The following article is from my senior colleague, A/Prof Michael Woodward, geriatrician at the Austin Medical Centre, Heidelberg Victoria. I wish to thank Michael for allowing me to post it on my website.
Springfield Advances in Alzheimer’s Treatment conference
Athens, March 2016
Conference report- Associate Professor Michael Woodward AM
As is both expected and delivered, this “first blush of spring” on the Alzheimer’s disease “A” list conference circuit delivered quite a punch. Some 600 delegates, including nearly all of the high-profile well-recognized stars (Winblad, Schneider, Morris, McKeith, Dubois, Masters, Villemagne, Nordberg, Sperling, Ashe, Duff, Gauthier, Hock, Jagust, Mandelkow, Trojanowski- I could go on), assembled in this most significant city to do what Athenians have been doing for about 2,500 years- contemplate the field, stretch the boundaries, define the future and worry little about the financial consequences.
I know many of you are time-poor so let me start this report with 3 key messages, refined by a select group of 6 in the Athina BA departure lounge as the dust settled:
- We have finally unified the amyloid and tau hypotheses. Both matter. There is almost (95%) universal deposition of neurofibrillary tangles (NFTs) in the medial temporal cortex by age 50 but this does not cause any clinical features. With increasing age, driven by risk factors including ApoE4 status, amyloid plaques appear (but not in the same site as the NFTs, initially) and this promotes the tau pathology to spread through the neocortex in the pattern well described by Braak. Only at this stage do clinical features appear- including memory decline. So, amyloid pathology drives tau pathology in likely all with late onset AD. Both matter, and both are potential therapeutic targets. Indeed we may need to target both in the majority of people developing AD, and the earlier the better. But we also need to resist the temptation of abandoning the anti- amyloid therapies, should they fail to deliver in this next round of results. Putting all our bets onto the tau- directed therapies could be equally disappointing. Let us await the results of anti-amyloid therapies used early and in sufficient doses before we completely jump ship.
- Earlier is better. We don’t try to solve the issue of car wrecks just by developing better cars- we try to prevent the crashes. Many folk spoke of the AD continuum but all agreed that neurodegeneration begins early in AD, and we need to apply our therapies early, ideally before this loss of nerve cells. The real issue Is that the earlier we go, the less toxic our therapy needs to be, and the more certain we need to be about the “at risk” population. Biomarkers are likely best at defining this risk of subsequent AD, but are not yet accepted as ideal surrogate outcomes. ARIA- E is more a sign of toxicity than the relatively benign ARIA-H, and probably is a good marker of target engagement in our current trials (no ARIA-E= no effect?) but we need to look for other adverse effects in our preclinical trials, especially with therapies that may have off- target effects.
- Sadly, Aβ may not be the therapeutic target, even in those with familial onset early AD. The failure of the ¥- secretase inhibitors in AD has, even amongst the βaptists, raised the issue of the validity of the amyloid hypothesis. The presenilin mutations are said to be causing an early onset AD that is β-amyloid driven so surely Aβ-directed therapy will work? Well, what if these mutations are causing AD by mechanisms other than by increasing Aβ levels. What if Aβ is a mere epiphenomenum? It may be that whilst these mutations cause a toxic gain of function of ¥-secretase on its APP target, these mutations may also cause the more usual loss of function with other targets- and who buys this mutation causing a gain of function anyway? ¥- secretase has many targets, Notch being one, but neuroprotective targets are most relevant here. If, for instance, these mutations are turning off processes that reduce the production of Brain- Derived Neurotrophic Factor (BDNF) or processes that affect calcium homeostsasis, then monoclonal antibodies against Aβ are unlikely to prevent the onset of disease in these most unfortunate folk. As a DIAN- TU site, where we are using monoclonal antibodies against Aβ in those with these mutations, I sincerely hope these alternative explanations of the effects of presenilin mutations are incorrect.
So to the meat, and it was indeed an almost carnivorous feast for those who are fed by the sustenance of theories, missed explanations, conspiracies and downright good science. Not to forget a driving passion to provide better therapies for our folk with, or at risk of, the various dementias. We have certainly come a long way from the days when this conference was largely devoted to the cholinergic hypothesis, and barely tolerated other therapies (I remember the hushed reverence when nearly 10 years ago, at this meeting at its usual Geneva base, Colin Masters presented ignored the cholinergic focus and presented results suggesting PBT2 may be effective, sadly not replicated). So, where to begin?
Preventing AD
ADAD (Autosomal Dominant AD). John Morris from St Louis reminded us that this accounted for only 1% of AD cases but provides a far greater proportion of the knowledge we have about AD. There is a clear over production of Aβ as opposed to the reduced clearance that occurs in most late onset, sporadic AD. He presented preliminary results from the DIAN (observational) study of 200 folk who mainly had the gene mutations that destine themselves to develop AD before the age of 50. Possibly the most intriguing result was that CSF Aβ is elevated some 30 years before they are expected to develop symptoms- remember this biomarker is reduced in the majority of folk with AD, as it is in these ADAD people about 10-15 years later- still 15 years before symptom onset. Cliff Jack may need to again tweak his curves. Unlike those with late onset AD, multiple pathologies are rarer in this ADAD group, although Lewy Bodies are found- an observation that Ian McKeith found tantalizing. In the DIAN treatment unit (DIAN-TU) some 210 have been enrolled (138 with a mutation- the rest do not have any such risk but are included for comparisons) and we now have 4 year cognitive endpoints, but so far only 2 year biomarker endpoints. The results to date seem to support Aβ- mediated pathology but as stated above that may not be the pathogenic mechanism.
Jessica Langbaum from Phoenix discussed the API-ADAD trial (to be distinguished from the API trial in ApoE4 homozygous individuals, using Novartis drugs and discussed later). There is a region of Colombia with about 25 families with the Presenilin-1 mutation that leads to the early onset ADAD. Of the ~5,000 living individuals in these families some ~1,000 carry the mutation and develop MCI by around age 44 and dementia by age 49. Volunteers- nearly all the cohort now- have been flown to Phoenix for amyloid PET scanning and other evaluation, a testament to how closely and effectively the investigators have worked with the community. Amyloid begins increasing at age 28 and hippocampal volume begins decreasing from age 35. Some are also flown to Boston for tau-PET scanning (more on this technique below) and this begins accumulating from about age 38, after the increase in amyloid begins and unlike in Late Onset AD but consistent with the theory that amyloid drives tau into the neocortex. The therapeutic trial is testing crenezumab, chosen by the subjects as it binds soluble Aβ and can be given subcutaneously, 2nd weekly. Some 300 have been randomized- 100 each with the gene (equal numbers on placebo and active therapy) and 100 non-gene carriers, all on placebo. Following the July 2014 Phase II results with crenezumab in AD it was decided to increase the dose- a common trend with the Roche mAbs as will also be further discussed below. The study will proceed for 5 years and the primary endpoint will be the API composite cognitive score but there will of course be biomarker secondary endpoints. There are many lessons already, including recruitment (TV and other ads were used) and having the participants in this very catholic country agree to the required contraception whilst in the study. And- build a local cyclotron, as they have, rather than fly potential subjects thousands of miles.
