An Ancillary Study of the STAMPEDE Trial Assessing the Clinical Qualification of Transcriptome Signatures for Advanced Prostate Cancer Starting ADT +/- Abiraterone Acetate and Prednisolone - Gerhardt Attard
October 3, 2022
Whole transcriptome profiling was performed using a clinical test (DECIPHER) on tumor index core mRNA from patients randomized 1:1 ADT vs. ADT + abiraterone acetate and prednisolone in STAMPEDE. Dr. Attard and colleagues selected 58 signatures for association with outcome, including four signatures shown to be prognostic in previous studies: AR-A, DECIPHER Genomic Classifier, PAM50, and PSC. Additionally, there were 54 signatures including capturing pathways of importance in prostate cancer biology (i.e., AR biology, immune signaling, cell cycle pathways, and cell of origin). A pre-specified statistical analysis plan was approved by the trial oversight groups. The primary objectives were to (i) establish the DECIPHER genomic classifier as a prognostic marker for OS in M1 patients and MFS in M0 patients and (ii) test the ability of AR-A, PAM50, PSC, and the DECIPHER genomic classifier to predict the treatment effect of abiraterone + prednisolone for OS. Cox models were fit with mRNA signature, abiraterone acetate and prednisolone (+/-), age, WHO performance status, pre-ADT PSA, NSAIDs/aspirin use, Gleason score, and disease burden (M0N0 vs. M0N1 vs. M1 low volume vs. M1 high volume) as covariates. Primary analyses included DECIPHER genomic classifier (continuous) for prognosis and AR-A (average vs. low) for prediction.
Dr. Attard addresses the results of the ancillary study of the STAMPEDE trial assessing the clinical qualification of transcriptome signatures for advanced prostate cancer starting ADT with or without abiraterone acetate and prednisolone focused on the following. He states that there was a consistent effect of the addition of abiraterone acetate + prednisolone on OS in the combined cohort. Also that the absolute benefit of adding abiraterone acetate + prednisolone to ADT was greater in patients with higher DECIPHER genomic classifier risk tumors, which has potential clinical utility, especially in localized disease, and that all four primary signatures were prognostic in M0 for MFS, with the DECIPHER genomic classifier prognostic in both M0 and M1 patients across all endpoints and finally that the exploratory analysis reveals different signatures are prognostic in M1 and M0, with an interaction with metastatic stage observed in immune-related signatures.
Gerhardt Attard, MD, Ph.D., FRCP, John Black Charitable Foundation Endowed Chair in Urological Cancer Research, University College London Cancer Institute, London, UK
Ashley Ross, MD, Ph.D., Associate Professor, Department of Urology, Northwestern University, Feinberg School of Medicine, Chicago, Illinois
Ashley Ross: Hi, and welcome to ESMO 2022 in Paris. I have the privilege of being here today with Dr. Gerhardt Attard. He's a professor and the John Black endowed charitable foundation chair of the urology oncology research division of the University College of London. He has a few presentations at his group has been giving today at ESMO, and we wanted to go over some of the data, particularly on some of the biomarker analysis from the STAMPEDE trial. So Dr. Attard, thanks for being here and I'm looking forward to our discussion. The presentation was looking at men from STAMPEDE and looking at some of the biomarkers that might be prognostic or otherwise potentially predictive, even of disease response to therapy. Can you elaborate a little bit about the arms of STAMPEDE that were being studied and what were some of the goals that you had going forward?
Gerhardt Attard: So STAMPEDE is a multi-arm study, which has incorporated what is effectively multiple phase three trials. And then today's presentation, we focused on the Abiraterone trial. That was 2000 men randomized between ADT versus ADT plus Abiraterone. The premise of STAMPEDE was to recruit patients who are starting long term androgen deprivation therapy, either because they had high risk localized, locally advanced disease. And that's about half of our patients and they were receiving ADT for three years and we mandated two years of Abiraterone plus radiation if that was clinically safe to administer. The other half of the patients had metastasis as defined on CT and bone scans, and that's a stratification factor. So we have two groups that are balanced across the randomization arms in this one phase three trial, which was presented first in 2017 back to back with a LATITUDE trial, which was only in high risk patients.
And that of course led to a change and the standard of care for metastatic patients. More recently, we presented data in that localized high risk population. So that was actually ESMO last year in the presidential session. And we showed two years of Abiraterone improves metastasis free survival that was a primary outcome measure and overall survival for those locally advanced patients. So as we know, the outcomes for prostate cancer are highly variables. Abiraterone is an effective therapy and indeed other second generation hormonal therapies. Apalutamide enzalutamide more recently, darolutamide has shown a similar efficacy in combination with docetaxel, but we have very variable outcomes. And we're not currently using molecular biomarkers to perform any risk stratification. So that underlying hypothesis here of this work was that transcriptomes will contain biologically relevant data and give us information that we could use to improve patient outcomes.
