Introduction
Portable Magnetic Resonance Imaging (pMRI) for neuroimaging research in remote field settings holds great potential to reach populations previously excluded from research, including minoritized and underserved communities underrepresented in current neuroscience databases.1 However, research conducted far from a medical institution and in populations that may not have adequate health coverage or established primary care relationships raises the question of how to manage incidental findings (IFs). These findings may warrant clinical work-up, sometimes urgently. Many prior guidelines on return of IFs to research participants condition return of results (RoR) and IFs on clinical actionability.Reference Wolf and Green2 However, in a population with poor access to clinical care, some commentators have questioned whether IFs are actionable as a practical matter and have suggested reduced or no return of IFs to research participants.Reference Ortiz-Osorno, Ehler and Brooks3 Other commentators have argued that historically disadvantaged participant communities have an equal or greater entitlement to their results and IFs.Reference Sullivan and Berkman4 With the emergence of pMRI facilitating neuroscience research in such communities, this debate cannot be avoided.
Prior ethical guidance has addressed return of results and IFs (or secondary findings, which we include under IFs in this discussion of neuroimaging research5) when a subset of participants may lack health care coverage and established care relationships; the dominant recommendation is to anticipate this problem and establish a pathway to referral and clinical workup.Reference Wolf6 Some studies may pay for that initial clinical workup as part of the research budget, though they will typically not pay for subsequent care.Reference Wolf7 However, moving the locus of neuroimaging research outside the hospital setting to field locations that may be hundreds of miles from a medical home raises the possibility that most or all of the research participants have been underserved and may lack health care coverage and a medical home. Some pMRI research will predictably raise this issue.8 Indeed, pMRI research in remote settings may involve data transfer to a cloud platform, identification of IFs enabled by artificial intelligence (AI), and interpretation by researchers and clinicians who are far from the location where scanning was conducted. Dealing with these issues in pMRI research in remote and underserved field settings thus requires ethical analysis and a plan for addressing these problems in partnership with the participant community. Researchers will need to articulate the agreed plan in their research proposal and protocol, and in seeking an adequate research budget.
The goal of this paper is to analyze the question of how to approach IFs in pMRI research in remote field settings and generate initial recommendations. Our focus here is on IFs identified by structural MRI, as there is currently less consensus on what would constitute an IF in functional imaging with MRI.Reference Soumaré9 Decades of work on managing IFs emphasize the inevitability of IFs in neuroimaging — with varying frequency and significance, depending on the age of the population and recruitment criteria — and the necessity of advance planning and transparency. We argue here that historically underserved populations that may be reached by pMRI research deserve no less: research in these populations warrants careful community consultation to devise an acceptable plan, consideration of the potential impact of an incidental finding in the context of inadequate health care coverage, and a focus on research that avoids exploitation and instead confers local value including by offering information of potential health importance.Reference Shen10
I. Past Work on Incidental Findings
The earliest writings on IFs from imaging the brain focused on disclosure based on urgency, with clinical actionability the dominant criterion used to determine whether to return an unexpected finding to a research participant.Reference Katzman, Dagher and Patronas11 Subsequent work took a more holistic perspective, considering data on IF incidenceReference e.g12 and the perspectives of potential recipients,Reference Kirschen, Jaworska and Illes13 as well as age and cultural factors,Reference Vander Wyst14 in an effort to develop recommendations and protocols. As consideration of offering IFs as well as research results to participants has unfolded over more than 25 years of work, criteria discussed to determine what findings to offer research participants have expanded to consider findings of potential personal importance (whether or not the findings are clinically actionable in the sense that they could be used by clinicians to alter diagnosis or care), and to contemplate offering all findings in some studies. Examples of potential personal importance can include reducing anxiety, enabling participants to communicate with family members, and informing decisions about health and life insurance.Reference Bacon15
The literature on return of results and IFs is now extensive and ranges from normative analyses to empirical studies, policy reports, and case studies. In the context of neuroimaging research, that literature generally converges on three principles: (1) findings in structural brain imaging can indicate pathology and may require urgent clinical attention, (2) increased imaging resolution will yield increased numbers and types of IFs, and (3) it is essential to establish a pathway for managing IFs in the earliest phases of research design, with appropriate provision for informed consent, expert consultation to evaluate potential IFs, and effective referral to clinical care when clinical pursuit of the IF is warranted. The introduction of pMRI capabilities take the already complex discussion of how to manage IFs to a new level: field strengths vary for pMRI, the support of a well-resourced medical center may not be available when pMRI scanning is conducted in a rural and remote community, and preferences for management of IFs as well as cultural norms for control of data may vary by participant community.
