Hello everyone, this is Dr Bishal Gyawali from Queen’s University, Kingston, Canada and I’m back with you all today to provide a round-up of the major trials and results from the ESMO 2020 virtual meeting. Before discussing about other trials first let me discuss about a couple of trials that were presented at ASCO and I had already summarised them in my ASCO round-up video. There were a couple of updates on those trials at the ESMO 2020 meeting.
First we are talking about the ADAURA trial in lung cancer. We knew that there was significant benefit in disease free survival from ASCO 2020 but at ESMO they also presented the benefit in terms of disease free survival benefit in CNS relapse and it was quite impressive as well. However, having said that, I had expected osimertinib to improve CNS relapse because of such good management of overall DFS benefit, it wouldn’t have happened without benefit in CNS. So, my take-home message from ADAURA does not change yet, OS data is still immature so we need to follow up on OS. There are a couple of points to highlight: we have the New England Journal publication of the ADAURA trial now but still we don’t know the percentage of patients that got MRI of the brain at baseline. So if this CNS effect of osimertinib is predominantly on patients who did not get MRI of the brain at baseline then that could skew the results. The other thing is about we also don’t know how many patients in the control arm got osimertinib at the time of relapse because osimertinib is the standard of care in first line treatment now for EGFR mutation positive patients. So if they have not received osimertinib at the time of relapse then that is another way in which the results may be skewed. We should also keep in mind that for this type of trials that have been stopped early for early signal of benefit usually the benefits are exaggerated so the hazard ratio, the true hazard ratio, is probably not as good. This looks so good because the trial has been stopped early.
The other trial that we discussed at ASCO as well is the KEYNOTE-177 trial of pembrolizumab versus chemotherapy in MSI high colorectal cancer patients. We saw an impressive benefit in PFS, nearly doubling of PFS with pembrolizumab versus chemo and at the time I had said that we also need to see what the effect is on quality of life. I had also expected overall survival to improve. We don’t know overall survival data yet but quality of life data was presented at ESMO 2020 and quality of life has been shown to improve with pembrolizumab versus chemo so I think that pembrolizumab should be the standard of care for first line treatment of metastatic colorectal cancer patients now.
Now, moving on to other trials, new trials, that were presented at ESMO, first let’s see some trials from lung cancer. We discussed about ADAURA, the other trial that was presented was the CROWN trial. It’s a phase III randomised controlled trial of lorlatinib versus crizotinib for ALK positive lung cancer patients. Again, there are some issues with the control arm here because the standard of care for ALK positive lung cancer patients now is not crizotinib, it is alectinib, but in this trial the control arm was crizotinib. So I don’t think this changes our decision making much. Lorlatinib is certainly one more option to have but it’s not a cheap drug either. So again I think that this just provides us with one more option but it cannot be called a standard of care because the control arm was not the standard of care. Having said that, lorlatinib is not exactly a very safe drug either – we saw high rates of increased triglycerides, hyperlipidemia, hypercholesterolemia, oedema and also there are some important effects in terms of cognition decline, so that should be kept in mind.
Interestingly, there was one more trial of a checkpoint inhibitor presented at ESMO. It’s called the EMPOWER-Lung1 trial of cemiplimab versus chemo. Now, this was done in Europe, Asia and the rest of the world, not in North America, but I’m quite unhappy with the control arm again. These are patients with PD-L1 more than 50%. These patients would never receive chemotherapy in the real world if they have access to. So I don’t know how many patients in this trial did not have access to pembrolizumab if they were not a part of the trial but if these are patients who would have had access to pembrolizumab if they were not a part of the trial but got randomised to chemotherapy simply because they were a part of this trial then that is quite unethical. I was not happy seeing that. Because we have also seen from this ESMO meeting that in KEYNOTE-024 the five-year overall survival rate for pembrolizumab has been 32% which is quite a significant achievement. We have seen this difference in overall survival despite such a huge crossover of nearly 66% crossover in the trial. So the theory that crossover takes away the overall survival benefit in trials is not entirely true. If the drug is really good and the survival effects are really pronounced then you will see a survival advantage despite crossover.
