Introduction
The nature and extent of information produced in genomic research studies that should be fed back to participants is hotly contested. Next-generation sequencing (NGS) technologies, including whole genome sequencing (WGS) and whole exome sequencing (WES) have supplemented panel analysis, and are now used extensively in genomic research.Reference Wolf and Green1 Each of these technologies produce substantial amounts of sequence data. WGS in research studies often comprises both germline and somatic testing, to enable comparison of germline and tumor sequence data and identification of germline variants versus somatic mutations.Reference Mandelker and Ceyhan-Birsoy2 There is considerable literature debating what findings should be returned to participants, particularly where those findings are not related to the primary indication for testing (incidental or secondary findings).Reference Vears3 Where such findings are clinically actionable, there is increasing consensus that they should be returned to the individual to enable autonomous choice about health management and avoidance of harm (where “clinically actionable” denotes there is some clinical action that can be taken, noting the potential expansiveness of this term).Reference Gornick4
The recently developed draft World Health Organization (WHO) “Guidance for Human Genome Data Collection, Access, Use and Sharing” stipulate principles of social justice, declaring that “Return of results to individuals should be considered in cases where: results are clinically relevant and could be validated; return is feasible within the local health setting; and the return of results is legally and ethically permissible.”5 Even in clinical practice, however, decisions regarding return of results are finely tuned, particularly in respect to the multiplicity of variants of unknown significance (VUS), and when they should be reported to patients.Reference Hoffman-Andrews6 There is even less clarity in relation to the return of raw genomic data. This is sequence data that has not undergone interpretation or annotation comprised in read files but not including call files. We define Binary Alignment/Map (BAM), Sequence Alignment/Map (SAM) and FASTQ files as raw genomic data. Although “raw” in the sense that they have not undergone interpretation or annotation, Variant Call Files (VCFs) have undergone an additional step of processing to extract specific variants that preclude their return to FASTQ status.Reference Johnston7 Internationally, there has been discussion as to whether raw genomic data should be returned to individuals, many of whom are recruited to research studies through clinician referral.Reference Thorogood8
Common practice in genomics research and medical practice is to limit information provided to participants who have undergone NGS. In the clinic where the primary purpose of sequencing is diagnosis, clinicians (rather than patients) are often provided with a summary report rather than raw data. Much genomics research melds the clinical/research paradigms by defining and stratifying cohorts based on clinical data from individuals identified through research as having known mutations or clinical indicators of disease.
Providing research participants with access to raw genomic data involves different considerations than returning secondary/incidental findings.Reference Evans and Wolf9 The return of secondary/incidental findings emanates from the duty to warn or the duty to rescue,Reference Richardson and Cho10 but raw genomic data has no immediate clinical validity or relevance as it contains no interpretive material or annotations. However, it embodies an individual’s whole genome and, accordingly, has significant potential implications for family members and close relatives.
Context
Patients have a clear right to access their clinical raw genomic data under privacy laws in many jurisdictions,Reference Liddell11 but access rights of research participants to research data are more equivocal. Under US law, analysis of the HIPAA Privacy Rule (45 C.F.R. § 160, 164 (2018)) suggests that it provides a legal right to research participants to access a copy of their raw genomic data held by HIPAA-regulated hospitals and laboratories, based on the fact the data could be used for follow-up testing and clinical care.Reference Barbara and Evans12 However, this right to access data from research laboratories for confirmatory testing in a clinical context does not constitute “clinical care,” which may only be provided by state-accredited clinical laboratories.13
Literature examining European law has concluded that Article 15(1) of the General Data Protection Regulation (GDPR) (Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 O.J. (L 119)) may provide research participants with a right to gain access to their raw genomic data.14 Where there is conflict between the GDPR and national laws, interpretation to accord with the GDPR is generally undertaken.15 However, Article 89 of the GDPR permits derogation from this right of access where data is processed for scientific research purposes, and allowing access would seriously impair achievement of those purposes. Member states retain discretion as to whether the access right applies to data generated in research settings in that state.Reference Narayanasamy16 Given that similar access rights to health data exist under UK law (Data Protection Act 2018, c. 12 (UK)), the same conclusion is likely. There appears to be no other basis under UK law for the provision of access to raw genomic data for research participants.Reference Kaye17
