On the reconceptualization of Alzheimer’s disease

Abstract In the hope of future treatments to prevent or slow down the disease, there is a strong movement towards an ever‐earlier detection of Alzheimer’s disease (AD). In conjunction with scientific developments, this has prompted a reconceptualization of AD, as a slowly progressive pathological process with a long asymptomatic phase. New concepts such as ‘preclinical’ and ‘prodromal’ AD have been introduced, raising a number of conceptual and ethical questions. We evaluate whether these new concepts are theoretically defensible, in light of theories of health and disease, and whether they should be understood as disease or as an at‐risk state. We introduce a pragmatic view on disease concepts and argue that an evaluation of the reconceptualization of AD should also take its aims and effects into account, and assess their ethical acceptability. The reconceptualization of AD is useful to coordinate research into preventive strategies, and may potentially benefit future patients. However, in the short term, early detection and labelling of ‘preclinical AD’ can potentially harm people. Since there is no treatment available and the predictive value is unclear, it may only create a group of ‘patients‐in‐waiting’ who may suffer from anxiety, uncertainty and stigmatization, but will never actually develop dementia. We conclude that only if the promise of preventive medication materializes, will the reconceptualization of AD turn out unequivocally to be for the better. Otherwise, the reconceptualization may do more harm than good.

a number of conceptual and ethical questions. We evaluate whether these new concepts are theoretically defensible, in light of theories of health and disease, and whether they should be understood as disease or as an at-risk state. We introduce a pragmatic view on disease concepts and argue that an evaluation of the reconceptualization of AD should also take its aims and effects into account, and assess their ethical acceptability. The reconceptualization of AD is useful to coordinate research into preventive strategies, and may potentially benefit future patients. However, in the short term, early detection and labelling of 'preclinical AD' can potentially harm people. Since there is no treatment available and the predictive value is unclear, it may only create a group of 'patients-in-waiting' who may suffer from anxiety, uncertainty and stigmatization, but will never actually develop dementia. We conclude that only if the promise of preventive medication materializes, will the reconceptualization of AD turn out unequivocally to be for the better. Otherwise, the reconceptualization may do more harm than good.

K E Y W O R D S
Alzheimer's disease, biomarkers, dementia, disease concepts, early diagnosis, pragmatism

THE C A S E
A 75-year-old man visits a memory clinic because of memory complaints over the past 6 months. His wife was the first to notice. He used to have a fantastic memory, but now he sometimes forgets conversations they had. He also has a tendency to repeat a story he has told not so long ago. The couple live independently and have an active social life. The 'patient' plays tennis in a near-by village, where he travels to by car or bike, depending on the weather. They have always done the financial administration of their household together.
This has not really changed, although he has become slightly insecure and lets his wife take more decisions than he used to. On formal cognitive testing he has slight memory impairment, but no impairment in other cognitive domains such as executive functioning, language or visuospatial functioning. This person clearly does not have dementia. In spite of some memory impairment he has almost normal and independent daily functioning. But the question is whether he is showing signs of a very early stage of a devastating brain disease that will eventually lead to dementia. Should we aim to label this person as having a disease at this point? And can we actually predict with sufficient accuracy if and when this person will eventually develop dementia? And if so, would this person benefit from this prediction?