Risa Sperling from Boston presented her “A” list, mainly A4 and A5 but with a mention of A3. The preclinical AD study, A4 (LZAZ), is treating those at risk of clinical AD, as defined by normal cognition but a large amount of Aβ on amyloid PET scanning, and is utilizing solanezumab, a monoclonal Ab (mAb) against amyloid, in those so at risk and aged 65-80. The trial is in progress and will severely shake the amyloid hypothesis if it returns negative results in 2 years- but the EXPEDITION 3 study, to be completed in November this year and using the same anti-Aβ antibody, will be almost as concerting if negative, albeit in mild AD cases (so can we can hear the “excuses” already- “we targeted folk too advanced so let us see what LZAZ will hopefully show”). Those who screen fail for A4 (ie no elevation of brain amyloid) are being offered the “LEARN” observational trial. The 4.5 year A5 trial of a β- amyloid cleaving enzyme inhibitor (BACE-I) is a Janssen- funded preclinical AD study (called EARLY) that extends A4 eligibility by 5 years, down to age 60. Because of the lower risk of progression in younger folk, even Aβ positive, those in the first 5 year of this extended age cohort are required to have an additional risk factor (eg family history of AD). Predicting progression, and the effect this risk has on clinical trials, was a recurrent theme at this conference. Too many trials across the AD spectrum enrol non-progressors and this makes it much harder to show an effect of the investigational product.
But once we have demonstrated that a therapy is effective in preventing the development of AD symptoms, does it matter if we then apply that therapy to those who, untreated, may progress slowly, if at all? Risa spoke of the “cholesterol wars” over a decade ago- when it was argued that treating all with elevated cholesterol means that you are over-treating: many of these folk will never develop any consequences of their risk factor. Yet, widespread treatment of this risk factor has been accepted and it is estimated that as a consequence cardiac mortality in the USA has fallen 28%- just through treatment of elevated cholesterol. Similar arguments could be applied to the treatment of hypertension- why await symptoms? Thus, once we have an effective anti-amyloid therapy, why wait for symptoms- why not treat all those demonstrated, perhaps by an amyloid PET scan, to have elevated brain amyloid? Indeed, Hilary Clinton has included in her election platform $2billion to fund amyloid PET scans of all turning 50, with the change used to develop an effective anti-amyloid therapy within 5-10 years.
Risa reiterated our latest conceptualization of the propagation of AD pathology. As stated above, nearly everyone reading this has NFTs in their medial temporal lobe and this may account for you forgetting who is writing this report (who am I?). But do not fear or offend, this may be normal ageing. We should only be concerned if we develop the Aβ pathology that drives tau seeds, and the subsequent NFTs, into the neocortex with resultant development of the clinical features of AD. Again though, this begs the question- do we treat Aβ (and in what form- monomeric, fibrillar, aggregating, plaques), tau (also in what form- normal, aggregating, hyperphosphorylated, truncated), or both? Then of course there are the minor inconveniences of inflammation and other processes that don’t fit into this conceptualization- fortunately they may well be downstream rather than primary.
Keith Johnson from Boston discussed the importance of serial tau-PET imaging to evaluate the effectiveness of new potential therapies in those with preclinical or prodromal AD. If we can demonstrate that these therapies, that may target Aβ or tau, prevent the neocortical spread of tau then we are likely to have an effective drug. It may be wise to first target those with excess medial temporal lobe tau- yes, almost all over 50 have tau there, but some have much more and these may be the group at greatest risk of neocortical spread. Tau-PET can demonstrate those with higher levels of medial temporal lobe tau, so perhaps we should be selecting those into our trials- not those with excess Aβ? He showed that tau-PET correlated well with CSF tau levels. In high amyloid normals (preclinical AD), tau-PET levels correlated much better with the rate of subsequent cognitive decline than in those with low amyloid. He postulated that if we use tau-PET to evaluate responses to therapies, we should concentrate on neocortical levels as these are more likely to decline than the already very high medial temporal lobe tau levels. He also pointed out that medial temporal lobe tau is asymmetrical with usually one side more affected that the other, but the neocortical spread occurs on both sides, not just the side of the most affected medial temporal lobe. A separates section, entirely devoted to tau PET imaging, came later and is described below.
Armand Savioz, Geneva, reviewed the overproduction of Aβ in ADAD, as opposed to the reduced clearance in late onset AD, and noted that this sadly may mean that drugs that are effective in ADAD may not work in the much more common late onset AD. He is a particular fan of retinoic acid, which increases activity of the α-secretase ADAM-10, so may well be effective in preventing ADAD. He noted that retinoic acid may also repair breaks in double stranded DNA but the relevance this has to ADAD is unclear.
Risk factors for AD- potential disease prevention targets
There were several free papers on this. Edwin Tan, ex Australia and now in Stockholm presented meta-analysis data showing that treatment of midlife hypertension reduces the risk of AD by about 9%, and this seems to be true across all antihypertensive therapy classes. We are however less certain of the benefits of treating hypertension later in life. Using the Kungsholmen SNAK study with data on 2701 older subjects, 283 of whom have developed dementia, he showed that those who did develop dementia were more likely to have lower diastolic BPs at baseline and more likely to be on treatment for hypertension, especially a diuretic or beta-blocker and especially if on higher doses of any agent. So is treatment bad- or are any treatment and treatment intensity just a marker of greater baseline risk? Higher baseline systolic BP did seem to protect against dementia in this very elderly population. Ingmar Skoog may not be happy!
Pathways to neurodegeneration
John Hardy from London reviewed the genetic information that we have obtained about AD and other neurodegenerative processes and how this can lead to other prevention targets. The Icelandic/Finnish mutation of the BACE target on APP certainly gives strong support for the rationale of BACE inhibitors- these fortunate folk produce little or no Aβ and consequently have a much lower risk of AD, so the current BACE inhibitor trials (there are about 6 agents in current trials with leading contenders produced by Merck, Janssen and Novartis) have a good chance of success. But there is a world of difference between lacking BACE activity from birth and turning it off, and then not completely, in later life and possibly when neurodegeneration has already begun. The BACE inhibitors may also have off-target effects, an issue that again distinguishes this approach from having the BACE target mutated. This author is however very excited by this treatment approach.
John pointed out that all the AD risk genes identified by genome- wide assessment (GWAS) can be placed into 4 categories- cholesterol metabolism, endosomal vesicle recycling, innate immunity and microglial activation. Of particular interest are variants of the TREM-2 gene that is associated with presenile AD. This gene controls microglial activation in response to lipid and likely Aβ deposition, freezing them in phagocytic mode and preventing them moving into inflammatory mode. Certainly new compounds targeting microglial activation are progressing into clinical trial stages- and may well work independent of amyloid or tau mechanisms.
