ISCTM 7th Annual Scientific Meeting – Session 1 Amyloid Hypothesis – Summary

Background

Unfortunately, to date the numerous amyloid-targeting drugs and vaccines developed to treat symptoms or modify the course of Alzheimer’s disease (AD) based on the amyloid cascade hypothesis have largely failed, causing some researchers to question the validity of the approach, as well as the clinical trials methodologies employed. This notion was recently reviewed in a session entitled “Do Clinical Results to Date Suggest That Drug Development Based on the Amyloid Hypothesis of Alzhiemer’s Disease is Dead” at the 7th Annual Scientific Meeting of the International Society of CNS Clinical Trials and Methodology (ISCTM) in Washington DC.

Summary

The session was led by industry veterans, academicians and regulatory experts.  Data for various amyloid drugs was reviewed and it was generally suggested that, despite the fact that most amyloid-targeting drugs/vaccines mechanistically were able to impact their intended targets in the predicted manner, these treatments have largely been clinically ineffective or in some cases even deleterious due to a variety of methodological and design flaws including under-powering, poorly chosen outcome measures, and short time frames.  However, one of the most salient factors contributing to these failed trials concerns subject inclusion criteria, as it was determined that that these amyloid-based interventions may have been administered much too late in the course of AD illness.  Moreover, the various mechanisms of actions of drugs targeting amyloid, suggest that the stage of the illness needs to be better matched to the pharmacological intervention. There was a general sentiment expressed that several good drugs may have been simply applied at the wrong stage of illness.  Therefore, investigating amyloid agents in patients who are in the prodromal (pre-clinical) or even “pre-prodomal” phases of illness would be more advantageous when assessing efficacy.  Although it was even suggested that the mere appearance of a measurable biomarker implied that it might be too late to significantly impact disease state, the consensus opinion supported the argument that having qualified biomarkers in the arsenal of CNS drug development tools is very useful.  In summary, it was concluded that despite decades of research in pre-clinical models, few strategies are adequately validated or predictive to effectively guide drug development.  This has been borne out by the negative results of the clinical trials completed to date.  Going forward, it was recommended that an earlier stage of the disease should be targeted and a combination of biomarkers and clinical measures should be used in clinical trials of disease-modifying treatments. 


Introduction: Ravi Anand, MD and Larry Ereshefsky, PharmD

Dr. Anand reviewed the development of the amyloid hypothesis of AD, from the original neuropathological findings of plaques and tangles made by Alois Alzheimer through the discovery that amyloid plaques correlated with the degree of dementia and contained amyloid beta peptide (Aβ).  The identification of the Amyloid Precursor Protein (APP) gene and the discovery that patients with early onset AD had genetic mutations leading to amyloid deposition formed the basis for the hypothesis that accumulated Aβ leads to a cascade of events resulting in neuronal death.  Based on this hypothesis, numerous therapies were developed targeting Aβ formation, clearance and toxicity, including β- and γ-secretase inhibitors, α-secretase enhancers, vaccines to enhance clearance of Aβ, and agents to inhibit Aβ aggregation. To date, none of these agents has shown significant improvement in cognitive function in AD patients, and some have been associated with significant worsening or toxicity.  Therefore, it is possible that Aβ metabolism is the wrong target.  Shortcomings of the amyloid hypothesis of AD include the fact that neurotoxicity of Aβ has not been demonstrated, and the amyloid load does not seem to correlate with the onset of dementia or cognitive deficits.  Although the altered metabolism of amyloid in AD is supported by a large body of evidence, there is no evidence that reversal of this process will lead to therapeutic benefit in AD patients.

Dr. Ereshefsky noted that secretase inhibitors (gamma and BACE1) appear to be engaging the presumed target for clinical effect since dose-related reductions in CSF Aβ40/42 and increases in soluble APP have been observed.  Similarly, vaccines targeting Aβ have resulted in increases in amyloid clearance, yet in all cases to date, clinical benefit is marginal.  For γ-secretase inhibitors, too much inhibition appears to be detrimental, with side effects demonstrating a lack of selectivity consistent with significant NOTCH engagement. Unfortunately, semagacestat demonstrated statistically significant worsening in cognition at the highest dose used in clinical trials in Alzheimer’s patients.  .Possible explanations for the disappointing results may be that the drugs were given at the wrong stage of illness (too late), that tau-opathies might be a more relevant disease target, and/or the measures used to assess efficacy, e.g., ADAS-Cog and the ADCS-ADL are susceptible to culture, language, educational, and socioeconomic bias.  Currently, global trials in ‘pre-mild’ Alzheimer’s Dementia are being initiated; however, reliable severity typing in International trials might also lead to confounding effects, thereby reducing the clinical trial’s signal detection sensitivity.


