Biogen Inc. (NASDAQ:BIIB) Aducanumab Phase 3 Topline Results Conference Call December 5, 2019 11:00 AM ET
Ron Petersen – Neurologist and Director, Alzheimer’s Disease Research Center at the Mayo Clinic, Rochester, Minnesota
Samantha Budd Haeberlein – VP, Clinical Development
Paul Aisen – Director, Alzheimer’s Treatment Research Institute, USC
Sharon Cohen – Behavioral Neurologist, Toronto
Steve Salloway – Professor, Neurology and Psychiatry, Brown University
Conference Call Participants
Well, good morning, everyone. Thank you so much for coming out to hear what I think is going to be a very enlightening presentation this morning and subsequent discussion. My name is Ron Peterson, I’m a Neurologist and Director of the Alzheimer’s Disease Research Center at the Mayo Clinic in Rochester, Minnesota, accompanied by an esteemed panel, whom I will introduce in just a moment.
First of all, I’d like to thank the CTAD leadership committee for allowing us this presentation this morning. So, Paul, Bruno, Jacques and Mike, thanks very much for this, because I think this is a pivotal time for us in the field. And I think what we are going to discuss this morning’s is going to be very, very important.
I’d like to introduce our panelists. First, we have Samantha Budd Haeberlein who is the Vice President for Clinical Development at Biogen, and she will be presenting the data. Next is Paul Aisen, who is the Director of the Alzheimer’s Treatment Research Institute of the University of Southern California here in San Diego; Sharon Cohen, a Behavioral Neurologist from Toronto who is a principal investigator in the studies we will be discussing; and Steve Salloway who is Professor of Neurology and Psychiatry at Brown University, also a principal investigator in the studies.
So, those are our shots. Those are who we are. Here are disclosures, and I’ll leave those up for you to read yourself.
I think, this is a pivotal time in the field, certainly with regard to the development of Alzheimer’s therapeutics. There have been a lot of ups and downs in recent months and years and particularly more downs than ups. And in many senses, the aducanumab data, which we will be discussing today, sort of characterizes, this is almost a caricature of the course that we’ve had in the field. As you know, these studies were motivated based on the Phase 1b study published five years ago on the potential efficacy of this compound, two international studies were launched, very exciting, going along well, and in March of this year, both studies were stopped because of the futility analysis. They met futility criteria. Subsequent data were collected, however, in a blinded fashion, and those data were reanalyzed. And perhaps a different interpretation of the results of these studies arose. Consequently, the Food and Drug Administration agreed to allow in October of this year, made the statement that the Biogen could in fact file for possible regulatory approval. So, where we are today is then how do we get there? What would these data show? And Samantha is going to share with us these data in just a moment.
So, the flow of this morning is going to be, I’m going to turn it over to Samantha in just a moment, and you’ll see up there that — and you saw this when you came in, we’re going to invite questions from you. So, please register your questions on your phone and we will collate them and try to put them together over the next hour or so to present them to the panel in terms of the ones that are most salient, most relevant to the discussion.
So, Samantha will present the data. Then, I’ll ask our panelists to give their brief impressions of the data. And then, we will open it for questions through the phone. So, I think this is going to be very interesting, very exciting. And we really appreciate your opinion and interpretation of these data. I turn over to Samantha.
Samantha Budd Haeberlein
Thank you, Ron, and good morning, everybody.
So, as you’ve heard, my name is Samantha Budd Haeberlein, and I work at Biogen. And I’m going to report this morning the topline data from EMERGE and ENGAGE, which were two identically designed studies to evaluate aducanumab in patients with early Alzheimer’s disease. And I’m doing this on behalf a passionate and talented team of clinicians and scientists who conducted and led the analysis of these studies and my co workers listed here.
Now, I know many of you will be interested to take pictures of the slides, and that’s much appreciated. But, we — just to let you know that we will post these slides at the end of the presentation on biogen.com. So, those will be accessible for you.
Just to say that I may be making forward-looking statement and that these are subject to certain risks. And I would suggest that if you want details on those that you look at our SEC filings.
One last disclaimer goes without saying that aducanumab is an investigational compound and is not yet approved.
So, starting with the design then. As I mentioned, these were two identically designed trials. They were 18 months in treatment duration. They enrolled 3,285 patients over the two studies and 348 sites in 20 countries, which are listed to the right. The population in these studies by design was approximately 80% MCI due to AD and approximately 20% mild Alzheimer’s disease dementia.
We investigated two dosing regimens in these studies, a low and a high dose, and these were randomized 1:1:1 with placebo. The primary endpoint was the main change from baseline in the CDR Sum of Boxes at week 78. The pre-specified secondaries were MMSE, ADAS-Cog 13, and ADCS-ADL-MCI. And there were sub-studies that evaluated the effect of aducanumab on disease-related biomarkers.