The position in Australia is more problematic. Australian legal and ethical frameworks are contradictory: the “National Statement on Ethical Conduct in Human Research” provides “[w]hile participants may have a strong interest in their own information, researchers are not expected to return raw genomic data to participants.” (para 3.3.28)18 This ethical position is, however, at odds with legal obligations for access to information under Australia’s privacy regime (Privacy Act 1988 (Cth) (Austl.)),19 whereby individuals may access health (including genomic) data held by relevant government agencies/organizations under right-to-information legislation. Research organizations conducting research using health data fall within the scope of privacy legislation.20 This conflict in ethical guidelines and the legal framework results in ambiguity in the standards guiding researchers.Reference Nielsen21
Importantly, these legislative regimes provide a right of access but not an obligation to return data in the absence of a request, particularly in relation to raw data generated through genomic research. The norm in most research studies is that raw genomic data are not returned,Reference Wright22 although some research programs aim to place data under the direct control of participants.23 Several ethical principles support the return of raw genomic data to research participants. Respect for autonomy and beneficence would dictate returning an individual’s raw genomic data in promoting an individual’s right to determine how and by whom their information is used.24 Returning raw data would enable research participants to pursue further interpretation and possibly targeted therapies where a research study has not yielded diagnostic or therapeutic options.Reference Shabani25 Participants may seek interpretation or reanalysis through another bioinformatician, a clinical genetics service or a third-party interpretive (TPI) service. This may be helpful where participants wish to save costs, reuse germline data, or have somatic data reanalyzed. TPI services have evolved to offer services to the public and health professionals: some now accept data generated by others in raw formats. Providing participants with the capacity to control the use of genomic sequence data may further a willingness to engage in researchReference Middleton26 and assist in meeting patient expectations.Reference Nelson27 Even where the utility of the information is still not fully understood or where the data is still being analyzed by researchers,Reference Lunshof28 support for return has been premised on the fact that participants’ data is “theirs.”Reference Schickhardt29
Ethical grounds for raw data return should be balanced with potential concerns. The veracity of research-grade results generated in nonaccredited laboratories may be questionable, and any data provided are not certified, limiting the extent to which it may be relied upon for downstream clinical decision-making.30 If shared by participants, data may be stored or used in ways that create privacy risks for participants and their families.Reference Guerrini31 Data may be misinterpreted and many TPIs are unregulated. These providers may give erroneous, incomplete, or poorly explicated advice,32 and a high rate of false positives has been reported.Reference Moscarello33 This could prove burdensome if anxious participants seek medical reassurance from treating clinicians who may lack relevant knowledge/skills to manage patients’ concerns.Reference Cusack34 These risks might be mitigated if data are returned to participants with adequate processes, safeguards and information,Reference Chad and Szego35 although this raises resourcing issues for research projects,36 especially in light of predicted increases in request numbers.
Study Aims
This program of work sought to interrogate the Australian regulatory framework to determine whether it provides sufficient guidance for clinicians, researchers, the public health system, and individuals where participants in a research project request the return of raw genomic sequence data. The initial phase comprised regulatory landscape mapping and exploration of ethical and legal issues associated with the return of raw genomic data. The second phase involved interrogating the understanding and attitudes of key stakeholders on whether raw genomic data generated in a research setting should be returned to study participants, and what an ethically defensible process for return would look like. This involved qualitative inquiry through semi-structured interviews with key stakeholders. This paper reports the results of this qualitative arm.
The collection of qualitative data was informed by two fundamental research questions: should raw research data be returned to study participants; and if returned, what factors should be taken into account in facilitating a process for return?
Methods
Study Design
Incorporating semi-structured interviews, we sought to generate knowledge focused on the particular expertise and experiences of our study participants.
Participants
Participants were recruited from two national cancer clinical trials, the Australian Pancreatic Cancer Genome Initiative (APGI) and the ZERO Childhood Cancer Program (ZERO), both of which are precision oncology initiatives utilizing NGS, whole transcriptome, and other sequencing/profiling to identify biomarkers for disease and treatment options. From ZERO, we recruited parents of children with cancer aged under 18, and from APGI we recruited adults with pancreatic cancer and family members of now deceased APGI participants. We also recruited treating clinicians, genetic counsellors, scientists, and bioinformaticians involved in the studies. Representatives from patient advocate groups and policy organizations were also included. Sydney Children’s Hospitals Network Human Research Ethics Committee (2021/ETH00733) and Queensland Children’s Hospital/Centre for Children’s Health (SSA/2023/QCHQ/94229) approved the project.
