Alzheimer's disease drug development pipeline: 2019

Introduction Alzheimer's disease (AD) has few available treatments, and there is a high rate of failure in AD drug development programs. Study of the AD drug development pipeline can provide insight into the evolution of drug development and how best to optimize development practices. Methods We reviewed clinicaltrials.gov and identified all pharmacologic AD trials of all agents currently being developed for treatment of AD. Results There are 132 agents in clinical trials for the treatment of AD. Twenty-eight agents are in 42 phase 3 trials; 74 agents are in 83 phase 2 trials; and 30 agents are in 31 phase 1 trials. There is an increase in the number of agents in each phase compared with that in the 2018 pipeline. Nineteen agents in trials target cognitive enhancement, and 14 are intended to treat neuropsychiatric and behavioral symptoms. There are 96 agents in disease modification trials; of these, 38 (40%) have amyloid as the primary target or as one of several effects. Eighteen of the antiamyloid agents are small molecules, and 20 are monoclonal antibodies or biological therapies. Seven small molecules and ten biologics have tau as a primary or combination target (18%). Amyloid is the most common specific target in phase 3 and phase 2 disease modification trials. Novel biomarkers (e.g., neurofilament light), new outcomes (e.g., AD Composite Score [ADCOMS]), enrollment of earlier populations, and innovative trial designs (e.g., Bayesian adaptive designs) are new features in recent clinical trials. Discussion Drug development continues robustly at all phases despite setbacks in several programs in the recent past. Continuing unmet needs require a commitment to growing and accelerating the pipeline.


Introduction
Drug discovery and development for Alzheimer's disease (AD) is arduous. There have been no new drugs approved since 2003, and there are no approved disease-modifying treatments (DMTs) for AD. The challenges of drug development have become more complex as potential trial populations have expanded to include preclinical and prodromal AD, as well as AD dementia [1][2][3]. The US Food and Drug Administration (FDA) has provided guidance for clinical trials in AD dementia and predementia AD including use of a single primary outcome in trials of prodromal AD, the role of biomarkers in staging preclinical and prodromal AD, and the use of Bayesian statistics and adaptive clinical trial designs [4][5][6]. A new research framework for the diagnosis of AD based on amyloid, tau, and neurodegeneration (ATN) biomarkers was introduced by the National Institute on Aging (NIA) and the Alzheimer's Association [7]. This framework allows more precise classification of stages of AD, especially predementia stages, and may facilitate clinical trials of DMTs in AD [8]. Progress in biomarkers relevant to clinical trials of AD include increased understanding of the role of tau positron emission tomography (PET) in characterizing and staging AD and development of new fluid biomarkers such as neurofilament light and neurogranin that are increasingly integrated into clinical trials [9,10]. These advances comprise the foundations for progress in drug development and demonstrate collaboration among key stakeholders including basic and translational neuroscientists, clinicianscientists, pharmacy benefit managers, regulators, the National Institutes of Health (NIH), advocacy groups, and participants and family members.
In our annual update on the state of the AD drug development pipeline, we build on prior contributions to discuss the current phase 1, phase 2, and phase 3 clinical trials in AD [11][12][13]. We describe clinical trials and experimental treatments for disease modification, cognitive enhancement, and neuropsychiatric symptoms of AD. We note changes from 2018 and discuss specific areas of interest including repurposed agents, immunotherapies, novel mechanisms, the use of biomarkers in drug development, and new trends in AD clinical trials. Our goal is to continuously learn from the drug development process, identify best practices, and provide an update and overview of the current state.

