Immunotherapy for Alzheimer’s Disease

Summary

In 1999 a vaccine approach was found to reduce amyloid deposits in transgenic mice overproducing the amyloid precursor protein. This was followed closely by demonstrations that vaccines or passive immunotherapy could rescue memory deficits in these mice. Initial human clinical trials revealed apparent autoimmune reactions in a subset of patients, but also some cases of cognitive benefit and amyloid clearance. Further work with passive immunotherapy in mouse models confirmed exceptional clearing abilities of anti-amyloid antibodies even in older mice. However, in parallel with parenchymal amyloid clearance was the appearance of microhemorrhages and increased vascular amyloid deposition. Additional clinical trials with passive immunotherapy confirmed occasional appearance of microhemorrhage and occurrence of vasogenic edema in some patients, particularly those with the apolipoprotein E4 genotype. Recent data with positron emission tomography demonstrates trial participants passively immunized with anti-Aβ antibodies have reduced signals with amyloid binding ligands after 18 mo of therapy. Several anti-Aβ immunotherapies have reached phase 3 testing and immunotherapy is likely to be the first test of the amyloid hypothesis of Alzheimer’s disease. Identifying antibody variants that retain amyloid clearance with fewer adverse reactions remains a major focus of translational research in this area.

INTRODUCTION

Alzheimer’s disease is the most common form of organic dementia. It affects roughly10% of the population over 65 and 40% of those over 85 years. In the US, 7–8% of all medical costs are related to dementia. Unlike other leading causes of death, mortality from Alzheimer’s disease is increasing. The number of cases is predicted to escalate dramatically over the next 2–3 decades as success in treating heart disease and cancer permit more individuals to reach the age of risk for dementias.

The definition of Alzheimer’s dementia is based on a case report by Alois Alzheimer (1907) based on a plaque and tangle pathology within the brain of a women succumbing to the disorder in her fifties. Until recently the definitive diagnosis of Alzheimer’s required either biopsy or postmortem histopathology [1]. The introduction of ligands labeling amyloid plaques with positron emitting isotopes and spinal fluid measurements of specific molecules combined with cognitive testing now permits more definitive diagnoses to be ascertained without requiring histopathology [2].

Although debated intensely for the past 25 years, there is now a reasonable consensus emerging regarding the pathogenesis of Alzheimer’s disease. The initiating factor, which appears necessary but not sufficient for Alzheimer’s disease, is the accumulation of amyloid aggregates consisting of the Aβ peptide. The genetics of familial forms of Alzheimer’s disease and Down’s syndrome cases (which result in precocious Alzheimer pathology) have as a common element overproduction of a longer C-terminal form of the Aβ peptide (42 amino acids in length). This form is more prone to form beta sheet structures and aggregate into oligomers and fibrils[3]. Evidence is now emerging that these amyloid deposits may be present up to a decade prior to the initiation of cognitive symptoms of the disorder [4, 5].

A second step in the pathogenesis of the disease is the formation of intraneuronal neurofibrillary tangles formed from hyperphosphorylated forms of the microtubule binding protein tau. Other neurodegenerative disorders can be formed by the tau pathology in the absence of amyloid deposits, but these differ from Alzheimer’s both in clinical presentation and in the location of the pathology regionally in the brain. In tau transgenic mouse models, the tau deposits can be precipitated by intracranial injection of amyloid [6] or breeding with mice producing Aβ deposits[7]. Moreover interrupting the amyloid deposition with anti-Aβ immunotherapy can diminish the progression of tau pathology in mice expressing multiple transgenes[8]. Moreover, the tau pathology appears to be more closely correlated with cognitive status than the amyloid pathology, consistent with it being more proximal cause of the mental dysfunction[9]. By uncertain mechanisms these pathologies result in the loss of synaptic function, synapses and ultimately a loss of neurons leading to considerable brain atrophy. There are multiple hypotheses regarding the mechanistic steps involved, some of which argue the intermediate sized aggregates of amyloid and/or tau, referred to as oligomers, may be more directly causing the toxicity, but demonstrations in vivo are still lacking. Even in the mildest stages of the disorder, the so-called “mild cognitive impairment” phase, there appears to be considerable accumulation of plaque and tangle pathology, and neuron loss[10]. Increasingly these observations suggest that treating the disorder at the earliest possible stages, much as is done with cardiovascular disease, will be essential to controlling Alzheimer’s.