That’s all for the lung cancer trials. Moving on to the GI trials we already talked about KEYNOTE-177 but there were a couple of very important practice changing potentially trials in gastric cancer and oesophageal cancer: the CheckMate 649 and ATTRACTION4. These two trials tested nivolumab plus chemo versus chemo in patients with gastric cancer. The CheckMate 649 was limited to patients with a PD-L1 CPS score of more than 5. This was a global trial including 25% Asian patients. This showed a remarkable improvement in overall survival for gastric cancer by 3.3 months which has been unprecedented for gastric cancer. However, in ATTRACTION4, another trial of nivolumab again but done entirely in Asian patients, the overall survival was not improved, only progression free survival was improved. So we have to probably look at why there have been differences in the outcomes between CheckMate 649 and ATTRACTION4 and why ATTRACTION4 failed to show improvement in overall survival while CheckMate 649 improved overall survival by 3.3 months. We have to look into that in detail but it is encouraging to see an improved overall survival in gastric cancer first line and especially encouraging to see that the median overall survival bar has crossed the one year landmark in this trial.
Interestingly, the chemotherapy backbone for these trials is GELOX or FOLFOX for CheckMate 649 and GELOX and SOX for ATTRACTION4; SOX is S-1 plus oxaliplatin. So in some parts of the world we may be using a different chemotherapy backbone but these trials of nivolumab in gastric cancer have been done with the chemo backbone of any 5FU plus oxaliplatin, that should be kept in mind. In the case of CheckMate 649 although the trial was overall positive in the intention to treat population as well, we have to keep in mind that more than 60% of the patients in this trial had a CPS score of more than 5 so whether this benefit holds for patients regardless of CPS score, for low CPS scores, I’m not exactly confident because the total population has been diluted by a majority of the population having a CPS score of more than 5.
With regards to oesophageal cancer we had the CheckMate 577 and KEYNOTE-590 trials presented at ESMO. Interestingly, both of these trials include both squamous cell cancer as well as adenocarcinoma patients. KEYNOTE-590 includes more of the squamous patients, 73% squamous and 27% adeno, while CheckMate 577 includes only 29% squamous and the rest of them were adenocarcinoma. KEYNOTE-590 was about the addition of pembrolizumab to the treatment of FP, 5FU plus platinum, and this showed an improvement in overall survival in all patients, including squamous cell cancer patients with a CPS score of more than 10, all squamous cell cancer patients as well as all patients.
CheckMate 577 was in the neoadjuvant setting of patients receiving CROSS followed by surgery and for those patients who still had residual disease who could not achieve pathological complete response, they were randomised to receive either nivolumab or placebo for eight cycles. The primary endpoint for this trial, interestingly, was disease free survival and not overall survival. So I’m not confident about the results of this trial because oesophageal cancer, gastric cancer, these types of cancers, all treatments have been changed based on overall survival alone. Disease free survival as a primary endpoint is not enough in this case so we still have to wait on overall survival to change our treatment based on CheckMate 577, it will still be too early. Although the improvement in DFS seems quite impressive, nearly double the improvement in median DFS time, but the quality of life remains the same, there is no improvement in quality of life. So we have to wait on overall survival results.
Moving onto ovarian cancer trials there was one trial from ovarian cancer that caught my interest, that was the GOG3015 trial. It was a trial of chemo plus bevacizumab plus atezolizumab versus placebo for 22 cycles. Both PFS and OS were measured and both of them were not improved so this was a negative trial – the addition of atezolizumab to chemo plus bevacizumab did not improve outcomes. One interesting thing I have noticed when I am describing most of these trials is that for immunotherapy trials we have usually seen overall survival data presented and overall survival data are usually a primary endpoint or at least a co-primary endpoint. I had previously done this surrogacy analysis showing that PFS may actually underestimate the benefit for immunotherapy and overall survival should remain the gold standard endpoint, especially for immunotherapy drugs. So that might be a positive change that these trials of immunotherapy are bringing because most of them are including overall survival as a primary or at least a co-primary endpoint.
Moving on to breast cancer, one of the tumour types that had a huge number of important trials that were presented at ESMO 2020 but also one of the tumour types that leads to the most controversial trials, a very heated discussion. So first among the trials of breast cancer let’s talk about the IMpassion130 and 131 trials. From IMpassion130 that was published a couple of years ago we knew that the addition of atezolizumab to chemotherapy for triple negative breast cancer improved progression free survival, did not improve overall survival, but when we looked at the PD-L1 positive subgroup only then there was a significant improvement in overall survival, nearly ten months’ improvement in overall survival. But the caveat was that this trial was not designed to look at the PD-L1 subgroup, it was an exploratory analysis, it was not the primary analysis, a, and, b, in fact, in that trial design the hierarchical testing actually prohibits us from testing the efficacy of the drug in a PD-L1 positive subgroup. But nevertheless the trial presented the subgroup analysis of PD-L1 positive and that led to a big enthusiasm among many oncologists and patients to have access to atezolizumab for triple negative breast cancer in combination with nab-paclitaxel. But even at that time I had written a blog on ecancer cautioning everyone that this is an exploratory analysis. Yes, the results look interesting but this cannot be taken as confirmatory. In fact, this subgroup analysis was not even supposed to happen based on the trial design. So this could just be a false positive signal, that was what we cautioned when IMpassion130 was presented.