Methods
Clinicaltrials.gov provides the source of information for this review. There are other clinical trial registries, and our review does not represent an exhaustive listing of every clinical trial in AD. However, the "Common Rule" governing clinicaltrials.gov mandates registration of all trials from sponsors with an investigational new drug or investigational new device [14,15] being assessed in the US. Compliance with the required trial registration is high [16][17][18]. The US has more clinical trials than any other nation, and thus clinicaltrials.gov includes most agents currently in clinical trials for AD.
We assayed clinicaltrials.gov as of February 12, 2019, and the tables and discussion provided apply to the information available at that time. We comment on terminated trials if the information has become publicly available but is not yet reflected on clinicaltrials.gov. We include all trials of all agents in phase 1, 2, and 3; if trials are presented as 1/2 or 2/3 in the clinicaltrials.gov database, we use that nomenclature in the review. Our trial database tracks trial title; trial number in clinicaltrials.gov; beginning date; projected end date; calculated trial duration; duration of treatment exposure; number of subjects planned for enrollment; number of arms of the study (usually a placebo arm and one or more treatment arms with different doses); whether a biomarker was described; subject characteristics; and sponsorship (a biopharmaceutical company, NIH, academic medical center, "other" entity such as a consortium or a philanthropic organization or a combination of these sponsors). We used the clinicaltrials.gov labeling and included trials that were recruiting, active but not recruiting (e.g., trials that have completed recruiting and are continuing with the exposure portion of the trial), enrolling by invitation, and not yet recruiting. We did not include trials listed as completed, terminated, suspended, unknown, or withdrawn. Information on these trials and reasons for their current status may not be publicly revealed. We do not include trials of nonpharmacologic therapeutic approaches such as cognitive therapies and caregiver interventions; we do not include studies of supplements and medical foods. We provide a table and brief discussion of new device trials (not included in Fig. 1). We do not include trials of biomarkers, although we note whether biomarkers were used in the trials reviewed. We include stem cell therapies among the interventions reviewed (not integrated into Fig. 1).
Drug targets and mechanisms of action (MOA) are important aspects of this review. MOA was determined from the information on clinicaltrials.gov or from a comprehensive search of the literature. In a few cases, the mechanism is undisclosed and could not be identified in the literature; we note these agents as having an "unknown" or "undisclosed" MOA. We grouped the mechanisms into symptomatic agents or DMTs. We divided the symptomatic agents into those that are putative cognitive enhancing agents or those that address neuropsychiatric and behavioral symptoms. DMTs were divided into small molecules or biologics including immunotherapies. DMTs were further divided into those targeting amyloid-related mechanisms, those that have tau-related MOAs, and those with "other" mechanisms such as neuroprotection, anti-inflammatory effects, growth factor promotion, or metabolic effects. The distinction between symptomatic and disease-modifying agents can be arbitrary, and some agents may have both properties. For purposes of this review, we chose what appears to be the principal MOA.

Overview
As of February 12, 2019, there were 132 agents in 156 trials of anti-AD therapies. Fig. 1 shows the universe of pharmacologic compounds currently in clinical trials for AD. Nineteen (14%) agents in trials target cognitive enhancement, and 14 (11%) are intended to treat neuropsychiatric and behavioral symptoms. There are 96 (73%) agents that intend to achieve disease modification; 38 (40%) of these have amyloid; and 17 (18%) have tau as the primary target or as one of several effects seen in nonclinical studies. Eighteen of the antiamyloid agents are small molecules, and 20 are monoclonal antibodies or biological therapies. Anti-tau agents include seven small molecules and ten biologics.
All compounds in AD clinical trials as of February 12, 2019 (the inner ring shows phase 3 agents; the middle ring is comprised of phase 2 agents; the outer ring presents phase 1 compounds; agents in green areas are biologics; agents in purple areas are disease-modifying small molecules; agents in orange areas are symptomatic agents addressing cognitive enhancement or behavioral and neuropsychiatric symptoms; the shape of the icon shows the population of the trial; the icon color shows the class of target for the agent.). Bolded names represent agents new to that phase since 2018.

