New multi-omic prostate cancer test 96% accurate

First this week, a potentially game changing prostate cancer screening test has been announced by a Cambridge-based company. Currently, there is no national screening programme for prostate cancer, despite it being the most commonly diagnosed cancer in England with more than 55,000 men diagnosed with it in 2023, Olympic cycling legend Chris Hoy among them. The new test uses a ‘multi-omics’ approach comprising the assessment of hundreds of biomarkers using an AI algorithm. I’ve been speaking to its creator, Professor Sir Chris Evans, from EDX Medical…

Chris E - It's a cancer that creeps up on men. 50,000 men get prostate cancer in the UK every year, 12,000 will die. 330,000 get it in Europe, 80,000 die. So it's a very serious problem for men between the ages of 45 and 85.

Chris S - But, diagnostically, what's the challenge? Because we do have ways to test for prostate cancer.

Chris E - The challenge we've got is that PSA is the main test. It's a single protein, a single biomarker, and it's about 43% accurate. And if its level is high, then you are referred on for an MRI, which of course is far more expensive and time-consuming, and you may be referred on for a biopsy. And that's based on the high PSA score. If it's a low PSA score, then you are not referred on because you are considered to be safe. And the problem there is that 14% of all low PSA score men can go on to get aggressive forms of prostate cancer. They don't know about it.

Chris S - But equally, we've got people who are getting a high level, and it isn't cancer at all, so we're then spending a lot of money screening and investigating and panicking the life out of people who haven't got prostate cancer.

Chris E - And that's about 70 odd percent of the men with high readings.

Chris S - What do you think you can do then to improve on our present situation?

Chris E - Our test has 100 biomarkers. It includes the PSA biomarker as one biomarker. So if the PSA, for example, is high, you have the rest of the 100-plus markers indicating whether or not cancer is highly likely to be present or it isn't. And it also indicates whether that cancer, if present, is likely to be slow-growing or very aggressive. And it also tells you whether you have hereditary genes increasing your risk of getting cancer in the prostate or not.

Chris S - In a nutshell then, you've got a large assemblage, 100-ish different markers, which are all associated with prostate cancer, and they give prognostic information about the likely outcome of a prostate cancer. When you bring all of those together, that paints a high-resolution picture of that person's situation, I presume.

Chris E - Exactly. It's a meta-analysis of all the biomarkers. They've been through tens and tens of thousands of prostate samples, and the algorithm that we've developed actually produces a very accurate reading and analysis of all of that data, which gives you the accuracy level above 96% that we've observed time and time again with samples in our own laboratories.

Chris S - PSA then, mid-40s in terms of accuracy, this is in the high 90s in terms of accuracy. So that is what you would describe an industry then as a potential game-changer, isn't it?

Chris E - Yes, that's why we think it is a game-changer. Obviously without going into prices, there's a difference between the cost of measuring one thing versus the cost of measuring 100 things. But the accuracy of 43% will compare to an ultimate accuracy of maybe between 96% and 99%.

Chris S - One of the most crucial things about any cancer is early detection, because that does literally make the difference between something potentially curative and something that you're going to live with rather than be able to get rid of. Do you think you can bring forward the pickup so a person would be detected earlier, and therefore their disease is almost certainly going to be more amenable to curative treatment with this?

Chris E - Yes, our test, for a number of reasons, has a very high sensitivity as well as specificity. We've already seen that we can pick up very early stage prostate cancers, and I think that is crucial because the treatment prognosis is excellent if you can catch these things early, and we can characterise the cancer from our test data as well. So we can help the MRI. The MRI scan that you will do shows you where it is, how big it is, its position in case you want to go in and do surgery later. But we produce a lot of genetic, genomic, proteomic data for that clinician before he even starts.

Chris S - And with that rich supply of data, can you feed it back in on itself so the machine is in a position to learn from its own activity? Because the surgeons now who are operating with robots are now training other surgeons using robotic data collected from other surgeons, for example, but equally we're training computer programs with the outcome of other computer programmes. Can you do the same thing and make your algorithm even better?

Chris E - Yes, we have already done that. This is a full AI-driven algorithm. Not only are we feeding the blood in and the urine in and looking at all of the markers and all of their ratios and all of their levels and their presence, we're feeding in what we call phenotypic variables; your age, the ethnicity of the individual, the size, obesity, family history, lifestyle, diet, alcohol, smoking, symptoms, all of that information is fed in and that's quite important because these things skew the accuracy of what's going to come out the other end as well. So we're combining very advanced biochemistry with all of what we call the phenotypic characteristics, into the algorithm, and it's learning all of the time it's improving. So we’re very optimistic actually about the ultimate outcome from this little test.

Chris S - Is it ready to go? As in, if I wanted to screen myself tomorrow, can I commission this test or is it somewhere yet down the track before we're in a position to use this commercially?

Chris E - No, you can't commission the test now because we will run for the next several months our final lap of validation studies where we will use different banks of prostate samples and we will look at a lot of different variables, just stress testing our own system. Then we will submit all of that information and data to the regulators, the MHRA, the FDA, etc. to get approval. So within, we expect about 12 months, we would hope to then launch the product into the marketplace.