Using data science to identify better therapeutic targets in cancer

A major goal of precision medicine in cancer is to identify treatments that will be effective in patients with specific mutations, so that, rather than treating patients based on the organ their cancer occurs in, they will be treated based on the mutations present in their tumours. The hope is that such targeted therapies will provide a better outcome for patients than standard chemotherapy, with better survival outcomes and fewer side effects. Many such targeted therapies are already in use – for example a drug called herceptin is widely used as a therapy for breast cancer patients whose tumours have a mutation of the HER2 gene. Similarly, a drug called lynparza is used to treat ovarian cancer patients whose tumours have a mutation of the BRCA1 or BRCA2 genes. However, many mutations that are common in patient tumours still have no associated targeted therapies. Consequently for many patients there is no targeted therapy available.

Many scientists are seeking to address this by performing experiments to identify new targeted therapies. A common approach is to perform experiments in cancer cell lines – cells taken from tumours and grown in the lab. Scientists can try adding different drugs, or knocking out different genes, to see what kills cancer cell lines with a specific mutation. Hundreds of examples of such new therapeutic targets have been identified but unfortunately many of them only seem to be effective in very specific contexts. What works in one tumour cell line might not work in another. In some ways this is to be expected – tumour cells can have many different mutations, not just the one you are trying to target, and these additional mutations can alter the efficacy of targeted therapies. However, if you want to develop a treatment that will work for a wide range of patients, you can’t rely on something that only works in very specific contexts.

In a new study, published in the journal eLife, Dr Colm Ryan (UCD) combined results from different studies to identify what therapeutic targets are identified again and again for different mutations. These ‘context-independent’ targets are much more likely to make useful new drugs for cancer then targets that were only found in a single study. He and his team identified 220 such targets. Furthermore, he states, “we found that we could predict whether a target identified in one study was likely to be found in additional studies. This is very helpful as it will allow other researchers to ‘pick winners’ earlier on in the drug discovery process instead of wasting time on drug targets that will only work in very specific contexts. Finally, we used our predictions to identify two new drug targets for mutations that are common in patients but that have no currently associated targeted therapies.”

Dr Colm Ryan is funded by the Irish Research Council under the Laureate Award and is based in Systems Biology Ireland in UCD.