Cancer Genomics: Enabling patient-rather than disease-driven research

---

By Olena Morozova and Marco Marra, PhD, FRSC

It is widely recognized that cancer patients with the same diagnosis often respond differently to the same treatment. While this observation has long been attributed to the individual differences among the patients, genomics tools for the first time shed light on what these differences entail.

Cancer has genetic roots

The association between abnormal genetic make-up of cells and cancer was first suggested in the early 1900s by Theodor Boveri1. Boveri observed that abnormal numbers of chromosomes led to improper development of sea urchin embryos. Based on this finding, he suggested that abnormal genetic composition might account for improper growth of cells seen in cancer.

Taking this idea even further, Boveri hypothesized that abnormal chromosomes contributing to tumour development came in two flavours: those stimulatory, and those inhibitory to cell growth. The growth stimulatory chromosomes were accumulated by the developing tumour, while the inhibitory ones were excluded1. Boveri’s prediction of stimulatory and inhibitory DNA material was later manifested in the concepts of oncogenes and tumour suppressors, collectively known as cancer genes. Oncogenes were discovered by searching for human homologues of retroviral proteins that control cell growth -- the discovery that earned Michael Bishop and Harold Varmus the 1989’s Nobel Prize in Physiology and Medicine. Tumour suppressors were characterized in the same year as genes that normally counteracted the action of oncogenes.

Cancer genetic complexity as an obstacle to targeted therapies

Decades of research, and several more Nobel Prize-winning discoveries later, the paradigm of cancer being caused by abnormalities in genetic material still holds. Moreover, the discoveries of oncogene cooperation, DNA damage checkpoint deficiencies, and genomic instability2 added to the extent of genetic abnormalities that may be present in cancer cells.

The complexity of genetic composition of cancers makes it difficult to distinguish cancer-causing genetic changes from those contributed by genomic instability, a process that produces mutations in DNA at a higher rate than that in normal cells.

Partly as a result of this issue, targeted cancer therapies that block the activity of specific molecules (e.g., specific oncogenes) exist for only a few types of cancer. In these few cases, the exact molecule causing the disease is known, and it is possible to make a clinical inhibitor against it. A famous example of such a therapy is the drug Imatinib or Gleevec, developed for the treatment of Chronic Myelogenous Leukemia (CML). Gleevec targets the specific oncogene BCR-ABL that is known to drive CML development. The main advantage of targeted therapies over standard chemotherapy is their specificity: targeted therapies go after cancer cells while sparing their normal counterparts.

In contrast to CML, most cancers are treated with chemotherapy, and radiation therapy regiments that are often very similar for different cancer types. The standardized anti-cancer treatments are aimed at killing all rapidly dividing cells with the hope that they would also kill most cancer cells. As a consequence, these therapies are often not discriminatory enough, resulting in the killing of large numbers of normal cells, and high toxicity to the patient. Moreover, many of these therapies are not effective when it comes to indolent or slow-growing tumours.

Cancer and its host: the context matters

The quest of developing targeted therapies for cancer became even more complicated after the discovery of the prevalence of normal human genetic variation, most of which is manifested as Single Nucleotide Polymorphisms (SNPs) and Copy Number Variations (CNVs).

SNPs or single base differences in the DNA sequence among individuals occur every 100-300 bases in a human 3-billion-base genome3. According to a recent study, CNVs are also common and occupy approximately 12% of a human genome sequence4.

The discoveries of SNPs and CNVs in human DNA implied that “the human genome sequence” produced by the Human Genome Project5 is merely the sequence of one or a few human individuals and may not be fully representative of the genomes of all humans. From the prospective of cancer research, this means that it is even more difficult to identify which genetic changes present in cancer cells represent causal mutations contributing to cancer growth, and which are examples of neutral human variation.

The observed human variation provided potential explanations for differences in clinical outcome of cancer patients with the same clinical history and treated with the same drugs, and motivated further research into personalized treatment approaches.

