Welcome to the NHS in Leeds

New breast cancer drugs could help more than previously thought

"Up to one in five women with breast cancer could benefit from a type of treatment currently only given to patients with a rare form of the disease," The Independent reports.

Research suggests around 20% of women with breast cancer may benefit from a new class of drug known as PARP inhibitors.

PARP (poly ADP ribose polymerase) inhibitors were designed to treat women with breast cancer related to inherited mutations in the BRCA1 and BRCA2 genes (the so-called "Angelina Jolie mutation", because of the film star's history of the mutation) which are thought to account for up to 5% of breast cancers.

But this latest research suggests that as many as one in five women with breast cancer could benefit from PARP inhibitors.

In the study, researchers designed a computer programme to recognise genetic "signatures" associated with problems in the body's ability to defend itself against cancer. These problems are linked to the BRCA1 and BRCA2 mutations.

The software looked for non-inherited genetic problems that were similar to the problems caused by inherited BRCA1 and BRCA2 mutations, which might mean they could be treated in the same way. Of the 560 people tested, the model found 90 people – around 20% of the group – had genetic problems similar to those caused by BRCA1 and BRCA2 mutations, suggesting they might also benefit from PARP inhibitors.

The next step would be to see whether using PARP inhibitors in these types of cases would be helpful.


Where did the story come from?

The study was carried out by researchers from more than 30 medical institutions, led by the Wellcome Trust Sanger Institute in the UK. It was funded by the European Community, the Wellcome Trust, Institut National du Cancer in France and the Ministry of Health and Welfare in Korea.

There are potential conflicts of interest as three of the researchers have a patent for the code and intellectual property rights of the algorithm used to identify BRCA1 and BRCA2 deficiency. Another researcher was involved in the invention of PARP inhibitors.

The study was published in the peer-reviewed journal Nature Medicine.

The Independent, The Sun and BBC News all gave accurate overviews of the research. However, reading the headlines, you would not realise that the drugs have not yet been tested in the groups that might benefit.

Also, The Independent's headline: "One in five breast cancer patients not receiving new treatment that could benefit them," suggests the drugs are being deliberately withheld, when in fact there had not previously been any suggestion they could help, and we still don't know whether they are helpful.


What kind of research was this?

This was laboratory research using whole genome analysis of tissue in the laboratory, and feeding it through computer algorithms. This sort of research is good at generating theories, but clinical trials are needed before we know the practical application of the results.


What did the research involve?

Researchers took DNA from breast cancer tumours removed from 560 people and did whole genome sequencing. They used DNA from people known to have inherited BRCA1 or BRCA2 mutations to develop a computer programme to recognise signature features linked to the mutations. The programme then compared these results with results from people with non-inherited breast cancer.

The programme created a model, called HRDetect, to spot people who didn't have the inherited BRCA1 or BRCA2 mutation, but did have DNA changes that made them deficient in BRCA proteins. HR refers to homologous recombination repair, a method by which BRCA1 and BRCA2 proteins defend cells against cancer.

They tested HRDetect on several groups, and looked to see if it could be used on biopsy samples, rather than large tissue samples from removed tumours. They tested it on ovarian and pancreatic cancers, as well as the initial breast cancer group.


What were the basic results?

The HRDetect model found that, in women with non-inherited breast cancer:

  • 69 out of 538 breast tumours were BRCA1 or BRCA2 deficient
  • 16 out of 73 ovarian cancer tumours were BRCA1 or BRCA2 deficient
  • 5 out of 96 pancreatic cancer tumours were BRCA1 or BRCA2 deficient

The researchers say the model showed 98.7% sensitivity (meaning that it missed less than 2% of tumours which were BRCA1/BRCA2 deficient). This is very high. They don't report on specificity – how many might have been wrongly identified as deficient (also known as a false positive result).

The model performed as well on samples of tumours from small needle biopsies as on large specimens removed during surgery. This suggests that the model could be used early in the clinical process, from the patient's first biopsy. This might help direct treatment from an early stage.


How did the researchers interpret the results?

The researchers said the model was "extraordinarily effective" as a predictive tool.

"If the tumours with predicted BRCA1/BRCA2 deficiency also demonstrate sensitivity to PARP inhibitors, this would unearth a substantial cohort of patients who could be responsive to selective therapeutic agents," they add. They say that this is "potentially transformative" of treatment and recommend use of their model in trials of PARP inhibitors.



Advances in genetic technology are happening fast, improving our knowledge about which treatments may be most suitable for which types of cancer. However, testing these theories takes time, which can be frustrating for researchers, when newspaper headlines suggest people should already be receiving new treatments.

This study potentially widens the pool of people who may benefit from targeted cancer treatment with PARP inhibitors, from around 5% to around 20%. That's clearly good news, but the potential for benefit needs to be tested in clinical trials.

The researchers express a great deal of confidence in the accuracy of their model. It would still be useful to see it externally validated in other groups of people, before we can know how well it performs in the real world. It would also be useful to see how specific the test is, as well as how sensitive it is.

One question remains about the 20% figure. It's not clear how the researchers selected people to take part in the study. They deliberately selected 22 patients who were known to have BRCA mutations. But we don't know whether the others in the study were randomly selected and representative of all people with breast cancer. If they were not randomly selected, then the 20% figure may not hold true for the wider population of people with breast cancer.

Most people with breast cancer don't have inherited gene mutations. Find out more about predictive genetic tests for cancer risk genes