Alzheimer's disease diagnostic criteria: practical applications

Alzheimer's disease (AD) can be identified prior to the occurrence of dementia by using biomarkers. Three phases of AD are recognized: an asymptomatic biomarker-positive phase, a phase with positive biomarkers and mild cognitive deficits, and a dementia phase. Codification of these phases was first accomplished in 2007 by an International Work Group (IWG) led by Bruno Dubois. The definitions relevant to the approach were further clarified in 2010. In 2011, the National Institute on Aging/Alzheimer's Association (NIA/AA) established three work groups to develop definitions and criteria for these three phases of AD. The criteria of the IWG and those of the NIA/AA have many similarities and important differences. The two sets of criteria concur in recognizing the onset of AD prior to dementia. The three phases of AD described in both sets of criteria embrace the same clinical entities but with different terminologies and emphases. IWG criteria emphasize a single clinico-biological approach that includes all symptomatic phases of AD and uses the same diagnostic framework across the spectrum of symptomatic disease; the NIA/AA criteria apply different diagnostic approaches to the three phases. Biomarkers are an integrated and required part of the IWG criteria and are optional in the NIA/AA approach. Both sets of criteria have substantial strengths, but new information demonstrates shortcomings that can be addressed in future revisions of the criteria. These new criteria have profound implications, including greatly increasing the number of people identified as suffering from AD and increasing the time that patients will spend with knowledge of the presence of the disease.

AD can be reconceptualized as a progressive disorder that advances from biological changes in the brain with no accompanying cognitive abnormalities, to a state of memory impairment with biomarker abnormalities indicative of AD, to mild, moderate, severe, and profound stages of AD dementia.

International Work Group criteria
Dubois and colleagues, of the International Work Group (IWG), developed new criteria for the diagnosis of AD as a clinico-biological entity [11] and produced a lexicon of terminology to address the complex issues associated with this marked revision of the way AD is conceived [12]. Th e author of this review is a member of the IWG. Th e new criteria have several important aspects: (a) the diagnosis of AD can be made in the living individual and is no longer dependent on autopsy confi rmation; this is a shift from the clinico-pathological paradigm to a clinicobiological paradigm; (b) the diagnosis of AD can be made with greater certainty and does not use the terminology of 'probable AD'; (c) other diseases can be excluded with biomarkers, addressing the poor negative predictive value of earlier criteria; (d) the diagnosis of AD can be made prior to the onset of dementia; (e) the identifi cation of AD pathology placing the person at risk for progressing to symptomatic AD can be made in asymptomatic individuals; (f ) the same clinico-biological approach and same criteria can be used for diagnosis of all stages of symptomatic AD regardless of whether dementia is present; (g) the most common phenotype and typical presentation of AD are a hippocampal type of memory abnormality with poor storage and more response to cuing as demonstrated on a Free and Cued Selective Reminding Test; (h) several biomarkers can support the presence of AD pathology and fulfi ll the criteria; and (i) pathophysiologic (amyloid imaging and CSF Aβ and tau measures) and topographic (MRI, FDG PET, and SPECT) biomarkers are recognized [11,12]. In the IWG criteria, the concept of mild cognitive impairment (MCI) is abandoned in favor of the more specifi c prodromal AD for those patients with symptomatic predementia AD. Th e IWG criteria do not off er any nomenclature for the general nonspecifi c syndrome of 'MCI' in favor of developing criteria that use biomarkers to identify that subset of MCI patients who are in the predementia phase of AD.
Patients without cognitive abnormalities and with positive biomarkers are identifi ed as having AD pathology, an AD risk state. Th e IWG describes AD as a clinicobiological disorder; if no symptoms are present, then this is not a disease state. Th e presence of the biomarker repre sents a risk factor for progressing to AD in the future. Th e proportion of people who progress, the time frame for progression, and additional risk factors for progression are currently not fully defi ned [13][14][15]. Th e update of the criteria expanded the identifi cation of diff er ent manifestations of AD dementia or atypical forms of AD dementia, including posterior cortical atrophy, the frontal variant of AD dementia, and the logopenic presenta tion of AD dementia [12].