At this ESMO we had results from IMpassion131. IMpassion131 was designed to test that particular hypothesis of whether or not addition of atezolizumab to a taxane, in this case paclitaxel, improves outcomes for triple negative breast cancer patients whose PD-L1 is positive. So that subgroup benefit we saw in IMpassion130, whether or not it was a real effect that was designed to be tested in IMpassion131. Surprisingly, we saw that there was no benefit in PFS. Actually the overall survival hazard ratio is over 1.0. So the results of IMpassion130 and 131 diverged and people were making comments that probably it’s because IMpassion130 used nab-paclitaxel but IMpassion131 used paclitaxel, probably that made a difference.
But I think there is a much simpler explanation. One is that there is no reason why a paclitaxel or nab-paclitaxel backbone should make that much of a different impact from the separate trials. We see that overall survival is not different between nab-paclitaxel or paclitaxel for breast cancer patients. But, more importantly, you have a trial, IMpassion131, which is designed to test the hypothesis that whether or not the addition of atezolizumab improves outcomes for patients with triple negative breast cancer with the PD-L1 positive. And you have a trial, IMpassion130 which was not even designed to test that hypothesis but one exploratory subgroup analysis showed that it may be positive in PD-L1 positive breast cancer patients. In that case what do you trust? Of course you have to trust the properly conducted phase III trial that was designed to test that particular hypothesis, rather than trust the results from an exploratory analysis from a trial that was not even designed to test that hypothesis. So that’s why IMpassion131 was a confirmatory trial. So we have to trust the results of this confirmatory trial and unfortunately the addition of atezolizumab does not improve outcomes. I had even seen some slides from some discussant showing that atezolizumab plus nab-paclitaxel should still remain the standard of care. I totally disagree with that. Atezolizumab should not be the standard of care for triple negative breast cancer patients because the properly powered, properly designed trial showed that the overall survival hazard ratio is, in fact, more than 1.0. So I don’t think it’s just a difference of paclitaxel versus nab-paclitaxel. We need to have a higher threshold of evidence before using a new drug when, in fact, a trial has shown that it may even be harmful.
Talking about atezolizumab, IMpassion031 was done in the neoadjuvant setting, again, the addition of atezolizumab plus chemo. It improved pCR rates which is good but we had done a recent analysis published in EClinicalMedicine about the validity of surrogate endpoints in breast cancer. We have seen that pathological complete responses are a very poor surrogate for event free survival and overall survival. So unless we have EFS or OS data this should not lead to any change in practice.
We also had the SOLAR-1 trial overall survival results. This is a trial of alpelisib plus fulvestrant in PIK3CA mutation positive breast cancer patients, hormone positive, HER2 negative patients, as a second line treatment after receiving prior aromatase inhibitor. Again, in this trial there was benefit in PFS which we had already known before but overall survival there was no benefit. Numerically there is a difference of 39 versus 31 months but the hazard ratio crosses over 1.0, the confidence interval reaches up to 1.15 with a p-value of 0.15, so I was surprised to see that some of us were making comments that this is a numerical advantage in overall survival – a clinically meaningful survival benefit, although not statistically significant. Again, we designed the trial to test this difference as significant or not and it is not significant so I don’t think we can make the claim that it is clinically meaningful but not statistically significant. In fact, as someone had mentioned, we never see someone make a claim that some results are statistically significant but clinically meaningless, we always see people making the claim that results are clinically meaningful but statistically not significant. In any case, the results were similar to what we saw with everolimus in the BOLERO trial for hormone receptor positive breast cancer patients – improvement in PFS, no improvement in OS. So in this case we need to think about a couple of factors – whether there was any difference in post-progression treatment or not, that we don’t know yet. So physicians and patients should take all these factors into consideration when making treatment decisions.
