Biogen Inc. (BIIB) Earnings Call Transcript & Summary

All right. Good morning. My name is Samantha Budd Haeberlein. And on behalf of my co-authors here, I would like to present the EMERGE and ENGAGE top line results. These were 2 identically designed Phase III studies to evaluate the effectiveness of aducanumab in patients with early Alzheimer's disease. Here are the disclosure of the authors. And I would like to remind you that aducanumab is an investigational compound and is not yet approved in any country. The aducanumab Phase III studies called EMERGE and ENGAGE were 2 18-month randomized, double-blind, placebo-controlled Phase III studies. They recruited 3,285 patients in 348 sites in 20 countries. The population that was enrolled in these studies was patients with MCI due to Alzheimer's disease and patients with mild Alzheimer's disease dementia with a baseline MMSE of 24 to 30 inclusive and a CDR of 0.5. There were 2 dosing regimens that were tested, a low and a high dose, and these were randomized 1:1:1 with placebo. The primary end point was the CDR sum of boxes changed from baseline at 18 months. And there were other key secondaries: MMSE, ADAS-Cog 13, ADCS-ADL-MCI, and there were several sub-studies from biomarkers of disease. The 2 dosing regimens that were selected were chosen at a time when we had understood from the Phase Ib that the safety finding of interest, ARIA, was potentially dose-dependent and was also greater in ApoE4 carriers than noncarriers. And so the doses for EMERGE and ENGAGE were both stratified by ApoE and had a titration phase before reaching target dose. So in the low dose then, if you were an ApoE4 carrier, you were titrated to 3 milligrams per kilogram, and I'm showing that here. And if you were an ApoE4 noncarrier, you were titrated up to 6 milligrams per kilogram. And this low dose regimen was maintained throughout the study. In the high dose regimen, at the beginning of the study, if you were an ApoE4 carrier, you were titrated to 6 milligram per kilogram, and that's the green line here. And if you were an ApoE4 noncarrier, you were titrated to 10 milligram per kilogram, the target dose in the high dose regimen, and that's the purple line. After the study initiated with these dosing regimens, we obtained additional data from the PRIME Phase Ib study that showed us that in a cohort of ApoE4 carriers who were titrated to 10 milligram per kilogram that titration reduced the incidence of ARIA in that population. And so when we investigated this data, we amended the protocols for EMERGE and ENGAGE in early 2017 to take the high dose ApoE4 carriers also up to the 10 milligram per kilogram. The protocol that we made that amendment is called Protocol Version 4. For patients then who were titrated to their target dose, 10 milligram per kilogram for the high dose regimen, then after titration and by week 78, those patients would maximally receive 14 doses of their target dose. The 2 studies, EMERGE and ENGAGE, took 35 months to recruit the entire population. ENGAGE started 1 month earlier than EMERGE and maintained slightly ahead of EMERGE throughout most of the study. There were several protocol amendments through these studies and two that were important to the aspect of dosing. The first of these was Protocol Version 3. And in this protocol, we modified the management of ARIA such that patients with some ARIA we're able to restart at the same dose or to titrate up to the originally assigned dose after pausing for ARIA. At the time that this protocol version was signed, ENGAGE was already 100 patients ahead of EMERGE. The second protocol amendment was Protocol Version 4 that I've just mentioned, where the ApoE4 carriers in the high dose regimen were taken up to the target dose of 10 milligram per kilogram. At the time this amendment was signed, ENGAGE was now already 200 patients ahead of EMERGE in recruitment. Of note, the population that was enrolled, the approximately 50% of subjects in the beginning of the study that were included in the futility analysis for EMERGE and ENGAGE, were complete for their enrollment by June of 2017. And you can see that that's pretty close to the Protocol Version 4 amendment. The implementation of Protocol Version 4 took more than 18 months, such that there were very few patients in the futility analysis who had been included in Protocol Version 4. Now I'm going to go through the EMERGE and ENGAGE top line results. I'll take them in order, first starting with EMERGE. And I'm going to describe the prespecified primary and secondary end points. First of all, just to reflect on the data sets. There is the futility data set, which was an opportunity to complete analysis, so patients who had the opportunity to complete their week 78 visit by December 26 of 2018. And this is approximately 50% of the population, exactly 49% in EMERGE and 57% in ENGAGE. The study continued to enroll patients, continued to collect data from patients in the study per protocol, while that futility analysis was being conducted. And as we know, post-futility, a larger data set became available, and this larger data set had divergent outcomes to futility. In the larger data set, 2 analyses were run: Both an opportunity to complete, so