Ethical Considerations in Communicating Alzheimer’s Disease Neuroimaging Biomarker Test Results to Symptomatic Individuals

Abstract

This article examines ethical issues associated with the return of AD neuroimaging results to cognitively symptomatic individuals. Following a review of research on patient and study partner reactions to learning the results of biomarker testing for AD, we examine ethical issues that will be of increasing significance as the field transitions to an era wherein disease-modifying treatments for AD become available. We first review the ethical justification for returning AD biomarker results to individuals who desire them. We then address a more novel question: whether, and to what extent, clinicians or clinical researchers should influence the decisions of individuals who are potentially reluctant to learn their AD imaging results. We argue that in many cases, it is ethically correct to explore, and sometimes alter, factors that may be inhibiting one’s desire to know these test results. Our argument is grounded in the premise that having more complete information about changes that may be happening in one’s brain will generally yield more informed participation in decisions about one’s own care, thereby promoting autonomy. Finally, on the assumption that we have established that it is frequently ethically correct to try to communicate testing information, we examine considerations regarding (not whether but) how this is best accomplished, discussing the concept of responsible transparency. We suggest that both (1) explorations of why one may or may not want to learn results of AD biomarker imaging and (2) the responsible return of such test results is best accomplished using a transactional model of communication.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13311-021-01047-0.

Introduction

Advances in neuroscience have yielded an acceleration in the field’s ability to detect, in vivo, biological markers of Alzheimer’s disease (AD) in the brain, cerebrospinal fluid (CSF), and blood. Of these techniques, neuroimaging-based markers offer the most direct evidence of AD pathology in the brain. Positron emission tomography (PET)-based testing includes the use of fluorodeoxyglucose (FDG) PET to characterize patterns of cerebral glucose metabolism and of radioligands that, depending on the agent, bind to either beta amyloid plaque or tau protein deposits. This enables the visualization and quantification of AD-associated pathologic changes in the brain. In addition to generating laboratory values like tests of CSF and peripheral blood, PET ligand testing creates an image of the brain that can potentially be shown to a research participant or a patient in clinical practice.

The possibility of offering to an individual direct, visual evidence of AD biomarkers in his or her brain has led to important questions about whether, when, and how to safely and effectively disclose these sensitive and potentially life-changing test results to affected individuals. Key assumptions within this body of discourse are that manifestations of AD are both devastating and, once underway, unstoppable. Recognition of AD as treatable yet fundamentally progressive and incurable lies at the core of concerns about psychological harm, social stigma, or discrimination as potential adverse consequences of learning one’s AD biomarker status.

The purpose of this article is to review the ethical issues associated with the return of AD neuroimaging results taking into consideration new developments in the quest for a course-altering treatment for individuals with symptoms of dementia who are biomarker positive for AD. Following a brief overview of the ethical considerations raised by the prospect of biomarker testing for AD, we will discuss two issues that we assert will be of increasing significance as the field of dementia care and research transitions into an era wherein a disease-modifying treatment for AD becomes clinically available.

First, we will review the ethics of disclosing test results to individuals experiencing cognitive decline who would like to know these results. This topic is addressed elsewhere [1, 2], and so we will move over it fairly quickly. Second, we will address a more novel ethical issue—to what extent clinicians or clinical researchers should try to influence potentially reluctant persons with cognitive symptoms to know their test results. Our central argument will be that, with the exception of persons with advanced dementia, uncontrolled mood disorders, or serious mental illness, it is critical to examine, and sometimes alter, factors that may be inhibiting a cognitively symptomatic individual’s desire to know biomarker test results. Third, on the assumption that we have established that it is frequently ethically correct to try to communicate testing information, we will examine ethical issues regarding (not whether but) how this is best accomplished.

For purposes of this paper, we will be mainly considering cognitively symptomatic persons who are participating in research studies where biomarker results indicating the presence or absence of AD pathology may, at least in part, explain their cognitive symptoms. That said, much of the same logic would apply to clinical contexts, and perhaps some, though not all of the considerations raised in this paper, would extend to asymptomatic individuals who have undergone testing in research settings. In addition, our discussion relies heavily on studies of psychological reactions to the return of amyloid PET results. While there is no single and/or unequivocal test for AD or AD dementia, amyloid PET is the context in which virtually all of the studies of psychological and behavioral reactions to non-genetic AD biomarker testing have occurred. Our review of these articles should not be mistaken for an endorsement of the amyloid hypothesis but rather a reflection of our position that findings from these studies have implications for the discussions undertaken in this paper. Indeed, it is reasonable to suspect that our main arguments concerning communication and responsible transparency will have applicability to other AD biomarker tests, including existing and well-established brain imaging approaches like FDG PET and multi-marker assessments which may represent the future of diagnostic and prognostic testing in AD and related dementias.