It is likely that a few common pathways lead to other neurodegenerative diseases and indeed this raises the possibility of a preventative treatment for AD working as prevention for other neurodegenerative diseases, but perhaps we are getting ahead of ourselves.
In Parkinson’s Disease (PD) there are only a few pathways but they are not all the same as those in AD. Inhibition of mitochondrial complex 1 seems to be a target for PD. Interestingly, GWAS shows different results for Dementia with Lewy Bodies (DLB) than for PD and probably PD-Dementia. This seems to be related to the different mechanisms leading to nigrostriatal Lewy Bodies in PD and cortical Lewy Bodies in DLB, and underlines the fact that these are different diseases.
Other neurodegenerative diseases have similar stories- many risk genes, common pathways. In Frontotemporal Dementia (FTD) the genes aggregate to the TDP-43/ lysosomal pathway, abnormalities of which particularly affect cortical pyramidal neurones. So anti-TDP-43 therapies show promise and may not be useful in AD, where GWAS has not demonstrated such a link, even though some cases of AD have been linked with TDP-43 aggregates. This predilection for pyramidal neurones, important to motor function, is of interest as there is a clear link between some cases of FTD and motor dysfunction, including the FTD/ALS disorder. Confusingly however GWAS in FTD/ALS has revealed different gene clusters, affecting autophagy, RNA metabolism and toxic aggregation. It seems likely that these genetic abnormalities affect removal of toxic aggregates via ubiquitin proteasome dysfunction, but it also seems that the causes of motor abnormalities in FTD/ALS are not the same as those causing motor abnormalities in non- ALS FTD. Indeed, as in the subtypes of AD (FvAD, “limbic” AD, PCA and logopenic aphasia) we now conceptualize at least 5 subtypes of FTD that are likely due to different genetic predispositions and disease processes (also a primarily amnestic “limbic” variant, two language variants, motor variant/s and the common behavioural variant) and may well have different treatments. So, whilst there may be a limited number of pathways to neurodegeneration, there are likely to be different pathways in the diseases that are unified just by name (AD, FTD, Lewy Body disorders).
In a later plenary lecture, Nikolas Robakis from New York reviewed mechanisms of neuronal death in ADAD and the therapeutic implications of this. He noted that Aβ is produced in normal folk- so why does it aggregate and become toxic in some? In ADAD it is likely that overproduction overwhelms the mechanisms that keep this peptide safe, but this isn’t true in the vast majority of older people who develop AD, where there is no over-production of Aβ. He is also concerned that AD may not be caused by Aβ itself but, as stated at the beginning of this report, by other mechanisms including the other targets of presenilin in ADAD or the non-Aβ peptides cleaved off the precursor APP protein in late onset AD. All these are potential therapeutic targets.
Microglia in neurodegeneration
Richard Ransohoff from Biogen reviewed everything we did not know about the importance of microglia to neurological degeneration and degeneration. Microglia are crucial to brain development and maturation. In adult life the microglia wrap their processes around the neuronal soma of excited cells, (attracted by ATP elevation), to sooth the cell, settle down excess neuronal activity and allow these neurones to be more “team players”- ie more effective components of neural networks. Aβ impairs the mobility of these processes and essentially isolates that cell from the network, reducing their effective functioning. Aβ sets up a vicious cycle that also prevents these microglial processes phagocytosing amyloid, as do TREM-2 gene variants (see above). This is quite relevant to Biogen (see below), as targeting Aβ may improve microglial activation and thus amyloid removal, and return of the neurone to the neural network.
Anti- Aβ immunotherapies
This remains a major focus of therapeutic trials and with the phase III solanezumab trial in mild dementia due to AD (LZAX, EXPEDITION 3) due for completion in November, with results soon after, we may yet see our first anti-amyloid therapy on the market by 2017, regulatory bodies permitting.
Jan Grimm from Switzerland reviewed the Reverse Translational Medicine approach that has produced arguably the most exciting monoclonal anti- Aβ antibody this year, aducanumab. This innovative approach isolates antibodies from slowly progressing old AD folk, or those who seem to be spared from AD despite their age, then screen these antibodies for their activity against Aβ and produce then a monoclonal antibody (mAb) from the screen successes that is fit for human use. When Biogen presented their PRIME Phase I study results on aducanumab in Nice at the ADPD meeting almost precisely a year ago there was a buzz of excitement- the first ever anti-amyloid therapy to both lower brain amyloid levels and show cognitive benefits, with subsequent analyses suggesting a dose- response effect- more amyloid reduction being associated with greater cognitive benefits. Their Phase III trial is currently recruiting although their insistence on using only certain amyloid ligands is causing difficulties, at least at one site. Aducanumab only binds amyloid plaques- not Aβ monomers. There is some oligomer binding, but no vascular amyloid binding.
Aducanumab is clearly hitting its target as the rates of Amyloid Related Imaging Abnormalities (ARIA) are high. Ping Chiao from Cambridge USA showed that the more therapy-specific oedema (ARIA-E) is occurring in 51% of ApoE4 positive subjects treated with the highest dose of 10mg/kg, and claimed that ARIA- haemorrhages (ARIA-H) are not related to the mAb- just to the disease itself. The Phase I study results have recalibrated dosages of other monoclonal antibodies including Roche’s gantenerumab which has a very similar mode of action to aducanumab, and which has to date been used at doses up to only about 20% of the equivalent of the most effective aducanumab dose. Indeed, this potential under dosage may well explain their Scarlet Road study of gantenerumab in prodromal AD being ceased due to futility- the dose was simply far too low. Roche in fact are extending that study using much higher doses, up to 1200mg (ie about 20mg/kg) but titrated up slowly to try to avoid as much ARIA as possible. So too with their Marguerite Road study in mild dementia due to AD. The DIAN-TU trial is also almost certain to increase the gantenerumab dose to similar levels.
Back to aducanumab, the same Reverse Translational Medicine has developed two other mAbs which target proteins important to dementia- BIIB054 against α-synuclein and BIIB076 against tau. Biogen is also developing the latter and if combined anti- Aβ and anti-tau therapies are required, Biogen could well be soon leading the field.
Other technologies have also developed anti-BACE mAbs and anti-DNA antisense agents that could prevent translation of the genes that produce Aβ.
Tau pathology and tau-based therapies
Whilst the first full day of the conference was dominated by amyloid, the next day restored the balance. But there was still only limited mention of the cholinergic system, despite this dominating earlier conferences.
Luc Buée from Lille reviewed tau transport. Tau is transferred both down and between neurones and the inter-neuronal transfer is clearly a therapeutic target- keeping tau locked in the medial temporal lobe may prevent the spread to the neocortex and the development of AD. I’d never heard of “tunnelling nanotubes” (TNTs) before but these tunnels apparently form across the synapse and allow transfer of substances, including propagating tau, to the next neurone- without any need to become extracellular. Other transfer techniques that could keep tau intra-membrane is via endosomes and exosomes- each with tau nicely bundled up inside a phospholipid layer and spared the ravages of any extracellular attack. However, anti-tau antibodies apparently can still cross into the intracellular space (remember they can cross the blood brain barrier, albeit in tiny amounts, <1% of that administered peripherally). Other approaches would be antibodies to receptors of the tau-containing exosome/endosome, the exosome protein itself or the TNT protein.
