Biology is like economics, participants at a European Commission meeting on personalized medicine in Brussels heard last week: they are both complex and neither is properly understood. The view struck a chord with attending scientists and health-care economists, who felt that personalized medicine should be happening, and didn't understand why, mostly, it isn't.

Personalized medicine aims to use the latest genomic knowledge and technologies to tailor treatments to individuals. Pivotal to the field are drugs that have been designed to hit a particular molecular pathway that has gone wrong in a disease. The European Medicines Agency has already approved around 15 such drugs for cancer therapy and is set to approve several more in the next year or so.

The personalized approach faces two major problems: complex biology and complex economics. The pathway involved is often not well understood, and most targeted drugs are so expensive that health-care systems and insurance companies don't want to pay for them, even if they reduce waste and should therefore save on overall treatment costs in the long term. The drug gefitinib, for example, costs around €20,000 (US$28,000) per patient and targets the EGFR pathway, which is disrupted in fewer than 15% of patients with lung cancer. What's more, targeted drugs need to be accompanied by diagnostic tests to identify suitable patients, yet many health-care systems have no mechanism to pay for the tests. The result is an absurd situation in which expensive drugs can be prescribed without testing, and therefore to some patients who will gain no benefit.

As arguments about the value of personalized medicine rage around the world, France has found its own solution.

As arguments about the value of personalized medicine rage around the world, France has found its own solution — at least for cancer, where molecular medicine is most advanced. In 2005, the country said it would pay for the treatment of every citizen shown to be likely to benefit from targeted drugs. Its National Cancer Institute set up 28 platforms for molecular genetics at university hospitals and cancer centres with expertise in both molecular and pathological analysis. Biopsies of cancerous tissue from patients all over France are sent to these platforms for a battery of 20 or so genetic tests. If the tissue displays a genetic signature in any molecular pathway targeted by one of the drugs, the patient gets treated with it. The platforms develop the tests themselves, and are already working on a test to accompany a drug that researchers hope will be approved this year for melanoma. Targeted drugs now account for 57% of France's cancer-treatment budget. The Czech Republic has a similar system.

The model seems to work. The French platforms have so far tested samples from around 15,000 people with lung cancer for alterations in the EGFR pathway. Just over 1,700 patients tested positive and were given gefitinib until they stopped responding (an average of 38 weeks). That has cost France €35 million. Had all 15,000 patients been given an eight-week course of gefitinib just to see whether they would respond, it would have cost the nation another €69 million — with no extra benefit.

Some assessments, however, have concluded that personalized drugs do not offer enough benefit to justify the cost. It will not be easy to persuade the spectrum of state health systems and health-insurance companies that personalized medicine makes economic sense. Understandably, they will want a lot more evidence that it works.

Much reluctance also seems to come from a medical profession unused to needing genetic tests to select patients and from inflexible bureaucratic systems. The European Commission's health directorate could help by encouraging European countries to harmonize their health-technology assessments, or even by issuing its own (non-binding) conclusions on which targeted drugs it considers cost-effective. And the commission's research directorate could provide greater support for efforts to translate the results of preclinical research on molecular pathways into the clinic, which it plans to do in its 2012 call for proposals.

Amid the excitement and attention paid to cancer, it is crucial to remember that other conditions — such as psychiatric disorders — carry just as great a societal burden, yet remain too poorly understood to benefit. The research directorate has enabled a great deal of fundamental research on animal models designed to understand such complex conditions, and it must continue to do so, in parallel with its translational efforts. We are at the beginning of personalized medicine in the clinic. But we are also just starting to understand the mechanisms behind most of the diseases that are likely to gain the most.