The ADAURA Trial of Adjuvant Osimertinib in NSCLC

Neal Navani, MRCP, MSc, PhD: Last year, as we know, we had the ADAURA trial published. Solange, perhaps you could give us some details about the ADAURA trial.

Solange Peters, MD, PhD: I’m very proud to say it was also presented at ESMO [European Society for Medical Oncology Congress] in 2020 for a second time. This trial has been presented at least 3 times. Some other time points I ignore, because it’s an evolving trial looking at the various parameters at that time. To come back to the previous discussion about the alectinib trial [ALEX], for example, we’re speaking about a never-met magnitude of difference between an experimental arm and a standard arm. Basically, the ADAURA trial is a trial assessing osimertinib as an adjuvant therapy for patients with EGFR-mutated non–small cell lung cancer. The stages are from stage IB to IIIA, but according to the seventh TNM classification, we need to be careful because it moves a little to stage IB. After complete tumor resections, the usual criteria help us when looking at complete resections. The idea of this trial was not to violate the usual standards of care, meaning that surgery was performed as routinely dictated by the standards of surgery. Adjuvant chemotherapy could be given by the investigator according to their local standards of care, which I must say are quite established everywhere. Patients had to have a disease with classical, actionable EGFR mutations, meaning they had an exon 19 or L858R exon 21 mutation, and they had to be fully assessed—including brain imaging—before randomization.

These patients are randomized 1:1, and after surgery and chemotherapy based on the standard of care. They’re given osimertinib 80 mg, which is classical, once daily for a maximum of 3 years. Others were given a placebo once daily for a maximum of 3 years. It’s important to say that patients were stratified according to stage IB vs II vs IIIA and that the patient was stratified according to the type of mutation—which is a large matter of debate: exon 19 or exon 21—and by race, because sometimes, in the field of toxicity of a TKI [tyrosine kinase inhibitor], that might make a difference. The primary end point was focusing on the high-risk subgroup, as we were thinking. It was DFS [disease-free survival] for patients at stage II and IIIA with a target superiority and a hazard ratio of 0.7. The secondary end point was looking at the overall population, thinking also about the inclusion of stage IB.

What must be stressed about this trial is that this was not a planned interim analysis. It was an IDM [independent drug monitoring team] looking at the data, and the IDM asked to unblind the trial because of this major signal of efficacy. At the time of unblinding, the study had already completed enrollment and all patients had followed up with their doctors for at least 1 year, removing some of the ethical considerations that you can have in these circumstances. Basically, 1 thing which is striking when you look at these data: Patients with EGFR-mutated lung cancer had a poor outcome after surgery.

We had some previous data that showed that, potentially, these diseases were behaving the same, or more aggressively than unmutated tumors. Here, what was very visible in the ADAURA trial was that in most patients with stage II and IIIA disease, the primary end point was relapsing. With this disease-free survival in the placebo arm of 19.6 months, there was a curve going down, which was surprising or even disappointing. Despite the best surgery and adjuvant chemotherapy when patients relapse, most of them relapsed. If you add in this subgroup’s primary end points, these patients at stage II and III, who were on osimertinib for up to 3 years, with the maturity of this set of the patients, the hazard ratio was 0.17 with a median DFS not reached for osimertinib. Keep in mind this 0.17. This 83% reduction in the risk of recurrence or death is something that we never or very rarely see in oncology. That is about the extent of the benefits.

This is a primary end point with an amazing separation of the curve. Of course, we have to keep in mind that the secondary end point includes all comers and those with stage IB disease. First, the placebo curve is strikingly disappointing. When you add osimertinib to the standards of care, of surgery, and of adjuvant chemotherapy, the hazard ratio now is 0.2. This is still something extremely important regarding the maturity, which more or less was 29%. Even when including the patients in stage IB, you can see this striking benefit. Then come the questions of subgroups, and you were exposed to forest plots trying to identify a subgroup that would be divergent in the benefits. We could not find this subgroup. Basically, all usual prognostic subgroups were benefiting from that—including 1 thing I would like to discuss with you: the adjuvant chemotherapy. The question of ADAURA was not to interrogate the role of adjuvant chemotherapy. Patients were not randomized or stratified in a way that allows us to compare patients with or without adjuvant chemotherapy. Remember, this was left to the investigator. The trial was not intended to look at the role of adjuvant chemotherapy, but based on the fact that patients were given adjuvant chemotherapy, and based on the investigator decision, we were not surprised about the administration of chemotherapy, which follows the usual description of the adoption of adjuvant chemotherapy. There was a very low-level in stage IB, but a high-level in stage II and IIIA. Importantly, it looks like, based on this investigator’s decision, osimertinib kept its beneficial effect with or without adjuvant chemotherapy. With adjuvant chemotherapy, there was a hazard ratio of 0.16; without adjuvant chemotherapy, that rate was 0.23. Based on this very correct adjuvant treatment decision by the investigator in terms of chemotherapy, osimertinib still plays its role in improving DFS. Be careful. The question was not about chemotherapy, so it would be too early to say we can skip chemotherapy because this question was not addressed. The benefit is clear irrespective of the decision about chemotherapy, and it’s quite important. I’m sure we will discuss the stage.