Frank LaFerla, PhD

Dr. LaFerla discussed the pre-clinical models used for AD and their appropriateness for predicting the success of Aβ-modifying therapies.  Agents that are currently being tested in AD patients worked in pre-clinical models; however, clinical trials of amyloid modifying agents to date have been uniformly disappointing, possibly due to the following: AD does not have a uniform path in all patients, co-morbidities must be taken into account, and treatment may be ineffective once Aβ and tau pathology has set in, with treatments being unable to reverse decades of pathological changes.  Also, animal models in mice may not adequately represent pre-AD, i.e., they do not mimic the neuronal loss and tau pathology seen in humans.  A triple transgenic mouse model, which includes the human APP gene and shows age-dependent build-up of plaques and tangles, along with significant cognitive impairment, was discussed.  Passive immunization to remove Aβ in this model did not improve cognition if there was no effect on tau.  Numerous negative studies (i.e. no effect on pathology or cognition) in mouse models with various therapies have been conducted but have not been published. 

A new approach to treating AD pathology using neural stem cells (NSCs) was discussed.  The advantages of using NSCs include 1) no ethical issues, 2) NSCs are transplanted into neural tissue, and 3) NSCs can differentiate into neurons, oligodendrocytes or astrocytes.  NSCs injected into the hippocampus of AD transgenic mice improved memory without affecting Aβ or tau pathology.  Most cells migrate to the dentate gyrus and become glial cells, not neurons.  

Improvement appears to be due to increased synaptogenesis via a neurotrophic mechanism, i.e., increased expression of BDNF.  However, many unresolved issues need to be addressed before using NSC transplants in humans with AD.  

Suggestions for the future included the need for developing better animal models of AD that mimic sporadic disease, not just the genetic disease, as well as the heterogeneity of AD seen in humans.  Mouse models that develop tau pathology and neuronal damage are needed.  In clinical trials, treatment should be initiated sooner to improve the response, and targets other than Aβ should be considered. 


Holly Soares, PhD

Dr. Soares presented a summary of findings from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) focusing on progress in biomarker research, especially as they apply to disease progression.  Biomarker-based clinical trials that reliably predict those patients more likely to progress to Alzheimer’s disease will help researchers identify the best study participants, better characterize disease progression, and reduce the number of subjects needed to detect efficacy signals.  From the myriad biomarkers utilized across the ADNI programs it was reported that, although some biomarkers such as beta amyloid, tau and PIB are prone to large variation, non-normal distributions, and have generally been associated with only modest longitudinal changes, other biomarkers such as FDG PET, have shown greater sensitivity to changes over time.  For example, it was suggested that prodromal patients with greater deficits in glucose metabolism are more likely to progress to AD.  Additionally, two cognitive markers reflecting delayed procedural memory and lateral temporal cortical thickness may also help predict rate of decline in AD.  This rate of decline (slope) of AD is influenced by age, baseline levels of cognition, apolipoprotein E (ApoE) status and various other biomarkers. This slope is variable and trials that have durations of 18 months or less may need to select patients who are rapid decliners.  In addition to ADNI, the Coalition Against Major Diseases has developed a new shared and standardized database.  This database containing information from approximately 4,000 Alzheimer’s subjects has also provided important information regarding the natural history of disease progression by looking at the placebo arms from eleven industry-sponsored clinical trials.


Lon Schneider, MD

Dr. Schneider presented an overview of recent clinical trials in AD and discussed ways to move forward.  Despite more than 20 years of clinical research on AD treatments, only cholinesterase inhibitors and memantine have been approved.  Promising therapies, based on animal models, have not translated into clinical benefit in humans.  Numerous treatments with varying mechanisms of action have failed in Phases 2 or 3 of development, including those targeting amyloid.  Failures may be related to a number of issues, including the fact that the cause of AD is not known, there are no validated drug targets, and the outcome measures we are using, e.g. ADAS-Cog, may not be appropriate.  Trials are getting longer (18 months is standard for amyloid-modifying therapies) and are dependent on the placebo group worsening, which is problematic as the placebo response for the ADAS-Cog varies greatly among studies, individual patients and the stage of the disease. 