similar to futility, but here, we see that we have 10% or more greater subjects who were able to complete week 78 visit; and an intent-to-treat population, which includes all data, all subjects. In the larger data set, we took the conservative position to censor all data after the March 20 date, i.e., the date -- the day before the futility announcement was announced. And of course, the intent-to-treat population listed to the right are all subjects who were enrolled in the study. At the end of last year, we had database lock for both studies. And so now we have the final data set. And the primary analysis was run on the intent-to-treat population, similar to the larger data set. And any minor differences between these 2 relate to QC and a small amount of data cleaning. So the final data set has the same number of subjects as the larger data set. And then finally, we have sub-studies that will run on key biomarkers of disease, an amyloid PET sub-study where approximately 30% of subjects were enrolled and similar sub-studies in CSF and tau PET. In this presentation, I'm going to focus on the larger data set in the sub-studies, but I will include 2 slides of the primary analysis on the final data set. First then, a reflection on the futility analysis that occurred in March of last year. We conducted a futility analysis that was based on conditional power, and that is the probability that the primary efficacy end point analysis would be statistically significant at final analysis. So it's a projection. It is a future probability. The futility analysis was based on prespecified criteria, and those were both dose arms of both studies needed to have less than a 20% conditional power to meet the primary end point at final analysis. Given that the studies were identical in design, our prespecified methodology included that the conditional power was calculated using a pooled data set from both studies. And that pooled data was to predict the future behavior of the remaining patients. Pooling as a methodology is believed to be a more powerful statistical methodology but does rest on the assumption that the studies are identical. Using this criteria, the futility criteria -- using this methodology, the futility criteria were met. At the time of the futility analysis, however, EMERGE as a study was trending positive, whereas ENGAGE was not. Okay. So now I'll take you through the Phase III top line results for each of the study. Starting with the baseline demographics for EMERGE, you can see that these are well balanced across arms, particularly for age, education and disease stage. The average age of patients at inclusion is slightly under -- at 71 years, which is consistent with this earlier stages of disease. By design, the clinical disease staging at inclusion was approximately 80% of subjects with MCI due to Alzheimer's disease and some 20% of subjects with mild Alzheimer's disease. And here are the baseline disease characteristics, the scores at relevant scores and inclusion, and these were as anticipated. And the average MMSE score at baseline was approximately 26. Here, I'm showing the patient disposition in EMERGE, where 1,643 subjects were randomized and 1,638 subjects were dosed with aducanumab. Between 15% to 24% of patients discontinued treatment through this study, and patients who discontinued treatment were encouraged to stay in the study and to contribute to the continued collection of data. And those visits, while off drug, were included in the primary analysis. Between 7% and 12% of subjects withdrew from the study, and this was a relatively low number for this size and duration of study and contributes to the overall quality of the study with a low level of missing data. This then, on the larger data set, is the primary and secondary outcomes for EMERGE at week 78. And if we focus on the right, on the high dose, we see on the primary end point, CDR sum of boxes, is 0.40 delta change from baseline, which is a 23% change versus placebo, which has a nominal p value of 0.01. MMSE has a 15%. ADAS-Cog has a 27%, and ADCS-ADL-MCI has a 40% difference from placebo in the high dose of EMERGE. If we look at low dose, and we see an intermediate effect on the primary and 2 of the secondaries, or -- however, this is not statistically significant. The placebo decline was 1.74 unit change on CDR sum of boxes over the 78 weeks, and this is well within the anticipated range for this patient population. This next table is the same primary analysis run on the final data. And it is this data set which will be the -- is the basis of statistical inference. And we can see only very minor changes, as I discussed, from QC and data cleaning. For example, the CDR sum of boxes has changed from 23% to 22%, and the MMSE p value has changed to approximately 0.05. And these now are actual p values. Let's take a look at the longitudinal change from baseline in these end points. And these graphs are all somewhat similar with placebo in the gray line showing that trajectory over the 78 weeks. The high dose is in the magenta or purple line, and you can see the change from placebo at week 78. You can see this continued separation over time, somewhat indicative of a disease-modifying mechanism of