Based on past empirical and theoretical research, procedural approaches to identification of IFs in research may take at least four forms: (1) no screening for IFs, and no offer of any findings; (2) no screening for IFs, but offering findings of potential importance that are accidentally discovered; (3) systematic review of those scans on which a suspected anomaly is spotted; or (4) routine review of all scans for findings of potential importance.Reference Graham, Hallowell and Savelscu16 Similarly, there is a range of substantive approaches to defining the set of IFs to be offered to research participants: (A) offering none; (B) offering only IFs suspected to be clinically actionable; (C) offering a broader set of IFs, including those of potential personal importance to research participants even if not clinically actionable; and (D) offering all findings. Table 1 depicts likely combinations of screening strategies and scope decisions.Reference Wolf and Evans17
Combinations of procedural approaches to identifying IFs (vertical axis) and substantive approaches to defining the scope of IFs to offer (horizontal axis) are marked with a check.
The trend has been away from options (1) and (A) — ignoring IFs and offering none — as studies overwhelmingly show participant interest in receiving them and ethical duties including respect and reciprocity have been seen to weigh in favor of identifying and offering these findings to participants.18 For example, a study by Kirschen et al. demonstrated overwhelming receptivity to disclosure of unexpected neuroimaging results regardless of expected degree of clinical actionability.19 A study by Wilkins and colleagues has argued that the scope of findings offered to research participants should be based not on clinical actionability, but on what information the participants value, which may extend beyond IFs and return of results to additional insights generated by the study.Reference Wilkins20 Based on a national survey of a diverse U.S. sample to ascertain their views on return of research results, those authors “found that participants across all demographics highly valued receiving information from research studies and were more likely to trust researchers and to volunteer if research information were returned.”21
In deciding what approach to take, researchers, their institutions, and review boards should consider participants’ right to control information about themselves, researcher duties to respect participant autonomy, and researchers’ ancillary care obligations.Reference Richardson22 Those undertaking and overseeing pMRI research should also prioritize the imperative to collaborate with the participant community in planning and executing research, including planning for the management of IFs.Reference Shen23 Numerous commentators have argued that there is a moral duty to offer information on IFs to participants who contribute their time and brain data,Reference Sadatsafavi24 that is, to return information about an IF that is identified, accompanied by referral to resources for further information and clinical follow-up.
However, the specific challenge this paper addresses is return of IFs in remote and underserved populations. McMahon and colleagues, analyzing researcher attitudes toward RoR in under-resourced settings, have questioned, “how the process of return would actually unfold, especially in under-resourced, under-served, and under-represented communities.”25 They note that, “[p]ersistent disparities in access to health care have raised additional questions of whether such results are, or can be, equitably incorporated into clinical care….”26 Graham and coauthors have expressed concern over return of IFs in health care systems that are publicly funded, where pursuing ambiguous IFs may consume scarce health care resources.Reference Murphy and Weijer27
II. Ethical Duties in Underserved Participant Communities
Arguments made by other commentators that researchers owe less of a duty to ascertain and offer return of IFs in underserved and under-resourced participant communities turn on several claims: (1) that participants lack the resources to pursue clinical follow-up that would yield benefit, (2) that participants would thus experience more burden than benefit from return of IFs, and (3) that efforts to ascertain, return, and offer initial clinical assessment of whether further clinical care is needed would unreasonably deplete research resources, thereby disadvantaging those who stand to benefit from the research knowledge sought by the study. We consider each of these claims in turn.
