Sequencing Cancer Genomes and Predicting Molecular Networks

Trastuzumab and beyond: sequencing cancer genomes and predicting molecular networks

DH Roukos1,2 Life diversity can now be clearly explored with the next-generation DNA sequencing technology, allowing the discovery of genetic variants among 1Personalized Cancer Medicine, Biobank, individuals, patients and tumors. However, beyond causal mutations catalog Ioannina University, Ioannina, Greece and completion, systems medicine is essential to link genotype to phenotypic 2 Department of Surgery, Ioannina University cancer diversity towards personalized medicine. Despite advances with School of Medicine, Ioannina, Greece traditional single genes molecular research, including rare mutations in Correspondence: BRCA1/2 and CDH1 for primary prevention and trastuzumab for treating Dr DH Roukos, Department of Surgery, HER2-overexpressing breast and gastric tumors, overall, treatment failure and Ioannina University School of Medicine, death rates are still alarmingly high. Revolution in sequencing reveals that, Ioannina TK 451 10, Greece. now both a huge number and widespread variability of driver mutations, E-mail: [email protected] including single-nucleotide polymorphisms, genomic rearrangements and copy-number changes involved in breast cancer development. All these genetic alterations result in a heterogeneous deregulation of signaling pathways, including EGFR, HER2, VEGF, Wnt/Notch, TGF and others.Cancer initiation, progression and metastases are driven by complex molecular networks rather than linear genotype–phenotype relationship. Therefore, clinical expectations by traditional molecular research strategies targeting single genes and single signaling pathways are likely minimal. This review discusses the necessity of molecular networks modeling to understand complex gene–gene, protein–protein and gene–environment interactions. Moreover, the potential of systems clinico-biological approaches to predict intracellular signaling pathways components networks and cancer heterogeneous cells within an individual tumor is described. A flowchart specific for three steps in cancer evolution separately tumorigenesis, early-stage and advanced-stage breast cancer is presented. Using reverse engineering starting with the integration of available established clinical, environmental, treatment and oncological outcomes (survival and death) data and then the still incomplete but progressively accumulating genotypic data into computational networks modeling may lead to bionetworks-based discovery of robust biomarkers and highly effective cancer drugs targets. The Pharmacogenomics Journal (2011) 11, 81–92; doi:10.1038/tpj.2010.81; published online 26 October 2010

Keywords: breast gastric cancer; trastuzumab; cancer genome; systems biology; drugs; biomarkers

Introduction

Cancer is a highly complex and heterogeneous disease.1–5 Cancer initiation, progression and metastases are driven by mutations. These DNA changes, mostly somatic, but also inherited, accumulating deregulated signaling pathways that Received 14 April 2010; revised 27 August 4 2010; accepted 30 August 2010; published cause cancer development and evolution. Epigenetic changes—chemical online 26 October 2010 modifications of DNA and its associated proteins may also contribute to Trastuzumab and beyond DH Roukos 82

tumorigenesis through gene expression deregulation.6,7 Breast and gastric cancer Over the past 3 years, massively parallel DNA sequencing platforms have become widely available. As the cost of With nearly 2 million new diagnoses and over 1.2 million reading and writing DNA in the last decade has dropped by a deaths each year worldwide,20 breast cancer and gastric million-fold, in the next few years less than $1000 will be cancer are a major health problem. In the United States, required for a whole-genome sequencing; international breast cancer is and will probably continue to be the most personal genomics projects are now underway to explore common malignancy among females, whereas gastric cancer the diversity of life and understand why each of us are incidence has dramatically been reduced in the United unique. The ‘1000 genomes’ project, the International States, but its global incidence and mortality still remain Cancer Genome Consortium (see http://www.icgc.org/ alarmingly high.21 The current highest mortality rates in home)5 and the International Human Epigenome Consor- non-early stages reveal that despite intensive research efforts tium,6 will deliver information about health and disease as and advances with surgery, radiotherapy and systemic it progresses, particularly about intractable diseases, such treatment death rates improvement is modest and very as cancer.6 slow over the last decade.21 Eight human genomes and several cancer genome Systemic treatment with chemotherapy has improved survi- sequences for individual patients with acute myeloid val and saved the lives of many patients with cancer.22 leukemia,8,9 lung cancer,10 melanoma11 and breast cancer12 Particularly for breast cancer, the addition of tamoxifen to have been most recently reported. Beyond point mutations, chemotherapy as a systemic treatment in endocrine-responsive, base substitutions, insertions or deletions of small or large estrogen receptor (ER)-positive and/or progesterone receptor segments of DNA –rearrangements, DNA broken and (PR)-positive patients has saved the lives of a substantial rejoined to a DNA segment from elsewhere in the genome proportion of women with a potentially curable stage of and copy number changes are implicated in cancer.5 Indeed, disease.23 Studies with longer than 5-year survival results are using new ‘paired-end’ sequencing technologies, a latest important for quality assessment of treatment effect on study has shown more genomic rearrangements than oncological outcomes. The 15-year survival results of a large- previously recognized. Most of them were tandem duplica- scale meta-analysis prove that the risk of recurrence and death is tions and remarkably, common in some breast cancers, but still very high; approximately 60% of patients with an early- essentially absent from others and may reflect a novel stage breast cancer recur at 15 years after surgery and adjuvant mutator phenotype.13 These current findings5–13 and evi- multimodal therapy.23 Similarly, in gastric cancer, the 15-year dence from recent extensive genetic studies,3,4 reveal the survival results of the Dutch trial, including stage I–III patients, high complexity and variability of genetic bases of cancer. haveshowna37%deathratebecauseoftherecurrenceafter This cancer heterogeneity underlines the importance of adequate D2 surgery alone.24 However, the failure treatment personalized medicine, one of the biggest biomedical rates for the most common stages II and III in the Western challenges. To achieve this goal, two steps are essential. world are much higher. Despite adjuvant perioperative che- First, to be completed the catalog of driver mutations. To motherapy for stages II and III, the overall treatment failure rate obtain such a catalog, for example mutations and genes of measured as 5-year recurrence or death rate was over 60% in a breast cancer, many sequencing studies with large numbers Western randomized controlled trial (RCT).25 Similarly poor are of patients’ samples will be required because the widespread theresultsfromtheUSAdespitestandardizationofadjuvant of the genetic variability. The second goal, which is to chemoradiotherapy.26 The reasons for the higher corresponding understand how the cancer genomes and individual tumors rates of 60% or more 5-year survival rate in Japan27 have not yet function as whole biological systems, is much more been clarified although the extend of lymphadenectomy may complicated. There is a need to acquire insights into how have a crucial role in this different outcome.28 intracellular signaling pathways network is deregulated in Although adjuvant systemic chemotherapy significantly cancer14 and to understand the interactions among hetero- improves survival, the net absolute survival benefit is less geneous cancer cells with different treatment response than 15%. In the metastatic, stage IV setting, the mean within an individual tumor.1–4 Similarly, exploration of survival is B26 months for breast cancer and B12 months tumor microenvironment2,15 and of a more global approach for gastric cancer despite major efforts, funding and to link individual tumors with their multiple host variables, development and use of newer chemotherapeutic agents including heritable causal mutations, environmental expo- combinations.22 How could survival substantially be pro- sures and lifestyle, are of fundamental importance and longed in the metastatic setting and cure rates be improved represent systems medicine approaches.16 in the adjuvant setting? Arguments in favor and against This article provides an overview for the advances of standard molecular biology and sophisticated systems standard single-gene molecular research and how current approaches are discussed. substantial limitations might be overcome. Given the widespread genetic variability and the molecular interactions complexity of breast cancer and gastric cancer, it is Single genes molecular research discussed how next-generation of molecular networks scenarios17–19 might lead to bionetworks-based discovery Over the last 15 years, important discoveries of single genes of biomarkers and biological agents. and proteins-based basic sciences have been made. But most

