Harpal Kumar: a cancer treatment revolution

Harpal Kumar, image courtesy of AstraZeneca

"Click here to listen to the full podcast episode"

Kat: One exciting application of CT DNA is in detecting cancer earlier. My old boss from Cancer Research UK, Sir Harpal Kumar, is now the President of GRAIL Europe, a company that has developed a blood test for multiple different types of cancer based on looking for DNA methylation patterns on fragments of circulating tumour DNA. These methylation marks are molecular tags attached to DNA that are associated with changes in gene activity, also known as epigenetic changes. So, how can we use this information to detect cancer, and what else can it tell us?

Harpal: What we've done over the years is to understand which regions of the genome that are aberrantly methylated can best distinguish cancer from non-cancer. And so we use that to essentially figure out who is likely to have cancer, who is not. But the beauty of methylation is it also enables us to say, where is that signal coming from with pretty high fidelity. So for example, we're able, first of all, to say, this person is likely to have cancer. And then secondly, we're able to say, and it's most likely a pancreas cancer or an ovarian cancer, or what have you.

Kat: That's really important because it's not really very helpful just to say to someone "Oh dear, it looks like we've picked up a signal of cancer". It would be helpful if you knew at least where to start looking for it.

Harpal: Indeed, we think that's critically important.

Kat: So you're looking at these epigenetic marks, these methylation marks on the DNA. You're not looking at changes to the DNA itself, mutations that we might more often associate with cancer. And so this is more just, are we seeing the presence of aberrant cancer cells based on these methylation marks rather than going, oh, it's got this mutation or that mutation.

Harpal: Correct. I have to commend the way that GRAIL went about this, because actually they started out by doing a completely unbiased discovery study to really understand which technological approaches gave the best signal in terms of this critical question of differentiating cancer from non-cancer. And so they looked at mutations and they looked at copy number variations, and they looked at every sort of approach you might think of. And actually methylation came out as giving the best differential signal, and indeed that none of the other approaches - mutations, copy number variations, et cetera - none of those added anything to that signal. So that's how GRAIL ended up with this methylation assay and then has further optimised it to look at the very specific regions that give the best signal to noise differentiation.

Kat: I do find this kind of technology amazing because I worked in the area of epigenetics when I was doing my PhD. This is sort of early 2000s. And we were getting into the idea of studying DNA methylation, these little flags that we find on the sequence of DNA, and the technologies we had then were not sophisticated. You know, the DNA sequencing technology wasn't very sophisticated, and then our ability to look at methylation marks and see whether things were more methylated or less methylated than we expected at the genes we were studying. That was really, really hard. So there must have been some kind of like quantum leap in the technology between then and now that enables you to do this.

Harpal: Well. I mean, as you rightly say, there have been very significant advances in technology over the last five, 10 years. And by the way, it isn't just the technology itself. It's also the price of that technology, and the fact that it's come down so much means that you can look in much greater depth on a particular DNA sample. And that enables you to detect signals that are present in actually minute quantities. So there's been advances in sequencing, there's been advances in chemistry. And then in terms of the overall workflow that enables you to do this at large scale for what we believe will be an acceptable cost.

Kat: The computing as well, presumably to interpret all this data. I feel like you're almost looking for needles in haystacks when you're talking about finding these tiny fragments of DNA and these little methylation marks, and are they more or less than we're expecting, that's a big statistical computational problem as well.

Harpal: Huge statistical computational problem. And that the volumes of data we generate through our work and through our clinical studies dwarfs pretty much anything else that's out there in genomics right now.

Kat: So thinking about applying this to cancer screening - so here in the UK, we have national screening programs for cervical cancer, for bowel cancer, for breast cancer. How are you thinking about applying this technology to cancer screening? Because obviously with screening, it's really important that you can pick up cancers early at a stage when they are more likely to be treated successfully, that you're not picking up these cancers that actually might not cause a problem, or maybe don't need treating now, and that, you know, it is useful and informative. So how are you testing this out?

Harpal: Yeah, we have from all of the studies, we've done already we know the test works. We know we can pick up cancers. What we're now doing is some very large clinical studies looking at the clinical utility. So we're in the midst of a very large randomised control study in the UK called the NHS Galleri study. It's 140,000 people. And the objective is to see, can we make a significant difference in terms of reducing the number of people who are only detected at late stage? Can we pick them up earlier? There are several critical things. When you think about the use of technology like this in population screening, you want to be able to detect as many cancers as possible, but you want to do it with a very low false positive rate. So you want to have, what's called very high specificity. Why is that important? Because you're screening what would otherwise be a healthy population. And so what you don't want to do is tell lots of people they might have cancer when actually they don't.

Kat: Oh, no, you do not.

Harpal: Yeah. And, and so because that not just causes anxiety obviously, but it leads to a whole bunch of follow on investigations that may not be necessary and may themselves have risks associated with them. So you wanna have a very low false positive rate. And you know, it looks like our false positive rate is an order of magnitude better than the current screening programs we have. So that's very promising.

