Prostate cancer impacts countless lives, but not every case requires treatment - making early, precise diagnosis essential. In this second instalment of our two-part series on UEA’s pioneering prostate cancer research, Professor Dan Brewer, Professor of Medical Bioinformatics and Genomics and Dr Sergio Llaneza-Lago, Senior Research Associate, share their work on the Tiger Test.
This groundbreaking diagnostic tool identifies aggressive "tiger" cancers, sparing patients from unnecessary treatments and their side effects while ensuring those with high-risk cancers receive timely intervention. Philanthropic funding remains critical to bridging the gap between research discoveries and clinical application.
If you missed the first instalment, catch up with Dr Rachel Hurst’s insights on the PUR test and the bacteria test here.
What does analysing tiger test samples and data look like day-to-day?
Sergio: We have samples from patients’ tumours that have different cells. Cells can have different biological markers in them. After samples have been processed in the lab, data from them is sent to me for analysis. I run different statistical analyses on all the data to look at these markers and produce results based on them. One of the things we look for is how much of a patient’s cancer shows a specific biomarker called DESNT – this biomarker suggests a cancer is a more aggressive, or ‘tiger’ type of cancer.
But it can be tricky to determine how much DESNT there is in a sample. Imagine you have a bag of pick-and-mix sweets, and each tumour sample is a different sweet. You're trying to find specific colours and each colour represents different types of cells. So, you may have in your bag a gummy bear which is completely green, for example, and you can say, OK, this is 100% green, but you may also have a gummy strawberry, where the majority of it is red and a small part of it is green.
We found before this test was developed that most techniques people used would say, ‘OK, that gummy bear there is green, I'll put it in a group of green.’ And then they would say ‘OK, this strawberry is majority red and a bit green. I’ll put it with red,’ and they would ignore all the green on the strawberries. The algorithms we use on the Tiger Test can tell you ‘This gummy bear is 100% green – it’s 100% tiger type cancer. This strawberry is 80% red, 20% green – it’s 20% tiger cancer and 80% other types of cancer.’
Dan: And it could be that having a little bit of green is really important to your diagnosis. Those patients would be missed if we don't know there's a small amount of green.
Sergio: We know that patients whose cancers are 100% green, or 100% tiger, have a very poor prognosis. But those who have 20% tiger cancer also have a poorer prognosis than people with less than 20% tiger cancer or no tiger cancer.
Dan: To make sure our results are robust, we don’t just run this analysis once – we run it 1,000 times.
Sergio: Yes – one important side of machine learning is that sometimes you can get a random outcome. One way we avoid that is by running the test 1,000 times. We also run it with three different setups to make sure that we are not capturing noise. We implement a lot of steps to make sure that we have very reliable data.
Once we have looked at a batch of samples 1,000 times, our algorithm sorts them into groups according to the characteristics of the cancer – so it will group all the tiger samples together and all the other samples together. Then we compare the samples in one group against the samples in the other group to see what makes each group different. It could be, for example, that a particular gene is much more active in all the samples in one group. That suggests a treatment targeting that gene might be an effective way to treat the cancers in that group. So the test can allow us to inform tailored treatments for patients.
How long does the process take?
Sergio: There are multiple stages. The first is building the model, which takes about a week. This involves identifying the 500 genes that vary most across the samples and modifying our algorithm to look at these genes. This requires a lot of computing resources and a lot of checking - a single mistake could mean losing a whole week of work.
After that we do the thousand runs – applying our algorithm to a set of samples 1,000 times to learn how genes are expressed in each sample. One of the gene expression patterns we look for is DESNT, which indicates tiger cancer. This also takes about a week. Finally we compare the sample groups against each other. The results provide a complex picture and understanding them can take anything from one day to four weeks of collaborative work, with each team member bringing their unique expertise.
To date we have run this process – building the model, running it on samples, interpreting the data - 30 times with different settings to make sure we are running the Tiger Test in the best possible way, so that the final, accredited Tiger Test can be optimised, relatively quick, and reliable.
What would Tiger Test mean to patients?
Dan: The importance to a patient is that at the time of diagnosis, the patient gets the optimal treatment. As most men grow older, they're going to get prostate cancer. But the majority of cases of prostate cancer are not harmful or dangerous. So the question isn't if you've got prostate cancer or not, it's whether you should treat it or not. You could say, OK, let's treat everyone, but if you do treat prostate cancer you can get bad side effects like impotence, incontinence, bowel trouble. So we want to make sure that we only direct radical treatment to the patients that need it.
And having a tiger cancer is a strong indication that you should have treatment. Whereas if you don't have it, then it's a good indication that you should just be followed up with regular testing instead. So it's really crucial at the time of diagnosis for choosing the best treatment for patients.
"The importance to a patient is that at the time of diagnosis, the patient gets the optimal treatment. Having a tiger cancer is a strong indication that you should have treatment. Whereas if you don’t, it’s a good indication that you should just be followed up with regular testing instead."
What does donor funding mean to your research?
Sergio: Funding is extremely important because it allows our team to keep working towards the Tiger Test, which can really change the diagnosis of prostate cancer. We can make an impact which will avoid unnecessary surgeries and spare patients from nasty side effects, and help patients who need treatment to get it.
"Donor funding has been crucial to get to the point where we are now. Without donations, we wouldn’t have got as far as we have. The funds that have been raised for the lab are essential for turning a scientific idea into a clinically usable test."
Dan: Donor funding has been crucial to get to the point where we are now. And it's absolutely essential in getting the test ready for clinical use. A lot of funders either fund the beginning parts of research, the discovery parts, or the very end parts, like a clinical trial. But there's this crucial middle bit that's very difficult to fund, where you go from a scientific idea to a clinically usable test. And the funds that have been raised for the lab are essential for that process to happen. Without donations, we wouldn't have got as far as we have. We're really thankful for the funds that we receive.
Learn more about our groundbreaking research and how it’s making a difference.
If you would like to talk to us about this project, please feel free to write to us or call us.