Combining imaging and gene analysis could transform breast cancer diagnosis
Scientists have developed a computer system that can automatically analyse images of breast cancer cells to look for telltale signs that the tumour will be aggressive. The technique, details of which were published this week in Science Translational Medicine, could be used to give women with breast cancer a more accurate prognosis and help tailor their treatment accordingly.
The standard process for diagnosing breast cancer involves a pathologist examining the tumour cells with a microscope, which can be time-consuming and subject to human error. The new digital system scans through microscopy images automatically, looking at the make up of the cells within a tumour, and can pick up important details the human eye would miss.
Scientists believe that combining these results with an assessment of the key genetic changes that are crucial for the development of breast cancer could give women a more accurate prediction of how their cancer is likely to behave.
Dr Yinyin Yuan, who jointly led the research at the Cancer Research UK Cambridge Research Institute, has recently founded the Computational Pathology and Integrative Genomics team at The Institute of Cancer Research, London, to further refine this new diagnostic tool.
Dr Yuan said: “The appearance of tumour cells under a microscope has been used to diagnose cancer since the 19th century, but by applying modern automated techniques we can hugely increase the accuracy of our analysis and also generate useful new information. Over recent years, genetic testing has become increasingly important in cancer diagnosis as well and by combining these two approaches we can create a full picture of an individual’s cancer to best personalise their treatment.
“My team at The Institute of Cancer Research are working to perfect this new technique and we also plan to test its usefulness in other types of cancer.”
In the new study, scientists used the imaging technology in combination with tumour gene information to predict whether women with a particular type of breast cancer called oestrogen receptor negative would survive for five years after diagnosis. A computer model that combined both sources of information made the right prediction in 86 per cent of cases, compared with just 67 per cent in a model that included just the genetic information.
The computer imaging system accurately assessed the number of white blood cells that had entered the tumour – an indication of the strength of the patient’s immune response against the tumour, which has an impact on whether they will beat the cancer.
The researchers also found a new way to predict survival based on the arrangement of supporting cells within the cancer. Known as the stroma, these cells play a role in encouraging the growth of breast cancers and are not detected in current tests.
Dr Florian Markowetz, the lead researcher at Cancer Research UK’s Cambridge Research Institute, said: “Cancers are a mix of different cells – not just cancer cells. They also include cells from the immune system and others that support the structure of the tumour. These cells play an important role in how cancers behave, but this information is not currently detected in tests that focus on the genetic make up of the cancer.
“By bridging the gap between the ‘two cultures’ of pathology and genomics our research has a huge potential for better understanding breast cancer.”