Dr. Alan H. Bryce, MD presented “Adjuvant Sunitinib for High Risk Renal Cell Carcinoma: Not Ready for Prime Time” at the 26th Annual Perspectives in Urology: Point-Counterpoint, November 11, 2017 in Scottsdale, AZ

How to cite: Bryce, Alan H. “Adjuvant Sunitinib for High Risk Renal Cell Carcinoma: Not Ready for Prime Time” November 11, 2017. Accessed Mar 2024. https://grandroundsinurology.com/adjuvant-sunitinib-for-high-risk-renal-cell-carcinoma/

Summary:

Dr. Alan H. Bryce, MD, argues that there is not yet enough substantive data and research to justify the FDA approving using adjuvant Sunitinib to treat high risk renal cell carcinoma.

Adjuvant Sunitinib for High Risk Renal Cell Carcinoma: Not Ready for Prime Time

Transcript:

Well, I guess I’m not unfamiliar with debating myself, so I’ll go ahead and talk to you about adjuvant therapy in renal cell carcinoma, adjuvant Sunitinib and whether or not, whether or not this should be approved.  This is a timely topic.  If you are not aware, this has recently been debated at ODAC because there is a positive study out there, and the question is are we going to see an FDA label change or not?  I’m going to speak to why I think we shouldn’t see a label change, and even though Dr. Crawford looked far and wide for someone to debate me, everyone was intimidated to the extent that no one took up the offer.  So ultimately vote for me at the end please.  Okay.  

Here is the background.  About 15 to 20% of all RCC patients will present with high-risk, locoregional disease, so stage 3 disease, where the five-year overall survival will be 50 to 60%, and you all know this.  I mean you see these patients.  We have therapy that has proven to extend overall survival, and we have therapy that’s proven to extend overall survival in the metastatic population so VEGF TKI, sunitinib, pazopanib, cabozantinib, Axitinib, sorafenib, all with proven overall survival advantage.  So in oncology the standard paradigm in advanced disease is prove a drug in the metastatic setting usually last line, and then start moving it forward, first-line therapy, adjuvant therapy, neoadjuvant therapy, with the general believe that if something is beneficial in the metastatic setting it should be beneficial in the adjuvant setting, and of course the point of adjuvant therapy is to improve cure rates, and it is conceptually what we are trying to do is treat micrometastatic disease, to eradicate that last little bit that wasn’t removed surgically and improve the cure rate in doing so.  

And this is a well-established paradigm in any number of malignancies, colon cancer, breast cancer, lung cancer, this is all standard of care.  Having said that, the fact of the matter is that it is not automatically true that what works in the metastatic setting will necessarily work in the adjuvant setting, and we have seen this with multiple drugs over time, adjuvant therapy studies that fail despite very effective drugs for the metastatic setting.  So we cannot extrapolate from metastatic care to adjuvant care.  We have to do the studies, and prove the point.  It would be ideal if we had some kind of retrospective analysis that tells us what is the criteria to let you know what is going to work and what is not, but so far that hasn’t played out, so what we have in our data set here is two large randomized, phase 3 studies of adjuvant sunitinib, one of them also has sorafenib, one of the studies that is, totaling over 1900 patients that have been completed.  

We have the ASSURE study, ECOG2805, over 1900 patients high-grade RCC you can see the criteria there, so post-nephrectomy and these patients were randomized on a three-arm study between sunitinib, sorafenib and placebo, and then we have the S-TRAC study, this was the industry phase 3.  Slightly higher risk population than ASSURE, only stage 3 patients post nephrectomy, 1:1 randomization sunitinib versus placebo.  These are not blinded studies in the sense that with the toxicity of these therapies you know exactly which arm the patient is on.  The primary endpoint of both studies was progression-free survival.  ASSURE was investigator assessed.  S-TRAC had central review.  So PFS was the primary endpoint, not overall survival.  