Phase 3
In phase 3, there are 28 agents in 42 trials (Figs. 1 and 2, Table 1). There are 11 symptomatic agents in phase 3; three cognitive enhancers and eight targeting behavioral symptoms. There are six biological therapies and 11 oral agents/ small molecules in phase 3 that target disease modification. All the biological therapies and four of the small molecules have amyloid as the primary or one of several targets. There is one anti-tau agent in phase 3: LMTX (TRx0237). A phase 3 trial of this agent failed to show a drug-placebo difference [19], and based on the results, a new phase 2/3 trial (LUCIDITY) was started in 2018 with a lower dose of LMTX as monotherapy. Other mechanisms represented among phase 3 DMT molecules include neuroprotection, anti-inflammatory approaches, and metabolic interventions. Of the DMTs, two are repurposed agents approved for use in another indication (losartan plus amlodipine plus atorvastatin; and levetiracetam). Of the drugs with amyloid targets, there were six biologics, two beta-site amyloid precursor protein cleavage enzyme (BACE) inhibitors, and one antiaggregation agent. Fig. 2 shows the MOAs of agents in phase 3.
There were six prevention trials enrolling cognitively normal participants; 14 trials in patients with prodromal AD/mild cognitive impairment (MCI) or prodromal-tomild AD; 12 trials of patients with mild-to-moderate AD; and 10 trials of patients with mild-to-severe AD.  1. All compounds in AD clinical trials as of February 12, 2019 (the inner ring shows phase 3 agents; the middle ring is comprised of phase 2 agents; the outer ring presents phase 1 compounds; agents in green areas are biologics; agents in purple areas are disease-modifying small molecules; agents in orange areas are symptomatic agents addressing cognitive enhancement or behavioral and neuropsychiatric symptoms; the shape of the icon shows the population of the trial; the icon color shows the class of target for the agent.). Bolded names represent agents new to that phase since 2018. Phase 3 trials included an average of 640 participants and had a mean duration of 246 weeks (including the recruitment and the treatment period). Mean treatment exposure period was 73 weeks. DMT trials were longer and larger than trials of symptomatic agents with a mean duration of 297 weeks including 112 treatment weeks, and included an average of 862 participants. The mean duration of cognitive enhancer trials was 88 weeks (17 treatment weeks), and they included an average of 333 participants. Trials of agents for behavioral symptoms had a mean duration of 187 weeks (15 treatment weeks) and included a mean of 311 subjects.
The average duration of treatment exposure for phase 3 DMTs is 112 weeks, and the mean period from trial initiation to primary completion date (final data collection date for primary outcome measures) is 269 weeks. This indicates that 157 weeks, more than the treatment period, is the average anticipated recruitment time. When examined by trial population, DMT prevention trials are 405 weeks in duration (192 treatment weeks); trials for patients with MCI/prodromal/ prodromal-to-mild AD are 263 weeks in duration (98 treatment weeks); and trials for patients with mild-to-moderate AD are 264 weeks in duration (57 treatment weeks). Planned recruitment periods for these three types of trials are 192, 130, and 191 weeks, respectively.

Phase 2
Phase 2 has a larger array of therapies and mechanisms that are being assessed than are represented in phase 3. There are 74 agents in 83 trials (Figs. 1 and 3, Table 2). Of these, there are 20 symptomatic agents; 14 cognitive enhancers; and six agents targeting behavioral symptoms. There are 53 potential disease-modifying agents in phase 2 trials; 16 biologics and 37 small molecules. One agent had an undisclosed mechanism. Twelve of the small molecules and eight of the biologics have amyloid reduction as one of the mechanisms observed in nonclinical studies (38% of DMTs). Four small molecules and six biologics in phase 2 target tau as one of their mechanisms (19% of DMTs). There are 24 small molecules and two biologics with neuroprotection as one of the mechanisms (49% of DMTs). Other mechanisms represented in phase 2 include anti-inflammatory and metabolic interventions as the primary or one of a combination of effects documented in animal models. There are six trials involving stem cell therapies. Sixteen of the DMT agents are repurposed agents approved for use in another indication.
Of the drugs with amyloid targets, there were seven immunotherapies, one colony-stimulating factor, two BACE inhibitors, and two alpha-secretase modulators. Two agents targeted synaptic activity, two were anti-aggregation agents, and two agents involved neuroprotection or a metabolic MOA. There were two agents targeting both amyloid and tau reduction. Fig. 3 shows the MOAs of agents in phase 2.
Three of the phase 2 trials were prevention trials; 36 trials involved patients with prodromal or prodromal and mild AD; 38 were trials for mild-to-moderate AD; two trials were for patients with severe AD; two included patients with mild, moderate, or severe AD; one included patients with MCI or healthy volunteers; and one trial was for prodromal or mild-to-moderate AD.
Phase 2 trials are shorter in duration and smaller in terms of participant number than phase 3 trials; phase 2 trials had a mean duration of 178 weeks, average treatment period of 45 weeks, and included an average of 143 subjects in each trial.

