predict prognosis (95, 96, 100-102), and predict have also indicated clear differences in susceptibility be- responses to treatment (103-106) and early recurrence (107). tween cancer subtypes. For example, a lung cancer locus Newer microarray platforms, such as miRNA microar- identified on 5p15.33 was strongly associated with adeno- rays, are also showing encouraging preliminary data for their carcinoma but not squamous or other subtypes (131). potential use as cancer biomarkers (108-113). MiRNA sig- Also, aberrations in the ovarian cancer locus BNC2 were natures have been used to stratify patients into prognostic associated with the serous subtype only (132). Recent cohorts and treatment subgroups. Finally, microarrays can findings indicate that certain mutations can predict pa- be used to assess epigenetic changes in cancer (DNA meth- tient response to treatment, as was shown for certain ylation, histone acetylation, serine phosphorylation), which EGFR mutations that can predict patient response to are implicated in tumorigenesis, and can be used to classify the drug gefitinib (Iressa). Recently, a genome-wide scan cancers and direct patient management (114). of SNPs found 20 SNPs that were associated with the effectiveness of platinum-based chemotherapy in small- Proteomics by mass spectrometry cell lung cancer patients (133). Also, a genome-wide Changes in the protein profiles of cancer cells can be im- methylation analysis in mantle cell lymphoma patients portant for determining new diagnostic biomarkers and may found that differentially methylated genes can be targeted also aid in the classification of tumors into unique subtypes for therapeutic benefit (134). Although the use of GWAS

www.aacrjournals.org Mol Cancer Res; 8(9) September 2010 1181

Diamandis et al.

has discovered many genetic loci and SNPs that are relat- patients. Care must be taken when interpreting published ed to cancer, more studies are required to identify how results due to the heterogeneity of the analyzed material, these alterations contribute to disease risk (135). which can be obtained from tissues, cell lines, and biological fluids, and when comparing different histologic types, Databases/bioinformatics stages, and grades of tumors. An increasingly important new tool to achieve PM is the Another important issue is the need for standardization availability of repositories of freely accessible databases. of testing. Standardization encompasses several aspects Information obtained from genomic, transcriptomic, and including the type of specimen to be analyzed, the appro- proteomic analyses is now deposited into large databases priate methods of specimen collection and storage, the for global access and data processing. Some of these choice of the target genes/proteins to be tested, the plat- include the Cancer Genome Anatomy Project (136), form to be used, optimal experimental conditions, and the Single Nucleotide Polymorphism Database (79), the the clinical interpretation of tests (cutoff values for positive Catalogue of Somatic Mutations in Cancer (24), the and negative results). There will be a need for quality Mittleman Database of Chromosomal Aberrations in standards for different laboratories, and test validations Cancer (137), the Stanford Microarray Database (138), will be required with external quality assurance protocols. the Roche Cancer Genome Database (139), the Protein It may also be necessary to implement Clinical Laboratory Information Resource (140), and the Human Protein Improvement Amendments–certified specialized laborato- Reference Database (141), in addition to many others. ries to perform particular tests to ensure quality. A recent Bioinformatic algorithms can then be used to integrate a report by the National Academy of Clinical Biochemistry patient's clinical information and the genetic profiles of Laboratory Medicine Practice Guidelines expounded on their tumor to predict the relationships of certain molecu- two main technologies commonly implemented for PM, lar changes to cancer, as discussed earlier. microarray and mass spectrometry, and developed recom- mendations on what must be done before for their appli- Personalized Medicine in the Clinic: Challenges cation into the clinical realm. and Controversies One of the most important considerations will be the higher cost of incorporating new, advanced technologies As we move into the era of PM, several challenges have to thatwillprovidemoredetailedpredictiveandprognostic be addressed. The first is to establish a significant clinical information about each individual patient. Setting up the utility that is superior to our existing parameters, and this infrastructure for PM will require significant spending. will lead to a “real” improvement in cancer patient care. This Prices are, however, becoming much more affordable com- requires a team approach, with collaborative efforts between pared with earlier years, as technology becomes more wide- clinicians, research scientists, computer experts, and biosta- spread. Following an initial capital investment, additional tisticians, to achieve a clinically meaningful outcome. Full running costs will become much less. The increased cost transparency in reporting results (especially the negative of these advanced diagnostic systems will be partially com- ones) should be emphasized to avoid selection bias for pos- pensated for by savings in patient care cost, such as reduced itive results reporting (35). It also has to be noted in this re- courses of unnecessary chemotherapy or radiotherapy and gard that statistically significant results in a research setting reduced hospital stays. are not always equivalent to clinically significant results. Ethical issues, including who will have access to this new Prospective trials on large patient cohorts are needed to so- detailed information, especially those at risk of developing lidify the clinical utility of new tests being offered to cancer cancer, should be adequately addressed. Individuals should