Studies of Active Immunotherapy in animal models

The gene responsible for the protein containing the Aβ peptide is the amyloid precursor protein (APP). APP was cloned in the late 1980’s following the sequencing of the vascular Aβ deposits by Glenner and colleagues[11]. Several years later it was identified that some families inheriting Alzheimer’s in an autosomal dominant fashion had mutations in APP near the cleavage sites releasing the Aβ peptide[12]. This led to a very substantial number of largely failed attempts to develop transgenic mice overexpressing mutant forms of human APP[13]. Finally two models emerged [14, 15] which demonstrated amyloid deposition in locations and with features common to that found in Alzheimer’s disease. Although some claim that certain of these APP transgenic models are superior and other APP models lacking, the remarkable feature is that the patterns of Aβ deposition are largely the same in all models in spite of different mutations and promoters (cortical and hippocampal diffuse Aβ deposits, compacted neuritic plaques and vascular amyloid are found in most APP models and Alzheimer’s tissue[16]). These transgenic mice are superb representations of amyloid deposition. Unfortunately, many authors argue that they are “Alzheimer’s mice” and that treatments effective in the mice should be effective in Alzheimer patients. Unfortunately, the APP mice lack a number of critical features of Alzheimer’s, including hyperphosphorylated tau pathology, and significant neuron loss. At best they resemble an early preclinical phase of the disease, nonetheless one which may be the optimal phase in which to initiate therapy that prevents the disorder.

The seminal paper in the field of immunotherapy for Alzheimer’s disease was published in 1999. Schenk and his collaborators at Elan Pharmaceuticals (nee Athena Neuroscience) demonstrated that treating mice with a vaccine against the Aβ peptide in an APP mouse model prior to the formation of deposits was effective in blocking amyloid accumulation as the mice aged. Within weeks of this publication, our research group in Tampa, Florida began treating APP mice at a later age, when amyloid deposits had already become manifest. One pathogenic component of Alzheimer’s disease pathology was argued to be the inflammation associated with amyloid deposits, and we had concerns that the immune activation associated with the vaccine might aide in clearing the amyloid plaques, but may also cause destructive bystander effects to the surrounding tissue. We also had recently detailed the time course of memory deficiencies developing in our APP+PS1 mouse model [17] as the amyloid accumulated[18]. After several months of repeated immunization we tested the mice for behavioral deficiencies, but found no deficits in either the treated or untreated mice. However, when we continued the treatment to an age when we expected memory deficits to be present, we found the vaccinated mice were protected from these deficits, while mice given a control vaccination were unable to learn the task[19]. A second research group working independently found qualitatively similar effects on memory in a different mouse model[20]. These observations that vaccines or antibodies against the Aβ peptide can rescue the memory deficits in these models have been replicated extensively, in some cases demonstrating that rescue can be seen within days of treatment (before clearance of plaque deposits can be detected; [21, 22].

Importantly, the efficacy of the vaccine in reversing amyloid deposits is not only found in transgenic mice. Nonhuman primates also develop amyloid deposits like Alzheimer patients and associated memory deficits (although like the mice, they do not develop tau pathology).[23]. These can be reversed within months using vaccination against the Aβ peptide[24]. Beagle dogs also develop diffuse amyloid deposits and memory dysfunction with age. Two years of active immunization were found to clear these diffuse (not compacted) deposits, but not have impact on the age-associated memory impairments in these animals[25]. It is important to recognize that most mammals develop age-associated memory impairments and the majority lack amyloid deposition of any type. Hence, memory deficiencies with age are not always associated with amyloid deposits.

The development of novel vaccine strategies and adjuvants against the Aβ peptide has been an area of intense creativity. In most instances, the goal has been to develop B cell activation and antibody production, with minimal T cell involvement (at least against Aβ), due to the adverse events found in human trials with vaccines against Aβ (see below). These impressive approaches have been carefully described in a recent review[26], and are outside the scope of the present review. However, long-term, assuming immunotherapy is successful against Alzheimer’s pathology, these may become standard approaches to preventing the disease.

Mechanisms of Immunotherapy Action in Reducing Aβ Deposits

The success of the immunotherapy approach to brain disorder surprised a number of investigators given the dogma that the brain is an immune-privileged organ with minimal immune surveillance. Nonetheless, the normal range of CSF IgG (5–50 mg/l) is roughly 0.1% of the plasma IgG concentration, with an average concentration of 0.1 μM. Studies in rodent with iodinated anti-Aβ antibodies confirm that 0.11% of the circulating antibody enters the CNS[27]. This has led to several proposed mechanisms regarding the reduction in Aβ deposition associated with anti-Aβ immunotherapy.