Now, at the time we started these studies, we had learnt from the Phase 1b that ARIA was — the incidence of ARIA was higher in ApoE4 carries and increased at higher doses. And so, the two dosing regimens that we studied in this study were stratified ApoE4 carriage and had a titration phase prior to reaching the dose.
I’m showing here the low dose regimen where ApoE4 carriers were titrated up to 3 milligrams per kilogram, and non-carriers were titrated up to 6 milligrams per kilogram. In the low dose, this regimen was maintained throughout the duration of the study. In the high dose regimen, initially the non-carriers were titrated up to 10-milligram per kilogram, that’s the purple line here; and the carriers were hydrated up to the lower 6-milligram per kilogram. However, after we learnt more about our ARIA in our Phase 1b study in the titration cohort, the titration could reduce the incidence and severity of ARIA. We amended the protocols such that ApoE4 carriers would also receive 10 milligrams per kilogram in the high dose group.
The amendment I’m going to refer, and it’s important to understand the context of these results, was in protocol version number 4. And if you were in the high dose group, the maximum number of doses of 10 that you would receive were 14.
Now, I’m going to talk a little bit about the implementation of that amendment, because it is important for the interpretation of the data that I’ll speak to subsequently.
Here are the enrollment curves for the two studies, ENGAGE and EMERGE. We started recruitment of ENGAGE one month ahead of EMERGE, and ENGAGE remains slightly ahead in enrollment throughout the study.
I’m going to build this. We recruited 1,643 subjects in EMERGE and 1,653 subjects in ENGAGE. And there were two protocol amendments that had an impact on dose in these two studies. First of all, protocol version 3, which went into effect in July of 2016, was to enable some subjects who had been dose suspended due to ARIA, to restart their treatment at the same dose or to titrate upto the originally assigned dose. At the time we implemented that amendment, ENGAGE was already 100 patients ahead in enrollment. And then, the second protocol amendment, and I just mentioned it, was protocol version number 4 that was implemented in March of 2017. And that was to take the ApoE4 carriers upto 10 milligrams per kilogram. And the time that this was implemented, ENGAGE was now ahead by 200 patients in the enrollment. The populations that we included in the futility analysis by design were the approximately first 50% patients who had the opportunity to complete the week 78 visit. And that cohort of individuals were fully enrolled by the June of 2017. You can see the vertical line here reflecting that. Of note, that is quite close to the initiation of the implementation of Protocol Version 4. The two studies completed their enrollment in July of 2018. And it took 35 months to recruit these studies.
I want to go a little bit further into the implementation of the Protocol Version 4. This is a different type of chart that you may not have seen before. What we are looking at is the individual patient consent to Protocol Version 4 in the 20 countries and 348 sites. The countries are listed on the left. And each vertical line is an individual patient timing of consent to Protocol Version number 4. You can see that it’s a very large undertaking. And overall it took 18 months for the complete implementation of this Protocol Version. You can also see if you look at the timeline at the bottom that by the time the futility cohort had been completed their enrollment, very few subjects will have had the opportunity to consent to Protocol Version 4.
Protocol Version 4 had a real impact on the dosing in these dosing regimens, in particular in the high dose group where the change occurred. I want to draw your attention to the right here where we are showing the median cumulative dose at week 78. In individuals after the implementation of Protocol Version 4 in the purple box, the median cumulative dose, actual median cumulative dose was 153 milligrams per kilogram. The target median cumulative dose for those who did receive 10, would have been 160-milligram per kilogram. So, you can see that it’s quite close. However, for the pre Protocol Version 4 population, that median cumulative dose was only 116 milligrams per kilogram.
So, let’s have a look at the impact of that on individual patient dosing. I appreciate this plot is really quite complicated. What I’m showing you for every individual in high dose regimens for EMERGE and ENGAGE, both their individual dose pattern and each dose that they took. Let me orient to you a little bit. The dark blue are doses of 10 milligrams to kilograms, the yellow are no doses, and on the left here the vertical gray or lilac bars, those are the titration doses through 1, 3 and 6-milligram per kilogram, or indeed if you see lighter colors to the right, those are individuals who did not receive 10 or had a dosing interruption.
Generally, looking at this, what you want to have is more solid blue. And in fact, if you look at the numbers at the top here, the reality is that only 29% of patients in EMERGE and 22% of patients in ENGAGE received the full possible 14 doses of 10 milligrams per kilogram. The majority of the population had a very heterogeneous dosing image with either not achieving 10 milligrams per kilogram or having dose interruptions and suspensions.