Table 3. A partial list of genome-wide association studies

Project Objectives Reference

Cancer Genetic Markers of Susceptibility Identify inherited susceptibility to prostate (128-130, 148) and breast cancers by examining SNPs The Consensus Coding Sequence of Systemic genome-wide scan of the coding (149) Human Breast and Colorectal Cancers sequence of breast and colorectal cancers The Cancer Genome Atlas Identify all functional gene mutations and (127) other abnormalities in common tumor types The International Cancer Genome Consortium Systematic study of more than 25,000 (150) cancer genomes at the genomic, epigenomic, and transcriptomic levels Human Proteome Project Catalogue and characterize all proteins (151) in the human body

1182 Mol Cancer Res; 8(9) September 2010 Molecular Cancer Research

Personalized Medicine in Cancer Management

FIGURE 4. A schematic representing the future use of PM through the course of a cancer patients' disease. Individual molecular analysis can be applied at each step to ensure that the patient receives the most appropriate and optimal treatment. The benefits of PM can also help predict which family members in high-risk groups may develop malignancy. be free to accept or decline susceptibility testing. The would replace the clinicopathologic designation system and access for insurance companies, other family members, would provide many pieces of information related to treat- employers, and other agents will need to be clearly stated ment simultaneously. This was, however, followed by a and new laws may have to be passed to ensure fair treat- dormant period when promising research results did not ment, protection, and privacy of the tested individuals. cross the bridge to be adopted into clinical applications. Misuse of genetic information could take place at the One important reason for this is that the new classifiers expense of those tested. Patients with known risks for created new “categories” of patients for which the clinical developing cancer, those who are expected to have poor significance was unknown. One famous example of this is outcomes, or those who will not respond well to treat- the basal-like phenotype of breast cancer that was created ments may be discriminated against or denied employment by microarray analysis (143). There are important reasons opportunities or health insurance coverage. Individuals as to why there has not been a rapid change in the who are expected to not respond to treatment may also approach to cancer with the introduction of PM. The be denied certain medications. Also, the cost of targeted process of developing a tumor marker for “clinical” use is therapies is high. In this case, although genetic testing a tedious and multistep process requiring extensive testing, may indicate one would receive benefits from a certain optimization, validation, and establishment of usefulness drug, they may have limited access due to wealth or insur- above other currently used methods (57). ance status. In this case, PM will add to the problem of After years of experience, we developed a more practical socioeconomic divisions. Furthermore, genetic testing view of using molecular genetic information to comple- may reveal traits for other serious illnesses (regardless of ment, rather than replace, clinicopathologic parameters. whether an individual chooses to find out about them or Molecular profiling will be able to answer specific focused not), which may also affect how they are treated by others. questions related to patient management for each particu- Good policy decisions will be crucial to reap the benefits of lar cancer. In some respects, great leaps have been taken PM (142). Luckily, in the United States, laws are already in with the introduction of PM approaches to cancer. Several place to protect patients from these types of issues (8). biomarkers have been very useful and are currently being marketed, as described earlier. Personalized Medicine: Fantasy or Reality? A Future Prospective The introduction of the concept of PM initially created a lot of promise in the scientific community. There was op- It is clear from recent literature that the concept of PM timism for an imminent revolution in our medical practice will have a significant impact on medical practice and is whereby a molecular “fingerprint” for each cancer patient gradually being included as an integral component of our

www.aacrjournals.org Mol Cancer Res; 8(9) September 2010 1183

Diamandis et al.

cancer management plans. It will inevitably result in more PM will be performed hand in hand with the usual histo- effective treatment and fewer side effects for patients and pathologic evaluation to bring together more specific de- will induce a proactive and participatory role for the patient tails about every individual cancer, including assessment in their own health care decisions. Patients will have the in- of risk, aggressiveness, and treatment options. This tumor centive to engage in lifestyle choices and health maintenance “fingerprint” will reduce unnecessary treatments and could to compensate for their genetic susceptibilities. also extend to other family members who are at risk of Before the era of PM, cancer diagnosis, prognosis, and developing the same or a related malignancy. subsequent treatment decisions were based on histopatho- Following the establishment of molecular fingerprints, logic parameters including the tissue of origin and the stage the examination and incorporation of one's phenotype into and grade of the tumor. Experience has shown morpholog- individual disease assessment must be considered (144). Of ic classification to be deficient in many aspects and that course, this comes with its own challenges including the patients with the same histopathologic diagnosis can have requirement of super-computers that can analyze these unexplained variable outcomes. Individual molecular mar- enormous amounts of data for one person. Integrating kers have been slowly added to ameliorate the accuracy of data from cancer genetics and functional data into molec- predicting prognosis and treatment efficiency. We now wit- ular networks could potentially approach a genotype- ness the accumulation of enormous amounts of genomic phenotype map (145-147). Although this may be a reality data generated by high-throughput analyses, which are in years to come, we need to approach PM in a stepwise augmented by tremendous advances in computer analytic manner and remember that the long-tem benefits are systems. PM is a revolutionary concept that challenges yet to come. our traditional classification and management of cancer. Although very appealing, it is unlikely that it will com- Disclosure of Potential Conflicts of Interest pletely replace traditional approaches, at least in the near future. It will be essential to confirm the validity of new No potential conflicts of interest were disclosed. biomarkers generated by PM technologies using prospective clinical trials in independent patient cohorts. This will Grant Support be an ongoing process that is expected to extend for many years to come. G.M. Yousef was supported by grants from the Canadian Institute of Health Research (grant 86490), the Canadian Cancer Society (grant 20185), and the The introduction of PM requires a very big initial invest- Ministry of Research and Innovation, Government of Ontario. ment, but with it comes the promise of a rewarding and The costs of publication of this article were defrayed in part by the payment of cost-effective future for medical practice. Figure 4 shows page charges. This article must therefore be hereby marked advertisement in one possible future scenario where molecular analysis can accordance with 18 U.S.C. Section 1734 solely to indicate this fact. be incorporated with the various steps of cancer patient Received 06/15/2010; revised 07/30/2010; accepted 08/02/2010; published management, from early detection to time of treatment. OnlineFirst 08/06/2010.

