Article Text
Abstract
Many healthy volunteers choose to take part in Alzheimer’s disease (AD) prevention studies because they want to know whether they will develop dementia—and what they can do to reduce their risk—and are therefore interested in learning the results of AD biomarker tests. Proponents of AD biomarker disclosure often refer to the personal utility of AD biomarkers, claiming that research participants will be able to use AD biomarker information for personal purposes, such as planning ahead or making important life decisions. In this paper, the claim that AD biomarkers have personal utility for asymptomatic individuals is critically assessed. It demonstrates that in the absence of clinical validity, AD biomarkers cannot have personal utility and do not serve research participants’ autonomy. Over the next few years, many research groups will be confronted with participants’ preferences to learn the results of AD biomarker tests. When researchers choose to make results available upon explicit request, they should ensure adequate information provision and education, notably on the uncertain clinical significance of AD biomarker information. Routine disclosure of AD biomarkers to cognitively unimpaired individuals in research settings cannot be justified with an appeal to the personal utility of AD biomarker information.
- clinical trials
- dementia
- research ethics
- autonomy
- genetic screening/testing
Statistics from Altmetric.com
Introduction
Alzheimer’s disease (AD) is a growing public health concern, its prevalence rising as the population ages. Despite decades of research, AD remains incurable, and effective disease-modifying treatment options are lacking. Researchers are still struggling to understand the pathophysiological processes underlying the development of AD which are believed to begin up to 20 years before the onset of symptoms of dementia.1 Although investigational drugs have not been proven to cure or ameliorate AD, it is hypothesised that when applied in asymptomatic individuals, they may prevent irreversible neuronal damage, and thus avert or delay the onset of clinical symptoms of dementia.
Concerted efforts are currently under way to test drugs administered much earlier in the pathophysiological process, in prevention studies.2 Prevention studies require research participants who are cognitively unimpaired, but are undergoing pathological changes in the brain that are associated with AD, a condition currently categorised as ‘preclinical AD’ or ‘asymptomatic at risk for AD’.3 4 Participants in these trials may be recruited from the general population or through existing longitudinal cohort studies (not necessarily focused on AD), in which they are examined inter alia for genetic and non-genetic AD biomarkers, and monitored over time. Through biomarker testing, healthy volunteers may thus be relabelled as having preclinical AD, thereby becoming eligible for prevention clinical trials.
It is ethically contentious to screen healthy individuals for ‘preclinical AD’ or ‘asymptomatic at risk of AD’ in the context of research. After all, in the absence of effective treatment or preventive options, informing people about the possibility of an increased risk of developing AD dementia in the future may bring more harm than good. However, it is generally acknowledged that research participants should know why they are invited to take part in research.5 In clinical trials, therefore, individual research results are often routinely disclosed to participants as part of the recruitment process. In longitudinal cohort studies, by contrast, especially those that are focused on general health, not AD, it is not yet clear whether or not AD biomarkers should be—actively—disclosed to research participants; most cohort studies do not disclose until research participants are selected for follow-on studies based on biomarker information.6
Proponents of active disclosure refer to the personal benefits of AD biomarker information (eg, on the APOE gene or imaging biomarkers) as a justification for risk disclosure in research.7 8 Also, many research participants prefer disclosure, often with an appeal to the personal utility of this information.9 10 But is this argument valid? In this paper, we examine whether policies for routine communication of AD biomarkers in research studies with healthy volunteers can indeed be justified with an appeal to the personal utility of biomarker-based AD risk information.
The question whether personal utility can serve as a justification for disclosure has immediate practical relevance, as over the next couple of years, many thousands of research participants will be invited to take part in large-scale longitudinal cohort studies or clinical trials on the prevention of AD. Also, in some countries, healthy individuals can register through online platforms, such as Join Dementia Research in the UK or Hersenonderzoek.nl in the Netherlands to become volunteers in dementia research. Although at the time of prerecruitment, it is not yet known whether the individual will be invited to take part in one or more studies, what those studies will be and what those studies’ policies for the return of individual research results will be, volunteers are encouraged to register by statements such as ‘Discover how healthy your brain is.’11 Many of those volunteers would likely be interested in learning about AD biomarkers.
