Personalized therapies in the cancer "omics" era
© Ocaña and Pandiella. 2010
Received: 5 May 2010
Accepted: 29 July 2010
Published: 29 July 2010
A molecular hallmark of cancer is the presence of genetic alterations in the tumoral DNA. Understanding how these alterations translate into the malignant phenotype is critical for the adequate treatment of oncologic diseases. Several cancer genome sequencing reports have uncovered the number and identity of proteins and pathways frequently altered in cancer. In this article we discuss how integration of these genomic data with other biological and proteomic studies may help in designing anticancer therapies "a la carte". An important conclusion is that next generation treatment of neoplasias must be based on rational drug combinations that target various pathways and cellular entities that sustain the survival of cancer cells.
A critical step towards defining a correct personalized anticancer therapy is the identification of the genes and pathways altered in the tumour of the patient, and the elucidation of their particular oncogenic role. The success of molecular studies in identifying potential molecular targets for therapeutic intervention is exemplified by the developments in the treatment of chronic myelogenous leukaemia (CML) . This disease is characterized by the presence of the Philadelphia chromosome, created by a translocation which provokes constitutive activation of the tyrosine kinase Abl. The identification of this molecular alteration fostered the development of drugs such as Imatinib Mesylate that inhibit Abl kinase activity and successfully control the disease . However, in most solid tumours multiple genetic lesions are expected to be required for tumour progression . In fact, the clinical experience in the treatment of oncologic diseases indicates that combinations of drugs that act on different cellular targets demonstrate superiority over single agent-based treatments . Yet, these combined treatments are mostly based on empirical trials and lack, in many instances, a biological rationale [5, 6]. Given the positive experience that molecular knowledge has offered for the treatment of CML, efforts have been made to define the molecular alterations present in the DNA of distinct types of tumours. Here we will comment how novel high throughput techniques may help in finding more adequate and less toxic personalized anticancer therapies.
Somatic alterations in the cancer genome
Somatic mutations in cancer exomes and pathways affected
Tumour type (sample size)
Average number of mutated genes/patient
Major pathways deregulated
Breast (n = 11)
RTKs, PI3K, NFκB,
Colorectal cancer (n = 11)
RTKs, PI3K, Cell adhesion, Cytoskeleton, Extracellular matrix
Colorectal cancer (n = 1)
Transcription (SPDEF), Metalloproteases (MMP28), PI3K, BRAF
Glioblastoma (n = 22)
RTKs, PI3K, Cell cycle, DNA damage (p53), Neuronal-type pathways (ionic channels), IDH1
Pancreas (n = 24)
KRAS signalling, DNA damage control, Cell cycle, TGFβ pathway, Wnt/Notch signalling, Cell adhesion and integrin signalling, MAPK signalling, Apoptosis
Glioblastoma (n = 206)
601 selected genes
RTKs, NF1, DNA damage, PI3K, Cell cycle, methylation, mismatch repair
Lung cancer (n = 188)
623 genes implicated in cancer
RTKs, DNA damage control, RAS (K and N), NF1, LRP1B (lipid metabolism), MAPK signalling, Wnt signalling, STK11 (Ser/Thr kinase)
Lung cancer (n = 371)
Cell cycle, PI3K, RTKs, Tyrosine phosphatases, cAMP, Angiogenesis, NKX2-1 (pneumocyte differentiation)
Lung cancer (n = 1)
Cell cycle (Rb), DNA damage (p53), DNA helicase CHD7
Mesothelioma (n = 4)
DNA damage, Extracellular matrix, Mitochondrial reductase activity, proteasome, Apoptosis
Diverse cancers (n = 210)
RTKs, JNK, MAPK,, BRAF, DNA damage control
Renal cancer, clear cell (n = 101)
Histone modifications (SETD2, JARID1C, UTX), VHL, NF2, HIF1A, PMS1 (DNA mismatch repair), WRN and NBN (DNA double strand repair)
Diverse cancers (n = 3131)
Kinases, cell cycle, NFκB, Myc, Apoptosis, Cell adhesion, DNA methylation, microtubule organization, transcription
Acute Myeloid leukaemia ((n = 1)
NRAS, NPM1, IDH1, CDC42, IMPG2, ANKRD46, LTA4H, FREM2, CEP170
Ovarian granulosa cell tumours (n = 4)
Drugs in clinical development against pathways identified in genomic/proteomic studies
Receptor Tyrosine Kinases
Histone Deacetylase inhibitors