References 1. Finishing the euchromatic sequence of the human genome. Nature 13. Ziogas D, Roukos DH. Genetics and personal genomics for person- 2004;431:931–45. alized breast cancer surgery: progress and challenges in research 2. A haplotype map of the human genome. Nature 2005;437:1299–320. and clinical practice. Ann Surg Oncol 2009;16:1771–82. 3. Frazer KA, Ballinger DG, Cox DR, et al. A second generation human 14. Roukos DH, Briasoulis E. Individualized preventive and therapeutic haplotype map of over 3.1 million SNPs. Nature 2007;449:851–61. management of hereditary breast ovarian cancer syndrome. Nat 4. Abrahams E, Ginsburg GS, Silver M. The Personalized Medicine Clin Pract Oncol 2007;4:578–90. Coalition: goals and strategies. Am J Pharmacogenomics 2005;5: 15. Gronwald J, Tung N, Foulkes WD, et al. Tamoxifen and contralateral 345–55. breast cancer in BRCA1 and BRCA2 carriers: an update. Int J Can- 5. Abrahams E, Silver M. The case for personalized medicine. J Dia- cer 2006;118:2281–4. betes Sci Technol 2009;3:680–4. 16. Byrski T, Huzarski T, Dent R, et al. Response to neoadjuvant ther- 6. Ginsburg GS, McCarthy JJ. Personalized medicine: revolutionizing apy with cisplatin in BRCA1-positive breast cancer patients. Breast drug discovery and patient care. Trends Biotechnol 2001;19:491–6. Cancer Res Treat 2009;115:359–63. 7. Allison M. Is personalized medicine finally arriving? Nat Biotechnol 17. Narod SA, Boyd J. Current understanding of the epidemiology and 2008;26:509–17. clinical implications of BRCA1 and BRCA2 mutations for ovarian 8. Ginsburg GS, Willard HF. Genomic and personalized medicine: cancer. Curr Opin Obstet Gynecol 2002;14:19–26. foundations and applications. Transl Res 2009;154:277–87. 18. Friedenson B. The BRCA1/2 pathway prevents hematologic can- 9. Fidler IJ. Tumor heterogeneity and the biology of cancer invasion cers in addition to breast and ovarian cancers. BMC Cancer and metastasis. Cancer Res 1978;38:2651–60. 2007;7:152. 10. Overdevest JB, Theodorescu D, Lee JK. Utilizing the molecular 19. Gallagher DJ, Gaudet MM, Pal P, et al. Germline BRCA mutations gateway: the path to personalized cancer management. Clin Chem denote a clinicopathologic subset of prostate cancer. Clin Cancer 2009;55:684–97. Res 2010;16:2115–21. 11. Perera FP. Environment and cancer: who are susceptible? Science 20. Kauff ND, Domchek SM, Friebel TM, et al. Risk-reducing salpingo- 1997;278:1068–73. oophorectomy for the prevention of BRCA1- and BRCA2-associated 12. Kadouri L, Hubert A, Rotenberg Y, et al. Cancer risks in carriers of breast and gynecologic cancer: a multicenter, prospective study. the BRCA1/2 Ashkenazi founder mutations. J Med Genet 2007;44: J Clin Oncol 2008;26:1331–7. 467–71. 21. Fan S, Meng Q, Auborn K, Carter T, Rosen EM. BRCA1 and BRCA2