Cancer and the OMICS world

The completion of human and model organism genome sequencing projects motivated the development of a set of new research areas, known as ‘omics disciplines’. Omics disciplines aim to understand the function of DNA sequences, their variations, and products on a genome-wide scale. Genomics is one such discipline that concerns itself with understanding the function of whole genomes. One common way of studying gene function on a large-scale is examining gene expression levels in different experimental conditions, termed ‘gene expression profiling’.

Gene expression profiling studies in cancers showed the presence of several cancer sub-types within what was previously thought to be the same disease. In addition, these techniques were also shown to be useful for discriminating between known cancer types. For instance, in a landmark study introducing gene expression profiling approaches in cancer, Golub and colleagues were able to differentiate between acute lymphoblastic and acute myelogenous leukemia on the basis of gene expression levels6.

Cancer genomics and the prospect of personalized medicine

Genetic differences among individual human genomes as well as variability among tumour types, revealed by expression profiling initiatives, highlighted the need for genome-wide approaches to understanding genomes of individual tumours. Recent sequencing studies comprehensively characterized sequence variation found in coding regions of individual cancer samples. These studies found that ~10% of all human genes could be mutated in cancers. Even more significantly, two individual tumours of the same type often did not share the same mutations. This finding underscored that cancers needed to be examined and potentially treated on individual basis.

The standard Sanger sequencing procedure used in the Human Genome Project did not allow for routine sequencing of individual human genomes due to the labour and cost requirements of such efforts.

However, the recent advent of a panel of new sequencing instruments, collectively known as “next-generation” sequencing technologies, for the first time provided a means for genome sequencing by individual research laboratories.

A testament to the promise of next-generation sequencing in cancer research is the first cancer genome sequencing study published in November 2008 in Nature7. The study by Ley and colleagues reported the full genome sequence of an acute myelogenous leukemia patient sample as well as a comparison of this sequence to that found in normal cells of the same patient. The authors found that this particular cancer sample had acquired mutations in 10 genes, most of which were not previously known to be involved in cancer.

One can envision that future cancer management, enabled by next-generation sequencing technologies, would include genome-wide identification of mutations discovered in a particular cancer case, followed by the selection of therapeutic options targeting these mutations. This patient-driven approach would be a manifestation of the concept of ‘personalized medicine’ that is expected to revolutionize current medical practice. Tailoring treatments to each individual patient based on their genetics is in contrast to the established medical procedures, centered on standardized therapies that often do not work for everyone.

References:
1.     Balmain, A. “Cancer genetics: from Boveri and Mendel to microarrays.” Nature Reviews Cancer. Oct 2001, 1(1): 77-82.

2.     ”Nature Milestones in Cancer.” Nature.com. 19 Jan 2009. <http://www.nature.com/milestones/milecancer/index.html>
3.     ”SNP Fact Sheet.” Genomics.energy.gov. 19 Sept 2008. Human Genome Project Information. 19 Jan 2009. <http://www.ornl.gov/sci/techresources/Human_Genome/faq/snps.shtml>

4.     Redon, Richard et al. “Global variation in copy number in the human genome.” Nature. 23 Nov 2006, 444:444-454.

5.     International Human Genome Sequencing Consortium. “Initial sequencing and analysis of the human genome.” Nature. 15 Feb 2001, 409:860-921.

6.     Golub, T.R. et al. “Molecular classification of cancer: class discovery and class prediction by gene expression monitoring.” Science. 15 Oct 1999, 286(5439):531-537.

7.     Ley, T.J. “DNA sequencing of a cytogenetically normal acute myeloid leukaemia genome.” Nature. 6 Nov 2008, 456(7218):66-72.

Marco Marra is the Director of the Genome Sciences Centre, BC Cancer Agency, 100-570 West 7th Avenue, Vancouver, BC, Canada V5Z 4S6; Olena Morozova is a PhD candidate in the Bioinformatics Graduate Program at the University of British Columbia and is supervised by Marco Marra