Empirical Background

Over the past decade, several research teams have published reports describing their experiences with returning AD biomarker test results to study participants. These reports have focused on communicating amyloid positron emission tomography scan results to three groups of individuals: (1) cognitively healthy, or asymptomatic, persons; (2) symptomatic persons who are experiencing cognitive decline and carry a diagnosis of subjective cognitive impairment, mild cognitive impairment, or dementia; and (3) family members or caregivers of cognitively symptomatic individuals. While critiques of these studies have described their varying levels of methodological rigor (some were retrospective, most lack control conditions) [3, 4], these studies represent important early work in understanding how study participants will respond to the opportunity to learn the results of AD biomarker tests conducted in research settings. A shared goal of these investigations is to test theoretically based claims that learning of one’s AD biomarker positivity may be psychologically harmful. As described above, such claims are grounded in the understanding of AD as a progressive and incurable disease with devastating implications for the quality of life among those affected. Investigations of the psychological impact of AD biomarker testing are typically accomplished by administering standardized measures of depression, anxiety, and/or test-related distress. Comparisons are conducted either (a) within group, by comparing psychological assessment scores before and after biomarker result disclosure, or (b) between groups, by comparing psychological assessment scores of those who receive positive results to those who receive negative results. While most of these studies have used standardized assessments of mood, a handful of relevant investigations have been qualitative in nature, meaning that open-ended interview questions are asked of participants, allowing them greater latitude to describe their personal experiences than would be possible with a questionnaire-based assessment.

Asymptomatic Persons

Regardless of the methodologic approach, the bulk of this research has focused on the first of the aforementioned three groups, healthy older adults who are not experiencing any changes in cognition. According to a systematic review of amyloid PET disclosure studies that were published through 2018, reports from this growing body of research have consistently concluded that disclosing amyloid PET scan results to cognitively healthy older adults has no discernible psychological impact on such individuals [3]. None of the studies in the 2018 review included control groups and several were quite small, with only one out of five such reports describing a sample of more than 50 participants. Since that time, findings from another observational study of amyloid PET disclosure in asymptomatic persons, with a considerably larger sample size, have been published [5]. Consistent with earlier reports, this larger investigation showed that participants’ scores on measures of mood did not significantly differ before and after amyloid PET result disclosure, thereby offering additional evidence for the psychological safety of returning AD biomarker test results to asymptomatic persons.

This set of studies offers preliminary assurance that it may be uncommon for cognitively healthy persons to have adverse psychological reactions to learning that they have AD pathologic changes in their brains. However, the context of the amyloid PET disclosures in this group of studies must be considered. There are currently no circumstances under which a cognitive healthy individual would have a clinical indication for an amyloid PET scan [6]. In each of the published reports of reactions to learning one’s brain amyloid status, cognitively unimpaired persons learned the results of testing that was conducted as part of a research protocol, not for a clinical indication.

Cognitively healthy individuals typically participate in research studies on AD and cognitive aging for altruistic reasons and/or to access the potential benefits of AD prevention research [7]. While such individuals may desire the results of their individual research brain scan results, and in some cases be intensely curious about those results, the receipt of this information is balanced with the highly positive feedback that these healthy control participants have received based on their performance on comprehensive neuropsychological research test batteries. In effect, these individuals enjoy the position having been deemed cognitively “normal” by expert clinical researchers. It also bears emphasis that these individuals explicitly opted in to the receipt of their research test results and were carefully pre-screened for anxiety or depression. In each of the above-described studies, persons with elevated scores on measures of anxiety or depression were excluded from participation. On the whole, these caveats suggest that caution is warranted in assuming that the reactions of these study participants reflect how a typical older adult from the general population would respond to the same information.

Symptomatic Individuals

Four published reports have described the psychological reactions of persons with mild cognitive impairment or a dementia syndrome to the receipt of amyloid PET scan results. Given that these individuals perceive their cognition to be declining, it is reasonable to speculate that learning of AD biomarker positivity, whether through a research test or clinical scan, may be distinctively impactful for this group of persons. Indeed, such individuals often enroll in clinical research to better understand the etiology of their perceived cognitive decline. Thus, receiving the results of one’s AD biomarker tests may have more immediate implications and lead to different psychological reactions than those observed in asymptomatic persons. The first two reports of psychological reactions to amyloid PET result disclosure in symptomatic persons were qualitative studies [8, 9]. Both reports described a mix of positive and negative emotional responses to amyloid PET result disclosure among participants. Both included instances of amyloid-negative individuals reporting feelings of relief that were tempered by lingering worry over what, if not AD, was causing their cognitive impairment. Fear, sadness, and worry were present in many but not all individuals who received positive scan results. And, in some cases, amyloid-positive individuals described feeling happy to have their suspicions confirmed and/or more information about the changes occurring within their brains.

The third and largest observational study of amyloid PET disclosure in symptomatic persons found that participants’ scores on measures of mood did not significantly differ before and after amyloid PET result disclosure [10]. Similarly, a recent randomized controlled trial found no differences, over time, in depression or anxiety scores between (a) those who did and did not learn their brain amyloid status and (b) those who learned of being amyloid positive as compared to those who received negative scan results [11]. However, subgroup analyses of scan group participants in this trial identified significant differences on a measure of test-related distress, or emotional upset, a psychological construct that is described as a more proximal or direct reflection of the perceived emotional impact of a given event (in this case, amyloid PET result disclosure). Persons receiving amyloid-positive results scored higher on this measure relative to their amyloid-negative counterparts, with scores averaging at or near the clinical cut-point for moderate distress at multiple follow-up points. These findings, while limited to a single study, raise the possibility that, in the context of cognitive symptoms, learning of one’s AD biomarker status may not be an entirely benign event.