Phase 1
Phase 1 has 30 agents in 31 trials (Fig. 1, Table 3). There are two cognitive enhancers being assessed in phase 1.    There are currently no agents addressing neuropsychiatric symptoms in phase 1. In addition, there are 13 small molecules and 13 biologics being assessed in phase 1. The MOA was not identified for two agents. Two of the small molecules and six of the biologics have amyloid as a primary target or one among several targets. Tau is targeted by one small molecule and four biologics in phase 1 studies. Other mechanisms represented in phase 1 include neuroprotection, metabolic, anti-inflammatory, and regenerative interventions. Phase 1 trials had an average duration of 141 weeks (recruitment and treatment period) and included a mean number of 58 participants in each trial.

Trial sponsors
Across all trials, 54% are sponsored by the biopharma industry, 35% by Academic Medical Centers (with funding from NIH, industry, or other entities), and 10% by others. Table 4 shows the sponsor of agents in each phase of development. Table 5 shows the biomarkers used as outcome measures in current phase 2 and phase 3 AD clinical trials as described in the federal website; not all trial descriptions in clinicaltrials.gov note if biomarkers are included in the trial.

Biomarkers
AD biomarkers served as secondary outcome measures in 16 phase 3 DMT trials and 29 phase 2 DMT trials. The most common biomarkers used were cerebrospinal fluid (CSF) amyloid, CSF tau, volumetric magnetic resonance imaging, and amyloid PET. Of the 25 phase 3 DMT trials, five trials (20%) used amyloid PET as an entry criterion, two (8%) used CSF amyloid, and eight (32%) used either amyloid PET or CSF amyloid. Ten (17%) out of 60 phase 2 DMT trials used amyloid PET as an entry criterion, seven (12%) used CSF amyloid, and six (10%) used either amyloid PET or CSF amyloid. Ten DMT trials in phase 3 and 37 in phase 2 did not require biomarker confirmation of AD for trial entry. Table 5. Biomarkers as outcome measures in phase 2 and phase 3 trials for agents in the Alzheimer's disease drug development pipeline (clinicaltrials.gov as of February 12, 2019)

Devices
A variety of approaches to brain stimulation are under study in clinical trials for AD (Table 6). These range from deep brain stimulation with implanted electrodes to surface application of light, electric current, and laser therapy. Most of the trials target cognitive enhancement; a few trials posit effects on amyloid, tau, inflammation, oxidative stress, or mitochondrial function [20,21]. Targets have varied from deep brain stimulation of fornix and memory-related structures to surface stimulation of parieto-frontal regions. The few completed studies have shown no consistent cognitive benefit; the techniques have been safe with acceptable adverse event profiles [22]. There are no FDA-defined phases for device trials, and most trials did not list the phase on clinicaltrials.gov. The stages of development for device studies can be divided into pilot, pivotal, and postapproval phases.

