MAST Trial Finds Metformin Ineffective in Preventing Prostate Cancer Progression - Neil Fleshner

June 26, 2024

Rashid Sayyid converses with Dr. Neil Fleshner about the outcomes of the MAST trial, which examined the effects of metformin on patients under active surveillance for prostate cancer. Despite promising preclinical evidence suggesting metformin's potential benefits against prostate cancer, the trial results were unexpectedly negative, showing no significant advantage of metformin in preventing cancer progression. Dr. Fleshner explains that the study, involving 407 men across Canada, did not confirm the hypothesized benefits of metformin, with some subgroups, especially patients with higher BMI, even showing potentially adverse outcomes. Despite these disappointing outcomes, Dr. Fleshner underscores the value of rigorous clinical trials in advancing medical knowledge and hints at future research possibilities using the trial’s comprehensive data sets to better understand prostate cancer progression.


Neil E. Fleshner, MD, MPH, FRCSC, Surgeon, Professor, Division of Urology, University of Toronto, POINT BioPharma, Toronto, ON

Rashid Sayyid, MD, MSc, Urologic Oncology Fellow, Division of Urology, University of Toronto, Toronto, ON

Read the Full Video Transcript

Rashid Sayyid: Hello everyone, and thank you for joining us today in this UroToday discussion. I'm Rashid Sayyid, a Society Urologic Oncology fellow at the University of Toronto, and I'm absolutely delighted to be joined by Dr. Neil Fleshner, who needs really no introduction to our audience. Dr. Fleshner is currently a professor of urology at the University of Toronto and has served previously as the division chair of the university for 10 years between 2012 and 2022.

Dr. Fleshner, thank you so much for joining us today.

Neil Fleshner: Pleasure to be here, Rashid.

Rashid Sayyid: Dr. Fleshner, we saw a great presentation by Dr. Anthony Joshua at the most recent ASCO meeting, where he presented the data from the MAST trial but looked at metformin in active surveillance patients. This is one of the unique studies in one of the trials in an active surveillance cohort, and we saw some very interesting if not unexpected results.

And so as one of the principal investigators yourself, we thought it'd be fantastic if you could go over the study results, explain the results that we saw, why we saw them, and maybe discuss what are some future directions with this study.

Neil Fleshner: It's a pleasure to present this work. This has been a large amount of work over the last 10 years getting this trial off the ground as the principal investigator and getting this through to completion.

What we did here is really exploit the quite considerable amount of preclinical evidence that suggested that metformin would have a benefit in prostate cancer. A lot of it generated from our labs, both our basic science as well as our epidemiologic studies, looking at diabetics treated with metformin in large populations. As well as understanding the metabolism of prostate cancer, which generally metabolizes glucose using oxidative phosphorylation, which is different than other cancers. Of course, metformin is generally an OXPHOS inhibitor.

There was lots of preclinical evidence to justify that this may work. We decided to look at this in the particular active surveillance population. This was a large effort. Many sites across Canada, 407 men. These are the men what we used to call ultra-low risk prostate cancer. They had to have Gleason six disease, one out of three cores or less involved, no more than 50% of length involvement of any of the cores, and a PSA less than 10. The men were randomized basically to 1700 milligrams daily of placebo, or metformin, and were followed for three years. Patients had to have a mandatory biopsy at 18 months and 36 months and progressions basically defined as any case that no longer fulfilled entry criteria. If you progressed to Gleason seven or higher, if you got more than 50% of a core, or if you had more than a third of cores involved, you were then deemed to have progressed.

This was the same design we used in the REDEEM study some years ago that we published in The Lancet. Participating centers: 12 sites. As you could see, it was a long effort. I think one of the interesting things perhaps for discussion is this study almost spanned the MRI era. MRI was not a requirement of the study. But as you can imagine, as time evolved, MRI was haphazardly used as practice patterns have changed. Many of the patients were recruited in the old, basically template, transrectal ultrasound-guided biopsy era.

You can see the baseline demographics of our patients, a typical surveillance cohort. Average age 62, vast majority T1c disease. PSAs in the five to six range. You can see that the basic volumes were in that range of about 40 CCs. That's prostate volume.