The Pharmacogenomics Journal Trastuzumab and beyond DH Roukos 83

importantly from a patient perspective, some of these have RCTs have been made available, how optimistic have we been translated into clinical practice in both the prevention been? and treatment setting. EGFR and trastuzumab Primary prevention The Erbb family has a crucial part in the development and The discovery of BRCA1 and BRCA2 (BRCA)29,30 and CDH131 evolution of cancer. This receptors family consists of four closely related type 1 transmembrane tyrosine kinase approximately 15 years ago has changed medical practice allowing personalized breast cancer and gastric cancer receptors: EGFR (or HER1), ERBB2 (HER2), ERBB3 (HER3) prevention, respectively.32,33 Individual women with heri- and ERBB4 (HER4). In contrast to HER1 and HER2, HER3 does not have tyrosine kinase activity. Each receptor table, high-risk mutations in BRCA and positive genetic comprises an extracellular domain at which ligand binding testing face a very high lifetime risk of roughly 75% for a developing breast cancer and 20–60% for ovarian cancer occurs, an -helical transmembrane segment and an in- 32 tracellular protein tyrosine kinase domain. Ligand binding (Hereditary Breast Ovarian cancer syndrome). Persons to these EGF family receptors phosphorylates and activates a carrying heritable mutations in CDH1 gene regulating complex intracellular signaling pathways network that E-cadherin protein face a B75 lifetime risk of diffuse gastric controls a range of cellular processes, including prolifera- cancer and B40% risk of lobular breast cancer (Hereditary tion, angiogenesis, cell cycle, survival and apoptosis diffuse gastric cancer syndrome).33 Prophylactic surgery or (Figure 1a).36 closed surveillance can save the lives of these high-risk HER2 persons.32,33 amplification and overexpression has a central role in initiation, progression and metastasis of some common cancers, including breast cancer and gastric cancer.35–38 Limitations HER2 status has been recognized as an important prognostic Table 1 summarizes the limitations of genetic testing in the factor. Patients with breast cancer or gastric cancer and prevention setting. Despite basic science advances and the HER2-positive disease have significantly worse survival than advent of genome-wide association studies, for the vast those with HER2-negative tumors.35–38 Thus, this pivotal majority of people in the general population, accurate receptor is a potential therapeutic target. 34 prediction of a person’s cancer risk is elusive. Indeed, Trastuzumab, a monoclonal antibody (Figure 1b), the these inherited mutations are very rare in the general binding of which to HER2 receptors, inhibits HER2 signaling population and mutation carriers account for 3–10% only pathway activity in tumor cells overexpressing HER2. In the of these cancer types. The predictive accuracy among metastatic setting, a previous RCT for breast cancer37 and a persons with positive genetic testing at the age of 50 years most recent international phase-III RCT for gastric cancer38 is less than 50% and ultimately 25% of the mutation carriers have provided evidence for the safety and efficiency of in BRCA1/2 or CDH1 genes will never develop breast or trastuzumab added to chemotherapy to prolong signifi- gastric cancer, respectively. This uncertainty challenges a cantly overall survival (OS) in patients with these two major tailored surgical or medical preventive intervention in cancer types and HER2-positive tumors. For advanced, carriers of these mutations with outsized cancer risk. metastatic or recurrent HER2-positive gastric cancer, trastuzumab-based systemic treatment with chemotherapy has Present generation of targeted therapy become a new standard of care.39,40 It was only recently that the concept of biological agents In the adjuvant setting, trastuzumab for early-stage breast has been incorporated into the pharmaceutical industry, cancer represents the triumph of bench-to-bedside molecu- approximately 50 years after the discovery of the epidermal lar research.41,42 Indeed, potential clinical success at this growth factor (EGF or ERB) and its receptor (EGFR or setting can save the lives of thousands of patients with non- ERBB).35 Targeting only cancer cells, but not healthy cells, metastatic disease. Five RCTs in HER2-positive early-stage when using biologics could dramatically improve the poor breast cancer comparing adjuvant systemic chemotherapy outcomes of cancer patients, while maintaining very low with and without trastuzumab in 13 493 patients have been adverse effects profiles. Now, as multiple sets of data from reported.43–49 All, these studies have demonstrated the

Table 1 Limitations of genetic testing and primary prevention of breast cancer and gastric cancer

Single genes Hereditary Rare incidence Predictive power among mutation carriers syndrome (% of all cancer cases) No cancer Cancer risk at the development (%) age of 50 years (%)

BRCA1/2 HBOC B7 B25 o50 CDH1 HDGC o3 B25 o50

Abbreviations: HBOC, hereditary breast-ovarian cancer; HDGC, hereditary diffuse gastric cancer.

The Pharmacogenomics Journal Trastuzumab and beyond DH Roukos 84

Figure 1 (a) Ligand binding and subsequent Erbb dimer formation initiates signaling through a complex array of intracellular pathways that initiate and control a range of cellular processes. Dimer formation results in the cross-phosphorylation of the dimer partners, creating docking sites that allow the recruitment of downstream signaling components and the formation of signaling complexes. Two key signaling pathways activated by the Erbb family dimers are the MAPK pathway, which stimulates proliferation, and the PI3K–Akt pathway, which promotes tumor cell survival (see the figure). Only signaling through these two pathways and some of the known outcomes are shown here for simplicity. (b) The antibody trastuzumab binds directly to domain IV of the extracellular region of ERBB2, suppressing ERBB2 signaling activity, preventing cleavage of the extracellular domain and marking tumor cells that overexpress ERBB2 for further immunological attack through antibody-dependent cell-mediated cytotoxicity. GSK3b, glycogen synthase kinase 3b; NF-kB, nuclear factor-kB; PDK1, pyruvate dehydrogenase kinase 1; PIP2, phosphatidylinositol biphosphate; PIP3, phosphatidylinositol triphosphate.

Table 2 Limitations of adjuvant treatment with trastuzumab in the early-breast cancer

Study Short follow-up Absolute net Adverse effects increased risksa Impact on cure survival gain

Meta-analysis50 o3 years E10–15% Congestive heart failure Unknown Central nervous system metastasis

Abbreviation: RR, risk ratio. aPatients receiving trastuzumab with chemotherapy had a higher risk for congestive heart failure (RR, 7.60; 95% CI, 4.07–14.18), left ventricular ejection fraction decline (RR, 2.09; 95% CI, 1.84–2.37), and for central nervous system metastasis as the first recurrence event (RR, 1.60; 95% CI, 1.06–2.40). superiority of trastuzumab in both disease-free survival and of trastuzumab was associated with increased risks for OS. An updated meta-analysis has showed a significant congestive heart failure, left ventricular ejection fraction reduction of risks of recurrence by 38% and death by 34%.50 decline and central nervous system metastasis as the first recurrence event. In addition, given the short-term follow- Trastuzumab limitations up it is unclear whether trastuzumab may have a long-term Table 2 summarizes adverse effects and resistance to efficacy leading to cure or it has simply only a delaying- adjuvant treatment of breast cancer with trastuzumab. recurrence occurrence effect. These limitations suggest the need for novel cancer research From a general point of view for all solid tumors, the strategies. Intrinsic and acquired resistance is the biggest present generation of biologics targeting a single compo- challenge. No response to trastuzumab has been reported in nent of a signaling pathway has substantial limitations. HER2-negative patients, who account for 475% of all breast Biological agents include two major groups of drugs: cancers, whereas the net response rate among HER2-positive monoclonal antibodies and small molecule tyrosine kinase patients with breast cancer is less than 15%. Administration inhibitors. The mechanisms of action of these two agents’