The first proposed mechanism relied upon the traditional role of antibody to opsonize antigens leading to macrophage phagocytosis and complement activation [28] (Fig 1). This approach assumes that sufficient antibody enters the brain and binds to the amyloid to trigger this phagocyte action of either resident microglia, or infiltrating monocytes/macrophages. Certainly our own work and that of others has demonstrated decoration of amyloid deposits after systemic administration of anti-amyloid antibodies [29]. One advantage of this mechanism is that the stoichiometry to the antibody to Aβ can be considerably less than 1:1 and still have a meaningful impact on deposition. It also may explain how antibodies selective to one form of Aβ can clear other forms, assuming that heteromeric fibrils containing these different forms are present [30, 31].

Figure 1.

Opsonization and Phagocytosis. Anti-Aβ antibodies bind to Aβ aggregates within the CNS. The immune complex then stimulates phagocytosis by Fc-gamma receptor bearing macrophages. The ingested Aβ is then digested and/or exported from the CNS. Reprinted from Ugen, K. and Morgan, D. DNA and Cell Biology 20:677–678

A second proposed mechanism assumes that the penetration of the antibodies into CNS is not a critical step, and that the presence of circulating antibodies creates a “peripheral sink” which alters the equilibrium across the blood brain barrier for Aβ to favor efflux due to reduced free Aβ concentration in blood [32] (Fig 2). Certainly many antibodies with high affinity for Aβ do dramatically elevate circulating concentrations. There also appear to be mechanisms by which circulating Aβ does cross into the CNS, which would be expected to decline as free Aβ dropped [33]. However, it has also been argued that the major cause of increased Aβ in circulation with antibody administration is due to reduced rates of clearance from the blood [34]. A variant of this idea is that the antibodies in the brain ventricular fluid might bind Aβ and act as a sink within the central nervous system.

Figure 2.

Peripheral Sink. Circulating anti-Aβ antibodies bind free Aβ in the blood and increase the efflux of Aβ down its concentration gradient and/or block the influx of Aβ from the circulation and back into the brain. Reprinted from Ugen, K. and Morgan, D. DNA and Cell Biology 20:677–678

A third option is that antibodies act catalytically to modify the secondary structure of the Aβ monomers into a conformation that is less likely to form the aggregates associated with oligomeric or fibrillar forms (Fig. 3). This was first proposed by Beka Solomon and colleagues before the capacity of antibodies to affect Aβ in vivo was demonstrated [35, 36]. She has demonstrated that stoichiometries of antibodies as low as 1:10 effectively blocked Aβ fibril formation in vitro. We have confirmed this and extended the results to indicate that stoichiometries of antibody to Aβ as low as 1:1000 can specifically modify the secondary structure of Aβ and block fibril formation using ATR-FTIR and atomic force microscopy[37]. Obviously this catalytic modification hypothesis benefits from being active at low stoichiometries and may be active in the absence of macrophage activation, a circumstance that might still lead to adverse bystander effects.

Figure 3.

Catalytic Modification of Conformation. Antibody binds Aβ and modifies the secondary structure to one which minimizes the formation of aggregates. Reprinted from Ugen, K. and Morgan, D. DNA and Cell Biology 20:677–678

A fourth mechanism has been suggested to be active at least for intracranially administered antibody. That is neonatal Fc receptor (FcRn) medicated efflux of antibody antigen complexes across the blood brain barrier[38]. This mechanism was shown to be saturable and inhibited by Fc fragments. One possible consequence is that excess antibody may lead to competition within this efflux mechanism by increasing transport of antibody without bound antigen. Our own work and that of others has found that excess anti-Aβ monoclonal antibody administered systemically can inhibit clearance of amyloid plaques [39, 40].

Critically, none of these mechanisms are mutually exclusive. Moreover, different antibodies may utilize different mechanisms to different degrees. For example, antibodies not binding fibrillar amyloid may be less able to induce phagocytosis, yet function exceptionally as a peripheral sink. Some antibodies upon binding Aβ may not modify the secondary structure, but still activate macrophages and FcRn mediated transport. The extent to which different antibodies utilize different mechanisms may also confer different adverse event profiles for each of the antibodies.