Here's really the punchline. If we look at some definitions first of all, pathologic progressions mean you have no longer fulfilled entry criteria. Of course, some patients on surveillance get fed up and want treatment despite not pathologically progressing. Of course, these days some can even pathologically progress and not want treatment. The therapeutic progressions include those men who want to go on and have treatment even though their biopsies are not particularly worse.

You can see here the main punchline on really the middle panel, I would say, which looks at the biology or the pathologic progression, there was no benefit to adding unfortunately metformin versus placebo in this particular patient population. There was a slightly higher amount of therapeutic progressions, P = .05. You can see it there. We're trying to explore that now to understand, and we have a lot of serially collected PROM data. We will have all of those to determine if there's a quality of life or some other reason to expect that, or also perhaps it was just a false positive. The overall, I think, basic punchline is unfortunately this hypothesis did not turn to be true, at least in this particular group of patients.

And then there were some other, I think, disturbing trends. One is looking at the BMI subgroups, and this was an a priori determined sub-analysis. The reason why we did this is it made logical sense that those men with more unfavorable metabolic states, those who are heavier as a raw measure, would actually benefit the most from metformin. That those perhaps who were slim didn't need its benefits from a metabolic point of view.

Of course, because science is often cruel for lack of a better term, we actually had a flip of the risk. You can see actually here that the men who were BMI 30 or higher actually had a higher rate of progression in terms of pathologic progression with the metformin over the placebo. For the longest time, I kept wondering if we had our groups misclassified in the dataset, but unfortunately, we didn't. This is the unexpected finding that actually there appeared to be, for lack of a better term, perhaps some harm to metformin in patients who have the higher BMI.

And then the other, again, somewhat surprising, but we could argue we've seen this before in all the five alpha reductase type trials, is there was also this bit of a disturbing trend in that the rates of high-grade disease, so Gleason eight or higher, were more prevalent in the metformin assigned group compared to the placebo assigned group. Again, the P-value is .08. It may just be a false positive, but certainly not something we'd like to see looking at these data. Again, it may just be a false positive, may be true biology. Of course, we'll probably never know. But nonetheless, these are, I would say, disturbing trends in the data.

Now, fortunately, I don't think there's much clinical implication of this. Remember we were taking non-diabetic men here and placing them on metformin, and I don't think there was ever an indication to give non-diabetic men metformin in the setting of surveillance because of the trial. And then I think conversely, I don't think there'd be a case to take a BMI-laden individual who's on surveillance off his metformin if he were a diabetic. I don't think the data would justify that at all. You can see the emphasis there.

I think the take-home message here is despite all the, if you will, tantalizing preclinical data and fairly strong scientific basis for its rationale, at least in this patient population, metformin did not prevent progression in patients with ultra-low risk prostate cancer to more severe disease. And then the disturbing trends, the exploratory subgroup analyses, again, high BMI patients seem to have a worse chance of progressing when assigned the exposure to the drug. Again, at least an asymmetric assignment of high-grade progression. By that, we mean Gleason eight or higher or ISUP four, five in those who receive the metformin.

Unfortunately, this is a classic example of the limitations of over-embracing, I think, phase two data or preclinical data or epidemiologic data as fact. I think because science is the search for truth, this would tell me that obviously randomized trials are so important and they should have a priori hypotheses. I mean, that's really the only way we can determine cause and effect. Of all the confounding and all the other work we've done, of course, you can never control for unrecognized confounding, unless you put it up against this type of study design, and hence the results.

Rashid Sayyid: Absolutely.

I want to take a second here and congratulate the whole team on this effort. Although at face value this may seem as a negative trial, we didn't learn a lot, but there's a lot to be taken from this study. The amount of effort that goes into, and the time as you touched upon before over a decade in order to conceive this study is no mean feat, especially in a population such as the active surveillance cohort.

Neil Fleshner: I think a good message, particularly for young individuals wanting to start their career, is the clinical trial business is a long and arduous process, right? I mean, you get an idea, you've got to pitch it, you've got to get a grant that usually takes two years. You then have to accrue patients to your trial, and then you have to have time evaporate for the events to occur. Not to mention we had COVID during all this, which largely shut a lot of it down for a couple of years in terms, certainly, of accrual.