The argument that individuals should not even be offered information of potential clinical importance if researchers have already concluded these individuals lack to resources to fund clinical follow-up that may yield benefit (the first claim) makes several questionable assumptions. It assumes that investigators looking at a research IF that has not yet undergone clinical evaluation already know the nature of the finding, what surveillance or interventions (if any) will be recommended, what course the patient will then choose, what costs are associated with that chosen course, and what funding options the patient will have (or be offered) to pursue that course. Of course, when an IF is initially identified in research, the investigator has little basis on which to make these assumptions. Nor will investigators generally have a sound basis for the second claim, that the participant will experience more burden than benefit. Individuals vary in what they consider a benefit and a burden. Moreover, the literature on return of research results and IFs recognizes a wide range of potential benefits, both clinical and non-clinical.28 These include knowledge about one’s risks and health, information that might motivate the individual to seek additional insight about the finding or to volunteer for a relevant clinical trial, and the opportunity to share the finding with family members or other loved ones. On the burden side, overwhelmingly studies on return of results and IFs have not found significant and lasting psychological harms, even when the finding presented to the participant indicates a serious and untreatable condition.29
When investigators fulfill their obligation to consult the participant community in co-designing the study30 — including co-designing the management of IFs — investigators have the opportunity to elicit community perspectives on what IFs should be offered, how initial clinical evaluation should proceed, and then how to offer a pathway to timely clinical care for those participants who need and choose it. The community can also work with the investigators to anticipate the types of IFs likely to arise in the planned study, in order to consider what pathways to care are needed and how they can be funded. Researchers should not deprive communities of the opportunity to collaborate on ensuring access to care for participants with IFs. On the contrary, investigators should partner with participant communities to assess what pathways will likely be needed and whether the research funder or others can help provide access. These investigator duties flow from obligations to avoid treating research participants as mere means, to avoid research that extracts knowledge from a community with no return of local benefit, and to respect the needs and preferences of participant communities, especially those communities that have historically been exploited or neglected in research.31
The third claim, that managing and returning IFs will deplete the research budget and reduce the ability to provide future benefit from research findings, is also problematic. A large literature has used a range of ethical arguments — including ancillary care, duty to warn, and reciprocity to research participants — to support researcher duties to manage IFs. These duties, like virtually all ethical duties that researchers shoulder, require resources and personnel effort. The fact that conducting research ethically takes time, effort, and resources is a truism, rather than an argument against those ethical responsibilities.
Some commentators have nonetheless questioned whether there is a limit to what proportion of the research budget should go to managing IFs, though there is no agreed formula or percentage limit.Reference Compare32 The problem with the question as often framed is its assumption that any resources spent managing IFs necessarily take away from resources to create new research knowledge. Duties to manage IFs responsibly devolve on both researchers and funders; research funders should consider enlarging the budget, if necessary to conduct the research ethically and in a way acceptable to the participant community (including with respect to managing IFs).33 A number of studies have included in their research an investigation of how best to manage return of results and IFs,Reference Green34 so that resources devoted to managing IFs contribute to the knowledge gained. That knowledge may be particularly lacking and needed in historically underserved participant communities.35
Ultimately, participant communities that have been underserved and neglected in neuroscience research have at least as strong a claim to the return of IFs as other more resourced communities that have not been excluded. We do not adjudicate among the differing ethical theories and arguments supporting return of IFs; together they offer a robust ethical architecture that has supported a sea change toward a new normal of return of results and IFs.36 Against that background, we argue that if those participant populations that have more resources are offered IFs, then underserved and under-resourced populations should not be deprived of this information because of their historical and comparative disadvantage.37 Indeed, considerations of justice, equity, and respect give the claims of such populations added weight. Moreover, the imperative to include previously neglected populations in neuroscience research to diversify neuroscience data sets and ensure that increasing knowledge of the brain includes all populations38 requires developing approaches to managing IFs that address barriers to care in underserved communities.
