Biogen Inc. (BIIB) Earnings Call Transcript & Summary

Welcome, everybody, to late-breaking readout, roundtable #8. I'm Steve Salloway from Brown University. And we have an exciting program to share with you today. The title is dose and time-dependent changes in plasma p-tau181 in patients treated with aducanumab in the ENGAGE and EMERGE trials. And we're very fortunate to have Oskar Hansson and others from outside the U.S. We lowered the immigration standard. He's back in. And thank God and to all of you, welcome back to the U.S. And so we have Oskar here from Lund University; Gil Rabinovici from UCSF; Jeff Cummings from the University of Nevada, Las Vegas. So I think it's going to be a really great program. None of the participants in the panel have received compensation for participating in this event from the sponsors of this program, Biogen or Eisai. And so the views that we are expressing are our own. The disclosures and each panelist can share theirs. My most relevant disclosures are that I was a site investigator for the ENGAGE trial and the PRIME trial and also the co-Chair of the Steering Committee for the Phase III program for aducanumab and have received consultation fees from both Biogen and Eisai. I don't think they conflict with any remarks I might make, and I'll let the other panelists comment on their own when they introduce themselves. So as I mentioned, Oskar will be presenting these new results, and I think they are exciting about the changes in plasma p-tau with treatment with an anti-amyloid antibody. And Oskar is in a great position to do that, having done really much of the seminal work on plasma biomarkers for Alzheimer's. So Oskar, please.

Thanks, Steve, for that kind introduction. And when it comes to disclosures, I've also been part of a Steering Committee for an ENGAGE and EMERGE. So I will present the talk called dose and time-dependent changes in plasma phospho-tau in patients treated with aducanumab in ENGAGE and EMERGE. And this is just to remind you all that plasma phospho-tau levels are clearly increased in both prodromal Alzheimer's disease in a mild AD dementia and there is accumulating evidence that the levels of phospho-tau changed very early during the preclinical stages of Alzheimer's disease just when amyloid PET imaging starts to become abnormal. And clearly, before tau PET can detect near cortical tau aggregates and well before a cortical atrophy. There's a lot of interest, of course, to try to better understand the relationship between a soluble Abeta and insoluble tau tangles. We did just one study together with Tom Beach, Eric Reiman and others looking at data from new pathology cohort, where that quantified amount of Abeta pathology immigrating as well as tau tangles. And in that study, plasma samples have been collected antemortem. And we found that association between Abeta plaques and the number of tau tangles [indiscernible], were to a very large degree, more than 70% mediated by the levels of soluble phospho-tau. This is just a cross-sectional study, so it doesn't prove any causality, whatsoever. But of course, it could be congruent with the idea that Abeta pathology somehow induces increased levels of soluble tau in the brain, which might facilitate the aggregation and potential loss of spread of insoluble tau. However, if that is true, we would anticipate that, of course, amyloid aggregates are associated with increased levels of phospho-tau, as we know. But if we reduce the number of Abeta fibrils in the brain, you're seeing a drug like aducanumab, the phospho-tau levels should go down. However, if that change is relevant for the disease, we would also anticipate that the change in phospho-tau levels should be associated with a slower worsening of the disease. So the data presented here are from the EMERGE and ENGAGE, just to remind you all that the patients here had either prodromal Alzheimer's disease or mild AD dementia were followed for up to 18 months. And this is just to again remind you that