1184 Mol Cancer Res; 8(9) September 2010 Molecular Cancer Research

Personalized Medicine in Cancer Management

as molecular targets for phytochemicals indole-3-carbinol and non-small cell lung cancer treated with epidermal growth factor re- genistein in breast and prostate cancer cells. Br J Cancer 2006; ceptor tyrosine kinase inhibitors. Clin Cancer Res 2006;12:4416–20s. 94:407–26. 43. Massarelli E, Varella-Garcia M, Tang X, et al. KRAS mutation is an 22. Peltomaki P, de la Chapelle A. Mutations predisposing to hereditary important predictor of resistance to therapy with epidermal growth nonpolyposis colorectal cancer. Adv Cancer Res 1997;71:93–119. factor receptor tyrosine kinase inhibitors in non-small-cell lung 23. Lips CJ, Landsvater RM, Hoppener JW, et al. Clinical screening as cancer. Clin Cancer Res 2007;13:2890–6. compared with DNA analysis in families with multiple endocrine 44. Amado RG, Wolf M, Peeters M, et al. Wild-type KRAS is required for neoplasia type 2A. N Engl J Med 1994;331:828–35. panitumumab efficacy in patients with metastatic colorectal cancer. 24. Forbes SA, Tang G, Bindal N, et al. COSMIC (the Catalogue of J Clin Oncol 2008;26:1626–34. Somatic Mutations in Cancer): a resource to investigate acquired 45. Lievre A, Bachet JB, Le CD, et al. KRAS mutation status is predic- mutations in human cancer. Nucleic Acids Res 2010;38:D652–7. tive of response to cetuximab therapy in colorectal cancer. Cancer 25. Girard N, Ostrovnaya I, Lau C, et al. Genomic and mutational Res 2006;66:3992–5. profiling to assess clonal relationships between multiple non-small 46. Allegra CJ, Jessup JM, Somerfield MR, et al. American Society of cell lung cancers. Clin Cancer Res 2009;15:5184–90. Clinical Oncology provisional clinical opinion: testing for KRAS 26. Cordes B, Williams MD, Tirado Y, et al. Molecular and phenotypic gene mutations in patients with metastatic colorectal carcinoma analysis of poorly differentiated sinonasal neoplasms: an integrated to predict response to anti-epidermal growth factor receptor mono- approach for early diagnosis and classification. Hum Pathol 2009; clonal antibody therapy. J Clin Oncol 2009;27:2091–6. 40:283–92. 47. Laurent-Puig P, Taieb J. Lessons from Tarceva in pancreatic cancer: 27. Davicioni E, Anderson MJ, Finckenstein FG, et al. Molecular classi- where are we now, and how should future trials be designed in fication of rhabdomyosarcoma-genotypic and phenotypic determi- pancreatic cancer? Curr Opin Oncol 2008;20:454–8. nants of diagnosis: a report from the Children's Oncology Group. 48. Evans WE, Relling MV. Pharmacogenomics: translating functional Am J Pathol 2009;174:550–64. genomics into rational therapeutics. Science 1999;286:487–91. 28. de Leval L, Gaulard P. Pathobiology and molecular profiling of 49. Rodriguez-Antona C, Gomez A, Karlgren M, Sim SC, Ingelman- peripheral T-cell lymphomas. Hematol Am Soc Hematol Educ Sundberg M. Molecular genetics and epigenetics of the cyto- Program 2008:272–9. chrome P450 gene family and its relevance for cancer risk and 29. Bullinger L, Valk PJ. Gene expression profiling in acute myeloid leu- treatment. Hum Genet 2010;127:1–17. kemia. J Clin Oncol 2005;23:6296–305. 50. Ingelman-Sundberg M. Polymorphism of cytochrome P450 and xe- 30. Silva FP, Swagemakers SM, Erpelinck-Verschueren C, et al. Gene nobiotic toxicity. Toxicology 2002;181–182:447–52. expression profiling of minimally differentiated acute myeloid leuke- 51. Lazarou J, Pomeranz BH, Corey PN. Incidence of adverse drug re- mia: M0 is a distinct entity subdivided by RUNX1 mutation status. actions in hospitalized patients: a meta-analysis of prospective Blood 2009;114:3001–7. studies. JAMA 1998;279:1200–5. 31. Verhaak RG, Hoadley KA, Purdom E, et al. Integrated genomic 52. Newman WG, Hadfield KD, Latif A, et al. Impaired tamoxifen metab- analysis identifies clinically relevant subtypes of glioblastoma olism reduces survival in familial breast cancer patients. Clin Cancer characterized by abnormalities in PDGFRA, IDH1, EGFR, NF1. Res 2008;14:5913–8. Cancer Cell 2010;17:98–110. 