Consequently, many research groups will be confronted with the question whether or not to disclose AD biomarkers to research participants, at what point in time and on what grounds. While current clinical guidelines recommend against the clinical use of AD biomarkers in asymptomatic individuals, as the prognostic value in these populations ‘remains investigational,’12 some degree of disclosure of AD biomarkers in the research context will be unavoidable. After all, from the fundamental right of access to personal data,13 it follows that on explicit request, individual research results should be disclosed to research participants. There also seems to be increasing support for the more active disclosure of AD biomarkers in research.14 This concords with broader arguments that it is ‘unduly paternalistic’ to withhold any individual research results from research participants.15 As disclosure of AD biomarkers in research is on the rise, it is important to critically assess the notion of personal utility as one of its primary justifications.
The clinical validity and utility of AD biomarkers
The pathophysiological process leading to AD dementia is associated with the accumulation of deposits of amyloid-β and tau proteins in the brain. This process, also known as the amyloid cascade hypothesis, is thought to begin long before the onset of clinical symptoms, possibly decades before. AD is currently understood as a continuum,1 ranging from a state in which individuals are cognitively unimpaired while disease pathology may already manifest in the brain, to the experiencing of subjective cognitive decline which may or may not subside, to mild cognitive impairment (MCI) ‘in the absence of a significant effect on instrumental activities of daily living,’16 which may or may not progress to AD dementia. The various stages of this continuum can be assessed using a combination of cognitive testing, evaluation of daily functioning and biomarker testing. According to current classifications, ‘preclinical AD’ or the ‘asymptomatic at risk for AD’ status can be established on the basis of amyloid-β, tau or other biomarkers,3 4 also before the onset of symptoms, in cognitively unimpaired individuals. There are several AD biomarkers that are currently used in research and/or clinical care to determine whether an individual either has (preclinical) AD or is at risk of developing AD dementia.
First, amyloid β1–42 and total tau or phosphorylated tau (P-tau) levels can be detected in cerebrospinal fluid (CSF), for which research participants or patients need to undergo a lumbar puncture. Although the predictive value of CSF biomarkers has not sufficiently been studied in cognitively unimpaired individuals, it has been studied in individuals with MCI. Results from observational cohort studies suggest that the combination of decreased amyloid β1–42 and increased (P-)tau levels detected in CSF can contribute to a better prediction of which patients with MCI will progress to develop dementia in the years to come.17 However, most of these studies were not designed as diagnostic accuracy studies and are prone to several sources of bias, limiting the accuracy of biomarker values as diagnostic or prognostic tools. From a clinical perspective, results of CSF analysis have little added value over and above clinical assessment of memory functioning to predict whether patients with MCI will or will not develop dementia over the next years.18 In cognitively unimpaired individuals, the clinical validity of CSF biomarkers is as of yet unknown, but expected to be even more limited than in individuals with MCI.
The same holds for amyloid plaques that can be detected using positron emission tomography (PET) scans. For the acquisition of PET scans, the intravenous injection of a radioactive tracer is required. Two tracers—flutemetamol and florbetapir—were recently approved by the Food and Drug Administration and European Medicines Agency for diagnostic use in adults presenting with complaints of cognitive impairment who are evaluated for AD dementia. While ‘amyloid PET negativity’ adequately rules out amyloid pathology (which is not equivalent to ruling our dementia in the near future), the clinical significance of ‘amyloid PET positivity’ is not yet clear, as not all patients with MCI who are biomarker positive on amyloid PET are—or will become—symptomatic. It is even less clear in the 20%–40% of cognitively unimpaired elderly people in whom elevated levels of amyloid-β can be found in the brain.19 Thus, although PET can be used in patients with memory impairments to establish whether they likely have MCI due to AD, these biomarkers have uncertain prognostic value in cognitively unimpaired populations, and are not currently recommended for routine use in these individuals or in individuals with subjective cognitive decline.
