Cambridge Institute Getty Science Photography
Combining information from different cancer scans could offer a way to predict if immunotherapy will work, according to a small unpublished study.
Researchers said the computer programme offers a possible way of gauging the likelihood of successful treatment without the need for an invasive tissue sample (biopsy).
The approach, presented at the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics, is called radiomics.
It is an emerging field in medicine that gathers large amounts of information from medical images (CT, MRI and PET scans) and uses computer programming to sort the tumour images in to groups based on shared ‘radiomic features’.
The computer can pick up patterns from hundreds of images that doctors can’t see. And those patterns are used to indicate a characteristic of the tumour, in this case the abundance of a specialised immune cell in the tumours.
Immunotherapies have changed treatment for some advanced cancers, but the treatments only work in a minority of patients. This is why specific markers are needed to tell doctors which patients are most likely to benefit.
Previous lab work has shown that the more immune cells found in a patient’s tumour the more likely it is to respond to immunotherapy drugs called checkpoint inhibitors.
So the team, from the Gustave Roussy Institute in Villejuif, France, used a catalogue of tumour images from patients with head and neck, liver, lung, and bladder cancers to identify 80 radiomic features that may indicate the abundance of immune T cells inside the tumour.
Using these features they developed a radiomic score, based on the number of immune cells inside the tumour, and applied it to the CT scans of 137 patients on a clinical trial testing out particular immunotherapy drugs.
They sorted the patients into two groups; those whose radiomic score was below the average from the initial analysis and those who scored above the average. They found that patients with a higher score were 1.5 times more likely to be alive than those who had a low score, during the course of the trial.
Dr Roger Sun, who led the research, said this new way of monitoring patients was noninvasive, cost-effective and could be used to monitor patients throughout the course of their disease.
He added: “We are very encouraged by our findings that a signature based on imaging features could reflect the tumour immune infiltration and could predict response to immunotherapy.”
Professor Martin Glennie a Cancer Research UK-funded immunotherapy expert from the University of Southampton, said: “One of the most difficult and frustrating problems with immunotherapy treatments, including anti-PD-1 drugs, is knowing which patients will respond and which patients will not.”
He said this work was interesting because it potentially provides a means of selecting the minority of patients who are most likely to benefit from immunotherapy.
But he, and the Gustave Roussy team, acknowledged that the study was small and further research with larger groups of patients were now needed
“It’s still early days and the work needs careful validation in other centres and with bigger groups of patients,” said Glennie.
Sun, R. et al. A Non-invasive Computational Imaging Approach May Help Predict Response to Immunotherapy. Presented at AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics, 2017.