Then we had the ASCENT trial of a new antibody-drug conjugate, sacituzumab govitecan. This was tested versus treatment of physician’s choice for triple negative breast cancer for patients who had already received more than two lines of therapy. There is an impressive benefit, there is a 5.5 month overall survival benefit. But, again, we have to be careful because this trial was stopped early for benefit and if a trial is stopped early for benefit there is always the chance that the benefit is over-exaggerated and is not a true benefit. But, more importantly, we see that in this trial 30% of the patients had not received platinum before but the control arm said treatment of physician’s choice but it does not include carboplatin as a choice. So even if it says treatment of physician’s choice it is actually not allowing physicians to provide carboplatin as one of the treatment options which we know is a better chemotherapy for triple negative breast cancer patients. So that also has to be kept in mind when interpreting the survival results from this trial.
Then we had two important breast cancer trials in the adjuvant setting, both from CDK4/6 inhibitors. The PALLAS trial of adjuvant palbociclib, it was stopped for futility, it did not show any difference at all so it has been a negative trial. This took many of us by surprise but also, interestingly, in both of these trials, adjuvant CDK4/6 inhibitor trials, although very high risk women were included, more so in the MonarchE trial of adjuvant abemaciclib, but all of these patients are not getting adjuvant chemotherapy and that concerns me. If we think that these patients are high risk enough to be enrolled into these trials why are we not giving them adjuvant chemotherapy? So that is one question that remains to be answered.
In the MonarchE trial of adjuvant abemaciclib it did improve disease free survival in high risk hormone positive, HER2 negative patients with a hazard ratio of 0.7. But overall survival is immature and the follow-up is quite short. So this disease free survival benefit, whether it collapses with time or whether it will lead to overall survival advantage as well is yet to be known.
We also had an interesting new drug called sotorasib, its results, presented at ESMO 2020. It is a new drug that is supposed to be a KRAS inhibitor. This is a big milestone scientifically in the sense that we never had a KRAS inhibitor before. Having said that, the results are exciting because we are able to drug KRAS for the first time but the results are not overwhelming. In fact, I was very underwhelmed with what I saw. There was only a 32% response rate in non-small cell lung cancer patients. For this type of patients with other targeted drugs we have seen a much better response rate. In fact, for colorectal cancer the response rate alone is 7.1% which is nothing at all. I had published a paper in the Journal of the NCCN this January showing that these response rates and durations of response may actually shrink when we test the same drug in a randomised trial.
The other interesting thing was the patients included in this trial had received more than two lines of therapy before and it will be interesting to see what the response was in those previous two lines of therapy, or what the duration of response was, just to figure out whether they were patients with a relatively indolent course, or whether they were responding well to other treatments as well before, or whether they had a very aggressive course and still we could find this response with the KRAS inhibitor. That would make a big difference.
Having said that, it is not entirely safe because 18.6% of the patients in this trial had grade 3 or higher side effects, luckily there were no fatal adverse events and that was quite reassuring.
Moving on to the final category of tumour type that I want to discuss today, genitourinary cancers. Firstly a study about prostate cancer; we had the overall survival results presented from the PROfound trial. Again, although these trial results will tell us that overall survival benefit was seen, we need to look at the control arm. The control arm is inferior to the standard of care because the control arm patients are getting abiraterone or enzalutamide after they have already progressed on the same drug. The other thing is people may highlight that there was an overall survival benefit even after adjusting for crossover but this was a trial where crossover should not have happened because olaparib had not been approved for prostate cancer before, this was its first trial. So, in fact, the control arm patients who had already progressed on abiraterone or enzalutamide and were still receiving abiraterone or enzalutamide, after progressing they did not get the chance to get a taxane, even after progressing, because they were crossed over to receive olaparib. So I think that has a huge caveat in interpreting the results. So I am not very excited about the results from the PROfound trial.
Then there was a trial of ipatasertib versus placebo in combination with abiraterone. This was called the IPATential trial. Again, in this trial only 18% had received prior docetaxel, now the standard of care for castration resistant prostate cancer would be to receive docetaxel up front with androgen deprivation therapy. But in this trial only 18% had received prior docetaxel. Interestingly, the primary endpoint was radiological progression free survival, it was not overall survival. In prostate cancer I don’t know a single drug that has been approved on the basis of progression free survival. The benchmark here is overall survival so this should not change practice because overall survival there was no difference.
I think all trials in prostate cancer should take lessons from the STAMPEDE trial, that is how to properly do a prostate cancer trial or any cancer trial, to be honest, a multi-arm, multi-stage trial with an overall survival endpoint.