In sum, in cognitively unimpaired research participants the value of AD biomarkers has yet to be established. As of yet, biomarkers cannot be used to distinguish reliably between individuals who likely will or will not develop clinical symptoms of AD in—say—5 or 10 years’ time. In other words, their predictive value is unknown, and consequently, they have unknown so-called clinical validity (see box 1), which is defined as the ability of a test to identify individuals with the intended phenotype—that is, a disease or a risk factor for a disease.20
Definitions of clinical validity, clinical utility and personal utility
Clinical validity: the predictive value of the biomarker, or the extent to which the biomarker distinguishes between those who will develop the disease and those who will not.
Clinical utility: the extent to which the biomarker test will affect clinical management and improve the individual’s health.
Personal utility: the extent to which the biomarker test has the potential to effect change on a (non-medical) personal level.
In contrast to non-genetic biomarkers, genetic biomarkers can be more clearly associated with an increased risk of developing AD dementia in the future. Notably, the APOE gene has been studied extensively. Having one copy of the APOE ε4 variant constitutes a 3.5-fold increased risk of AD, while having two copies of the APOE ε4 variant (which occurs in less than 2% of the general population) constitutes an up to 15-fold risk for AD21: this biomarker may have clinical validity, especially for those who have two copies of the APOE ε4 variant. Yet, there are carriers of two copies of the APOE ε4 variant who never develop AD dementia.
There are currently no proven effective strategies to prevent AD dementia. Some modifiable risk factors have been identified, including cardiovascular risk factors and diabetes, lifestyle factors (eg, physical inactivity) and psychosocial factors (eg, low educational level), which together account for up to a third of AD cases worldwide. It is not clear, however, whether and to what extent modification of these risk factors will prevent or delay the onset of AD dementia in individuals who are free from symptoms but biomarker positive.
A biomarker test has clinical utility when it helps clinicians in making treatment decisions conditioned on the result of the test that effectively improve patients’ health (see box 1). This means that if clinicians do not have different or better treatments to recommend to biomarker-positive than to biomarker-negative patients, the test has no clinical utility. In the absence of options to prevent the development of cognitive impairment in biomarker-positive individuals, it follows that AD biomarkers cannot lead to health benefit and therefore do not have clinical utility in cognitively unimpaired individuals.
Defining personal utility
Various authors have suggested that although there may not be clinical utility in knowing about an increased risk of AD, there may instead be personal utility.22–24 Studies have shown, for instance, that people are interested in APOE susceptibility testing for AD for a variety of medical and non-medical reasons, ranging from arranging one’s financial affairs or advance care planning to preparing one’s family members for the possibility of AD or making altruistic contributions to scientific research.23 In fact, APOE testing for AD risk is often mentioned as a paradigmatic example of the personal utility of predictive tests19 20: knowledge of one’s AD risk is thought to be useful personally, rather than clinically, and is thought to affect people’s personal lives, and their social, practical, financial and lifestyle decisions. People have been found to value ‘the information itself, rather than its potential to improve their health.’23 Similar results have been found in relation to non-genetic biomarker testing, particularly among individuals who see themselves as at higher risk of AD, who have a family history of disease or who have cognitive complaints.10 25
Elsewhere, a stringent definition of personal utility has been proposed,20 that is—in line with definitions of clinical utility—focused on the change a biomarker test may effect in people’s personal lives (see box 1). A test can only have personal utility when (A) it has clinical validity; and (B) it can ‘reasonably be used for decisions, actions or self-understanding which are personal in nature.’20 This means that when biomarker tests cannot accurately estimate the likelihood of developing a disease within a given time frame, they cannot have personal utility. Also, biomarker tests can only have personal utility when they can be used on a personal level, for example, for one’s view of oneself or one’s life in the past or in the future and/or for personal decision-making. Biomarker tests can be said to have personal utility when they allow individuals to understand themselves differently—which may be of value to them—or to make different (reasonable) choices based on the result. Furthermore, these choices should have the potential to effect positive change in people’s lives.