When STAMPEDE was designed, patients were asked to consent to analysis of their tissue. So we collected that tissue post randomization. We were successful in collecting tumor samples from about 70% of patients partnering with DECIPHER. Now, Veracyte, we extracted RNA, ran that on the raise and generated whole transcriptomes for 781 patients. And that biomarker population matches the full trial cohort, which you'd expect these large numbers. So then, or in parallel, we defined a statistical analysis plan, which had three parts. We aimed to generate data for 58 signatures and our priori, we stated which would be our primary signature and the primary question. And there were two primary questions. The first was, is DECIPHER prognostic. So there's a DECIPHER signature score is that prognostic and locally advanced and is a prognostic in metastatic patients. And the second question was, do we have a predictive effect? And we use a close test there, we redefined four signatures.
The primary one was an air activity signature. And we were looking to test a hypothesis that actually arose from analysis of the TITAN trial, that the AR activity, low cancers derived greater benefits. If that was positive, we would then cascade to three other signatures. So for prediction, we see consistent effect with Abiraterone across molecular signatures. So the relative benefit of Abiraterone added to ADT is consistent. So we see no evidence of an interaction between treatment and specific molecular subtypes. And that may make sense, because here we're not testing no treatment versus hormone therapy. We're testing, adding abi to ADT. The next question was prognosis and DECIPHER's highly prognostic. So per 0.1 unit increase and the signature score, we have a hazard ratio of between 1.18 and the metastatic patients 1.2 and the non-metastatic patients. And in fact, across the 58 signatures we studied, DECIPHER was the most prognostic across both disease groups.
What's its potential value and I think the greatest value and the potential clinical implications are in locally advanced because at 96 months of treatment, we can see a very large difference in the absolute benefits derived from additional of Abiraterone and in this population, especially this competing risks of death from other causes. And increasingly as we use drugs like Abiraterone and other second generation hormonal therapies in this localized population and potentially earlier lower risk patients, we're going to start overtreating. So we really need to have better risk stratification. And from this data, we can see DECIPHER, the DECIPHER signature is giving us that. We will need to validate that. We will need to test other cohorts, but that's the initial indication.
The third aim was to test all 58 signatures and their association with prognosis and we set a threshold P value cutoff of 0.001. And we see, we have a lot of interesting results and there are signatures that are prognostic in M1 and the ones that have been described previously on a single gene basis and the strong biology to support this such as the loss of P 10 signature, loss of P 53 signatures increase cell cycle signatures.
And then we have a number of signatures that are prognostic in non-metastatic patients that primarily relate to immune activity and these are not prognostic in the metastatic patients. So we formally tested that. We formally tested for an interaction between treatment and metastatic stage. And we see one. So we're observing this dynamic tumor immune relationship that has an influence on outcome in localized patients, but not in metastatic patients.
Ashley Ross: So thank you. That's a lot of work and a lot of information. So just to unpack it a little bit, first off, as you mentioned, the most prognostic molecular signature was this DECIPHER genomic classifier, which is commonly used at least in the states in my personal practice. So for me, when I'm presented with the metastatic hormone sensitive patient, that patient, as you mentioned, the data is very robust that they should usually be getting in almost every case intensified therapy with something like abiraterone and ADT. And it sounded like the big place where there's decisional uncertainty, and maybe we have to balance over treatment with oncological outcomes is in your N0 M0 high risk or N1 M0 space.
In both settings however, the DECIPHER classifier risk stratified hazard ratios were similar, but it's it just the outcomes among the patients that have metastatic disease uniformly are worse or higher stage. So am I understanding that correctly?
Gerhardt Attard: Correct.
Ashley Ross: And so there, you would say a utility of this, particularly as it gets validated in future studies is when you're presented with that patient who's M1 and you're deciding if you should intensify or not, DECIPHER low patients may have fairly good outcomes. And even though there was a difference in the survival curves, the absolute reduction is small enough that in selected populations, you might be able to save them the morbidity of adding on the additional therapy.
