The Uromigos Episode 135: The STAR Study

Dr. Janet Brown discusses intermittent VEGF TKI therapy in the context of the STAR clinical trial.

Episode Transcript

Tom:
Hey everybody, we’re joined here with Janet Brown, who today is going to present The STAR study as part of our ESMO series. STAR explored intermittent versus continuous sunitinib in a randomized phase 3 non-inferiority trial. Janet, do you want to introduce yourself, to have a quick issue around any conflicts of interest, and then dive in and tell us a bit about just the top line of what the study showed.

Janet Brown:
Okay. My name is Janet Brown and I’m a medical oncologist and a professor at Sheffield in the UK. I’m the CI on the phase two/three STAR trial, which is what we’re discussing today in terms of particularly the phase 3 data. I’ve been involved as a consultant in treatment of kidney cancer for the last 16 years.

So, my conflicts, really, are research grants from Amgen and Bayer, and personal fees from Amgen, Novartis, BMS, Ipsen, Sanzari, MSD, Bayer, and conference registrations from it Ipsen. So that’s everything. So that bit’s ticked off, so we can move on to the study.

So really where this came from is, and we’ve got Brian who’s on the line as well who’s also been involved in an intermittent study, and I think he’s about to introduce himself, so I’ll let him do that. And then I’ll maybe talk a little bit about things study. Or do you want me to do that now?

Brian:
Yeah, go ahead and dive in of a rationale on the basic study design.

Janet Brown:
Okay. So, where this really came from is, once TKIs became standard therapy, many of us were picking up that, particularly the patients who had perhaps been on them for a long time and maybe a period of years rather than months, some of them were asking to have breaks. They might be going on holiday. There might be a reason for them to stop treatment.

And certainly, I noticed with some of mine where they had things like leg infections and they were off it, and it was healed, they didn’t seem to come to any harm with the treatment breaks. And, and I think we were all building up cases like this. So, I know there’s been a range of ways of approaching this.

What we decided to do in the UK is, we’ve got something called NIHR, which tend to fund the trials that perhaps pharma don’t want to do, where it might be looking at less drug and treatment breaks and treatment holidays. And we were able with the national studies group in the UK to put forward a trial design, which involved answering the question, “Do we need to give TKIs continuously?”

And this was really initially with sunitinib, and then when selpercatinib came online, we needed to offer both drugs within the trial, so we adapted it for either drug. And the question really is, rather than giving these drugs continuously till progression radiologically, could we give a fixed amount of treatment and then a treatment break in the experimental arm, which would hopefully give less toxicity, cost savings, and not be too detrimental in terms of survival, compared to continuous.

So that was really the question we were wanting to answer, and I’m sure Brian, and obviously Tom will come in. When this trial was actually discussed and designed in 2009, 2010, there was actually quite a lot of concern at that time with clinicians about treatment breaks. So really the phase 2 part was looking at, could we recruit a safety signal in terms of time to strategy failure that these patients who were coming off weren’t all immediately progressing and having problems. And this was a run in for the numbers counted to a phase three.

So, what was interesting about this study is it was a non-inferiority trial. So, it was powered not only to look at survival, but also qualities as in the co-primary endpoint. And so that was felt to be quite novel for health service resources, and that was what was quite unusual about this trial. So, I think I’ve probably said enough about that as a sort of outline.

Brian:
So, talk about some of the mechanics. So, patients were randomized at baseline …

Janet Brown:
Yeah.

Brian:
… and then they got 6 months of treatment and then went on the assignment, so to speak?

Janet Brown:
Yeah. So, there was a lot of discussion with this trial whether patients should be randomized at baseline or at the time they were going to take up a treatment break. It was felt clinically and statistically from the national groups that it was cleaner to randomize it at baseline and prepare patients to take up that randomization.

So, it was actually done at baseline. Similarly, nobody quite knew how much treatment to give before a treatment break. We had discussions about whether it should be 2 cycles of treatment, and we talk about cycle as 6 weeks, so 3 months, or whether it should be 3 cycles or 4 cycles.

