|Home | About | Journals | Submit | Contact Us | Français|
The progress achieved in the field of genomics in recent years is leading medicine to adopt a personalized model in which the knowledge of individual DNA alterations will allow a targeted approach to cancer. Using pancreatic cancer as a model, we discuss herein the fundamentals that need to be considered for the high-throughput and global identification of mutations. These include patient related issues, sample collection, DNA isolation, gene selection, primer design, and sequencing techniques. We also describe the possible applications of the discovery of DNA changes to the approach of this disease and cite preliminary efforts where the knowledge has been translated into the clinical or preclinical setting.
Twenty-eight years after Frederick Sanger made his first attempts to sequence small pieces of DNA, the National Human Genome Research Institute (NHGRI) announced the completion of the Human Genome Project [1-3]. After 13 years of international collaborative effort at a cost of $3 billion, an essentially complete human genome sequence was finished and the world was ushered into a “genomic” era. In 2007, less than 4 years later, the first genome sequence of a single individual, James Watson, was deciphered in only 2 months at a cost of less than $2 million. Project “Jim” was initiated by a private company in collaboration with the Baylor College of Medicine-Human Genome Sequencing Center (BCM-HGSC) [4,5]. This project demonstrated the feasibility of characterizing virtually all of the small scale variation in a single individual using a next generation sequencing technology thereby advancing the “personalized” genomic medicine model one step closer to a reality.
Cancer has long been recognized as fundamentally driven by genetic mutations [6,7]. Although opinions vary as to whether a comprehensive inventory of the genetic changes in tumor cells can lead to fundamental new insights that will further the diagnosis and treatment of this cancer, pilots for large-scale projects already have been initiated [8-10]. Sjöblom et al have shown that the somatic mutations playing a role in the multistep progression of carcinogenesis are far from completely identified, even in the most studied cancer types (breast and colon) . The Tumor Sequencing Project Consortium (TSPC), formed between the NHGRI funded Genome Centers at BCM, Washington University, and the Broad Institute has initiated a pre-pilot project on non-small cell lung cancer (1,000 genes) to demonstrate the potential of a systemic approach to tumor genotyping by DNA sequencing . The benefit of this large-scale project already has been proven by the identification of a novel candidate proto-oncogene . The Cancer Genome Atlas pilot project (TCGA), launched by the National Cancer Institute (NCI) and NHGRI, is now attempting the identification of all genomic alterations significantly associated with cancer. This includes detecting genomic loss or amplification, mutations in coding regions, chromosomal rearrangements, aberrant methylations, and expression profiles. Three tumor types (glioblastoma multiforme, squamous cell carcinoma of the lung, and ovarian carcinoma) have been selected initially by TCGA, with the scope of expanding to all major cancer types .
Some success in cancer care has already been obtained with the targeting of specific genetic alterations. Among the first effective applications of targeted therapy was the use of imatinib to inhibit the tyrosine kinase activity of the Bcr-Abl fusion protein formed by the chromosomal t(9:22) translocation in chronic myeloid leukemia , and the use of transtuzumab, a recombinant monoclonal antibody against HER2, in women with metastatic breast cancer with HER2 amplification and overexpression . Other breakthroughs in personalized therapies include the treatment of gastrointestinal stromal cell tumors, in which mutations in KIT and PDGFRA were found to predict a response to imatinib [15,17]. Additionally, mutations in EGFR predicted a response to gefitinib and erlotinib in lung adenocarcinoma [18-20]. The fact that specific mutations in multiple loci are a major determinant of the response to targeted therapies suggests that DNA sequencing is likely to provide an effective diagnostic and therapeutic approach to cancer in the future.
This review focuses on the requisites to identify, validate, and confirm mutations, as well as the possible applications the discovery of new DNA changes can have for the characterization and treatment of the disease. Some technical information is provided for those who may be interested in initiating similar projects, and some examples of preliminary attempts to apply current knowledge in a clinical setting are given with the goal of attracting new clinical investigators to bridge the gap between genomics and its application to everyday patient care.