An overview of anti-Aβ therapies in development, including Aβ immunotherapy (antibodies, immunoglobulins, vaccines), γ-secretase inhibitors and modulators, BACE inhibitors and fibrilogenesis inhibitors was provided.  Results of Phase 2 studies for amyloid-modifying agents, including an Aβ antibody immunotherapy (Bapineuzumab), a γ-secretase modulator (Tarenflurbil), a γ-secretase inhibitor (BMS 708163) and a drug that inhibits Aβ oligomer formation (Scyllo-inositol) were discussed.  In each case, decisions were made to proceed to Phase 3 studies, despite limited evidence of efficacy (cognitive benefits) and/or significant safety issues noted at higher doses in the Phase 2 studies.  In general, the results from Phase 2 studies have not been predictive of the outcomes in Phase 3 trials.

It was recommended that in moving forward, we need to better understand the disease process and the pharmacology of the drugs being tested, and need to design better trials using multivariate outcomes that are more appropriate.  The causes of variability also need to me minimized.  “Prevention trials” in AD must be initiated earlier, before Aβ and tau pathology is evident, as this may be too late.


David Henley, MD

Dr. Henley, presented the background information and the results from failed Phase 3 trials of a γ‑secretase inhibitor (Semagacestat) conducted by Eli Lilly and Company.  By inhibiting γ‑secretase, semagacestat was expected to decrease levels of Aβ in the brain, and the formation of amyloid plaques.  In Phase I studies, semagacestat lowered levels of Aβ1-40 in plasma of healthy volunteers and AD patients, and dose-dependently decreased the CSF Aβ synthetic rate.  Despite this biomarker evidence of effects on the target enzyme, two large Phase 3 trials in mild to moderate AD patients were stopped following an interim analysis which showed that patients on semagacestat had significantly greater worsening on clinical measures of cognition and activities of daily living.  In addition, the drug was associated with an increased risk of skin cancer, possibly related to its interaction with the Notch receptor.  Other drugs targeting amyloid that had robust Phase 2 biomarker data, e.g. bapineuzumab, flurbiprofen and tramiprosate, have also failed to show clinical benefit. 

Possible reasons for the failure of γ‑secretase inhibitors may relate to the fact that there are six different types of γ‑secretase, and over 50 different substrates for the enzyme have been identified; therefore, it is not know which of these targets are being affected.  Amended protocols are ongoing in patients who stopped the medication to determine if cognitive worsening is reversible and to continue collecting biomarker data.


George Grossberg, MD

Dr. Grossberg discussed proposals for designing the next generation of disease-modifying trials in AD.  Some of the weaknesses in the current approaches include defining the target population, e.g. prodromal AD (pre-Clinical) vs. MCI vs. mild AD, lack of agreement on diagnostic criteria, need for better assessment of other risk factors and co-morbidities, e.g. vascular pathology, and need for use of biomarkers for inclusion.  In particular, with regard to amyloid-modifying therapies, the optimal time in the disease process and where in the amyloid cascade to intervene is not known.  In addition, it is not known if the formation of amyloid plaques and neurofibrillary tangles are related and whether they are the cause of AD or a reactive process.  A revised amyloid cascade hypothesis incorporating the effects of aging and other risk factors needs to be developed.

An important question is whether treatments targeting removal of Aβ from the brain in AD patients are viable.  There is evidence that Aβ may have neuroprotective effects and that removal of Aβ may result in oxidative stress.  Earlier intervention may have greater benefit, as mitochondrial injury related to Aβ has been observed prior to plaque formation.  Other proposed alternative approaches include inhibition of Gamma-Secretase Activating Protein (GSAP), enhancing brain metabolic activity, blocking Amyloid Beta-Derived Diffusible Ligand (ADDL), which affects synaptic signaling, and anti-amyloid “combination therapy”, e.g. combining a BACE inhibitor with a γ‑secretase inhibitor and using lower doses of each.  Possible approaches targeting tau pathology include reducing hyperphosphorylation of tau (kinase inhibitors), reducing tau aggregation, promoting clearance of hyperphosphorylated tau, and promoting microtubule stabilization.