| INTRODUC TI ON
In the hope of future treatments to prevent or slow down the disease, there is a strong movement towards an ever-earlier detection of Alzheimer's disease (AD). It is believed that brain changes presumed eventually to lead to dementia start to develop years to decades before clinical symptoms of dementia occur. 1 It seems attractive to develop treatments that can stop or slow down these changes as early as possible, even before symptoms of cognitive impairment arise. It is now possible to detect certain biomarkers -proteins in cerebrospinal fluid (CSF) and protein depositions on neuroimaging -that are presumed to reflect the early brain changes that may eventually lead to clinical dementia, in persons with no or only mild cognitive impairment. Biomarkers are thus increasingly seen as means for early detection of the disease. 2 In line with these scientific developments, a reconceptualization of the notion of AD is taking place. Instead of being defined by the clinical syndrome of dementia, AD is more and more depicted as a well-defined slowly progressive pathological process with a long asymptomatic phase. A new lexicon has even been proposed and new concepts and definitions have been introduced, such as 'preclinical' and 'prodromal' AD in those with no or only mild symptoms. 3 While currently mainly used in research, these new concepts of preclinical and prodromal AD, as well as the use of biomarkers that define these 'conditions', which were originally intended for research purposes, are now gradually entering clinical practice.
This reconceptualization of AD, and the concomitant emerging use of biomarkers aiming to detect biological evidence of a presumed pathological process, raise a number of conceptual and ethical questions. First, from a theoretical perspective it is questionable whether a state with abnormal biomarkers but without overt clinical symptoms should be considered as a disease, or rather as an at-risk state.
For the persons it concerns, the distinction between risk prediction and a very early diagnosis of a much-feared disease may not be that clear. This is further complicated by the fact that it is still contested how accurately biomarkers can predict future symptomatic disease.
Hence, the exact meanings of biomarker-based categorizations of preclinical and prodromal AD are unclear.
Second, from a moral perspective, the unclear predictive abilities of current biomarker tests and the current lack of meaningful treatment and prevention options -a lack of actionability -make it questionable whether the categorization of people as having an asymptomatic early phase of AD is defensible. It could create a group of 'patients in waiting' who may suffer from anxiety and fear for the future, but may never actually develop dementia.
Third, the motives for aiming at early detection are not always clear. Of course there is the intrinsic motivation of clinicians and researchers alike to stop the progression or even prevent the onset of dementia. Another driving factor is constant technological advancement, leading to the ability to detect molecular disease markers in body fluids as well as on neuroimaging. This technological advancement brings inevitable financial incentives, both for those who develop the biomarker tests ('diagnostic tests') and for those who foresee a large target population of people with an early stage of a disease for their new drugs. The persons it concerns are often driven by fear or the hope that something may be done to prevent a feared future with dementia; this may render them vulnerable to misconceptions and unrealistic expectations.
In this article we aim to address these important conceptual and ethical questions that result from the recent reconceptualization of AD. We will discuss whether the new concepts of preclinical and prodromal AD are theoretically defensible and ethically desirable and consider their implications for medical practice.

