Questions and Panelists' Answers—Posted 6 October 2011
Question: What evidence will convince you that the amyloid hypothesis is wrong? A similar question was asked six decades ago to his countryman JBS Haldane about Darwin's theory.
Eric Karran: Proving a negative is normally pretty difficult—I guess the same can still be said of the Darwin/Wallace theory of speciation. The problem is that the most complete test is probably not feasible clinically. However, if it were possible to reduce Aβ production very significantly in AD "at-risk" patients prior to Aβ deposition in the brain, and those patients went on to develop tau pathology and a dementing disease in the absence of Aβ plaques, I think the field may conclude that the genetic forms of the disease do not mirror the sporadic forms.
Question: Rinne et al., 2010, showed that bapineuzumab treatment clearly reduced Aβ load, but cognition decline continued. What does that tell them?
Karran: It is unwise, I think, to conclude very much from the Phase 2 data on small numbers of patients. The full test will come in the Phase 3 trial. However, I also refer to our Nature Reviews Drug Development paper, where I lay out various scenarios for testing the Aβ hypothesis. If one thinks that the data support an Aβ trigger scenario, then it would also predict that Aβ-centric therapies used beyond this point would not work. Thus, it is not inconsistent that bapineuzumab may clear Aβ from the brain (or more accurately, reduce PIB binding) but not affect the disease process.
Question: What about symptomatic therapies? One of the striking aspects of this year's AAIC in Paris was the abundance of serotonergic compounds in early development by big pharma. Do you see a shift in research?
Karran: If Aβ or other disease-modifying therapeutics work, it may increase the numbers of patients with AD-related cognitive impairment. I don't see a shift currently in research; many companies are investing in both symptomatic and disease-modifying therapeutics.
Question: In the April 27, 2011, Alzforum Webinar Reimagining Alzheimer's Disease—Time for Bright New Ideas?, Dr. Herrup at Rutgers on Slide 8 stated that “Site of plaques ≠ site of neurodegeneration.” Today Dr. Karran noted that the location of amyloid plaques and tau tangles do not correlate with the location of neuronal death. Can you discuss why the amyloid hypothesis would have any validity?
Karran: This is addressed in the Nature Review DD article, albeit obliquely. If Aβ acts as a trigger or threshold for another mechanism, then there is no need for it to correlate with tau pathology. It just needs to be present as "aggregate stress" at a sufficient concentration. Then, another mechanism (e.g., complement activation, or inflammatory cytokines, etc.) is able to trigger tau pathology in sensitive neurons. Of course, this is still a massive gap in our understanding, but does remove the conundrum of this lack of correlation. And, the fact that the duration of disease in FAD is not different from sporadic disease, while the age of onsets are radically different, is in keeping with a trigger mechanism (in my view).
Question: How do you account for the long lag time between amyloid-β accumulation and the onset of disease if amyloid-β functions as a trigger? Why is the trigger effect so delayed?
Karran: I do not think the delay is that great. It seems like tau abnormalities (increased pTau) show up in CSF relatively quickly after the drop in Aβ42 in CSF is seen that heralds (presumably) deposition into the brain. However, the very long lag time is between the pathology causing dementia, which seems to be about 10 years. As with the destruction of neurons in Parkinson's disease, the considerable redundancy in brain function permits normal cognition for some time despite the loss of neurons.
Question: Eric Karran’s presentation has a curve for tau pathology to the "right" of the amyloid curve. How might his views change if the tau curve were to the "left" of amyloid, such as that presented by Charles Duyckaerts based upon experimental data (Duyckaerts, 2011).
Karran: Very good point. Please remember that these are diagrams, not data. However, I view this relationship to mean that some amount of tau pathology will continue in the absence of Aβ in some people, but this can be massively accelerated, or initiated, by "aggregate stress."
Question: Could Eric Karran comment further on the balance between Aβ1-42/43 and the shorter forms that was mentioned?
Karran: Basically, it seems from examination of human Aβ in plaques and from preclinical studies that the longer forms of Aβ, which are very much more hydrophobic, will aggregate more readily in brain, and that the shorter forms are able to prevent or protect against this. How is not at all clear.
Question: If Aβ is a trigger, then how will it be beneficial to pursue clearance or reduce production of Aβ?
Karran: It is a question of timing. As shown in Figure 5 of the NRDD article, if Aβ can be affected prior to the trigger point, then it would be expected to have a very dramatic effect.
Question: Why is the genetic disease association of PS1 and PS2 so strong, but undetectable for BACE1?
Karran: There are no easy answers to this. However, the most likely reason is that the genetic variation around BACE is insufficient for it to be picked up. Also, in BACE+/- knockouts, Aβ production is affected very modestly at around a 10 percent reduction. This tells you that the enzyme does not act near Vmax in vivo. Thus, increasing BACE activity might have very little effect on Aβ production in native systems.
