Commentary on “Recommendations from the National Institute on Aging‐Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease.”

The National Institute on Aging and-Alzheimer Association Workgroups undertook a very ambitious task when they attempted to operationalize the definition of Alzheimer disease (AD). This attempt was mandated because of the accumulation of data on the clinical syndrome we call AD, the definition of mild cognitive impairment as a clinical entity, and the availability of biomarkers—molecular, anatomic, and functional, which are increasingly being used in the diagnosis of AD. These important conclusions will definitely be used as guidelines by clinicians as well as researchers and employed by the industry when planning drug studies. Having uniformly-accepted diagnostic criteria will be very important for having a common basis and ensuring that we know what someone means when they talk about AD. In particular, the reference to mixed dementia should be applauded, although full recognition of the contribution of mixed cases is still needed [1]. Beyond these technical issues of consensus, however, lies a much more serious point, of what do we mean when we speak about AD. The basic issue is not defining AD, but rather defining the abstract concept of a "disease." It is likely that the authors evade this problem intentionally. However since this reflects on the whole issue of AD and senile dementia research, it should be addressed. One definition of AD, which the authors might agree upon, is “a progressive neurodegenerative disorder, caused by abnormal amyloid-beta (Aβ) metabolism or trafficking, with characteristic amyloid and tau deposits manifested as a clinical phenotype of initial hippocampal dysfunction causing episodic memory impairment and progressing over months or years to full-blown dementia.” This definition includes the pathophysiological basis, the clinical phenotype, and the eventual pathology. However, it does not include reference to the etiology, which is heterogeneous. Importantly, just having a common clinical syndrome, biological markers, and pathology is insufficient to define a disease. Thus, we do not consider hepatic cirrhosis a disease, although it fulfills the above requirements, because it can be caused by many different causes (genetic, infective, toxic, autoimmune, etc.). However, what we call AD is also heterogeneous in its etiology. Not only are some cases genetic, but the sporadic form is commonly regarded as being multifunctional in origin. This fundamental issue is of practical importance. Immunizing against hepatitis virus will not prevent cirrhosis from alcohol abuse, and therefore the assumption that AD is a unitary disorder is erroneous. The pathogenesis may be due to excessive Aβ synthesis, as in Down syndrome patients or carriers of APP mutations. However, Aβ misprocessing occurs following ischemia or trauma to the brain, and in sporadic AD it may be due to abnormalities of clearance [2]. It is unclear whether, in their guideline, the authors define AD by the underlying process of Aβ metabolism, by the pathology, or by the clinical phenotype. Definitely, these do not always coincide. Thus, elderly people who have Aβ and neurofibrillary tangle (P&T) load in their brains are defined by pathologists as having AD even if they are non-demented. This depends on the unproven assumption that P&T will eventually always lead to dementia. The use of cerebrospinal fluid (CSF) Aβ levels or cortical Aβ deposits to establish the efficacy of an intervention is particularly problematic when dealing with interventions specifically designed to alter Aβ processing. The well-known discrepancies between amyloid burden and clinical phenotype means that when evaluating therapy aimed at Aβ processing, Aβ can only be used to show that the intervention works biologically but not whether it will have a cognitive effect. Technically, the authors favor CSF Aβ42 levels as biomarkers, overlooking the fact that some investigators employ CSF Aβ40 or Aβ42:Aβ40 ratio. Clearly identifying which of these parameters is more sensitive or specific is an important goal, as well as comparing these to Aβ:tau ratios or perhaps another combination of the two molecular parameters. As the authors remark, the optimal cutoff point in any of these parameters is as yet unknown and may differ depending not only on the stage of the disease but possibly on other factors (apolipoprotein status, level of education, etc.). In addition, it is important to consider the purpose of the relevant study, where recruitment into a drug study may need to employ different cutoff points than a prospective study when the natural history of the disease is explored. The anatomical progression of the disease is still not well understood, as well as the sequence of biological processes leading to it. The authors seem to favor the old Braak & Braak staging of pathological cascade [3], while neglecting other brain areas such as the brain-stem changes which may well precede entorhinal changes, such as the locus ceruleus [4]. The hypothetical model of AD pathophysiological sequence fails to consider a critical factor, that of resilience and compensatory mechanisms, such as brain plasticity which are probably extremely important in determining the speed of cognitive deterioration and may be a critical factor explaining why some people harbor considerable brain pathology without evidence of cognitive decline. Clearly, much attention needs to be devoted to identify these processes and any biomarkers for their existence. The very early stages of cognitive impairment, for which the presently favored term now is mild cognitive impairment (MCI), is a central topic of discussion. As the authors clearly state, this issue carries societal and ethical repercussions beyond the scope of their discussions. In MCI cases with presumed underlying AD pathology, the authors prefer the term "MCI due to AD" although in many cases we do not initially know what is the underlying process. The use of biomarkers, while having certain research advantages, cannot be used in most cases because of cost and lack of availability of positron emission tomography or magnetic resonance imaging machines. Clearly, there is need to have a simple and easily accessible neuropsychological method, and several such are available [5]. Finally, these papers were written primarily by US investigators, with a minimal input from European dementia specialists and none at all from other geographical areas. Clearly, a wider representation would have added to the substance and contributed to the dissemination and acceptability of the new criteria.