aducanumab very clearly reduces the amount of Abeta fibrils in the brain as detected by a serial amyloid PET imaging. As you can see, both in EMERGE and ENGAGE in both a low-dose and high-dose groups, show in green and in blue. Two years ago, data were presented on the CSF levels for both phospho-tau and total tau in the small sub cohort that had undergone serial lumbar puncture, so CSF had been collected. And the data was very encouraging, showing that aducanumab reduces the levels of phospho-tau in the low-dose and high-dose group, which was significant then in EMERGE, and you also see similar tendencies in ENGAGE. Further, in a small sub-study that undergo serial tau PET imaging, treatment with aducanumab was associated with a slower accumulation of tau aggregates as then detected by tau PET. So the objective of this study was now to see the effect of aducanumab treatment on the plasma levels of phospho-tau181. And the individuals that were selected for this analysis all had plasma samples collected at a 78-week visit and had baseline samples. So this meant more than 6,900 samples were analyzed in these studies. It was a huge effort. And the Quanterix Simoa assay was used to detect phospho-tau181 levels. And as you can see, more than 1,800 individuals in EMERGE and ENGAGE were included in these analysis. So these are just the baseline characteristics of the different groups that we're including this analysis, and there is no differences between these groups, overall, that population selected here were very similar to the overall EMERGE and ENGAGE studies when it comes to baseline demographics. So here are the results. So in the placebo group, where as expected, an increase in the levels of phospho-tau with around 8% to 9% in both EMERGE and ENGAGE. However, we in low dose group has a clear reduction in the phospho-tau181 levels, what was significant in both studies. Further, it was even more accentuated reduction in the high-dose group. So when doing an association analysis or correlation analysis between the change in amyloid PET from baseline to week 78 in amyloid PET imaging compared to the change in phospho-tau over the same time period, these associations were clearly significant in both studies. We have a correlation coefficient of around 0.4. So the more of amyloid PET was reduced, more phospho-tau levels were reduced. And this included both the placebo group and the low-dose and the high-dose groups. However, when stratifying, the treated groups only is here -- the placebo group is not included, but the low-dose and high-dose groups are included. Those individuals that had a normalized amyloid PET scan at the 78-week visit, that is SUVR values florbetapir below 1.1. They have a more accentuated decrease in plasma phospho-tau181 levels compared to those that had SUVR values above 1.1, and this was also clear in both studies. So I guess this might be the most important slide, at least to me, which I think is very exciting, and I think we can discuss a lot. So the greater the reduction in plasma phospho-tau in the treated patients due to then aducanumab, the slower worsening in cognition. And this was found significant for all the 4 clinical outcomes, and this was significant in both EMERGE and ENGAGE studies. So even though the correlation coefficient are not huge, it's still very clear result from this analysis. So to just summarize, so when analyzing about 7,000 samples for plasma phospho-tau181, it was a clear reduction in phospho-tau levels by treatment with aducanumab. Further, the greater the reduction in amyloid PET imaging, the greater reduction in plasma phospho-tau levels. And there's also a significant association between the change in plasma phospho-tau in the treated group with a slower worsening on all 4 cognitive outcomes in both trials. So with that, I want to thank all co-authors and everyone working here with this study and especially, of course, all the patients and caregivers that have participated in these trials. Thank you.

