Future screening for incipient Alzheimer's disease--the influence of prevalence on test performance

Much effort has been made to identify and verify diagnostic biomarkers for early stage Alzheimer's disease (AD). The need for this is often advocated by possible future disease-modifying treatments, likely to be most effective if initiated early in the disease process. Since the neurodegenerative process probably starts many years before the first onset of symptoms, such future drugs are likely to invoke a need for screening presymptomatic individuals. Here, we speculate on the performance of currently available AD biomarkers in hypothetical screening programs of different designs. We note that many diagnostic tests will have an excellent ability to exclude upcoming AD. However, even the best tests will suffer from poor positive predictive values given the relatively low disease prevalence in populations with no or very few symptomatic individuals, also when taking future converters to AD into account. The magnitude of this problem, which is common among most screening programs, will depend on the efficacy, safety and cost of the future anti-AD drugs. A number of tentative solutions to the problem, apart from better tests, are discussed.