Okay. So with that backdrop where dose is important and a complexity in this study I would like to take you through the top line results. First of all, I would like to walk you through the datasets that I’m going to refer to in the following presentation.
First of all, the utility and which I’ve already mentioned is when approximately 50% and precisely 49% and 57% of patients had the opportunity to complete the week 78 visit by December 26th of 2018. After the futility analysis, a larger dataset became available, and that larger dataset showed a different outcome from futility. And from the basis for the subsequent analyses that we have conducted to understand the difference from futility and the difference between the two studies. And in that larger dataset, those additional analyses have been conducted in close consultation with external advises and with the neurology division of the FDA.
In the larger dataset, we have conducted analyses on the primary analyses, which is the intent to treat population and also the opportunity to complete population. I want to point out that in that dataset, data after the March 21st announcement of utility were censored to remove the potential of bias introduced by that futility announcement. The p-values for the larger dataset should be considered nominal.
Most recently, really very recently, we have had database lock on EMERGE and ENGAGE, and we now have the final dataset. And we have run the pre-specified primary and secondary analyses on final dataset. And today I will be sharing those data with you. The final dataset is the basis for statistical inference for future regulatory submissions and these are actual rather than nominal p-values.
Last but not least, I mentioned that there were sub-studies, and those are listed here for the disease related biomarkers. Of note, final dataset and larger dataset, both have the same number of individuals, but the large dataset has small administrative alterations due to data cleaning, central review and QC. And both datasets, as I mentioned, are censored for the after March 20th announcement.
Okay. So, for the remainder of this presentation, the majority of the data I’m going to present are from that larger dataset. There are just two slides that I am going to show from the final dataset, given that these are very recent.
Okay. First then, I would like to present the futility analysis that we announced in March. The futility assessment was based on conditional power, which is a probably calculation that the primary efficacy endpoint would be statistically significant at final analysis. We had pre-specified criteria for the futility analysis and that was that both dose arms of both studies would have a less than 20% conditional power to meet the primary endpoint of final analyses.
This pre-specified methodology, given that the two studies were identical in design, was based on the calculated pooled data from both studies to predict the future behavior of each study. Pooling was used, because pooling is believed to be a more powerful statistical methodology on the assumption of homogeneity. Using this methodology, the futility criteria were met. And as you’ve heard, we discontinued the studies. However, at the time of the futility analysis, EMERGE was trending positive while ENGAGE was not.
As I mentioned, subsequently, additional data became available and we assess this additional data using the pre-specified endpoint that were pre-specified prior to the futility analysis.
First, looking at the baseline demographics, I’m going to take you through each study one-by-one, and I’m going to start with EMERGE. Here are the baseline demographics for EMERGE that are generally balanced across age, gender. And you can see that their age here is on average 71 years, which is consistent with that earlier patient population. The studies have approximately two-thirds of patients who have an ApoE4 gene, and that is consistent with what we anticipate from the Alzheimer’s disease population.
As I mentioned, by design, the baseline clinical stage was approximately 80% MCI due to AD and 20% mild Alzheimer’s disease dementia. The amyloid PET SUVR baseline in the subgroup were consistent across each of the arms of the study.
Moving then to the baseline disease characteristics. And these were generally well balanced across each of the arms of the study and consistent with the protocol for inclusion. The average baseline MMSE is around 26, again reflective of the intended patient population.
Here is the patient disposition for EMERGE. 1,643 patients were randomized and 1,638 patients were dosed. Between 15% to 24% of patients discontinued treatment across the arms. And while individuals may discontinue treatment, we made great efforts to encourage these individuals to stay in the study and to continue their assessments post discontinuation, despite not having treatment. And those additional visits are included in the primary analysis. The number of individuals who withdrew from the study completely were between 7% and 12%, which is a relatively low level of withdrawal from a study.
I’m showing here then the primary and secondary endpoints from EMERGE and ENGAGE. These were analyzed on the intent to treat as demand, which are all patients, all data. The method was the mixed models for repeated measures. And starting with the high dose then, on the primary endpoint, there was an advantage of aducanumab over placebo of minus 0.40. That’s a 23% change from baseline with the p-value nominal of 0.01. On the MMSE, there was a minus 15%, on ADAS-Cog minus 27%, and on the ADCS-ADL-MCI scale, there was a minus 40% difference from placebo at week 78.
In the low dose group, we can see a numerical advantage of aducanumab over placebo. And this is lower than that seems in the high dose group and did not achieve statistical significance.