I think this is not to discourage young investigators. I think science, again, is the pursuit of truth.

I think the other goldmine, at least in this, is this is now going to be one of the most well-characterized cohorts of surveillance, at least for low-risk disease. We have baseline biopsy samples. We have baseline serum samples. We have a pile of epidemiologic data that I think will help us really develop a better understanding of who progresses on surveillance and who doesn't.

We already have just had a large grant funded looking at interrogating some of the baseline bloods from, in this case, a proteomic approach and as well as an AI approach. We will utilize these data well.

Rashid Sayyid: Yes, you touched upon this, Dr. Fleshner, very briefly. Although this is a negative trial, there are some interesting opportunities that come from having this data set. I know that the team already is moving on to the next steps in terms of blood-based tests, also working with AI teams, genomic tests, etc., in order to better understand who progresses and why, etc.

And so can you touch upon that a bit more about some of the ideas the team has in order to encourage our active surveillance patient? This isn't the end of the road when it comes to the metformin idea, but this is just the start for them.

Neil Fleshner: Yeah, no, I think, look, I think we're going to interrogate the blood samples. Obviously, the technology for interrogating in a multi-omic way biofluids, tissue, which we have, unstained slides on all patients from all centers, as well as today just image from an AI and ML point of view, really gives us the ability now to have such a tight data set. Obviously, the data from randomized trials is always cleaner than those from in-house cohorts because of the meticulousness and oversight in terms of data acquisition, data entry. It is, we feel, a little goldmine that will help us.

I think one of maybe the limitations of this is it is limited to Gleason six disease, low volume disease. Unfortunately, the reality is we don't really see it that often anymore. The only time we really see, I would say, these patients are probably those who have a PI-RADS-3, four, rarely a five lesion that gets biopsied and it's negative. But you pick up a little six elsewhere that's... We don't see these patients as much anymore. But I like to believe that whatever causes progression in these patients probably also could be translatable to others.

Rashid Sayyid: Yeah, absolutely.

Going off a bit on a tangent here, do you think this is the end of the story for metformin in the prostate cancer space? Do you think there's still some potential for metformin in the advanced prostate cancer space, such as patients on ADT, cancer resistant? Where do you see the story going with metformin in prostate cancer?

Neil Fleshner: That's a great question. There is a lot of also evidence from trials, particularly in the CRPC, HSPC space, some of the VA data out of the US suggesting metformin may benefit men with more advanced disease, and it may.

Remember, as prostate cancer gets more aggressive, it often undergoes a metabolic switch to more traditional glycolytic type of metabolism. And that's why sometimes you get the FDG avid PET scanning. There is the UK STAMPEDE arm, which includes metformin, will actually probably report this fall.

I think that's going to be the final story, either good or bad, because we'll have an early stage and a late stage trial. I think if that's the case, we're going to have to wait. It won't be long. We'll eagerly await those data. If they're negative, I don't think that means we shouldn't try to exploit some of the unique metabolic properties of prostate cancer and try to perturb it as a therapeutic strategy, but certainly an easy, off-the-shelf, cheap, which was really exciting, demonstrated safety profile of this approach for decades would've been a real home run for the field. But unfortunately, truth gets in the way, which is just great because that's what we're here for.

Rashid Sayyid: Absolutely. Again, congratulations on this fantastic study. I thank you for taking the time today.

Do you have any final words for our audience?

Neil Fleshner: I think the main one, again, is if a patient comes to you and says, "Hey," and I've had many over the years, "I think you should give me metformin because I read that it slows down cancer growth..." This is not just metformin. It could be any drug that there's a bit of literature. It could be a natural product. It could be anything else. I just encourage you to resist doing that because you may cause harm. I think that's why we have this high bar called randomized trials for cause and effect.

Please don't write off-label prescriptions for medications that only appear to be good in a certain level of evidence, I would say.

Rashid Sayyid: Absolutely.

Well, Dr. Fleshner, again, thank you so much for taking the time today. Thank our viewers as well for taking the time to listen to our discussion today.

Neil Fleshner: Pleasure. Thanks, Rashid. Thank you. Have a great one.