The results of these studies, looking at them side by side on an intent to treat basis in assure you had 647 patients on sunitinib in S-TRAC you had 309.  Disease free survival was roughly similar, somewhere in this 5.5 to 7-year range, 5-year disease free survival like you see there fairly similar 55% in ASSURE with adjuvant sunitinib, 59% with S-TRAC even though that was a slightly higher-risk population, and you see that the hazard ratio for disease-free survival was negative.  In fact in ASSURE the survival curves for the three arms overlapped, no space between them whatsoever, but in S-TRAC disease-free survival was statistically better on the sunitinib arm than on the placebo arm so the hazard ratio there is 0.76, confidence interval as you can see remains less than 1, so this is certainly a valid, statistically significant result, P value of 0.03, but then you look at overall survival here, and five-year overall survival in ASSURE, roughly around 80%, here on S-TRAC really no difference whatsoever between sunitinib and placebo.  Now, on S-TRAC the final overall survival follow-up is not scheduled until 2019, so this data may yet mature, but what we can say at the moment is with available data we have a progression-free survival advantage, no overall survival advantage, and the debate then, and this is a debate that played out at ODAC, is that PFS has been accepted as an endpoint in other malignancies.  In the adjuvant setting, PFS is accepted for some cancers.  Shouldn’t it be accepted for the RCC population?  That is the central debate.  Of course when we had multiple studies, we like to put them together in meta analyses, and I’m just showing you the sunitinib data, there is also a pazopanib study, but it largely tracks what we are talking about, and this meta-analysis the things that pop out, I mean you look here, here is the composite hazard ratio, okay, for progression free survival, 0.89, it crosses 1.  I mean this is certainly not statistically significant okay, maybe it starts to favor sunitinib a little bit, but you’re talking about 2500 patients ultimately here.  This is plenty of data to try and reach a conclusion.  

The toxicity certainly matters.  You are talking about adjuvant therapy, so the patients who otherwise would not go on any therapy and sunitinib carries a grade 3/4 toxicity rate over the period of treatment, over 60% with four sunitinib related deaths.  Okay, so sunitinib the TKIs are pills, and I think one of the myths we often deal with in oncology is people think, well, if it is a pill, it’s no big deal, it’s not chemo.  Chemo is IV.  If it’s a pill, it’s not chemo, and really the first comment about that is number one, chemotherapy of course is not a scientific term, doesn’t actually mean anything at all.  It’s a colloquial term, and mostly we take it to mean things that make you feel sick.  Sunitinib in terms of its toxicities I would say is right on part with chemotherapy.  Now, one of the problems we have when we talk about toxicity is the common toxicity criteria were really developed to talk about the toxicity of therapy.  That is acute events, so the CTC really look at how severe a toxicity is at its worst.  It does not look at duration of toxicity.  So if a patient gets hospitalized with neutropenic fever and pneumonia, is in the hospital five days, on oxygen gets IV antibiotics, it’s a grade 3 neutropenic infection, okay, and if a patient has 5 to 6 stools a day over baseline, diarrheal stools a day, that is a grade 2 toxicity.  

So the question is would you rather get chemotherapy and get one grade 3 neutropenic fever for five days or would you rather have a year of 5 to 6 stools diarrheal stools a day on sunitinib?  I would take the hospitalization.  Right?  Have it and be done with it.  This is a problem with how we categorize toxicity.  Chronicity is left out of the consideration.  Yes, regardless of that there are always going to be a few treatment-related mortalities because these are not benign drugs.  AR, so the investigator assessed disease free survival on S-TRAC and this is the point that comes up.  The investigator assessed endpoints for DFS, you know, don’t always translate very well when you have central review.  And on S-TRAC if you limit it to the investigator assessment the hazard ratio then crosses one whereas with central review it was positive so the argument that gets made is assure, if you had central review perhaps you would have a positive study.  Perhaps, perhaps not we don’t know, but on assure they play out the fact that you’re looking at a higher-risk population, only 30% of the patients on, or I’m sorry S-TRAC is on a higher risk.  Only 30% of the patients on ASSURE would have made it onto S-TRAC so perhaps what we should do is disregard the assure data which is dragging that meta-analysis towards non-significance and focus only on the S-TRAC data since that is the one that’s positive.  And we understand that argument, right, they are different studies, you try not to do cross-study comparisons, inclusions criteria are different, so what do you do with this, and the fact of the matter is the reason this is a pro/con debate is because when ODAC voted on this, they split 6/6 and so reasonable people can disagree.  You have a drug, we would all like to use it, one of the traps we get caught in, there are a couple of insidious traps in oncology, and one is the argument that we have high-risk disease, we know this is going to recur, you have to do something, anything, just do something, right, and that is an unhealthy way of thinking, that’s not a helpful way of thinking when we’re talking to our patients but it drives a lot of decisions, right, the desire to do something, I would argue that it is never the right answer.  The right answer is to do the right thing, and that should be data driven.  The other trap we get caught in is the argument that you have to give the patient hope.  At least if you are giving them something they feel some degree of hope.  That argument was made by some of the members of ODAC that voted on the positive side to say we should approve this drug.  The comment was actually made, I’m not sure it’s working or not but at least it gives the patient hope.  