Additional studies carried out in glioblastoma multiforme (GBM) [12, 13] and pancreatic cancer , in addition to describing the somatic alterations in the DNA, advanced to more precisely defining the molecular pathways altered in these pathologies. Interestingly, in these tumours the number of mutated genes was lower (around 40 mutations/tumour) than in breast or colorectal cancer. By using similar and complementary techniques to search for point mutations, as well as gains and losses of genetic material, the two different reports analyzing the genomes of patients with GBM satisfyingly came to common conclusions [12, 13]. These reports observed frequent alterations in three major signalling pathways that control cell proliferation: the receptor tyrosine kinase-PI3K route, the p53, and the retinoblastoma (RB) tumour suppressor pathways. Interestingly, some of these pathways were also found to participate in pancreatic as well as colon and breast cancer tumours, indicating overlapping of signalling pathways that may be critical in the genesis/progression of solid tumours. Interestingly also was the fact that mutations in the genome of GBM patients accumulated in patients treated with the alkylating agent temozolomide, a chemotherapeutic used in this pathology which is also highly mutagenic. Moreover, another study in which the genes coding for kinases of patients with several types of cancer were analyzed also showed that the highest prevalence of mutations in the kinases corresponded to GBM patients treated with temozolomide . This indicates that resistance mechanisms may develop in these patients due to additive molecular alterations that favour the development of clones of cells resistant to the action of classical treatments. The sequencing efforts carried out in GBM also identified isocitrate dehydrogenase 1 (IDH1) as a protein that could act in the pathogenesis of this disease [13, 21]. Interestingly, IDH1 was also found to be mutated in the genome of AML patients .
In lung cancer, using single nucleotide polymorphism (SNP) arrays, Weir et al.  analyzed the presence of copy-number alterations of chromosomal regions in a large proportion (n = 371) of lung tumours, and found frequent gains or losses of chromosomal regions. Some genomic alterations, such as copy number gain of chromosome 5p occurred in a high number of patients (60%). These alterations affected genes known to be frequently involved in lung cancer, such as EGFR/HER1, CCNE1, or KRAS. In addition, the paper describes a novel player in lung cancer pathophysiology, the NKX2-1 gene product, a transcription factor implicated in the formation of lung pneumocytes. Knocking down the expression of this protein in NCI-H1925 lung cancer cells decreased their ability to grow in an anchorage-dependent manner, indicating that this protein may represent a novel lung cancer promoting oncogene, and an interesting novel therapeutic target. Another report analyzed somatic mutations of 623 genes in 188 lung cancer samples . These genes were selected for their already known implication in oncogenesis. KRAS or EGFR were mutated in a substantial proportion of patients; but, in addition, several other genes not formerly associated to lung cancer were identified, and included tumour suppressors (NF1, RB, ATM, and APC), as well as tyrosine kinase genes (ERBB4, ephrin receptor genes, KDR, FGFR4, and NTRK). Of note, 132 of the 188 tumours had at least one mutation in genes that participate in MAPK signalling. Also, mutations in multiple genes of the Wnt pathway were observed in 29 of the 188 samples. Frequent mutations were also detected in DNA damage response genes, including TP53, and ATM. This clustering of mutations in certain pathways which play a role in oncogenesis points to the possibility of interfering with them for the treatment of lung tumours. Moreover, from a more ample perspective the molecular studies mentioned above open outstanding possibilities in terms of better focusing on certain targeted therapies for these tumours, based on drugs that target components of the identified signalling pathways. Another report confirmed the relevance of p53 and Rb in lung cancer . Here the researchers sequenced the whole genome of a lung cancer cell line and found 22,910 somatic mutations, of which 134 were included into coding exons, 94 of them causing changes in the primary sequence of the coded proteins.