This means that community consultation is key in resolving how to handle IFs in remote and under-resourced populations. Respect for participant and community preferences and careful attention to how they see the benefits and risks of different IF strategies are essential. This is especially the case for research involving participants who may not have a medical home, who have historically had poor access to medical care, and who may hold community-based and culturally specific views about wellness that differ from narrowly biomedical explanations, or who may have experienced past adverse relationships in prior research. Moreover, transparency is vital for remote field research with pMRI as it may involve data transfer to a cloud platform, AI screening, and potentially major geographical, financial, structural, and cultural barriers to access follow-up clinical care. We consider these variables next.
III. Challenges in Managing IFs in Remote pMRI Research
Portable MRI research data collected in a rural or remote region will not reside locally in most cases. Instead, the data will likely be transferred to a cloud platform and ultimately to a research institution that is far from the point of collection.39 This situation is not unlike that in biobanks and archived data, which have become crucial engines of genetic and genomic research.40 However, because identification of IFs requires access to and analysis of data, a research study’s approach to IFs must be embedded in a broader approach to consent, data management, and control. This too requires community consultation and the development of a data management plan that is acceptable to the participants and community. Some communities may require that they maintain control of data — for example, through data sovereignty and data ownership41 — while others may negotiate shared governance or protection for participant data access and control.42 The data framework developed for the study will affect the flow of data and feasibility of any IFs plan.
When identification or reidentification of an individual participant is possible and the community agrees to a research design involving an offer of IFs to participants, investigators should ensure that they can: (1) establish a process for timely review of brain scans, (2) specify the criteria for identifying an IF for potential return to the participant, (3) reliably reidentify the individual participant, and (4) implement a process acceptable to the community for recontacting the participant (see Table 2 ). This is the case whether scans are screened using AI methods, by humans with radiology expertise, or both. AI is proving to have highly accurate diagnostic capabilities for many, but not all communities and conditions.Reference Oren, Gersh and Bhatt43 As AI improves and bias is reduced with more inclusive training sets, the benefits of better detection of clinical actionable findings may well accrue to research in rural and remote locations, even for scanners with field strength that is limited to 0.5T.
Summary of key steps for researchers to take in planning the management of incidental findings (IFs) for pMRI research in remote field settings, so that the plan can be incorporated into the research protocol.
* Return of genomic IFs and results, for example, has involved the U.S. Food and Drug Administration’s Investigational Device Exemption (IDE) process. See, e.g., E. Venner et al., “Whole-genome Sequencing as an Investigational Device for Return of Hereditary Disease Risk and Pharmacogenomic Results as Part of the All of Us Research Program,” Genome Medicine 14 (2022): 34, doi: 10.1186/s13073-022-01031-z. The use of investigational neuroimaging devices may also trigger regulatory requirements for management of IFs.
While the flexibility of pMRI has great potential to democratize brain imaging research and advance much-needed inclusion in neuroscience research, conducting research in remote and underserved communities may challenge three dimensions of the workflow for IFs: researchers’ ability to ensure (A) timely review of scans by expert radiologists or neuroradiologists, (B) a timely offer of information and counseling concerning IFs to the affected research participants, and (C) participant access to follow-on clinical evaluation and care. We consider each of these in turn.