Family Members of Symptomatic Persons

There are two published reports of family members’ psychological reactions to learning a spouse or relatives’ amyloid PET scan results. The first was a qualitative study of telephone interviews with family caregivers of patients who had undergone a clinical amyloid PET scan within, on average, the past 8 months [9]. Sixteen out of 17 caregivers of individuals who were scanned in that study learned that their spouse or relative had positive scan results. Similar to patient reports from the same study, several caregivers (n = 6) expressed relief, but half described feelings of sadness and despair. The second published report of caregiver reactions to receiving the amyloid PET scan results of a spouse or relative was the above-mentioned randomized controlled trial [11]. Like the patient participants, caregivers of amyloid-positive individuals reported significant and sustained emotional distress. In addition, caregivers of participants in the scan group were found to endorse less confidence in their ability to cope with their loved ones’ syndromes at follow-up, a finding that subgroup analyses revealed to be driven by those learning of positive scan results.

Taken together, the empirical research on AD neuroimaging biomarker result disclosure suggests that the potential for emotional distress post-disclosure is greatest in individuals who are experiencing cognitive changes and learn of a positive test result. However, it is critical to note that the studies informing this conclusion involved highly select populations with limited racial, ethnic, and socioeconomic diversity, and actively excluded persons exhibiting signs of depression or anxiety. Given that the potential for distress that follows a positive biomarker test may be heightened in persons experiencing concurrent mood symptoms, which are highly prevalent in the general population of older adults [12, 13], this paper treats psychological harm as a risk to be considered in ethical analysis of AD biomarker result communication.

This overall conclusion that the risk of psychological harm upon receipt of AD biomarker test results may be greater for those who are experiencing cognitive symptoms than for those who are not warrants a fresh analysis of the ethical implications of such result disclosure for two main reasons. First, initial concerns regarding the ethical challenges and psychological sequelae of AD biomarker testing focused on asymptomatic individuals and did not address the unique considerations of studies involving cognitively impaired persons [2, 14, 15]. Second, the more general discourse on the return of individual research results to participants has been dominated by concerns regarding incidental findings [16]. A finding of biomarker positivity in an individual who is experiencing cognitive changes and participating in research on AD cannot be treated as incidental.

Disclosure of Test Results to Actively Interested Symptomatic Persons

In this section, we briefly review the ethical dilemma of whether to share positive biomarker test results to symptomatic individuals who desire to know their results. We first review the operative fundamental bioethical principles, then summarize the argument from Lingler and Klunk [3] that it is ethically appropriate to share results in these circumstances.

Relevant Bioethical Principles

Beginning with the Belmont Report [17], it has been generally accepted that bioethics should be primarily concerned with supporting autonomy, making healthcare decisions in accordance with justice, and maximizing net benefits to patients while avoiding harm. This last principle was originally combined under the heading of “beneficence”, but was later—see Beauchamp & Childress [18]—divided into the positive requirement of “beneficence” to do good and the negative imperative of “nonmaleficence” to do no harm. While issues of justice no doubt touch on systemic factors such as who can participate in research studies and how they are done, since for present purposes, we are restricting our analysis to cases where the patient is already enrolled in a study and the test results are already known by the researcher, then (at least for the most part) questions of justice more generally fall by the wayside. This leaves the ethicist in the (somewhat familiar) position of having to balance autonomy, beneficence, and nonmaleficence.

On the surface, autonomy—the principle that patients should be allowed and empowered to make their own healthcare choices—speaks strongly in favor of sharing results whenever possible. While there are some arguments that a more paternalistic approach can foster autonomy [19, 20], for the most part, these focus on the process of decision-making rather than questions of information sharing. Part of freedom to decide is having the relevant information (Beauchamp and Childress [18] list this as a “threshold requirement”), which would speak in favor of sharing test results with research participants whenever feasible. While the general argument might appear simplistic, it carries great weight in the case of individuals who want to know their results. Perhaps more surprisingly, we will also argue that in some circumstances, respecting people’s autonomy actually speaks in favor of, to some limited extent, resisting their inclination not to learn potentially disturbing information.

Informing Cognitively Symptomatic Individuals Who Want to Know

Lingler and Klunk [1] make a compelling case that we should communicate amyloid PET scan results to individuals with mild cognitive impairment (MCI) who want to know them. Their argument is based on the scientific validity of the results, the respect for persons incorporated in sharing the results, and the possible benefits of knowing the results. It is important to note that to the extent that their argument was compelling in 2013, it may be even stronger now given subsequent advances in the field. While the issue of respect for persons might be unchanged, several AD biomarker tests (particularly in combination) are more informative now and there is more positive use to which the results can be put. We expect the possible benefits of knowing to continue to increase, particularly as disease-modifying treatments become available [21].

The case for respect for persons is similar (though not quite identical) to the basic argument from autonomy discussed above. People can make more genuinely free decisions when they have all the information available. This is especially true when—as in the present case—individuals desire to know medical information. Absent an extremely compelling reason not to provide it, we should do so. This is the case even when the information is potentially disheartening or even emotionally upsetting. While 50 years ago, it might have been deemed appropriate to withhold disturbing information about risk factors for serious, life-threatening illnesses such as cancer [22], in which position has long been discredited [23]. While there are some disanalogies between that case and the biomarkers under discussion, the relevant ethical considerations are mostly symmetrical and are warranted in light of reports that physicians are reluctant to disclose AD diagnoses [24]. The main difference might include the connection between potential cognitive impairments and a loss of a sense of agency that one would not find in the case of risk factors for cancer. The relevance of such factors is beyond the scope of this paper, though it is an issue to which we would like to return in future research.