An ample study of somatic copy number alterations performed on 3131 tumour samples, and also using high resolution SNP arrays identified a median of 12 gains and loses in each patient tumour . Frequent alterations in the control of cell cycle progression, apoptosis, DNA damage control, or kinase activity were reported.
The contribution of functional genomics to cancer therapy
A strategy that has been recently developed to massively identify potentially useful therapeutic targets is based on functional genomics studies using RNA-interference (RNAi) screenings. In one such functional screen based on the knockdown of 2,924 genes selected for their potential implication in tumour generation/progression, Schlabach et al. identified between 80-150 gene products important for tumour survival/proliferation . Of them, 19 genes were shared by the three distinct tumoral cell lines analyzed (two from colon cancer, and one from breast cancer). Interestingly, the list of the identified proteins showed significant overlap between distinct tumoral cell lines, but were different from those identified to be essential for the survival of normal breast epithelial cells . This indicates the existence of qualitative differences between normal and malignant cells with respect to proteins that support their respective viabilities. This is highly relevant from the therapeutic point of view, as it indicates that the targeting of proteins that specifically support survival of cancer cells is realistic, and may spare normal cells, therefore representing an efficient and likely safe therapeutic strategy. In addition, these authors also found a certain degree of overlapping between proteins that support viability in tumours from the same tissue origin. However, they also found proteins whose function was required to sustain viability of one tumour cell, but did not play a critical survival role in others. Thus, knock down of the protein phosphatase PP1 seriously affected survival of HCC1954 colon cancer cell line, but did not substantially affect DLD-1 colon cancer cells. These results support the concept of the distinct susceptibility of different tumours, and stress the importance of adequately selecting protein targets to achieve therapeutic success. Analogous functional genomic studies in haematological malignancies led to the identification of the IRF4 and genes regulated by this protein as potential therapeutic targets in multiple myeloma , and the demonstration that tyrosine kinases play an important role in sustaining AML vitality . This last work is particularly interesting as the authors used fresh samples from patients to identify, by also using RNAi techniques, potentially relevant drug targets. More ample RNAi screening techniques, based on the knockdown of up to 17,000 different genes , have allowed an almost universal analysis of the participation of the proteome in tumour survival, and have identified pathways and proteins that participate in tumour proliferation/survival .
In addition to finding potentially useful drug targets, these RNAi-based functional genomic screens have also been used to uncover mechanisms of drug resistance. Bernards and colleagues explored proteins involved in trastuzumab resistance in breast cancer and identified PTEN as one of the principal proteins whose lack of function was linked to trastuzumab resistance , in agreement to previous reports . Moreover, preclinical data indicated that combination of agents that target HER2 and the PI3K route reverts resistance to trastuzumab .
Biological findings support the use of combined targeted therapies
An important aspect of the functional genomics assays is the finding of individual proteins whose targeting is deleterious for the tumoral cell, opening the door to the development of drugs that by interfering with their action may be therapeutically relevant as single agents. However, as mentioned above, the clinical experience suggests that efficient cancer treatment usually relies on the combination of agents that target distinct oncogenic networks. The fact that combination of agents may be more efficient than single agent treatment may be due to the heterogeneity of tumours, in some cases promoted by some anticancer treatments that either act as mutagens, as mentioned above in temozolomide-treated patients, or by the mutational heterogeneity within a single tumour. This heterogeneity has been recently proposed as a limit for the value of high throughput sequencing efforts in individual tumours . The cross-talk between intracellular signalling pathways mainly activated by RTKs represents another reason to explain the efficacy of combination strategies. It has been reported that inhibition of certain pathways, such as the mTOR route may lead to increased MAPK activity . Therefore, ablation of signals through both routes is required for efficient antitumoral action. Moreover, combined inhibition of PI3K and MAPK routes has shown superior antitumoral effect compared to individual targeting of either pathway .
