Chapter 9.  Translational Research and New Approaches 135

Genomics

Genomics involves the study of organisms’ genes including the determination of the entire DNA sequence. Complete genomes have been developed for many organisms including humans. Genomics can be divided into either structural or functional genomics. Structural genomics involves the characterization of the physical nature of whole genomes, and corresponds to the genetic background of organisms. Using high-throughput technologies that allow for the simultaneous determination of a large number of components from a clini- cal sample, one can evaluate the variation of DNA sequence within a population.This enables the systematic study of dis- ease-correlated genomic variations.

Biological variations can vary from a single base change termed a Single Nucleotide Polymorphism (SNP), to small insertions or deletions of sequences called indels, to largescale chromosomal translocations. In some instances, individual mutations can produce a disease state as is seen in sickle-cell anemia where a single base change in the b chain of hemoglobin (adenine to thymine) causes the replacement of glutamine with valine.Huntington’s disease is caused by an expansion of the polyglutamine repeat (CAG) in the gene that encodes Huntington protein. The Philadelphia chromo- some is the result of a wholesale translocation in chromosomes 9 and 22, resulting in the production of a constitutively active tyrosine kinase that can lead to chronic myelogenous leukemia.

It is more common that a genomic variation is found to correlate with increased susceptibility to a given disease. First a specific genomic variant is identified in an individual with a disease. When the same variant is identified in another indi- vidual without evidence of the disease, the SNPs can be tested for association with susceptibility to a variety of diseases.

When the frequencies of the genotype are compared in populations of cases and controls, a higher frequency in patients with the disease is thought to be sufficient evidence that the genetic variation is associated with increased risk of disease.

136 D.P. Foley

One example is the mutation in the BRCA1 and BRCA2 genes.Women with both mutations have an 80% lifetime risk of breast cancer compared to 12% lifetime risk in the general population. In some instances prophylactic mastectomies are performed in high-risk patients due this correlation between genomic variation and clinical prognosis.

One example demonstrating the rapid translation of pre- clinical molecular findings in genomic studies into the clinic has been seen in ALK (anaplastic lymphoma kinase) gene inhibition in non-small cell lung cancer (NSCLC). ALK encodes a tyrosine kinase normally expressed only in neuronal cells. In a rare subset of anaplastic large cell lymphomas, inter- stitial deletion and inversion within chromosome 2p result in fusion of the N-terminal portion of the protein encoded by the echinoderm microtubule-associated protein-like 4 (ELM4) gene with the intracellular portion of the ALK receptor tyrosine kinase. While genetic alterations involving ALK have been seen in other malignancies, thus far, the ELM4-ALK fusion gene appears unique to NSCLC. In less than 3 years after these findings,studies ofALK inhibition yielded dramatic results in patients with NSCLC. In a pretreated population that generally has a 10% response rate to conventional chemo- therapy, treatment with the oral ALK inhibitor, crizotinib, yielded an overall response rate of 55% and an estimated

6-month, progression-free survival rate of 72%.2

Gene-Expression Profiling

Functional genomics involves the overall patterns of gene expression, and the targets of this research are RNA, pro- teins, and metabolites. Functional genomics allows the detec- tion of genes that are turned on and off at a given time based on environmental factors. Transcriptomics consists of the study of all transcribed mRNA species at a given time. Multiplex oligonucleotide or complementary DNA microar- rays are platforms that can be used to determine mRNA abundance of hundreds to thousands of genes simultane- ously. The principles of these technologies involve the

Chapter 9.  Translational Research and New Approaches 137

following.Oligonucleotide or complimentary DNAs (cDNAs) for specific mRNA species are immobilized on a surface (glass slide, or nylon membrane). The target mRNA is iso- lated from the sample of interest, converted to cDNA, labeled, and allowed to hybridize to the oligonucleotides or cDNA fixed to the solid surface. The intensity of hybridiza- tion on each probe is proportional to the gene expression level.

To perform array experiments, one needs the probes to detect the RNA, reproducible and sensitive techniques for quantification of RNA levels, and standardization procedures and databases for analysis. Array methods differ based on which probes are used (cDNA or oligonucleotides), the tech- nology used to fix the probes to the solid surface probes, and the labeling technologies for the mRNA targets. When changes of mRNA abundance are identified and they are small in number, the microarray findings can be validated by using real-time polymerase chain reaction (PCR). PCR is more sensitive and is cheaper than using microarrays when smaller numbers of genes are studied.The two techniques are usually used in parallel fashion to validate the findings.

The most difficult aspect of employing microarray tech- nology is not the acquisition of the data but rather the analy- sis.While a detailed discussion of statistical analytical methods used in genome-wide studies is beyond the limitations of this chapter, a few analytical techniques will be discussed.

Clustering is a data mining technique used to group genes having similar expression patterns. Hierarchical clustering and k-means clustering are widely used techniques in microar- ray analysis.The concept is such that genes of similar function areco-expressedtoproduceaspecificphenotype.Hierarchical clustering is a statistical method for finding relatively homog- enous clusters of similarly expressed genes. This clustering consists of two separate phases.A distance matrix containing all of the pair-wise distances between the genes is calculated. Pearson’s correlation or Spearman’s correlation are often used to estimate dissimilarities among genetic clusters.

Analyses are also termed either as supervised or unsupervised. Unsupervised analyses involve analyzing the

138 D.P. Foley

most differentially expressed genes that are clustered to look for patterns. In supervised analysis, the most commonly used analy- sis, samples are grouped prior to analysis using existing knowl- edge and then they are clustered and analyzed. For example, samples are grouped by good or poor prognosis,and then genes are clustered.This approach identifies genes that are potentially linked to prognosis in a training set.

A commonly used statistical method for the analysis of microarrays is SignificanceAnalysis of Microarrays (SAM) that was adapted byTusher et al.3 SAM identifies genes with statisti- cally significant changes in expression by assimilating a set of gene-specific t-tests. Each gene is given a score on the basis of its change in gene expression relative to the standard deviation of repeated measurements for that gene. Genes with a score greater than a predefined threshold are deemed potentially significant.The percentage of such genes identified by chance is the false discovery rate (FDR). In other words, the FDR is the proportion of genes that were wrongly identified by chance as being significant. It is calculated by dividing the median of the number of falsely called genes by the number of genes called significant. In addition, if a preliminary set of expression data are available,SAM can also estimate the number of microarray chips required to reach a defined level of significance.

There are two major areas where functional genomics can impact medicine. One is by identifying molecular markers or differentially expressed genes that may be important in bio- logical functions as their expression differs based on environ- mental and genetic factors.The other application is the ability to identify a signature profile that can be used as a detailed molecular phenotype. These predictors could be comple- mented in the future with changes in structural genetic varia- tions that are identified at the DNA level. For example, with this technology researchers who study trauma biology and sepsis can develop molecular signatures for inflamed tissues and specific cell populations. This technology is currently being used to characterize the progress of disease in patients with trauma, burns, and sepsis. Applications can also extend to tumor biology and identifying those patients who may respond to treatment better than others. In solid organ