Although these criteria represent a conceptual advance, they can be improved. Since the publication of these criteria, it has become evident that the physiological markers are more specifi c to the AD bio logical process; topographic markers are more sensi tive to disease progression but less specifi c to AD [13]. Atrophy, hypometabolism, and hypoperfusion occur in a variety of disease states, whereas amyloid abnormali ties are present in a very restricted number of disorders, predominantly AD. Moreover, the absence of Aβ abnor malities is strong evidence against AD as the cause of the cognitive changes, whereas negative fi ndings on MRI, FDG PET, or SPECT do not have the same negative predictive value.
Requiring a pathophysiologic biomarker to support the presence of AD will improve diagnostic accuracy. In addition, the IWG criteria recognize atypical forms of AD dementia but not atypical manifestations of prodromal AD. Extending this diagnostic framework backwards into prodromal AD will improve diagnostic continuity and sensitivity. Th e issue of mutation carriers can also be clarifi ed in the criteria. Th e presence of a hippocampal type of amnestic disorder in a patient who has a known causative mutation and who presents at an age compatible with the mutation expression can be diagnosed as prodromal AD or AD dementia depending on the presence of the accompanying functional defi cits. Merely being in a family known to carry the mutation, however, should not be suffi cient to make this diagnosis, as there are unusual non-AD causes of hippocampal amnesia and these can occur in non-carriers of mutationbearing families. Revisiting and adjusting the criteria in this way would reduce the opportunity for diagnostic ambiguity in their application.

National Institute on Aging/Alzheimer's Association criteria
After the two publications of the IWG criteria, the National Institute on Aging (NIA) and the Alzheimer's Association (AA) convened three working groups to develop criteria for the asymptomatic, minimally symptomatic, and dementia phases of AD [16][17][18]. Th ese criteria share many features with the IWG criteria, including recognition of an asymptomatic biomarker-positive phase of AD pathology, identifi cation of a predementia symptomatic phase of AD, retention of criteria for AD dementia, integration of biomarkers into some approaches to the diagnostic process, and categorization of biomarkers into two types: one identifying Aβ abnormalities and one for neurodegeneration.
A notable aspect of the NIA/AA criteria for the preclinical asymptomatic phase of AD is the description of a sequential appearance of abnormalities from stage 1 with abnormalities of amyloid only, to stage 2 with bio markers of both amyloid abnormalities and neuro degenera tion, to stage 3 with both types of biomarkers and minimal clinical decline not meeting criteria for MCI [16].
Th e predementia sympto matic phase of AD retains the concept of MCI in the NIA/AA approach [17]. Th is has the advantage of being able to be implemented in clinical practice where biomarkers are unavailable. Non-AD as well as AD types of MCI are recognized in the criteria. Th ere is some ambiguity in the use of 'MCI' versus 'MCI due to AD' in the criteria. MCI is defi ned by using the original criteria of Petersen and colleagues [19]. Biomarkers are not required for the diagnosis of MCI, although the diagnosis of MCI due to AD would require biomarker support given the heterogeneous nature of the MCI syndrome [20,21]. For research application, biomarkers are intro duced and stratifi ed. If biomarkers of both amyloid abnormalities and neurodegeneration are present, the term 'MCI due to AD -high likelihood' is applied; if only one type of biomarker is collected and is positive, then the term 'MCI due to AD -intermediate likelihood' is used. If no biomarker is collected, the results are ambi gu ous, or two biomarkers are contradictory, then the bio markers are uninformative and the relationship to AD is uncertain. Th e NIA/AA MCI approach does not establish a continuity with the criteria suggested for preclinical AD in which both types of biomarkers are implied to be positive by the time the patient becomes symptomatic [16].
Th e stratifi cation of biomarkers off ered in the NIA/AA approach is challenging to apply in practice since biomarkers are not strictly and consistently related to each other. Patients with MCI and positive amyloid imaging but with an MRI that does not reach a predefi ned cutoff for excessive atrophy would receive a diagnosis of MCI due to AD of uncertain likelihood since the biomarkers are inconsistent with each other, although there are few competing diagnoses for this set of observations. Likewise, a patient with MCI and hippocampal atrophy on MRI and negative amyloid imaging would receive the same diagnosis of MCI due to AD of uncertain likelihood, although the diagnosis of AD in this case would be biologically unlikely.
