SNPs as predictive markers for cancer treatment

ED04-02

A major goal of pharmacogenetics and genomics in oncology is to provide tools to individualize anticancer therapies by maximizing efficacy and minimizing toxicity for each patient. The rapid technological developments in molecular pharmacology and genomics is allowing powerful new approaches to these ends, such that an individual’s germline and tumor DNA sequences will become an integral determinant of drug therapy for cancer. Patients treated with the same drug exhibit wide variations in both response and in the incidence and severity of side effects, likely due in part to individual genetic variation. Given the narrow therapeutic index for most chemotherapy agents, identifying genetic variants that contribute to susceptibility to cytotoxic effects will provide important tools by which to tailor therapies and decrease adverse events. Furthermore, identifying genetic aspects of tumors within specific histological subtypes that confer chemotherapy sensitivity or resistance to particular agents will help maximize efficacy, though this is challenging given the degree of genomic heterogeneity in tumors. Thus, a comprehensive analysis of DNA sequence polymorphisms and copy number variation in individual human genomes and somatic tumors has the potential to optimize and target anticancer therapeutics while minimizing host toxicities. Although the practical application of such personalized genomic strategies remains in the future for most common malignancies, a number of experimental approaches are being employed to identify genetic variants contributing to cancer therapy outcomes.
 Inherited functional single-nucleotide polymorphisms (SNPs) can affect both the activity and expression of host genes that direct chemotherapy drug distribution, metabolism and drug-target interactions. Somatically alteration of SNPs due to loss-of-heterozygosity and other events associated with genomic instability can further contribute to drug availability, efflux and activity in tumors. Most work in cancer pharmacogenomics has been to explore genetic variation in candidate genes or gene pathways thought to be critical for these physiologic processes. Genes are selected that are known or thought to affect the pharmacokinetic and pharmacodynamic pathways of a particular drug, and putative functional SNPs genotyped, often in a case-control series of patients with differing responses or toxicities to a particular drug. This approach has yielded important and clinically relevant insights into candidate gene variation, particularly in those involved in drug metabolism, transport and DNA repair. Examples include allelic variants of thiopurine methyltransferase with decreased enzymatic activity that affect 6-MP toxicity in ALL, polymorphic promoter alterations that affect expression levels of UDP-glucuronosyltransferase 1A1 and alter irinotecan metabolism and subsequent neutropenia, and SNPs in the DNA repair genes ERCC1 and XRCC1, and glutathione-S-transferase metabolism genes and cisplatin efficacy in solid tumors.
 A potential limitation to the candidate gene approach is that it may not detect the contribution of other genes, or discover genes not known to affect the drug pathway of interest. The tremendous advances in genomic knowledge emerging from the Human Genome Project and the International HapMap project now allow investigators to interrogate large sets of SNPs across the entire genome in samples differing in a particular phenotype. This genome-wide association study (GWAS) approach maybe used to identify new loci associated with variability in drug responses, though the particular SNP may not be functionally involved, but rather in linkage-disequilibrium with a nearby gene that is causative. Studying the frequencies of many hundreds of thousands of SNPs in patient samples in an unbiased approach (that is, without focusing on a particular gene or pathway) results in a large number of false-positive associations, thus requiring rigorous validation efforts. However, the fact that SNPs do not segregate independently, but rather may be grouped into haplotypes, necessitates fewer informative SNPs for broad genome coverage, and current and next generation SNP genotyping platforms provide extensive genome wide coverage. GWAS were initially performed to identify disease susceptibility alleles, but can also be exploited to identify genetic variants associated with chemotherapy sensitivity or even host toxicity. For example, lymphocyte cell lines from HapMap individuals from different ethnic populations and other collections have been highly genotyped, many for more than 6 million SNPs each. In vitro assessment of cell line drug sensitivity in these hundreds of cell lines has been performed and a number of genomic loci have been identified that are associated with response to cisplatin, etoposide, 5-FU, docetaxel and others. Further studies will be necessary to identify the genes linked to these SNPs and validate their role in drug responses and toxicities. GWAS should also be valuable to analyze germline DNA from large collections of patients that have been carefully phenotyped for various drug adverse events to identify predictive SNPs.
 Many challenges remain to translate results from both candidate and genome-wide approaches to identify SNPs predictive of chemotherapy drug efficacy and toxicities to a clinically relevant setting. Significant work remains to validate and incorporate these approaches into ongoing trials of novel drugs and multiagent chemotherapy regimens, and to simultaneously interrogate both host and tumor genomes for variation. The role of copy-number variation in predicting drug responses may equal or exceed that of SNPs and remains to be explored in depth. There is no doubt that the tools emerging from the genomic revolution will soon allow for germline genetic signatures predicting susceptibility to the toxic effects of multiple chemotherapy agents and perhaps efficacy to be defined in individual patients, and will contribute to the goal of personalized cancer therapeutics.
 References
 1) Marsh S, McLeod HL. Pharmacogenomics: From bedside to clinical practice. Human Molecular Genetics 2006;15:R89-R93.
 2) Hartford CM, Dolon ME. Identifying genetic variants that contribute to chemotherapy-induced cytotoxicity. Pharmacogenomics 2007;8:1159-68.
 3) Giacomini KM, Brett CM, Altman RB et al. The pharmacogenetics research network: From SNP discovery to clinical drug response. Clinical Pharmacology and Therapeutics 2007;81:328-45.