Others have implicitly or explicitly used broader definitions of personal utility,26 27 especially in the context of direct-to-consumer personal genome testing. They claim, for example, that if people value the test procedures or the test results for whatever reason, even, for instance, because they find them amusing, or interesting to talk about at birthday parties, the tests have utility. This type of definition makes utility synonymous with a subjective sense of satisfaction, which (although it may be closer to the original Benthamian view) we believe is too broad. Say, for instance, that someone would like to learn his earwax type on the basis of a genomic test. The test, however, has little clinical validity or predictive value. That means that the test cannot reliably tell apart those with wet and dry earwax types. We would conclude that although the test may be fun or interesting, it does not have personal utility. Incidentally, as there are few actionable options or significant self-understandings associated with learning one’s earwax types, it is doubtful whether a test for this phenotypic trait can have any utility at all, even if it were clinically valid.
Likewise, when someone pursues direct-to-consumer personal genome testing in order to find out whether or not she will develop atrial fibrillation, which runs in the family, she may be happy to learn that in comparison to the general population, the test provider estimates her risk to be slightly decreased. She may (mis)interpret this risk information as ‘I will not develop atrial fibrillation.’ Although she may find satisfaction in ‘knowing’ about this decreased risk, given the limited clinical validity of the test, she is actually experiencing false reassurance rather than (personal) utility.
This need not imply automatically that tests that lack personal utility should not be offered; there can be other reasons to offer or allow tests, such as the fact that people like taking them. However, we want to point out that if people have a specific reason why they value an AD biomarker test—if they expect it to have a specific usefulness for them, for example, because they wish to plan ahead—then the test must provide risk information with a sufficient degree of certainty, so that it is indeed reasonable to base one’s future plans on this information.
A classic example of personal utility is clinical genetic testing for Huntington’s disease, which, in contrast to many other conditions, is fully penetrant: if a mutation is found in the huntingtin gene, the individual has a 100% chance that he or she will develop the disease. Although the clinical validity of presymptomatic testing for Huntington’s disease is thus very high, as the disease cannot be cured or prevented, testing has little clinical utility and will not improve the individual’s health. At-risk individuals indicate that they opt for presymptomatic testing because of its relevance to the planning of one’s private life, including reproductive decision-making and the planning of one’s professional life, and because of psychological benefits, including the elimination of uncertainty and the hope to lack the mutation.28 These benefits have very little to do with clinical management of the heritable condition, but are of personal value to at-risk individuals. At the same time, it should be noted that many at-risk individuals decide not to undergo presymptomatic testing for reasons that are not always clear. Personal utility is highly personal in nature.
Thus, in parallel to the criterion of clinical utility, the notion of personal utility can be used to distinguish between useful and useless biomarker tests—useful or useless on a personal level, or for personal purposes. The notion of personal utility, however, is only useful itself as a criterion when it is stringently defined. We argue that personal utility requires both clinical validity and personal actionability or improved self-understanding.
Is there personal utility in AD biomarkers?
Information about whether or not one will develop Alzheimer’s dementia—and when this will happen—could, in theory, have tremendous personal utility. From qualitative studies it seems that people would use this information for all kinds of purposes: respondents cite a wide range of perceived benefits of APOE testing: improving one’s lifestyle, arranging one’s financial affairs, advance care planning, preparing oneself or one’s family members for the possibility of AD, participating in clinical research for altruistic or non-altruistic reasons, or even making plans for euthanasia.29 People presume that when they know they will likely develop AD dementia, they will start making different decisions or plans, such as moving plans closer into the future. By offering such valuable options, some researchers feel, AD biomarker information will thus serve people’s autonomy,27 and offering this information is an expression of respect for their autonomy. However, autonomy, we contend, is not served by offering healthy volunteers something they falsely believe to be accurate risk predictions.
Let us consider the case of a 67-year-old woman taking part in research and undergoing CSF biomarker testing for AD because she wants to know whether she will develop AD dementia. In the consent discussion with the physician-researcher, she indicates that she wants to learn the test result. When the results come in, the researcher informs her— carefully—that a reduced level of amyloid-β has been found in the sample, and that this means that she is possibly at increased risk of developing AD dementia in the future. While this may not be untrue, we have seen that at present, CSF biomarkers are of very little clinical validity in cognitively unimpaired individuals. Still, the research participant feels that she is now ‘at risk’ of developing AD dementia. This changes her perspective on life and on herself, and as time goes on, it affects her social relationships. She begins to use this ‘information’ to make long-term care and end-of-life arrangements. But can we truly say this ‘information’ has personal utility for her?