Gerhardt Attard: Okay. So in N1, I agree. So [inaudible 00:09:13] nonmetastatic population, so nonmetastatic by conventional imaging, we increasingly will need to identify who is going to benefit from these therapies. And there's several other trials that are both testing AR antagonists for this population, and then further left as well. There's going to be a point where the absolute benefit for a group of patients is too low to justify the increased toxicity. And it's clearly going to be increased toxicity. Now we can either use NCCN alone or we can use NCCN plus DECIPHER or we can move to DECIPHER alone. And I think that's something we're going to have to debate as a community. I think adding on extra information to current risk stratify is important.
Ashley Ross: And I think one thing that it brought up for me is that even if I took a strategy where upfront the patient and myself decided, even though the DECIPHER was low, we were going to pursue bimodal therapy in this non-metastatic setting. And if there was early signs of intolerance, it could give us a lot of confidence that we could drop the therapy as long as other parameters looked okay and continue with the intensified strategy. So that's excellent. And again, for our viewers, you've seen continuous large effects in the metastatic hormone sensitive setting for intensification of therapy and really where these markers are going to be helping us quite a bit is in the slightly lower stage, meaning N0 M0 N as a Nancy 1 M0 setting. Were there any surprises to you when you were looking at the other signatures, like the antigen response signature, the classifier of cell origin signatures, luminal versus basal, was that data at all surprising for you or somewhat expected? And where do you think that they'll go there as they develop that DECIPHER grid?
Gerhardt Attard: So the first AR signaling score, we find that increased AR signaling is protective, which I can't say I expected before we perform the experiment. It's counterintuitive that as a [inaudible 00:11:36] becomes cancerous, AR signaling increases and then as it becomes progressively more aggressive that signaling decreases. And we do find a negative association between AR signaling, a metastatic burden. So the higher the metastatic burden, the lower AR activity. So there's some biology to interrogate there. In terms of prognostication, I've just said air activity is prognostic, especially in M0, but not as accurate, not as significantly associated with survival as DECIPHER. Pam 50 and and there's another basal luminal phenotype signature we use called PSC, which is newer and we'll have to explain this better, generate more data. But the signal's there with Pam 50 are not very strong. So we do not see need a treatment interaction nor associations with long term outcome.
For PSC we see that some associations with outcome. So the luminal patients attending to do better. And how we will use that in clinical practice when we already have DECIPHER saying such a stronger signal, that's something to tease out. And I think let's keep our eyes on the docetaxel question. We're discussing this earlier that in metastatic patients, the key question is who should get chemo? So we have this backbone, ADT and NHA, so second generation hormonal agent. A proportion of patients are probably going to benefit from addition of docetaxel. We don't have randomized data to show that, but we have non-randomized data suggesting improved outcomes, but the effect is likely to be heterogeneous. We're seeing this across trials. So can we identify that subgroup that really get benefit from docetaxel and it is of course really interesting data for Pam 50 from the CHAARTED trial, suggesting we could have molecular subtyping that helping that decision, and that is ongoing work I hope we'll present next year.
Ashley Ross: That's wonderful. And it actually shows that how much the things like DECIPHER is useful across the spectrum, even in that metastatic setting is you're not going to go down to unit into just ADT alone, but when deciding between bimodal and tri modal therapy, if you will, it might play a large role. The final thing to unpack is this is very interesting and rich information that Dr. Perry presented about the immune signatures and that differential between M0 M1 space. I found it academically provocative in hypothesis generating. I can't say I've wrapped my mind around what's the step for me? So I might have you do my work for me and tell me, what do you think the next step would be either around trial designs or further analysis to sort of understand, is this an immune surveillance phenomena? Is it something predicting response to a certain therapy? I wonder what your thoughts are there.
Gerhardt Attard: So the ongoing work to unpack it, and this is going to, I think, take a while, but the implication is that the biological processes driving the disease are different and the locally advanced versus the metastatic. And therapy development has tend to be prove efficacy for very end stage disease and then slowly move this back to localized disease. And maybe we need to rethink that if we're targeting different biological processes. And of course, immune checkpoint inhibition and other immune modulating strategies haven't been successful in unselected metastatic patients. And I think we need to really think carefully about how we use these signatures potentially in localized disease to select patients for immune modulation. So I think that's what is currently really exciting is that potential.
Ashley Ross: Well, thank you again. It was great work. A few studies presented by you and your team here at ESMO. And I think it really is continuing to usher in the era of molecular and sort of risk stratification, molecular signatures that help us with treatment choices. And it is been very exciting to see it's all been happening in the last decade and now really coming to fruition. So thanks for your time, Dr. Attard and I hope you enjoy the rest of the meeting.
Gerhardt Attard: Agreed. Thank you.