| FROM ALOIS AL ZHEIMER TO CURRENT D IS E A S E CRITERIA
Historically, the notion of AD has changed considerably since Alois Alzheimer first described the case of Auguste D who developed 'presenile dementia' in her early fifties, in 1907. For decades, AD was considered a rare form of dementia with an onset at relatively young age. This as opposed to the common 'senile dementia', which was considered to be attributable to cerebral atherosclerosis. It was not until the 1970s that neurologist Robert Katzman first suggested that most dementia cases, also in later life, should be considered as AD. 4 This led to a radical change in disease concepts and a dramatic increase in interest of the scientific world in dementia. In the 1980s the proteins of which the plaques and tangles Alois Alzheimer described consisted were identified as amyloid-β and tau, respectively. This further enhanced research and fueled the hope for earlier diagnosis and potential treatment.
In parallel, clinical concepts started shifting, and the early disease stages, before the onset of dementia -which is defined by impairment in daily functioning -got more attention. In the late 1990s, 'mild cognitive impairment' (MCI) was introduced as an entity defined by cognitive impairment, but not severe enough to impair daily functioning. 5 It became technically possible to assess the presence of amyloid-β and tau protein in CSF and even using nuclear imaging of the brain. Combining early clinical detection with detecting these biomarkers has triggered a completely new way of thinking about  According to these frameworks, AD is defined by a pathological process characterized by a specific sequence of brain changes, which develop over a long period of time and may or may not become symptomatic during life. Resulting from these frameworks, a new lexicon and new criteria have been proposed to capture these changing conceptualizations. Biomarkers, which can be detected using CSF analysis, magnetic resonance imaging (MRI) scanning and positron emission tomography (PET) scanning, play an important role in these definitions (Table 1). The main difference between the two conceptual frameworks is that according to the NIA-AA criteria one can have AD in the absence of any symptoms, whereas, according to the IWG, people without symptoms can only be labelled as 'at risk for AD'. Both frameworks consider people with MCI and abnormal biomarkers as having AD, irrespective of whether their cognitive impairment is progressive and will lead to dementia.
So over the last century AD seems to have changed from a rare presenile form of dementia, clinically defined and characterized by the presence of cognitive impairment leading to functional decline, of which a certain diagnosis could only be made post-mortem, to a common biologically defined condition in older people with or without cognitive impairment. The current views as represented in Table 1 illustrate the conception of AD as a slowly progressive pathophysiological process that will eventually lead to symptoms that will worsen and ultimately result in full-blown dementia. This process can presumably be detected early on (before any clinical symptoms are present) and is -also presumably -unidirectional. The often-invoked metaphor of a cascade (as in the amyloid cascade hypothesis 7 ) indicates necessary progression in a single direction: inevitable decline.
The current amyloid cascade model is, however, hypothetical.
The biological mechanisms are still insufficiently understood and it is not certain that the brain changes that define the condition do inevitably lead to symptoms. It is known that a considerable proportion of people dying in old age without cognitive impairment have a substantial load of amyloid-β and abnormal tau protein, considered to be the hallmarks of AD, in their brains. 8 In contrast, these changes are not always found in those dying at a greater age with clear symptoms of dementia, which was clinically diagnosed as Alzheimer's dementia. 9 Moreover, of those with MCI, only 5-15% develop dementia per year, depending on the population under study. Those with MCI who have these 'AD biomarkers' in their brain, do not always develop further cognitive decline, and after several years a considerable proportion has not developed dementia. 10 For people without cognitive impairment, data are scarce, but do suggest that abnormal biomarkers are common and do not indicate imminent dementia in the majority of people. 11 This brings us back to our main question: how should we evaluate the reconceptualization of AD? Are the new disease concepts theoretically defensible? And are they ethically desirable? We will start with the first question -which will eventually lead us to the second.

| AL ZHEIMER ' S D IS E A S E AND THEORIE S OF HE ALTH AND D IS E A S E
In order to consider the defensibility of the new disease concepts proposed in the AD field, we take a look at how health and disease in  Medicine, 360, 2302Medicine, 360, -2309 Of course one could argue that those with Alzheimer changes without dementia simply died before they had the chance to develop symptoms and those with a clinical diagnosis of Alzheimer's dementia who did not have AD pathology were misdiagnosed during life.
Still, these data, as well as the data on MCI, cast doubt on the validity of the amyloid cascade hypothesis.

| A PR AG MATIC APPROACH TO D IS E A S E CON CEP TS
The debate about definitions of disease in the philosophy of medicine has been focused primarily on the dispute between naturalist  A shortcoming of a purely social-constructivist approach, however, is that it is merely descriptive and not normative. It gives an account of how concepts of disease and disease classification are developed. As such, it offers no criteria to decide whether a certain conceptualization is a good one, or to determine whether one conceptualization, definition, or classification is better or worse than another. We therefore propose a pragmatic approach to concepts and classifications of disease. 27 Philosophical pragmatism understands definitions, concepts and classifications as tools.
From this perspective, it makes sense to ask what new definitions or concepts are supposed to do -what their goal or aim is -and whether they do this well. 28 Moreover, it makes sense to look at the, perhaps unintended or unforeseen, effects that the introduction and usage of these tools has, and take these into account in the evaluation. As Kingma says: 'the reason we should care about ideas, concepts and classifications is that they have effects', 29 and these effects can be good or bad, desirable or undesirable. Conceptions and classifications of diseases do not exist in a vacuum, but they influence practices, create new realities and have consequences for people's lives. As such, they are ethically relevant and hence they can, and should, be evaluated from a moral point of view.
Taking a pragmatic approach thus implies that a moral evaluation of the effects of a newly proposed concept, definition or classification, becomes an integral part of the evaluation of its defensibility. Whether the reconceptualization of AD is defensible is therefore not merely an epistemic, scientific question, but also a normative, moral issue. So, the questions become: what does the reconceptualization of AD aim to achieve and does it do so well? And what are the practical effects and consequences of this reconceptualization and are they desirable?