Recruitment and Inclusion
Potential interviewees were initially emailed to ascertain interest and were then provided with written information about the study via the secure web application Research Electronic Data Capture (REDCap) prior to consenting to participation.
Selection of participants was undertaken according to the following inclusion criteria: (a) adults enrolled in the APGI study and their family members; (b) patients enrolled in the ZERO study; (c) health professionals providing clinical care to these groups; (d) scientists involved in genomic analysis and interpretation of genomic data in these studies; (e) representatives of patient advocacy groups associated with ZERO and APGI. Participants were excluded on the following grounds: (a) participants without a conversational level of English or willing to have an interpreter present; (b) parents of children enrolled in ZERO where the child had a life expectancy of less than six weeks from the date of recruiting, or where the child had died within the last six months.
Participants were recruited following identification of suitable candidates by treating clinicians involved in ZERO and APGI. Health professionals were purposively identified based on their specialization/involvement with the ZERO and APGI cohorts. Advocacy/policy representatives were identified through our existing investigator networks. Policy organizations were defined as government-funded bodies involved in patient support and policy development. An initial list of possible interview recruits was compiled and included APGI patients (n=3); policy representatives (8); scientists/researchers (13); clinician/researchers (9); bioinformaticians (5); and genetic counsellors (3). We are unaware how many ZERO patients/families were approached for interview as clinicians discussed the study during routine clinical appointments. At least two patients approached for interview declined.
We were aware that requests for raw genomic data had been made by patients involved in the ZERO and APGI studies, but our recruitment strategy was not limited to participants who had been involved in making or responding to requests.
The Interview Guide
All team members were involved in preliminary identification of issues for exploration in interviews. JN and CJ developed interview topics based on these issues. Topics were deliberately broad to facilitate narrative interviews. The interviews commenced with questions about interviewees’ backgrounds and went on to explore experiences with requests for return of raw genomic data, views on return and possible processes for return, with questions customized to interviewee category. The interview guide underwent iterative revision as early interviews highlighted certain concerns about the return of raw genomic data as being common to many interviewees. These included primarily safety/security concerns, with the result that these nuanced issues were explored in depth with all interviewees.
Analysis
We employed an inductive content analysis (ICA) approach (Vears and Gillam 2022). ICA is a method of analyzing text-based data through iteratively coding for themes. Data collection commenced in June 2022 and concluded in April 2024. Data collection was protracted due to unexpected delays obtaining site-specific approvals. Interviews were conducted by JN and CJ via web conferencing and ranged from 30 to 110 minutes. Permission to record interviews was sought from each participant. Recordings were transcribed verbatim and de-identified transcripts were coded and reviewed by JN and CJ, using nVivo software, to develop an explanatory matrix of interviewee experience relevant to our research questions. The coding framework was initially developed around the broad research aims, then inductively refined into themes and subthemes during coding. These themes included experience with returning raw genomic data/genomic sequencing results, views regarding return of raw genomic data, utility of raw genomic data, relevance of the consent process to return of raw genomic data, possible risks/liabilities arising from return of raw genomic data and what a process for return should look like.
Findings
A total of 36 interviews involving 40 interviewees were completed (four interviews involved two interviewees). All ZERO interviews were conducted with parents without children present. Two interviews were conducted with both parents present. Interviewees were categorized according to their primary roles (Table 1):
Categorization of Interviews and Interviewees
Table 1. Long description
From the top row downward, the table columns are: Categories, Abbreviation, Number of Interviews, and Number of Interviewees. The first row lists Scientist/researcher (bench scientist, no direct patient care), abbreviation S forward slash R, with 9 interviews and 9 interviewees. Next is Clinician or clinician/researcher, abbreviation C forward slash R, with 7 interviews and 7 interviewees. Genetic Counsellor, abbreviation G C, has 3 interviews and 3 interviewees. Bioinformatician, abbreviation B I, has 2 interviews and 2 interviewees. Advocacy, abbreviation Adv, has 5 interviews and 6 interviewees. Policy forward slash admin, abbreviation Pol forward slash Admin, has 3 interviews and 4 interviewees. Patient forward slash family (Z E R O 5, A P G I 2), abbreviation Pat forward slash Fam, has 7 interviews and 9 interviewees. The final row totals are 36 interviews and 40 interviewees.
