Can genes predict drug response?

Behind the Headlines

Monday March 1 2010

Genetic tests may help to personalise treatment

Researchers have identified “a genetic signature that can determine whether breast cancer is likely to respond to a common treatment”, The Times reported. It said researchers had found that measuring the activity of six genes could forecast whether a breast tumour was sensitive to paclitaxel (Taxol), a chemotherapy drug.

This study looked at whether the activity of the genes, identified as being related to paclitaxel response in a previous study, could predict whether a breast tumour was sensitive to paclitaxel. It found that their activity was a good predictor of how women with a particular type of tumour, called triple-negative tumours, would respond to paclitaxel.

The findings of this study will need to be replicated in other groups of women, particularly as there were only a few paclitaxel-treated women with triple-negative tumours in this study. Such research will also need to confirm how many false positives and negatives might be expected from the test. These studies would need to take place before the test could be considered for trials in clinical practice.

 

Where did the story come from?

This research was carried out by Dr Charles Swanton from Cancer Research UK London Research Institute, Nicolai Juul from the Technical University of Denmark and colleagues from other institutions in Europe and the US. The study was funded by charities including the UK Medical Research Council, Cancer Research UK and the National Institute for Health Research, as well as the European Commission. The study was published in the peer-reviewed medical journal The Lancet Oncology.

The research was reported by The Times, BBC News and the Daily Express, all of which gave reasonably accurate coverage of the study.

 

What kind of research was this?

This was a retrospective cohort analysis of pooled study data, investigating whether the activity of a group of six genes could predict whether a woman with breast cancer would respond to paclitaxel treatment. Paclitaxel is often combined with other chemotherapy treatments to shrink tumours before they are surgically removed. However, only 15-25% of breast cancer patients fully respond to preoperative chemotherapy. If individuals with a particular genetic makeup can be shown to be more or less likely to respond to certain chemotherapeutic agents, genetic testing could be used to tailor chemotherapy treatment to maximise their chances of working.

Treatment that is tailored to people's genetic makeup is a growing area of research, aiming to increase the likelihood that the selected treatment works, while reducing the risk of adverse effects. This type of study - looking at how well a new test predicts a treatment response in a group of people whose outcome is already known - is an important step in seeing whether the test might be a useful one. Once a study identifies particular genetic variations that are associated with response or adverse effects, these findings need to be replicated in other populations to ensure that they work, before they can be tested in clinical practice.

 

What did the research involve?

The researchers pooled data from five different studies, in which women with breast cancer had chemotherapy to shrink their tumours before surgery. Some of the studies involved paclitaxel-containing regimens, while others did not. While they were undergoing surgery, the patients were assessed to see if the chemotherapy had been successful (the tumour had responded).

This research measured the activity of six genes in these women’s tumours in tissue collected before they had any chemotherapy. The researchers then looked at whether a particular pattern of activity of these genes was associated with the likelihood that a woman’s tumour had responded to chemotherapy.

Previous research had identified these six genes as likely candidates for affecting the response of paclitaxel. In that study, the researchers examined the effects of 829 genes in triple-negative breast cancer cells in the laboratory. Triple-negative tumours lack oestrogen receptor, progesterone receptor, and human epidermal growth factor receptor-2 (HER2), and this type of tumour is associated with a particularly poor treatment outlook if there is residual disease after chemotherapy. Based on the results of their analyses, the researchers selected four genes with roles in cell division (BUB1B, CDC2, AURKB, and TTK) and two genes with roles in metabolism (breakdown) of a compound that promotes cell death (UGCG and COL4A3BP). Based on this study, higher activity of the cell division genes was predicted to be associated with sensitivity to paclitaxel, and higher activity of the metabolism genes predicted to be associated with resistance to paclitaxel. The test they developed involved measuring the difference in activity of the two groups of genes, which they called the “paclitaxel response metagene”.

The tumours’ response to the drugs was determined during surgery, and was defined as no evidence of residual invasive cancer in the breast or lymph nodes. To determine whether their test was a good predictor of response, the researchers used standard statistical tests for assessing the predictive abilities of diagnostic tests. They carried out analyses looking at all women, and also only at women with triple-negative tumours (57 women treated with paclitaxel and 203 women not treated with paclitaxel). They also carried out analyses that took into account other predictors of paclitaxel response (potential confounders), including oestrogen receptor status, HER2 status, tumour grade, and whether the tumour had spread to the lymph nodes.

 

What were the basic results?

The activity of the paclitaxel response metagene was a good predictor of response to paclitaxel in all women and in women with triple-negative disease. However, the metagene was not a good predictor of response to non-paclitaxel chemotherapy.

When all women were analysed together, the metagene was associated with a significantly increased likelihood of paclitaxel response in analyses that did not take potential confounders into account, but this association was no longer significant after adjustment for potential confounders. However, when only women with triple-negative tumours were analysed, the metagene was associated with a significantly increased likelihood of paclitaxel response in both unadjusted and adjusted analyses.

The odds of complete response with paclitaxel-based treatment in the patients with a high paclitaxel response metagene score was over five times the odds of complete response in the patients with a low paclitaxel response metagene score (odds ratio 5.65, 95% confidence interval 1.67 to 19.11).

 

How did the researchers interpret the results?

The researchers conclude that these genes show promise as predictors of which triple-negative breast tumours will respond to paclitaxel treatment. They say that the study highlights the ability of functional genomics (looking at dynamic aspects of gene function) to identify markers that can predict drug response.

 

Conclusion

Identifying ways to tailor treatment based on an individual’s genetic makeup is a growing area of research, aiming to increase the likelihood of a treatment response while reducing any adverse effects. This study has identified a panel of genes whose activity could help to predict those women with triple-negative breast cancers who will respond to paclitaxel treatment before surgery.

These findings will need to be replicated in other groups of women, particularly as there was only a small number of paclitaxel-treated women with triple-negative breast cancers in this study. Such research will be needed to confirm how many false positives (proportion of women that the test predicts will respond to treatment that do not) and false negatives (proportion of women that the test predicts will not respond to treatment that do) the test has.

These further studies will need to be completed before tests in clinical practice could be considered.

Analysis by Bazian

Edited by NHS Choices

Ratings

How helpful is this page?

Average rating

Based on 0 ratings

All ratings

Add your rating