| E VALUATING THE CON CEP TS OF PRECLINIC AL AND PRODROMAL AL ZHEIMER ' S D IS E A S E
Evaluating the reconceptualization of AD in relation to scientific evidence is primarily a task for scientists, and one that is presently being taken on in the Alzheimer's research literature. we will focus on evaluating the aims, goals and effects -both intended and unintended -of the concepts of preclinical and prodromal AD, referring to a state in which there is no dementia, but abnormal biomarkers presumably related to AD are present.

| Aims and goals of the reconceptualization
The first evaluative question to be asked from a pragmatic perspec- tive is: what are the aims of the new concepts and are these attained?
From the articles published by the NIA-AA and IWG it is apparent that the first aim of introducing the new concepts and classification is to enable and support research in this area. In order to perform research into the pathophysiological processes assumed to underlie AD, to test the amyloid hypothesis and to assess the predictive value of various biomarkers, it is necessary to use a common vocabulary Alzheimer's dementia. Whether this goal will be successfully attained, the future will tell. For the moment, no interventions aiming to prevent or modify clinical symptoms of dementia have been successful.

| Effects of the reconceptualization on individuals
The next important evaluative question regards the effects of the reconceptualization. These can be both intended or unintendedbut sometimes foreseeable -and can manifest at either the individual or societal level.
A first concern here is what happens when the terminology that was introduced in the context of research, and intended primarily for communication between researchers, also gets employed in the clinic and in communication with research participants, patients and their family members, as is often the case. The emotional and social effects of terms chosen to communicate with lay-people can be considerable; being told one is 'at risk' for developing AD is different from being told one has preclinical or asymptomatic AD -although the situations these terms aim to describe may be exactly the same.
Likewise, being told one is in the early stages of AD is different from being told that biomarkers have been detected that may indicate that one is at increased risk of developing dementia within 10-15 years. In dealing with research participants and patients, terminology has different connotations and effects than in communication among researchers. Terms chosen should convey a truthful image of the condition of the person, and not cause confusion, unnecessary anxiety or misconceptions. 32 Giving someone a diagnosis can be understood as a 'speech act': it turns a healthy person or research participant into a patient, which has considerable psychological and social consequences and may be harmful, particularly in the case discussed here, where the person it concerns may never develop any symptoms. 33 A second concern is that defining these new conditions of 'being at risk' or having preclinical and prodromal AD -whatever the exact terms that one uses to describe them -is not necessarily in the best interest of the people involved. Another worry is that people may come to have a different perception of themselves once they know about their condition, start to worry more over normal memory lapses or even develop a nocebo reaction, i.e., show cognitive decline as a result of getting labelled. 37 Friends and family may also change their attitude or behaviour towards those persons and this may negatively interfere with their relationships.
On the other hand, it has been argued that people do have a right to know their risks, if they want, and that this would enable one to plan and prepare for the future -financially, mentally, with regard to life plans or lifestyle. Even though a risk prediction is not 'actionable' in the sense of available preventive or treatment options, it is often claimed it can still be used for such non-medical purposes. 38 However, as we have argued elsewhere in more detail, as long as the prognostic value of the diagnosis 'preclinical AD' is so unclear and hence risk prediction is inaccurate, this argument has a limited validity. 39 At present, there is insufficient knowledge about the actual frequency of these potential harms and burdens and about their severity. Little is known about the psychological impact of learning one's risk status, or learning that one has 'preclinical' or 'prodromal' AD.