Prevention strategies for AD are the ultimate goal; however, at a recent NIH State-of-the-Science Conference on Preventing AD and Cognitive Decline, the consensus was that there is currently no evidence to support use of any pharmaceutical agents or dietary supplements in the prevention of AD.  Going forward, primary prevention trials for disease-modifying agents must identify the population of very early AD patients to target, possibly using biomarkers to identify” at risk” individuals before clinical symptoms are present.  Diagnostic criteria, validated surrogate biomarkers for following disease progression, and sensitive outcome measures will need to be defined for these studies.  The use of enriched patient populations and stratification of subjects based on biomarkers should be considered to enhance signal detection and help identify responders.


Russell Katz. MD and Karl Broich, MD

Russell Katz and Karl Broich from the FDA and EMA regulatory agencies, respectively, summarized the circumstances under which a biomarker or unvalidated surrogate measure could be adopted as a primary efficacy variable.  These represented personal views and not those from their respective agencies.  A “surrogate marker” can be defined as “…a laboratory measurement or physical sign that is used in therapeutic trials as a substitute for a clinically meaningful endpoint that is a direct measure of how a patient feels, functions, or survives and is expected to predict the effect of the therapy”.  The FDA may grant marketing approval for a new drug product on the basis of adequate and well-controlled clinical trials establishing that the drug product has an effect on a surrogate endpoint that is reasonably likely, based on epidemiologic, therapeutic, pathophysiologic, or other evidence, to predict clinical benefit, or on the basis of an effect on a clinical endpoint other than survival or irreversible morbidity.  It was acknowledged that there is no accepted definition or clear threshold of evidence supporting the term “reasonably likely” which is subjective and open to interpretation.

Surrogates are sought after as they should ideally permit a significant decrease in both the duration and size of studies, shortening the development timelines and saving money, and accelerating approval.  However, surrogate markers remain not yet sufficiently understood and there are none validated for use as a sole primary measure of effectiveness in definitive trials of CNS investigational drugs.  Surrogates are particularly useful when the clinical benefit of the drug is likely to be well in the future and when there are no other therapies; they may not be as useful when clinical effects are easily measured in a reasonable time frame and therefore, surrogates are often proposed as the most realistic way to support a claim for disease progression in AD,an ambitious goal for any CNS drug development company.

It was implied that the disappointing AD trial results to date with amyloid-modifying therapies have failed to support the utility of amyloid-targeted surrogates in AD trials.  In the absence of compelling information, an ideal but as yet undefined indication for testing an amyloid based unvalidated surrogate would be in the setting of very early stage of AD.  In this case subjects would be included in trials even though they are essentially asymptomatic, but may be at high risk for AD at a later time based on some combination of risk factors including but not limited to family history, ApoE genotype status, medical history, etc.  In this trial setting the assessment of traditional AD outcome measures such as the ADAS-Cog or CGIC/CIBIC would be irrelevant.  Instead, a correlation between the effects on a surrogate marker and an appropriate clinical outcome (such as cognition either as a single scale or domain) could be considered for a disease modification claim in AD in which a slowing of progression is seen.  Prevention trials were viewed as too cumbersome and almost unattainable.  It was also suggested that a defined change in a biomarker and cognition (even with no global measure provided) may be adequate for approval, pending an advisory meeting to support the validation of the biomarker.

European regulators also identified the need for a link or plausible correlation between a biomarker (such as a PET ligand that labels beta amyloid plaque in the brain or MRI of the medial temporal lobe) and a desired clinical outcome.  To facilitate this, a two-step approach was outlined, which shows a delay of progression based on signs and symptoms initially, followed later by a correlation with biomarker data to support a disease modification claim in AD.  Because a “disease modifying” effect cannot be established conclusively based on clinical outcome data alone, such a clinical effect must be accompanied by strong supportive evidence from a biomarker program.  As this is difficult to achieve without an adequately qualified and validated biomarker, a two-step approach may be more suitable. If in the first step, delay in the natural course of progression of the disease based on clinical signs and symptoms of the dementing condition can be established, this may be acceptable for a limited claim, e.g. delay of disability. If these results are supported by a convincing package of biological and/or neuroimaging data, e.g. showing delay in the progression of brain atrophy, a full claim for disease modification could be considered. (http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09/WC500003562.pdf).