MCI is distinguished from dementia by the presence of abnormalities of activities of daily living (ADLs). Th is is an arbitrary distinction based on clinical judgment and is subject to both clinician biases and patient or caregiver reporting errors. Studies of ADL rating scales in MCI demonstrate that impairment of at least some high-level instrumental ADLs is common in MCI [22]. Morris [23] analyzed the eff ect of the NIA/AA criteria on diagnostic divisions. Th e author showed that, if these criteria were applied, many conditions currently classifi ed as mild AD would be identifi ed as MCI. Th is shift would have many ramifi cations, including allowing patients with a more advanced pathology into MCI trials.
Th e NIA/AA criteria for AD dementia retain most of the features of the past diagnosis of probable AD [1] despite the limited positive predictive value and poor negative predictive value of these criteria [4]. Diagnostic standards for all-cause dementia are provided, and 10 categories of dementia of the AD type -including probable AD dementia, possible AD dementia, probable or possible AD dementia with evidence of the AD pathophysiological process, and pathophysiologically proved AD dementia -are established [18]. Probable AD dementia meets clinical diagnostic criteria without support ing biomarker evidence of AD. Biomarkers are integrated into the criteria to describe probable or possible AD dementia with pathophysiological evidence of AD and are stratifi ed as of high likelihood of being due to AD if amyloid abnormalities and evidence of neurodegeneration are both present and of intermediate likelihood of being due to AD if only one type of biomarker evidence is present.
Th e terminology of 'probable AD dementia of intermediate likelihood of being due to AD' may be confusing for clinical and research application. When NIA/AA criteria are used, the patient with the clinical syndrome of probable AD, no available structural imaging, and positive amyloid imaging would be receive a diagnosis of probable AD of intermediate likelihood of being due to AD, although there are few (if any) alternatives to the diagnosis of AD in this case. A patient with an AD-like phenotype, atrophy on MRI, and negative amyloid imaging would receive a diagnosis of probable AD with uncertain likelihood of AD, although this person is unlikely to have AD. Th e person with the AD dementia phenotype and no biomarkers would receive a diagnosis of probable AD.
Th e NIA/AA criteria are an improvement over the preceding diagnostic approach of the NINCDS/ADRDA criteria and are a step toward electively integrating new biomarker information into a diagnostic framework. Like the IWG criteria, the NIA/AA criteria can be improved. Th e use of three diff erent approaches to diagnose the three phases of the same illness fails to capture the evolving idea of a single disorder occurring along a spectrum of severity. Th e concept of a single stereotyped sequence of biomarker and clinical progression in the preclinical phase of AD does not refl ect the heterogeneity of fi ndings observed in clinical practice [24]. Th e ambiguity of MCI versus MCI due to AD can be clarifi ed, and the diagnostic complexity of the approach to AD dementia off ered can be simplifi ed to assist in clinical application. Th e criteria for identifying someone as manifesting MCI rather than mild AD may not be suffi ciently stringent [23]. Th e stratifi cation of biomarkers used in both the MCI and AD dementia approaches is challenging in clinical implementation; the data support the diagnostic specifi city of amyloid biomarkers over biomarkers of neurodegeneration and these observations can be incorporated into improved criteria.

Implications of new criteria
Two pivotal reconceptualizations of AD have occurred in the history of our understanding of this disease, and both have led to dramatic increases in the recognition of the number of people aff ected. Th e fi rst was led by Robert Katzman, who in 1976 [25] drew attention to the identity between 'presenile' AD and 'senile dementia' . He recognized that AD is the major cause of progressive cognitive impairment in the elderly and is bound to grow to epidemic proportions with the aging of the world's population. Th e second is the pioneering work of the IWG. Th is work synthesizes the available evidence into a cohesive framework for recognition of AD as a clinicobiological disorder that can be diagnosed in life and embraces a spectrum of severity from very mild symp toms to profound dementia [11,12]. Th is change in paradigm also has the eff ect of greatly increasing the number of individuals recognized as manifesting AD. Th e projection of 80 million cases of AD by 2040 [26] is based on projections of AD dementia and will be multiplied by a factor of at least 2 if predementia cases of AD are included.
