Personalised medicine – often, and more precisely, called stratified medicine – is defined as the tailoring of medical treatment to a group of patients based on data from genetic or other biomarkers. It is growing in utility and popularity and has potential applications in many therapeutic areas. Almost certainly, however, it is in cancer treatment where the personalised approach is currently most widely used.

The well-established and successful drugs imatinib (Glivec®) and trastuzumab (Herceptin®) were developed to treat subsets of patients, those carrying the Philadelphia chromosome translocation and those over-expressing the HER2 receptor respectively. It is therefore hardly surprising that oncology drugs and their development featured extensively in a recent, prestigious meeting in this area, Personalised Medicine and Diagnostics Europe, held in London on March 9 and 10.

Personalised Medicine and Diagnostics Europe was organised by the UK division of pharmaceutical events and information company eyeforpharma. During the two days, a total of 19 speakers drawn from industry, clinical research and regulatory affairs discussed many aspects of the development, registration and use of drugs for stratified patient populations and their associated diagnostic tests.

It is self-evident that even when personalised medicines are available, they cannot be delivered to "the right patient at the right time" without access to sophisticated diagnostic tools; the often-used term "companion diagnostic" refers to any tool or test that is used alongside a drug to determine whether a patient is likely to benefit from a drug. Although many types of diagnostic procedure are available, it is molecular diagnostics, particularly for cancer, that are most often thought of alongside personalised medicine. David Gonzales de Castro, head of molecular diagnostics at the Royal Marsden Hospital and the Institute of Cancer Research, London, UK, gave a presentation describing the translation of molecular diagnostics research into clinical use in oncology.

The kinase inhibitors imatinib and gefitinib (Iressa®) are classic examples of cancer drugs that only work in patients with specific mutations. Mutations in the EGFR gene were first associated with gefitinib response to non small cell lung cancer (NSCLC) in 2004, and five years later this drug became the first-line treatment for this tumour only in patients with EGFR mutations. Responses even differ between patients carrying different mutations in this gene.

Gonzales de Castro went on to discuss the mechanisms and challenges of implementing molecular diagnostics in the clinic. All developed countries are establishing standards and accreditation bodies for molecular diagnostic tests; in the UK, this is Clinical Pathology Accreditation (UK) Ltd. Specialist diagnosis, particularly for rarer cancers with few patients diagnosed annually, is best performed in a network centred around a single specialist laboratory for each cancer type. There are still challenges in implementing this in routine clinical practice, particularly revolving around test sensitivity and turnaround times.

Hans Winkler of Janssen Oncology R&D in Beerse, Belgium (now a subsidiary of Johnson & Johnson) gave a developer's perspective on the use of biomarkers in decision making for drug development. The goal of patient stratification is to identify the segment of the patient population – maybe as low as 5-10% - in which a drug has the highest probability of success.

The "pharmacological audit trail", a series of questions asked in pre-clinical and early clinical development that determine whether a compound is taken further, can be expanded to include questions about the availability of predictive biomarkers. As patients differ in metabolic efficiency as well as in drug response, biomarkers can be used to extend the dose range for Phase II trials beyond a single optimum dose level calculated from Phase I results.

It is essential to develop and validate a biomarker hypothesis and associated assay as early as possible during a drug discovery programme, at the very latest at the start of Phase II trials. The most appropriate biomarkers will have robust predictive power, ideally with a predictive accuracy of over 70%: sadly, this is not yet the case even with widely used tests for HER2 receptor expression such as the HercepTest™. They should also be easily accessible as sample accessibility and quality are critical. Circulating biomarkers that can be measured precisely from simple blood samples are ideal.

Changes in DNA expression patterns resulting from reversible chemical changes to the DNA – epigenetic changes that do not affect the DNA base sequence – are often drivers of cancer development; increasing DNA methylation is the most efficient way to shut down a gene. James Clark, vice-president for research and development at MDxHealth, based in Durham, North Carolina, USA, described how DNA methylation patterns can be used as biomarkers for predicting cancer drug activity. One well-described example of this concerns temozolomide, a glioblastoma drug that has achieved blockbuster status. Patients with tumours in which the enzyme O6-methyl guanine methyltransferase (MGMT) has been deactivated by DNA methylation respond much better to this drug, with median overall survival improved by about nine months over similar patients without this methylation. It is now straightforward to detect DNA methylation patterns using second-generation sequencing techniques. MDxHealth is building a workflow to sequence 1000 complete "reference epigenomes", which will be used to discover markers for cancers and other diseases, and modifying its technology so it can be used to test circulating DNA in readily accessible body fluids.

Karol Sikora, medical director of the privately-run network of cancer treatment centres, CancerPartnersUK, gave a sobering account of conflicts and challenges in prescribing cancer drugs. The development of targeted therapies – often very expensive drugs that are only effective in or available for relatively small subsets of patients – is one of the most important reasons why he believes that the cancer market is heading for "financial meltdown". The pipeline of candidate drugs for cancer is full, but many of these will not make it through Phase III clinical trials, and cost-based rationing is bound to prevent patients from accessing many of those that will be registered. This is already starkly true in the UK, where drugs that are quite readily available in continental Europe are often rejected by NICE on cost-effectiveness grounds. Conversely, some drugs are being licensed for use in indications where survival benefit is measured in weeks, which can raise patients' and carers' expectations unrealistically.

Sikora presented several scenarios for cancer diagnosis and treatment in the next 10-20 years. Despite the gloomy economic outlook, these were fairly optimistic. Molecular diagnostics dominated them all, but to differing extents. By 2020, it is possible that biomarkers will be used to identify healthy people at risk of cancer for chemoprevention programmes, and to adjust doses based on predictions of toxicity. By 2050, it is possible that 95% of cancers will be detected early enough to be controllable, but costs will ensure that rationing – however unpalatable – will have to be introduced, probably based on age.

The final talk was given by Ansar Jawaid, Global Diagnostics Brand Manager at pharmaceutical giant AstraZeneca, developers of gefitinib. He illustrated the challenges of delivering a personalised therapy worldwide using this drug as an example. Iressa® was first licensed for the treatment of non-small cell lung cancer in Japan in 2002, and it is now available in over 64 countries. Jawaid explained that the barriers to the launch and eventual delivery of the drug to all patients who could benefit were different in each country. Although it was first associated with EGFR mutations in 2004, it was only the results of the large IPASS trial, published in 2009, that showed the strength of the association. This led to the drug's approval in many more countries alongside a diagnostic test. The company has had to understand and work with a different medical culture and infrastructure in each country in order to maximise test take-up and make sure that its drug can reach "the right patients, at the right dose, at the right time": perhaps a paradigm for the introduction of many more personalised oncology therapies in the years to come.