Generating genomic information on a large scale requires strict criteria to protect patient’s rights and confidentiality . Institutional Review Board (IRB) oversight and informed consent are unambiguously required. Genome sequencing should be explained to potential participants in the informed consent process. Specific consent should be obtained for the future storage and use of collected samples. For most federally-funded studies, broad data sharing is required. (NHGRI Data sharing policies) Explanation of the anticipated scope of data sharing, along with the risks and benefits of broad data release, should be provided. Under some circumstances (e.g., if data sharing is not required to achieve the primary goals of the project and data release may impede research participation), it may be appropriate to offer an opt-out provision for public and/or restricted data broadcast [21,22].
Studies involving extensive genome sequencing raise an additional concern about whether individualized research results should be shared with study participants. Convincing arguments have been made that if the research reveals validated data of known clinical relevance, it should be reported to participants [23-27]. The language used to explain return of research results should be carefully crafted to avoid potential legal liability.
The patient information collected should include demographics, exposure, family history, symptoms and physical findings at presentation, laboratory values, diagnostic imaging test results, details of the surgical treatment, histology from preoperative and operative specimens, pathologic staging data, details of chemotherapy and radiation treatment, response to treatment in terms of follow-up imaging, disease-free survival and overall survival, and quality of life survey data.
The data should be entered and stored in a password-protected, HIPPAA-compliant database. To assure patient confidentiality, the specimen should be logged into the database and then assigned a new serial number for use in the laboratory. Limited access by clinicians and biostatisticians will allow later correlation with clinical data. The tubes containing the samples also should be bar coded to achieve automatic assignment, increase speed of processing, and protection from mixing the samples.
The number of patient samples to be sequenced depends on the number of available samples and the size and statistical power of the intended study. We have noticed that the number of modification event detected increases when a larger panel of patients is used in the discovery process (11 patients versus > 40 patients) (Table 1). The issue of statistical power has been raised in reviews of Sjöblom’s paper [11,28,29]. Much larger samples than the ones used in that paper are required to detect cancer genes, since in small sample sizes, some candidate genes are expected to display false-positive results. The validation and specificity processes should also be planned with a statistically valid number of patients.
It is necessary to collect both tumor and matched normal tissue from each patient for the comparison of germline and tumor genotypes in order to verify that any detected mutations are of somatic origin. Matched normal tissue can consist of surrounding normal tissue and/or blood. Collecting the blood of the patient as the primary source of normal germline DNA is preferable. If the blood is not available, a sample of “normal” tissue taken from an area adjacent to the tumor can be used as an alternative. This choice is less attractive because a “normal” sample may not be entirely normal and may contain some invasive neoplastic cells or normal cells that share alterations with the cancer . In such a case, the sequencing result might be identical between the tumor and “normal” tissue, representing an inherited germline mutation or polymorphism, rather than a somatic mutation.
Samples should be identified as primary tumor tissue or metastasis or abnormal non-cancerous tissue. Strict requirements of quality, quantity, purity, and avoidance of necrotic tissue should be met in order to create an ideal tissue bank. Tissue quality issues encountered by large projects like TCGA can assist in better planning for future projects [31,32]. The degree of homogeneity of the tumor sample to be sequenced is also important. First, it is known that the tumor itself is not homogenous, since it consists of subpopulations of cells with different phenotypes and genotypes that determine different immunity, aggressiveness and metastatic behavior [33,34]. In addition, the contamination of tumor samples by surrounding tissue must be considered. This has led to the use of a tumor purity threshold of at least 80% as a criterion by TCGA and others [13,35,36] and to the development of Laser Microdissection Technique, especially if the cell population of interest is scant .