So, there was quite a lot of, and I’m sure everyone’s used to this, discussion amongst the various experts. And we came up with the pragmatic 4 cycles of treatment at, at 24 weeks. So partly, therefore, there was an intermediate scan to look at for response, but also that seemed a reasonable amount of time to then stop treatment.

Brian:
Yeah. I agree. Then the criteria to go on to the assignment, obviously a patient’s progressed at 2 months or something, they came off. But were there other criteria at that four-cycle mark?

Janet Brown:
Yeah. So, 2 of the other criteria obviously were if they’d already gone down the standard dose reductions for selpercatinib or sunitinib, you know, there’s 2 dose reductions allowed, then clearly if they would be coming off treatment as well, whether they’d progressed or not. So, they weren’t included in that remaining number.

The other thing is, we had to pragmatically make a decision about if patients had a long time off treatment, whether they should still be included. Because it probably mattered if you had more than a cycle off, treatment because you had a toxicity or a complication you admitted. So basically, we had a 28-to-six-week window. And if patients didn’t resume their treatment allocation in that time, they didn’t also proceed past the six-month end point. So, they were the major ones that we obviously looked at.

And I think the good thing is, although this trial was done a while ago, I’m now quite happy about that in terms of the randomization, because it avoided all of these, being in a COVID situation, because our patients were randomized with the phase 2 starting in 2012, and we finished, in 2017, our recruitment.

Brian:
And about half the patients made it to the continuous versus intermittent. Is that right?

Janet Brown:
Yeah.

Brian:
Roughly?

Janet Brown:
So, just over half. So, our criteria, 1 of the extra criteria from the phase two, apart from recruitment numbers and sharing an efficacy signal, was that at least 50% had to get to the six-month mark. And maybe you’ll comment on this, but I was actually initially quite surprised at that. But if you look at standard populations, we think about patients that stick in our mind, who’ve maybe had 4 lines of treatment, but actually quite a lot of them are dropping out on first-line treatment within the first 6 months.

Brian:
Yeah. And we had the same experience in our small phase two. I think it was like 56% of the initially enrolled patients made it too. We had, I think, the same criteria, 4 cycles and stable disease or better and lack of toxicity went on to intermittent.

Tom:
Janet …

Janet Brown:
Yeah?

Tom:
Janet, so 50% of the patients stop therapy after approximately 6 months, providing they’ve got a tailwind and things are going okay, they stopped therapy. How do they then restart therapy?

Janet Brown:
So basically, what we did in this study is, if they had clinical progression or radiological progression, and the vast majority were a radiological progression, so they had regular scans every 10 to 12 weeks. And when they progressed, they went back on treatment [crosstalk 00:08:45]

Tom:
How did you define progression?

Janet Brown:
So, in this particular trial, we did it slightly differently. Because in the phase two, we use resist reporting just to get a handle for the 250 in the phase two, on the number of patients who had an ongoing partial response at 6 months. And it was only about 3 in the study. So, after that, it was felt reasonable to do local reporting, according to resist, but to record the actual measurements.

But it was felt pragmatic that we would use resist progression as the criteria for restarting treatment. And obviously we were able to look at the 250 patients from the phase 2 to check that this was a reasonable strategy. I know with Brian’s single center, and it was quite nice to be able to do, I think you looked at when it had gone an increase of 10%.

Brian:
That’s right, yeah.

Janet Brown:
Obviously with this one, at the time we started it, we looked at the central reporting. We’ will have to look at later the tumor burden measurements, but we felt pragmatically for a 60-center study in the UK, which I think importantly include quite a number of small sites that wouldn’t normally take place in early phase renal trials, we needed something that was understood by the majority of individuals. So, we had the run into the phase 2 to look at the central reporting with more measurements, and then for the phase two, we also did checking of the measurements.

I think the important thing was, obviously when you go back on treatment, and Brian, you probably know this from your trial, you have to have a new baseline as to which to report it to. So that also had to be very carefully designed within the CRS for this study, because you’re not just reporting it to the original baseline, you’re reporting it to the new baseline when the people in the intermittent arm have gone back on treatment. [crosstalk 00:10:38]

Brian:
So, 50% of patients stopped therapy in the study arm. What proportion of those patients were able to go back and restart therapy?