Another eff ect of making the diagnosis earlier in the disease and including predementia cases in the diagnosis of AD is the dilemma it raises for patient care [27]. Patients deserve to know what the clinician knows and to be informed about the limits of our knowledge. Th e diagnosis of prodromal AD is often -but not alwaysassociated with transition to AD dementia over the ensuing few years. Hertze and colleagues [28] found that 71% of patients with the syndrome of MCI and AD-type CSF fi ndings, including low Aβ and high tau, progressed to AD dementia within 5 years. Twenty-nine percent of such patients did not develop a dementia syndrome and were cognitively stable even after half a decade of observation [28]. Th is represents uncertainty for the physician and hope for the patient and patient's family. When individuals are given a diagnosis of prodromal AD, they will carry the diagnosis of AD for many years and must be resigned to the uncertainty of the duration of this phase of the disorder. Helping patients live with AD will be a major theme of coming research and care.

Comment
Rapidly evolving information about the biology of AD and biomarker windows on the disorder are reshaping our understanding of the disease at an accelerating pace. Th e IWG and the NIA/AA work groups support the diagnosis of AD prior to the onset of dementia and point to how best to integrate biomarkers into diagnostic criteria. Both sets of available criteria can be revised to better capture current data on the role of amyloid as a diagnostic hallmark and early indicator of the presence of the AD.
An important goal of developing new criteria is to support clinical trials and advance the new therapeutics for AD. So that optimal cognitive function is maintained and the progression of AD into the dementia phase is deferred, a diagnostic framework for predementia AD is needed. Many trials have involved patients with diagnosed MCI and none has led to the development of a new treatment or extension of currently available treatments into the MCI phase of AD. Th e heterogeneity of MCI and the lack of uniform biological targets in this non specifi c syndrome may have contributed to these failures [20,21]. IWG criteria for prodromal AD have been successfully implemented in current phase 2 clinical trials for small molecules, immunotherapies, and medical foods and have been accepted by the European Medicines Agency [29,30] for use in AD clinical trials. A conundrum in the clinical trial application of research diagnostic criteria is how to make them specifi c enough to capture a population with a homogeneous biology but not so narrow that they exclude many patients who have AD but who lack a specifi c phenotype or biomarker profi le.
Th e preclinical phase of AD (NIA/AA criteria) or AD risk state (IWG criteria) sets the stage for the imple mentation of prevention trials of individuals who do not have symptoms of AD but who harbor the amyloid changes that have been associated with the presence of the disease. Th e goal of these studies will be to delay the onset of cognitive and functional decline. Such studies will be facilitated by refi ned diagnostic criteria.
Th e emergence of biomarkers is the basis for both justifying and formulating new diagnostic criteria. Without progress in biomarkers, there would be no need for new criteria. Th e IWG criteria require both clinical and biomarker evidence to identify the diff erent phases of AD and AD pathology. In NIA/AA criteria, use of biomarkers is optional and clinical information alone can suffi ce for diagnosis of MCI or probable AD dementia. Among biomarkers, amyloid abnormalities may have a particularly informative role in AD risk states and prodromal AD.
Regulatory language used with the recent approval of fl orbetapir emphasizes that negative amyloid imaging in persons with cognitive impairment is inconsistent with identifying neuritic plaques as the cause of the cognitive decline and therefore inconsistent with a diagnosis of AD as the cause of the cognitive syndrome. AD is not currently excluded with negative amyloid imaging in either set of diagnostic criteria. It will be important to align regulatory and clinical criteria in future versions of these diagnostic approaches.
Positive amyloid imaging is less informative than negative imaging and occurs in patients with AD, Lewy body dementia of the AD type, or amyloid angiopathy and in cognitively normal persons [31,32]. Th e combination of careful description of the phenotype in conjunction with the use of biomarkers is critical to accurate diagnosis and will be assisted by diagnostic criteria.
Both sets of criteria emphasize the research nature and need for studies of the sensitivity, specifi city, and positive and negative predictive value of the criteria. Th e urgent need of clinicians for guidance in how to apply biomarkers and improve diagnostic standards means that new technologies and approaches will be incorporated rapidly into clinical practice where the appropriate biomarkers are available. Rapidly advanc ing research on biomarkers provides new information with which to revise the criteria and enhance their clinical utility. Th is is a necessary step toward advancing new and desperately needed treatments for AD.