The analysis that follows is organized into four primary themes in relation to fielding requests for raw genomic data: (1) whether raw genomic data should be returned; (2) reasons for seeking access; (3) risks in returning raw genomic data; and (4) processes for return.
Whether Raw Genomic Data Should Be Returned?
Just one scientist/researcher interviewee recommended not providing raw genomic data to study participants, on the basis that raw genomic data can be harmful if shared with someone who knows little about its power. All other interviewees believed that study participants are entitled to receive their raw genomic data:
[I]f I wanted it, and I asked the hospital for it,… I would be a little bit shocked if they said, “No, you can’t have it.” I think there’s an expectation there, an assumption that we should have access to it, because it’s my son’s personal data. [Pat/Fam6]
This view was held despite the fact the data were generated in a research (versus clinical) context, and irrespective of participant age. Some likened the situation to the return of other medical results such as x-ray imaging, where data might have direct implications for a patient: “… ultimately if the patient is saying ‘I want this,’ I feel like they do have the right to ask.” [S/R1]
Around two-thirds of interviewees considered that raw genomic data should only be provided upon request. One parent observed, “I don’t think it’s wrong to not offer it, but I do feel that if it’s asked for, that it should be easily accessible and provided.” [Pat/Fam6] Some interviewees also recognized that this could engender inequity.
… [O]nly if the patient demanded it … So really that then relies on the patient being informed and so there might be a disparity there between the sort of patients that might ask and those that don’t ask. [Clinician/Researcher 5]
Patient advocates also acknowledged that participants must have some understanding of the nature of genomic sequencing to request raw data, suggesting that in some cases “… that level of sophistication within the patient community is not there.” [Adv3] On the other hand, particularly in relation to younger patients,
I see a lot of people who are educated and they’re bringing questions to their doctors in a way that they haven’t done before and sharing these massive, long emails of the research … that they’ve found around the world and asking … how they will have access to this type of thing. So I think it’s very differently spread. [Adv1]
Several interviewees addressed the difference between germline and somatic (tumor) data. One clinician/researcher reported that there is a qualitative difference between germline and somatic data that differentiates the extent to which each should be considered personal information:
[T]here is so much benefit that a family can get from [the germline data] … So … it’s almost unethical to not offer the option … [The tumor data], it’s completely different … It’s so specific to the customised pipeline that we apply … And that’s why … balancing between potential harms and potential benefit of that is completely different … So do the families have the right to have that result? I’m not sure. [C/R 6]
Further, interviewees explained that tumor pathology changes over time, potentially rendering the data of questionable future value: “Your germline DNA is not going to change, but your tumour DNA is … These cancers evolve, so that a tumour sample is a snapshot in time.” [S/R 6] Clinicians/researchers also shared that there are different clinical implications in returning somatic findings:
[I]f there are germline changes, then that mandates genetic counselling. So anything that is inheritable should have genetic counselling … as the system stands, we don’t really have genetic counselling for want of a better word, for somatic mutations. [C/R 1]
Others expressed few concerns regarding returning somatic data, but viewed the sharing of germline data as more problematic, given its longer-term implications:
If you start talking about the germline data, I think that’s really where it becomes–where it gets really tricky … for me, somatic is a no-brainer … I think the germline may affect more things that we don’t know enough about right now, but in the future, we might. Yeah, being able to predict how that might affect the individual, as well as the family … I guess that’s much trickier. [S/R3]
Reasons for Seeking Access
We asked patients/families whether they had contemplated making requests for raw genomic data. Few patients/families we interviewed had made such a request. One family had made a request at the urging of an extended family member with relevant expertise and received tumor samples and genomic data. Another family had received an eight page report from their clinician which identified a somatic mutation and contemplated inquiring about the availability of further data. This interviewee indicated they were open to utilizing TPI websites to interpret genomic data and had used them previously. Both families sought this data to investigate additional treatment options for their children. No other patients had requested raw genomic data as it had not occurred to them, or they had found a treatment option through their involvement in the research program that negated the need:
Maybe because we had success. Maybe that was enough … But if we hadn’t had success, that’s a bit of a tricky one. If it was unsuccessful, I don’t know if I would’ve bothered. I might’ve wanted to just bury it and not go there. [Pat/Fam 3]
One family member commented they would not contemplate asking for the raw data because they did not think they would understand it, but added:
I guess the one thing that I would want to know is that if there was another study … we have the choice for them to have access to the data. [Pat/Fam5].