We welcome your -- can you hear me? We welcome your questions. So please come up and use the microphone. And while we're waiting for people to come up, let me ask the panel a couple of questions. I guess I'll start with Gil is, what's your main takeaway from this data? And how do you envision plasma biomarkers like p-tau being used both for screening and as a marker of disease progression and potentially treatment response?

Thank you, Steve. So I'll just start with my most relevant disclosure, which is that I have consulted for Eisai in the past, unrelated to aducanumab. So I think the data that Oskar just presented are very encouraging and fill a gap in that we had very little data from the original top line results from EMERGE and ENGAGE about how amyloid reduction impacted tau biology. The number of participants who had CSF studies or tau PET were very small and not randomly selected. And so I think as we were initially digesting the data, we noted that gap and we noted that plasma biomarkers might really help us here because the plasma had been stored and could be analyzed post hoc. And I think what we see is very consistent with the results that were presented earlier from Trailblazer in the sense that we see that reductions in amyloid PET correlate also and correspond with changes in plasma phospho-tau. I think that this fits very well with the underlying biological conceptual model that we're working with, which is that amyloid in and of itself induces changes in tau phosphorylation and secretion and therefore, reductions in amyloid burden should translate into changes in tau biology. And I think that it's very encouraging to see that and also very encouraging to see that the results align very nicely between the 2 studies, even though the ultimate clinical outcomes were not -- were -- have discordant results, positive in EMERGE, negative in ENGAGE. These biomarker results are very consistent. And furthermore, the correlations with cognition in both studies are very consistent, albeit modest. So I think these are maybe not unexpected but confirmed not only target engagement, which we already knew from the PET sub study, but now modification of downstream tau biology. And we're starting to fill the gaps in terms of how biomarkers might correlate with clinical outcomes by showing the correlations with between phospho-tau and changes in cognition. And so I think we're still missing some boxes here. So I think we would very much like to know how changes in soluble levels of phospho-tau might impact fibrillar tau that associates in tangles. And I think we'd want to know about neurodegeneration and how that was impacted either through bio -- plasma or imaging biomarkers, but we're starting to fill some of these gaps. In terms of how plasma might be used in the future, I think we've seen a lot of data at this conference that suggests that plasma may be very useful as an enrichment strategy, as an initial screen for clinical trial populations. I think these data on the Trailblazer data suggest that someday plasma may be also very useful for looking at target engagement in patients in the clinic and treatment responses. But I don't think we're quite there yet. So if you noticed the coefficients of variation of this test was ranging around 8% or 9%, and the overall magnitude of p-tau reduction was 13% to 16%, and so I think that the signal of p-tau reduction is just beyond the noise of the test itself. And that makes me wonder whether this particular test is yet ready for prime time, something that I can order as a neurologist to monitor potentially in the future of treatment response to anti-amyloid therapy. But I think blood biomarkers, both phospho-tau and some of the Abeta data that we also saw from the lecanemab study are very encouraging and suggest that blood biomarkers are on their way. And of course, it would greatly make biomarker use much more accessible and cost-effective for treatment monitoring if we were able to at least initially monitor our patients with blood rather than CSF or PET.

Maybe, Jeff, I'll ask you this similar question about your initial response -- your main takeaway from this data and how you see the field moving forward with plasma biomarkers like this one.

Yes. Thanks, Steve. Thanks for all of our audience for being here. What a terrific observation this is? I think we have wanted to get to a blood test for so long. And now at CTAD, we have seen 217, we've seen Abeta 42 to 40. Now we see 181, and we see that they're behaving consistently and that they're behaving across different molecules, and I think they're giving us increasing confidence that we'll be able to use it to look at the response to treatment. I think the implication is that they will be useful in diagnosis. We still have some work to do there. They might have prognostic importance in terms of higher tau levels, predicting faster progression or earlier onset of progression. That would be really useful both for trials and for clinical care. So I see the door opening here, Steve, for a wide application of these, and I agree completely with Gil that we're not quite there yet. I think methodologically, analytically, statistically, we still need a lot more observations, but these preliminary observations are really encouraging.

Can you envision a time where we -- again, we make progress, get more data, like you're saying, both of you, where we could use the change in p-tau -- plasma p-tau, the lack of change over whatever, 12 months or 18, whatever the appropriate time period as a marker of nonresponse?

As a marker of nonresponse, no change. Yes, I think that's a conceivable outcome that you would -- that we should be able to begin to expect a corresponding response to a successful intervention. And if that response does not occur, it would predict an unsuccessful intervention.

Okay. Oskar, does targeting amyloid change tau pathology?

Yes, it comes down a little bit to semantics, what we mean with tau pathology in some degree, but it definitely changes the soluble levels of phosphorylated tau, which I personally believe is probably very important in the aggregation process of tau. So I would tend to say yes. But when it comes to exactly the -- and also, I guess they are accumulating, data also showing that these therapies have an effect on serial tau PET imaging, at least ] slowing the accumulation of tau aggregates in the brain. I think it would be extremely interesting to see if these therapies can really even hold or even reduce, but at least what we see, and our experts, seem to be a slowing of tau aggregation over time, especially in areas that are affected later in the disease as was shown earlier today.

Thank you. Your question, [ Anton ]?

So I want to just throw a question to the panel. Obviously, we've had a lot of controversy about the provisional approval based on amyloid lowering properties and the connection of amyloid lowering to clinical outcomes. Where are we at for the possibility of provisional approval based on impact on tau? You showed a correlation between tau reduction and clinical outcomes. So what else would we need to see for that to become a viable pathway forward at some point maybe in the near future?