Discussion
In 2018, the FDA approved 59 novel pharmacotherapies across all therapeutic areas, breaking the 1996           record of 53 drug approvals [23,24]. There were 42 small molecule therapies and 17 biological therapies approved [24]. Eight new neurological drugs were included among the new therapies: 3 migraine treatments (all were calcitonin gene-related peptide receptor antibodies), 2 for seizures in Dravet syndrome (1 included Dravet syndrome and Lennox-Gastaut syndrome), two for hereditary transthyretin-mediated amyloidosis, and 1 for Fabry disease. The latter three agents can be regarded as DMTs, the others provide relief of symptoms albeit lifethreatening symptoms in the case of the epilepsies. The therapies for transthyretin-mediated amyloidosis are RNA-based interventions (antisense oligonucleotide or interference RNA) representing a new approach to neurological disorders. Oligonucleotide-based therapies have shown initial promise in Huntington's disease and may have applications in other neurodegenerative disorders including AD [25]. There is one RNA-based treatment in the AD pipeline (IONIS MAPTRx). Several agents have completed clinical trials since the time of last year's pipeline analysis and shown no drugplacebo difference. LTMX targeted tau pathology in AD and did not establish efficacy [19]. Azeliragon is a receptor for advanced glycation end products inhibitor and was found to produce no drug-placebo difference in a trial of mild-tomoderate AD. Crenezumab is a monoclonal antibody that targeted oligomeric forms of amyloid-beta protein (Ab) [26]. It failed to show a drug-placebo difference at the time of a futility analysis in two large clinical trials, and development of the agent was halted. Similarly, aducanumab trials were recently stopped after a futility analysis. Verubecestat is a BACE inhibitor whose development was halted for futility in a mild-to-moderate AD clinical trial [27]. A trial of verubecestat in patients with prodromal AD defined by clinical and amyloid PET measures was halted after a futility analysis suggested that the agent could not succeed. Similarly, lanabecestat did not meet the criteria to continue after a futility analysis. Atabecestat is a BACE inhibitor being assessed in preclinical AD; the trial was discontinued when elevated liver enzymes were observed among some trial participants. Intranasal insulin was assessed in mildto-moderate AD and showed no drug-placebo difference [28]. Pioglitazone, an insulin sensitizing agent was stopped for futility in a preclinical AD trial. A trial of a FYN inhibitor (AZD0530) used fluorodeoxyglucose PET as the primary outcome and showed no drug-placebo difference on the biomarker or any clinical measure [29]. Fluorodeoxyglucose PET performed well in this multisite trial suggesting it can be used in multicenter trials to show drug-placebo differences with agents that affect brain metabolism. ITI-007 is a multitransmitter agent being developed for the treatment of schizophrenia and was tested in a clinical trial to determine its effect on agitation in AD. No drug-placebo difference was observed in the trial. Of the 17 phase 3 DMTs listed in our 2018 review, eight have been terminated.
GV-971 is a multitargeted molecule that completed a phase 3 clinical trial in China in 2018 [30]. GV-971 has nonclinical evidence of effects on neuroinflammation, amyloid plaques, neurofibrillary tangles, mitochondrial function, and cholinergic function [30]. In a phase 3 trial conducted in China by an international contract research organization, GV-971 showed a statistically significant benefit over placebo on the Alzheimer's Disease Assessment Scalecognitive subscale [31]; a trend toward improvement was noted on the clinical interview-based impression of change [32]. There was no impact on functional and behavioral measures. The outcomes appear to have met the criteria required for approval by the Chinese FDA, and the agent is under review.
Biomarkers play an increasingly important role in AD drug development. Participant selection, target engagement, disease course prediction, evidence of disease modification, and side effect monitoring all involve biomarkers [33]. The NIA-Alzheimer's Association established the biomarker- Other combinations 0 6 (7) 2 (5) Abbreviation: NIH, National Institutes of Health.  based ATN framework for the diagnosis and characterization of AD [7]. This framework will assist in trials with both accurate diagnosis of AD and biological staging of AD relevant to matching the trial population to the MOA of the agent being assessed [8]. Of the disease-modifying trials currently in the AD pipeline, 52 use amyloid imaging and/ or CSF to support the diagnosis, 20 have amyloid imaging as an outcome, ten have tau imaging as an outcome of the intervention. In addition to the specific biomarkers included in the ATN framework, evidence is accruing that the plasma amyloid 40/42 ratio corresponds to the presence of cerebral amyloidosis [34] and that plasma neurofilament light is indicative of neurodegeneration [35]. These increasingly available biomarkers will facilitate screening for clinical trials and may have a role in course prediction and assessing treatment outcome.
Design innovations are evident in recent trials of AD therapeutics. Futility analyses were used to terminate development programs for pioglitazone, verubecestat [27], crenezumab, ITI-007, LY3314814 (lanabecestat), and aducanumab. Futility analyses are conducted when the trial is incomplete but when sufficient data are available to predict if continuing the trial could meet prespecified criteria [36].