Khalid Iqbal, Mr Tau himself- from Staten Island-noted it takes 7-15 years from the beginning of tau’s movement into the neocortex for neurodegeneration to occur. The pathogenic phospho-tau binds to normal tau, converting it into more phospho-tau and the creation of the NeuroFibrillar Tangles (NFTs) that are the hallmark of AD. So this prion-like process proceeds at a rate similar to other prion diseases and also over a similar timeframe to preclinical AD as defined by Aβ accumulation. Interestingly, the forms of phospho-tau differ in different regions of the brain so it is not a single “prion” that is propagated.
Khalid pointed out that there are numerous clinical tau-opathies, including AD, Guam Parkinson’s Disease, Progressive Supranuclear Palsy, Cortico Basal Degeneration and many (other) FTD subtypes. He lay the pathogenesis at the feet of hyperphosphorylation, not the rarer tau mutations (which can occur- eg in Parkinson’s Disease with FTD, FTDP-17), and noted tau mutations are not found in AD. Again, each clinical tau-opathy is associated with a particular type of tau hyperphosphorylation. Tau immunotherapies need to not be specific for the particular hyperphosphorylated tau type found in that condition or affected brain region, but more “generic” to be most effective. Indeed it may also be that Aβ immunotherapies also need to be generic to be most effective, and not target a particular stage of Aβ polymerization , aggregation or indeed Aβ length. He also cautioned against targeting non-phosphorylated tau as that certainly has useful functions and is best left alone.
Khalid went on to describe how his lab had developed anti-tau mAbs (77E and 43D seem most promising to date) and found that in mouse models they also reduced amyloid plaque levels, as well as improving memory. Higher doses reduce not just phosphor-tau levels but also normal tau so lower may be better. He also favours antibodies to the proximal, amino end of tau.
Eckhard Mandelkow from Bonn reviewed tau chemistry- it is, like Aβ, very insoluble and usually not folded. Propagation of tau requires aggregation and he noted this requires more than just the presence of phospho-tau as tau concentrations are very low. Also, there is a “fuzzy coat” around the microtubules (he alikened it to the bristles of a toilet brush) that would prevent pathogenic tau reaching the normal tau that is on the surface of the microtubule. But he feels there is evidence of extracellular tau aggregation somehow precipitating intracellular aggregation possibly through a mediating mechanism- a work in progress.
Eva Mandelkow, in a different session, reviewed the broad spectrum of ways to develop tau- directed therapies. Firstly we can aim to reduce the unwanted effects of tau by stabilizing microtubules, activating proteasomes, increasing autophagy of the lysosomal system or activating chaperone proteins. Secondly we could aim to prevent the pathological modification of tau by inhibiting protein kinases (but she felt this kinase-mediated hyperphosphorylation is no longer a sufficient explanation of tau toxicity), activating phosphatases (? similar concern- we had a negative trial here in Melbourne of sodium selenate that was targeting this enzyme), enhancing a protective modification of tau (? making it less likely to “seed”), inhibiting tau acetylation and inhibiting relevant proteases. Thirdly we could aim to lower tau concentration/production by using anti-sense oligonucleotides, small interfering RNAs or microRNA approaches. Fourthly we could aim to inhibit tau aggregation biochemically including using low molecular weight inhibitors, peptide inhibitors or blockage of what she described as tau “residues” (methylene blue derivatives work this way- and this is being tested in the TauRx trials). Finally, tau aggregation can be reduced immunologically and she felt this was the most promising approach. This final angle includes mAb’s to neuronal tau, antibodies to extracellular tau (eg by preventing microglial uptake or preventing cell-cell extracellular transmission of tau) and neutralizing tau using conformation-dependent antibodies.
Karen Duff reviewed her mouse transgenic tau model- these express the (human) pathogenic tau only in medial temporal lobe areas when young but as they age there is spread to the neocortex and also the amygdala. In these affected areas she is now seeing cell loss which is proportionate to the degree of behavioural and memory changes. She said these changes were equivalent to those in humans classified as Braak Stage III, the earliest stage where symptoms occur. Her lab is also focussing on cell to cell tau propagation and noted that cells have a (pathogenic) tau clearance mechanism, which is only overcome when there is a lot of tau accumulation. Ubiquinated tau, ie tau tagged for clearance, is degraded by the proteasome and, setting up a vicious cycle, this proteasome degradation becomes less effective when the proteasome is exposed to pathogenic tau aggregates/ fibrils. Rolipram increases proteasome function and restores this function when the proteasome is exposed to pathogenic tau, so is clearly a therapeutic approach.
John Trojanowski reiterated some of the above. He said there was no evidence of tau pathology being transmitted by extracellular diffusion. Interestingly, when pathogenic tau is injected into the locus coeruleus, the only site directly connected to it where the normal tau does not undergo prion-like transformation to pathogenic tau is the hippocampus. We are certainly understanding more about how and where tau pathology is transmitted, tau clearance and other modulating effects, the nature of the pathogenic effects of tau and most importantly potential therapeutic targets in this complex chain of events.
Tau Imaging
The Grand Dame of PET (we in Heidelberg, Australia lay claim to the Grand Master!), Agneta Nordberg from Stockholm, reviewed the tau ligands that have been developed for PET imaging. As with amyloid ligands, their specificities vary greatly. Many brain areas show poor correlation between the 3 approaches (FDG, amyloid and tau). The frontal cortex shows best correlation between tau PET and FDG PET (but poor correlation between amyloid PET and FDG PET suggesting amyloid is not the main cause of neurodegeneration that turns off metabolism). It is not simply that tau is a better marker of neurodegeneration and subsequent hypometabolism as many brain areas show poor correlation between tau PET and FDG PET. We need to be aware that the ligand in tau PET scanning is not binding normal tau, only pathogenic tau- the ligand is an aromatic, flat compound that can only insert itself into flat planes between stacks of aggregated tau, just as amyloid ligands are probably binding the pathogenic amyloid plaques. So it is likely that tau-PET is “lighting up” the important areas. Agneta feels FDG PET is highlighting areas more important to cognition than is tau PET. She feels we need to tease out tau PET further, as we have amyloid PET, by correlating the scans with autopsies performed soon after the scan, and with CSF tau results. She somewhat contradicted earlier pathologists by saying that neocortical tau, at least as indicated on PET scanning, can be seen in those with low levels of amyloid. The Grand Dame is even grander as she also moves into astrocyte ligands and imaging inflammation by PET, as well as ligands that are binding to various neurotransmitter systems including 5HT, dopamine and Ach (receptors, both nicotinic and muscarinic, but also related enzymes such as AChE). We may yet find there is no limit to PETting although we need more than foreplay in this desperate endevour. Certainly tau and other PET imaging shows promise as a diagnostic approach and in monitoring responses to new therapies, be they anti- amyloid, anti-tau or anti- astrocyte/inflammation.
