The type of statistics used to analyse crime in cities can also help identify immune ‘hotspots’ within lung tumours. Crime map image source: BBC
This entry is part 24 of 30 in the series Science Snaps
To tackle crime, you need to know where it’s happening and why.
London’s Metropolitan Police collect and analyse data on crimes committed in different neighbourhoods and map local crime rates compared to the average across the capital. The stats highlight ‘hotspots’, like in the map above, that can focus efforts and resources.
Research has shown that cancers are like a complex city, with bustling neighbourhoods of different cells, some of which are more dangerous than others.
In the mix are immune cells, which operate much like a city’s police force. And that’s why scientists are turning to crime-mapping tech as a new way to look at cancers.
We visited a researcher who’s applying these statistical methods to analyse lung tumours. And in the hunt for ‘hotspots’ of immune activity, she hopes to produce maps that could predict how patients will respond to treatment.
Mapping cancer evolution
Dr Yinyin Yuan, a Cancer Research UK scientist based at The Institute of Cancer Research, London, pulls up a picture on her computer. It’s an image of a patient’s lung tumour sample taken using a microscope. They’ve volunteered to have their cancer analysed as part of the Cancer Research UK-funded TRACERx study, which is giving our scientists the greatest possible understanding of the biology of lung tumours.
As part of this study, Yuan is developing an artificial intelligence program to allow researchers to probe the cellular neighbourhoods inside the tumour.
Their aim is to understand how the body’s immune system responds to cancer as it evolves and develops.
“Cancer cells don’t live alone, and we need to understand how they interact and the evolutionary pressure they’re under, to be able to stay one step ahead and anticipate cancer’s next move,” she says.
Crime scene investigation
Stained in colourful shades of pink and purple, but with no discernible features, it’s difficult to see how the images of cancer samples on Yuan’s screen could tell us anything about a patient’s disease. But as she flips the image, it’s transformed into a map that clearly points to one region being different to its neighbour.
And it’s the transformation of images like these, as shown below, that are revealing a snapshot of the body’s immune cells interacting with different parts of the lung cancer.
“Here we’ve applied the same rules used to identify crime hotspots in London,” says Yuan. “Police look at the number of crimes committed according to the population of a given area. And we look at the same within a tumour.”
Computer scientists in Yuan’s lab are developing an algorithm that can identify different types of cells within a tumour. This will help to build of picture of where each individual cell sits and how it interacts with neighbouring cells.
The team don’t just want to spot the regions carrying the most immune cells. Instead, they want to see where the number is high compared to the number of cancer cells.
“It’s in these immune ‘hotspots’ where the body’s immune system is recognising and interacting with cancer cells, and where we think immunotherapies are more likely to work,” she adds.
The challenge of lung cancer
Delving into the intricacies of lung cancer is no small feat and TRACERx involves the collaboration of over 200 researchers and clinicians based at centres across the country, including staff at the Cancer Research UK Lung Cancer Centre of Excellence and UCL.
The study has already taught us that no two parts of a lung tumour are genetically identical – different regions evolve separately, creating a mess of different tumour characteristics that make the disease very difficult to treat. And this is reflected in different levels of immune activity.
While some parts of the tumour are packed with immune cells, others appear to be completely deserted. And according to Yuan, there’s lots to gain from studying these areas too.
“We’re just as interested in parts of the tumour that are ‘immune cold’,” Yuan says. “These regions may have evolved to hide from the body’s natural defences and we want to find out how.
“If we can understand the processes that lead to immune evasion, we can look for and develop new targeted treatment options.”
Bringing AI to the clinic
Despite pushing the boundaries of new technologies, Yuan’s lab is grounded in the clinic.
Pathologists from the University of Leicester and the Francis Crick Institute in London are teaching Yuan’s computer system to identify 8 different types of immune cell. “It’s like we’re downloading the brain of a pathologist,” says Yuan. “We want the computer to copy their thinking process.”
“In the lab, pathologists spend hours identifying and counting different cell types, but with our algorithm, this painstaking task can be completed almost instantly.”
Mapping out cancer’s immune hotspots in this way gives us an entirely new way to look at tumours. In the future this could allow doctors to predict how well a patient will respond to certain treatments, and even help personalise care.
“What we’re hoping for is the perfect marriage of artificial intelligence and clinical practice,” adds Yuan. “As our algorithms become smarter, we’ll be able to analyse tumour samples very quickly and in immense detail, giving doctors the best picture of a patient’s tumour and informing the most suitable course of action for them.”
Kathryn is a senior science media officer at Cancer Research UK