Steve Connor: Most drugs don't work on most people – for now

It is an open secret within the pharmaceuticals industry that most of the drugs it sells don't work on most of the people who end up taking them. Patients with chronic illnesses have an instinctive feel for this "trial and error" approach when they go to the doctor and find that they have to try two or more medicines before receiving the one that actually works.

It is estimated for instance that drugs for Alzheimer's disease work in fewer than one in three patients and those for cancer are only truly effective in about a quarter of patients. Drugs for migraines, osteoporosis and arthritis work in about half the patients, and most drugs work in fewer than one in two patients, mainly because of inherited differences in the DNA code of individuals.

If it were possible to devise simple blood tests that could determine which DNA mutations someone was carrying, and whether these mutations have any effect on the effectiveness of drugs, it would open the door to a future where patients would only be given the drugs that actually work – avoiding unnecessary side-effects and the expense of prescribing useless medicines.

The study by Charles Swanton and his colleagues into breast cancer patients who are likely to respond to paclitaxel is one example of the wider attempts to bring about this era of "personalised medicine". In this case, however, the genetic differences between patients are not the result of inherited variations in their DNA, but due to variations in the DNA of their tumour cells, the result of a tumour's growth and development.

Nevertheless, a genetic test that can distinguish between these differences in tumour DNA is a critical development along the path to a more personalised approach to cancer treatment, an area of medicine that has long recognised that patients with certain cancers may need to be treated differently to others with the same disease.

Personalised medicine could in the long run save money for the NHS because it would mean that expensive drugs are not taken by people who receive no benefit. More importantly, "non-responders" to a drug could be spared the unnecessary suffering of enduring the side-effects of a medicine they should never have been given.

The application of human genetics to the development of drugs, called pharmacogenetics, is a hugely important area of research and one that could transform medical care. However, it is still in its early stages of development and there is of course the concern that many drugs companies will see it as a threat to their income stream: it could cut the size of the customer base for many of their drugs.

However, the days when it is commercially justifiable to sell as many drugs as possible to the greatest number of people may be coming to an end. In future, personalised medicine and pharmacogenetics may mean that we are only given the drugs that we need and, more importantly, that will work for our genetic make-up.