Fresh tissue is the best source of DNA. However, if the DNA cannot be extracted immediately, freezing remains the best way to preserve tissue. Standard operating procedures for the collection of fresh frozen tissue samples have been developed and used in the European Human Frozen tumor Tissue Bank (TuBaFrost) . Formalin-fixed, paraffin-embedded (FFPE) tissue also can be used, but with limitations [39-41]. In such material DNA is scant, and cross linking and degradation result in sequencing failure. FFPE tissue constitutes a valuable source in terms of the number of available samples, diversity, and global patient information that has been collected over decades. When using such a source, the DNA has to be whole genome amplified and only short amplicons (~100-200 bases) can be sequenced. As a consequence, FFPE tissues can be used in the validation process, but are not preferred for mutation/SNP discovery. Cell lines are not an optimal DNA source since mutations associated with long term in vitro culture and unrelated to the in vivo development of cancer can be acquired over time.
In cases when the amount of the provided human specimen is limited and the extracted DNA is scarce, whole genome amplification (WGA) is required to overcome the problem. By this method, the original DNA sample is amplified in a specific way from nanogram concentrations to microgram, while the sequence representation of the template is conserved. Of the different methods studied, multiple displacement amplification (MDA) results in DNA products of high molecular weight (up to 12kb) and generates the least bias [42-44]. Original unamplified samples can then be saved to validate the identified mutations.
There is a wide range of genes that can be chosen for a sequencing study: genes with mutations already known to be associated with the tumor by medical literature and data sources [46-49]; genes with mutations associated with familial genetic syndromes that increase the risk of cancer [50,51]; genes known to be differentially expressed in the tumor [52,53]; genes linked to several other cancer types and cellular pathways important in cancer (e.g., oncogenes, tumor suppressor genes, DNA repair genes, cyclins, kinases, phosphatases) . CancerGenes is a valuable source of information for gene selection and prioritization in the elaboration of a gene list .
High-throughput DNA sequencing methods have been used on data sets of 200 to 1000 genes in 150-200 patients. In these large scale-up studies, considerable effort is expended in finalizing the gene list [12,14]. In an approach to exon sequencing, Sjöblom et al expanded the list to 13,023 genes by limiting the study to a single direction in only 11 patients per tumor type . The power to detect rare mutations was therefore limited in this study, and yet 365 mutations in 236 genes were found. As the ability to comprehensively sequence tens of thousands of exons in a single experiment continually increases, the problem of gene selection diminishes.
This sequencing ability has been further improved by another novelty: the replacement of PCR by capture arrays (Fig. 1). Until now, regions of interest to be sequenced were selectively sampled through a labor-intensive process whereby each fragment was individually amplified using PCR. This required the parallel design and execution of thousands of reactions. This will now change as NimbleGen (Madison, WI) developed seven custom arrays that contained 204,000 exons, allowing the entire exonic coding genome to be captured, enriched, and sequenced on the 454 or Genome Analyzer platform upon release (Fig. 2) [56,57]. Such enrichment will eliminate the need for large numbers of specific primers, thousands of specific PCR reactions on individual or pooled samples presently associated with the PCR-based large-scale sequencing approaches, thus considerably reducing cost and time expense. This will give superior results, allowing for the identification of polymorphisms/mutations on the whole exonic genome.
Currently, when performing an exonic study, specific amplification by PCR of the selected gene exonic regions has to be performed prior to sequencing and specific primer pairs need to be designed (Fig. 1). A pipeline can be established that links several steps for the integrated automating design of primers (Table 3).
The DNA sequencing revolution started in 1977 with Sanger sequencing that soon became by far the most frequently used sequencing technology (Fig. 4) . The Sanger sequencing method, a termination technology, is still widely used to perform individual amplicon sequencing (400-500 bases/read in a 384-well plate = ~150,000 bases/plate). However, the interest in deciphering the whole genome of organisms and large-scale DNA sequencing projects recently prompted the development of other technologies and platforms accommodating high-throughput sequencing.
