Why do so many clinical trials of therapies for Alzheimer's disease fail?

Alzheimer’s disease is an irreversible, progressive brain disorder that accounts for about 50–75% of all cases of dementia. Alzheimer’s disease is characterised by the presence of amyloid plaques (amyloid β) and neurofibrillary (tau) tangles, plus the loss of connections between neurons in the brain. The damage to the brain induced by abnormal deposits of amyloid β and tau tangles is believed to start a decade or more before a decline in cognitive function is apparent. Three stages in disease progression are commonly recognised: the preclinical stage without symptoms (cognitively normal), mild cognitive impairment, and the final stage of Alzheimer’s disease that is stratified into mild, moderate, and severe phases. To measure the stage of disease in patients, a focus of research is to define the relations between three groups of variables. These variables consist of scores in cognitive tests, concentrations of specific biomarker proteins (eg, tau and amyloid β) in cerebral spinal fluid, and brain scans to measure brain volume changes and protein deposition (MRI and PET scans), each of which might serve as an endpoint in the design of clinical trials of possible therapies. The Alzheimer’s Disease Assessment Scale-Cognition subscale (ADAS-Cog) is the most widely used general cognitive measure in clinical trials of Alzheimer’s disease. Research is focused on the development of therapies to delay or halt the progression of Alzheimer’s disease. However, no disease-modifying drug for Alzheimer’s disease has been approved, despite many long and expensive trials. A recent failure in phase 3 was β secretase (BACE) in patients with mild-to-moderate Alzheimer’s disease. Other large phase 3 trials of antiamyloid approaches with disappointing results include semagacestat, bapineuzumab, and solanezumab. Many explanations have been proposed for the failures of trials of disease-modifying drugs for Alzheimer’s disease, including starting the test of therapies too late in disease development, incorrect drug doses, wrong treatment target, and an inadequate understanding of the biology of Alzheimer’s disease. All may be true, but could there be a simpler explanation based on the choice of the clinical endpoint for the trials and associated variability in measurement of endpoints within or between individuals? The importance of this variability in all three groups of measurements—cognitive tests, biomarkers, and brain scans—can be tested with clinical trial simulators. Clinical trial simulators use computational approaches based on mathematical models of disease induction and progression to explore the potential outcomes of a clinical trial under various trial designs and endpoint choices. They are powerful tools for increasing understanding of how the pharmacokinetics and pharmacodynamics of a drug influence the choices of the clinical endpoint, sample size, patient recruitment criteria, and trial duration. For Alzheimer’s disease the mathematical model of disease progression can be based on the description of the probability (a transition probability) that a patient moves between healthy and diseased states (eg, from cognitively normal to mild cognitive impairment or from mild