We argue that in the absence of a sufficient level of clinical validity, there can be no personal utility in a biomarker test for AD. When people misunderstand or overinterpret the clinical significance of an AD biomarker, they can believe something that is not the case and make decisions based on disinformation. The 67-year-old woman may feel that biomarker testing has been an important and helpful experience. She may even feel satisfied or relieved that now she finally knows her risk, she can act on it. We contend that there can thus be a discrepancy between perceived utility and actual utility. While it can be argued that arranging long-term care insurance is always a sensible thing to do and that research participation, in this case, has provided the impetus to do so, thus effecting positive change in the woman’s life, this is both circumstantial to research participation and based on highly uncertain ‘information’. It does not mean that the biomarker has personal utility.
We have said that in order for an AD biomarker test to have personal utility, there should be a sufficient level of clinical validity. But what is ‘sufficient’? Just like other medical tests, biomarker tests will not be 100% sensitive and specific, and their positive and negative predictive values will not be perfect. When is an AD biomarker informative? This is a normative question to which there is no single answer. Future research should focus on finding out how much certainty people require to make meaningful choices, and to what extent this depends on personality characteristics of people, the purposes for which they will use the biomarker, and the context they are in. Say, for example, that someone learns through a research PET scan that he has an elevated level of amyloid in his brain, and in response, he considers quitting his unsatisfactory job. This decision will have less of an impact when he has savings in the bank and a partner who can pay the mortgage, than when he does not. And when he does not, his preferred threshold of clinical validity will likely be higher. The more far reaching a decision, the more certainty one usually prefers regarding the information on which it is based. End-of-life decisions clearly qualify as far reaching, but—arguably—so does mentally preparing oneself or one’s partner or family members for a future with dementia. When research participants take personal actions, they should do so based on adequate, valid and reliable information.
Other reasons to disclose AD biomarkers
There may be reasons other than personal utility to disclose AD biomarker information to cognitively unimpaired research participants. First, the ethical requirement of informed consent entails that research participants should know and understand all relevant aspects of a proposed research study, and are free from coercion, undue influence or deception. The authoritative Council for International Organizations of Medical Sciences states that recruited research participants should inter alia be informed about the reasons ‘for considering the individual suitable for the research.’5 This means, for instance, that when research participants in population-based longitudinal cohort studies are invited to take part in an AD prevention trial based on specific AD biomarker information indicating an increased risk of AD, they should learn this information at the time of enrolment. In principle, enrolment in clinical trials based on existing individual research data should be transparent. There are exceptions: some studies use blinded enrolment designs for scientific reasons,30 which is acceptable as long as the enrolment design is understood and agreed on by research participants beforehand as part of the informed consent process.
Second, research participants are generally acknowledged to have a ‘right of access to their clinically relevant data obtained during a study on demand.’5 On their explicit request, research data should be made available to them.13 The General Data Protection Regulation aims at increasing individuals’ control of personal data, including research data, and is expected to place higher demands on biomedical researchers in European countries with regard to data access. The Regulation states, for instance, that research participants should ‘receive their information in an appropriate form without hindrance’ to improve data portability.31 In practice, however, it will not always be easy for research groups to provide research participants with individual research data, as it may require effort to present data ‘in an appropriate form’. While research groups may discourage research participants from requesting or insisting on access to individual research data, they may not be allowed to refuse.
There may be other reasons to disclose or not to disclose AD biomarker information to research participants, including reasons related to the scientific validity of the study. It is beyond the scope of this paper to specify under what conditions disclosure should or should not take place in particular (types of) research studies.