Discovered in the late 1990s, pyrosequencing technology is based on the real-time monitoring, by bioluminescence (conversion of luciferase into oxyluciferin), of a 4-enzyme sequencing reaction . This technology recently has been adapted to a high throughput setting by 454 Life Sciences in which whole genome or targeted segments can be analyzed in a single run . The technology is based on emulsion PCR of individual DNA fragments captured on 28-micron beads, at a resolution of one DNA molecule per bead, resulting in a 107-fold amplification of the initial DNA copy per bead, followed by pyrosequencing-by-synthesis of each clonally amplified template in a fiber optic slide (Fig. 5). Presently, approximately 500,000 reactions can be performed in parallel in the new FLX system and about 125,000,000 bases are sequenced per run (with each read currently at 200-300 base pairs long, the expectation is that in late 2008, 500 base pair reads will become available). Whereas patient samples and amplicons are pooled making it less expensive and more rapid for discovery, the physical segregation of the DNA-carrying beads in an emulsion during the in vitro amplification (clonal amplification) results in the detection of specific mutations in low tumor content samples without the need for tumor cell enrichment by laborious methods. The extreme sensitivity of the 454 technology has been recently demonstrated in a study in which it enabled the detection of mutations in low tumor content samples for which conventional Sanger sequencing has failed to detect any of these mutations . The call for somatic mutation and the percentage of patients carrying a mutation are then estimated during validation by genotyping the matching normal and tumor samples individually for each patient using a different technology.
In addition to the 454 technology, there are several new high-throughput sequencing methods that are being explored. Among these, the Solexa system (recently acquired by Illumina (San Diego, CA) and renamed Genome Analyzer) is a new, massively parallel sequencing platform in which millions of single molecules are covalently attached to a planar surface and amplified in situ by a “bridge amplification” process [66-68]. Sequencing by synthesis is then carried out by adding a mixture of four fluorescently labeled reversible terminators and DNA polymerase to the template on the Solexa flow cell. The nucleotide sequences are determined by the fluorescent signals. After removing the fluorophore and reversing the blockage group and the terminator, the terminator-enzyme mix is added to start a new cycle. The whole process is then repeated until the end of the run (Fig. 6). One nucleotide sequence is read out for a given molecule at each cycle. One Solexa run, on average, generates about 35-45 million reads with read lengths of 36 bases. Between 1.2 and 1.5 billion bases are sequenced per run, about 10 fold higher than the 454 technology throughput. However, the error rate for a single read generated from the Solexa platform is about 1.5% for 36 cycles. As a result, the detection of mutations in samples with low tumor content might be hard to differentiate from the error rate background. The challenges of the technology also include a massive data storage and computional load to manage. Presently, due to the short read length, Solexa has better use in a long-range PCR-based approach (>3 kb) or a whole genome approach.
Other important characteristics associated with tumor development and progression are the variation in gene copy number due to heteroploidy and chromosomal loss. This genomic aspect cannot be detected by sequencing but with techniques such as CGH and SNP arrays [69-70]. The SKAP2/SCAP2 gene was found to be amplified and associated with the development of pancreatic cancer using this technology .
The data derived from sequencing can be aligned to identify mutations by parallel comparing of the tumor DNA sequence, the patient normal DNA, and the reference sequence deposited in GenBank (Table 4). This comparison can reveal single nucleotide polymorphisms (SNPs), germline, and somatic mutations. Modifications from the GenBank reference sequence in the patient samples (sequence identical in the normal and tumor samples) are germline mutation or SNP. Somatic mutations are modifications specific to the tumor and not found in the blood or other normal tissue of the patient.
It is now believed that polymorphic variations in the DNA sequence also can be related to population-attributable cancer heritability. SNP is defined as a genomic locus where two or more alternative bases occur with a frequency of at least 1% in a population. SNP accounts for more than 90% of the total variation in the human genome . There are as many as 7 million common SNPs with a minor allele frequency of at least 5% [73,74]. SNPs in close chromosomal proximity can be inherited together on haplotype blocks due to underlining linkage disequilibrium (non random association of alleles from one generation to the next) . SNPs indeed have been associated with predisposition to cancer, prognosis, and response and toxicity to chemotherapy or radiotherapy [76-78]. SNPs in xenobiotic metabolizers, hormone metabolizers, DNA repair genes, genes involved in angiogenesis and cell cycle are under scrutiny in several cancers .