It's a very good question, I guess. And it, of course, depends a little bit on the requirements you want for a surrogate biomarker for predicting a clinical outcome. To really approve it, I guess, you would like to target the same molecule through another mechanism and see that if you affect that target with another mechanism, you also have a clinical improvement. So for example, if you reduce, let's say, soluble tau levels, we have antisense or nucleotide or something. And if that also associates a change in tau, associates with a positive outcome, I think you would be even more certain that this could be a good surrogate marker. But I guess there's accumulating evidence here, but it might be, definitely. We don't have data suggesting the other way around.

I agree with Oskar. I think it's still early. I think the data are accumulating. I think many of us in the field would feel more comfortable with tau as a surrogate biomarker for cognitive outcomes because of the correlations between the distribution and burden of tau and clinical measures. Neurodegeneration are much tighter, of course, than they are with amyloid beta, at least at the symptomatic stage of disease. I want to say, though, that the magnitude of the correlations was quite modest, right? So the correlation coefficients were in the range of 0.15, 0.2. Now that -- maybe that's not unexpected because we're still -- when we're talking about soluble forms of phosphorylated tau, we're still many degrees from brain function, synaptic function, the biology that matters in terms of cognition. So I don't think it's surprising. I think it's encouraging that we do see the correlations, but we have to remember that we're still measuring that it's something that's quite upstream of cognition and function ultimately in the pathophysiological cascade, at least as we understand it currently.

[ Anton ], of course, always asks the most challenging questions. I would just say that accelerated approval is based on a change that is reasonably likely to predict clinical benefit. That's the actual FDA wording for accelerated approval. And I think we agree that actually tau has been more likely to predict clinical benefit than -- and clinical change than Abeta has historically. So certainly, we need all of the additional data that we're talking about here, but I expect other biomarkers to be predictive and are likely to be the basis of accelerated approval under appropriate circumstances.

Excellent. Question over on the right, my right.

This is a great talk regarding phospho-tau181, 217 as a predictive marker. But one of the major question in Alzheimer's field we have seen, it's a multifactorial disease. It's not Abeta tau, it's alpha-synuclein and TDP-43. Do you guys have seen any effect on those aggregates in these patients? Or if they have some alpha-synuclein and TDP-43 aggregates as well in AD?

It's a very good question. The problem, I guess, it was still we don't have the methods to evaluate that. So we are emerging assays for measuring alpha-synuclein pathology in the sense of seeding aggregation assays. But currently, you need cerebrospinal fluid or you need skin biopsies maybe for that. And also it's not usually quantitative, just binary, yes, no. So it's difficult. TDP, we don't have any at all. So -- but in the future, when these markers are around, it would be interesting to study exactly what you say, but we are not there.

And I think it's a great question, and it also underscores the importance of developing methods to follow clinical trial participants to autopsy. Because ultimately, the way that we'll be able to determine -- hopefully, we'll have useful in vivo biomarkers for alpha-synuclein and TDP-43. But ultimately, what we have now to see how the biology has been affected and will still be valuable regardless of biomarkers is to study individuals after they pass away and understand how the treatment may have modified the biology of the disease microscopically. So I think that's a very important call to the field. And I think it's been said before, but I'll take this opportunity to say it again, we need registries and methods to follow individuals to autopsy.

Question on the left side, my left.

So [indiscernible]. So I think this is really encouraging to see. And to get this link to patient-level treatment effects that are then linked to cognition, we've really been missing that. But I wonder, have you looked even further downstream to look at biomarkers of neurodegeneration, look at NFL, for example?

Exactly. So yes, it hasn't been presented here, but it will be presented later on, but there -- we have such data. And I guess there will be also -- I don't know if it's ready yet, but I guess also other biomarkers like for GFAP, for example, and so on, would be extremely interested to see. And I think that will be analyzed for all of these ongoing studies -- or all these completed studies were plasma samples are available. And I feel we -- it's extremely important.

Stay tuned for those answers. I want to pick up -- I don't see anybody at the mic at the moment. I want to pick up...

Steve?

Sorry. Go ahead.

Yes, this is a question on statistician from Osyu. Although the correlation from a p-value perspective is very strong, but the magnitude generally is about below 0.2. So I'm wondering if the panel can comment on why is the correlation in magnitude is small? And also, is there any kind of a subgroup analysis, a progressive or nonprogressive analysis? Is there any stronger correlation in some of those subgroups?

