Each of these stages can be targeted for specific drug treatments, requiring different trial designs and outcomes:

• a healthy elderly population can be tested with a safe antioxidant such as Ginkgo biloba versus placebo over five¹ to seven years using incident dementia as the primary outcome;²
• an aMCI population could be tested with a symptomatic or a potential DMD versus placebo over three years using time to diagnose dementia as the primary outcome;³
• patients with mild to moderate AD must be offered the usual standard of care to which a potential DMD or placebo is added over 12–18 months, with a slower decline on cognitive, global, or functional autonomy as primary end-points;⁴ and
• patients with severe AD are usually in a nursing home setting requiring shorter six-month studies with appropriate cognitive, functional, and behavioral outcomes.⁵

Disease milestones have also been defined in AD⁶ and can be used as endpoints in a survival analysis. For instance, α-tocopherol delayed loss of residual autonomy and progression to severe dementia or death in one study by the Alzheimer Disease Co-operative Study group (ADCS),⁷ thus influencing clinical practice to use vitamin E in all stages of AD (at least in the US until the negative results in the Memory Impairment Study),⁸ and a meta-analysis of clinical studies using high-dosage vitamin E suggesting a higher mortality rate.⁹

Symptomatic domains in dementia include cognition, activities of daily living (ADL), and behavior. In many patients, early changes in mood and anxiety precede the formal diagnosis of AD. Cognitive, functional, and global decline are relatively linear over time, whereas neuro-psychiatric symptoms peak midway into the disease course and improve spontaneously through the severe stage as mobility becomes impaired with emerging Parkinson-like physical signs.¹⁰ These natural changes in the symptomatic domains through the stages of AD have an impact on trial design and outcomes.¹¹

Symptomatic Clinical Trials using Cholinesterase Inhibitors and Memantine
The modern treatment for AD was initiated by the report that tacrine improved some aspects of cognition and daily life.¹² The follow-up confirmatory studies used cross-over and parallel-group designs. The latter design group offers the possibility of short-term (minimum of three-month) studies comparing the efficacy of different doses of the drug with placebo.

The primary analysis is carried out on outcomes at end-point, using the last observation carried forward (LOCF) or intent to treat (ITT) to compensate for missing values in case of drop-outs. Although LOCF/ITT has been favored for symptomatic studies, there is a trend for using observed cases (OC), e.g. completers, for studies of 12 months or longer. For practical purposes, both types of analysis are performed. Although ‘cognitive enhancement’ was the main hope for cholinesterase inhibitors (CIs) as a therapeutic class, the reality that has emerged from six-month studies with open-label extensions and the one-year placebo-controlled Nordic study¹³ is that although there is a small but statistically significant improvement in cognition peaking at three months with CIs, the most clinically relevant finding has been the stabilization of cognitive decline with ‘return to baseline’ at nine to 12 months for the actively treated groups at the higher therapeutic doses compared with placebo-treated groups, who decline steadily. These results were considered sufficient for CIs to become ‘standard treatment’ for AD in the mild to moderate stage,^14–16 notwithstanding the challenges to the effectiveness of CI¹⁷ using the arguments that the trial designs were flawed;¹⁷ a pragmatic study over three years comparing donepezil with placebo was negative for the primary end-point of delaying entry to institutional care and progression of disability;¹⁸ and there was insufficient cost–benefit, at least in mild stages.¹⁹