action. And we can also see the low dose, which is the intermediate between placebo and high dose. Here are each of the 3 secondaries for EMERGE, each showing a similar trajectory, particularly for the high dose, a divergence from placebo over time. Changing then to the sub-study on measurements of amyloid plaque as measured by amyloid PET, and we use a previously published composite of SUVR in composite of cortical regions with whole cerebellum as reference region. And we measured PET at week 26 and week 78. And here, you can see that in the placebo group, there is essentially no change, while we have a dose and time-dependent reduction in amyloid plaque in the high and low dose, with -- at week 78, a change in the high dose of minus 0.272, which is a similar degree of reduction as we saw in the PRIME 10 milligram per kilogram flat dosing at 12 months. We also had CSF biomarkers of tau pathology and neurodegeneration in a quite small sub-study. And here on the left, you can see the CSF phospho-tau measures, where essentially no change in placebo but a dose-dependent and statistically significant change in the low and high dose. And you can see the n numbers underneath, the plots here showing you that this really is quite a small study but, nonetheless, a clear outcome there. However, over in the total tau measure, while we see the same patterning of dose dependency, there is no statistical significance of these results. Okay. So I'll now take you through the second of the 2 studies, ENGAGE, in a similar pattern. First then, the baseline demographics equally are well balanced across arms with perhaps a slight difference from ENGAGE in the number of baseline disease medications used. Equally, the baseline disease characteristics are somewhat similar to EMERGE and are well balanced across arms here. For the patient disposition, ENGAGE randomized 1,653 patients and dosed 1,647 patients. Those who discontinued were between at 17% to 27%, which is similar but a little higher than we had in EMERGE. And the numbers of subjects who withdrew from the study was between 10% to 14%, also still a quite low number and similar to that we observed in EMERGE. Here then are the primary and secondary end points in that larger data set. And if we look over at the high dose, the first thing you see is that we did not see a difference from placebo in CDR sum of boxes. There is a 2% effect there. Equally, in MMSE, we did not see a difference from placebo. ADAS-Cog and ADCS seemed to have some difference from placebo, but there's no statistical significance there. The low dose, on the other hand, we see a minus 12%, minus 6%, minus 11% and minus 18% difference from placebo. And these are not statistically significant, but these values are quite similar to the low dose that we saw in EMERGE. And so out of the 4 arms of these 2 studies, the high dosing in ENGAGE appears to be the aberrant result. Here then the primary analyses run on the final data set. And similar to EMERGE, there really are only very small differences from the larger data set, as would be expected. Here, following the same schema, we have the longitudinal change from baseline in the primary end point, CDR sum of boxes. And the gray is the placebo arm; magenta, the high dose; and the low-dose is in blue. And you can see the high dose does not separate from placebo and is, if anything, slightly above. And here, we have the 3 secondary end points: MMSE, ADAS-Cog and ADCS-ADL-MCI for ENGAGE. If we look at the PET result from ENGAGE. It looks very similar to that which we saw with EMERGE with a dose and a time-dependent reduction in amyloid plaque. In the high dose, this is 0.23 unit difference from placebo, which -- and I'm putting up the high dose results here to the right. There is a difference in the high dose between ENGAGE and EMERGE on amyloid plaque reduction, while the low doses are very similar in both studies. And we already learned from our Phase Ib that, indeed, amyloid plaque reduction was exquisitely associated with the exposure in the Phase Ib. And so we were concerned to understand if there was a difference in ENGAGE high dose exposure. And indeed, if we look in the PET subgroup, at the median cumulative dose by week 78, then we can see a difference in these 2 studies. In EMERGE, it was a median of 140 milligram per kilogram, while ENGAGE, it was the slightly lower number of 126 milligram per kilogram. Equally, if we look at the CSF biomarkers of tau pathology and neurodegeneration in ENGAGE, in the phospho-tau measure on the left, we see statistically significant difference from placebo in both low and high dose, but we have lost that dose dependency that we saw with EMERGE. And this degree of reduction is similar to the low dose that we saw in EMERGE. Overall, in the total tau, the low dose appears to have a similar numerical difference from placebo as we saw in EMERGE. In the high dose, we do not see the same dose dependency as we saw in EMERGE. We also conducted towards the end of the studies a small tau sub-study. And this was using the MK6240 tau PET ligand. And since this sub-study was quite small with 36 subjects, we pooled the data across the 2 studies to assess the change in tau PET measures. Also since this