William Jagust from Berkley, himself no wallflower in the PET imaging landscape, reviewed his tau ligand, T807/AV 1451. It reveals greater tau in older folk and a Braak- defined spread both with age and in AD. On tune with other speakers and with what I feel was a major message of the conference, he noted that amyloid seems to drive tau into lateral temporal lobes, a pattern not seen in ageing individuals without excess amyloid. Memory impairment correlates best with tau levels in the medial temporal lobe in those who are amyloid PET positive and negative. Global cognitive scores correlate more with neocortical tau but only in those who are amyloid PET positive. These findings support the hypothesis that normal age-related memory loss is amyloid-independent and related to medial temporal (and presumably non pathological tau) but the more global cognitive changes we see in AD are related to amyloid-driven spread of now pathological tau into the neocortex. A publication in Brain that came out the week of the conference (Ossenkoppele et al) presents tau PET findings in 20 AD cases, each categorized as one of the 4 main AD variants- PCA, aphasic, frontal/dysexecutive and amnestic. As expected, tau is distributed differently in these variants and likely represents spread along the most affected neural networks in each variant- the posterior temporal, anterior temporal, frontal and limbic/default mode networks respectively. This is different to amyloid PET distribution and does suggest the spread of tau pathology correlates better with the clinical phenotype than does amyloid pathology- again suggesting amyloid is the trigger, but not the most pathogenic process. Unlike Agneta, he feels tau PET correlates with FDG PET but that tau lights up in larger areas than those showing hypometabolism on FDG PET- so he and Agneta are not fully at odds about whether the two techniques match/don’t match in the areas they demonstrate as affected. He surmises that hypometabolism follows pathogenic tau deposition and that eventually the tau-affected areas will become fully hypometabolic. In the logopenic variant there is left/right asymmetry in tau binding but only in those higher MMSE scores- as the disease progresses, this asymmetry is lost. This has been suspected by neuropathologists but not well described until tau PET- a case of PET predicting the “gold standard”, pathology- and subsequently confirmed. These differences between amyloid, tau and FDG PET all become less as the disease progresses- the end stage of all AD variants looks pretty much the same on PET and clinically eventually all affected individuals do show amnestic, visuospatial, behavioural and aphasic features.
Makoto Higuchi from Chiba, Japan, reviewed the potential role of tau-PET in FTD. He noted that in FTDP-17 the 4RT tau is more widespread thoughout the neocortex than the other tau forms and this both may make the tau PET more positive than in AD but also suggests this tau mutation may spread even without amyloid “driving “it. In PSP the tau is found more in the white matter than in cortical areas but this is still detected by tau-PET scanning, but perhaps best by tau ligands that bind greater to white matter. The tau PET binding tends to decrease as tau levels decrease and this could occur with an effective anti-tau therapy but perhaps also with cell death as a result of neurofibrillary formation, blurring the interpretation of less tau PET signal when such therapies are used.
Finally Keith Johnson from Boston reviewed the early findings with the 18FT807 tau ligand in AD. Consistent with most other speakers in the conference he reiterated that neocortical tau is only found when there is amyloid pathology but, as we see with FDG PET, tau is not found in the sensorimotor cortex. He said frontal tau comes quite late, presumably in the non-frontal variants of AD at least. Nigrostriatal cells and the basal ganglia bind this tau ligand as do areas of micro haemorrhage- he is unsure why. He also feels early tau distribution reflects the phenotype- eg more in posterior regions in PCA.
The interaction between amyloid and tau
Lilly have keen interest in this and sponsored a session with a panel of stars including our own Victor Villemagne.
Mike Hutton from Windlesham in the UK noted that ApoE4 is associated with reduced amyloid clearance- yet another of the many postulated mechanisms of this clear risk factor for AD, but clearly ApoE4 is a therapeutic target and indeed some drugs do modulate ApoE gene expression. He then concentrated on the complex interactions between tau and amyloid in AD, noting that in one APP transgenic mouse model the mice have impaired memory and amyloid deposition, at least in blood vessels, but no increase in neocortical tau and no evidence of neurodegeneration. However usually we do see both pathologies and he postulated that Aβ modulates a complex series of cellular processes that may increase tau pathogenesis. Aβ also initiates an innate immune response and may facilitate tau “seeding”. He felt that rather than separately targeting tau and Aβ we may be able to develop drugs that modulate their interaction.
Victor then drew upon our increasingly rich Australian Imaging Biomarkers and Lifestyle database and noted that only healthy controls that are accumulating amyloid have a relationship between cognitive decline and current amyloid level. Once the SUVR, a measure of the amount of the target pathology on PET, is above 1.4 this correlation becomes apparent- not in those with an SUVR below 1.4. He noted that brain amyloid SUVR is not affected by ApoE4 status unless the individual has Mild Cognitive Impairment (MCI) or is a healthy control accumulating amyloid. He feels this suggests that those who are ApoE4 carriers accumulate amyloid earlier but by the clinical AD stage the ApoE4 negative folk have caught up with the carriers, at least with respect to amyloid accumulation. Whilst we understand the amyloid- centric position of the Heidelberg folk, and of Lilly who have vast resources invested in anti-amyloid therapies (both solanezumab, a mAb , and now their BACE inhibitor in the AMARANTH trial) it does seem that amyloid is integral to disease progression, with higher brain amyloid being highly associated with such progression.
Michael Devous from Philadelphia then moved the discussion back to tau, and the AVID tau ligand. Much of what he said is already documented in this report, from other speakers, including the early asymmetrical deposition of tau and the link between tau SUVR and CSF tau (although he did describe this as a work in progress), but he did add some unique insights. Younger folk who are Aβ positive have greater rates of neocortical tau deposition than older Aβ positive folk, suggesting that older is better with respect to the progress of core pathology in AD, although older folk do more commonly suffer the additive effects of usually multiple pathologies that impact cognition. Medial temporal lobe tau levels predict ADAS-Cog performance, and he feels we can use tau SUVR as a surrogate endpoint in clinical trials as well as a selection criterion into trials, to better select those who are more likely to be “rapid progressers”.
Surrogate Markers of Drug Efficacy
Tau- PET was being touted at the last session, reported just above, as a “surrogate” marker of efficacy of new AD therapies, and this session drilled down further into what that actually means. A surrogate is a complete substitute and are we ready for that yet? Do we want drugs to be judged, and marketed, on biomarker changes even if they do not have evidence of effect on clinical endpoints?
