Mining medical records to improve drug safety

Following the 1995 Audit Commission into the standard of hospital records within the NHS, the UK government invested heavily in information technology. This investment has helped change radically the way in which information on routine clinical care is recorded. Now, James Mackay and Ailsa Taylor of The Doctors Organisation for Clinical Studies based in London, say that, despite technical issues with the original IT technology, the system has accumulated a vast amount of potentially invaluable data.

The coupling of this medical data resource with e-Science technologies built with Medical Research Council funding presents a unique opportunity to carry out large-scale pharmacogenetic studies of drug effectiveness and safety for important diseases, such as cancer.

“Many of the ingredients required to start the process of obtaining new knowledge from routinely collected clinical data are already in place,” the researchers explain, “However, the vast majority of clinicians and other health professionals are completely unaware of the potential and what progress might be made in a very short time, rather than waiting decades.”

Pharmacogenetics is one of the fastest growing areas of medical research because individuals respond to different drugs in different ways. Tiny variations in a person's genetic make up, known as SNPs, single nucleotide polymorphisms, change the structure of key metabolic enzymes in subtle ways. Such changes mean that not all drugs work in the same way for everyone.

For instance, people with a SNP associated with the liver enzyme cytochrome P450 cannot process the common painkiller paracetamol into its active form in their bodies. This makes the drug totally ineffective in those people. SNPs also exist for the metabolic enzymes and protein receptors associated with chemotherapy drugs for cancer and can have subtle or devastating effects.

Doctors would like to know in advance whether particular patients with different genetic backgrounds or ethnic groups have a SNP associated with inactivity of such drugs. Moreover, the presence of certain SNPs in particular patients can make them more susceptible to a drug's side effects or potentially lead to adverse drug reactions. Knowing in advance how a patient might respond to a drug, would give doctors the opportunity to choose an alternative drug or therapy that is not doomed to failure in those patients.

Mackay and Taylor detail the various known SNPs associated with particular drugs. Variations in the cytochrome P450 that makes paracetamol ineffective for some people, also has an effect on how quickly patients respond to a wide range of antidepressants as well as the anticoagulant warfarin used to prevent clotting in heart patients. The team also explains how more than half of Caucasian HIV/AIDS patients who suffer side-effects caused by the antiviral drug abacavir carry a SNP that affects their white blood cells. Testing for the SNP in advance of prescribing the drug could improve outcomes for many AIDS patients.

There are also several examples of how IT coupled with genetics has allowed medical scientists to home in on particular SNPs associated with a positive response to a drug, such as Glivec in chronic myeloid leukaemia. The researchers suggest that, “The rapidly advancing field of genetics has opened up a number of opportunities in medical research likely to significantly impact future clinical practice, including the prescription of drugs based on a patient's genes.” They add, “Opportunities to learn new knowledge from routinely collected clinical data held within NHS databases, are very valuable, and should be encouraged.”