Janet Brown:
So, the majority of those patients were able to go back and restart therapy. The overall was 1 treatment break, but the way the study was run, we only mandated one because at the time the trial was run, clinicians, some of them were not comfortable doing multiple breaks. And obviously we’d change that if we were doing it now.

But interestingly around 42-43% actually had multiple breaks, and the highest number was nine. And I think that will be an interesting group to look at. Can we predict the ones that are going to benefit the most from the breaks and have frequent breaks?

Brian:
Janet, when you say that, I’m sorry to interrupt. When you say that the overall break was one, does that include all the patients, say, who progressed at 2 cycles or 4 cycles? Is that being that the whole population, or just the, those randomized to intermittent, those who’ve made it to the intermittent phase?

Janet Brown:
Yeah. So, if you look at the ones that made it to the intermittent phase at 24 weeks, the average break was 1 break, but it’s actually nearer 2 in terms of statistical analysis. But if you look at those numbers, 42% actually had more than 1 break.

Brian:
Okay.

Janet Brown:
What we don’t know is that some, because we mandated one, some centers interpreted that, “Right, we’ve done what the trial says, they’ve had 1 break. We’ll not give them another one,” and others, particularly the centers that got used to the trial and patients obviously were liking the breaks, offered more breaks. So, what we do know for future trials with TKIs or immunotherapy is we could probably in the future mandate if they’re stable again, they have to go on another break. But for this, it was a step too far.

Tom:
Janet let’s do the [inaudible 00:12:37]. You recruited over a thousand patients into the study in 60 centers. There were primary end points with a non-inferiority efficacy endpoint. There was also a second co-primary endpoint associated quality of life. What did your primary end point show?

Janet Brown:
So, there was 920 patients actually went into the study, which was fairly matched between the continuous and the intermittent arm. So, the overall primary analysis showed that for the overall survival, the graph actually looks completely overlapping. But that the non-non-inferiority was met for the ITT population, but narrowly missed for the per protocol population.

For the quality of life, non-inferiority was made for both the ITT and the per patient population. But looking at the graphs, they’re pretty much superimposed on another.

Initially we went for an 80% power. It was a good thing that some of these patients actually survive longer than expected, possibly because of other secondary treatments, and so the event rates were slightly lower than expected. So, our power is about 77% rather than 80%. So, whether that had an input into that …

Tom:
Janet, I’m just going to summarize that if I may. So essentially, there was an attempt to look at non-inferiority. You set quite strict limits of non-inferiority, including the need to hit it in both the ITT and the per protocol population. You did not meet that. And therefore, you’re not able to say statistically, for the purists in the room, it’s not inferior. Is that correct?

Janet Brown:
Yes. I think for the purist in the room and the statisticians, you need both to absolutely conclude non-inferiority. But as a clinician, as we all are, if you look at those graphs, you could trace one on top of the other. And I think the other thing is, as 1 of the secondary things was looking at the cost benefit, and there is a significant cost saving in the intermittent arm.

So, I think you have to take all of those into consideration. So, I what think our trial showed is that patients and clinicians were willing to tolerate a treatment break, and there didn’t seem to be a major effect on overall survival, and that there was actually cost savings and less TKI-related toxicity in the intermittent arm.

Brian:
Janet, I’ve I have 2 questions. One is the non-inferior non-inferiority results you just mentioned. Those are for the entire population, not just the patients who’ve made it to the random assignment, is that correct?

Janet Brown:
Yes, because my understanding, and again, my stats are a little bit simplistic in a medical oncology person, is that if you randomize at baseline, that’s based on the baseline numbers.

Brian:
Yeah.

Janet Brown:
Because it’s the intention to treat.

Brian:
Sure. And do you think because of that, that may have affected your outcome? I.e., half the patients were treated identically, they never even made it to the random assignment. And even if you made it to intermittent, you only got an average of 1 break. So that the arms really weren’t that different. Do you think that maybe impacted the results?

Janet Brown:
I think it’s always difficult. And we had a lot of debate whether to randomize it at baseline or so many cycles into treatment. One of the factors is, we wanted to pick a representative population for RCC rather than a group who’d done well just with the first cycles of treatment, because we wanted to know how general applicable this was across different sorts of scores and performances and things like that. So, I think we certainly have done that.

