Anna Wilkins: Thank you so much for having me.
Andrea Miyahira: So first, can you just describe the CHHiP trial, the background, and its objectives?
Anna Wilkins: Yes, of course. So the CHHiP trial was a big trial testing different radiotherapy fractionation schedules, so a kind of seven-and-a-half-week schedule versus a four-week schedule, used for the curative treatment of prostate cancer. And it recruited just over 3200 men earlier this century. And it really has been a key trial in changing the global standard of care for curative radiotherapy for prostate cancer, such that it's now given over four weeks, rather than the longer, much more expensive seven and a half weeks.
Andrea Miyahira: Thank you. And could you also describe the multimodal AI biomarker that you were testing in this trial, and what your objectives were?
Anna Wilkins: Yes, of course. So the multimodal AI biomarker was developed by Artera, initially using a series of US RTOG trials, and it combines the images from H&E-stained diagnostic prostate biopsies, which are kind of bread and butter of prostate cancer diagnosis, age, T-stage, and the presenting PSA. And there is a combination of unsupervised and supervised machine learning to create this MMAI score.
Andrea Miyahira: Thank you. And could you describe the major findings and conclusions from your study?
Anna Wilkins: Yes, of course. So I suppose what we aim to do was give the multimodal AI test a tough challenge. And I think we were able to do that quite well with CHHiP, because we really, really wanted to say, "Does the MMAI test actually truly add value to what we already know?" And so we collected over 2000 biopsies in the CHHiP trial. We did that sort of more than 10 years ago, with no idea that any of this kind of research would be possible. So it's really exciting that this digital pathology AI thing has kind of come along as a whole new thing. The major findings really were that, when we compared the MMAI test results to how we currently risk-stratify localized prostate cancer, so whether we say this is a low, moderate, or high-risk tumor, we found using a number of different kind of gold-standard performance metrics that the MMAI test really added predictive value for how any individual patient's tumor will behave. The reason why kind of clinically this is important, I think, I would highlight two things.
So what we saw was that the MMAI test was particularly good at predicting the reasonably small proportion of men who sadly would recur quickly. And it picks out that proportion very clearly. So that's important, because these patients need more aggressive therapy, they need treatment intensification. But they're a minority and the treatments that we give have improved. So radical radiotherapy and then also surgery, which this trial wasn't testing, but they have improved, but they come with important side effects. And the second kind of clinically important result was that the MMAI test identified quite a large group of men with low-risk tumors. And that low-risk group is much bigger than what we see with our existing current tests. And that is important for men, because arguably these men could have less treatment. So for example, they probably don't need the hormone treatment that we give, which is the form of chemical castration and comes with important side effects. So I think using the MMAI test effectively in the clinic, we can intensify a small group of patients, and then treatment de-escalate quite a large group of patients.
Andrea Miyahira: Well, that's really exciting, and also very important for patients.
Anna Wilkins: This is a very common disease, localized prostate cancer is, and globally is projected to double by 2040. And I suppose the beauty, really, of the MMAI test is its pragmatism, and how simple it is to use. And so I think there's real global applicability.
Andrea Miyahira: So what do you think is necessary next for the clinical development of the MMAI test?
Anna Wilkins: So I think from the work we're doing, we're running a bit more robustness analysis, just to really augment what we've already done, prior to publishing our full results. We are debating how best to do a randomized trial using the MMAI test results. And initially that was quite a simple thing to plan. And then, because the retrospective data is so compelling, it's not totally clear whether an interventional trial is really needed. And I think that the MMAI test is likely to start coming into practice based on the strength of the data that we've already got, from both the US and the UK. And so CHHiP was very much a localized prostate cancer trial, but there are a series of other trials that the UK has performed in higher-risk, locally-advanced prostate cancer. And I think work from STAMPEDE and other studies has suggested that the MMAI test, although it was not initially developed in this more advanced prostate cancer, it can be really effective. So there's a lot of other important clinical questions that can be asked about addition or removal of other drugs, and radiotherapy dose escalation or de-escalation. So hopefully carry on using further UK trials for exciting AI digital pathology research.
Andrea Miyahira: Okay, wonderful. And how do you envision the use of this MMAI tool in the clinic to help patients have better outcomes?
Anna Wilkins: I hope that we will, before too long, be in a situation where every patient has a digital scan of their H&E biopsy, and the best possible AI algorithm is run quickly. And this forms part of the conversation between each patient and the relevant clinicians as to the best management approach for them. And this, in quite a lot of cases, can mean hopefully safely advising men that they don't need treatment at all, and we can just carry out active surveillance. So I hope this becomes a really routine part of every man's treatment. And I suppose part of that is because there is a lot of, in the UK, a lot of disparities nationally on the quality of care that men are able to access. And I think one of the benefits of the AI approaches is that it standardizes it, and it reduces a lot of that treatment disparity.
Andrea Miyahira: Thank you so much, Dr. Wilkins. It's really incredible to hear this progress, and congratulations again.
Anna Wilkins: Thank you very much.