Gil, do you want to discuss it?

Yes. So I think I may have mentioned, but I agree. I think the magnitude is quite modest. But when we're thinking about the pathophysiological cascade and where cognition sits and compared to phospho-tau, phospho-tau is very upstream. And so if we're correct in our conceptual models of disease, phospho-tau may contribute to neurofibrillary tangle spread, which then conversely leads to synaptic changes and changes in brain function, which we then measure with our relatively course measures of cognition and function. And so I don't think it's a big surprise that the correlations are modest. And I would say that the correlation between amyloid lowering and phospho-tau lowering was stronger, right? So those correlation coefficients were around 0.4. These are more closely linked in the pathophysiological cascade. And so I think at least in terms of ranking the magnitude of the correlations, things have gone in the expected direction. Abeta lowering and phospho-tau lowering are more tightly related to each other than phospho-tau lowering is to cognition. But I agree that ideally, we would have biomarkers that would have even stronger correlations with cognitive outcomes.

I can just add that subgroup analysis are -- we can just mention that these data have only been around for a short time, 3 weeks, something like that. So the subgroups analysis are on its way, and we'll present it at future conferences.

I just wanted to echo this question about subgroups because I think looking at the severity of the amyloid deposition at baseline and how high the magnitude of change at baseline of ptau181, also very important to determine really what we're seeing and what is the relationship of the baseline of the patient when we look at these changes.

No other questions -- oh, yes?

One of the burning questions that's consumed the field is why high-dose aducanumab and ENGAGE didn't show any signal at all. Actually, a slight decrease. And I assume that in the analysis so far of p-tau181 decline, there is no hint at the high-dose ENGAGE behaving any differently than the high-dose EMERGE. Is that right? Or did I miss that?

No, you're right. So in both studies, a great reduction in plasma phospho-tau, the slower the worsening of the disease. And that was clear in both studies. So when there is an effect on phospho-tau levels, there seem to be associated with a slower progression in both studies.

In point of fact, the high-dose ENGAGE didn't achieve its end points and was neutral basically in terms of not showing a change in the clinical study, right?

Exactly. And that is not explained by these findings, I would say. No.

And I think it's important maybe, Oskar, you could explain how the participants were selected for this analysis because I do think it's very important to understand that these were all completers. So if you could explain just...

Yes, exactly. So as mentioned, so the ones that we included were everyone that had plasma samples at 78 weeks baseline. So those were the individuals. So everyone that had completed the study. It's not a full study is cyber.

So that could explain the difference. One of the other observations that's interesting in the CSF data from the 2 studies is that 42 levels rose with treatment in EMERGE. And even though it was a small number of CSFs, I think all of those subjects we got LPs in EMERGE showed 42 levels that went up and the 40 level didn't really change during the 78-week trial. In ENGAGE, that wasn't the case. There were about half the CSFs that were sampled show return of 42 to the normal range and the other didn't. And one wondered again whether that was associated with a very slight difference but meaningful difference in Abeta lowering across the full trials of ENGAGE versus -- there was maybe 12% or 13% less amyloid lowering by PET in ENGAGE and EMERGE. So it's interesting that there are some differences in biomarkers, but it's an important point that in this dataset, which are completers, there's no difference in the effect in ENGAGE versus EMERGE.

Correct.

Correct. So I think we're at the end of our time. I just want to echo a comment that Jeff made about -- I think we're at the beginning. This is my bias, not related to any company, but we are at the beginning of a new treatment era for Alzheimer's. For a long time, we wanted drugs that could lower amyloid. We wanted to be able to measure it first now with PET, and now with a series in looking in plasma and Abeta and in phospho-tau. They're moving in the right direction, and they're giving us an opportunity to really build on this. So I think that the next steps are -- the key is to figure out the next. We're out of time today. But the key is to figure out the next steps, so we can continue to move the field forward. But I think we're in a much different position than we were 10 years or 20 years ago when we were planning and hoping for this to be at this moment. So I want to thank the panel and thank Oskar for presenting and thank the panel for excellent comments.