53. Yamaguchi H, Hishinuma T, Endo N, et al. Genetic variation in 32. Wiechmann L, Sampson M, Stempel M, et al. Presenting features of ABCB1 influences paclitaxel pharmacokinetics in Japanese pa- breast cancer differ by molecular subtype. Ann Surg Oncol 2009;16: tients with ovarian cancer. Int J Gynecol Cancer 2006;16:979–85. 2705–10. 54. Johnatty SE, Beesley J, Paul J, et al. ABCB1 (MDR 1) polymorph- 33. Normanno N, De Luca A, Carotenuto P, Lamura L, Oliva I, D'Alessio isms and progression-free survival among women with ovarian A. Prognostic applications of gene expression signatures in breast cancer following paclitaxel/carboplatin chemotherapy. Clin Cancer cancer. Oncology 2009;77 Suppl 1:2–8. Res 2008;14:5594–601. 34. Dahinden C, Ingold B, Wild P, et al. Mining tissue microarray data to 55. Frueh FW, Amur S, Mummaneni P, et al. Pharmacogenomic bio- uncover combinations of biomarker expression patterns that marker information in drug labels approved by the United States improve intermediate staging and grading of clear cell renal cell food and drug administration: prevalence of related drug use. cancer. Clin Cancer Res 2010;16:88–98. Pharmacotherapy 2008;28:992–8. 35. Arsanious A, Bjarnason GA, Yousef GM. From bench to bedside: 56. Macgregor PF, Squire JA. Application of microarrays to the analysis current and future applications of molecular profiling in renal cell of gene expression in cancer. Clin Chem 2002;48:1170–7. carcinoma. Mol Cancer 2009;8:20. 57. Kulasingam V, Diamandis EP. Strategies for discovering novel 36. Dave SS, Fu K, Wright GW, et al. Molecular diagnosis of Burkitt's cancer biomarkers through utilization of emerging technologies. lymphoma. N Engl J Med 2006;354:2431–42. Nat Clin Pract Oncol 2008;5:588–99. 37. Druker BJ, Talpaz M, Resta DJ, et al. Efficacy and safety of a spe- 58. Kim HJ, Shen SS, Ayala AG, et al. Virtual-karyotyping with SNP mi- cific inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid croarrays in morphologically challenging renal cell neoplasms: a leukemia. N Engl J Med 2001;344:1031–7. practical and useful diagnostic modality. Am J Surg Pathol 2009; 38. Tuveson DA, Willis NA, Jacks T, et al. STI571 inactivation of the 33:1276–86. gastrointestinal stromal tumor c-KIT oncoprotein: biological and 59. Conlin AK, Seidman AD. Use of the Oncotype DX 21-gene assay to clinical implications. Oncogene 2001;20:5054–8. guide adjuvant decision making in early-stage breast cancer. Mol 39. Baselga J, Norton L, Albanell J, Kim YM, Mendelsohn J. Recombi- Diagn Ther 2007;11:355–60. nant humanized anti-HER2 antibody (Herceptin) enhances the anti- 60. Wheeler DA, Srinivasan M, Egholm M, et al. The complete genome tumor activity of paclitaxel and doxorubicin against HER2/neu of an individual by massively parallel DNA sequencing. Nature overexpressing human breast cancer xenografts. Cancer Res 2008;452:872–6. 1998;58:2825–31. 61. Schuster SC. Next-generation sequencing transforms today's biol- 40. Paez JG, Janne PA, Lee JC, et al. EGFR mutations in lung cancer: ogy. Nat Methods 2008;5:16–8. correlation with clinical response to gefitinib therapy. Science 2004; 62. Ronaghi M, Karamohamed S, Pettersson B, Uhlen M, Nyren P. Real- 304:1497–500. time DNA sequencing using detection of pyrophosphate release. 41. Pao W, Miller V, Zakowski M, et al. EGF receptor gene mutations Anal Biochem 1996;242:84–9. are common in lung cancers from “never smokers” and are associ- 63. Thomas RK, Baker AC, Debiasi RM, et al. High-throughput oncogene ated with sensitivity of tumors to gefitinib and erlotinib. Proc Natl mutation profiling in human cancer. Nat Genet 2007;39:347–51. Acad Sci U S A 2004;101:13306–11. 64. Stephens PJ, McBride DJ, Lin ML, et al. Complex landscapes of 42. Janne PA, Johnson BE. Effect of epidermal growth factor receptor somatic rearrangement in human breast cancer genomes. Nature tyrosine kinase domain mutations on the outcome of patients with 2009;462:1005–10.