Some interviewees also raised the alternative prospect of centrally maintained platforms to facilitate patient-led data sharing, to enable re-interrogation.
Health professionals made similar observations that patients requested raw data to explore further prognostic/treatment options. Many interviewees spoke of patients nearing the end of standard treatment being motivated to learn as much as possible about their cancer, particularly where “… we haven’t found anything that’s targetable.” [BI1] While requests for access to raw genomic data had been encountered infrequently, they currently occur where:
[T]hey know someone or they’ve read something or they’re very higher level-trained or someone has approached them about this as an option. The average person participating in research I don’t think would have had that interest or they wouldn’t know what to do with it. [S/R1]
As observed by one interviewee in an advocacy role, patients enrolled in research are “really motivated” to use their raw data to seek further interpretation, particularly upon exhaustion of treatment options [Adv2].
Regardless of whether further interpretation is initiated, some health professionals also conjectured that participants may envisage storing the data with no current use in mind while retaining the option of future use. Some interviewees referred to increased genomic literacy amongst lay persons, and a sense of “ownership” over genomic data.
[I]t may not be useful right now but maybe in the future it could be useful again so they could have a … safe record of it themselves … Other people will request having copies of things and maybe … they’re just like why can’t I have my genomic file in my filing cabinet? [GC 3]
Aligned with possession of data is a sense of control over that data. As one bioinformatician noted:
I just know that power that you would probably feel knowing you’ve got that in case — you just never know, you might meet a bioinformatician and say “Hey, can you look at this for me?” You never know. Doubtful, but … [BI1]
Health professionals believed that participants should not require a basis for requesting return and stressed the reason should not be the determining factor in permitting access, but would help to guide the process for return. A clinician/researcher commented:
[I]f a patient or a parent requests so-called raw genomic data the answer, it’s yes … we should not basically try to be obstructive … I think for me the most important thing is to understand why they want it. [C/R 5]
Risks in Returning Raw Genomic Data
Almost two thirds of interviewees raised the prospect of feasibility constraints, in the sense of (i) participants’ support needs upon the return of data to protect those participants and their families against data misuse/misunderstanding, and/or (ii) resourcing constraints. This section canvasses risks identified by interviewees in facilitating safe return of data.
Risks to Participants/Families
While most health professionals were supportive of returning raw genomic data, an overriding theme was that such return could generate risks. The primary harm identified was that participant vulnerabilities could be exploited by so-called “experts” or TPI services purporting to provide accurate interpretive advice. Several interviewees had directly encountered this scenario.
I’ve got a couple of scenarios where people have chosen to obtain and analyse their own data through these websites, and in my experience most of the time … we’ve actually shown clinically that they don’t have that specific variant that was reported, there’s a high number of false positives particularly in the cancer genetics sort of sphere … patients have become highly anxious having received these results and then through confirmatory testing we’ve actually confirmed no, that was a false positive. [S/R 1]
Health professionals highlighted the importance of ensuring that participants understand the potential dangers of consulting these sites and other “experts.” The importance of verifying their legitimacy was considered critical: “I think we also have the responsibility to make sure they are not going … to “a so-called expert,’ but this is not even an expert, they just found someone on the internet.” [C/R 5]
One bioinformatician explained the expertise and context involved in the interpretive process:
[We’ve got the clinical background, we’ve got the context, we’ve got the clinician’s notes, we’ve got the histopathology, we’ve got … as much of the full picture as we can get that we then see things and we go oh well that’s nothing because of this or that, this is nothing because of this. Whereas you pass it to one of those service providers that then does these tests they don’t have that information. [BI 1]
Some interviewees were in favor of participants sending data to other known research groups but believed that this should be done using institutional mechanisms such as Data Transfer Agreements. Security of data transferred to others was also identified as an issue, particularly if there is a commercial aspect to the operations of a recipient. Data transferred outside strict institutional processes presented security risks:
[W]e have very defined clinical processes for when and how we share information about blood relatives … [t]he fundamental difference here is the second you put that information in the hands of a patient those very strict confidentiality processes could go by the wayside … [S/R1]
Also, data generated in a research context focuses on a specific pipeline, is not functionally validated, and in the case of tumor sequence data, is static. It was recognized that these shortcomings could in themselves be problematic for participants who do not understand the nature of the data.