As I mentioned, we now have the final data set, and I’m going to show you the same table with that final data set, which has the full QC and the full data included. And essentially, as you can see, there are very small changes between that larger dataset and the final data set. Two numerical differences are that the CDR Sum of Boxes has moved from 23% to 22%, and the MMSE p-value has moved from 0.06 to approximately 0.05. So, very, very little difference between those two tables.
I’d like now to show you the longitudinal plots of each of these endpoints, starting with the change from baseline in CDR Sum of Boxes. And each of these chart is going to have exactly the same schema. So, I’m going to take you through it a little bit. In gray, what we have are the placebo, which in EMERGE declines by 1.74 units over the 18 months of the study, which is within the range that we anticipate from this patient population. And then, in the blue line, you have the low dose, and in the purple line or magenta, you have the high dose. And you can see that these diverge from placebo over time. At the bottom on each of these plots, you can see the patient numbers that were included in the analysis. And given that we had an early termination of the study, at week 50, we have about 80% of the data that we would have had. And at week 78, we have about 55% of the data. And you’re going to see that trend in each of the plots that I’m going to present.
Next is the longitudinal change in MMSE, similar format as before. Here, I’m showing the longitudinal change from baseline in ADAS-Cog 13. And here, the last, secondary, showing the longest change from baseline in the ADCS-ADL-MCI.
Moving then to amyloid plaque reduction as measured by a cortical composite in amyloid [Technical Difficulty] you’ve seen similarly that we’ve previously — and the colors are similar to the scheme I showed before, gray is placebo, blue is low dose, magenta is high dose. And essentially, you can see a dose and time dependent reduction in amyloid plaque in EMERGE. And the change from baseline in the high dose group is minus 0.272 units. This is quite similar to what we have seen previously in our PRIME study.
Additional biomarkers of disease and neurodegeneration, i.e. phospho-tau and total-tau were measured in the CSF. I want to point out that the sample sizes are at the bottom here. And this sub study was smaller than we would have liked and quite small for this methodology. But nonetheless, what we’re looking at here is in the phospho-tau in EMERGE, you see a dose dependent and statistically significant change in phospho-tau. And in total-tau, while this is dose proportionate, the reduction from placebo, it did not reach statistical significance.
Okay. So, that was EMERGE. Now, I’m going to take you through the same set of analyses with ENGAGE. Starting with the baseline demographics, which were well-balanced within groups and ENGAGE and across the 2 studies, perhaps with one distinction that the baseline disease medications in ENGAGE was slightly higher. Here are the baseline disease characteristics, also well-balanced across arms and between the two studies. And in fact, the baseline demographics and disease characteristics, while there may be small changes between the 2 studies, those changes are not significant enough to be the basis for the difference between the 2 studies.
Here’s the patient disposition for ENGAGE — excuse me a second. It’s hot up here. Here’s the patient disposition. 1,653 patients were randomized and 1,647 patients were dosed. Those who discontinued were between 17% to 27%, very similar to EMERGE. And those who withdrew, were between 10% and 14%, also quite low, adding to the robustness of the data that we have in the primary analysis and not different from EMERGE. Going to the primary and secondary endpoints, which were pre-specified, in the high dose group, we do not see an advantage of aducanumab over placebo in CDR Sum of Boxes. In fact, there was a plus 2% change, in MMSE a plus 3% change, in ADAS-Cog, a minus 12% change, so in advantage over placebo and in ADCS-ADL-MCI and minus 18% advantage.
In the low dose group, we do see a small advantage over placebo of a magnitude very similar to the low dose in EMERGE. However this also did not achieve statistical significance. So, ENGAGE has a different outcome than EMERGE.
Here is the primary data for completeness for ENGAGE. And once again you see very minimal differences between that larger data set and the final data set.
Here are the longitudinal profiles for each of those endpoints in ENGAGE, starting with CDR Sum of Boxes, the MMSE, ADAS-Cog 13, and ADCS-ADL-MCI.
Here is the amyloid PET reduction in ENGAGE and it looks very similar to what I showed you in EMERGE with the dose and time dependent reduction in amyloid plaque. However, I would like to point out that in the high dose group, the delta from placebo was smaller than that we had in EMERGE. And I’m showing that here on the right for comparison. And indeed in our Phase 1b PRIME study, we have observed and learned that amyloid PET is quite reflective of the dose that somebody achieves. And so, with that understanding, we looked at the difference in dosing of these two cohorts. And what you can see is that reduction in amyloid PET difference between the two studies is also associated with the difference in dose with ENGAGE having a median cumulative dose at week 78 in this cohort of 126 milligram per kilogram compared to the higher 140 milligram per kilogram in EMERGE.