Although the average was only 1 break, I think we mentioned because it was mandated, but 42%, is a significant proportion who had more than that. I think one of the things that will be useful to tease out of the trial is, can we predict early on who these individuals are, based on their criteria, who are most likely to be at the point to take up the break and benefit from the break? Because that’s really what we’d all like to be able to do, isn’t it.

Tom:
Janet, what actually was the hazard ratio for survival, if you look at the answers?

Janet Brown:
So, if we’re looking at the hazard ratio, for the ITT population, it was 0.87, not 0.83, and 1.12 were the intervals. And the PP was 0.94 with the interval 0.8, and 1.09. So, it was the 0.8 was met, so it’s that, yeah. So that was …

Tom:
So, 13% or 6% worse, depending on the population, and not able to say non-inferiority, but on 1 of those 2 populations. Let’s just talk a little bit about progression-free survival and time to treatment strategy failure.

Janet Brown:
Yeah?

Tom:
They’re different end points. The time of treatment strategy failure was when you came off sunitinib and started something else. And that was better in the study arm, but I guess that was because you allowed progression in that arm. But the progression-free survival was less good because patients who stopped therapy progressed faster. Is that correct?

Janet Brown:
Yes, cause by definition to go back on treatment on the intermittent arm, you had to have progressed according to resist. So, it was pre-specified in this trial that PFS wasn’t going to be a very good measure, if you like, to time to event.

So that’s why we used strategy failure, which is progression deaths, new systemic therapy, as well as summative progression-free interval, which is the number of periods in the treatment breaks where it’s progression free. So, I know it’s slightly more complicated than a normal PFS trial, but a PFS doesn’t [inaudible 00:18:50] information.

Brian:
I think what’ll be interesting is when you have all the tumor measurements. And if your time to failure is later, which it inherently is when you allow progression in the arm, did those patients end up with a greater tumor burden when they started their next therapy? Right? I think that’s what Tom’s getting at is, when you have those tumor burden measurements, you can look at that level of detail.

Janet Brown:
Yes.

Brian:
I know you don’t have it now, but that’ll be interesting.

Janet Brown:
Yes. So certainly, for the phase two, we’ve got that for all sites of tumor, because they were all centrally reported. For the phase three, we’ve got the target lesion measurements.

Brian:
That’s still pretty good.

Tom:
Yeah.

Janet Brown:
Yeah, the first 250 we would have, yeah.

Brian:
Janet, can I just …

Tom:
Sorry, you go, Brian, then me.

Brian:
I was going to ask a question about toxicity. So, as I remember in the presentation, there was significantly less, I don’t remember if it was SAEs or some toxicity measurement in the intermittent arm, which of course you would suspect. So, could you just go over those results? And then if that’s the case, and this is really the fundamental question is, why did that not translate into a quality-of-life advantage?

Janet Brown:
So, I think there’s 2 things. So, what we’ve presented in the slides is, you’re right, there is less events attributable to the tyrosine kinase inhibitors. But what we perhaps, and hindsight’s a wonderful thing, is that the patients in the intermittent arm were on the strategy for longer.

So, things that are events that are secondary to their cancer, you’re actually getting them for longer than in the CCS arm because they have had less time on strategy, and they’re only then on the survival part of the trial, if that makes sense.

Brian:
Mm-hmm (affirmative).

Janet Brown:
So, it’s actually quite a difficult scenario. I don’t know if you had the same, Brian, with yours, but there’s less TKI-related toxicity, but actually quite difficult to divide any adverse event or serious adverse event that might be non-drug related by the amount of time you’re on the trial.

Tom:
Janet, the pessimists in the room might say that if you can’t confidently say that the efficacy is not worse, and you can’t confidently show that the combination is better tolerated from a quality-of-life perspective, it’s not a strategy we should be pursuing in the future. What would your come-back to that be?

Janet Brown:
So, I would say this is a large trial. The survival curves are pretty much super imposed. Certainly, that those of us who were involved in putting many of the patients in on the ground saw significant numbers of patients who greatly benefited from that extra break of treatment. Quality of life obviously was assessed every 6 weeks, and it’s quite subtle, sometimes, to pick up some of those changes. The strategy, we feel from a safety point of view, the overall survival, looking at those curves, isn’t really any different.