Active Immunization (Vaccine) Clinical Trials

Given the remarkable success of vaccination in animal models and the absence of any alternative disease modifying therapy for Alzheimer’s disease, Elan partnered with Wyeth to initiate an active immunization trial against the Aβ peptide in Alzheimer patients. The vaccine consisted of the full length peptide, which contains both B and T cell epitopes, aggregated into fibrils. Initially about 60 patients were treated with one or more doses of the vaccine and the cases monitored in a phase 1 safety trial. One of the initial observations was a variable antibody response, with many patients failing to develop detectable titers against the antigen. Approximately half way through the trial there was a change in the adjuvant to QS-21 in an attempt to enhance this response. There were no adverse responses detected during the trial portion of the phase I study[41].

This led Elan and Wyeth to pursue a phase 2a trial with the AN1792 formulation including the QS-21 adjuvant. The goal was to immunize patients to a preset antibody titer using multiple inoculations. Within a short time of initiating the trial however, the immunizations were halted due to a fraction of the patients developing aseptic meningoencephalopathy, an inflammatory reaction in the CNS, which in most instances responded well to steroid therapy. Subsequently, one case from the phase 1 trial also developed these symptoms of CNS inflammation and died from pulmonary embolism a short time later[42]. A second case from the phase 2a study was also collected[43]. In both cases, a T cell infiltrate of the CNS was apparent with signs of meningeal inflammation. The T cells did not appear directed to the amyloid deposits, and there was no relationship between antibody titer and incidence of the adverse event, but it was concluded the response was an autoimmune reaction caused he vaccine[44].

In spite of the interruption of the trial, some cases did develop significant antibody titers. In the cohort evaluated in Zurich, these titers were measured using a brain amyloid histology assays (TAPIR) without breaking the blind[45]. Cognitive function tests in this cohort (35 overall) indicated that those patients with the highest titer declined less over the ensuing year than those without detectable antibody titers. Some of these cases remained stable for years after the original treatments. While admittedly anecdotal, these observations implied that the immunotherapy approach may have benefits in spite of the adverse events observed.

A more complete follow up of the patient cohort studied in the truncated phase 2a vaccination trial has been conducted at 4–5 years post immunization. This study examined cognitive test performance of 25 original cases that responded to the immunization (reaching a titer of 1:2200) and had 129 cases overall. A significant difference in Disability Assessment for Dementia (an activities of daily living measure) and the Dependence Scale were found in the initial responders indicating a benefit of the treatment. Although no differences in the rate of brain shrinkage by MRI or cognitive performance were observed, these results are encouraging that some benefit could be obtained from a trial that was suspended before reaching the intended duration of therapy.

Studies of Passive Immunotherapy in Animals

The combination of low responsiveness to vaccines, and the appearance of T-cell dependent adverse events with active immunization quickly led multiple groups to evaluate monoclonal antibody treatments to clear brain amyloid deposits. The first publication of the success of this approach[28], was followed shortly by others[32]. Our own work in this area was facilitated by a collaboration with a small biotechnology start-up in the San Francisco are, Rinat Neurosciences. Rinat shared with us antibodies they had developed against specific Aβ epitopes which we then tested in our murine models of amyloid deposition (using research support from the NIA/NIH). Our first studies examined the effects of intracranially applied antibodies, and detailed a rapid time course of changes in amyloid deposits and microglial activation. In series of papers we found that diffuse amyloid was cleared within 1 day, while congophilic deposits, associated with microglial activation, were cleared within 3 days. By one week the regions infused with antibody were largely cleared of deposits and free of antibody[46, 47]. We further found that systemic administration of antibodies in aged amyloid depositing APP mice (18–22 mo; analogous to 65–75 Y old humans) caused a time dependent clearance of pre-existing deposits that after 3 mo reduced the congophilic compacted plaques by 90% compared to mice given a control antibody[29]. However, a prior brief report had indicated that passive immunotherapy caused appearance of microhemorrhage in aged amyloid depositing mice[48]. When we stained sections for hemosiderin, we found a time-dependent increase in the number of microhemorrhages associated with the antibody-mediated removal of the parenchymal amyloid deposits[49]. Moreover, we were the first to report an increase in the number of vascular amyloid deposits, and association of the hemorrhages with these deposits. However, remarkably these aged mice that had held amyloid in their brain from over half their lifetimes and endured microhemorrhages, still obtained the benefit of restored memory performance with the antibody treatment. Subsequently, others confirmed that treatment with anti-Aβ antibodies in older mice that already possess deposits when treatment is initiated (a therapeutic rather than prophylactic model) leads to microhemorrhage development [50, 51].