Discovery of the germline mutations (mutations found in every cell of the individual) predisposing to hereditary cancer syndromes was the trademark of translational cancer research in 1990s. Genes such as BRCA1/BRCA2 in breast and ovarian cancer, APC in colon cancer with adenomatous polyposis coli and CDNK2A in melanoma are examples of mutated genes with high penetrance leading to cancer and studied in family pedigrees [80-82].
Based on observation made by Loeb and others, at least six different metabolic or signaling pathways must be altered to lead to cancer [83-85]. By these alterations cells express insensitivity to growth inhibitory signals, escape apoptosis and acquire limitless replicative potential and sustained angiogenic, invasive and metastatic abilities . As normal mutation rates cannot by themselves account for the multiple mutations found in cancer cells, it is believed that special mutations called mutator mutations lead to genetic instability and increase the inherent rate of genetic change, thus exhibiting a “mutator phenotype” [85,87].
The whole list of cancer associated genes includes oncogenes, tumor supressor genes and stability genes. Oncogenes can be abnormally activated by intragenic mutation, chromosomal translocation or gene amplification. Tumor suppressor genes can be inactivated by a missense mutation, nonsense mutation, deletion or insertion of various sizes; by epigenetic silencing; or by amplification of regulatory inhibitors. Finally, the function of stability genes including mismatch repair genes, nucleotide excision repair and base excision repair genes, as well as mitotic recombination and chromosomal segregation genes also can be impaired by a mutation .
These DNA modifications can be found in coding (exon) or non-coding (intron and untranslated exonic) regions. Exonic mutations in the coding region can be synonymous (silent) in which the change in base does not affect the amino acid call, or non synonymous resulting in a different amino acid (missense) or protein termination (stop codon, non sense). The non-synonymous mutations can have drastic effects on the protein function and structure. On the other hand, the impact of the silent polymorphisms in the MDR1 and the Lamin A genes have proved that synonymous mutations should not be overlooked [88,89]. Mutations in non-coding regions located in introns, promoters, splice junctions or untranslated regions of the gene may also contribute to cancer by changing the regulation, exon splicing, mRNA stability, or conformation of the protein [90-93]. Table 5 illustrates the different types of modification events we found in pancreatic adenocarcinoma using Sanger and 454 technology.
Although many somatic mutations can be detected in tumors, it is essential to distinguish between driver mutations which actually contribute to cancer and passenger mutations which randomly happened but are not responsible for cancer [11,94]. In the case where the function of the genes is still not clearly known, the impact of the mutated gene on the development of cancer will have to be evaluated by functional studies.
Each sequencing technology has its own limitations and pitfalls. After the completion of sequencing and the discovery of mutations, a second technique based on different principles should be performed in order to validate the obtained genotype. There are a number of genotyping methods including TaqMan SNP Genotyping Assay (Applied Biosystem), MIP (ParAllele), SNPStream (Beckman-Coulter), iPlex Gold Assay (Sequenom), GoldenGate Assay (Illumina) and for genome wide studies GeneChip (Affymetrix) and Infinium (Illumina) [95-101]. Our strategy was to use the Sanger or TaqMan SNP Genotyping Assay to validate 454 pyrosequencing and low throughput pyrosequencing (Biotage) to validate Sanger results. For example, 454 high throughput pyrosequencing detected base modifications, but individual sample validation done by Sanger technique allowed the comparison between tumor and matched normal tissue and thus the differentiation between germline and somatic events (Table 6). We also used differential gel migration of amplicon to validate large deletions.
In order to confirm that the identified mutations are specific to the malignant tissue, other tissues should be genotyped. As an example, we intend to genotype pancreatitis, precancerous lesions, benign cystic tumors, and neuroendocrine tumors to discriminate between mutations associated with pancreas cancer and non-cancerous or other cancerous conditions.
