Biomarkers: yesterday’s tomorrow today

The whole rationale behind biomarkers is to be able to follow the common mantra of providing ‘the right medicine, at the right dose, to the right patient at the right time’ by being able to use the biomarkers to predict a patient's optimal path for treatment when presented with a particular disease.

One of the critical issues in drug discovery today is to identify a small molecule or an antibody (or other protein based reagent) as a therapeutic to treat a disease. In the early 1990s we saw the advent of large combinatorial libraries of small molecules together with the increased necessity for robotics platforms to run screens with these reagents. The state of the art molecular biology at that time allowed over-expression and purification of the target of interest (e.g. a specific kinase) or the development of cell lines that specifically over-expressed a membrane bound receptor (a GPCR for example).

These high throughput screens produced a number of hits and these hits were optimized and progressed through tox studies to pre-clinical and clinical testing only to find out that the efficacy of the therapeutic in a broad patient population was much lower than anticipated.

The reason for this revolved primarily around variations in the genetic make up of the patients within the population being treated. Such variations can be at the level of SNPs (single nucleotide polymorphisms) which alter the responsiveness of the modified target to the drug or can be at a larger scale where the expression levels of single or multiple genes involved in response pathways or metabolism of the drug are altered.

The naivety of the pharmaceutical industry was based on the notion that a single drug could cure all patients with a particular disease. Often a drug with pronounced efficacy in a particular disease (e.g. breast cancer) would be evaluated for efficacy in additional cancer types.

Clinical trials are very expensive to run and the costs for obtaining approvals for these therapeutics in the new indications could be very high (estimates of $500-800 million have been made for a drug to make it to market). Consequently, as the industry sought to expand markets by moving into new indications it became necessary to reduce the number of patients enrolled in a trial while maximising the chance of success with the drug of interest. It was an excellent business model for the industry if all patients with a particular form of cancer (for example) were prescribed a particular drug even though the success rate for the drug was limited. Consequently, the aim of biomarker discovery is two-fold: 1) to allow pre-selection of patients for clinical trials (hence improve the chances of success for the trial and diminish the costs associated with gaining approval for the new indication). 2) To utilise biomarkers when the product hits the clinic to ensure that it is being administered to those patients most likely to respond and also to possibly identify adjunct therapies that can increase the responsiveness of the patient to a therapeutic regimen.