After confirming our observation that Rinat’s antibody caused microhemorrhage, we cautiously advised them of this significant concern. This led them to modify the antibody to minimize interaction with effector proteins such as Fc-gamma receptors and complement proteins (by enzymatic deglycosylation). When we tested this antibody using intracranial administration, it was capable of removing amyloid deposits without causing activation of microglia[52]. Further tests using systemic administration showed dramatically reduced formation of microhemorrhages and a smaller increase in vascular amyloid deposition with the modified antibody. Although there was a slight reduction in amyloid clearance, both the modified an native antibodies rescued the memory deficits in the aged amyloid depositing transgenic mice[53].

One potential explanation for the increase in microhemorrhage is that antibodies capable of binding amyloid fibrils in the vasculature can cause a local inflammatory reaction. Ultimately this may weaken the endothelial cell barrier and result in vascular leakage. A second possibility is that the opsonization of parenchymal Aβ deposits leads to macrophage phagocytosis which then redistributes the phagocytosed material to the vasculature. Fiala and colleagues used an in vitro blood brain barrier to demonstrate that macrophages with engulfed amyloid die when attempting to pass through the endothelial cell barrier[54]. This leads to local deposition of the amyloid material they have phagocytosed. By reducing the activation of effector proteins, the deglycosylated antibody used in our mouse work may diminish the formation of vascular deposits and microhemorrhages by reducing the extent of microglial/macrophage activation and phagocytosis. However, effects mediated by the peripheral sink and the catalytic modification of Aβ would be retained. Our results with Rinat’s modified antibody led to their acquisition by Pfizer. A humanized version of this antibody is presently in phase II clinical testing (ponezumab; Table I)

Table 1.

Immunotherapeutic Approaches in Clinical Development

Company	Approach	Abeta Epitope	Biological	Stage
Janssen/Wyeth (Elan)	passive	N-terminus	Bapineuzumab	Phase III
Eli Lilly	passive	central domain	solanezumab	Phase III
Baxter	passive	IVIg mix	Gammaguard	Phase III
Janssen/Wyeth (Elan)	active	N-terminus	ACC-001	Phase II
Novartis	active	N-terminus	CAD106	Phase II
Pfizer	passive	C-terminus	Ponezumab	Phase II
GSK/Affiris	active	Abeta mimetic	Affitope AD1 and AD2	Phase II
Roche	passive	N-terminus + central domain	Gantunerumab R1450	Phase I
Merck	active	conformational	V950	Phase I
GlaxoSmithKline	passive		GSK933776A	Phase I
Janssen/Wyeth (Elan)	passive	N-terminus	Bapineuzumab s.c.	Phase I
Eisai/BioArctic	passive	protofibrils	BAN2401	Phase I
Abbott	passive	conformational		preclinical
Elan/Wyeth	passive	conformational	AAB-002	preclinical
Genentech/ACImmune	passive	conformational		preclinical
BiogenIdec/Neurimmune	passive			preclinical
Boehringer/Ablynx	nanobodies	Abeta		preclinical

Passive Immunization in Alzheimer Patients

Given the issues associated with active immunization, several companies have started clinical trials using monoclonal antibodies against the Aβ peptide. By far the most advanced of these is an N terminal specific antibody from Elan-Wyeth referred to as bapineuzimab (now in development by Wyeth-Pfizer and Janssen). The phase 1 trial of this antibody used single administrations of 0.5, 1.5 or 5 mg/kg of the antibody[55]. Patients were tested cognitively both before administration and 16 weeks later. With respect to safety, 3 of 10 patients at the highest dose developed vasogenic edema by MRI. One patient developed a punctuate MRI hyperintensity that was not present prior to the treatment and remained 12 mo later. This was interpreted as a microhemorrhage. Although not designed to measure efficacy, exploratory analyses found that the MMSE values improved in the 0.5 and 1.5 mg/kg doses, with no benefit found in the highest dose. This is similar to our dose-response data mentioned above, with benefits to APP mice at 3 and 10 mg/kg of antibody and no benefit at 30 mg/kg[39]. The causes for these atypical dose responses are not clear, although saturation of FcRn transport is one option. A similar inactivity at high doses was also reported previously in abstract form[40].