Question: Only Randy Nixon has mentioned the genomewide studies, which do not point well to Aβ. Can he expand on this, or others comment?
Nixon: GWAS in AD are identifying a high frequency of genes involved in endocytosis and endocytic sorting. This is not surprising, in my opinion, in light of the important influences the major AD genes have on these same processes, as pointed out in my presented slides (also, Nixon and Yang, 2011). ApoE4, APP (mutations, duplications), and presenilin-1 have effects on endosomes and endocytic function and signaling that likely have pathogenic significance, some of whose effects may be mediated directly by βCTF. Apart from disrupting diverse neuronal functions through these mechanisms, endosomal-lysosomal abnormalities can also potentiate amyloidogenesis. According to the AlzGene top results, some of the strongest genetic associations revealed by the GWAS include BIN1, PICALM, and CD2AP, which have important endocytic roles. Also, CLU and ABCA7, which influence lipid processing, might, like ApoE and cholesterol, promote the endosomal dysfunction seen in early stages of AD through an βCTF-rab5-dependent mechanism, although this needs to be investigated. While it is possible to tie nearly any genetic finding in some way to an influence on Aβ, there is increasing evidence that the genetics of AD are pointing to additional mechanisms relevant to the development of AD dementia, as well as supporting an association of β-amyloidogenesis as a feature of this particular form of dementia.
Comment by: A. David Smith
Posted 27 September 2011
Selkoe’s article is timely and provocative. I am sure that many will agree with his view, “Of salient importance is testing in the mild phase of dementia—or even earlier.” But there appears to be a contradiction between the statement, “Importantly, alternative disease-modifying targets, especially tau and neuroinflammation, should also be vigorously pursued” and the schema in Figure 2, where he proposes that compounds should only move into clinical testing if they have met all the preclinical criteria, including lessening AD phenotypes in mouse models, which are based upon Aβ. Selkoe himself admits that the mouse models do not produce the full spectrum of Alzheimer's disease phenotypes. If we adopt this approach, there is a risk that truly novel therapies will be missed if they have to satisfy an amyloid mouse model.
In spite of disappointments (e.g., over statins), I believe we should not ignore approaches from epidemiology. But I support Selkoe that such approaches should be targeted at those with MCI. A recent example is our VITACOG trial of homocysteine-lowering B vitamins in subjects with MCI. A two-year treatment with folic acid, and vitamins B6 and B12 led to a slowing of the rate of brain atrophy and, in those with elevated baseline homocysteine, to a slowing of cognitive decline and an improvement in clinical status as shown by CDR and IQCODE (Smith at al., 2010; De Jager et al., 2011). The parallel effects on brain atrophy and on cognition suggest that this treatment is modifying the disease process. The results are consistent with an earlier homocysteine-lowering trial in patients with AD, where a slowing of cognitive decline was found in those with mild AD, but not in those with moderate AD (Aisen et al., 2008). Neither of these trials was predicated on prior animal experiments, but arose out of hypothesis-generating epidemiological studies.
Aisen PS, Schneider LS, Sano M, Diaz-Arrastia R, van Dyck CH, Weiner MF, Bottiglieri T, Jin S, Stokes KT, Thomas RG, Thal LJ, Alzheimer Disease Cooperative Study. High-dose B vitamin supplementation and cognitive decline in Alzheimer disease: a randomized controlled trial. JAMA. 2008 Oct 15;300(15):1774-83. Abstract
de Jager CA, Oulhaj A, Jacoby R, Refsum H, Smith AD. Cognitive and clinical outcomes of homocysteine-lowering B-vitamin treatment in mild cognitive impairment: a randomized controlled trial. Int J Geriatr Psychiatry. 2011 Jul 21. Abstract
Smith AD, Smith SM, de Jager CA, Whitbread P, Johnston C, Agacinski G, Oulhaj A, Bradley KM, Jacoby R, Refsum H. Homocysteine-lowering by B vitamins slows the rate of accelerated brain atrophy in mild cognitive impairment: a randomized controlled trial. PLoS One. 2010;5(9):e12244. Abstract
Comment by: Dennis Selkoe
Posted 4 October 2011
Dr. Smith makes good points. I concur that some potentially disease-modifying trials (such as the promising approach of lowering homocysteine levels) that are not necessarily directed at the underlying histopathology of AD may not meet the criteria I set out in Fig. 2, but are nonetheless worthwhile pursuing. Regarding anti-tau approaches, I believe that tau accumulates abnormally and alters its phosphorylation state in some "bigenic" (mutant APP + mutant tau) mouse models. Therefore, tau-modulating agents could be tried preclinically in those models, or even in tau-only transgenic mice, as the agents may be efficacious whether the tau abnormality arises de novo from the tau gene/protein itself or secondary to AAβ accumulation.