www.aacrjournals.org Mol Cancer Res; 8(9) September 2010 1185

Diamandis et al.

65. Emmert-Buck MR, Bonner RF, Smith PD, et al. Laser capture 89. Huang YT, Heist RS, Chirieac LR, et al. Genome-wide analysis of microdissection. Science 1996;274:998–1001. survival in early-stage non-small-cell lung cancer. J Clin Oncol 66. Strausberg RL, Simpson AJ, Wooster R. Sequence-based cancer 2009;27:2660–7. genomics: progress, lessons and opportunities. Nat Rev Genet 90. Sun X, Li F, Sun N, et al. Polymorphisms in XRCC1 and XPG and 2003;4:409–18. response to platinum-based chemotherapy in advanced non-small 67. Maher CA, Kumar-Sinha C, Cao X, et al. Transcriptome sequencing cell lung cancer patients. Lung Cancer 2009;65:230–6. to detect gene fusions in cancer. Nature 2009;458:97–101. 91. Kujawski L, Ouillette P, Erba H, et al. Genomic complexity identifies 68. Morozova O, Marra MA. Applications of next-generation se- patients with aggressive chronic lymphocytic leukemia. Blood quencing technologies in functional genomics. Genomics 2008; 2008;112:1993–2003. 92:255–64. 92. Ding L, Ellis MJ, Li S, et al. Genome remodelling in a basal- 69. Ley TJ, Minx PJ, Walter MJ, et al. A pilot study of high-throughput, like breast cancer metastasis and xenograft. Nature 2010;464: sequence-based mutational profiling of primary human acute 999–1005. myeloid leukemia cell genomes. Proc Natl Acad Sci U S A 2003; 93. Tada M, Kanai F, Tanaka Y, et al. Prognostic significance of genetic 100:14275–80. alterations detected by high-density single nucleotide polymor- 70. Tyner JW, Erickson H, Deininger MW, et al. High-throughput phism array in gastric cancer. Cancer Sci 2010;101:1261–9. sequencing screen reveals novel, transforming RAS mutations in 94. Simon R. Microarray-based expression profiling and informatics. myeloid leukemia patients. Blood 2009;113:1749–55. Curr Opin Biotechnol 2008;19:26–9. 71. Zhao Q, Caballero OL, Levy S, et al. Transcriptome-guided charac- 95. Kim MJ, Ro JY, Ahn SH, Kim HH, Kim SB, Gong G. Clinicopatho- terization of genomic rearrangements in a breast cancer cell line. logic significance of the basal-like subtype of breast cancer: a com- Proc Natl Acad Sci U S A 2009;106:1886–91. parison with hormone receptor and Her2/neu-overexpressing 72. Beroukhim R, Mermel CH, Porter D, et al. The landscape of somatic phenotypes. Hum Pathol 2006;37:1217–26. copy-number alteration across human cancers. Nature 2010;463: 96. Dybkaer K, Iqbal J, Zhou G, Chan WC. Molecular diagnosis and 899–905. outcome prediction in diffuse large B-cell lymphoma and other 73. Beck J, Urnovitz HB, Mitchell WM, Schutz E. Next generation subtypes of lymphoma. Clin Lymphoma 2004;5:19–28. sequencing of serum circulating nucleic acids from patients with 97. Golub TR, Slonim DK, Tamayo P, et al. Molecular classification of invasive ductal breast cancer reveals differences to healthy and cancer: class discovery and class prediction by gene expression nonmalignant controls. Mol Cancer Res 2010;8:335–42. monitoring. Science 1999;286:531–7. 74. Nickerson ML, Jaeger E, Shi Y, et al. Improved identification of von 98. Alizadeh AA, Eisen MB, Davis RE, et al. Distinct types of diffuse Hippel-Lindau gene alterations in clear cell renal tumors. Clin Can- largeB-celllymphomaidentifiedbygeneexpressionprofiling. cer Res 2008;14:4726–34. Nature 2000;403:503–11. 75. Drmanac R, Sparks AB, Callow MJ, et al. Human genome sequenc- 99. Bittner M, Meltzer P, Chen Y, et al. Molecular classification of cu- ing using unchained base reads on self-assembling DNA nano- taneous malignant melanoma by gene expression profiling. Nature arrays. Science 2010;327:78–81. 2000;406:536–40. 76. Service RF. Gene sequencing. The race for the $1000 genome. 