The fact that genomics led to deeper understanding and more targeted approach to CML and solid cancers like breast, gastrointestinal and lung cancer is indisputable. In the case of pancreatic cancer, translation of information from the DNA level to the clinical practice is still preliminary and experimental. Effective screening or diagnostic genomic tools have not been established. Targeted therapies based on mutations are just beginning to be tested.
A recurrent pattern of genetic changes associated with pancreatic carcinogenesis has been identified . The genetic changes include the inactivation of CDKN2A (p16) and an activating point mutation of K-RAS in codon 12 as early events, as well as inactivation of the SMAD4 (DPC4), TP53, and BRCA2 genes as later events [103,104]. K-RAS mutation and CDKN2A inactivation are detected in over 90% of the tumors; however, SMAD4 and TP53 inactivation can only be found in 55%, and 50-75% of the adenocarcinoma, respectively .
While tobacco smoking is a high risk factor for the development of pancreatic adenocarcinoma, numerous associations have been found between smoking behavior and genetic polymorphism in genes responsible for nicotine metabolism . Indirect association of K-RAS mutation and activation with occupational exposure to dyes, organic pigments and other agents has been suggested [107,108]. Moreover, a metanalysis evaluating folate intake and genetic polymorphism of 5,10 methylene tetra hydro folate reductase (MTHFR) found a MTHFR variant associated with an increased risk for pancreatic adenocarcinoma . A polymorphism in glutathione S-transferase gene, affecting detoxification of carcinogens and anticancer agents in the human pancreas, was found to confer a protective effect against pancreatic adenocarcinoma .
Other gene polymorphisms that may determine the risk for pancreatic adenocarcinoma include Aurora-A and CDKN2A (p16) polymorphisms, which were associated with diagnosis of pancreatic adenocarcinoma at an early age . Finally, polymorphisms in DNA repair genes XPD and XRCC2 have been studied as genetic risk modifiers for smoking-related pancreatic adenocarcinoma [112,113], while an XRCC1 polymorphism had a significant interaction with the APE1 or MGMT polymorphism in modifying pancreatic adenocarcinoma risk .
A growing body of evidence suggests that some of the aggregation of pancreatic adenocarcinoma in families has a heritable genetic basis and that as many as 10% of pancreatic adenocarcinoma could be hereditary . Several familial genetic syndromes already have been associated with an increased risk of pancreatic adenocarcinoma such as hereditary pancreatitis, the hereditary nonpolyposis colorectal cancer syndrome (HNPCC), breast cancer, ataxia-telangiectasia, the familial atypical multiple mole-melanoma syndrome (FAMM), and Peutz-Jeghers syndrome. Already, germline mutations in BRCA2, CDKN2A, STK11, FANCC, PRSS1, and palladin (PALLD) have been shown to predispose to pancreatic adenocarcinoma, although with incomplete penetrance [115-117].
Researchers already have used genetic markers separately or concomitantly to analyze cellular material from pancreatic juice and fine needle aspirates [118-122]. One of these studies has indicated that the analysis of K-RAS mutation and TP53 and SMAD4 inactivation can complement traditional cytology and clarify the diagnosis of patients with atypical biopsy samples . EUS-FNA biopsy specimens also are beginning to be examined for some of the mutations associated with pancreatic adenocarcinoma. In one study, analysis of K-RAS point mutations improved sensitivity from 44% to 82% . Another similar study examined the utility of immunohistochemistry for TP53 expression and found that the sensitivity of EUS-FNA was improved from 76% to 90% .
Unfortunately, K-RAS mutations are not specific to pancreatic adenocarcinoma and also have been detected in 25% of pancreatitis samples, nonneoplastic exocrine pancreatic lesions of smokers, and pancreatic intraepithelial neoplasia (PanIN) [122-128], which proves the need for discovering of a broad panel of mutations that would most likely increase diagnostic specificity.
Some studies have shown an association between patient survival and somatic mutations in CDKN2A (p16), TP53, SMAD4 (DPC4), or germline mutations in XRCC2, XRCC3, RecQ1, Rad54L, ATM, and POLB [129-136]. On the other hand, gene copy number of the epidermal growth factor receptor (EGFR) did not have prognostic value in pancreatic adenocarcinoma .