| Societal effects
Finally, we should consider the effects of the reconceptualization on a societal level. 40 Depending on the context and the level of public awareness, a large proportion of the elderly population may eventually prove to be 'at risk for AD' or even to have (an early stage of) AD.
Although population-based screening programmes for those over 65 do not seem very likely at this moment, other levels of screening for cognitive impairment may become reality, such as in the context of comprehensive geriatric assessments, which are increasingly popular, or opportunistic screening in those who are presumed to be at risk for cognitive impairment due to clinical or demographic characteristics.
Labelling large numbers of people as such will inevitably raise questions about medicalization of aging and age-related memory problems, since a growing number of people will be diagnosed with The reconceptualization is taking out a mortgage on the future success of a specific research agenda, but it may do so at the expense of current research participants, patients and older people in general. This is not necessarily unethical but we should at least be aware of it, weigh the pros and cons and minimize the negative effects. One way to do this is to be very careful with the vocabulary used to address research participants. As Alzheimer Europe states in a recent report: 'Careful attention should be paid by researchers to the terminology surrounding what is currently defined as pre-clinical AD and to its possible impact on research participants and the general public.' 43 They recommend, for example, that research participants falling in the preclinical AD category, should be described as 'being at risk of AD' rather than as 'having preclinical AD', and that researchers should speak of 'disclosure of risk status', rather than of 'diagnosis'. 44 Moreover, we believe clinicians who are also involved in research should be acutely aware of these issues and should take utmost care not to conflate research language with clinical language when speaking to patients. We should consider the reconceptualization and the proposed terminology and classification as research tools, not as referring to clinical 38 For example, Vanderschaeghe, G., Schaeverbeke, J., Vandenberghe, R., & Dierickx, K.
entities. The AD research community should take responsibility to prevent terms like 'preclinical AD', 'asymptomatic AD' or 'at risk for AD' from prematurely entering the consulting room and the public domain. 45

| CON CLUS ION
While the reconceptualization of AD, and especially the introduction of the notion of preclinical or asymptomatic AD, might seem attractive for research into preventive strategies, and may have the potential to benefit future patients, it will not benefit individuals in the short term. It may lead to diagnosing a pre-symptomatic condition that in a considerable proportion of cases will never become symptomatic.
This can be harmful for individuals and their caregivers. It is important to consider the possible harmful effects of creating these new, uncertain and unclear conditions of pre-dementia AD in evaluating the defensibility of the proposed reconceptualization. A fundamental shortcoming in the current scientific AD debate is that illness is overlooked and that the disease is being oversimplified by characterizing it as the presence of biological abnormalities which in themselves correlate poorly with the clinical symptoms of cognitive impairment.
We conclude that the reconceptualization of AD is legitimate and meaningful for usage within a narrowly defined research community studying a clearly defined biological condition, namely the presence of specific measurable biomarkers, but that translation to clinical practice poses various ethical and communication problems. It is too early to move those concepts from research into the clinical setting, since they are based on a disputed hypothesis and since attempts to do so may actually be harmful to the people concerned. The distinction between research and clinical practice may be difficult to maintain, however, and it appears as if the use of biomarkers is slowly creeping into clinical practice, without proper scientific underpinning. 46 Whether it is a good idea to move toward ever-earlier diagnosis of AD, or of detecting at-risk states for AD dementia, is a complex question. A good predictive value and actionability are important preconditions for the ethical implementation of predictive testing. With regard to the first condition, biomarker tests for AD currently fall short, while, for the time being, the second is limited to 'preparing for one's future'.
Only if the promise of preventive medication materializes, will the reconceptualization of AD turn out to be unequivocally for the better.
However, if the attempts to develop such medication continue to fail, the reconceptualization may do more harm than good.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

ACK N OWLED G EM ENTS
This work is part of the research programme 'Early diagnosis of Alzheimer's disease -conceptual and ethical issues' with project number 731010012, which is financed by the Netherlands Organisation for Health Research and Development (ZonMW).