LAUSR

Optimism was high that the first disease-modifying antiamyloid medications would be effective in slowing the rate of decline in those with symptomatic Alzheimer disease (AD). When these studies failed, it set off conversations that still continue about whether the medications were ineffective, the disease was too advanced to be slowed by such agents, or amyloid was the correct pathology to target. The operationalization of mild cognitive impairment (MCI) led to targeting earlier symptomatic cases of the illness and treatment strategies based less on pathology and more on a chance to halt or slow decline than there would be earlier in the disease.1 With the development of amyloid imaging, MCI due to AD diagnosis was refined,2 and early-stage AD was extended further to include preclinical AD,3 wherein a positive amyloid positron emission tomography (PET) scan or diagnostic low levels of cerebrospinal fluid β-amyloid (Aβ) indicated the presence of pathology in people who were cognitively normal. Data from patients with familial AD with autosomal dominant mutations4 and longitudinal observation of individuals with positive Aβ PET scans and normal cognition led to the observations that amyloid plaque pathology could be present as long as 2 decades prior to the emergence of the first clinical symptoms of AD. Armed with biomarkers of Alzheimer pathology, studies of therapeutics to delay the emergence of symptoms could be undertaken, with the rationale that delay in symptom onset approaches the idea of prevention of dementia. The problems, of course, were many, including (1) using medications with some degree of risk in cognitively unimpaired persons; (2) balancing benefit and risk, as perhaps one-third to half of people with positive Aβ PET scans will not develop dementia5; (3) determining the effective doses for use in a long trial; (4) determining the duration of treatment and follow-up to ascertain that the treated group had a delay in the time of emergence of symptoms—a formidable logistical and statistical issue for clinical trials; and (5) funding the substantial cost of such longtermtrialsundertakenwithoutpriorlong-termevidenceandwith alowlikelihoodforsuccess.Apurepreventiontrialwouldrequire a long follow-up time and would have to show a clinically significant delay in emergence of cognitive symptoms. We are not yet at a time when a biomarker for Alzheimer pathology (such as Aβ PET, τ imaging, or magnetic resonance imaging volumes) could stand in or substitute for cognition as an efficacy outcome for clinical trials. Even in its latest draft guidance for early-stage AD, the US Food and Drug Administration has emphasized that a medication could not be approved for marketing solely by its effect on a biomarker, that it also must demonstrate an effect on cognition, and that the validity of any biomarkers used would be evaluated at the time when the administration was considering marketing approval.6 Recognizing the difficulties in such trials and the desperate need for preventive therapies, the US Food and Drug Administration would allow demonstration of potential benefit using a single cognitive measure or scale or neuropsychological battery for provisional approval until clinical meaningfulness could be demonstrated postmarketing.6 With the complications of prevention trials, findings from epidemiological studies have been used to examine the various medications or health habits, eg, diet, exercise, comorbid diseases, use of anti-inflammatories, antioxidants, estrogens, and HMGCR inhibitors (statins), that might have effects on the development of AD. Results from these observational studies indicated that some of these agents were associated with fewer cases of dementia or AD. The prevention clinical trials have been initiated with 3 primary types of design. The first—and most difficult—is to recruit a new population of people with normal cognition (and sometimes those with MCI as well) and observe them at regular intervals until a given number of incident cases of dementia have emerged, which would allow analysis to determine if the therapy was statistically effective. The Ginkgo Evaluation of Memory study7 is an example of this type of design, recruiting more than 3000 people and observing them for 7 years. Outcomes showed no benefit for ginkgo biloba extract, which has strong antioxidant properties, in slowing the dementia incidence rate. The GuidAge study8 in France had a similar design and used ginkgo biloba as well, although this study recruited people with mild memory complaints from their primary care physicians; findings from this study were also negative. In both studies, there were no significant adverse effects of ginkgo biloba use. The Alzheimer’s Disease AntiInflammatory Prevention trial9 also recruited a new population of cognitively unimpaired elderly persons who had a relatively higher risk for dementia by age and family history for a clinical trial of 2 anti-inflammatory medications, celecoxib and naproxen. This trial was stopped prematurely because of the emergence of data suggesting adverse cardiac effects of the anti-inflammatory drugs used. While the degree of risk of nonsteroidal anti-inflammatory drugs is still debated, the trial highlighted the fact that lower levels of risk are requisite for healthy individuals in such studies.9 Other prevention trials for dementia piggyback a cognitive or dementia outcome on a clinical trial of a drug used for another Related article Opinion