CFS biomarkers were also assessed. And we see a statistically significant reduction in phospho-tau but which is not dose proportionate in ENGAGE and has a magnitude very similar to the low dose that we saw in EMERGE. In total-tau, there is a numerical difference in low dose, but neither of these had any significance.
We, as I mentioned, did conduct a tau PET substudy in EMERGE and ENGAGE. We’re going to put up this image, which does not come from a study. This is from a previous publication on MK6240, which is [Technical Difficulty] and that we used in this sub-study. The sub-study was conducted at 14 sites in the U.S. and was at the end of the two studies. And so, it’s a small sub-study with 36 patients and we have pooled patients across the two studies for the analysis that I’m about to show you.
We used MK6240 as a second generation tau PET ligand, which has a high affinity for neurofibrillary tangles and low off target binding.
We looked at the three composite regions, I’m going to just spell this in the tau PET sub study. And these were pre-specified and loosely follow the stages of tau disposition in Alzheimer’s disease. On the left is the medial temporal composite where we see both low and high dose having a statistically significant reduction compared to placebo. And then, in the middle, the temporal composite and on the right the frontal composite in both of these, the high dose showed a statistically significant reduction, and the regions comprising those composites are listed below the graph.
I wanted to show something in relation to the video temporal composite reduction in tau PET. The graph on the right is now showing for those individual patients, the reduction in tau in that composite and its association with cumulative dose received by that individual at week 78. And so, what we’re looking at in gray placebo, blue low, magenta high is a correlation between the degree of reduction in tau and the dose received.
And finally, on the tau sub-study, I do believe it’s fair to say that we are still learning how best to analyze tau PET. We know that there are regions that will increase over time and then there is spread and how best to capture this is not necessarily a set science. So, we wanted to show you some representative images from patients. On the left are three patients who were in the placebo arm of the studies; and on the right are three representative patients who were in the high dose group in the PET sub-study. And essentially on the left what you can see here is that from baseline to follow up, there is an increase in the tau PET signal in the amygdala hippocampus and some of the temporal regions in these individuals. And on the right, in contrast from baseline to follow-up, you have a reduction in the tau signal, particularly in the those temporal regions. I forget to mention that because this study was terminated, the time duration between baseline and a follow-up is actually in average of 14 months for these 36 individuals and range from 9 to 20 months.
I’m going to take you now through the top-line safety results from these studies. First of all, a summary of the adverse and serious adverse events, which were well-balanced across arms and across studies. I’m now showing you both studies at one time. The AEs that were reported were consistent with those in Alzheimer’s disease with the exception of ARIA. Ad ARIA was one of the reasons that patients could or would discontinue treatment in the study. And indeed our protocol under certain conditions mandated that discontinuation.
There were 16 deaths across the two studies, a third of these were in the placebo arms and none of these deaths were related to an event of ARIA.
If we look at the adverse events with an incident greater than 10%, ARIA-E is the highest incidence. And the average incident of ARIA was 25% in the low dose and 35% in the high dose. And the next most common being headache, ARIA-H, microhemorrhage and nasopharyngitis.
Taking a closer look at our adverse event of interest, which is ARIA, what we do see is that ARIA is dose dependent, particularly in the high — in the ApoE4 carriers. If you look at EMERGE here, you compare the 29.8 with the 42.5, and as we’ve previously reported, ARIA is greater and equally for carriers than it is in non-carriers. You can compare in the high dose of 42.5 versus the 17.9. The incidence of ARIA in each studies is consistent across the two studies.
The majority of ARIA was asymptomatic. On average, 75% of patients did not report symptoms during their ARIA event. When patients did report symptoms, those are generally mild and they’re listed here, such as headaches, dizziness, visual disturbances, nausea and vomiting. ARIA, as measured by MRI, was transient and generally results within 4 to 16 weeks. And most patients who experienced ARIA were able to continue their investigational treatments. The characteristics of ARIA in EMERGE and ENGAGE are very consistent with what we have reported from the PRIME study.
So, last, I have a couple of slides directed at addressing the key question, which is what is different between EMERGE and ENGAGE. Why was the high dose group in ENGAGE not positive. We have on the basis of believing that having consistent high dose as an important factor, we have been looking at trying to understand if there are subjects in ENGAGE who did have a response.
And I’m going to put up the definition of the population I referred to earlier, which is you can define a population based on their having consented to Protocol Version 4. And the advantage of looking at those patients selected by consent to Protocol Version 4 is that in that population then, since everybody is titrated to 10-milligram per kilogram, we would be assessing the treatment effect under the intended dosing regimen and the intended ARIA management. This then would also preserve the representative population, which is the balance of ApoE4 carriers across the placebo and the Protocol Version 4 population. It would also preserve randomization in this population.
