We propose a general framework with biomarker criteria for the study of the preclinical phase of AD; however, more work is needed to clarify the optimal CSF assays, PET or MRI analytic techniques, and in particular, the specific thresholds needed to meet these criteria. There are significant challenges in implementing standardized biomarker “cut-off” values across centers, studies, and countries. Work to standardize and validate both fluid-based and imaging biomarker thresholds is ongoing in multiple academic and pharmaceutical industry laboratories, as well as in several multicenter initiatives. These criteria will need to be validated in large multicenter natural history studies, or as provisional criteria for the planning of preventative clinical trials. For instance, it will be important to establish the test–retest and cross-center reliability of biomarker measurements, further characterize the sequence of biomarker changes, and the extent to which these biomarkers predict subsequent clinical decline or clinical benefit. In particular, there is an important need to evaluate methods for determining “amyloid-positivity” because it remains unclear whether there is a biologically relevant continuum of Aβ accumulation, or whether there is a clear threshold or “cut-off” value that could be defined on the basis of predictive value for subsequent clinical decline, as has been suggested in several CSF studies [28
]. It also remains unknown whether these thresholds should be adjusted for age or genotype. After these thresholds are established, it may be most feasible to select research cohorts for large studies solely on the basis of “amyloid-positivity” on CSF or PET amyloid imaging, and to use additional biomarker and cognitive measures for post hoc analyses to determine additional predictive value.
Although recent advances in biomarkers have revolutionized our ability to detect evidence of early AD-P there is still a need for novel biomarker development. In particular, although the current biomarkers provide evidence of Aβ deposition, an in vivo marker of oligomeric forms of Aβ would be of great value. Imaging markers of intraneuronal pathology, including specific markers of specific forms of tau/tangles and alpha-synuclein, are also needed. In addition, more sensitive imaging biomarkers that can detect early synaptic dysfunction and functional and structural disconnection, such as fMRI and diffusion tensor imaging, may one day prove to be useful to track early response to amyloid-lowering therapies. Finally, we may be able to use the currently available biomarkers as a new “gold standard” to re-evaluate simple blood and urine markers that were discarded on the basis of excessive overlap between clinically normal and AD patients. The significant proportion of clinically normal individuals who are “amyloid-positive” on both CSF and PET imaging may have confounded previous studies attempting to differentiate “normal” controls from patients with AD.
Similarly, additional work is required to identify and validate neuropsychological and neurobehavioral measures to detect the earliest clinical manifestations of AD. We need to develop sensitive measures in multiple cognitive and behavioral domains that will reveal evidence of early synaptic dysfunction in neural networks vulnerable to AD pathology. We also need to develop measures of very early functional changes in other domains, including social interaction, mood, psychomotor aspects of function, and decision making. These measures would allow us to link better the pathological processes to the emergence of clinical symptoms, and may be particularly useful to monitor response to potential disease-modifying therapies in these very early stages.
The proposed criteria apply primarily to individuals at risk by virtue of advanced age because inclusion criteria for trials in autosomal dominant mutation carriers and homozygous APOE ε4 carriers will be likely defined primarily on genetic status. Trials in genetic-risk populations might use these criteria to stage individuals within the preclinical phase of AD. In genetic-risk cohorts, it may even be possible to detect an even earlier stage of presymptomatic AD, before the point when there is already detectable cerebral amyloidosis. Several FDG-PET and fMRI studies have suggested that evidence of synaptic dysfunction may be present in young and middle-aged APOE ε4 carriers (see ), and there may be other biological alterations that are present before significant deposition of fibrillar forms of amyloid that would be preferentially responsive to presymptomatic intervention.
The emerging concept of preclinical AD and the role of biomarkers in the detection and tracking in this stage of the disease have important implications for the development of effective treatments. Therapies for preclinical AD would be intended to postpone, reduce the risk of, or completely prevent the clinical stages of the disorder. As recently noted, the use of clinical endpoints in clinical trials of such treatments would require large numbers of healthy volunteers, large amounts of money, and many years of study. Researchers have raised the possibility of evaluating biomarker endpoints for these treatments in cognitively normal people at increased risk for AD because these studies might be performed more rapidly than otherwise possible. Subjects enrolled in these studies could include individuals with autosomal dominant mutation carriers (with essentially a 100% chance of developing clinical AD) or those at increased risk of developing sporadic AD (e.g., APOE ε4 carriers or subjects with biomarker evidence of preclinical AD pathology). The use of biomarkers rather than clinical outcomes could accelerate progress in these trials; however, regulatory agencies must be assured that a given biomarker is “reasonably likely” to predict a clinically meaningful outcome before they would grant approval for treatments tested in trials using biomarkers as surrogate endpoints. Research strategies have been proposed to provide this evidence by embedding the most promising biomarkers in preclinical AD trials of people at the highest imminent risk of clinical onset to establish a link between a biomarker effect and the onset of clinical symptoms of AD. We envision the time when the scientific means and accelerated regulatory approval pathway support multiple preclinical AD trials using biomarkers to identify subjects and provide shorter term outcomes, such that demonstrably effective treatments to ward off the clinical stages of AD are found as quickly as possible. There are several burgeoning efforts to design and conduct clinical trials in both genetic at-risk and amyloid-positive older individuals, including the Dominantly Inherited Alzheimer Network (study of familial AD), the Alzheimer Prevention Initiative, and Anti-Amyloid Treatment in Asymptomatic AD (A4) trial being considered by the Alzheimer's Disease Cooperative Study.
Finally, the ethical and practical implications surrounding the issues of future implementation of making a “diagnosis” of AD at a preclinical stage need to be studied, should the postulates put forth previously prove to be correct. Although at this point our recommendations are strictly for research purposes only, the public controversy surrounding the identification of asymptomatic individuals with evidence of AD-P raised several important points that the field must consider. In particular, the poignant question of “why would anyone want to know they have AD a decade before they might develop symptoms, if there is nothing they can do about it?” should be carefully considered well before any results from research is translated into clinical practice. First, there may be important reasons, including social and financial planning, why some individuals would want to know their likelihood of developing AD dementia within the next decade, even in the absence of an available disease-modifying therapy. It is our hope, however, that the advances in preclinical detection of AD-P will enable earlier, more effective treatment, just as nearly all of therapeutic gains in cancer, cardiovascular disease, osteoporosis, and diabetes involve treatment before significant clinical symptoms are present. It is entirely possible that promising drugs, particularly amyloid-modifying agents, will fail to affect the clinical course of AD at the stage of dementia or even MCI, when the neurodegenerative process is well entrenched, but may be efficacious at the earliest stages of the AD-P, before the onset of symptoms.
The definitive studies to determine whether the majority of asymptomatic individuals with evidence of AD-P are indeed destined to develop AD dementia, to elucidate the biomarker and/or cognitive endophenotype that is most predictive of cognitive decline, and to determine whether intervention with potential disease-modifying therapies in the preclinical stages of AD will prevent dementia are likely to take more than a decade to fully accomplish. Thus, we must move quickly to test the postulates put forth previously, and adjust our models and study designs as new data become available. Because potential biologically active treatments may be associated with small but significant risk of adverse side effects, we will need to determine whether we can predict the emergence of cognitive symptoms with sufficient certainty to appropriately weigh the risk/benefit ratios to begin treatment in asymptomatic individuals. It is clear that many questions remain to be answered, and that there may be additional factors which will influence the probability of developing clinical AD. However, the considerable progress made over the past two decades now enables a strategic path forward to test these hypotheses, move the field toward earlier intervention, and ultimately, toward the prevention of AD dementia.