was implemented towards the end of the study, the average duration of individuals in this tau PET study from baseline to follow-up was 14 months but with a range from 9 to 20 months, so not a fixed 18-month as we had originally intended. So there were 3 regions that we assessed with tau PET imaging, and these are composites. And the areas that are included in each of those composites is listed below the graph. And the medial temporal and frontal composites were selected based on the approximate spreading of tau pathology as it is understood in Alzheimer's disease. You can see that these are small sample sizes with an n of 11 in placebo, 14 in the low dose and 11 in the high dose. In placebo, you can see in each of these regions a continued increase in the degree of tau PET SUVR. Whereas in the medial temporal composite, you can see dose-dependent and statistically significant reduction from baseline in tau PET on those subjects treated with aducanumab and a similar outcome in temporal and frontal composites but with statistical significance only at the high dose. Here are some representative images or representative patients from those tau PET sub-study. On the left, we've got 3 patients in the placebo arm and images of their baseline and follow-up. And you can see, if you look at patient 1, for example, that there is an increase in the hippocampal and amygdala regions from baseline to follow-up. Whereas if you look at the high dose individuals on their baseline and follow-up and across all 3, you see generally a reduction in the tau PET SUVR signal in -- particularly in the temporal regions, as apparent in these three. Okay. So that concludes the efficacy results from the Phase III. And I'm going to take us through the safety results from both studies. And here, I'm going to look at both studies side by side. Starting then with a general safety summary. We can see that the AEs and the SAEs are well balanced across placebo and each of the active dosing arms. In general, the AEs that we were -- that we observed, with the exception of ARIA, were those that one would expect in a population of patients with Alzheimer's disease. And listed here are the number of patients who permanently discontinued treatment due to an AE. And ARIA is one of the reasons that patients might have discontinued treatment as per protocol under certain conditions that was mandated. There were 16 deaths in total through these studies, and you can see where they occurred in EMERGE and in ENGAGE, and none of these deaths was considered due -- to be due to ARIA. If we look at adverse events with an incidence greater than 10%, ARIA-E is the adverse event at the top of that list with an incidence of approximately 35% in the high dose and 25% in the low dose. And then the next AEs in this list include headache, ARIA-H, nasopharyngitis and superficial siderosis. Looking more closely at the incidence of ARIA. Here, we've listed and broken it down by ApoE4 carriage. And generally, what we see is that ARIA-E is dose-dependent, and this is clearer in subjects who do have an ApoE4 genotype and less clear in the noncarriers. You can see the 18% in the low dose of EMERGE and 17% or almost 18% in the high dose for the noncarriers. Whereas in the ApoE4 carriers, you see a greater incidence of 29%, almost 30% in the low dose and 42.5% in the high dose, so a greater incidence in ApoE4 carriers and a clearer dose dependency in the ApoE4 carriers. And this was similar across both EMERGE and ENGAGE. If we assess the degree of symptoms related to ARIA, on average, ARIA-E was asymptomatic with approximately 75% of individuals who had ARIA detected by MRI not reporting any symptoms in association with that. When symptoms were present, they were generally mild, and the type of symptoms include headache, dizziness, visual disturbances, nausea and vomiting. By MRI, the ApoE -- the ARIA-E episodes generally resolved within 4 to 16 weeks. And the majority of patients who did experience ARIA were able to resume their treatment with aducanumab. So now that we have looked at the primary analysis, and we can see the divergent outcomes of the 2 studies, what followed for us was an investigation of the potential differences between the 2 studies. And as we had changed the protocol to change the dose midway through the study, and since we also understood from the PET results that there was a slight difference in ENGAGE, we had a focus to understand what was the impact of dose in these 2 studies. One of the analyses that we conducted was using a population defined by their consent to Protocol Version 4, or PV4. Patients selected using the cut-off related to PV4 would then actually assess the treatment of aducanumab under the intended dosing regimen. Since people consented to Protocol Version 4, both ApoE4 carriers and noncarriers would be at the intended target dose. It would, since this is then based on a randomization, continue to both preserve randomization properties, but we would also ensure that we had appropriate representative population, i.e., if the PV4 population ApoE4 carriers continued to consist of the approximately 2/3 of the anticipated population in Alzheimer's disease. So if you look at the schema below, we defined