Lingler and Klunk [1] also argue that issues that may have spoken against sharing amyloid PET results in the past had (already as of 2013) become moot points. Where there had been concern about the scientific validity of amyloid PET, these have progressively become less relevant as the science has advanced and amyloid and tau PET radiotracers have received Food and Drug Administration approval. Moreover, even in the absence of therapeutics to prevent or cure AD, Lingler and Klunk [1] argue that there are still actions that cognitively symptomatic persons with positive test results can do to improve the quality of their future lives. The prospect of a disease-modifying anti-amyloid agent gaining Food and Drug Administration approval in the near future only bolsters this argument [25]. Thus, in addition to respect for persons/autonomy, the principle of beneficence would require that we share validated test results with symptomatic research participants who, after being counseled regarding the pros and cons of learning their AD biomarker status, would like to receive them.

One might be concerned that countering the considerations from autonomy and beneficence is a worry that receiving disconcerting test results could harm patients in a way that would violate nonmaleficence. While this concern is tempered by the empirical evidence presented above, we recognize key limitations of the available evidence and the need to proceed with caution as we work to further our understanding of who is and who is not at risk for emotional upset after learning of their AD biomarker positivity status. We will return to this issue later on, where we will argue that even if individuals are reluctant to receive their results, it might still sometimes be worthwhile to create a context where they are open to them. If we can successfully argue against concerns of harming the research participant or patient in cases where they do not want their results, then a fortiori there should not be significant concerns in cases where they do.

Communicating Test Results to Hesitant Patients Pt. I

Suppose that one grants that researchers should share test results with participants who want to receive them, it would be natural to think that precisely the same considerations clearly speak against making any effort to communicate test results to participants who are hesitant to hear them. Respecting autonomy would on the surface suggest that we should not burden individuals with information they are hesitant to hear. Indeed, we grant that in cases where the participant is adamant that they do not want to receive the information it would be a clear breach of autonomy to force it onto them. However, we argue that if the participant is just somewhat reticent but not wholly decided against hearing the test results, then it might make sense in some circumstances to try to create a context where they would accept the results after all. This is especially true in instances where the participant presents in the company of a study partner (typically a relative or close friend) and the source of the reluctance is unclear. While the distinction between those who are somewhat reticent and those are adamantly opposed may not be readily apparent, in most cases, a skilled clinical researcher will be able to differentiate the two. Both the strength and the source of the reticence may be explored through a series of follow-up questions and the use of standard techniques such as active listening and validation as the individual and their study partner are invited to expand upon a stated disinclination to learning their results.

In the last 10 years, bioethicists have come to question whether respect for autonomy requires immediately acceding to all of a patient’s immediately stated preferences. Levy [19, 20] argues that—properly construed—respect for autonomy requires not just a provider doing what a patient says they want but rather a provider respecting patients’ fundamental values and trying to align those with their (the providers’) actions. To take a simple example, if a patient is addicted to a narcotic, trying to quit, but suffering from a craving, then it would probably be more respectful of their autonomy to withhold the narcotic than to provide it. This is obviously an extreme case, but it suggests the logical possibility of respecting patients’ autonomy in ways that involve departing from their immediately stated preferences.

Moving from providing a narcotic to providing information, the ethical question becomes the role of the researcher or clinician in communicating with the individual. Historically, the picture of the communicative process was that the provider (or, in this case, researcher) was simply responsible for making information accessible, with the question of what to do with that information left to the discretion of the patient (or, in this case, research participant). If this is the correct view of the interaction, then it is difficult to see how it would ever be justified to directly influence what information an individual wanted to receive. However, we will argue (following Cargill [26]) that this is not the best way to view provider/patient communication and that a more sophisticated model might suggest sometimes influencing an individual to accept information to which they might not be immediately amenable.

Interlude: the Nature of Patient/Provider Communication

In this section, we evaluate approaches to conceptualizing the nature of provider/patient communication. We follow Cargill [26] in asserting that the more traditional “transmission model” is insufficient.

The Transmission Model

Since Claude Shannon’s landmark 1948 “A Mathematical Theory of Communication” [27], the dominant model of how communication occurs has been (as the name suggests) a mathematical treatment of how information is conveyed by a signal measured against how much uncertainty or “noise” there is. The central idea is that we can measure what is communicated in terms of how much it reduces uncertainty about the world within the recipient.

There are many advantages to Shannon’s picture of communication. Perhaps the most valuable is that it is exceptionally amenable to quantification—Shannon and his followers provided precise tools for measuring how much information is conveyed in a variety of contexts. Second, it clearly articulates ideas such as “bandwidth” (how much information can be conveyed by a particular channel), “entropy” (how much uncertainty there is over the content of a message), and—particularly important for our purposes—“noise” (how much interference obscures the intended content of the message).

However—as is often the case with mathematically modeling a complex behavioral phenomenon like communication—Shannon’s vision involves abstracting away from elements of the communicative context that one might think are relevant—particularly in the healthcare context. That is, when we treat communication as a mathematical construct only, we omit a host of details regarding the particular communicators and communicative acts. Aside from their level of preexisting uncertainty regarding the content of the information, very little is said about the recipient. Specifically, nothing is said about their interest in the message, possible barriers to receiving the message at their end (as opposed to in how the message was sent), and any feedback loop between the listener and the speaker.