For my part, I’m not a purveyor of false hope, I only give treatment because I believe there is an actual chance it is going to help.  Now, by the time ODAC met they had the overall—they had updated overall survival data, but even still here it is negative.  And I would argue and this is the heart of the argument that overall survival is what matters.  We expect, we expect if you read between the lines of the various comments that were made, I mean FDA is not bound by ODAC, but when it is split, it means FDA can do whatever it wants, and I think there is an expectation that we will see this approval, but I don’t know.  We will see.  So the pitfalls of this, the argument that gets made is disease free survival is an appropriate endpoint because it is appropriate in other diseases, it is used in melanoma, in colon cancer, in breast cancer, overall survival takes too long to achieve, we have this data today in 2017, should we wait until 2019 when there’s going to be thousands of patients between now and then that could take advantage of this data?  We understand that argument.  Okay?  It is important to be as efficient as possible in getting positive data and positive treatment outcomes to our patients without any question at all, but the problem, you know, I’ve got children, this argument that everyone else is doing it why can’t I is of course never a legitimate argument.  Okay.  It’s not just the disease but the mechanism of action matters, and I would point out we have these conversations at Mayo when these studies initially came along, I was frankly uninterested in opening these studies because there has never been a single VEGF directed therapy that worked in the adjuvant setting in any malignancy.  VEGF therapy so anti-vascular endothelial growth factor exerts most of its effect by acting on angiogenesis.  Ultimately most of the effect of these drugs is not directly on the cancer cells, it’s more on the tumor micro environment.  Now, perhaps in RCC with activation of VHL, you know, there is a reason why there is a mechanistic effect on RCC cells but nevertheless, mechanism of action matters, you can’t extrapolate what works for a chemotherapy study and apply it to a tyrosine kinase inhibitor, okay?  

Overall survival data is currently negative and currently pending.  We have already seen in the cancer world, drugs that got approved based on the PFS data and then the overall survival data matured, we see no advantage and then those approvals were turned around, we see this in bevacizumab in breast cancer, and this was a scandal to take the drug away after it had already been approved.  Now you had the Susan G Komen Army marching on Washington saying this is unconscionable, you can’t take patients off drug, it’s working for them, despite the fact that you now had a negative study, so there is harm done by jumping too quickly and saying let’s go ahead and approve this drug because it’s awfully hard to reverse something after it’s already been approved.  But why should it be overall survival?  One of the things I argue is it is not how you start the race it’s how you end that matters.  What might patients really want is they want me to treat them in the way that gives them the longest survival with the highest quality, and the patient doesn’t care at the end of the day if you achieve that with one drug or twelve whatever adds up to the highest quality for the longest time is what they want, okay, and in other malignancy surrogacy has been proven, the great example here is colon cancer.  In colon cancer adjuvant studies there’s been now dozens of these studies so they can do meta analyses of tens of thousands of patients, and it’s been proven that disease free survival is predictive of overall survival.  