not just the PV4 population based on the actual consent of PV4 but the timing of consent, since we wanted that population to have a full access to the 14 anticipated doses of 10 milligram per kilogram after PV4. So post-PV4, then that population truly should have had the ability or the opportunity to receive 14 doses. And in that population then, the median cumulative dose at week 78 was now 153 milligram per kilogram. Whereas the pre-PV4 population included both individuals who are ApoE4 carriers who had already been randomized to that lower dose, so that's the green bar here, and would only have the opportunity to receive 6 milligram per kilogram. It would also include the ApoE4 noncarriers who already had 10 milligram per kilogram before the PV4 consent, and that's because they, throughout the study, had that dosing. And then there's the intermediate group of the ApoE4 carriers who consented to Protocol Version 4, but they had less -- the opportunity for less than 14 doses. So they had anywhere between 0 to 13 doses of 10 milligram per kilogram. In that group, the median cumulative dose by week 78 was a much lower 116 milligram per kilogram. Dosing has been complex in these 2 studies, and let me take you through this schema, focusing first on the 2 top boxes. What this chart is showing you here with each line is an actual patient individual dosing. So the dark blue are doses of 10 milligram per kilogram. The yellow are doses -- or are no doses, so either having been suspended from dose or skipping doses. The lighter colors, which are in vertical bars to the left, represent the 1, 3 and 6 in the titration schema up to the 10 milligram per kilogram. And those lighter colors, as they appear further than week 20, are due to down-dosing to those lower doses, potentially for individuals who never made it up to 10 milligram per kilogram or needed to down-dose due to incidence of ARIA. And you can see that in this chart that what you really want to achieve is to get to a solid dark blue trajectory. That's somebody who had titration followed by the full complement of 14 doses of 10 milligram per kilogram. But in both studies, only a minority of individuals actually succeeded in having this profile. When we now look at these charts, the top is the pre-PV4 population, and the bottom is the post-PV4 population. You can see the impact of that protocol amendment. In the pre-PV4 patients, only 21% in EMERGE and 15% in ENGAGE actually have that dark blue, the full possible 14 doses of 10 milligram per kilogram. Whereas post that protocol amendment, there's much less heterogeneity and a much larger proportion of subjects, 51% in EMERGE and 47% in ENGAGE, received those full possible 14 doses. If we then look at the impact of the population who did have the opportunity to receive the full 14 doses, we should compare the original outcome for both studies. And here, I'm showing the primary end point, CDR sum of boxes, for both EMERGE and ENGAGE at week 78. And you will recall that there was a 23% and a 2% difference in those 2 studies. And if we now look at the patients who had the opportunity for the full 14 doses in EMERGE, they now have, in the high dose, a 30% difference versus placebo. And in ENGAGE, where we did not have an outcome in the overall analysis in the PV4 population, the high dose has a 27% difference from placebo. And so in these populations, a much more similar outcome can be observed. And here are the longitudinal trajectories of the PV4 population in both EMERGE and ENGAGE. You can see EMERGE on the left, that divergence from placebo, and equally with ENGAGE on the right, both low and high-dose now having a separation from placebo at week 78. So with that, I would like to summarize the aducanumab Phase III top line results. Following study termination based on futility, an analysis of the larger data set showed that in EMERGE, high dose aducanumab did reduce clinical decline as measured by both primary and secondary end points. In ENGAGE, however, aducanumab did not reduce clinical design (sic) [ decline ]. In a post hoc analysis, data from a subset of patients, the PV4 population who had the opportunity to be exposed to high dose, did support the positive findings of EMERGE. In sub-studies, aducanumab showed an effect on disease-related biomarkers, both in CSF and in PET imaging studies. The most common AEs that we observed were ARIA-E and headache. We have now had site activation in the U.S. for the redosing study, which we call EMBARK, and we are proceeding to initiate that study also in the other countries where previous studies of aducanumab were conducted, with the aim to offer access to aducanumab for eligible patients who were previously enrolled in the aducanumab clinical studies. I would like to sincerely thank all the patients and family members who participated in our studies with aducanumab and the investigators and staff who helped us to conduct these very significant studies and equally our DSMB and Steering Committee members who helped us with the conduct design and oversight of these studies. It has been a large collaboration, and there are many individuals beyond the authors who should be thanked for their work for these studies. Thank you very much.