Ensuring timely review of scans — whether the researchers and community have agreed to narrow review (e.g., only scans where a concern has been noted) or broader review (e.g., routine review of all scans) — requires that the research team has engaged radiologists or neuroradiologists.44 Providing them with rapid access to scans may require a workflow markedly different from the process in conventional MRI research within a major hospital. If scanning is done in a remote setting,Reference Deoni45 data captured will likely be transferred to a cloud platform for interrogation and interpretation.46 Creating a workflow attentive to IFs will take planning before scanning begins and should be reviewed by an oversight committee such as an institutional review board (IRB) with independent authority and relevant expertise. Because a subset of IFs may require urgent clinical evaluation, the timeliness of expert review is crucial. In remote field settings, timely screening may also be needed to avoid losing the participant to follow-up.
Because some IFs will require rapid clinical work-up, participants who have agreed to receive information about their IFs should be offered that information in a timely way. In addition, participant mobility and the risk of losing participants to follow-up call for a timely offer of information. This too will require careful planning, appropriate personnel, and funding.
Perhaps the biggest challenge will be determining how to offer timely clinical evaluation and a pathway to follow-up care when a participant has elected receipt of an IF requiring clinical attention. Even providing indicated clinical care outside of a research protocol can be difficult in remote and under-resourced populations. For example, Harding et al. interviewed and surveyed neurosurgeons, neurologists, and other specialists across Canada to assess their perspectives on patient access to neuromodulation for drug-resistant pain, epilepsy, mental health, and movement disorders in remote areas of the country.Reference Harding47 The mixed-methods study found discordance between the perceived imperative of providing possibly lifesaving interventions for the four conditions, and the likelihood that these interventions would be available.
As pMRI moves into research use, a range of responses may emerge to discovery of IFs requiring clinical attention. One option is to offer the participant transport to a clinical center that can provide the evaluation needed. While this may be an appropriate option in some circumstances, burdens may include the disruption of asking a participant to leave their community for travel to a potentially distant and unfamiliar site. In addition, air transport for urgent clinical evaluation may be costly for the research budget and impossible for the participant to pay.Reference Young48 Much work is needed to understand the types of IFs that would need to be addressed by travel to a medical center, who would pay for the response, and the views of participants and their communities about the risks and benefits of transporting a person far from home for this purpose.
However, transporting the research participant to a clinical center is only one option. As pMRI research becomes more common in remote field settings, creating outposts offering clinical evaluation may be worth considering, especially for multiple research projects operating in overlapping areas. Portable MRI scanners with the demonstrated and approved capacity to provide clinical diagnosis could be situated strategically and used as needed. Even if the researchers are using a low-field or ultra-low-field scanner, cooperatively positioning scanners with proven diagnostic capacity may be part of the answer to the IFs challenge.
IV. Responsibilities in Conducting pMRI Research in Field Settings
With pMRI research emerging now, the time is ripe to address these issues as IFs are sure to arise in the field research that pMRI allows. As the systems begin to roll out and hands-on experience is gathered, this is the moment for researchers and professional organizations to attend to both the benefits and challenges. Partnering with the participant community and collaboratively devising solutions will be key. For example, in research involving First Nations, Native American, or Alaskan Native populations, using principles and approaches grounded in both Western bioethics and Indigenous methodsReference Stevenson49 will be important to address the IFs issue responsibly, with full understanding of IFs that could be identified; potential opportunity and burden for research participants, their family, and the community; and collaborative problem solving. Community leaders, Elders, and community members should not only be consulted, but should be full partners in addressing these issues.
Where research participants face significant barriers to accessing clinical evaluation and care — whether those barriers are a function of geography, economic hardship, reduced access to clinicians and care facilities, or other obstacles — addressing the reality that neuroimaging may uncover IFs can catalyze advocacy for access and innovative strategies. Clinicians at major medical centers, as well as providers at nursing stations and medical facilities in or near communities where research may take place, must be prepared to receive research participants who may become prospective patients through pursuit of IFs. This requires planning for and with a community well before the protocol starts, evidence-based analysis of the likelihood of IFs in the cohort studied, as well as plans for any broader return of results. The alternative of giving up from the start and denying participants the opportunity to receive information that wealthier and more resourced participants would be offered raises serious equity concerns. Such refusal to address the IFs challenge would only deepen health inequity. Under-resourced participants would be contributing to neuroimaging research and knowledge while being systematically deprived of information they would receive if they had more resources.