We feel that certainly because a significant proportion of those in the DFS had more than 1 break, they must have liked it, because it wasn’t just up to them, it was also up to the treating clinician. And we also feel that there is significant health economic savings. So, I think that I would argue it the other way round, if that makes sense?

Tom:
Brian, what’s your take on it?

Brian:
I agree with Janet. I think what we learned from this trial is that it’s clearly not a strategy for every single patient, right? There’s going to be patients who progress early, have toxicity, can’t go on treatment. Right? All those were excluded here. There’s going to be patients where their natural biology just doesn’t allow a break of this magnitude.

Certainly, doesn’t allow a break until resist PD. But I don’t think it means that we should never be giving any breaks with TKIs. I don’t want people to come home with that message. And maybe for some patients, I think the average break was about 12 weeks if I’m right, Janet?

Janet Brown:
Yes.

Brian:
So that’s a pretty long period of time for most patients. In our study it was about 8 weeks, which is still a pretty decent period of time. But I think taking small breaks, whether it’s 3 days or 2 weeks or 4 weeks, I think depending on a patient’s biology is absolutely okay. And it’s just knowing the patient and knowing their tox and knowing their biology. But to think you have to take pills every day I don’t think makes any sense whatsoever.

Tom:
I’m going to ask you both have a really direct question. So, let’s say we have a 65-year-old individual who has liver and bone metastasis. They’ve been on therapy for 6 months. They have stable disease, and they fulfill the eligibility criteria for this trial. Are you comfortable stopping therapy on this patient for up to 6 months?

Brian:
So, can I ask a couple of questions?

Tom:
Yes.

Brian:
So stable disease with tumor shrinkage? Or stable disease with tumor growth?

Tom:
20% shrinkage.

Brian:
20% shrinkage.

Tom:
Yes.

Brian:
And when you say, “Are you comfortable stopping for 6 months”?

Tom:
Yes?

Brian:
Well, I wouldn’t stop for 6 months.

Tom:
Okay. So how long would you?

Brian:
I’d probably stop for 6 or 8 weeks and re-scan them. I think your liver and bone are nastier sites, right? They’re generally more aggressive sites. So, I think you’d probably do a short interval scan.

Tom:
And does this data support you with doing that, Brian?

Brian:
I don’t know if this data does, but our other trial and my personal experience does.

Tom:
And Janet, what’s your answer to that question? How often would you be scanning these patients? Would you be stopping regularly? How would you approach this?

Janet Brown:
So similarly, if they’re stable at 6 months, I would want to be scanning them at least every 3 months. The other thing with this study is, we tried to make it pragmatic and something that could be used in routine practice. What we didn’t want is to increase the scan frequency, so it was every 3 weeks, or once a month.

So, I feel on the experience of looking at all the data that’s coming to me [inaudible 00:24:57], I would feel confident. And obviously it’s a discussion with the patient, as it always is, to stop at 6 months if things were stable. And obviously I would be keeping an eye on the scans regularly to assess if I needed to restart treatment at any stage.

Brian:
And Tom, I would add 1 thing. I think it entirely also depends on the patient’s toxicity. So, when patients, as you well know, get to 4 to 6 months, they’ve kind of settled into what tox they’re going to have, management strategy, supportive care. And for some patients, they’re doing pretty well and there’s less motivation to stop. I don’t think we can say there are disease-controlling benefits to stopping. Whereas other patients are having a tougher time. Just like with nivo maintenance, or something. I think it’s the same conversation.

Tom:
So, let’s say this person’s on, I’m going to [inaudible 00:25:44] question after this, but there’s this person who’s doing quite well on therapy on 37.5, happy, stable. The doubter in the room might come back and say, “Well, hold on a second, the survival data, the hazard ratios are less than one, so it’s somewhere between 6 and 13% worse, depending on that analysis, number one. And it looks like once you’re established and stable, the quality of life isn’t a great deal better. Why takes the risk?”

Brian:
So, fine. Keep them on drug.

Tom:
Janet, what’s your answer to that question?