100. Beer DG, Kardia SL, Huang CC, et al. Gene-expression profiles pre- Science 2006;311:1544–6. dict survival of patients with lung adenocarcinoma. Nat Med 2002; 77. Cargill M, Altshuler D, Ireland J, et al. Characterization of single- 8:816–24. nucleotide polymorphisms in coding regions of human genes. Nat 101. Potti A, Mukherjee S, Petersen R, et al. A genomic strategy to refine Genet 1999;22:231–8. prognosis in early-stage non-small-cell lung cancer. N Engl J Med 78. Reich DE, Gabriel SB, Altshuler D. Quality and completeness of 2006;355:570–80. SNP databases. Nat Genet 2003;33:457–8. 102. Dhanasekaran SM, Barrette TR, Ghosh D, et al. Delineation of prog- 79. Sherry ST, Ward MH, Kholodov M, et al. dbSNP: the NCBI database nostic biomarkers in prostate cancer. Nature 2001;412:822–6. – of genetic variation. Nucleic Acids Res 2001;29:308 11. 103. Bonnefoi H, Potti A, Delorenzi M, et al. Validation of gene signatures 80. Orr N, Chanock S. Common genetic variation and human disease. that predict the response of breast cancer to neoadjuvant chemo- – Adv Genet 2008;62:1 32. therapy: a substudy of the EORTC 10994/BIG 00-01 clinical trial. 81. Maresso K, Broeckel U. Genotyping platforms for mass-throughput Lancet Oncol 2007;8:1071–8. genotyping with SNPs, including human genome-wide scans. Adv 104. Spentzos D, Levine DA, Kolia S, et al. Unique gene expression – Genet 2008;60:107 39. profile based on pathologic response in epithelial ovarian cancer. 82. McCarthy MI, Abecasis GR, Cardon LR, et al. Genome-wide J Clin Oncol 2005;23:7911–8. association studies for complex traits: consensus, uncertainty 105. Frank O, Brors B, Fabarius A, et al. Gene expression signature of – and challenges. Nat Rev Genet 2008;9:356 69. primary imatinib-resistant chronic myeloid leukemia patients. 83. Pastinen T, Raitio M, Lindroos K, Tainola P, Peltonen L, Syvanen Leukemia 2006;20:1400–7. AC. A system for specific, high-throughput genotyping by allele- 106. Potti A, Dressman HK, Bild A, et al. Genomic signatures to guide specific primer extension on microarrays. Genome Res 2000;10: the use of chemotherapeutics. Nat Med 2006;12:1294–300. 1031–42. 107. Hartmann LC, Lu KH, Linette GP, et al. Gene expression profiles 84. Syvanen AC. Toward genome-wide SNP genotyping. Nat Genet predict early relapse in ovarian cancer after platinum-paclitaxel 2005;37 Suppl:S5–10. chemotherapy. Clin Cancer Res 2005;11:2149–55. 85. Kemp Z, Carvajal-Carmona L, Spain S, et al. Evidence for a colo- 108. Calin GA, Croce CM. MicroRNA signatures in human cancers. Nat rectal cancer susceptibility locus on chromosome 3q21-24 from a Rev Cancer 2006;6:857–66. high-density SNP genome-wide linkage scan. Hum Mol Genet 109. 2006;15:2903–10. ChowTF,YoussefYM,LianidouE,etal.Differentialexpression 86. Nam RK, Zhang WW, Loblaw DA, et al. A genome-wide association profiling of microRNAs and their potential involvement in renal cell – screen identifies regions on chromosomes 1q25 and 7p21 as risk carcinoma pathogenesis. Clin Biochem 2010;43:150 8. lociforsporadicprostatecancer.ProstateCancerProstaticDis 110. Liu CG, Calin GA, Volinia S, Croce CM. MicroRNA expression – 2008;11:241–6. profiling using microarrays. Nat Protoc 2008;3:563 78. 87. Gemignani F, Landi S, Moreno V, et al. Polymorphisms of the 111. Metias SM, Lianidou E, Yousef GM. MicroRNAs in clinical oncology: dopamine receptor gene DRD2 and colorectal cancer risk. Cancer at the crossroads between promises and problems. J Clin Pathol Epidemiol Biomarkers Prev 2005;14:1633–8. 2009;62:771–6. 88. Damm F, Heuser M, Morgan M, et al. Single nucleotide polymor- 112. Schaefer A, Jung M, Mollenkopf HJ, et al. Diagnostic and prognos- phism in the mutational hotspot of WT1 predicts a favorable tic implications of microRNA profiling in prostate carcinoma. Int J outcome in patients with cytogenetically normal acute myeloid Cancer 2010;126:1166–76. leukemia. J Clin Oncol 2010;28:578–85. 113. Sotiropoulou G, Pampalakis G, Lianidou E, Mourelatos Z. Emerging