And at the bottom here, I’ve got a graphic describing who is included in each of those populations. And again, if you look at the right, importantly, in the post Protocol Version 4 population, this population did have a higher median cumulative dose of 153-milligram per kilogram. And we’ve defined them by not just that they have consented to Protocol Version 4, but they did so early enough, early enough such that they then had the opportunity for the full 14 doses of 10. That means that the pre-Protocol Version 4 population does include a mixture of individuals. Those who never received 10 and went to 6, that’s the green bar; those who were ApoE4 non-carriers and went to 10, and those who did not consent to Protocol Version 4 early enough.
If we look at those two populations using the heat map I showed you before, here is the pre-Protocol Version 4 population. And again, you see this rate heterogeneity, and very little solid the blue in the heat map. And indeed only 21% of patients in EMERGE and 15% of patients in ENGAGE in that population received the full possible 14 doses of 10.
Looking at those in the post Protocol Version 4 population, and this is smaller because it’s less patients who went all the way out to week 78. You see quite a change. There’s much more solid blue, much less heterogeneity, and indeed an increase to 51% of patients in EMERGE and 47% of patients in ENGAGE in that population had the full possible 14 doses of 10. And that’s why, they had that higher median cumulative dose.
Okay. I want then to show the efficacy in that sub-population that we have formed. First of all, on the table here, this is results I’ve already shown you. This is the overall ITT population of the larger datasets, and we’re looking at the primary analysis on the primary endpoints, CDR Sum of Boxes. If you recall, we had a minus 0.40 in EMERGE, and we had a 2% in the wrong direction in ENGAGE.
Now in the post, PD 4 population at week 78, you have a minus 0.53, which is a 30% change from baseline and CDR Sum of Boxes in EMERGE, and a 27% change from baseline in ENGAGE. Now, this is a sub-group. So, we’re showing here the 95% confidence intervals rather than p-values. I’m also not showing you the secondaries. However, for the sake of time, because this has been a lot of data, but the secondaries are each consistent with this outcome that if you select the population in this manner, you do see an effect in ENGAGE, which is equivalent to that in EMERGE. Here is the longitudinal change from baseline in the post-Protocol Version 4 population in both EMERGE and ENGAGE. And you can see the low and high doses diverging over time from placebo.
Finally, then, to summarize the aducanumab Phase 3 top-line results. Following study termination based on futility, we analyzed a larger data set. And this showed that in Emerge, the high dose reduced clinical decline as measured by the primary and secondary endpoints. In ENGAGE, aducanumab, did not reduce the clinical decline. In a post-hoc analysis, data from a subset of patients exposed to the high dose of aducanumab, support the positive findings of EMERGE.
In subsidies of biomarkers, aducanumab showed an effect on those disease related biomarkers. The most common AEs were ARIA-E and headache. We are currently finalizing the details of a re-dosing study, such that patients who were in our previously enrolled trials will have access to aducanumab, or eligible patients previously enrolled in those studies.
Last but not least, I really must thank an enormous team of individuals and collaborators but most importantly the patients and families who dedicated their time and effort to help us with this undertaking. And listed here are the many groups who supported us in this analysis. Thank you.
Thank you very much, Samantha, for a complete — this is Ron. Can you hear me alright? Thanks Samantha for a very complete recapitulation of the data that were acquired and I think an exposition of what the challenges were and how you dealt with those questions.
So, please continue to ask questions for the panel. At this point, I’d like to ask our panelists for their initial impression of the data and its meaningfulness. Let me start with Paul.
Thanks, Ron. The data is complex. I think, that the futility decision was highly unfortunate and puts us in the situation of interpreting complex data. But clearly, the EMERGE final analysis is positive. The primary analysis of EMERGE is positive. And the analysis of all the key secondaries was consistent and positive. And as important, the biomarker evidence supports the mechanism with significant amyloid reduction leading to tauopathy, a reduction in tauopathy as indicated by both and tau PET and CSF phospho-tau. I think that is a hugely important result and it represents a major advance for the field.
Now, the ENGAGE dataset. The results of the final analysis of ENGAGE are negative. Is this a truly discordant, can we understand this? I think, it’s challenging. But, I believe that as Samantha has presented, when we consider the difference in the timing of enrollment into ENGAGE, relative to the protocol version change and the resulting reduced exposure to the effect of high dose of aducanumab, I think the data from ENGAGE and EMERGE can be considered consistent. And so, I think we are left with an overall positive interpretation of the data from the aducanumab studies, that therefore represents a truly major advance for the field.
Thanks very much, Paul. Let me turned to Sharon. As a clinician, as a PI in this study, what’s your impression?