The Pharmacogenomics Journal Trastuzumab and beyond DH Roukos 85

categories differ. For example, the monoclonal antibodies methodological flaws of this retrospective study suggest the cetuximab or panitumab used in the treatment of metastatic need for awaiting results of Phase-III RCTs for definitive colorectal cancer binding to extracellular portion of EGFR conclusions.61 block the ligand–receptor dimerization inhibiting signal transduction from cell outside the nucleus. By contrast, the Multigene assays EGFR tyrosine kinase inhibitors, erlotinib and gefitinib, The advent of high-throughput sequencing technology has developed for the treatment of non-small-cell lung cancer revolutionized screening approaches allowing the identifi- and other tumors compete reversibly with ATP to bind cation of profiling expression of hundreds of genes at a time. to the intracellular domain of EGFR thus, inhibit EGFR Several gene-expression profiling studies using the first autophosphorylation and downstream signaling.14,36 generation of microarrays have resulted in the development Despite positive results with phase I–II studies, phase-III of both a molecular classification of breast cancer and 62 RCTs have provided negative results.51,52 The initial over- multigene assays. It was thought that such a multigene enthusiasm could not be confirmed by comparative-effec- approach could lead to more individualized therapeutic tiveness research essential for the translation of new agents decisions in a mean of personalized medicine. into clinical practice.53,54 After this failure assessment, Among many multigene assays developed and published efforts have focused on tumor genotyping for the identifica- to have prognostic or predictive value for many solid tion of mutations that might guide patients’ selection for tumors, including breast cancer, two gene signatures have anti-EGFR treatment. Indeed, phase III trials have demon- the most rapid progress. The 21-gene assay and the 70-gene strated a significant benefit in survival among patients with signature have already been commercialized. The 21-gene a specific mutations status. However, even with this Recurrence Score was developed for identifying high-risk genotyping-based patient selection, there are limitations. patients of distant recurrence among those with node- 63 For example, phase-III RCTs showed that the treatment with negative, ER-positive early breast cancer. It is believed that cetuximab of patients with metastatic colorectal cancer this 21-gene assay (Oncotype DX) may improve decision on without KRAS mutations55 or the treatment with gefitinib of the addition of adjuvant chemotherapy to tamoxifen patients with advanced EGFR mutations non-small-cell lung among women with ER-positive, node-negative early breast cancer,55,56 was associated with a benefit only in progres- cancer. The 70-gene signature developed in the Netherlands sion-free survival, but not in OS. However, evident cancer for avoiding chemotherapy in early, node-negative breast genetic heterogeneity even within individual tumors with cancer has been approved by the Food and Drug Adminis- 64 variability of cancer cells population in therapeutic tration. However, there has been scepticism for both, the response57 with anti-EGFR and anti-VEGF agents,52,58,59 arises original studies used for the development of these genetic valid concerns for the use of progression-free survival as a tools and the subsequent retrospective validation studies, 65–67 robust marker of treatment response. Complete eradication because of methodological flaws. Given the emerging 2–19 of all cancers cells is better measured with OS improvement. evidence for the highly complex heterogeneity of cancer, it is rather an oversimplification and naive to predict the New markers tested in clinical trials clinical value of these genetic tests before the results of underway Phase-III RCTs in the United States (Trial Assigning Currently, approximately 30% of patients with early-stage, Individual Option for Treatment (AILORx)) and in Europe node-negative breast cancer are receiving unnecessary (Microarray in Node-Negative Disease Avoids Chemotherapy 65 adjuvant chemotherapy and some other patients miss a (MINDACT) trial) became available. Gene expression signatures useful chemotherapy treatment. To improve patients selec- based on the first generation of microarrays have substantial 65–68 tion on the basis of the standard clinicopathological limitations for incorporation into clinical practice and hope features, including age, tumor size, histological grade and has been moved to the next generation of DNA microarrays, 62 node status and molecular markers, such as hormone including microRNA arrays. receptor (ER/PR) status and HER2 status, two categories of new markers have been developed and reached the level of Future personalized medicine perspectives evaluation in phase-III RCTs. Micrometastases and isolated tumor cells in axillary lymph nodes and multigene assays. To increase the efficacy and safety of cancer therapeutics, data, information and evidence accumulating suggest an Micrometastases and isolated tumor cells urgent need for the development of biomarkers to tailor the As a next step of traditional node metastases with tumor best possible treatment to the right patient at the right deposit 42 mm (N1 disease), micrometastases (0.2 mm4 time.69 These biomarkers are of paramount importance to N1mi o2 mm) and isolated tumor cells (N0[i] o0.2 mm) in improve response rates and reduce adverse effects for the sentinel and non-sentinel axillary lymph nodes have been currently available agents. Similarly, new therapeutics recently suggested as prognostic markers. Although a large- strategies consider this concept of the development of both scale recent study has reported that women with early, novel drugs and biomarkers as predictors of response for node-negative (N0) breast cancer, but micrometastases and/ patient selection.69 or isolated tumor cells in sentinel and non-sentinel axillary The era of personalized medicine based on new sequen- lymph nodes were benefited from adjuvant chemotherapy,60 cing technologies for genotyping and identifying cancer

The Pharmacogenomics Journal Trastuzumab and beyond DH Roukos 86

Table 3 Targeting ERBB2 and ERBB3; Trastuzumab, lapatinib and other new agents for the treatment of breast cancer, gastric cancer and other solid tumors

Agent Molecular Mechanism of action Highest development status (protocol Licensed indications target number or NCT ID number)

Antibody-based agents Trastuzumab ERBB2 Suppression of ERBB2 signaling, Launched for breast cancer ERBB2-positive metastatic (Herceptin; Genentech) ERBB2 stabilization, marks cells for ERBB2-positive gastric breast cancer, ERBB2- immunological attack cancer—Phase III (BO18255) positive early-breast cancer Pertuzumab ERBB2 Dimerization inhibitor, marks cells Breast cancer—Phase III None (Genentech/ for immunological attack (NCT00567190) Hoffmann-La Roche) Ovarian cancer—Phase II (NCT00096993, NCT00058552) Trastuzumab-DM1 ERBB2 Targeted delivery of a potent Breast cancer—Phase III None (Genentech) anti-microtubule cytotoxic agent (NCT00829166) Ertumaxomab ERBB2 Bispecific affinity allows Breast cancer—Phase II None (Fresenius Biotech recruitment of T cells (NCT00351858, NCT00522457, GmbH) NCT00452140) AMG 888 or U3-1287 ERBB3 Not yet defined Phase I (NCT00730470) None (Amgen)

TKIs Lapatinib (Tykerb; ERBB2 TKI Launched for breast cancer In combination with GlaxoSmithKline) ERBB2-positive gastric capecitabine for advanced cancer—Phase III (NCT00486954, ERBB2-positive breast cancer NCT00680901) previously treated with an NSCLC—Phase II (NCT00528281) anthracycline, a taxane or Head and neck cancer—Phase II trastuzumab (NCT00490061, NCT00387127, NCT00424255) Colorectal adenocarcinoma—Phase II (NCT00574171) HKI-272 (Wyeth) EGFR, ERBB2 Irreversible TKI Breast cancer—Phase III None (NCT00777101) ARRY-334543 (Array EGFR, ERBB2, Reversible TKI Breast cancer—Phase II None BioPharma) ERBB4 (NCT00710736) BIBW-2992 EGFR, ERBB2 Irreversible TKI Breast cancer—Phase II None (Boehringer (NCT00425854, NCT00826267, Ingelheim) NCT00708214) NSCLC—Phase III (NCT00425854, NCT00826267, NCT00708214) Head and neck cancer—Phase II (NCT00514943)