The argument that IFs should not be offered to less-resourced participants or communities because they cannot afford to take clinical action similarly entrenches inequity and fails to consider access options. As noted above, some studies provide clinical evaluation on the research budget.50 Moreover, major studies (or smaller studies working in concert) may have the leverage to create pathways to clinical evaluation and care. In addition, research participants may value being offered their results and IFs for a wide range of reasons — not just to access clinical interventions to alter diagnosis and treatment. When researchers offer a participant desired information about their findings, the researchers are demonstrating respect for the preferences and values of that participant. Indeed, communities may refuse to allow researchers access unless the researchers agree to share such findings.Reference Morello-Frosch51 Moreover, some legal frameworks will require researchers to offer participants access to their individual specific findings52 and HIPAA-covered entities in the United States have duties to provide a participant with their research data at their request.53
In neuroimaging research, the incidence of IFs may make the problem more tractable than it first appears, though incidence varies by population and age group. IFs are generally found in less than 1% in people who are under 21 years (although these tend to be the most serious findings when detected), the incidence of clinically actionable neuroimaging IFs remains low in the adult population, and despite the high percentage in people older than 65 years (as much as 50%) these IFs are generally age-appropriate and require only routine follow-up.Reference Katzman, Dagher and Patronas54 However, large-scale prospective studies on the incidence of IFs in research participants from underserved and under-resourced remote settings are needed to form a complete picture of the scope of the IFs challenge.
This calls for more research in remote field settings. Such research will require partnership with these communities, including on the return of results and IFs. Federal, state, provincial, and private funders should support this work, including responsible management of IFs. Failure to fund and plan for IFs risks deepening health inequities and perpetuating the exploitation of underserved populations by conducting neuroscience research without offering IFs that would be routinely offered in other studies. Portable MRI research has enormous potential to make progress in addressing inequities and historical exclusion from research. It also offers the potential to use pMRI to bring clinical evaluation and care closer to those in settings far from major medical centers.
Conclusion
Both bioethics and neuroscience are increasingly confronting past exploitation of marginalized communities and demonstrating a commitment to equity, justice, and respect.Reference Fabi and Goldberg55 Brain imaging research using pMRI has the potential to significantly broaden research participation and create more representative neuroscience databases yielding a fuller understanding of human diversity. The World Health Organization (WHO) advises that neurological disorders are the leading cause of disability worldwide and the second leading cause of death globally, accounting for approximately 9 million deaths per year.56 Ensuring attention to underserved populations and analysis of the disorders those populations face is essential to progress.
Breakthroughs in discovery research can directly impact the translational care pipeline. When studies are conducted in historically underserved communities and lead to therapeutic interventions, the likelihood that they will be applied in those communities is greater today than ever. Portable MRI will enable that. Health outcomes may not immediately improve in areas where medical care is lacking, but an increase in research on health that is inclusive and culturally competent in such areas will foster trust where it has not existed in the past or has been squandered. Making return of IFs work effectively in field research with historically neglected communities — with respect for local perspectives and priorities, creation of a feasible and supportive pathway to care, individual participant choice, and robust protections for participant confidentiality and data control — is a crucial next step.
Acknowledgments
Preparation of this article was supported by National Institute of Mental Health, National Institutes of Health grant RF1MH123698 (Shen, Wolf, Lawrenz, PIs). J.I. is supported by the UBC Distinguished University Scholars Program and the North Family Foundation. The views expressed in this article are those of the authors and not necessarily those of the funders. Thanks to Peter Lyon, J.D. candidate at the University of Minnesota Law School, for helpful research assistance.
Disclosures
The authors report no relevant disclosures.
Open access