SNPs have been studied regarding the pharmacodynamics and pharmacokinetics of gemcitabine (GEM) in the treatment of lung and other cancers , but the knowledge is valuable for pancreatic adenocarcinoma as well. Variants in the promoter region of ENT1, the transporter that brings GEM into the cells influence gene expression and probably GEM chemosensitivity . A haplotype of cytidine deaminase (CDA), responsible for the detoxification of GEM was found to lead to decreased clearance and a high incidence of neutropenia, while SNPs in the 5′ regulatory region of the deoxycytidine kinase gene were found to predict GEM sensitivity .
The efficacy and toxicity of other drugs used in treating pancreatic adenocarcinoma like 5-FU and platinum also are affected by polymorphisms. Mutations in dihydropyrimidine dehydrogenase (DPYD) gene, and thimidylate synthetase promoter region, involved in the 5- FU catabolism and pharmaceutical effect respectively, may affect drug toxicity and patient survival [141,142]. Moreover, platinum therapy results on survival were found to be affected by polymorphisms in DNA repair genes like XRCC1, ERCC1, and ERCC2 [143-145].
Finally, in a preclinical study, sensitivity to cross-linking (mitomycin C, cisplatin, chlorambucil, and melphalan) chemotherapeutic agents was affected by the presence of mutations in BRCA2/Fanconi anemia gene .
K-RAS mutations, which are found in 70-90% of pancreatic adenocarcinoma tissues, have been the target of several therapeutic approaches. K-ras must be farnesylated to be active. Although no successful clinical trials have been reported, inhibitors of the enzyme farnesyl-transferase have been developed [147,148]. An immunotherapy approach to K-RAS mutations has also been proposed. K-RAS mutations currently are being studied as a target for immunotherapy with the use of yeast vectors called Tarmogens (Targeted Molecular Immunogens) in a phase II clinical trial, in post-resection, pancreatic cancer patients . One other key downstream target of the Ras family, the phosphoinositol 3-kinase (PI3K), may play a role in drug resistance. In a preclinical study, treatment with PI3K inhibitors enhanced apoptosis induced by GEM .
Another targeted therapeutic approach of pancreatic adenocarcinoma is the inhibition of EGFR by tyrosine kinase inhibitors like erlotinib. A phase III study for first-line treatment of advanced pancreatic adenocarcinoma showed that the addition of erlotinib to GEM offered some improvement to survival compared to GEM alone and led to FDA approval . Recently, EGFR intron 1 polymorphism was found to influence postoperative patient survival and in vitro erlotinib response .
The potential impact of gene sequencing studies on cancer treatment is enormous. Results from large sequencing projects deposited in databases accessible to the scientific community will serve as a referral for mutations associated with cancer. As the different cancer sequencing projects progress, the importance of mutations in cancer development, progression, and metastasis in unsuspected genes will be uncovered. With these discoveries, the list of genes of interest will expand. It will then be possible to orient such knowledge toward epidemiology and familial genetic studies to determine the importance of these mutations in the propensity to develop cancer. Eventually, screening tests and early detection for high-risk relatives and population will follow [153,154]. Functional studies will evaluate the impact on the biologic function of the identified mutated genes. DNA changes will be evaluated as biologic markers for diagnosis, prognosis (disease-free survival after surgery and overall survival), and therapy (including response to or toxicity from chemotherapy or radiation).
The study was supported by a grant from the Effie and Wofford Cain Foundation. We would like to acknowledge Mrs. Katie Elsbury for editorial support, Mrs. Sally Hodges for her assistance with patient-related issues and all the people at the Human Genome Sequencing Center who made this work possible.
This work was presented at the Molecular Surgeon Symposium on Personalized Genomic Medicine and Surgery at the Baylor College of Medicine, Houston, Texas, USA, on April 12, 2008. The symposium was supported by a grant from the National Institutes of Health (R13 CA132572 to Changyi Chen).