Heat-shock protein inhibitors 17-AAG (Bristol-Myers HSP90 Inhibitory activity reduces the Multiple myeloma—Phase III None Squibb) stability of ERBB2, causes abrogation (NCT00514371) of ERBB2 signaling Breast cancer—Phase II (NCT00817362) IPI-504 (Infinity HSP90 Inhibitory activity reduces the Multiple myeloma—Phase II and III None Pharmaceuticals) stability of ERBB2, causes abrogation (NCT00514371)) of ERBB2 signaling Breast cancer—Phase II (NCT00817362) NSCLC—Phase II (NCT00431015) Melanoma—Phase II (NCT00087386) Ovarian cancer—Phase II (NCT00093496)

Abbreviations: EGFR, epidermal growth factor receptor; HSP90, heat-shock protein 90; NCTID, National Clinical Trials Identifier; NSCLC, non-small-cell lung cancer; TKI, tyrosine kinase inhibitor. genomes variants responsible for response or resistance is currently more than 30 cancer stem cells R&D prog- here.70,71 But, yet myriad problems exist for achieving major rammes in progress, around 50% of which are at Phase I clinical applications of personalized medicine.71–73 or beyond.74 But most pharmaceutical and biotechnology To overcome resistance to breast treatment, various companies and academic research institutions have been strategies have been developed. For example, there are focussing on agents that target HER2, EGFR and most

The Pharmacogenomics Journal Trastuzumab and beyond DH Roukos 87

recently HER3 signaling pathways. As a result, many agents Completing cancer mutations catalog are undergoing a wide range of evaluation from preclinical Despite advances with massively parallel sequencing tech- to phase-III RCTs. These new therapies include antibodies nologies and decreasing costs, the way to complete a (pertuzumab and Ertumaxomab), tyrosine kinase inhibitors comprehensive catalog of genetic alterations in breast (lapatinib) and heat-shock protein inhibitors that interfere cancer and gastric cancer is long with multiple challenges. with the formation of ERBB2–ERBB3 dimmers.36 Table 3 First, these new sequencing techniques can primarily summarizes preclinical and clinical studies that evaluate identify point mutations, including base substitutions and these agents. Further efforts are underway to evaluate small insertions or deletions. The first nearly-complete potential combinations that target multiple or pivotal genomes sequencing in a patient with lobular, ER-positive players in oncogenic signaling cascades. For example, breast cancer has been recently reported.12 Hundreds of promising results have been reported for PI3K hyperactiva- patient’s samples for each classification category, which is tion-mediated resistance to lapatinib and trastuzumab, currently used in clinical practice, will be required to be which was overcome by the addition of a dual inhibitor of completely sequenced in order to obtain a full catalog both mToR and PI3K.74 Although it may present a more of point mutations. However, given that more than rational approach, the current evidence for complexity of 100 000 000 000 base pairs of DNA sequence will probably interacting cancer cells and signaling pathways suggest the be required to identify causal single-nucleotide polymor- need for networks approaches and global understanding of phisms (SNPs) that imply at least 20-fold sequence coverage cancer genomes. of the cancer genome, that increases the costs. Second, beyond point mutations, genomic rearrangements and copy number changes also have a crucial role Cancer heterogeneity and complexity-based direction in cancer and thus these should also be identified. Yet, the techniques for reliable results have not been standardized. 13 Emerging solid evidence by using new sequencing techno- For example, most recently, Stephens and colleagues using logies reveals that cancer nature is much more complicated a new technique have mapped the chromosome rearrange- than we imagined. Somatic and heritable mutations drive ments in human breast cancers at high resolution. They tumor development, growth and metastasis through deregu- found more rearrangements than previously recognized, lation of signaling pathways. Apart from EGFR and VEGF and tandem duplications were remarkably common in some pathways, several other downstream pathways, including breast cancers but essentially absent from others, suggesting Wnt/Notch, Hedgehog, TGF and others may have a crucial also potential breast cancer phenotypic diversity related to role in cancer initiation and evolution. Available evidence rearrangements. Presently, using similarly high-resolution 75 from both low-throughput and high-throughput analysis analysis (Affymetrix 250K Sty I array), Beroukhim et al. suggests that presently B400 cancer genes (http://www. analyzed 3131 cancer specimens for identifying somatic sanger.ac.uk/genetics/CGP/Census/) have been identified copy-number alterations. Among 243 breast cancers and 23 with estimates to indicate a huge number of genetic and gastric cancers, the genomic landscape showed an average of genomic alterations. B75 somatic copy-number alterations of which there were 75 Next-generation DNA sequencing technologies promise to B45 amplifications and B30 deletions. identify all these DNA changes and improve the under- Third, mitochondrial DNA (mtDNA) mutations, beyond standing of cancer genes function. Indeed, with a trend the nuclear DNA mutations reported above, may improve towards faster and cheaper whole- or partial-targeted understanding of the cancer nature. Indeed, providing genome sequencing, genome-wide RNA, serial analysis of insights into the nature and variability of mtDNA sequences gene expression, microRNAs, protein–DNA interactions and may have implications for cancer biomarker development. A 76 comprehensive analyses of transcriptomes and inter- most recent study by the Vogelstein team using massively actomes70 important insights into molecular mechanisms parallel sequencing-by-synthesis approaches found wide- underlying cancer development and metastasis will emerge. spread heterogeneity (heteroplasmy) in the mtDNA of However, the distance from this general and imprecise normal human cells and cancer cells. Moreover, the statement to reach clinical success with robust biomarkers frequency of heteroplasmic variants varied considerably and biologics is huge. Multiple problems await solution. between different tissues in the same individual. In addition to the variants identified in normal tissues, cancer cells harboured further homoplasmic and heteroplasmic muta- Challenging in moving forward tions that could also be detected in patient plasma. Given If we accept the variability and network complexity of the higher concentration of mutant mtDNA molecules in cancer, then single genes molecular research has major the plasma than that of the mutant nuclear mutations, limitations in the discovery of druggable targets. These mtDNA mutations could reliably be used as tumor-specific limitations in both prevention and treatment setting are biomarkers.76 showed in Tables 1 and 2, respectively. A rational strategy In summary, to obtain a complete mutations catalog, includes two major challenges, the completion of mutations new studies using massively parallel sequencing approaches catalog and the understanding of an individual’s tumor with large patients’ samples will be required given breast whole systems function. cancer and gastric cancer variability. Challenges include

