Scientific image of kidney tissue with silver staining

Blending a new approach to genetic testing for cancer

Published 17 June 2020

Conventional tissue sampling for cancer diagnosis tests as little 0.0005% of a tumour. Is vital information missing? For better insights into tumour genetic mutations, scientists enlisted the aid of an unusual tool - a kitchen blender.

Two cancer patients receive the same diagnosis and the same treatment plan. One survives, the other does not.

The reality is that no two tumours are exactly the same. Some harbor genetic mutations that can be tricky to find - and treat - using traditional methods.

Group photo of RMS scientists on representative sample sequencing project
The RMS representative sampling team

With conventional tissue sampling for cancer diagnosis, as little as five in a million cells - or 0.0005% of a tumour - are tested, and what remains is typically incinerated. By throwing away the remaining tumour, are we missing vital information that could ultimately determine the best treatment options – including cancer immunotherapy - for each patient?

To get better insights into biological forces driving each tumour, scientists from Roche Molecular Solutions (RMS), in collaboration with the Francis Crick Institute and The Royal Marsden NHS Foundation Trust, have enlisted the aid of an unusual tool - a kitchen blender.

Household appliance meets groundbreaking science

A blender, you say? According to a study published in Cell Reports, this common household appliance could significantly improve genetic testing for cancer treatment. Instead of sampling thin slices of tumour and discarding the rest, the tumour is homogenised in a blender and DNA sequenced. (Watch this video to see how it works).

The concept dates back to 2014, when RMS scientists were puzzling over how to test an entire tumour rather than just one sample.

How many samples would you need to fully understand that tumour, wondered Nelson Alexander, the research project’s principal investigator at RMS. For the answer, he turned to biostatistician colleague Kate Leith.

“Kate put together a very nice model to answer this question and that was 1,500 tissue slides,” Nelson recalled, compared to the single slide used for diagnosis today.

That wasn’t going to work. Nelson asked Kate to tweak the model.

‘Get yourself a blender’

The smallest number of slides estimated was 20 - still not possible. “I told him, ‘you can’t get there from here,’ ” Kate recalled of a meeting with Nelson.

She did have one suggestion: “Get yourself a blender.”

“We both laughed,” Nelson recalled. “It was a ridiculous thing to say. But we talked about representativeness - how one sample could truly represent the entire tumour - and by the time I got back to my desk, I knew she was 100% right.”

Nelson and the team brought in kitchen blenders, and early experiments led to some “monumental failures.” Then one day, Nelson dropped human tonsil tissue in the blender to see what would happen. To their astonishment, the research team saw intact cells and tissue chunks under the microscope.

“There are rare moments of discovery in your life, and this was one,” Nelson said.

We suspect there could be many new targets for therapy hiding in tumours.”
Nelson Alexander Roche Molecular Solutions

Proving their theory

“I still get goosebumps thinking about it,” senior scientist Stacey Stanislaw added. “We saw this was important, and we could stop worrying about looking like weirdos blending up tumours. Now we needed data sets to prove our theory.”

The blending moved into high gear. Stacey brought in heaps of grocery store meat, piled it into a blender and added a tumour the size of the tip of his pinkie.

“Then I hit go. The idea was that in every sample of this homogenised tissue, I should find a tiny bit of tumour,” Stacey explained. And that’s exactly what they found. The Roche team refined the process to homogenise and then DNA sequence the material. This method, called representative sampling, creates a more robust DNA sample for sequencing, which may benefit all sequencing platforms.

In 2015, the team took their idea to their tumour heterogeneity research partners at the Crick Institute in London.

Perfect partnership

The folks at Crick were not taken aback by the idea. They, too, had been considering homogenising tumours for the same purpose.

Photo of Samra Turajlic
Samra Turajlic

“Both teams had been headed in this direction and this seemed a perfect partnership,” said Samra Turajlic, a medical oncologist at The Royal Marsden in the UK, and a senior fellow at Crick. Researchers there had been piloting their own techniques. One had even ordered a blender.

“It was fantastic to have had both teams invested in thinking about the problem and arriving at this possible way of taking it forward and evaluating it,” Samra said.

The partners have worked together to test the theory using existing data and to gain insights into more than 200 patients with kidney, lung or bladder cancers.

Then, in collaboration with The Royal Marsden through extensive testing on a case of kidney cancer, this method gave identical genetic results 95% of the time, compared to only 77% consistency with the current methods.

Game-changing possibilities

“This method could be a game changer for tumour sampling in hospitals and in research,” said Kevin Litchfield, study lead and bioinformatician in the Translational Cancer Therapeutics Laboratory at Crick. The researchers hope their findings will ultimately lead to improved cancer treatments.

“We suspect there could be many new targets for therapy hiding in tumours,” Nelson said. “They have been invisible until now because no one has had the ability to tease them apart. It’s the coolest, craziest, most awesome science I will do in my lifetime.”

Tags: Science, Diagnostics, Oncology