The Alzheimer's Disease Composite Score (ADCOMS) [37] has been introduced as a cognitive outcome in several development programs including BAN2401, elenbecestat, and xanamem. The ADCOMS is an analytic approach whose score is based on combining scores on items derived from the Alzheimer's Disease Assessment Scalecognitive subscale, clinical dementia rating, and Mini-Mental State Examination [38] after these tools are administered in the standard way. ADCOMS constituents were derived from trials of patients with MCI that showed the most change over a one-year period to develop a score that is most likely to show a drug-placebo difference in trials of patients with early-stage disease and very limited cognitive deficits.
Bayesian adaptive designs are being implemented in AD trials. These have been used broadly in non-AD trials including development of cancer and diabetes therapies [39,40]. Adaptive trials are being used in the BAN2401 trial, Dominantly Inherited Alzheimer Network-Treatment Unit [41], European Prevention of Alzheimer Disease initiative [42], and the Intranasal Oxytocin for Fronto-temporal Dementia (FOXY) trial of intranasal oxytocin for frontotemporal dementia [43]. The trial of ABT-089 pioneered the use of an adaptive design in AD [44]. Bayesian designs use data derived from the ongoing trial to inform dose allocation, trial duration, sample size, or response to adverse events; decisions are prespecified before trial initiation. Dose-adaptive designs are participant-centric in that they allow study subjects to be assigned to the doses most likely to succeed or least likely to produce adverse events.
It has been argued that there is no "pipeline" of drug development because agents often do not proceed systematically from phase 1 to phase 3 and irregularities are common [45]. We use the word "pipeline" to categorize agents in early, middle, and late-stage trials. In AD drug development, agents tend to proceed from phase 1 single and multiple ascending dose studies to phase 2 proof-of-concept (POC) studies, and then to phase 3 registration-type trials. Testing for POC in phase 2 depends on dose ranges and safety established in phase 1 and provides the foundation for phase 3. Repurposed agents may have irregular pathways going from approved status for one indication to phase 1 or phase 2 to define dose and POC before testing in phase 3 for the AD indication [46,47]. The concept of "pipeline" applies as an imprecise but generally accurate overview of drug development for AD.
There are more agents in the AD pipeline in 2019 than was observed in the 2018 pipeline. There are 28 agents in phase 3 (compared with 26 in 2018), 74 agents in phase 2 (compared with 63 in 2018), and 30 in phase 1 (compared with 23 in 2018).
The lack of success in AD drug development has given rise to nihilism with regard to the ability of the field to develop agents that meaningfully modify the progression of AD. Suggestions to abandon the amyloid hypothesis, focus exclusively to combination therapies, place more emphasis on lifestyle interventions to prevent AD or reassess our assumptions and build new models to drive drug development are all voiced, and each of these perspectives have merit. Reviews of the pipeline show that lessons are learned from all trials; even negative and futile outcomes are highly informative and provide guidance for future trials. The overview of trials document a shift toward more diversification of targets between phase 3 and phase 2, the entry of combination therapies into the pipeline, and the use of biomarkers to allow early assessments of the impact of candidate interventions on disease biology.
Several agents have shown no drug-placebo difference, and the development programs have been discontinued. A few programs successfully demonstrated drug-placebo differences in phase 2 and are advancing. Progress depends on innovation and learning from exploration of new targets, assessment of new candidates, and implementation of new trial features. As in other chronic disease such as cancer, human immunodeficiency virus (HIV), and cardiovascular disease, a learning phase preceded periods whose sequential incremental successes led to meaningful treatments.

RESEARCH IN CONTEXT
1. Systematic review: There is a high rate of failure of drug development for Alzheimer's disease. New treatments are urgently needed, and review of the drug development pipeline can improve our understanding of how best to advance new therapies. We reviewed all drugs currently in clinical trials for Alzheimer's disease listed in the federal government database clinicaltrials.gov.
2. Interpretation: We showed that there are 132 agents in clinical trials for the treatment for Alzheimer's disease. Ninety-six of these drugs are diseasemodifying agents intended to change the underlying biology of Alzheimer's disease. Nineteen of the drugs are intended to be cognitive enhancing agents, and 14 are being developed for the treatment of neuropsychiatric and behavioral symptoms. We provide an overview of drugs currently in clinical trials for Alzheimer's disease.
3. Future directions: Progress is being made in terms of defining new targets for the treatment of Alzheimer's disease, developing new agents, introducing innovative clinical trial designs, incorporating a broader range of populations in clinical trials, and developing new biomarkers that provide insight into the impact of emerging therapies. Improvements in drug development success rates are anticipated.