Kat: So we're gonna wait for the results of that. And that will be very interesting to see how this goes. But we talked earlier in the podcast about the use of circulating tumour DNA for studying the mutations in cancer, studying how cancer's growing in the body and then using this in the context of things like clinical trials and in the drug development process as well. So how are you working with AstraZeneca to use your technology for the kinds of questions that they're trying to ask about how can we find better treatments for cancer?

Harpal: Sure. So most of what we've talked about so far has been about how we apply this technology in people who don't yet know they have cancer. But of course the same kind of technology could be very applicable for people who do know they have already had a cancer diagnosis and how can we manage them better? And there are two or three different respects or different opportunities in this arena. One is knowing that CT DNA can be predictive of outcome, can we use that to essentially to assess prognosis for an individual?

Kat: So that means that by looking at a sample of their blood, you might have an idea of how their cancer's likely to go - is this gonna be a bad one? Is their outcome likely to be good, their sort of survival chances?

Harpal: Correct. And what that gives you the opportunity to do is to say for those who look like they might have a worse prognosis, should we invoke treatment earlier or more intensively? And particularly when we are picking up early stage cancers, you want to be really sure about who you should give the more intensive treatment to and who you shouldn't, but it offers the opportunity for what are called neoadjuvant or adjuvant studies in those early stage cancers. And of course up until now, we haven't detected them early enough to be able to do such studies. But now we do have those opportunities. So that's one whole arena.

Kat: Another arena is can we then use this kind of technology to monitor how patients are doing when they are on treatment? Can we figure out those who are responding or not responding? Can we understand if people are having recurrences before that may be evident clinically through imaging or some other technology? And indeed, can we see if someone's actually developing another cancer? Because we know people who've been diagnosed previously have a higher risk of developing another cancer. So all of these things, all of these areas are opportunities and they're areas that we are exploring in partnership with AstraZeneca.

Kat: Anyone listening to this who's had cancer, or knows someone they love with cancer will know that even once you've been told it's "OK, I think we've think we've got it. We, we think you're doing alright. We think you're all clear." There's always that doubt in your mind, is it going to come back? Is it coming back? Every sort of lump, bump, symptom, something strange. You're like, is it coming back? Could these kinds of technologies, whether it's the GRAIL test or other circulating tumour DNA tests, be a way of, you know, you can have a blood test every six months or every year and know really are you in the clear,? Because that would be incredibly reassuring for patients and their families.

Harpal: I mean, I think you've described it beautifully and I'm not gonna repeat everything you've just said there, but yes, that is a significant opportunity. And look, let's just be clear. No test is absolutely perfect. So we'll never be able to give people a hundred percent guarantee, but what we can do is to significantly improve the information we have. So one specific example of what you've just described is what we call minimal residual disease. So, you know, we may have removed a tumour with surgery, but there might still be some tumour that is invisible through our current imaging technologies. We know through work, that's been done with circulating DNA that we can detect those many, many months, potentially even a year or more before they're visible clinically. And that offers the opportunity to catch it and deal with it before it becomes a problem. And so this notion that you can monitor how things are going, either give reassurance to the patient or indeed say, "look, we think there's something still there, we need to give you a bit more treatment" is another really important opportunity for the field.

Kat: Given how far we've come in what feels like a very short length of time from understanding that, you know, the underlying genetics, the epigenetics of cancer to potentially having tests where we can take a blood sample and we can diagnose cancer, we can understand the mutations in it and select the right treatments, we can monitor it. How do you feel that this technology, access to this technology is going to shape and transform cancer care over the years to come?

Harpal: Well, I'll just pick up on the word that you just used, that this will transform cancer care. You know, whether it's this specific test or whether it's other tests that use the same sort of general concept, the notion that you can use blood tests that look at circulating DNA to understand whether people do or don't have cancer and how they're likely to do will totally transform cancer. I'm quite confident of that. And it will transform cancer in the near to medium term, actually, because it will mean that we can find cancers earlier when we know we're more likely to be able to treat them successfully, but also then that we can manage them better as they're going through their treatment. And all of that offers the opportunity for vastly improved outcomes from where we are today.

Kat: That’s Harpal Kumar, from GRAIL. And you can find out more about the NHS-Galleri trial for cancer screening that Harpal mentioned at nhs-galleri.org.

References:

• Ofman J et al. GRAIL and the quest for earlier multi-cancer detection. Nature Portfolio.

• NHS England. Screening and earlier diagnosis.

• NHS Galleri Trial. About the Trial.

• Jenkins H. The Galleri multi-cancer blood test: What you need to know. Cancer Research UK. 2021 September 13.

• Imyanitov E and Yanus G. Neoadjuvant therapy: theoretical, biological and medical consideration. Chinese Clinical Oncology; 2018 December 7(6).

• Demarco C. Secondary cancers: Why they occur and how to catch them early. MD Anderson. 2022 January, 6.

• American Cancer Society. Understanding Advanced Cancer, Metastatic Cancer, and Bone Metastasis.

• American Cancer Society. If Cancer Treatments Stop Working.