Moving on to the kidney cancer, there were results from the CheckMate 9ER trial presented. This trial compared nivolumab plus cabozantinib versus sunitinib and it showed an improvement in overall survival. However, there is a big caveat here that patients who were randomised to sunitinib, only one-third of them got any PD-1 inhibitor in subsequent lines. We already know that we had already seen the trial of nivolumab improving overall survival in second line for renal cell cancer patients. So whether these patients who got sunitinib, whether they would have similar overall survival if they received a checkpoint inhibitor in a subsequent line, we don’t know that. To compare CheckMate 9ER with CheckMate 214 which tested nivolumab plus ipilimumab, it has better follow-up data, it has long-term data, and there is some hint of in CheckMate 214.
Another problem is when you are using nivolumab plus cabozantinib up front there are two extra factors to consider. One, what are you going to use when the patient progresses in the second line? Because if you are using nivolumab plus ipilimumab first line then you have the option to use a VEGF TKI in the second line. But if you are using nivolumab plus cabozantinib up front then you don’t have that option or you don’t know what is a good option. The other thing is that with nivolumab plus ipilimumab, ipilimumab is given just for the first four cycles but with nivolumab plus cabozantinib we are going to give these two drugs for a long time. So there will be toxicities, especially longitudinal toxicities, for a chronic duration, that should be kept in mind as well. Nivolumab second line was approved in 2015 so it’s really a shame that only 30% of the patients in this trial got any PD-1 inhibitor in subsequent lines.
Finally, moving on to bladder cancer there was one trial that caught my eye, it was the DANUBE trial in first line metastatic urothelial cancer testing durvalumab versus durvalumab plus tremelimumab versus chemotherapy. Both durvalumab or durvalumab plus tremelimumab did not improve outcomes compared to chemotherapy alone so this was also a negative trial.
Conclusions from ESMO 2020
We had a lot of important trials presented at this ESMO 2020 but my impression was that the biggest unmet need in oncology was not a particular drug or a particular tumour type but the biggest unmet need in oncology is actually critical appraisal skills, the skills to critically appraise the trials, to critically evaluate the trials. Because, as I have discussed until now, you must have realised that there were many trials that had fundamental design problems – they had an inferior control arm, they had an inappropriate crossover, they had used a surrogate endpoint that was not tested for that tumour type and they were not measuring quality of life. So these are some fundamental issues in clinical trial design that have to be always considered.
About the post-progression therapy, many of the patients we don’t know what they got post-progression or they got post-progression therapy that was not the standard of care. So all these things have to be kept in mind when evaluating a trial before jumping onto the bandwagon and saying that this is a game-changer, ‘I’m excited to use this treatment from tomorrow for my patients,’ or ‘Congratulations to Dr XYZ for this wonderful trial.’ Before jumping onto that bandwagon we should actually critically evaluate all the details and the pros and cons of each therapy and its issue in the trial.
In fact, we actually had a session at ESMO 2020 about this particular issue. It was the ESMO MCB session, the Magnitude of Clinical Benefit skills session. In that session I had a talk about these critical flaws in trial designs that can skew the MCB scores, that can make a drug look better than it actually is. So I would encourage all the listeners to check out that session, the ESMO MCB session, at this virtual meeting. But, most importantly, we should all be vigilant and we should always note whenever a trial is presented, we should always note what the primary endpoint is, if it is a surrogate whether it’s a valid surrogate or an invalid surrogate, whether the results, when they have been described, whether the trial was powered to detect significance in that result or not. If it failed to detect significance despite being appropriately powered whether the conclusions are misleading in that the conclusions say it was clinically meaningful but statistically insignificant or some other misleading language. And to look at the control arm, whether the control arm was receiving appropriate care, whether you, as a physician, would provide that control arm to your patients in a routine clinic or was it different, substandard. And what happened to these patients in the control arm after they progressed? Did they receive appropriate therapy and if there was crossover was that crossover supposed to happen or was the crossover not supposed to happen? So we have to take all these factors into mind while evaluating trials. Plus we need to also be careful about interpreting the results from exploratory or subgroup analysis. If the results from a subgroup analysis are contradicted by a properly powered trial to answer that specific question, we need to trust the properly powered clinical trial and not be attracted or seduced by the results of an exploratory analysis because that can be a false positive.
Thank you very much for listening to this roundup of major trials presented at the ESMO 2020 virtual meeting. Please check out other resources that are available in the ecancer site and elsewhere about these trials and do let us know your feedback, what you think about this.