The Pharmacogenomics Journal Trastuzumab and beyond DH Roukos 88

distinguishing of all these alterations into somatic and Bionetworks for breast cancer inherited, as well as into causal (driver) and non-causal Figure 2 delineates a flowchart towards a new horizon of (passengers) mutations.5 networks-based identification of key targets for breast cancer. For both, research and clinical reasons a classifica- Bionetworks towards genotype–phenotype MAP tion into tumorigenesis, early-stage disease and advanced- exploration stage cancer is meaningful. Differences in the number of genetic alterations, the degree of the impact of environ- Individual human cancers are either homogenous tumors or mental exposures, treatment intervention and prognosis in they contain a complex mixture of cancer-interacting cells each category make, nearly essential, a separate research with variability of mutations of their cancer genomes.1–13 In strategy if we aim at achieving clinical implications. contrast to good prognosis of patients with sensitive homogenous tumors, mostly in early stages, mixture tumors Tumorigenesis-based tailored prevention containing aggressive, non-sensitive cancer cells to current It is thought that normal breast epithelial cells are converted treatment, mostly diagnosed at advanced-stage disease, into cancer cells as a result of genetic and environmental have a poor prognosis.3,4,18,57,71 Even if future studies factors. Therefore, efforts to identify these two risk factors reach the completion of mutations and genes catalog for categories have been performed. Many lifestyle and envir- major cancers, such as breast or gastric tumors, the onmental risk factors for breast cancer have been identified bigger challenge will be to understand how an individual (age at menarche/menopause, parity, age of first and tumor functions within a certain patient as a complex subsequent pregnancies, breastfeeding, exogenous estrogen whole clinico-biological system.18 Such an in-depth analysis use, body mass index, height and alcohol consumption).77 requires advances in systems biology, medicine and Several genetic risk variants have been identified using oncology.16–19,57 conventional low-throughput sequencing.3 On the basis of Rational steps towards this direction include exploration the high-throughput sequencing technologies and biochips of the kind of cells within an individual mixture tumor, developed, genome-wide association studies have identified their interactions and the impact of the individual host many SNPs, which, however, confer a small breast cancer 78 that is strongly related with the environment and lifestyle risk. The concept, therefore, to combine traditional risk of this individual person. Genetic ancestry and the magni- factors and recently identified genetic variants (SNPs) for tude—small, moderate, high of genetic susceptibility to a women’s risk stratification appear promising. A latest study specific cancer type also has a crucial role in cancer involving 5590 case subjects with breast cancer and 5998 initiation and evolution. Therefore, completing the catalog control subjects, the addition of information on 10 newly of somatic and inherited mutations represents a substantial identified genetic variants (SNPs) to a standard clinical progress, but a part only in the puzzle of cancer nature. breast-cancer risk model (Gail model, including reproduc- In most cases, cancer development and progression is tive history, family history of close relatives with breast driven by complex genes–genes and genetic–environmental cancer and previous breast biopsies) provided disappointing factors interactions. The inference of these networks results without significant predictive power of SNPs.79 defines the cancer phenotype rather than a simple linear These current negative results in women’s personalized- link between a single mutated gene and cancer. There- risk-based prediction for primary prevention are not surpris- fore, sophisticated systems medicine approaches16 will be ing. Efforts to risk prediction on the basis of an incomplete required to acquire insight into how genotype–phenotype breast cancer mutations catalog and without considering operates and how using network modeling could predict complex genes–genes and genes–environment interactions the oncological outcome, that is, cancer development, will likely have modest clinical success. The magnitude of metastatic recurrence/death or cure after a specific complexity and challenge is revealed by the running therapy.17–19 Breakthrough Generations, a study into the genetic and environmental causes of breast cancer that has recruited Tumor microenvironment and metastasis 100 000 British women and has received d12-million in An individual primary breast tumor or gastric tumor is start-up funding from the Institute of Cancer Research consisted from cancer cells and non-cancer cells termed as and the Breakthrough Breast Cancer charity. The plan is stroma or microenvironment comprising numerous cells, to collect detailed health information over the next 40 years including endothelial cells of the blood and lymphatic to improve understanding of the causes and prevention circulation, stromal fibroblasts and a variety of bone of cancer. marrow-derived cell. The development of cancer cells from Alternatively, instead of waiting for several decades, normal cells (tumorigenesis). This microenvironment predictive network modeling might have faster clinical influences tumor growth at the primary site and develop- implications. Given the nonlinear relationship between ment of occult micrometastases or clinically evident causal mutations and breast cancer phenotypes computa- metastases at distant sites (secondary tumors).15 Therefore, tions network strategies are required to predict cancer consideration of the interactions between cancer cells development among individual women in the general and stromal cells is also required during the designing population. Women with strong family history and high- bionetworks strategies.15–19 risk heritable BRCA mutations represent a distinct group and

The Pharmacogenomics Journal Trastuzumab and beyond DH Roukos 89

Normal cells Tumorigenesis Early-stage cancer Advanced-stage cancer

• >20 mutations Mean values • 7 to 20 causal mutations Complex interactions • 80-100 genetic alterations-wide • Environment/lifestyle variability factors • Complex interactions of genetics and environment • Homogeneous tumors • Heterogeneous tumors • Few but varied deregulated DRIVE transformation of • Low metastatic potential • High metastatic potential signaling pathways normal to cancer cells • Good prognosis • Aggressive subpopulation of cancer cells • Poor prognosis • Very aggressive tumors • Poor prognosis

Bionetworks. Integration of both phenotype (clinical) and genotype type data into network modeling. Using reverse engineering method first are integrated into computational networks traditional standard clinico-pathologic features, environmental and lifestyle factors as well as treatment and follow-up data. Then the available and emerging genotyping data including point mutations, rearrangements and copy- number variants.

Biomarkers • Biomarkers • Bioassays to screen individual tumor • Biologically targeted agents tissue for assessing which signaling Tailored risk-based prevention intervention pathways are deregulated

• KEY TAGRETS based on Low risk High risk interactions (networks) predictions • Surveillance • Agents or Surgery • Biomarkers • Biologics (combination)

Figure 2 Flowchart on bionetworks-based approaches to discover key cancer targets (biomarkers/biological agents) adapted to genetic bases of cancer, heterogeneity and complex interactions. should be studied separately. Two sets of data required for for networks-based prediction of high-risk women may be networks integration models: genetics and clinical data. Yet shorter than awaiting the full gene catalog to be completed. major challenge is the threshold of the number of causal mutations required for conversion of normal cells into cancer cells. Earlier estimates and experimental data have Early-stage breast cancer suggested roughly seven ‘hits’ or five to sixmutated Despite advances with endocrine therapy for hormone- genes,80,81 but recent reports considering latest sequencing receptor-positive patients and trastuzumab for HER2-posi- data show that this number is higher with approximately tive tumor, many of these patients recur because of at least 20 genes to be involved in tumorigenesis.82 As resistance and die.23,50,83,84 No endocrine therapy can be cheaper sequencing techniques became available many used for one-third of patients who have hormone-receptor- more causal mutations, including SNPs, rearrangements negative tumor and two-thirds who have HER2-negative and copy-number changes, will be identified by ongoing tumor cannot benefit from trastuzumab, as it is not effective and new studies. in these tumors. As a result, these patients can receive only Many traditional factors, including age, reproductive adjuvant chemotherapy, with an overall, poor prognosis. factors, family history, breast biopsies and breast density One of the most promising R&D strategies to overcome on mammographic screening have been identified as this substantial resistance to current treatment is the field of standard risk factors for breast cancer. Although detailed molecular networks. Locoregional and distant recurrences information on all these variables (for example, age at are driven by complex genetic, molecular and cellular menarche, number of children, breastfeeding, exogenous interactions rather than a simple linear genotype–pheno- estrogen use, body mass index, height, alcohol consump- type relationship. Therefore, understanding of these net- tion, diet and several other lifestyle factors), is unable to works could be considered as nearly essential to predict identify high-risk women, combination with genetic factors recurrence risk and therapeutic response. may increase this predictive potential. Integrating all these Prospective studies to link oncological outcomes (recur- empirical data along with emerging next-generation sequen- rence/death vs no recurrence/cure) with somatic causal cing platforms data into computational-mathematical mutations are the most rational scientific way for identify- model may lead to the identification of high-risk women ing robust biomarkers and effective drugs. But such an in the general population. Given the long time required to approach will require several decades for completing the complete the mutation catalog of breast carcinogenesis, mutation catalog of point mutations, genomic rearrange- using reverse engineering methods integrating first the ments and copy-number changes and subsequently to await completed traditional data and then the incomplete, but 5–10 years follow-up results to define causal responsible progressively accumulating genetics data, the time needed mutations for treatment failure.