If one thinks of communication on the model of transmission, then there is little room for the possibility that researchers or providers should be actively involved in working with the participant or patient to figure out how best to share information. Thus—on this model—if the individual is reluctant to hear test results, then the researcher’s role is complete. However, we will see that this is not the only—and likely not the best—way to conceive of communication in a healthcare context.

The Transaction Model

Cargill [26] argues that it is a mistake to think of communication in the healthcare context as working on a transmission model. One strong piece of evidence for this is that the persistent pattern of thinking of providers as primarily/only interested in transmitting information has led to massive lacunae in what patients actually understand. Cargill makes this argument in the context of the literature on informed consent in research, where it is very well documented that despite innumerable attempts to shape how information is conveyed (i.e., on the transmission model), patients fail to understand even the most basic aspects of the studies in which they participate [28]. For example, there is the classic “therapeutic misconception”, where many research participants think that the treatment being tested is intended and/or likely to help them, when in fact the goal of a research study is to see what works for the benefit of future patients [29].

Cargill [26] identifies the problem in the failure of the transmission model to account for various factors that could interfere with communication other than the transmission model’s formally defined notion of “noise”. They trace this insight back to an “interaction model” developed by Schramm in 1954 [30] and modify it into a “transaction model” for which they advocate.

There are several relevant differences between the transmission and the transaction model. The transaction model views communication as continuous rather than as a set of discrete packets of information. It sees the process of communication as bidirectional, with questions and involvement from the audience affecting how the interaction unfolds. Most importantly for present purposes, the transaction model considers the ways in which aspects of the patient’s personal, psychological, and social context can inhibit their receipt of information. Just as communication can fail because the message is noisy, it can also fail because there is some barrier(s) preventing the people involved from optimally communicating.

For example, if one asks prospective research participants in the context of informed consent whether they understand an intervention, there can be a variety of reasons they say “yes” independently of whether or not they actually do. They could be so overwhelmed that they do not know what to ask, be too stressed not to want to continue the conversation, or be subject to a social norm that it is irresponsible to waste a provider’s time. If there might be contextual, psychological, and social factors that interfere with a patient or research participant’s ability to be an effective communicative partner, then part of communicating might involve not just sending information but working with the individual to make sure they are in the proper place to receive it. This might be the case even if the barrier is the person’s own hesitancy to receive the information, particularly if that hesitancy can be attributed to factors that do not accord with their underlying values. This opens the door to the possibility that good communication might sometimes make it incumbent upon researchers and providers to work with participants and patients to get them to a place where they are open to receiving information that they might otherwise be reluctant to receive. Symmetrically, one might wonder whether there are ways in which it is the patient’s job to help the provider communicate. However, as this paper’s intended audience is providers, we put that issue to the side.

Communicating Test Results to Hesitant Patients Pt. II

The Importance of the Transaction Model

In this section, we will argue that it is possible that there are circumstances where cognitively symptomatic individuals might be hesitant to receive test results, but where it is ethically preferable for researchers or providers to work with them to help them get to a place where they are more receptive. We intentionally frame the issue in terms of what is “ethically preferable”—invoking a scalar notion of proper action—rather than addressing the separate question of what is or is not required (that is, we reserve judgment on whether the ethical improvement is or is not obligation-generating). When and whether it would be ethically preferable to work with hesitant individuals will depend on a variety of factors, including what is preventing one from wanting to hear the results (which ties into autonomy), what benefits there would be in the individual hearing the results (which ties into beneficence), and what potential harm there would be in the individual receiving the results (which ties into nonmaleficence). Note that answering all of these questions will require a mixture of theoretical ethical reasoning and empirical inquiry into (among other things) the consequences of getting or not getting information about biomarker test results. We will argue that if the transaction model is the best model for healthy communication, then it is not always enough to blindly accept that participants or patients do not want to know something, but rather explore whether there is some problematic and/or surmountable barrier preventing them from wanting to know. Finally, for the purposes of this discussion, we will assume that individuals with mild cognitive impairment or early stage dementia have the capacity to make an informed decision about whether to receive their results.

The idea is that—on the transaction model of communication—social and psychological barriers to receiving information on the part of the recipient are treated very much like noise originating from the sender. Part of ensuring maximal communication is addressing such factors if they seem to be a barrier to an individual’s values and autonomy. This will require asking a question that might sometimes go overlooked in discussions of disclosures—not just what the information recipient says they want to hear, but also why they say what they do.

This raises the crucial question of what a researcher can actually do to improve communication. The first step would of course be to remove any and all roadblocks that impede successful interaction. Sankar systematically discusses ways in which researchers can confuse the situation, such as by implying that a phase 1 trial is intended to assess benefit [31]; it goes without saying that such (intentionally or otherwise) misleading presentations must be eliminated, but the present work goes farther than simply requiring an end to (hopefully) unintentional duplicity. The more important but subtle point Sankar makes is the ability of the researcher to influence the environment through presentation and framing. It is well established in behavioral economics (most famously in Tverskey and Kahneman [32]) that how we present a decision can affect how people think about a problem. More recently, Richard Thaler and Cass Sunstein [33] have demonstrated how simply affecting choice presentation and architecture can lead to improved outcomes. Another important aspect of the disclosure environment concerns who is doing the disclosing. Given what is known about cultural variations in the nature and pervasiveness of AD stigmatization [34, 35], there may be value in ensuring that the discloser is either of the same racial/ethnic background as the participant or is trained to provide culturally congruent care. None of these “manipulations” involve any sort of deception, and all would have the potential to overcome psychological and social barriers that might be the result of not having facts put forward in an optimal way.