Janet Brown:
Yeah, I don’t feel that that’s my experience on the ground. I always talked to the particular patients about it. I think even the overall environment has changed, and I actually get more people asking for treatment breaks than we ever did before. I don’t know where some of that’s also partly due to COVID and other issues, and peoples change in take on life.

But certainly, most of my patients in that situation would quite likely ask for a break in treatment. Some of the toxicities can be quite subtle, but generally they would like a treatment break, and they’re happy enough to be followed up on scan. I very rarely get the situation where they say, I don’t want to come off treatment, and whatever. So that’s not something I seem to get that very often at the moment. But that might just be my practice.

Brian:
Yeah, I agree.

Janet Brown:
And the number of patients in this trial, I think is also very, very large. And I think the other thing is, there were slightly less events. And looking at those curves and looking at clinical presentations, they pretty much overlap. And if I was showing that to my patient and saying, what would you do if you have a treatment, break, this is what your curve looks like, I think they would agree with that.

Tom:
So, what I’m hearing from both of you, it swings the dial perhaps a little bit towards, “It’s okay to have treatment breaks.” But the detail of how long and which patients is uncertain, and it’s certainly not a black and white issue.

Brian:
Yeah.

Tom:
My next question is, of course, this was-

Brian:
Hold on. I just want to make a couple of comments on what Janet said. Patients love being off drug, right? And they definitely feel better off drug. There’s absolutely no question about it, that a patient off drug feels better than a patient on drug. And to me, it just reinforces how poor our quality-of-life instruments are that they can’t capture that. I don’t think that’s anxiety of being off drug. I don’t think it’s any of that. And again, we’ll find out when we get some more detail.

Tom:
I’m super interested in the subset analysis of that quality-of-life data. The next question I wanted you to ask, which I think is important, it’s our last question is, that we don’t use sunitinib or selpercatinib very much anymore in the frontline setting. What does this actually mean in 2021?

Janet Brown:
Shall I start with that? Are you okay?

Brian:
Sure.

Janet Brown:
So, I think there are some patients, obviously, who still do have those agents. I think this trial has also been important though as an exemplar trial with this kind of approach. And I think it’s led to, even on the newer therapies, of questions being addressed about intermittent strategies and frequency of those kinds of treatments. I guess the other question is, in the IO/TKI combination setting, there are trials that are going on where there might be stopping the immunotherapy.

Would this increase the chance of then also stopping the TKI? Because if that was causing the residual, most of the toxicity, the patients aren’t really getting a break from treatment. So, I think that would be 1 thing. And the second is, a lot of these patients will still get a TKI second line if they’ve had it first line in combination, or they’ve had immunotherapy. So, I think it’s still relevant. The amount of TKI being used is still considerable.

Tom:
Brian, over to you.

Brian:
Yeah, I agree. I think in the refractory setting, the approach is still valid, although maybe with a little extra caution, simply because those patients have higher tumor burden. They’re the ones not controlled or cured with frontline therapy. So, I still use it in the second line or later settings, but maybe with even stricter criteria. And I totally agree for IO/TKI. Giving these drugs forever makes no sense.

I’d love to see studies that stop TKI early. I’m not sure we’ll ever see them, but I’m convinced the TKI is more important upfront. And I think, especially as we get to doublets, triplets, that being able to hold 1 drug, whether it’s for a short or longer period, will be important.

Tom:
Janet, this is a terrific effort. It’s an investigator-initiated thousand patient trial. They don’t come along very often, and you and the team need to be congratulated for that. I also think that the rigor of your non-inferiority, it may have worked against you a little bit, but I think that rigor also needs to be respected. I think that it’s obviously a huge and fantastic effort. And thank you very much for joining us today. Brian, have you got any last comments?

Brian:
No, thanks Janet. And again, congratulations. It’s a big effort, and I think it’s really important for the field

Janet Brown:
Thank you very much. And just also to thank all the patients and their carers who’ve been involved in the study. I think 1 thing it has done is, it’s opened up smaller sites, certainly within the UK, because it was quite a pragmatic study to doing renal trials in the future, which has been really useful. And thank you for all the sites and the trials unit [inaudible 00:31:21] much.

Tom:
Janet, thank you.

Brian:
Thanks, Jan. See you soon.