29 February 2008
The media have given a lot of attention this week to recent research reporting that anti-depressants are ineffective in the majority of patients (see Anti-depressants' 'little effect', BBC news). A paper in the open-access journal PLoS Medicine reports a meta-analysis (a statistical technique for combining. the findings from multiple independent studies) of clinical trial data submitted to the US Food and Drug Administration (FDA) for four of the most commonly prescribed anti-depressants [Kirsch I et al. (2008) PLoS Med 5(2): e45]. These drugs were all of the newest type known as selective serotonin uptake inhibitors (SSRIs), generally used as the first line of treatment for moderate to severe cases of depression, and had all previously been approved by the FDA.
Clinical trials are required to demonstrate that a new drug has acceptable efficacy and safety, ie. shows reasonable success in treating the condition for which it is intended, without significant adverse side-effects. Trial drugs are compared with placebo (inactive) compounds. However, when the researchers compared the data from the 47 separate trials, they found that the clinical benefit of all four drugs fell below the recommended level for clinical significance in all but the most severely depressed patients; the drugs did show more effect than the placebos, but not as much as would typically be required to demonstrate efficacy. Among severely depressed patients where the difference was significant, it was suggested that a generally poorer response to the placebo drug might be the reason, rather than efficacy of the anti-depressant. Lead researcher Professor Irving Kirsch from the Department of Psychology at the University of Hull has commented: "Given these results, there seems little reason to prescribe anti-depressant medication to any but the most severely depressed patients, unless alternative treatments have failed to provide a benefit".
These outcomes have led to a great deal of debate; firstly, about the most appropriate therapies for depression (although it is perhaps not altogether surprising to hear psychologists advocating alternatives to medication such as ‘talking therapies’); and secondly, about the regulation of drugs and how useful the current system of clinical trials are in demonstrating whether drugs work or nor. However, media coverage has largely skipped over the point that, in fact, even the most effective drugs do not work in a substantial proportion of patients. Efficacy rates for the vast majority of common drugs typically vary from 30-50% (see Glaxo chief: Our drugs do not work on most patients). That is, the drugs work very well in some patients, and poorly or not at all in others. One potential reason is that depression is not a single clinical entity, but rather may result from different disease pathways; drugs that alleviate one pathological process may have no effect on another. Another key point is that, just as common diseases arise from a combination of multiple different environmental and genetic factors, variation in drug response is also affected by all sorts of individual factors such as age, gender, diet, co-morbidities (other diseases present), co-medications (other drugs being taken) – and genetics.
Pharmacogenetics, the study of the genetic factors that underlie variation in drug responses, is a rapidly expanding field that has great potential to improve the drug development and approval process, and benefit patients. Understanding pharmacogenetics and pharmacogenomics could, it is hoped, eventually lead to drug prescription being linked to information on which of the relevant genetic variants a given individual has. If enough of this information were recorded and stored electronically, a GP could consider it along with the other individual factors, in choosing the most appropriate type and dose of drug for a given condition, in a specific patient.
For example, warfarin is an effective anticoagulant drug that is used to thin the blood and prevent thrombosis (blood clots) in many disorders. However, very careful monitoring of patients is necessary to achieve the correct dose which provides a reduction in the risk of clotting without going to far and leading to haemorrhage (serious uncontrolled bleeding). Doctors know that different people require very different doses to achieve this effect, and pharmacogenetic research has shown that certain common genetic variants (polymorphisms) in the CYP2C9 and VKORC1 genes (that encode metabolic enzymes) can affect individual responses to warfarin and potentially guide dosing [McClain MR et al. (2008) Genet Med. 10(2): 89-88]. Similarly, polymorphisms in the ABCB1 gene, which encodes a drug transporter molecule, could potentially guide the choice of anti-depressants [Uhr M et al. (2008) Neuron 57(2):203-9].
There are also prospects pharmacogenomic knowledge to inform and improve drug development and clinical trials; for example, if there is a genetic factor that significantly affects the response to a new drug, comparing trial results based on which variant is present might show that a drug is highly effective in one patient sub-group but ineffective in another, as opposed to an average result of moderate efficacy overall; it might then be approved for use, but only among patients with the relevant genetic variant.
So, will Drugs that don’t work become a headline of the past? Not in the immediate future; although research is providing new insight into pharmacogenetic factors, in most cases the evidence is not yet strong enough to warrant genetic testing associated with drug prescription. Recently the US Evaluation of Genomic Applications in Practice and Prevention Working Group issued a recommendation statement on the use of cytochrome P450 (CYP450) genetic testing in adults with non-psychotic depression beginning treatment with SSRIs, saying that there was insufficient evidence to support it (see previous news). Another major barrier is the currently prohibitive cost of genetic testing, and the associated delay in prescription if test results would be required to select the best treatment. However, such ‘personalised medicine’ has great potential.
Summary: Pharmacogenomics may help us to achieve a better match between individual patients and drugs that will be effective.