Rx-Dx: the way ahead for drugs

As Sarah Houlton explains, the use of personal diagnostics to ensure the patient is compatible with the drugs they are taking is set to be the next 'blockbuster'.

Many medicines are likely to be tied in with diagnostic tests in future, thanks to developments in pharmacogenetics.These Pharmaco-Diagnostics (Rx-Dx) linkages provided one of the major themes of this year's BioFinland 05 congress in Helsinki. This first step along the road to personalised medicines is certain to become an important strategy for pharma companies in the next few years.

The pharma industry has been under fire in the past few months over the safety of its products, so any way in which the risks could be reduced will be welcomed by the regulators. Many drug side-effects result from the fact that different people react to or metabolise medicines in different ways.

predictive tests

If it were possible to screen patients' genetic profiles beforehand to establish whether they would be good responders or liable to develop severe side-effects, then that would instantly make many drug problems a thing of the past.

The recent COX-2 inhibitor saga has highlighted the precarious nature of the business, as all it takes is a handful of safety concerns over a big selling product for company profits to take a dive. The problems with the COX-2 inhibitors came to light only once the products had been in use in huge numbers of patients for some time; what would have happened if some form of predictive test had been possible?

This is already becoming a reality: patients with acute myeloid leukaemia are already screened to establish whether they are genetically likely to respond to Novartis's Glivec (imatinib), before treatment begins, and we are likely to see a huge increase in this type of screen in the future (other examples are given in Table 1).

'For the first time in history, health and disease can be defined by molecular "fingerprints", which has important implications for diagnosis and treatment,' Robert Pietrusko, senior vice president, worldwide regulatory affairs, at Millennium Pharmaceuticals told the conference. 'Ultimately, individuals will receive medicine tailored to their molecular profile; most early advances will be in oncology, but other areas such as cardiovascular, diabetes, Alzheimer's disease and rheumatoid arthritis will also be affected.'

effective medicines

It's not just once a drug has reached the market that pharmacogenomics will provide a contribution, however. As Ruth March, senior principle scientist, r&d genetics at AstraZeneca told the BioFinland conference, a perfect medicine is effective in all patients at the same dose, with no side-effects, and costs nothing. However, real life is not like that. 'Real medicines are effective only in some people,' she said. 'The dose varies for different patients, and may have side-effects in some of them.'

She said that all patients can, effectively, be classified in one of three groups for a particular medicine: safe responders, poor responders, and those who are likely to experience adverse events. 'If we can screen out those who are poor responders or who will be adversely affected, this would leave only those who show good safety and efficacy.' If only those patients in whom a product is safe and effective are given it, then the likelihood of a company being hit by a catastrophic withdrawal would be very slight.

genotype correlation

Pharmacogenomics can be used to explore pharmacokinetics, and also to look for those essential biomarkers that would be needed if a diagnostic were to be developed.

There are many applications of pharmacogenetics in pharmacokinetic (PK) studies. Prospective PK studies can be used to investigate poor and extensive metabolisers of a drug. Retrospective analysis can be carried out on subjects in a PK trial in order to explain observed variations. Subjects in trials can be genotyped in different regions to see how response varies, and those who are known to be poor metabolisers can be excluded from certain studies to aid data interpretation.

March cited the example of AZ's proton pump inhibitor omeprazole. Because of differences in cytochromes P450 profiles, some patients metabolise the drug in different ways, leading to wide ranges of plasma concentrations of the drug in patients who have been given the same dose, depending on how well they metabolise it. One study suggested that this might have a clinical effect, although it had a very small sample size.

'The Japanese regulatory authority asked for a correlation of genotype with reponse,' March said. 'AZ's larger dataset of information from more patients showed that genotype actually had no effect, so it is safe in all groups.' So, although some patients were poor metabolisers of the drug, in this instance it had no effect on safety.

genetic factors

But will pharmacogenomics be able to deliver smaller and faster clinical trials? 'I believe they will only be smaller and faster if you have a validated biomarker to select patients prospectively for Phase III trials,' March told the conference; and therein lies the problem. Pharmacogenetic biomarkers will only rarely be validated by the end of Phase II trials because the numbers of subjects involved just aren't big enough.

Another example where pharmacogenetics may prove to be the saviour of a drug is AZ's Iressa (geftinib) for the treatment of non-small cell lung cancer. Trials initially showed that its efficacy was debatable, but a trial published in 2004 suggested that those patients whose tumours shrank by more than 50% had the same mutation in the kinase domain of the EGFR gene.

However, the picture isn't simple as that: around a fifth of responders had no detectable mutations, and there are several other genetic factors that may be having some effect.

'Screening for mutants takes several weeks, and interdependence is complicated,' March said. 'Interpretation of the data can be complex, not least because the mutations may be lost in metastasis, so tumour analysis may give an incomplete picture.'

However, in early development, the numbers of subjects in trials is usually low. This makes it difficult to pinpoint potential biomarkers in an accurate manner: the power to detect them will be low, and the false positive rate will be high, but data from early clinical results may still provide a useful starting point for tracking down relevant genes.

Steven Burrill of Burrill & Company believes that diagnostics will play a much bigger role in the pharma industry in future. 'They will be at the centre of the new world, and we will see incredible value attached to Dx companies,' he told the conference.

'We're heading from Genes R Us to Dx R Us. But the key is to build sustainability, rather than just jumping on the bandwagon. In the short term, drugs with an Rx-Dx linkage such as Glivec will have shorter approval times, and those that don't will take longer. Glivec was the fastest ever FDA approval, and speed is likely to be replicated for drugs with such a clear genetic basis.'

cutting bills

We are likely to see a move towards genotyping for health risk analysis - being able to predict that a particular person is likely to develop, say, Alzheimer's disease or rheumatoid arthritis, and then take steps to prevent it occurring. 'If you are predictive, you will then be preventative,' Burrill said.

Diagnostics also offer the potential to cut drugs bills: 'Lipitor is a US$10bn drug, but it doesn't work in half of patients,' Burrill said. 'If you can eliminate that $5bn of waste a year, that's a lot. So if there is a diagnostic that will eliminate the half of patients for whom it won't work, those who pay the drugs bill will obviously say "yes". Payers have economic incentives.'

He also believes that such 'theranostics' will be easier to develop in countries - like Finland - where there is a relatively homogeneous gene pool, unlike the US. As Leena Peltonen Palotie of the University of Helsinki told the conference, in a country like Finland there is a higher general degree of homogeneity among the population. 'It is easier to spot disease mutations,' she said. 'Those commonly found in the Finnish include multiple sclerosis, combined hyperlipidaemia, asthma and lactose intolerance. But a large biobank can be a trash bank without detailed clinical and epidemiological data. DNA is worth nothing alone, and most biobanks won't be useful for 10 or 20 years.'

Wherever in the world the biomarkers are discovered, they should be transferable elsewhere, importantly the US with its huge, diverse population and vast healthcare spend. This provides a huge opportunity for smaller biotech companies to create value. 'When I talk to my friends in Big Pharma, they say they will work on both Rx and Dx in future,' Burrill claimed. 'I don't share that view.

'They are much more likely to license the technology from specialist small companies. I think we will see a revolution in the Rx-Dx side, and should not assume that, as far as pharma companies are concerned, the dinosaurs we see today will have to undergo dramatic change to survive.'

So is this the end for the blockbuster model of medicines? Does it mean that the future pharma market will consist of more products with a smaller market? 'Blockbusters are not over,' Burrill claims. 'They are just going to change.'

And diagnostics are set to become the biggest new growth area in the pharmaceuticals industry.