Kaj Blennow from Mӧlndal, Sweden reviewed the relation of CSF markers to drug effects in AD trials. He firstly reviewed the accuracy of CSF markers- now over 140 trials have shown that the typical AD CSF effects (reduced Aβ and increased tau, especially phospho-tau) are robust. He reviewed lab variation and now successful attempts to limit this. It may not be that long before CSF biomarkers are indeed an accurate and well- accepted surrogate endpoint in AD trials. Certainly CSF tau tends to fall in anti-Aβ mAb trials including the Roche Scarlet Road trial of gantenerumab, even though retrospective analysis has suggested the dose was too low. Interestingly in that trial CSF Aβ did not change but that may due to the mechanism of action of this particular mAb- it targets the amyloid aggregates and not the monomers/oligomers. Then he spoke of the concordance between CSF and amyloid PET (87% at baseline in the Merck BACE-inhibitor trial of avagacestat). However in the gantenerumab trial the higher dose group did show a reduction in brain amyloid as measured by PET SUVR, by about 10% overall and up to 20% in those regions most affected by AD pathology such as the precuneus, without any CSF Aβ change.
He went on to note that there are a few individuals who are CSF profile positive for AD but amyloid PET negative- these usually if not always show amyloid accumulation later, but often at an advanced age, and he took this as suggesting that an AD-like CSF profile may precede a positive amyloid PET scan, and thus be an earlier predictor of subsequent AD. A number of synaptic proteins have been evaluated as future biomarkers. These include neurogranulin, increased in AD, and which may predict future synaptic loss and cognitive decline. Target binding of new therapies, including BACE inhibitors in AD, may also be predicted by changes in CSF biomarkers and can be useful if found to correlate with those who respond greater to therapies.
Giovanni Frisoni from Geneva moved the discussion to MRI. Using ADNI data, he is attempting to create a composite MRI score, just as we have composite cognitive/ functional scores. He has analysed 24 month serial MRI data of 150 very mild MCI subjects and included measures of microstructure, atrophy and connectivity using the local vendor sequences- not research acquisition sequences. A work in progress.
Similarly the various PET ligands can be used as surrogates, Agneta Nordberg reiterated as she took the stage again. Interestingly, astrocyte ligands show reduced binding as amyloid levels rise, perhaps reflecting amyloid’s toxic effects on astrocyte function.
Panteleimon Giannakopoulos, delighted to return to his native country but now based in Geneva, reviewed the less universally accepted EEG surrogate markers. Techniques include quantitative EEG (qEEG), sensory-evoked potentials and event-related oscillations, event-related potentials and coherence measures. In the cognitive “n-back” task, which utilizes cortical working memory, not just hippocampal stored memory, MCI decliners can be detected using a combination of these measures with 90% accuracy. Don’t discount EEG yet- it certainly has been used as an endpoint surrogate, even if not the primary, in a number of trials including those using Fortasyn Connect (Souvenaid).
Finally we had the guru of all therapeutic trials perspectives, Lon Schneider from LA, posing the simple question- “Measuring drug effects- do we know what we are doing?” He rightly urged caution in fully abandoning clinical endpoints, and using biomarkers as complete surrogates- but is there such a thing as a partial surrogate? The new consensus criteria for AD now include biomarkers for diagnosis so we can’t ignore their use as outcome markers but we need to be sure of what we are actually measuring and what it means. Nick Fox found this in the original active immunization trial using AN1792 when the expected reduction of hippocampal atrophy wasn’t found- indeed the actively immunized group showed a greater rate of atrophy- and that lead to a reconstruction of what may be actually occurring (reduced-amyloid related inflammation and a reduction of the associated “swelling”).Nevertheless as trials move to earlier stages of the disease process, where the odds of disease modification or delayed progression seem greater, we are more reliant on biomarkers for diagnosis (indeed entirely so in preclinical AD) and presumably as markers of drug engagement and outcomes. Biomarkers can also be used to define disease subtypes- eg the variants of AD and FTD- to better select subjects into trials where the agent may only be effective in some subtypes. But what of those with a clinical diagnosis where the biomarker is absent or inconsistent- eg SNAP (Suspected Non-Alzheimer’s Pathology) in AD. How do we use biomarkers to measure outcome here? We clearly need different biomarkers- eg those of neurodegeneration- but it is quite circular to define a disease with biomarkers then search for other biomarkers if the disease doesn’t show the expected biomarker profile. He noted that several preclinical AD trials now in progress, or planned, use biomarkers for entry- A4 using amyloid PET and trialling solanezumab, DIAN-TU using gene mutations and trialling solanezumab or gantenerumab, TOMMORROW using gene variants (ApoE and TOMM-40) and trialling pioglitazone and the other API trial also using ApoE4 homozygosity to trial the Novartis active immunization compound CAD106 and later also a BACE inhibitor. Biomarkers have also been used to assess futility at trial midpoints, as they will be in the API trial at 2 years. Biomarkers can also be used as surrogate measures of at least risk of being positive for another biomarker- eg ApoE4 status strongly predicts positivity for Aβ. To date regulatory authorities have permitted biomarker use for trial selection/enrichment but have not yet given guidance on their use a an outcome surrogate in preclinical AD trials, so may well reject them if the drug was filed for marketing entirely on biomarker outcome measures. The Lancet Neurology Commission has just released a white paper cautioning too great a dependence on biomarkers if there is no linked clinical outcome.
Emerging Therapeutic Targets
5-HT6 receptor antagonists are likely the next treatments off the block. The closest to trial completion is the Lundbeck drug idalopridine, with three 6 month duration Phase III trials near recruitment closure. The Axovant drug, RVT-101, purchased from GSK, is moving ahead at breakneck speed. There is also a third drug in this class being developed by a small biotech company in India.
Inge de Jong presented the scientific rationale for this treatment approach, and likely underlying the very encouraging Phase II results of idalopridine. She showed that idalopridine potentiates the effect of donepezil on desirable gamma oscillation in regions of the brain important to cognition. This drug also affects other neurotransmitter systems, including glutamate and GABA and these are likely to contribute to its effects- ie the drug is not just working through 5HT6 and potentiating AChEI actions on cholinergic systems. Functional MRI also shows that when idalopridine is used with donepezil there is increased brain activity in regions of the brain underpinning cognition, including the hippocampi, amygdala and striatopallidal regions. These are actions not seen with donepezil alone and suggest that the combination has a unique and hopefully effective mode of action.
The Axovant drug , because it has cholinergic actions, may well work alone- ie not requiring co-administration of a cholinesterase inhibitor. Geetha Ramaswamy, from the company, presented the phase II results from the GSK trial and it is encouraging that both this and idalopridine have positive results in their phase II RCTs. As their drug has cholinergic activity it is also being trialled in DLB where there is a profound cholinergic deficit. It has 5HT2A activity and this receptor may underpin visual hallucinations, further supporting the potential benefit of RVT-101 in DLB. Only the data will tell though.