However, questions of framing only scratch the surface of how to set up a social and psychological context where a client is free to pursue their genuine preference. For example, one vital issue is making sure a client is being empowered to communicate even when they might have social or psychological difficulty doing so. The American Speech-Language-Hearing Association communication Bill of Rights [36] is written with an eye towards individuals with disabilities, but the central insights about ensuring an open ability to communicate can help alleviate social pressures to “go along” with certain plans.

We should note that there might appear to be a bit of a paradox in our view—on the one hand, we seem to be arguing that:

1. Researchers should rig a conversation to get a hesitant participant to agree to receive test results they might not want to receive.

2. We should do everything possible to empower individuals to make their true preferences known.

However, the paradox is only apparent—what we are arguing should be done is that researchers should do everything in their power to give results if and only if doing so is in accord with the participants’ ultimate values. As Levy [19] controversially but persuasively argues, individuals’ immediately stated preferences might not be their true values. Our goal should be to make sure that the communicative context makes room for a person’s real values to be communicated and manifested.

One might wonder how this fits into broader questions of medical paternalism. Such questions are well beyond the scope of this paper, but we would like to comment that the extent to which arranging a context where someone is more likely to want to hear results counts as paternalistic is more complicated than it may seem, and depends in large part on a variety of questions having to do with distinctions among paternalism. A careful examination of distinctions among paternalism and the justifications against it [37] reveal that most do not apply in this case. It is clearly not an issue of government action (forbidden by an injunction against “narrow paternalism” [37]), and it is at least intended not to be a case of trying to alter someone’s truly held final ends (as forbidden by an injunction against even “weak paternalism” [37]). Most remarkably, this sort of “interference” might even be allowed by the “soft paternalist” permissible to staunch libertarian John Stuart Mill:

To use Mill’s famous example of the person about to walk across a damaged Gerald Dworkin Fall 2020 Edition 5 bridge, if we could not communicate the danger (he speaks only Japanese) a soft paternalist would justify forcibly preventing him from crossing the bridge in order to determine whether he knows about its condition [37(pp. 5-6)].

This is supposed to be an example where almost everyone should agree that some degree of “paternalism” is acceptable, and it is surprisingly similar to the cases under consideration at present.

Finally, one may wonder whether the duty to warn might already require constructing a context where a patient is willing to receive their results, so as to avoid future harm to themselves or others. This might or might not be the case, but the main advantage of the present work is that it neatly sidesteps entirely the contentious issues involved in navigating between the duty to warn and questions of autonomy (e.g., Robin and Farmer [38] lay out some of these issues (pp94-5); Gallagher & Ewer [39]) relate it specifically to incidental research findings (p253))—we argue that even without a duty to warn, there are reasons to try to get patients to accept information regarding which they are initially hesitant. A fortiori if there is a duty to warn it just makes our case that much stronger.

Possible Sources of Hesitancy to Receive Biomarker Test Results

There are innumerable possible reasons a participant or patient may indicate a disinclination to receive biomarker test results, and research is needed to better understand these reasons, especially within diverse populations. Our contention is that each case must be dealt with individually—there is no general algorithm for researchers or providers regarding how to effect optimal communication. In an ideal world, would-be results disclosers would receive training in assessing individuals’ values in motivations, though we do not in this paper determine the precise way to go about training providers given more limited resources. However, in this section, we discuss two likely sources of hesitancy and why it might be worth a researcher or provider considering digging deeper rather than just accepting a participant’s or patient’s disinclinations at surface value.

Concerns About Futility of Information Acquisition

One potential source of hesitancy regarding receiving news pertaining to one’s likelihood of having or developing AD is a general sense that nothing can be done with the information anyway. Notably, such an impression may be held, not only by participants and their families, but by clinicians and researchers, who may make statements that directly or indirectly introduce or reinforce this message. Just as clinicians and researchers should stay abreast of new research into treatment developments and preventative strategies, there is a need for continuing education about moving beyond questions of the clinical utility of AD biomarker information to raise awareness of the need to help AD biomarker candidates consider the potential personal utility of such information.

Regardless of the source of the belief, in the case that participants and families express the belief that AD biomarker information holds no possible value to them, we argue that the hesitancy is at least potentially based on an objectively false belief, and so, there is some responsibility on the researcher or provider to try to help the participant or patient reassess that belief. By analogy, if a person is refusing a vaccine because they believe it causes autism, it seems uncontroversially ethically appropriate at least to try to convey that the underlying belief is simply false [40]. As Wu and McCormick [41] succinctly note, “when patients express false or misinformed beliefs, it is professionally and ethically appropriate to attempt to correct and redirect the patients so that they can hopefully use evidence-based information to make an informed decision about their care” (p1055). The belief is false because there are now things we can do to improve quality of life based on receiving positive biomarker results. As noted above, Biogen has recently completed a license application for a groundbreaking treatment for AD [25]. However, even in the absence of available treatments, there are advantages to possessing the information, as Roberts, Dunn, and Rabinovici [2] explain:

One could even make the case that improved AD treatments need not be available to warrant access to one’s amyloid imaging results in certain scenarios…Studies of asymptomatic individuals who have undergone AD risk disclosure, with disclosure of their APOE genotype status, suggest they have found it useful for a variety of reasons, including informing advance planning, encouraging monitoring of AD treatment and research, gaining psychological relief (in the case of negative test results) and satisfying a perceived need for risk information (p223).