The Pharmacogenomics Journal Trastuzumab and beyond DH Roukos 90

An attractive alternative is bionetworks approach. To predict has two major implications: first, to prolong survival of genotype–phenotype map, both genotypic and phenotypic these patients. Second, even success of only few months OS data are required. Currently, high-quality clinical (phenotypic) benefit due to new agents, it may save the lives of many data are available from large-scale RCTs and meta-analyses. patients in the adjuvant setting. Indeed, clinical success of With the general term, clinical data are included myriad drugs at metastatic setting can usually be translated with variables. These are clinicopathological features (age, tumor drugs efficacy in the adjuvant setting of women with size, node status and histological grade), standard molecular potentially curable stages I–III disease. As the largest number markers (ER/PR/HER2 status), environmental-lifestyle factors of both causal mutations (B100) and deregulated signaling some of which may continue to affect recurrence, treatment pathways (B10), as well as the widest heterogeneity has (surgery, radiotherapy, chemotherapy, endocrine therapy, been found among patients with stage IV breast cancer, and trastuzumab, toxicity and adverse effects) and patients out- the highest grade of host immunity-tumor deregulation is come. Despite the large number of variables, short- and long- met at this tumor stage, molecular networks of systems term follow-up results define two only events, that is, medicine might represent the most promising direction for recurrence or death and survival without recurrence. The task future clinical success. Similar molecular networks predic- will be to link these outcomes events with the inference of tive models using reverse engineering approaches, as clinical and genetic interactions. described above in the adjuvant setting, but integrating Yet, breast cancer genotypic data are incomplete, but these clinical data of stage-IV patients could also here lead to the are rapidly accumulating with the next-generation sequencing identification of crucial breast cancer targets. These bionet- platforms for point mutations and copy-number changes and works targets can then be used for the development of new techniques for rearrangements identification. biomarkers and biologics. Therefore, using reverse engineering methods to build network modeling starting with the integration of pheno- Intellectual innovation typic (clinical) data completely available appears a powerful The landmark achievement of the first draft sequence of the strategy. Further integration of genetics data, which will human genome a decade ago raised excitement for a dramatically be increased over the next few years into revolution in both personal genomics and medical practice. powerful mathematical and computational modeling stra- Now, 10 years later, every clinician knows that the genomics tegies may lead to the discovery of bionetworks-based breast explosion has not, at least yet, translated into a day-to-day cancer targets. These targets represent DNA changes in a clinical practice. Distinguished scientists agree that the nonlinear correlation relationship and can be used for the expectations for personal genomes-based diseases risk pre- development of the next-generation of biomarkers and diction and prevention and appropriate treatment of biological agents. The ultimate goal is to tailor the best complex diseases, such as cancer for achieving cure were responsive combination of drugs to individual patients on too high set.87–94 The leaders of the public and private the basis of the networks prediction of clinical data, efforts for the completion of human genome sequencing, genetics–genomics data and their interactions. Francis Collins and Craig Venter, both say sequencing Signaling pathways network prediction is also a very explosions brought ‘not much’ in health.88,89 Collins promising approach with therapeutic implications. Given emphasizes that personalized medicine passes through the huge number of causal mutations as compared with the genotyping information88 and Venter points out the relatively small number of pathways deregulated by these importance of phenotype and the need to link clinical data mutated genes components of the pathways, screening with genetic variation.89 Looking forward at the next techniques provide exciting opportunities. Indeed, new decade, as the cost of full genome sequence drops rapidly, high-throughput analyses are being developed for screening perhaps less than $1000 in the next few years, and the 70 breast cancer samples for characterizing which signaling quality of genomics data increases, challenges should pathways, apart of well-known EGFR and VEGF, such as be overcome so that the astounding technological and Wnt, Notch, TGF-beta, Hedgehog and others are deregulated intellectual advances will be applied into medical practice 4,35,36,54,57,74,85,86 in individual samples. Integrating all these and improved health.87 data into molecular networks, we might link phenotypic Yet, however, the way to reach faster personalized cancer pre- drugs resistance and recurrence with deregulated signaling vention and treatment of breast, gastric and other major cancers pathways networks for individual patients. This networks has not been defined. Perhaps systems biology and medicine prediction can then be used for discovering novel robust may lead to clinical success but beyond technology, inspired assays for predicting recurrence risks and response or scientific intellectual should have the predominant role. resistance, as well as a new generation signaling pathways networks-based biological agents. Conclusions Advanced-stage cancer Metastatic, recurrent or locally advanced unresectable The explosion in sequencing technology provides now cancer is an incurable disease. Development of new evidence that common diseases, such as cancer, responsible biologics tailored to novel biomarkers-based selection of for high-mortality rates worldwide, are highly complex and responder individual patients at this final stage of disease heterogeneous.

