This study used the publically available clinicaltrials.gov database to assess the historic trends of AD drug development and to put the current pipeline of agents in perspective. The results demonstrate that detailed interrogation of clinicaltrials.gov can provide insight into longitudinal trends in drug development. A comprehensive database of all clinical trials registered in clinicaltrials.gov, the Aggregate of ClinicalTrials.gov (AACT), has become available [12] and may facilitate further analyses.
In the decade of 2002 through 2012, 244 compounds were assessed in 413 trials for AD. Of the agents advanced to Phase 3 (and excluding those currently in Phase 3), one was advanced to the FDA and approved for marketing (1.8%). Excluding the 14 compounds currently in Phase 3, the overall success rate for approval is 0.4% (99.6% attrition). This is among the lowest for any therapeutic area [13, 14].
The developmental time lines for conducting two Phase 3 trials needed to satisfy FDA requirements is substantially shorter for symptomatic agents than for disease-modifying compounds. For symptomatic cognitive enhancers in Phases 2 and 3, the trials were 20 and 34.6 weeks in duration, whereas trials of disease-modifying agents were 47.6 and 90.9 weeks in Phase 2 and Phase 3, respectively. The total duration of a trial is the length of time devoted to recruitment plus the treatment period; in some cases, the trial length included an open-label extension. The period of recruitment varies and is often longer than anticipated by the sponsor, because recruitment of AD patients is slower than expected for many trials [15]. The total time that a compound resides in any phase of the development pathway is a combination of duration of all the trials performed (some may be done concurrently) and time for analysis and decision making.
Progression through the pipeline is not necessarily sequential. Not all compounds tested in Phase 2 or 3 would have been assessed in previous stages. For example, a repurposed compound could be tested in Phase 3, based on data generated in populations with other indications, without necessarily be assessed in Phase 1 or in Phase 2 for AD. Rosiglitazone is an example of such a compound; it was tested in Phases 1, 2, and 3 for diabetes and in Phase 3 for AD. A repurposed compound entering the pipeline in Phase 3 might require testing in Phase 1 (for example, drug-drug interactions studies with antidementia agents in healthy volunteers). Dimebon is an example of this reverse sequencing; this agent had simultaneous Phase 1 trials assessing drug-drug interactions and Phase 3 trials for efficacy.
As drugs progress through the development pipeline, trials become longer and larger; this is especially evident in the programs for disease-modifying compounds. The mean duration of trials in Phase 2 is 47.6 weeks, and the mean duration of Phase 3 is 90.9 weeks. The average number of patients in Phase 2 is 142, and the average number in Phase is 833. The resource requirements for developing disease-modifying agents are greater than those required for symptomatic agents. Phase 2 has been substantially smaller and shorter than Phase 3 for most agents. More-robust Phase 2 programs with better understanding of the molecule might contribute to improving the success rate in Phase 3.
The attrition rate for AD treatment is high, with 72% of agents failing in Phase 1, 92% failing in Phase 2, and 98% failing in Phase 3 in the period observed. If these rates are applied to the current pipeline, 6.4 of the agents in Phase 1 and 4.7 of the agents in Phase 2 will be advanced to the next stage. Of the 14 drugs currently in Phase 3, the data predict that only a very limited chance exists of any being advanced for regulatory review. Predictions of this type will remain conservative until a breakthrough first-in-class agent recalibrates the expectations.
The one agent approved during the decade reviewed (memantine) is a symptomatic cognitive enhancer. Cognitive-enhancing agents are an active area of investigation with 151 of 413 trials in the 2002 through 2012 period devoted to this class of agents.
Two-hundred twenty-one agents have been assessed for disease-modifying potential, and none has shown a drug-placebo difference in favor of active treatment on primary outcomes, although a few agents (seven) are in on-going trials of this class of agent, and their outcome has yet to be determined. Failures in trials may be based on lack of efficacy, excessive side effects, or challenges in trial execution. Trial-conduct failure is suggested by a lack of decline in the placebo group, no effect in an active-treatment comparator arm of the study, or excessive measurement variability. The reasons for trial failures suggest means of enhancing the success of trials, including improved rating strategies, enhanced training, and better patient-selection approaches [16, 17]. New means of predicting drug toxicity may reduce the attrition rate attributable to lack of safety [18, 19].
Reasons for lack of efficacy in well-conducted trials must also be interrogated to improve the success rate for AD drug development. It has been suggested that use of antiamyloid agents may be optimized by intervening earlier in the disease process before nonamyloid processes prevail and neurodegeneration begins [20–23]. Identifying new disease pathways more amenable to pharmacologic manipulation, improved understanding of the complex neurobiology of AD, and use of combinations of therapies may provide new approaches to AD therapy [24–26].
Most disease-modifying trials have some form of Aβ protein as the pharmacologic target (that is, four of six current Phase 3 compounds of disease-modifying agents target the amyloid-beta protein). One-hundred forty-five (65.6%) of 221 trials of disease-modifying agents registered in the 2002 through 2012 period were directed at this target. The target is unvalidated, and no class of agents has shown efficacy for this target in human clinical trials. Many animal models of amyloidosis have shown biological and behavioral benefit from anti-Aβ agents, creating a “translational gap” between human and animal studies [27–30]. Development of animal models more predictive of success in human trials, diversification of targets within AD, use of rational combinations to address multiple disease pathways simultaneously, and optimizing the selection of patients more likely to be responsive to antiamyloid therapies may all enhance success in AD drug development.
The current AD pipeline is relatively modest, given the enormous challenge posed by this disease. AD is more expensive to the U.S. economy than cardiovascular disease or cancer [31]. Currently, 108 clinical trials for AD therapies are being conducted. This compares with 1,438 ongoing trials for oncology agents. The success rate of development of oncology compounds is 19% [32], encouraging biotechnology and pharmaceutical companies to invest time, effort, and funds in oncology drug testing. Similar successes are needed to spur AD drug development.
The high rate of attrition of compounds requires a constant supply of new approaches (new chemical entities, immunotherapies, repurposed drugs, devices) that can be assessed for efficacy in AD. The pipeline is dependent on a complex drug-development ecosystem of academic laboratories, federal funding agencies, biotechnology companies, venture capital, philanthropy, trial sites, contract research organizations, pharmaceutical companies, advocacy groups, and regulatory agencies. This ecosystem must be supported, grown, and coordinated to improve the success of AD trials and development of desperately needed new AD therapies.