Souvenaid’s phase III trial in prodromal AD, the LipiDiDiet study, is now completed and the results were released at the conference. Tobias Hartmann from Hamburg presented the rationale for multi-nutrient intervention in prodromal AD- a condition where the brain is under amyloid/tau “attack” and rapidly runs out of the neuronal membrane/ synaptic precursors, exacerbated by the common dietary deficiencies of these in AD patients. Fortasyn Connect, which is the only component of Souvenaid, supplies these polyunsaturated fatty acids, uridine and other nutrients to enable the neurones to form synapses to potentially slow down the effects of the neurodegenerative process. Hikka Soininen from Kupio presented the trial results-this 2 year EU- funded RCT had approximately 300 randomized patients and is one of the first completed treatment trials in prodromal AD, using Dubois criteria. Disappointingly it was negative on the combined neuropsychological endpoint (the NTB) but there were positive secondary endpoints including memory and an almost significant CDR-SB (p= 0.051). The results were more positive in those with higher MMSE scores at baseline, but the numbers were becoming small when one added this subgroup analysis. There was a statistically significant reduction in the rate of hippocampal atrophy in the treatment group, supporting a biological rationale for the cognitive and functional improvements. Reassuringly the placebo and treatment curves seemed to be separating at 2 years on the functional and cognitive endpoints and as the study remains blinded, 3 and even 4 year results will be most interesting- especially to see if the original primary, the NTB, becomes positive. The preparation seemed entirely safe although there were, in absolute rather than percentage terms, a slightly higher number of strokes and other cardiovascular endpoints in the treatment arm. So, the doubters will feel vindicated by the primary but other folk will feel that the encouraging secondaries and the 2 year trend suggest the treatment will be a major step in prodromal AD management.
Other targets that have been subject to the harsh light of clinical research include the pyroglutamate modification of Aβ and animal studies have used inhibitors of glutaminyl cyclase or monoclonal antibodies against this. Amyloid aggregation remains a target using tramiprosate- sadly the phase III trial results were negative but in Apo E4 heterozygotes the results are more encouraging. Iinflammation remains a robust target and masitinib has been used as another approach here. Anavex’s drug, targeting the sigmoid 1 mu- receptor, continues its non-blinded Phase IIa trial with a 52 week open label extension. Preliminary results demonstrate a favourable safety, bioavailability, dose-response curve and tolerability/risk profile. Analysis seems to indicate a cognitive benefit associated with ANAVEX 2-73. Cogstate, MMSE and EEG/ERP improved significantly at 5 weeks of treatment. The effect of changing from low to high dose was statistically significant with MMSE-Δ (p=0.0285) and ERP-Δ (p=0.0168) respectively.
Stem cells
These were darling of earlier AAIC meetings but seem to have dropped off the radar. However at this meeting Tristan Bolmont from Lausanne reinvigorated some enthusiasm, presenting results of trials where mesenchymal stem cells were given into transgenic mice bred to over-express amyloid. There was a clear reduction of Aβ42 deposition. But mouse dementia seems a much easier condition to treat than human AD so watch this space and do not encourage your patients to seek unproven stem cell therapies just yet, despite the current interest and unprofessional promotion of such approaches.
What will be the Best Treatment for AD?
This question was posed to an extremely distinguished panel on the last day of the conference. John Morris returned to introduce the topic and suggested that the numerous treatment trial failures to date may be explained by beginning therapy too late or not targeting all of the several pathologies at play even from an early stage. Bruno Dubois from Paris posed the elephant in the room question- is the amyloid hypothesis wrong? But maybe it is correct and our therapeutic approaches to date have been misguided- for instance, have our therapies been releasing toxic Aβ oligomers from the plaques where they were safely locked up and inactive? The Pfizer/ J&J bapineuzumab phase III trial showed the agent reduced brain amyloid but did not affect cognition or function, and the earlier active immunization trial with AN1792 also showed we can completely remove amyloid, but all died of end-stage clinical dementia. He did however feel that we have a greater likelihood of success if we treat earlier and used the Lilly solanezumab Phase !! EXPEDITION II trial data that post-hoc showed a benefit in those in the milder stages of dementia due to AD. His strength of course is defining diagnostic and research criteria and he noted that such criteria have to account for those with AD-type lesions in the brain but who are, and seem to remain, cognitively intact. Should we ever be treating these individuals, but how do we predict their future stability? And just how great a proportion of the population should we be treating- and at what cost? Some 50% of healthy controls (ie cognitively intact) at age 75 have both amyloid plaques and neurofibrillary tangles. At $20,000 a year, or more, for mAb therapy the figures are startling- and probably explain the pharma interest in the area despite all the negative trials to date. He feels that the best treatment population will be those that have both types of AD lesions but also another factor that drives them or marks them at risk of more quickly developing symptoms- this may be changes developing on FDG-PET, subjective cognitive concerns or ApoE4. In his INSIGHT-AD trial in 218 cognitively normal older (mean age 76) subjects some 27% are Aβ positive on PET scanning but only 2 have converted to clinical AD over an 18 month period- he astutely noted this (relatively well- educated) group would be a disaster in a preclinical AD prevention trial.
It is worth noting that “his” latest criteria now combine prodromal AD and dementia due to AD into a category of “clinical AD”, and many trials now allow both populations, albeit with limits on the numbers drawn from each group (eg in the current Biogen trial of aducanumab, which has closed recruitment to mild dementia due to AD but is still recruiting those with prodromal AD). So the three AD stages are now two- preclinical and clinical, and this makes sense as detecting the arbitrary transition into dementia, based on the onset of functional impairment, was always almost more an art than a science.
Bruno also touched on safety- a huge issue if we are going to subject a very large proportion of the healthy population to anti-AD therapies. Most of the field has been concerned about ARIA but there is consensus now that ARIA-H is a marker of AD, not the therapy (unless it becomes very extensive) and even ARIA-E is usually asymptomatic and can be reduced by slow dose up-titration initially, then if it occurs it can be managed by dose reduction or breaks and then usually resumption of therapy at a lower dose.
Bengt Winblad from Stockholm also returned to the stage and reminded us that non-pharmacological care will always be a part of the best treatment for AD. He also noted that we now have 5 studies published that show a decreasing incidence of AD (from USA, Holland, Sweden, Denmark and the UK) that likely reflect improved lifestyle factors and perhaps more aggressive and effective treatment of cardiovascular risk factors- so these changes need to also be in the equation for “best treatment”- it is prevention! But he did agree that any prevention is best started early- no later than mid-life to have the best chance of success. He did predict that the next generation of effective AD therapies would target amyloid and tau, and some may target both. But there are other targets that hold promise- inflammation, mitochondria, cholesterol pathways, and neuroprotection. He is intensely keen on Nerve Growth Factor trials, having delved into this approach some years ago.