Yes. I share with Paul, the fact that the positive results first time for Alzheimer’s disease-modifying agent being positive for primary, all secondary, and biomarker endpoint is exhilarating, not just to the scientific community but to our patients as well. And in terms of clinical meaningfulness, I speak to as a clinician, always patient-facing and forever thinking of myself as in the trenches, the ability to hold on to activities of daily living, if you think about a 40% reduction in decline on the ADCS- MCI-ADL, you are talking about people at a mild stage of disease still being able to work, bank, shop, travel, enjoy leisure activities for longer. And I submit to you that this matter is a lot more to our patients than what score they get on a memory test. And I also believe that results on the functional scale of daily activity, translates better to our non-specialists colleagues in medicine, our family physicians who bear the brunt of Alzheimer’s care as well as to the non-medical communities who struggle to understand the clinical meaningfulness.
And I should add — sorry, Steve, one more phrase. The community at large fears Alzheimer’s disease and rightly so, largely because of the loss of autonomy that comes with this disease. So, slowing loss of autonomy for me as a clinician is key.
Thanks very much, Sharon. I appreciate those comments from the frontline. Steve?
Yes. Thanks, Ron. I’m excited to be here discussing the aducanumab results at a pivotal turning point in Alzheimer’s research. I followed 60 patients on aducanumab with many on high dose, open label treatment for upto five years. We also follow the patients clinically. So, we know the families and how they’re doing day-to-day. And I also know a lot about ARIA and how to manage that. The PRIME and ENGAGE studies were very positive experiences at our site for our patients, families and staff, with very low dropout over five years. And we salute the value in patients and their study partners and I ask you to join me in a round of applause for their important contribution.
Now, to the hard question. How do we — that’s easy. The thank you so much to that. How do we make sense of this rich and complicated dataset? And as you heard two important things happened during the trial that makes interpretation difficult. The study was terminated early and the dose was increased for a large number of participants. We know that dose matters with this drug, and probably with this class. And there was substantial dose-related lowering of plaque with positive results across all of the outcomes for one study and not the other. But continued exposure to high dose benefited patients with early Alzheimer’s disease. And this sits with our experience.
As a group, our staff and I — my staff and I felt that our patients on open label treatment were doing better-than-expected over an extended period of time. And we need to drill down and understand who is most likely to respond. So, my feeling is it’s sufficiently positive after careful review of the full aducanumab program, this could represent the first treatment that targets the core pathology and open an era of precision medicine for Alzheimer’s disease. And since use of the drug will require amyloid testing to determine who is likely to benefit, it could also have the positive effects of leading to approval for amyloid PET, which would definitely benefit our patients. So, overall, I think, this is a — Samantha presented a very complicated dataset beautifully. And I think this is a milestone achievement for our field.
A – Ron Petersen
Thanks very much, Steve. And I appreciate everybody submitting questions. We’ve been sort of going through them now and trying to sort them out for a group that seems to coalesce with regards to a particular issue. And a couple are for Samantha with regards to the conduct of the study. Several have asked about Protocol Version 4. That amendment seems to be critical in the overall interpretation of these data. What motivated that, why was that instituted at that point in time?
Samantha Budd Haeberlein
Thanks, Ron. As I mentioned, when we started the studies, we had stratified the dose, such that ApoE4 carriers had the lower dose of 6-milligram per kilogram. But from our PRIME study, we had shown that the greatest benefit was at 10-milligram per kilogram. So, going into these studies, we did believe that achieving that higher dose would be important for efficacy. But, we did not have sufficient information on ARIA to be confident that we could take ApoE4 carriers up to 10 milligrams per kilogram. And it was when we received that data in August of 2016, from the titrating cohort in PRIME that titration itself did lower the incidence of ARIA in exactly ApoE4 carriers. That gave us the confidence. We discussed that with our DSMB and they agreed that we could safely take ApoE4 carriers up to 10 milligrams per kilogram. And I would not recommend changing dose in the middle of a Phase 3 trial. But, it did turn out to be important for this particular study. And if we had not done so, we would not have the results that we have today. So, I do believe that we knew dose was important and we made those changes to get to that dose.
Question about the international nature of the study and where the regional differences, country differences, et cetera. Can you comment at all on that with regard to the pattern of the results?
Samantha Budd Haeberlein
We are not commenting on how anything at that level may play out. What’s important is that in totality as we’ve been discussing in EMERGE, irrespective of region or other slices of the data, the study is overall a positive study.