You wouldn’t be surprised to see that the benefits of adding osimertinib to the treatment are larger when you’re at a higher risk of relapse. Patients were going from 0.12—if I remember well enough—from stage III to something like 0.5 for stage IB. Is a 0.5 DFS hazard ratio for stage I significant? We can discuss it later, but to me 0.5 is a reduction that divides the risk of relapse in 2. What about survival? First, in the trial—remember, it’s an unplanned interim analysis—the only thing we should say is survival is not mature. The maturity at the time—we have seen the data across meetings—is 5%, so nobody can conclude anything. The hazard ratio appears to be 0.4. With such maturity, we should not make any conclusions. With this magnitude of difference in DFS rates, there may be benefits in overall survival. I would guess that there is, but it’s a guess. It’s too early to say that survival rates will be positive or that the goal has been met. Why am I convinced about the strategy? It’s basically the description we had at ESMO about the site of relapse. When you give osimertinib to patients, most of the relapses will be observed locally. That’s quite surprising, but most of the relapses are local recurrences. If you don’t give patients osimertinib, most relapses are distant recurrences. These distant recurrences might be in the lung but might also be in the brain. Obviously, there’s a huge difference in brain relapse between the placebo arm vs those treated with osimertinib. Osimertinib is a drug that penetrates the CNS [central nervous system], meaning that by giving osimertinib you might protect the CNS.

The reduction of brain relapses, shown by Masahito Tsuboi and colleagues at ESMO—even if they’re not mature data—was already 82%. Altogether, there are fewer relapses. But more important, there are no brain relapses, which affects all aspects of the life of a patient psychologically but also physically. It’s a very important point that convinced me about the idea for the time being: We are waiting to see the rates of survival and for longer follow-up. That’s something I will discuss. With everything I’ve explained to you, with all my patients I am resecting EGFR-mutated non–small cell lung cancer from stage IB as defined by the seventh TNM classification. We’re always weighing the risk-benefit ratio of everything we do, and sometimes in these discussions the patient will be central in deciding alongside you about these treatment strategies. You have everything there, and I’m happy to see if you have other questions about the trial.

Neal Navani, MRCP, MSc, PhD: Thank you, Solange. That was very comprehensive, indeed. You mentioned 1 interesting point about the DFS rate in the control arm. You mentioned that there was a high relapse rate in patients with an EGFR mutation after surgery. The DFS rate in the control arm of the ADAURA trial was around 44% at 2 years, which is quite low if you compare that with the LACE trial’s meta-analysis, for example; that was around 60% for the same time point. How do you interpret that? Do you think patients with EGFR-mutation lung cancers are more likely to relapse after surgery? Could it be that there are issues around the quality of the staging in the trial that perhaps we don’t have enough information about?

Solange Peters, MD, PhD: It’s a modern trial, so it’s probably not the quality of the staging. Remember that the criteria of full resection, as stated by Ramon Rami-Porta in 2005, was rephrased recently last year. The application of correct staging and the correct resection, according to complete resection criteria, are better met than before. The LACE meta-analysis is an example. I would not say the surgeon is guilty or the staging or the radiologist is faulty. We have a question here: Is an EGFR mutation a negative prognostic factor in resection in early non–small cell lung cancer? It’s a question mark.

We don’t have formal comparable data, but we have some publications. If I remember well, we have 2 or 3 publications. I remember 1 in the Annals of Oncology and 1 in Oncotarget in 2017 and 2018, showing that there is a prognostic impact for surgical patients when some of the mutation can be found: Patients with EGFR, KIP1, or PIK3CA mutations will experience more relapses, as seen in this subgroup. The ADAURA trial teaches us something, of course. We don’t have any standard arm or an unmutated arm to make sure about it, but they suggest that patients with EGFR-mutated non–small cell lung cancer might relapse more than you would expect patients with wild-type EGFR to relapse. I don’t know your feelings on this curve. That was my first feeling when I saw this curve: “God, they relapse fast.” The majority do relapse.

Neal Navani, MRCP, MSc, PhD: It’s very interesting how a significant portion seem to relapse, and relatively quickly as well. It’s very interesting to know more data about that, and maybe other EGFR-mutation cohorts might help with that in historical surgical cohorts if we could go back and look.

Transcript Edited for Clarity