Clinicians also reported instances where patients requested access to certain drugs that had been recommended by overseas companies purporting to provide analysis of raw data. One clinician emphasized the importance of NATA-accredited testing to inform treatment plans, and concerns regarding clinician willingness to supervise any drug combination recommendation sourced offshore: “[T]hey’d … have to find an oncologist who’d be willing to prescribe a lethal combination of drugs. I doubt if there’s an oncologist in Australia willing to do that.” [C/R3] Another clinician highlighted the opacity of these sites and the reports/treatment recommendations produced:
[I]t’s like, 5,000 dollars, and I’m like, “There’s no evidence for that. That wouldn’t change anything I do.” Several times, I’ve tried to look up the website on the company, and it’s just completely opaque what they’re actually doing, or whatnot … Because they’re commercial … A lot of times, they are websites that are targeting the patient. [C/R2]
Other interviewees were concerned about participants receiving recommendations for treatments that are unavailable in Australia for that participant’s clinical indication, or not available at a reasonable price. An interviewee in an advocacy role suggested this may guide bioinformatic analysis in determining the extent of the panel, and whether biomarkers identified include only those that are treatable or, more extensively, those that are not indicated for treatment [Adv3]. Clinicians highlighted similar issues:
Yes, and sometimes they’re not satisfied with me saying something like, well, I just can’t get you that target. And even if I could, I don’t know whether it will work in pancreatic cancer or in bowel cancer or whatever the cancer is, and that’s very dissatisfying … But the reality is that … we just don’t have that proof. So it’s really, really challenging. [C/R1]
One concern identified amongst interviewees involved in pediatric research was the issue of consent. As a genetic counsellor indicated, returning raw genomic data challenges a central tenet of clinical care:
[T]he core principles when I trained is that you wouldn’t even support carrier testing for children for recessive conditions because you encourage the parents to hold off until that child had the choice themselves around whether they want to know their status. [GC1].
This remains the case under current guidelines (eg Human Genetics Society of Australasia, 2023).37
Risks to Institution/Research Study
Potential risks to the institution or research study were identified by some interviewees. Tight controls over data are maintained while it is under the custodianship of a research study, and giving up control means the institution is unable to protect participants/relatives from ensuing harm, which in turn exposes the institution/study to potential harm. The primary concern here was the fact that research studies focus on a narrow, discrete pipeline, and tumor sequence data generates a static report that does not reflect the changing nature of disease and can be out of date within weeks [eg C/R 2]. Subsequent analyses using a different pipeline may report previously undiscovered variants in genes, or variants earlier classified as benign by the research study. One researcher stated that
… we have not gone and validated the performance, the accuracy, the precision, the sensitivity of 99 per cent of the genome. We’re focusing on the bits that relate to helping with their disease. [S/R5]
With particular reference to germline findings, a bioinformatician noted:
… if initially there is a finding that is considered not of much significance, and that’s the advice that we give the patient, future studies could find that this is actually a significant factor. Or vice versa. [S/R7]
This was identified to be a reputational risk that could potentially increase the obligation of research studies to engage in ongoing review of data (discussed further below). One clinician also highlighted the sometimes-high failure rate for sequencing where sample quality is poor, in which case a negative result may not really be a “negative result.” [C/R4]
One clinician/researcher referred to the:
moral, legal and ethical issue…if you cause someone to have a misadventure and they make an extremely disadvantageous decision because you gave them the raw data [C/R 3].
A further implication of participants looking for alternate treatment options is impact on the doctor-patient relationship:
I guess another thing that we have to remember is the whole interfering with clinical management with their treating doctor. That’s a contractual agreement between that patient and that doctor and I don’t know what the medical legal aspects are that we’re enabling potentially the discontinuity of that contractual agreement between those two parties. [S/R8]
Interviewees highlighted the institutional resources required upon handover of raw genomic data. It was difficult to reach consensus, but generally involvement of multiple personnel would be required, and the best estimate of the time required to be spent with a participant (generally by genetic counsellors) was one to two hours. The physical actions of reidentifying, locating, and transferring individual sequence data onto a USB (or similar) was not considered by interviewees to be overly difficult/onerous.
Interviewees also highlighted the health system resources needed should a participant seek further interpretation and require follow-up medical triaging (including appointments with general practitioners/specialists), confirmatory testing, and genetic counselling. Interviewees associated with research studies (scientists, oncologists, bioinformaticians) had also devoted time to counselling and reanalysis where participants sought clarification after seeking external interpretation.