Elegant biological studies supporting that adequate targeted drug combinations based on genomic profiling may be effective in cancer treatment has been offered by reports from the Massagué group [34, 35]. These researchers used genomic and imaging techniques to analyze a particularly relevant property of tumoral cells, i.e. metastatic dissemination. Their strategy was based on the injection of a breast cancer cell line into the bloodstream of mice, and the selection of tumoral clones that homed to several organs. In their studies, genomic profiling identified signatures of genes that were particular of the cells that metastatized to a certain tissue. The researchers then selected some of these genes for further biological analyses, and suspected that four of them, Epiregulin, COX2, MMP1 and MMP2, represented therapeutic candidates in the case of lung metastases. In fact, experiments of gene knockdown confirmed the importance of these genes in breast cancer tumour growth in nude mice. Interestingly, single knockdown of only one of the coding mRNAs had little effect on the growth of these tumours as compared to double knockdowns. But the most impacting results were obtained by quadruple knockdowns. Reduction of the expression of the four mRNAs coding for the respective proteins fully abrogated tumour growth. These genomic-functional studies were complemented with targeted therapies aimed at neutralizing the activity of the four selected protein targets. Again, single agent treatment was less effective than combinations in preventing extravasation and lung colonization by the tumoral cells. Also, combination of three targeted agents, expected to neutralize the function of all four proteins, was more effective than dual combinations, and the triple combinations almost fully prevented lung invasion by the tumoral breast cancer cell population. Another recent analogous study identified a set of genes that mediate breast cancer metastasis to the brain . Interestingly, some of them (COX2 and HB-EGF) share molecular identity or properties with those formerly described to participate in metastatic spreading to the lung. However, other proteins such as the sialystransferase ST6GALNAC5 was specific for cells that metastatized to the brain, and could be involved in facilitating cell passage through the blood brain barrier. The findings of Massagué and colleagues come in support of the well established clinical concept that drug combinations are superior in efficacy to single agents to treat solid tumours. The added value of their work is the use of molecular and biological tools to define and verify adequate targets for therapeutic intervention.
Combined targeted therapies guided by proteomic studies
Probably the genomic portraits are insufficient in guiding the selection of antitumoral therapies. Indeed, studies on receptor tyrosine kinase (RTK) targeting in cancer have offered insights into the convenience of complementing the genomic data with proteomic studies to establish efficient anticancer therapies. Stommel et al.  analyzed the phosphotyrosine content (indicative of activation) of 42 different RTKs in GBM. This disease may present alterations of the EGFR that result in its constitutive activation. However, response rates to agents that target exclusively the EGFR are poor. Stommel et al. found consistent activation of three or more RTKs in 19 out of 20 GBM cell lines. These RTKs included the EGFR, ErbB3/HER3, platelet-derived growth factor receptor α (PDGFRα), and MET. These researchers then explored the value of targeting these activated receptors using several in vitro models. Inhibition of the activity of individual receptors had a marginal effect on the growth properties of the GBM cell lines. However, combination of drugs that targeted two receptors was superior to single agent treatments. Furthermore, triple combinations were more efficient than double combinations, and practically abolished tumour growth in soft agar colony forming assays. The authors conclude that coactivation of multiple RTKs may sustain cell proliferation in GBM, and that adequate treatment of this pathology must include a careful evaluation of the RTKs activated.
In line with those findings, two additional reports showed the importance of targeting multiple RTKs in breast  and lung cancers . Both studies attempted to identify the mechanisms of resistance to therapies that target HER (Human EGFR-like Receptors) receptors. In the breast cancer study, resistance to gefitinib or erlotinib, agents that act on the EGFR/HER1, was accompanied by increased tyrosine phosphorylation of ErbB3/HER3 . Similarly, gefitinib resistance in lung cancer was also found to be accompanied by ErbB3/HER3 signalling . This up-regulation of ErbB3/HER3 signalling was due to amplification of MET. Concomitant treatment with gefitinib and the MET inhibitor PHA665752 provoked a decrease in cell survival not obtained by single drug treatments. The fact that multikinase inhibitors have reached the oncology clinic adds value to these experimental results and paves the way for the development of multikinase targeting strategies guided by proteomic analyses of the activation state of the kinome.
