DESTINY-Breast05 was really a follow-up study to an earlier landmark study, the KATHERINE trial, that had shown us that patients with HER2+ early breast cancer who had received our best neoadjuvant chemotherapy but were found to have residual invasive cancer did a lot better with T-DM1 than they did with trastuzumab. That was important because two reasons. One, it really established neoadjuvant therapy as a standard of care because that treatment identified a population of patients who would derive substantial benefit from T-DM1 that you would otherwise never see. So that’s why it was such a crucial study – for the first time we had something to do for the patients who did not have a complete disappearance of all the cancer in their breast.
When we looked at the outcome data at the early initial evaluation, the 3-year invasive disease free survival, we had really very robust outcomes for half of the population. But the group of patients who had an advanced cancer presentation – T4 primary tumour, N2, N3 nodes, or patients who still had positive nodes in their axillary specimen – already at three years were having a risk of relapse that suggested we’re going to need to do something beyond T-DM1. They had a 50% improvement but, because they started out at such high risk, a long way to go.
Around that time we were seeing the first reports about this new drug, T-DXd, that just showed remarkable activity in heavily pretreated patients, an amazing waterfall plot. Basically with that data we launched the DESTINY-Breast05 in those high risk patients to see if we could do better with T-DXd than we did with T-DM1. Not that T-DM1 didn’t help, it just didn’t help enough.
What was the study design?
The trial was straightforward, we basically took the KATHERINE eligibility criteria and added that we wanted the subset at high risk. Patients who had had the locally advanced, inoperable disease when they presented or had positive nodes were then randomised to either receive standard 14 cycles of T-DM1 or 14 cycles of T-DXd. When we looked at developing the sample size for this study we thought that, given the potential toxicities of T-DXd, particularly the interstitial pneumonitis, we’d want to see a substantial risk reduction from it. So we powered it to see a 0.675, to be precise, hazard ratio. It took 1,600 patients, gave us 80% power to do that.
As we always do in these studies we built in an early interim evaluation in case the results were better than what we had targeted for. And that was the case. We planned our interim analysis at about 70-75% of the events that it would take to see something for a hazard ratio of 0.675. IDMC said, ‘Guys, you’re there now,’ and that’s when we saw the data, our hazard was much better, it was 0.47. So we had a 53% improvement with T-DXd relative to T-DM1 which was obviously very gratifying.
The other aspect that we saw as we followed up KATHERINE was that the drug, T-DM1, unfortunately did not help us with the problem of brain metastases which is a big issue with HER2+ early breast cancer, particularly higher stage disease. As T-DXd was developed it was evident that it has a lot of activity against brain metastases so we were also pleased to see that early on in the study numerically the T-DXd arm does have fewer patients with CNS metastases than the T-DM1 arm. So it’s still early, it was 25 patients in the T-DM1 arm, 18 in the T-DXd arm. So numerically better but we’re going to need follow-up for that one to firm up. But that was another encouraging sign we had in the trial.
What toxicities were experienced?
In terms of toxicity, in general the T-DXd produces side effects that we expected to see from the development of the drug in metastatic. We do see more GI toxicities, you have to use antiemetic therapy. The real critical thing that we were of course watching was for the problem of interstitial lung disease. One of the aspects of moving the drug from the metastatic into the early setting is we added low-dose non-contrast CT scans at fixed intervals. We did it at baseline, 6 weeks after starting treatment and then every 12 weeks while on treatment with a final one 40 days after the last dose of drug, solely for the purpose of identifying hopeful early ILD where it was just showing up on imaging, minimally symptomatic where the symptoms may be mild enough that you might miss it. So that was an important part of the study.
What we saw was that overall with that we had 9.6% of the patients, so it’s essentially one in ten patients, developed interstitial lung disease or pneumonitis. We did add a step on the trial to really try and get very objective expert calls on those things. What we did was when adverse events were reported we had 56 terms related to lung problems that would trigger, ‘Let’s pull those scans in and review them.’ If we had a diagnosis of pneumonitis based on a scan local read we’d pull them in and read them. So we had an adjudication panel of experts who looked at the scans not knowing if they were getting T-DM1 or T-DXd. The site doctors knew that, so there’s always that question of bias a little bit. This way we got independent unbiased experts and that where these adjudication numbers came from.
Most of them were grade 1 or grade 2, 8.5% were grade 1 or grade 2, we just had seven patients with grade 3. We did have two patients who died from interstitial pneumonitis so it would be grade 5. When we talk about the gradings we’re always describing the most severe toxicity the patient experienced. So we had no grade 4s, presumably the patients who died at grade 5 passed through that but that’s how those numbers are kept, it’s always the highest grade toxicity.
The thing that we are still looking at very hard to try to have more details by San Antonio is the critical question of recovery. We do know investigators were asked is this adverse event recovered, is it recovering or is it not recovered in their judgement. The majority had said patients had recovered but there are questions then what does that mean? Does that mean the CT normalises? Does that mean symptoms disappear? So it’s that kind of granularity. Because a critical aspect in early will be the reversibility and, ideally, the total reversibility of this side effect. Because potentially if you leave somebody with chronic lung problems, it’s been a high price to pay. So this is still a part of it but we are dealing with numbers in this high risk group with a large benefit where it’s clearly worthwhile but when you start thinking about the earlier setting more broadly, you really do want to know. And it helps to talk to patients about risk, benefits and things of that type.
What are the next steps?
Next steps for the study, of course, is early follow-up. We will update the analysis some time next year when we get to the originally planned 207 events. That’s where the endpoints will be updated. In the meantime we’re going to be looking at the quality of life instruments that we’re collecting. We’ll start the translational work on it, I know we’re very interested in looking at ctDNA clearance in the trial; we collected specimens to allow for that. And getting a lot of granularity on the pulmonary toxicities, the recoverability.
We had some pre-specified groups where we did subset analyses, there are some other subsets that are of interest that we’re going to be reporting on – HER2 IHC status. In KATHERINE the 2+ patients who were FISH positive didn’t seem to get the same magnitude of the benefit as the 3+. We’re interested in seeing how T-DXd does because we know it’s active in HER2 low, so with that lower HER2 expression. So these are things that we’re going to be working on and we are going to have an oral abstract at San Antonio with more information.
What is the take home message?
T-DXd has really been just a transformative, remarkably active antibody-drug conjugate in the HER2+ metastatic breast cancer space. I think DB05 provides the important long-term invasive disease-free survival endpoint that really tells us, yes, there will be value for patients, benefit for patients with early cancer. The challenge is going to be identifying, getting back to the pulmonary toxicity, what is it? Do we just stay in the high risk? What about intermediate risk? These are things that are the next steps. But this is a critical point where we do move into now we can say, yes, there is evidence and these long-term endpoints.
We heard from the DESTINY-Breast11 neoadjuvant study that the drug for four cycles can increase pathologic complete response rates. So this is beginning this process of really figuring out where we really need the drug, where the patients will benefit uniquely from the drug, when can we do fine without it. The nice thing about the HER2+ space is we have so many effective drugs, it really gives us an opportunity to think more about an individualised approach. Because we had been working a lot on the idea… The other thing KATHERINE did is it allowed us to think about instead of just having to have everything up front to try to get that pathologic complete response as high as possible, maybe if we can back off the therapy and get most people there we can cure some patients with less; those who need more we identify them. So this is all part of the treatment paradigm that’s evolving and this is another important step in that overall process.