A recent study of persons with MCI who were contemplating learning their amyloid PET status found that these same reasons motivated decisions in favor of the testing [42]. The point about the possible benefits of acquiring the information is actually twofold. First, it suggests that respect for autonomy implies that researchers and providers should try to overcome barriers to information acquisition based on false beliefs. Second, the possible benefits of communicating the information imply that beneficence also prima facie supports doing so.

Concerns About the Harms of Information Acquisition

A second possible source of hesitancy is the concern that getting test results would be actively harmful. Possible sources of this concern are that receiving the information would be psychologically traumatizing or that it would lead to invidious stigma and/or discrimination.

Beginning with the concern about the psychologically damaging nature of the information, research suggests that this concern is in some contexts overstated. In a study of patients with MCI who had no untreated mood disorders or serious mental illness, some showed emotional upset upon receiving this sort of information, in which upset did not generally rise to clinically concerning levels of depression or anxiety [11]. There is currently a lacuna in research regarding patients who have active mood disorders or serious mental illness, and so, there is at present no way to reliably gauge the harms in sharing that information with those individuals.

Even if one is not concerned that the information would be traumatizing, individuals might still be concerned that their lives will be diminished by the receipt of bad news. However, behavioral economics indicates that people are far more adaptive to such developments than they might think. For example, Gilbert [43] describes this as a failure of affective forecasting—our ability to predict our own mental states. He attributes the phenomenon to the fact that when we picture future possibilities we focus on the information at hand—in this case the fact that one will know one is at high-risk for developing AD—rather than the more accurate prognostication that in a year’s time we are more likely to have returned focus to the ins and outs of daily life.

On a different note, persons experiencing cognitive changes might be concerned that having or being at increased risk of developing AD might lead to stigma and/or discrimination. It is admittedly concerning that while there are legal protections against discrimination on the basis of genetic predispositions to diseases [44], there are no equivalent protections regarding biomarker test results. While the theoretical risk for stigma and/or discrimination has been raised by several commentators, studies have not yet characterized the extent of such phenomena. In medical settings, this information should (in the USA) be protected by HIPAA; however, the extent to which HIPAA protections extend to research data is less clear. Indeed, a review of existing law by Arias and Karlawish concluded that current legal and regulatory mechanisms are insufficient to protect research participants from discrimination by employers or insurers who gain access to their AD biomarker research tests [45]. Perhaps even more importantly, any such legal protections would be unlikely to offset the risk of experiencing the social stigma associated with AD and related disorders [46, 47]. This suggests that, ultimately, broader public health campaigns and community-based education may be needed to minimize the risk of stigmatization over the long term.

Note that—just as in the case of concern regarding possible benefits—the ethical lesson of the relative unlikelihood of harm is twofold. First, it suggests that respect for autonomy involves aiding communication by trying to explore patients’ potentially misleading concerns regarding possible harms. Second, it suggests that nonmaleficence—at least in the subset of cases we have delineated—does not require us to withhold information.

Responsible Transparency

In this final section, we will outline the rationale for and features of the concept of, responsible transparency. We acknowledge that this phrase has been used in other contexts like business, foreign relations, and healthcare pharmaceuticals [48–50], although a formal definition or outline of its features across various contexts is lacking. We propose that responsible transparency, as applied to AD biomarker status disclosure, represents an approach that prioritizes a researcher’s obligation to maximize transparency, while minimizing the potential for harm, in instances where it is ethically preferable to communicate results. Some may argue that existing constructs like informed consent are, when taken seriously and properly implemented, sufficient to guard against irresponsible transparency in the return of test results. However, informed consent occurs before one decides to have a test and/or learn its results and does not address the post-disclosure responsibilities of the discloser, which are critical considerations in the return of AD biomarker results. Stated another way, it is conceivable to imagine a scenario in which the ideals of informed consent were upheld at the pre-test stage, but in the days following the receipt of positive amyloid PET scan results, a participant experiences an acute, adverse psychological event that was neither detected nor addressed by the research team. Having fully and authentically engaged the individual in informed consent ahead of time does not release the team of any responsibility for the participant’s post-disclosure psychological outcomes. Hence, the need to articulate the basic features of responsible transparency in the return of potentially sensitive and upsetting research test results. We will first review current approaches to AD biomarker status disclosure in terms of both content and process, and then discuss possible strategies for extending responsible disclosure practices to the growing range of applicable scenarios.

Content

Several research teams have published commentaries and protocols describing the type of information that should be included in an AD neuroimaging biomarker result disclosure session [51–53]. This literature reflects a general consensus that communicating such results ought to entail (1) an overview of the testing that was performed, (2) a clear statement of the test results, (3) an interpretation of what those results mean, and (4) recommendations, or next steps, for the affected individual. Supplementing verbal communication with visual information displays has been recognized as a best practice in provider-patient communication and may be of particular value in AD biomarker disclosure sessions [54]. For example, a recent trial used pictographs to communicate the risk of progressing from MCI to AD dementia based on amyloid status [11].