Conclusion
Many research participants are motivated to take part in AD research because they want to know whether they will develop dementia.10 Yet, in the absence of clear clinical validity, AD biomarkers in asymptomatic individuals have no personal utility, and routine disclosure of AD biomarkers to such individuals in research settings cannot be justified with an appeal to personal utility. However, there are two situations in which active disclosure is—in principle—ethically required: (1) when research volunteers are recruited for prevention clinical trials based on known biomarkers and (2) when they explicitly request to receive the results of an AD biomarker test. In these situations, research participants should be provided with their AD biomarker test results. The reasons for offering this information are the ethical requirement of informed consent, and a fundamental right to access to and/or control of personal data, respectively, not personal utility. At the same time, if research participants do not want to have biomarker information disclosed to them, for whatever reason, it should not be forced on them.
Researchers should anticipate and set up adequate procedures for the disclosure of AD biomarkers on explicit request of research participants. Disclosure places high demands on information provision, education and communication. Researchers should ensure, as part of the informed consent process, that participants are informed about the uncertain clinical significance of AD biomarkers and the risks and implications of receiving this information.6 Most importantly, research participants will need to understand that taking part in research will not provide an answer to the question whether they will develop dementia. When participants misunderstand or overinterpret AD biomarker information and assign it too much clinical validity, they may use this—highly uncertain—information for personal purposes, possibly to their detriment. When, on the other hand, they come to understand that uncertain ‘information’ neither serves autonomy nor has personal utility, they might even choose not to receive the biomarker results. In fact, when participants learn about the uncertainties associated with most AD biomarkers, their interest in learning their individual research results decreases substantially.25 Further study should elucidate the effects of AD biomarker disclosure on cognitively unimpaired research participants and the extent to which AD biomarker information is (perceived to be) personally useful for these participants.
A stringent definition of utility implies a sufficient level of clinical validity. In keeping with this definition, CSF and PET imaging biomarkers for AD are currently of too little clinical validity. Over the next couple of years, researchers will be confronted with participants seeking access to their individual AD biomarker test results. Although there can be valid reasons to offer disclosure of these results, it will not be personal utility.
Acknowledgments
The authors thank the ELSI work package of EPAD for the extensive discussions that have contributed to the development of this manuscript.
References
Footnotes
Contributors EMB and MHNS had originally developed the ideas underlying this paper. RM and ER contributed to the further development of these ideas. EMB drafted the manuscript. MHNS, RM and ER critically revised previous versions of the manuscript.
Funding This work was funded through the Ethical Legal and Social Implications work package of the European Prevention of Alzheimer’s Dementia (EPAD) study and the project ‘Early diagnosis of Alzheimer’s disease: conceptual considerations and ethical guidance’ which was funded through the Netherlands Organisation for Health Research and Development (ZonMw, project number 731010012). EPAD receives support from the Innovative Medicines Initiative Joint Undertaking under grant agreement number 115736, resources of which are composed of financial contribution from the European Union’s Seventh Framework Programme (FP7/2007-2013) and EFPIA companies' in-kind contribution.
Competing interests None declared.
Patient consent Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Other content recommended for you
- Exploring cognitive and biological correlates of sleep quality and their potential links with Alzheimer’s disease (ALFASleep project): protocol for an observational study
- Personal utility in genomic testing: is there such a thing?
- Retinal imaging in Alzheimer’s disease
- What is the clinical impact of cerebrospinal fluid biomarkers on final diagnosis and management in patients with mild cognitive impairment in clinical practice? Results from a nation-wide prospective survey in France
- New diagnostic criteria for Alzheimer's disease and mild cognitive impairment for the practical neurologist
- Moral motivation regarding dementia risk testing among affected persons in Germany and Israel
- Plasma biomarkers for Alzheimer’s disease: a field-test in a memory clinic
- Remote data collection speech analysis and prediction of the identification of Alzheimer’s disease biomarkers in people at risk for Alzheimer’s disease dementia: the Speech on the Phone Assessment (SPeAk) prospective observational study protocol
- Two-year outcome of MCI subtypes and aetiologies in the Göteborg MCI study
- Nutrient intake and brain biomarkers of Alzheimer's disease in at-risk cognitively normal individuals: a cross-sectional neuroimaging pilot study