The Pharmacogenomics Journal Trastuzumab and beyond DH Roukos 91

A nearly linear genotype–phenotype relation is rare in 16 Roukos DH. Systems medicine: a real approach for future personalized human life explaining why a only few individuals benefit oncology? Pharmacogenomics 2010; 11: 283–287. from the discovery of BRCA1/2 and CDH1 in prevention 17 Schadt EE. Molecular networks as sensors and drivers of common human diseases. Nature 2009; 461: 218–223. setting and trastuzumab for HER2 amplification in the 18 Roukos DH. Novel clinico-genome network modeling for revolutioniz- treatment setting of breast cancer and gastric cancer. ing genotype-phenotype-based personalized cancer care. Expert Rev At the end of the first post-genomic era and looking Mol Diagn 2010; 10: 33–48. forward, the number of complete sequence of healthy 19 Rockman MV. Reverse engineering the genotype-phenotype map with natural genetic variation. Nature 2008; 456: 738–744. and cancer personal genomes will dramatically be grown. 20 Kamangar F, Dores GM, Anderson WF. Patterns of cancer incidence, But this genomic revolution providing a huge number of mortality, and prevalence across five continents: defining priorities to point mutations, genomic rearrangements and copy-num- reduce cancer disparities in different geographic regions of the world. ber variants alone will only modestly improve oncological J Clin Oncol 2006; 24: 2137–2150. 21 Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ et al. Cancer statistics outcomes of patients with breast, gastric and other major 2009. CA Cancer J Clin 2009; 59: 225–249. cancers. It appears nearly essential to focus on how to link 22 Chabner BA, Roberts Jr TG. Timeline: chemotherapy and the war on all these genotype data with clinical data of the phenotypic cancer. Nat Rev Cancer 2005; 5: 65–72. cancer diversity. Bionetworks modeling represents one of 23 Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on the most promising fields for understanding and predicting recurrence and 15-year survival: an overview of the randomized trials. genotype–phenotype map. Although this is a new way Lancet 2005; 365: 1687–1717. towards personalized medicine, and perhaps beyond of the 24 Songun I, Putter H, Kranenbarg EM, Sasako M, van de Velde CJ. current state-of-the-art, technological advances and inter- Surgical treatment of gastric cancer: 15-year follow-up results of the randomised nationwide Dutch DLD2 trial. Lancet Oncol 2010; 11: national networks of high intellectual scientists may lead 439–449. to the next-generation, bionetworks-based development of 25 Cunningham D, Allum WH, Stenning SP, Thompson JN, Van de Velde robust biomarkers and drugs key cancer targets. CJ, Nicolson M et al. Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med 2006; 355: 11–20. 26 Macdonald JS, Smalley SR, Benedetti J, Hundahl SA, Estes NC, Conflict of interest Stemmermann GN et al. Chemoradiotherapy after surgery compared with surgery alone for adenocarcinoma of the stomach or gastro- The author declares no conflict of interest. esophageal junction. N Engl J Med 2001; 345: 725–730. 27 Sakuramoto S, Sasako M, Yamaguchi T, Kinoshita T, Fujii M, Nashimoto A et al. Adjuvant chemotherapy for gastric cancer with S-1, an oral References fluoropyrimidine. N Engl J Med 2007; 357: 1810–1820. 28 Roukos DH. Genome-wide association studies and aggressive surgery 1 Heppner GH. Tumor heterogeneity. Cancer Res 1984; 44: 2259–2265. toward individualized prevention, and improved local control and 2 Hanahan D, Weinberg RA. The hallmarks of cancer. Cell 2000; 100: overall survival for gastric cancer. Ann Surg Oncol 2009; 16: 795–798. 57–70. 29 Miki Y, Swensen J, Shattuck-Eidens D, Futreal PA, Harshman K, 3 Wood LD, Parsons DW, Jones S. The genomic landscapes of human Tavtigian S et al. A strong candidate for the breast and ovarian cancer breast and colorectal cancers. Science 2007; 318: 1108–1113. susceptibility gene BRCA1. Science 1994; 266: 66–71. 4 Jones S, Zhang X, Parsons DW. Core signaling pathways in human 30 Wooster R, Bignell G, Lancaster J, Swift S, Seal S, Mangion J et al. pancreatic cancers revealed by global genomic analyses. Science 2008; Identification of the breast cancer susceptibility gene BRCA2. Nature 321: 1801–1806. 1995; 378: 789–792. 5 Stratton MR, Campbell PJ, Futreal PA. The cancer genome. Nature 2009; 31 Guilford P, Hopkins J, Harraway J, McLeod M, McLeod N, Harawira P 458: 719–724. Review. et al. E-cadherin germline mutations in familial gastric cancer. Nature 6 Time for the epigenome. Nature 2010; 463: 587. 1998; 392: 402–405. 7 Li M, Chen WD, Papadopoulos N, Goodman SN, Bjerregaard NC, 32 Narod SA, Foulkes WD. BRCA1 and BRCA2: 1994 and beyond. Nat Rev Laurberg S et al. Sensitive digital quantification of DNA methylation in Cancer 2004; 4: 665–676. clinical samples. Nat Biotechnol 2009; 27: 858–863. 33 Ziogas D, Roukos DH. CDH1 testing: can it predict the prophylactic or 8 Mardis ER, Ding L, Dooling DJ, Larson DE, McLellan MD, Chen K et al. therapeutic nature of total gastrectomy in hereditary diffuse gastric Recurring mutations found by sequencing an acute myeloid leukemia cancer? Ann Surg Oncol 2009; 16: 2678–2681. genome. N Engl J Med 2009; 361: 1058–1066. 34 Roukos DH. Genome-wide association studies: how predictable is a 9 Ley TJ, Mardis ER, Ding L, Fulton B, McLellan MD, Chen K et al. DNA person’s cancer risk? Expert Rev Anticancer Ther 2009; 9: 389–392. sequencing of a cytogenetically normal acute myeloid leukaemia 35 Gschwind A, Fischer OM, Ullrich A. The discovery of receptor tyrosine genome. Nature 2008; 456: 66–72. kinases: targets for cancer therapy. Nat Rev Cancer 2004; 4: 361–370. 10 Pleasance ED, Stephens PJ, O’Meara S, McBride DJ, Meynert A, Jones D 36 Baselga J, Swain SM. Novel anticancer targets: revisiting ERBB2 and et al. A small-cell lung cancer genome with complex signatures of discovering ERBB3. Nat Rev Cancer 2009; 9: 463–475. tobacco exposure. Nature 2009; 463: 184–190. 37 Slamon DJ, Leyland-Jones B, Shak S, Fuchs H, Paton V, Bajamonde A 11 Pleasance ED, Cheetham RK, Stephens PJ, McBride DJ, Humphray SJ, et al. Use of chemotherapy plus a monoclonal antibody against HER2 for Greenman CD et al. A comprehensive catalogue of somatic mutations metastatic breast cancer that overexpresses HER2. N Engl J Med 2001; from a human cancer genome. Nature 2010; 463: 191–196. 344: 783–792. 12 Ding L, Ellis MJ, Li S, Larson DE, Chen K, Wallis JW et al. Genome 38 Van Cutsem E, Kang Y, Chung H, Shen A, Sawaki F, Lordick J et al. remodelling in a basal-like breast cancer metastasis and xenograft. Efficacy results from the ToGA trial: a phase III study of trastuzumab Nature 2010; 464: 999–1005. added to standard chemotherapy (CT) in first-line human epidermal 13 Stephens PJ, McBride DJ, Lin ML, Varela I, Pleasance ED, Simpson JT growth factor receptor 2 (HER2)-positive advanced gastric cancer (GC). et al. Complex landscapes of somatic rearrangement in human breast J Clin Oncol 2009; 27(Suppl): 18s (abstr LBA4509). cancer genomes. Nature 2009; 462: 1005–1010. 39 Roukos DH. Targeting gastric cancer with trastuzumab: new clinical 14 Hahn WC, Weinberg RA. Modelling the molecular circuitry of cancer. practice and innovative developments to overcome resistance. Ann Surg Nat Rev Cancer 2002; 2: 331–341. Oncol 2010; 17: 14–17. 15 Joyce JA, Pollard JW. Microenvironmental regulation of metastasis. 40 Petrelli NJ, Winer EP, Brahmer J, Dubey S, Smith S, Thomas C et al. Nat Rev Cancer 2009; 9: 239–252. Clinical cancer advances 2009: major research advances in cancer

