Biomarkers boost R&D

Many of the problems affecting the European pharmaceutical industry's pipelines have been ascribed to a failure of the 'development' phase to keep pace with the output of 'research'.

A huge opportunity to address this problem lies in the use of biomarkers to identify responsive subjects for initial human testing and optimising dosage. Biomarkers are objectively measurable characteristics that are evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacological responses to a therapeutic intervention.

By assisting in the rational discovery of novel targets/pathways, and enabling adverse events to be avoided, they are also anticipated to help reduce the attrition rate in drug development, since only one in 15 drugs entering phase I ever reach the market, with the great majority of compounds failing in phase II.

Efficiency gains are needed, along with better drugs. Why, then, are biomarkers not already embedded in the R&D cycle from the earliest target identification to life cycle management?

Part of this is the fact that the industry's business model is built strongly around blockbusters. Like the music industry, commercial success of the few is used to underwrite development of the many.

This process is often wasteful, needs huge amounts of cash, and does not necessarily generate products of optimal quality.

Also, as the tenability of blockbusters wanes, there remain fears that the reduction in market size that biomarkers would force might not be compensated for by the increased value to the target market - a central element of the value proposition made for biomarker research. However, this need not be the case; rosiglitazone is an example of a drug with an 11 per cent market share, but which could benefit up to 40 per cent of the diabetic population.

Indeed, the primary reason for biomarkers being somewhat overlooked in the R&D cycle is probably down to their validation. At what point does a suspected `association' become strong enough to be scientifically meaningful? Given the potential commitments of time and money this association represents, it is clearly a key issue. As things stand, very few of the biomarkers that have the potential to be used as indicators of toxicity or efficacy in trials have yet been validated to Food and Drug Administration (FDA) standards. This undeveloped regulatory environment is clearly a source of much confusion.

~

The regulatory position
Regulatory agencies have a pivotal role to play in centralising, standardising and disseminating biomarker research.

In the US, the FDA's stated mission to 'protect and advance public health... by helping to speed innovations that make medicines and foods more effective, safer and more affordable' led to the development of its 'Critical Path Initiative' (CPI), with the explicit aim of stimulating innovation in the pharmaceutical industry.

The FDA's intentions were made starker by an announcement from the agency in March 2006, which focused the attention and scope of the CPI unambiguously on two key areas: clinical trial design and biomarkers.

The FDA is keen to reassure that the CPI is alive and well, but acknowledges that much development work and standardisation of the biological, statistical and bioinformatics methods must occur before these techniques can be easily and widely used.

In this, the agency has made clear the need for more sharing of the type of information that many pharmacaeutical companies are already beginning to amass, such as markers of liver, kidney, heart and muscle toxicity identified in animal models.

Resources are finite, however, and the industry's co-operation is essential to keep the CPI on track. More support from governments and healthcare organisations is also needed.

The European Medicines Agency (EMEA), which has a bilateral agreement with the FDA, has also been grappling with these points; it is considering the issue of regulatory submission of data relating to DNA variations prior to a full validation of the association, and the extent to which associations need to be confirmed in ethnic groups outside a study.

As with the FDA, the limits of acceptability for a genetic association to become `valid' is a fundamental issue.

~

New target potential
Despite all this, biomarkers are in fact already being used by several companies. Bristol-Myers Squibb (BMS) is actively using them in the life cycle management of drugs in its pipeline, providing a good example of three key ways in which biomarkers can benefit medicine R&D: differentiation, therapeutic extension and risk reduction (click on the box).

Johnson & Johnson (J&J), where work is underway to identify markers of CNS disease, is another healthcare company already using biomarkers. The firm has collaborated with Genset and Signalgene to identify markers in the D-amino acid oxidase (DAAO) gene, which J&J is now pursuing as a target for schizophrenia.

In addition, a large-scale genetic study in Alzheimer's disease is being carried out in collaboration with Genaissance, with results now being published. Some 89 candidate genes have been identified, although the company is frank in stating that it is likely to be many years before drugs based on this information come to market.

GlaxoSmithKline has obtained informed consent from all patients in phase I, II and III trials for collection of DNA samples for biomarker analysis since 2003 (and phase IV trials from 2006).

The importance of doing this is huge as the evolving science and regulation of the field are increasingly likely to necessitate retrospective screening, which can save huge amounts of time and even rescue a drug from the scrap heap.

GlaxoSmithKline has practical experience to demonstrate the impact of biomarkers; in the case of Tykerb (lapatinib), to predict the occurrence of side effects in patients. Around 15 per cent of 107 patients treated in phase I/IIa trials of lapatinib experienced diarrhoea and mild rash, leading to three withdrawals.

In preclinical studies, however, a strong correlation had been established between variations in the metabolising enzyme CYP2C19 and instances of rash or diarrhoea. Notably, of the three individuals who had withdrawn from trials, all were homozygous for CYP2C19*2.

Side effects of the drug are, therefore, predictable, and the dose can be adjusted, as required, in susceptible patients.

~

Toxicity testing
Toxicity is another area where biomarkers will impact on drug development. Abbott, for instance, has been keen to reach a situation where initial safety testing of lead compounds can be carried out with less than 10g of drug over 1 to 3 days in rats.

This would be impossible with traditional methods, but in collaboration with Rosetta Inpharmatics, a US-based wholly-owned subsidiary of Merck & Co, Abbott has compiled a liver toxicity gene expression database using 52 known hepatotoxicants and 10 non-hepatotoxicants, which has been refined to 40 markers.

Prediction accuracy for the liver toxicity of select kinase inhibitors has proved to be 88.5 per cent.

Abbott has also explored cardiotoxicity in collaboration with California-based Iconix, using its DrugMatrix database.

For its part, Rosetta has stated that molecular profiling technology is ready and that predicting compound toxicity using transcriptional profiling at early time points is now routinely feasible.

Pfizer has also been using biomarkers to optimise the dose of irinotecan given to cancer patients. Like GSK, the firm now routinely obtains informed consent from all patients and volunteers participating in clinical trials to take DNA samples for retrospective analysis.

Ethics and the future
Despite the increasing popularity of using biomarkers, however, practical and ethical issues still need to be resolved.

For instance, what if biomarkers are used to deny treatment to all patients lacking specific genotypes? This might be an acceptable situation for less serious conditions - not so with terminal cancer, for example, where all other treatment options had been exhausted and no others existed - where there was a small chance in some patients of a beneficial response in spite of the biomarker evidence.

Moreover, what might the impact be on an individual's insurability if they carried a disease susceptibility gene? Many of the ethical considerations are covered in a Nuffield Foundation report available free on the web (www.nuffieldfoundation.org).

In any case, the cost of genotyping still needs to fall, and management of huge amounts of data that biomarker research generates is an ongoing concern.

In general, big pharma has not been proactively developing targeted therapies, with the big exception of oncology.

Roche's Herceptin - which benefits only those 20 per cent of patients with Her2-positive breast cancers - in fact still serves as the best example of how biomarkers can benefit the whole drug life cycle, right from the earliest stages of development.

It was estimated that from the eight-year acceleration of Herceptin's approval, Roche saved $35m in clinical trial costs and generated $2.5bn of income, with 120,000 patients gaining access to the drug before they might otherwise have done.

The figures speak for themselves.

The author
Dr Peter Robins is editorial and content manager for drug information at Thomson Scientific. This article is based on data taken from the Thomson Pharma database (www.thomsonpharma.com) and includes elements from a Biomarker World Congress `06 report, authored by Dr David Paterson