Identifying and monitoring memory changes in individuals that are at risk of Alzheimer's disease (AD) dementia necessitates serial cognitive testing in a memory clinic. As clinical trials move to test novel therapeutics much earlier in the disease, two factors are critical for their success: 1) very large sample size to detect subtle memory changes, and 2) multiple testing centers across cities and countries, which requires highly-standardized tests. Annual and in-clinic cognitive assessments are conventionally used to monitor memory changes in studies, as this is considered a well-validated and accurate method of gauging cognitive performance. This method, however, is extremely time-consuming, labor-intensive, and difficult to standardize across centers and countries. It also reduces results in the recruitment of only those individuals who are willing to travel in to a memory clinic and do hours' worth of testing; as such, studies are likely to recruit participants who are highly motivated, well educated, physically healthy, and/or low in racial diversity, which can skew estimates of cognitive change. This is particularly problematic for clinical trials, as they normally gauge the efficacy of a drug based off of cognitive change estimates gleaned from these biased observational studies. Inaccurately estimated cognitive trajectories can have major implications for the success of these trials, as the success and failure of drug trials are often predicated on a cognitive endpoint. In addition, the most comment method for gauging cognitive change is via annual assessments, which introduce issues of signal-to-noise and regression to the mean, such that random error may also obscure the rates of change.
Recent work in computerized testing has reintroduced the possibility of more frequent assessments. Further, in-clinic studies of high-frequency testing have shown that this method is reliable and more accurate than annual assessment. What is unclear, however, is whether high-frequency testing can be translated to the internet. In this situation, individuals would regularly self-administer their own cognitive tests for researchers to monitor remotely.
The objective of this proposal is to assess the feasibility and reliability of online, high-frequency cognitive testing in a large group of clinically-normal adults. This proposed study will capture "burst design" monthly memory performance over a period of 12 months in a group of 400 participants (40-65 years; 200 APOE 4 carriers/200 non-carriers), and, in a separate group of 400 participants, annually (at baseline and after 12 months). All participants will be recruited from the Australian Healthy Brain Project (www.healthybrainproject.org.au), a large (n=5000) online study of brain health in clinically healthy Australian middle-aged adults between 40-65 years. By examining high-frequency, online cognitive testing over the span of a year, this study will determine the applicability of this method to better track cognitive trajectories in comparison with annual cognitive testing. Further, this study will determine whether high-frequency cognitive assessments can detect the well-characterized cognitive decline trajectories seen in those with genetic risk for Alzheimer's disease sooner than shown in annual cognitive assessments.
In an effort to improve measurement of cognitive change in middle-aged adults, the proposed study provides an unprecedented opportunity to rethink the way cognitive decline trajectories are currently measured and estimated. Outcomes from this study will aid in the development of new approaches to measuring the very earliest cognitive changes in those at risk for Alzheimer's disease. In addition, this has consequences for the ability of observational studies to inform clinical trials on likely cognitive trajectories in populations of interest and has the potential to lead to more accurate estimations of treatment effects.
Harvard University (USA) and Florey Institute of Mental Health (Australia)