A phase II clinical trial with bapineuzimab has also been completed. The study included 230 patients and 4 doses of the drug (0.15, 0.5, 1 or 2 mg/kg) receiving 6 infusions 13 weeks apart[56]. Vasogenic edema was found in roughly 10% of the cases treated with bapineuzimab, with half of these being symptomatic. There was propensity for the vasogenic edema to be more common in the higher dose groups and in apolipoprotein E4 carriers. Intention to treat analysis failed to identify a significant cognitive benefit of bapineuzimab (all dose groups combined), but there were significant benefits in patients that completed all 6 antibody infusions and in patients that were non-apolipoprotein E4. Thus, in spite of absence of a dose response effect, a phase III trial is presently underway, with a separate analysis of E4 carriers and non-carriers. It should be noted that prior studies also identified less efficacy of treatments in apolipoprotein E4 carriers for tacrine [57] or rosiglitizone[58].

Pathological Analyses of Vaccinated Patients

Roughly 8 years have passed since the first patients received anti-Aβ vaccines. A number of these cases have now come to autopsy. The Southampton group led by James Nicoll have collected tissue from a number of patients (up to 12 at present) who were in the phase I trial and have issued multiple reports on analyses from these specimens[59–62]. Based upon the time post vaccination, they describe the following sequence of events in the small patient population they have. At relatively short intervals (less than 1 Y) they observe the presence of ragged “moth-eaten” amyloid deposits and the presence of microglia with ingested immunopositive amyloid. Between 1 and 3 years post immunization they report a partial clearance of amyloid deposits (compared to age-matched Alzheimer samples that were not part of the study), and the appearance of increased numbers of vascular amyloid deposits and increased microhemorrhage. By 4–6 years post-immunization they identified two cases that have a virtually complete removal of amyloid deposits, including vascular deposits, and no evidence of residual microhemorrhage. Nonetheless, there was no resolution of tau pathology. Perhaps most disturbing was that no cognitive benefits of the amyloid clearance could be detected in these end stage cases[59]. This implies that a) amyloid has no direct impact on cognitive status in humans, b) these patients had some other form of dementia to begin with or c) that the damage is so extensive by the time that cognitive symptoms are apparent, that rescue is not feasible. In any event, if these 2 cases are representative, this does not bode well for any anti-amyloid approach in patients that already have the disease. It should be noted that even in the MCI phase, prior to overt clinical dementia, considerable neuron and synapse loss is evident[10].

The Way Forward

One of the more exciting developments of the last two years in Alzheimer’s research is the identification of methods to positively identify Alzheimer dements from other disorders without requiring autopsy. One method to achieve this is positron emission tomography (PET) using amyloid binding ligands. The first of these was a [11C]-labeled compound referred to as Pittsburgh compound B[63]. Additional ligands are in various stages of development using [18F] labeling, a longer lived isotope permitting a broader application of the methods (i.e. not requiring an adjacent cyclotron and radiochemistry facility). An increasing number of publications demonstrate a reasonable correlation of the PET ligand signals and amyloid deposition in brain. Perhaps most important, there now appear to be individuals who are cognitively normal, but carry positive PET amyloid ligand signals. A recent report monitoring these individuals has found a high predictability of the PET signals in normals and subsequent conversion to dementia[4]. In addition, there are now cerebrospinal fluid analyses that appear to specify Alzheimer cases, and these also appear in some individuals that do not yet have dementia. Together these raise the possibility of defining, with some precision, a set of individuals with dramatically increased risk of dementia. At the recent ICAD meeting in Honolulu, it was suggested these signals may precede disease onset by up to a decade[5].

This raises the possibility of attempting to prevent the development of Alzheimer’s dementia. Screening high risk populations for these amyloid signatures using PET scans and spinal fluid analyses may identify those with amyloid deposits before substantial neural damage has occurred. Treatments could then be attempted which reduce the amyloid in anticipation of slowing the onset of the disorder or preventing the disease altogether. In this context, a subset of the patients in the phase 3 bapineuzimab trial were screened for amyloid deposits using PET scans before and after the 18 mo treatment with the agent[64]. The antibody treatment significantly reduced the intensity of the amyloid signals over the 18 mo of the trials compared to pateitns receiveing placebo therapy. Even if the issue of vasogenic edema complicates the use of bapineuzimab, there are a large number of other immunotherapy approaches also in clinical or preclinical development (see Table 1). We remain hopeful that one or more of these will succeed in reducing amyloid presymptomatically, and that this will be sufficient to avert the onset of the cognitive problems associated with Alzheimer’s dementia, permitting us to prevent Alzheimer’s disease from occurring.