An example of this kind of predictive efficacy of a therapeutic is with herceptin. Patients with breast cancer that also over-express HER2 show remarkable responses to treatment with herceptin that selectively targets this protein and, consequently, this test is administered before deciding to place a patient on a herceptin regimen. Even with single drugs being administered to a single patient, issues such as the dose and timing of administration were difficult parameters to determine. Many candidate therapeutics have been identified by the industry for individual molecular targets with high specificity and potency through the HTS screens run in pharma, biotechs and academic labs and there is a long history of these agents that failed in trials. In many cases the mechanism of disease is not simply due to over-expression of a single target, hence the best inhibitor against this target will never act as a curative agent where there is a need to inhibit both this target as well as another (or many others). As our knowledge of disease biology and the roles that genetics and gene expression play in the etiology and progression of disease improve, it may be possible to identify the roles of multiple targets in the disease mechanism. Then we could prescribe on an individualised basis, the correct cocktail of therapeutics for optimal efficacy in an individual. As we progress in our understanding of the processes we will see cocktails of two drugs being administered and, at some time in the future, cocktails of three or more components may be administered and this cocktail may be altered temporally based on the response of the patient to the treatment. These cocktails could be in the form of mixtures of small molecules, antibodies and other agents, however, in recent years we have seen a remarkable increase in our knowledge of the mechanism of a powerful technology - that of RNAi (RNA inhibition). Large libraries of synthetic molecules (siRNAs) able to silence individual genes have been developed and these resources now allow the examination of the roles individual genes play within a disease phenotype. There is a lot of hope that in the future siRNAs may have therapeutic potential. Key issues around their use as therapeutics will revolve around delivery to the tissues where the diseased cells exist and the ability to gain entry of these molecules into the diseased cells. An advantage of these reagents is that they are all structurally very similar and consequently, they may be able to be administered with a single formulation that targets many species of siRNA to the same tissues. With small molecules and antibodies, the variation in hydrophobicity and charge on each molecule being evaluated results in a significant amount of development time being devoted to formulations, which are highly individualised for each molecular type. While we wait for companies to demonstrate the applicability of siRNAs as a therapeutic modality, our group is focused on using these libraries of siRNAs to identify novel targets for therapeutic intervention as well as identifying predictive biomarkers for patient selection. We are able to examine the role that each gene plays in the growth and proliferation of a ‘diseased’ versus ‘normal’ cell by exposure of these cells to each siRNA in high throughput. This mechanism uses much of the same processes and automation that was used for small molecule discovery. Careful attention to optimization of transfection parameters and exposure time as well as cell culture conditions allows the identification of genes which, when silenced by the individual siRNA increases the vulnerability of that cell. In oncology, the aim is to kill the cancer cell while having little or no effect on an equivalent ‘non-cancer’ cell. Using cell viability as an endpoint together with exposure to a library of siRNAs against the ‘druggable genome' we have identified genes essential to growth and proliferation of cancer cells. The proteins encoded by these genes may be good targets for therapeutic intervention in their own right. In addition, we can examine the sensitivity of the cells upon exposure to the siRNAs in combination with exposure to a therapeutic in late stage clinical trials. We can identify 3 types of genes from these studies: 1) A sensitizer gene that when silenced in the cancer cell preferentially makes this cell type more sensitive to the drug of interest. A specific inhibitor against this target protein may increase the efficacy of the candidate drug and/or diminish the concentration required for a therapeutic effect (an important issue since many drugs are toxic to non-cancer cells at the concentrations needed for a therapeutic benefit). 2) We can also identify genes which, when silenced, reduce the sensitivity of the cancer cells to the therapeutic being tested. These genes may have the potential to act as biomarkers during selection of patients in a clinical trial - those over-expressing these genes relative to the whole population may show diminished responsiveness to the drug being tested and would be excluded from the trial. 3) The majority of the silenced genes do not have any effect on sensitivity of the cells being tested. This paradigm may short-circuit the ability to migrate a drug from one indication to another (e.g. moving from lung cancer to colon cancer, the gene expression profiles between the two cell types will differ and alter their relative sensitivities to the drug).

Running an HTRNAi screen to examine the roles of these genes in both cell types will identify common targets and cell-specific targets that would benefit from adjunct therapy. In addition, biomarkers will differ between the two cell types.

With the sequencing of the human genome and the advent of new technologies allowing multiplexed gene expression monitoring (e.g. microarrays) we have been given tools to understand the roles and interdependencies of genes within the cell. Tissue samples from multiple patients can be screened to determine the relative expression levels of each of the genes within the genome expressed in that tissue. SNP analysis across patient populations is allowing the identification of the distribution of the variations in genetic makeup within a disease. These observations are gradually filtering down into the way that screens are run and hits are progressed. The final piece of the puzzle is then to match data obtained form the screens with the patients that receive the treatments (a pharmacogenomics approach).

The true power of biomarkers will be in their ability to identify patients suited to particular therapeutic regimens and, additionally, to determine that patients on these regimens are showing responses indicative of therapeutic benefit. Biomarkers may be different for each of these points and their method of detection may be genomic (e.g. array-based or multiplexed PCR based technologies). The expression patterns may not measure the responses of the genes themselves but may use surrogate endpoints – e.g. expression patterns in peripheral blood may be able to be used to identify patients and examine the course of treatment. In addition to the genomic approaches there will also be a role for proteomic approaches. Mass spectrometry is probably one of the more sensitive technologies available and being able to examine alterations in profiles of proteins and metabolites may serve as a valuable alternative or adjunct to the genomic approaches discussed above.

Hopefully there will be increased adoption of these technologies and applications to help improve therapeutic outcomes for patients with life-threatening illnesses such as cancer and, over time the sensitivity of these approaches will allow earlier diagnoses and intervention in these diseases. Eventually it may be possible for a routine test to be performed at your doctor’s office at each regular visit that, instead of measuring just vital signs, also provides an indication of health and potential impacts to health before they become major issues.

Intervention earlier in disease will help save more lives, diminish costs to the healthcare system and to patients, reduce the impact on an overwhelmed hospital system and benefit those companies who have embraced the combination between therapeutic delivery and the diagnostic and predictive potential of biomarkers.