Several questions have come in regarding the clinical responses in biomarker subgroups and I think those are reasonable questions important questions, but I think the subgroup was sufficiently small for both the CSF and clearly the tau PET to not draw any meaningful conclusions from those. I think the biomarkers were in the expected direction as suggestive. But again, the size of the samples was probably insufficient to really be definitive. Couple of comments have come in. Sharon, you and Steve have commented on the clinical meaningfulness. Is this a big deal? I mean are these — short term 18-months, but I guess the question would be, if we drop the CDR Sum of Boxes by 23% or 22% over 18-months. Is that a big deal?
It is a big deal. Sorry, Steve. I think, you’d agree. Sorry, I was so excited to jump in. I think, it is a big deal. Those of us who know this disease well, know what it means to lose yourself slice-by-slice. And anything you can hang on to and do well is a triumph. And if we’re talking about 25% flowing again on cognitive tests, that’s not the measure I love, although I’m pleased that the CDR Sum of Boxes, which include some functional report as well, caregiver inputs. But, I really go to the functional scale where we’re looking at what people can do. And if I don’t have to give up driving I have an extra year or 2, absolutely, that is meaningful. There’s absolutely no doubt. If I can continue to work and support my family, is that meaningful, I think we all know, it is. You have to think about this disease as a long, slow disease. And if you can slow it, you are winning out. And we start with 80% of people in a prodromal phase of Alzheimer’s, meaning by definition, they were independent in our ADL, maybe a little less efficient, they are doing everything they used to do, but still independent?
Yes. So, this is — we’re looking for a biological foothold against Alzheimer’s that we can build on. And so, these effects are small, but I think they are meaningful. And I hope they’re at the beginning of a process where we can that we can add to.
Paul, as a trialist?
Yes. So, I think there are two points I would make about the magnitude of the effect. One is that as Samantha showed, the exposure was highly variable, of course, across the population. We’re looking at an overall effect size on the outcome measures of 20% to 40% slowing. But I think that we can expect that with more consistent exposure to an effective dose, the response would have been even more favorable.
I would also say that all the data, including the biomarker data, and I agree with small numbers, but I think highly supportive of the hypothesis. All of the data suggests that this is a disease modification, the effect of this drug. And that means that the impact of treatment will continue to accrue with time. The separations from untreated individuals will continue to get larger with length of exposure. That would be the expectation.
Samantha, with regard to those comments, going back to the PRIME data, again, subsets of subsets of subsets in the PRIME data, who remained in the study, but there was divergence, right, of the curves over time. So, the longer — so what Paul is suggesting is that, while there may be perhaps a clinically meaningful difference at 18 months, the curves may be diverging such that if we were to follow those people 24, 36 months, it would be more impressive.
Yes. What we saw in the Prime study, the placebo controlled period was only 12 months. So, we only have the placebo really to look out for the 12-month period. However, those individuals who were in the high dose group of PRIME, so the 10-milligram per kilogram, they set off on a trajectory of improvement versus placebo, and they maintained that trajectory through the 36 and 48 months time points. So, we would hope that the same mechanism of action and trajectory would be apparent in this study too.
Several questions have come in with I think the last comment on your summary side of where you’re going forward now with these participants trying to get them back. What’s the plan there and what kind of data will you be looking for?
Samantha Budd Haeberlein
It’s unfortunate that we terminated the studies. Right? So the patients will have a period of being off of drug, so that we won’t have this continuous period of exposure to aducanumab to assess just the question we were talking about, which is the longer term benefit of aducanumab over time. But we do want to bring back patients such that we are able to look at, despite that period of drug, what are their longer term benefits on efficacy and biomarker endpoints. Plus, there are still a number of individuals who were on placebo at the time that we terminated the studies, and that’s about 400 patients. Those individuals will now have the opportunity to have the 10-milligram per kilogram dose and represents a new population exposed to aducanumab for us to assess those same efficacy and biomarker endpoints.
Just to take the temperature of the participants, which is pretty important, given their contribution. When we inform them about the redosing study, most to a person, every patient and family is very enthusiastic about returning and being part of that. So, they were delighted. So that has not — we’ve many studies that have terminated early recently and we haven’t had that response with any other treatment.
Samantha Budd Haeberlein
And this is exactly the case in Toronto at our site as well. People are more than egger to go back on the drug to come from once a month, and inconvenience aside, there is an overwhelming desire to back on the drug.
Well, I think we’re about out of time here. And again I want to appreciate everybody’s schedule for the rest of the meeting. And I want to thank the panel. I want to thank Samantha for her superb presentation of the data. I’m sure it’s going to be discussed extensively over the next few days as it should. I want to thank Paul, Sharon, Steve for their opinions, and again, CTAD for allowing us. I think this, as I was saying earlier, is a pivotal time for the field. And I think, it will remain with you to interpret these data and draw conclusions from them as you think appropriate. So again, thanks to CTAD. And thank you very much for coming this morning.