Christoph Hock from Zurich, who has a lot of research and other assets invested in aducanumab, began by showing magnificent in-vitro movies of Aβ aggregating, but did emphasize that it is much more complex in the brain. He reviewed what he saw as promising data from the Biogen PRIME trial of aducanumab showing we can both reduce brain amyloid and improve cognition, and that the two correlated in that trial. Indeed, he seemed to say that this data alone should be enough to have a drug approved for marketing but sadly the authorities still want clinical data! He reiterated that gantenerumab and high-dose bapineuzumab had also been shown to reduce brain amyloid, so we have a clear proof of target engagement across at least 3 monoclonal antibodies now. He also was not concerned by ARIA, and pointed out that ARIA-E, which was found in over 50% of those treated with the highest dose, is both of little clinical concern and also likely is a marker of target engagement of the drug. He is heartened that now 8 mAbs have come to trial, although are 2 now not being developed for AD (bapineuzumab dropped completely and ponezumab diverted to CAA- cerebral amyloid angiopathy). He noted the uniqueness of crenezumab, the only human IgG4 mAb but he also said they all differed in a number of ways, including how and where they bind Aβ. He concurred with others that we need to treat early and felt we had an 8 year therapeutic window, the time from the beginning of Aβ accumulation to the beginning of neocortical tau deposition, as seen in several studies including the Colombian Presenilin 1 mutation familes. He reiterated we need to begin before neurodegeneration, which is caused by tau, not Aβ. He feels therapies (and he favours mAbs) will target the toxic protein across several neurodegenerative conditions- tau, Aβ, α synuclein, TDP-43, C9Orf, SOD and others. It’s the protein, stupid!
Valentine Mantua from Rome and the European Medicines Agency showed how well they understood and were working with the field to develop “the best therapy for AD”. The agency is well aware of the huge amount of resources that to date have been consumed by negative trials, and some of the failures of methodologies so far. The Agency is unconvinced by the artificial boundaries between the 3 stages of AD but does accept that a preclinical stage can be defined. The Agency is not yet ready to accept biomarkers as surrogate outcomes. They are keen on pharma and agencies sharing data and learnings. For instance, this shared data has confirmed that both an amyloid and a tau marker are needed to ensure the correct diagnosis of AD and if CSF tau is not elevated, even though amyloid PET is positive, progression of disease is unlikely so these folk should not be selected into trials and are unlikely to become a treatment population for a marketed drug. Trials that select subjects based on just one positive biomarker have 30% of subjects fail to progress over 2 years so at least 2 biomarkers should be used for trial selection- an important lesson for current trials, almost all of which require only one biomarker. The Agency has also learnt more about the weaknesses than the strengths of the highly- lauded delayed start trial designs- the duration of the delay and the non-inferiority margins are highly problematic. The Agency agrees that ARIA-E can be managed by slow dose titration up and she feels it is more common with mAbs that activate microglia. Whilst most ARIA-E is asymptomatic and reversible, a small proportion is severe and we need to better predict this. She is also concerned that the plethora of new cognitive, functional and composite endpoints need to be validated in independent populations before they are “trial ready”. She is convinced that we are not yet ready to abandon the placebo group, and that to demonstrate that an agent is ready for regulatory approval we need two separate trials across two stages of disease severity- but each trial does not need two primary endpoints. If only one trial is completed, there must be follow-up (~phase IV) data. Unlike the FDA, the Agency does allow pooling of data from different AD stages, perhaps reflecting her earlier comment about the artificial boundaries between these stages. The Agency wants prevention trials to enrich the population through biomarkers or other risk factors and they need to be long enough, unlike treatment trials that can well be shorter if well designed. The goal of prevention trials needs to be delaying cognitive decline, not a biomarker change. She also touched upon BPSD trials- they can target symptom complexes but only if there is a biological rationale for this pooling. There needs to be background non-pharmacological therapy in both arms. She reiterated that the agency is working closely with funders/pharma and this will increase the likelihood that a successful trial, defined by this collaborative approach, will more likely proceed rapidly to marketing approval. Hallelujah!
Lon Schneider returned and to this author’s mind had the final say. He began by simply saying the best treatment will be the one that works. But how well does it need to work? It is noteworthy that trials are being constructed, and reconstructed, to be statistically significant, and thus get a drug to market, even with small and potentially clinically insignificant effect sizes. Indeed the Lilly EXPEDITION 3 solanezumab trial, to be completed in November this year, has just dropped one of its 2 primary endpoints and one could somewhat cynically argue that this is to give it a greater chance of success- easier to improve one than two primary endpoints, as many other completed trials that only hit one endpoint have shown. But Lon would always see himself as a realist, and is no doubt visionary, rather than a cynic. He noted that we now have 390 AD therapy trials completed, in progress or about to commence, testing some 200 agents, and the trend is to design them to show the smallest possible benefit. Back to Lilly- he believes they have secured an undertaking that a 2 year ADAS-Cog improvement of 1.5 points, over placebo, would be sufficient for them to file for listing of solanezumab as an AD- disease modifying drug. Remember the AChEIs, including donepezil, were marketed on a 3-4 point benefit over placebo over 6 months. But they were far more progressed subjects. In question time Ezio Giacobini, the co-founder of this meeting, felt new drugs should be aiming at a 6 point ADAS-Cog benefit over placebo, and he felt that was imminent. In his opinion, the best drug would be better than donepezil and would simply be the first drug next approved by the regulatory authorities.
Lon noted that over 100 Phase IIb/III trials have proceeded on the basis of a negative Phase IIa trial albeit with a post-hoc rationale for the failed trial, and with no clear understanding of the drug mechanism of action- this somewhat gun-ho approach to trial progression is almost unique to this clinical research area. Again, EXPEDITION 3 follows failed EXPEDITION 1 and 2 trials. But this is his view, and not necessarily the author’s! His prediction is that within a few years we are likely to be using one or more of 5HT6 antagonists, LMTN (methylene blue derivatives), solanezumab and still recommending Souvenaid- and here the author is in agreement. He does believe we can develop drugs with large clinically significant effects if we better design trials (which need to be shorter, but better select subjects), pool data (as placebo data is currently being pooled between 6 large pharma and 9 regulatory agencies in an exciting initiative) and of course design better drugs that strongly engage their target. He notes that AD is not a single disease and is too mixed with normal ageing- we need to define the appropriate target/s in an individual and use drugs that engage these targets, probably in combination. He feels we can develop drugs that reduce AD risk in Aβ/tau positive preclinical individuals by 25-30% and that this would be clearly clinically important. He fears that we may lurch away from amyloid as the target, to again waste years on tau as the new target- we need to learn from our amyloid experience if it does indeed disappoint us.
Conclusion
Again an exciting and rich Springfield- albeit the last (in future it will be combined with ADPD). I strongly believe that in the next 5 years these meetings will be focussing on bettering current disease-modifying drugs, and new symptomatic therapies- and will move away from the tau versus amyloid debate that to date has been fuelled by the lack of any effective preventative therapy for this most devastating disease. As one of Bengt Winblad’s patients said, what a shame Alzheimer himself died at 51- if he’d lived a few decades longer he may have cured the disease that was first described by him. Hope springs eternal.