Limitations of targeted therapies based on genomic profiling
Another factor that limits the efficacy of targeted agents is the presence of parallel activation of downstream pathways. A well characterized example is the preclinical identification of the lack of antitumor activity of agents against RTKs when mutations at the K-RAS and PI3K genes were present [43, 44]. Translation of these studies to the clinic confirmed lack of activity of anti-EGFR antibodies like cetuximab or panitumumab in colon cancer tumours with K-RAS mutations [45, 46]. These studies have been of much importance in selection of anti-EGFR therapies in colorectal cancer, as patients whose tumours express mutated forms of K-RAS are now excluded from treatments based on anti-EGFR drugs. In addition, among colorectal tumours carrying wild-type KRAS, other additional mutations in genes participating in EGFR signalling may cause resistance to anti-EGFR therapies. Thus, mutation of BRAF or PIK3CA or loss of PTEN expression may result in resistance to EGFR-targeted monoclonal antibody treatment [47, 48].
Another important tumour component that is gaining importance as a targettable cell type is the stromal compartment. In multiple myeloma, a disease characterized by the accumulation of tumoral plasma cells, the interaction of these cells with the stroma is expected to be critical for sustaining survival of the myelomatous cells, and to provide drug resistance . Moreover, treatments that target both the myeloma cell and the bone marrow microenvironment, such as the proteasome inhibitor bortezomib, or the immunomodulatory agents derived from thalidomide, have shown anti-myeloma activity and have reached the myeloma clinic . Interestingly, genetic studies have demonstrated the presence of genetic alterations in the bone marrow mesenchymal cells of patients with multiple myeloma, supporting the concept that the stroma in the vicinity of the tumour may be cancerized . In other neoplastic pathologies such as breast cancer, mesenchymal stem cells have been shown to promote metastatic dissemination of the tumoral cells . Moreover, recent studies perfomed in myelodisplastic syndromes support the concept that initial alterations in the stromal compartment may favour the generation of secondary leukemias in a mouse model, further supporting a potential role of the stroma in the pathophysiology of at least certain tumours . Therefore, efficient antitumoral therapies must also take into consideration these variables in order to achieve long lasting remissions.
The power of high scale "omics" analyses coupled with the ample portfolio of targeted drugs under clinical development and already approved offers hope for a more effective and individualized anticancer therapy. It is rewarding to observe that sophisticated genomic, proteomic and biological studies reinforce the already known clinical strategy of using drug combinations to treat cancer patients (Figure 1). The molecular and biological data herewith commented also offers clues as to how many targets, pathways or functions should be attacked. The preclinical evidence that targeting from one to four proteins may impede proliferation, survival or invasion properties of cancer cells indicates that these cells depend on a limited number of proteins to carry out these biological functions. That the studies mentioned above have correctly targeted some of these molecules is beyond doubt. However, it is possible that targeting other proteins could have a similar impact. In fact, few overlap exists between the molecules targeted in each of these individual papers, especially those that used functional genomics to identify potential targets. This can be interpreted as to say that these papers have found "a set", (but not a unique "set") of targets that are therapeutically relevant. Finding a correct algorithm for the combination of drugs that target a set of proteins critical for sustaining cancer cells should be pursued.
In conclusion, recent results using new "omic" techniques support the future use of targeted drug combinations for the treatment of solid tumours. In our opinion to increase the efficiency of this process, it is critical to reach a close collaboration between academia, the pharmaceutical and biotechnological companies as well as regulatory authorities to develop rationale studies with drug combinations based on strong preclinical data.
We thank Prof. I. Tannock for the critical reading of the manuscript, and suggestions. Work in AP lab receives support from the Ministry of Science and Education of Spain (BFU2006-01813/BMC and BFU2009-07728/BMC), from the ISCIII (RD06/0020/0041), and from the European Community through the regional development funding program (FEDER). AO receives support from the Spanish Association Against Cancer (AECC), from the FISCAM grant program PI2007/41 and Proyectos de Investigación en Salud (PS09/02144). The CIC receives institutional support from the Fundación Ramón Areces.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.