The Pharmacogenomics Journal Trastuzumab and beyond DH Roukos 92

treatment, prevention, and screening—a report from the American 63 Paik S, Shak S, Tang G, Kim C, Baker J, Cronin M et al. A multigene assay society of clinical oncology. J Clin Oncol 2009; 27: 6052–6069. to predict recurrence of tamoxifen-treated, node-negative breast 41 Burstein HJ. The distinctive nature of HER2-positive breast cancers. cancer. N Engl J Med 2004; 351: 2817–2826. N Engl J Med 2005; 353: 1652–1654. 64 van de Vijver MJ, He YD, van’t Veer LJ, Dai H, Hart AA, Voskuil DW et al. 42 Hortobagyi GN. Trastuzumab in the treatment of breast cancer. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 2005; 353: 1734–1736. N Engl J Med 2002; 347: 1999–2009. 43 Smith I, Procter M, Gelber RD, Guillaume S, Feyereislova A, Dowsett M et al. 65 Roukos DH. Twenty-one-gene assay: challenges and promises in 2-year follow-up of trastuzumab afteradjuvantchemotherapyinHER2- translating personal genomics and whole-genome scans into persona- positive breast cancer: a randomised controlled trial. Lancet 2007; 369: 29–36. lized treatment of breast cancer. J Clin Oncol 2009; 27: 1337–1338. 44 Perez EA, Romond EH, Suman VJ, Jeong NE, Davidson CE, Geyer S et al. 66 O’Shaughnessy JA. Molecular signatures predict outcomes of breast Updated results of the combined analysis of NCCTG N9831 and NSABP cancer. N Engl J Med 2006; 355: 615–617. B-31 adjuvant chemotherapy with/without trastuzumab in patients 67 Huang CC, Bredel M. Use of gene signatures to improve risk estimation with HER2-positive breast cancer. J Clin Oncol 2007; 25: 512. in cancer. JAMA 2008; 299: 1605–1606. 45 Romond EH, Perez EA, Bryant J, Suman VJ, Geyer Jr CE, Davidson NE 68 Eichhorn PJ, Baselga J. HER2 signatures in breast cancer: ready to go to et al. Trastuzumab plus adjuvant chemotherapy for operable print? J Clin Oncol 2010; 28: 1809–1810. HER2-positive breast cancer. N Engl J Med 2005; 353: 1673–1684. 69 Partnership to Improve Patient Care. Individualized Patients Personalized 46 Piccart-Gebhart MJ, Procter M, Leyland-Jones B, Goldhirsch A, Care 2009. PIPC: Washington, DC http://www.improvepatientcare.org/ Untch M, Smith I et al. Trastuzumab after adjuvant chemotherapy in individualized-patients-personalized-care (Accessed April 17, 2009). HER2-positive breast cancer. N Engl J Med 2005; 353: 1659–1672. 70 Shendure J, Ji H. Next-generation DNA sequencing. Nat Biotechnol 47 Slamon D, Eiermann W, Robert NJ, Pienkowski T, Martin M, Pawlicki M 2008; 26: 1135–1145. et al. BCIRG 006: II interim analysis phase III randomized trial comparing 71 Roukos DH. Bionetworks-based personalized medicine vs comparative- doxorubicin and cyclophosphamide followed by docetaxel with doxo- effectiveness research or harmonization of both in cancer manage- rubicin and cyclophosphamide followed by docetaxel and trastuzumab ment? Expert Rev Mol Diagn 2010; 10: 247–250. with docetaxel, carboplatin and trastuzumab in Her2neu positive early 72 Valid concerns. Nature 2010; 463: 401–402. breast cancer patients. The 2006 San Antonio Breast Cancer Symposium; 73 Goldstein DB. 2020 visions personalized medicine. Nature 2010; 463: 27. December 14–17, 2006; San Antonio, Texas. 74 Eichhorn PJ, Gili M, Scaltriti M, Serra V, Guzman M, Nijkamp W et al. 48 Slamon D, Eiermann W, Robert NJ, Pienkowski T, Martin M, Pawlicki M Phosphatidylinositol 3-kinase hyperactivation results in lapatinib resis- et al. Phase III randomized trial comparing doxorubicin and cyclo- tance that is reversed by the mTOR/phosphatidylinositol 3-kinase phosphamide followed by docetaxel with doxorubicin and cyclo- inhibitor NVP-BEZ235. Cancer Res 2008; 68: 9221–9230. phosphamide followed by docetaxel and trastuzumab with docetaxel, 75 Beroukhim R, Mermel CH, Porter D, Wei G, Raychaudhuri S, Donovan J carboplatin and trastuzumab in HER2 positive early breast cancer et al. The landscape of somatic copy-number alteration across human patients: BCIRG 006 study. The 2005 San Antonio Breast Cancer cancers. Nature 2010; 463: 899–905. Symposium; December 8–11, 2005; San Antonio, Texas. 76 He Y, Wu J, Dressman DC, Iacobuzio-Donahue C, Markowitz SD, 49 Joensuu H, Kellokumpu-Lehtinen PL, Bono P, Alanko T, Kataja V, Asola R Velculescu VE et al. Heteroplasmic mitochondrial DNA mutations in et al. Adjuvant docetaxel or vinorelbine with or without trastuzumab for normal and tumour cells. Nature 2010; 464: 610–614. breast cancer. N Engl J Med 2006; 354: 809–820. 77 Barnett GC, Shah M, Redman K, Easton DF, Ponder BA, Pharoah PD. 50 DahabrehIJ,LinardouH,SiannisF,FountzilasG,MurrayS.Trastuzumabin Risk factors for the incidence of breast cancer: do they affect survival the adjuvant treatment of early-stage breast cancer: a systematic review and from the disease? J Clin Oncol 2008; 26: 3310–3316. meta-analysis of randomized controlled trials. Oncologist 2008; 13: 620–630. 78 Pharoah PD, Antoniou AC, Easton DF, Ponder BA. Polygenes, risk 51 Bergh J. Quo vadis with targeted drugs in the 21st century? J Clin Oncol prediction, and targeted prevention of breast cancer. N Engl J Med 2009; 27: 2–5. 2008; 358: 2796–2803. 52 Roukos DH. Mea culpa with cancer-targeted therapy: new thinking and 79 Wacholder S, Hartge P, Prentice R, Garcia-Closas M, Feigelson HS, new agents design for novel, causal networks-based, personalized Diver WR et al. Performance of common genetic variants in breast- biomedicine. Expert Rev Mol Diagn 2009; 9: 217–221. cancer risk models. N Engl J Med 2010; 362: 986–993. 53 Gandara DR, Gumerlock PH. Epidermal growth factor receptor tyrosine 80 Miller DG. On the nature of susceptibility to cancer. The presidential kinase inhibitors plus chemotherapy: case closed or is the jury still out? address. Cancer 1980; 46: 1307–1318. J Clin Oncol 2005; 23: 5856–5858. 81 Schinzel AC, Hahn WC. Oncogenic transformation and experimental 54 Stafford RS, Wagner TH, Lavori PW. New, but not improved? models of human cancer. Front Biosci 2008; 13: 71–84. Incorporating comparative-effectiveness information into FDA labeling. 82 Beerenwinkel N, Antal T, Dingli D, Traulsen A, Kinzler KW, Velculescu VE N Engl J Med 2009; 361: 1230–1233. et al. Genetic progression and the waiting time to cancer. PLoS Comput 55 Van Cutsem E, Ko¨hne CH, Hitre E, Zaluski J, Chang Chien CR, Makhson Biol 2007; 3: e225. A et al. Cetuximab and chemotherapy as initial treatment for metastatic 83 Goldhirsch A, Ingle JN, Gelber RD, Coates AS, Thu¨rlimann B, Senn HJ, colorectal cancer. N Engl J Med 2009; 360: 1408–1417. Panel members. Thresholds for therapies: highlights of the St Gallen 56 Mitsudomi T, Morita S, Yatabe Y, Negoro S, Okamoto I, Tsurutani J et al. international expert consensus on the primary therapy of early breast Gefitinib versus cisplatin plus docetaxel in patients with non—small-cell cancer 2009. Ann Oncol 2009; 20: 1319–1329. lung cancer harbouring mutations of the epidermal growth factor 84 Roukos DH, Ziogas D. From tumor size and HER2 status to systems receptor (WJTOG3405): an open label, randomised phase 3 trial. oncology for very early breast cancer treatment. Expert Rev Anticancer Lancet Oncol 2010; 11: 121. Ther 2010; 10: 123–128. 57 Bastiaens P. Systems biology: when it is time to die. Nature 2009; 459: 85 Peterson RT. Drug discovery: propping up a destructive regime. Nature 334–335. 2009; 461: 599–600. 58 Roukos DH, Tzakos A, Zografos G. Current concerns and challenges 86 Arora PS, Ansari AZ. Chemical biology: a notch above other inhibitors. towards tailored anti-angiogenic therapy in cancer. Expert Rev Nature 2009; 462: 171–173. Anticancer Ther 2009; 9: 1413–1416. 87 The human genome at ten. Nature 2010; 464: 649–650. 59 Ellis LM, Reardon DA. Cancer: the nuances of therapy. Nature 2009; 88 Collins F. Has the revolution arrived? Nature 2010; 464: 674–675. 458: 290–292. 89 Venter JC. Multiple personal genomes await. Nature 2010; 464: 676–677. 60 de Boer M, van Deurzen CH, van Dijck JA, Borm GF, van Diest PJ, Adang 90 Weinberg R. Point: hypotheses first. Nature 2010; 464: 678. EM et al. Micrometastases or isolated tumor cells and the outcome of 91 Golub T. Counterpoint: data first. Nature 2010; 464: 679. breast cancer. N Engl J Med 2009; 361: 653–663. 92 Human genome at ten: the sequence explosion. Nature 2010; 464: 670–671. 61 Roukos DH. Isolated tumor cells in breast cancer. N Engl J Med 2009; 93 Abbott A. Human genome at ten: the human race. Nature 2010; 464: 361: 1994–1995. 668–669. 62 Sotiriou C, Pusztai L. Gene-expression signatures in breast cancer. 94 Check Hayden E. Human genome at ten: life is complicated. Nature N Engl J Med 2009; 360: 790–800. 2010; 464: 664–667.

The Pharmacogenomics Journal