Process

Because AD neuroimaging result disclosure protocols have typically been implemented in research settings, their authors have worked from the assumption that the information would be conveyed during a real-time personal communication, often a face-to-face office visit. Face-to-face visits provide the opportunity for dialog that can minimize the potential for “noise” to interfere with messaging during result disclosure encounters, whether that “noise” takes the form of more traditional transmission problems or the present concerns about (for example) cognitive and/or psychological barriers to communication. In the context of cognitively symptomatic populations, face-to-face encounters provide a critical opportunity for the results discloser to conduct both verbal and nonverbal assessments of comprehension of the information being conveyed, and to immediately clarify key points when warranted. Paralleling these formal and informal assessments of comprehension are real-time gauges of emotional reactions that allow disclosers to offer support and make note of individuals who may require further check-ins or support post-disclosure. In addition, the vast majority of research scenarios require that the participant be accompanied to the disclosure session by a research study partner, typically a family member. This practice provides an extra layer of protection against misunderstandings and failures of recall and increases the likelihood that distressed individuals will receive emotional support in the days and weeks following the encounter. Note that this sort of psychological support might be precisely what is required to help interactive communication.

Responsible Disclosure

Careful attention to the content of, and processes for, AD biomarker results communication, as described above, represent key tenets of the construct of responsible transparency. We maintain that the potentially sensitive nature of AD biomarker testing and the resultant risk of psychological distress (especially in those with active depression or anxiety who would have been excluded from extant published reports) require us to outline the conditions of an ethically responsible approach to the return of these results across a range of scenarios. If we accept the above articulated position that there are a growing number of situations in which AD biomarker test results communication is ethically correct, we should also acknowledge that not all disclosure sessions can be as tightly controlled as those which have been conducted as part of research protocols. If we consider a continuum that is anchored, on one end, by an encounter where the content and process of communication adhere to the descriptions above, and at the other end, a test result is simply delivered to an individual, we might label those ends of the continuum responsible and irresponsible, respectively. In terms of the communication models described in previously, the former corresponds to a transactional model and the latter with a transmission model.

As the practice of AD biomarker disclosure accelerates, it will be critical to develop and support disclosure practices that promote transactional communication, not only during the pre-test decision-making period, but during the communication of test results. Looking ahead to an era where amyloid PET results, for example, may be more commonly returned in research, and sought in practice, there is a great need for readily accessible patient education materials that contain the core information described in the content paragraph of this section. Regarding process, several considerations should be kept in mind as the field works to promote transparency with respect to AD biomarker testing. We propose the following as initial steps toward the development of recommendations for best practice.

First, the preferred manner of results receipt should be elicited from the individual at the time that the option for testing (in clinical practice) or for the receipt of results (in research) is discussed with the individual. During this discussion, potential result outcomes and their implications can be discussed, which will aid the individual in making an informed choice about whether and by what means to pursue results disclosure. Results may be communicated through face to face visits, telephone or videoconference calls, mailings, or electronic health record (EHR) postings. While limited pre-disclosure discussion of results scenarios may be acceptable if face-to-face disclosure is planned, significantly more advanced discussion may be warranted if EHR will serve as the communication vehicle. Regardless of the communication mechanism, patient education materials should accompany the results.

Second, the need for follow-up should be explicitly considered. Given that not all biomarker-positive individuals will present with companions (e.g., study partners in the research setting, or family members in clinical practice), at least one follow-up contact with such persons should occur. The follow-up should include assessments of comprehension of the information communicated and psychological well-being. Similar follow-up assessments may be indicated for individuals with concurrent mood disorders and those who have demonstrated potential for misunderstanding their results, whether during pre-scan discussions or results return sessions.

Given findings from the Imaging Dementia-Evidence for Amyloid Scanning (IDEAS) study [55] showing that, in most cases, amyloid PET scan results alter the clinical management of cognitively symptomatic persons, follow-up contact may be occurring in many of these cases. Follow-up contact will be all the more likely should FDA approve an anti-amyloid agent. Incorporating assessments of comprehension and mood into such discussions should be reasonable.

Finally, we note a need to be intentional with regard to terminology in the professional discourse on returning AD biomarker results to research participants or patients. The phrase “results disclosure” is common in publications on this topic, including in this article. However, to the extent that “disclosure” implies a unidirectional transfer of information from one person to the other, there is a risk for this term to promote a transmission model of communication. The phrase “results communication,” on the other hand, better aligns with a transactional, interactive model of communication that we have advocated for in this section on responsible transparency. We therefore recommend adopting the phrasing of “results communication” in the discourse on return of AD biomarker results and in discussions with research participants, patients, and families.

Conclusion

The receipt of amyloid PET scan results may elicit a variety of psychological reactions among cognitively symptomatic research participants and their family members. While concern exists for psychological harm associated with sharing positive AD biomarker results, the question of when to communicate results—and in particular how to communicate with patients who show some hesitancy to receive them—proves more nuanced than it might appear at first glance. Researchers (and providers) should be sensitive to the contextual and/or psychological factors that might impede successful communication. This includes understanding the degree to which individuals desire to know this information, as well as their own personal motivations. In the case of undecided or hesitant individuals, it may be worthwhile for providers/researchers to evaluate the communication context in order to create a more receptive environment where these results can be shared. When it is ethically preferable to communicate results, we outline several principles of “responsible transparency” for how best to do so.

Supplementary Information

Below is the link to the electronic supplementary material.