In that setting it is absolutely legitimate to say we’re going to use DFS as a surrogate, right?  Once you’ve looked at 10s of 1000s of patients you can make that statement, we do not have that data in RCC.  We have to worry about the fact that cross-resistance between VEGF therapies occur, and so it’s possible that the placebo patients even if you hold off the disease in the beginning it’s possible the placebo patients are going to get the same benefit later on, so this gets to the point of my patients don’t care how many therapies or how we get where we’re going as long as we get the best result in the long run.  If you can get the same outcome with less therapy, it means you report has higher quality of life.  I want to give the patient the minimum amount of therapy that is necessary to achieve the optimal outcome, and this has been seen in multiple other cancer, breast cancer, ovarian cancer, where over treatment is a real problem.  In lung cancer, randomized clinical trials have proven that in stage 4 disease early palliative care leads to longer life than ongoing chemotherapy.  

So if you can get the same outcome but start therapy years later, why wouldn’t you do that?  I think most people would.  All right, now this is one of my real problems with these studies, when does a metastasis become a metastasis?  Disease free survival is characterized by the appearance of a new lesion, but this depends on size criteria and this depends on the power the ability of your test to find very small disease.  Anti-VEGF therapy decreases neovascularization.  We know that RCC is a highly vascular tumor, and when you resect an RCC it is not as if you are talking about 100% tumor cell content.  A lot of what is in that tumor is in fact connective tissue, blood vessels, benign healthy tissue.  

So it is entirely possible that all you do by giving early anti-VEGF-directed therapy is you delay the time until the tumor is large enough to be visible on a scan.  You’ve made it look smaller.  You still got a billion cancer cells, but instead of having half the content being blood vessels you only have 10% of the content being blood vessels and therefore DFS is improved, but you haven’t done anything to actually alter the course of the malignancy.  So I might be that all we are doing is slowing the maturation of the first metastasis well beyond—which increases DFS and ultimately has no impact on overall survival whatsoever.  This is a very real possibility and that is why OS matters, so DFS may be a radiographic artifact without any clinical significance.  

So I don’t think sunitinib is ready as adjuvant therapy in my clinic today.  I tell patients nothing has been proven to improve your survival.  Now, if we’re increasing the cure rate they should absolutely appear as an overall survival advantage.  If you are not improving overall survival you are not improving the cure rate.  DFS is not a surrogate for that at all.  

At the moment, there is no demonstrable clinical value.  DFS is only a number, not an outcome.  My patients are really not saying on what day is the first spot going to pop up, that is not what they are losing sleep over.  They want to know how long are they going to live?  Quality of life matters as does treatment related mortality, these are not benign therapies.  Adjuvant sunitinib is chronic therapy with chronic toxicity.  Like I say the top line data about toxicity when you talk about these oral drugs are really under representative of the true toxicity and the quality of life impact that the patient feels, so our obligation is to explain these details to the patients, make them really understand, get past all of the fluff and say we want to get down to the clinical impact.  I want to get down to what this does to your life, maximum quality for maximum time.  That is really the only criteria that matter.  

That is why I would say hopefully we don’t see an FDA approval for this.  I fall on that slide of the 6/6 split, but obviously reasonable people can disagree.  Questions?  Yes, sir?  

MALE VOICE:  – – hazard ratio means, and why the shift to hazard ratio – – ?  

  1. BRYCE:  So this gets to some extent to the slides I was talking about yesterday.  How do you define a benefit on overall survival, right, and if you choose an arbitrary point and say 50% of patients are alive at 12 months, or you know 18 months or how many percent of patients are alive, or what is the time point for 50% overall survival, and the problem is you are choosing one point to interpret the entire outcome.  Hazard ratio is at a normalization of the true split between the curves across the entire time interval.  Now, certain misstatements get made that I think make this confusing so you know, if you have a hazard ratio of 0.7, they will say there is a 30% reduction in mortality, and you would think that means you have a 30% higher cure rate, and that’s not really what that means, that must means that within the period of follow up the overall impact was 30%, so hazard ratio, statisticians won’t like me for saying this, but think of it if you will like you think about an area under the curve.  It’s a normalization of the entire space between the two survival curves to say how large is this space, how significant is this impact.   Other questions?  

All right, so everybody vote for me.  I won the debate.  Thank you.