1186 Mol Cancer Res; 8(9) September 2010 Molecular Cancer Research

Personalized Medicine in Cancer Management

roles of microRNAs as molecular switches in the integrated circuit 132. Song H, Ramus SJ, Tyrer J, et al. A genome-wide association study of the cancer cell. RNA 2009;15:1443–61. identifies a new ovarian cancer susceptibility locus on 9p22.2. Nat 114. Esteller M. Epigenetics in cancer. N Engl J Med 2008;358:1148–59. Genet 2009;41:996–1000. 115. Wulfkuhle JD, Liotta LA, Petricoin EF. Proteomic applications for 133. Wu C, Xu B, Yuan P, et al. Genome-wide examination of genetic the early detection of cancer. Nat Rev Cancer 2003;3:267–75. variants associated with response to platinum-based chemothera- 116. Rogers S, Girolami M, Kolch W, et al. Investigating the correspon- py in patients with small-cell lung cancer. Pharmacogenet Geno- dence between transcriptomic and proteomic expression profiles mics 2010;20:389–95. using coupled cluster models. Bioinformatics 2008;24:2894–900. 134. Leshchenko VV, Kuo PY, Shaknovich R, et al. Genome wide DNA 117. van der Merwe DE, Oikonomopoulou K, Marshall J, Diamandis EP. methylation analysis reveals novel targets for drug development in Mass spectrometry: uncovering the cancer proteome for diagnos- mantle cell lymphoma. Blood. Epub ahead of print 2010. tics. Adv Cancer Res 2007;96:23–50. 135. Schadt EE. Molecular networks as sensors and drivers of common 118. Weissenstein U, Schneider MJ, Pawlak M, et al. Protein chip based human diseases. Nature 2009;461:218–23. miniaturized assay for the simultaneous quantitative monitoring of 136. Strausberg RL, Buetow KH, Emmert-Buck MR, Klausner RD. The cancer biomarkers in tissue extracts. Proteomics 2006;6:1427–36. cancer genome anatomy project: building an annotated gene index. 119. Zhang H, Kong B, Qu X, Jia L, Deng B, Yang Q. Biomarker discov- Trends Genet 2000;16:103–6. ery for ovarian cancer using SELDI-TOF-MS. Gynecol Oncol 2006; 137. Knutsen T, Gobu V, Knaus R, et al. The interactive online SKY/M- 102:61–6. FISH and CGH database and the Entrez cancer chromosomes 120. Miller JC, Zhou H, Kwekel J, et al. Antibody microarray profiling of search database: linkage of chromosomal aberrations with the ge- human prostate cancer sera: antibody screening and identification nome sequence. Genes Chromosomes Cancer 2005;44:52–64. of potential biomarkers. Proteomics 2003;3:56–63. 138. Hubble J, Demeter J, Jin H, et al. Implementation of GenePattern 121. Romaschin AD, Youssef Y, Chow TF, et al. Exploring the patho- within the Stanford Microarray Database. Nucleic Acids Res 2009; genesis of renal cell carcinoma: pathway and bioinformatics 37:D898–901. analysis of dysregulated genes and proteins. Biol Chem 2009; 139. Kuntzer J, Eggle D, Lenhof HP, Burtscher H, Klostermann S. The 390:125–35. Roche Cancer Genome Database (RCGDB). Hum Mutat 2010;31: 122. Siu KW, DeSouza LV, Scorilas A, et al. Differential protein expres- 407–13. sions in renal cell carcinoma: new biomarker discovery by mass 140. Wu CH, Yeh LS, Huang H, et al. The Protein Information Resource. spectrometry. J Proteome Res 2009;8:3797–807. Nucleic Acids Res 2003;31:345–7. 123. Kohnke PL, Mulligan SP, Christopherson RI. Membrane proteomics 141. Keshava Prasad TS, Goel R, Kandasamy K, et al. Human Protein for leukemia classification and drug target identification. Curr Opin Reference Database—2009 update. Nucleic Acids Res 2009;37: Mol Ther 2009;11:603–10. D767–72. 124. Zangar RC, Varnum SM, Bollinger N. Studying cellular processes 142. Collins F. Has the revolution arrived? Nature 2010;464:674–5. and detecting disease with protein microarrays. Drug Metab Rev 143. Jumppanen M, Gruvberger-Saal S, Kauraniemi P, et al. Basal- 2005;37:473–87. like phenotype is not associated with patient survival in estro- 125. Wulfkuhle JD, Edmiston KH, Liotta LA, Petricoin EF III. Technology gen-receptor-negative breast cancers. Breast Cancer Res 2007; insight: pharmacoproteomics for cancer-promises of patient- 9:R16. tailored medicine using protein microarrays. Nat Clin Pract Oncol 144. Venter JC. Multiple personal genomes await. Nature 2010;464:676–7. 2006;3:256–68. 145. Roukos DH. Novel clinico-genome network modeling for revolution- 126. Bodovitz S, Joos T, Bachmann J. Protein biochips: the calm before izing genotype-phenotype-based personalized cancer care. Expert the storm. Drug Discov Today 2005;10:283–7. Rev Mol Diagn 2010;10:33–48. 127. Garber K. Human Cancer Genome Project moving forward despite 146. Roukos DH. Systems medicine: a real approach for future person- some doubts in community. J Natl Cancer Inst 2005;97:1322–4. alized oncology? Pharmacogenomics 2010;11:283–7. 128. Zheng W, Long J, Gao YT, et al. Genome-wide association study 147. Roukos DH. Bionetworks-based personalized medicine versus identifies a new breast cancer susceptibility locus at 6q25.1. Nat comparative-effectiveness research or harmonization of both in Genet 2009;41:324–8. cancer management? Expert Rev Mol Diagn 2010;10:247–50. 129. Yeager M, Orr N, Hayes RB, et al. Genome-wide association study 148. Hunter DJ, Kraft P, Jacobs KB, et al. A genome-wide association of prostate cancer identifies a second risk locus at 8q24. Nat Genet study identifies alleles in FGFR2 associated with risk of sporadic 2007;39:645–9. postmenopausal breast cancer. Nat Genet 2007;39:870–4. 130. Yeager M, Chatterjee N, Ciampa J, et al. Identification of a new 149. Sjoblom T, Jones S, Wood LD, et al. The consensus coding prostate cancer susceptibility locus on chromosome 8q24. Nat sequences of human breast and colorectal cancers. Science Genet 2009;41:1055–7. 2006;314:268–74. 131. Landi MT, Chatterjee N, Yu K, et al. A genome-wide association 150. International network of cancer genome projects. Nature 2010;464: study of lung cancer identifies a region of chromosome 5p15 asso- 993–8. ciated with risk for adenocarcinoma. Am J Hum Genet 2009;85: 151. Pearson H. Biologists initiate plan to map human proteome. Nature 679–91. 2008;452:920–1.

www.aacrjournals.org Mol Cancer Res; 8(9) September 2010 1187

Personalized Medicine: Marking a New Epoch in Cancer Patient Management

Maria Diamandis, Nicole M.A. White and George M. Yousef

Mol Cancer Res 2010;8:1175-1187. Published OnlineFirst August 6, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-10-0264

Cited articles This article cites 149 articles, 40 of which you can access for free at: http://mcr.aacrjournals.org/content/8/9/1175.full#ref-list-1

Citing articles This article has been cited by 14 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/8/9/1175.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/8/9/1175. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.