A Process for Return
Many interviewees supported the development of a protocol, guidelines, or framework to guide and expedite the process of returning raw genomic data:
So we have to think about the speed of this process as well, so if we are going to hand this data over to the patients it needs to be done quickly, because handing that data over six months later after you’ve gone through however many hoops that we’ve jumped through is going to be meaningless, especially for high-risk kids. [S/R 6]
Some interviewees thought a generic protocol could serve a useful purpose, preferably endorsed by a body such as the Clinical Oncology Society of Australia or the Expert Advisory Group on Genomics Australia [S/R4], or the Medical Services Advisory Committee [C/R4]. Others believed that any protocol would need to be study- or institution-specific, generally built into the study from the outset; indeed, the next phase of the ZERO study has now adopted such a protocol.Reference Barlow-Stewart38
The notion that “support,” “explanation,” “education,” “counselling,” “caveats,” “checks and balances,” and “provisos” should accompany return of raw genomic data was ubiquitous. Interviewees stressed that education is key to explaining genomic data and “appropriate use” [GC1] to discourage handover of data to disreputable sources “and avoid … interfering with clinical management with their treating doctor as well.” [S/R8]
Interviewees reported that a critical aspect of providing adequate information was to ensure participants understand limitations of the data and the constraints researchers face in being bound by study protocols. The targeted nature of research projects means that particular genes constitute the focal point, particularly in germline sequencing:
So they’re fulfilling the requirements of that research study protocol, but it may not find anything that is of particular assistance to the individual participant. (Adv 3)
Interviewees engaged in the research process emphasized they are constrained by the scope of ethics approvals and “there is a lot we don’t look at …” [Pol/Admin3]:
[W]hen we interpret the germline data for these patients … we’re only allowed to look for cancer causing genes, … we don’t have ethics to look at everything in their germline … [S/R6]
This can sometimes have associated implications for clinical care; one clinician/researcher’s patient had undergone sequencing using a targeted panel which failed to find a relevant variant revealed by subsequent testing:
They went back and looked at their raw data, and said yes, this variant was actually above the calling threshold, however it was not called because it wasn’t one of the bioinformatic program canonical variants … it’s just such a fine-line right now between the clinical and the research setting … all I really want to know is the variant, maybe the variant allelic frequency, so I can see how abundant it is, and you know, I don’t really want much more information. [C/R 4]
In implementing a protocol/guideline, health professionals felt that there was scope to educate and empower participants through “plain English materials” such as a letter, information sheet, or booklet, either on paper or online. Others suggested a video might cater to diverse populations. According to interviewees, information should inform participants regarding the nature of both germline and somatic sequence data and what it reveals about an individual/their relatives, the scope and implications of testing undertaken during research, the role of bioinformatic analysis, the limitations of the data and/or analysis, and the utility/pitfalls of seeking further interpretation.
[Y]ou could create a letter that says … please find attached your raw genomic data. This relates to a biopsy taken from this lesion on this day. There are several limitations of genomic data. There is intra- and inter-tumoral variability. There are errors in the creation of genomic data. There are … not always targetable treatments available for the genomic data,… we could list all of the limitations. In light of all the above limitations, we suggest that any interpretation of this data be done with an expert both in cancer genomics and in … cancer treatments. [C/R1]
There were differences of opinion as to how comprehensive the support provided to participants should be. Interviewees suggested that scientists, bioinformaticians, oncologists, and genetic counsellors (for germline data) could be involved in the process:
I’m really talking about making sure that the patient has a clear line, obviously, to genetic counselling and also the treating clinician, so that they have both the medical and the scientific aspects really conveyed to them … But I think it really does come down to the genetic counselling … [S/R7]
This raised questions around resourcing and the potential to incorporate provision for return of raw genomic data in funding requests and study design. Reference to “up-fronting” the resourcing issue was mentioned by a number of health and policy professionals.
Virtually all interviewees emphasized the importance of a conversation with participants requesting raw genomic data, although there was variance on what form this might take. For example:
I feel like it’s almost like not so much just returning the raw data to them directly, I feel like the conversation or the workflow should really be more around exploring what they want to do with it and actually helping or working with them to do that…[F]rom a workflow perspective, yes it’s easier to simply say here you go, USB stick, you know, there’s your 6 million base pairs of DNA in a sequence. But I just don’t know if that’s going to be that helpful for the person. [R/S1]
Reduction in risks featured as an important premise, with interviewees stressing that “[T]here is probably some element of protection that would be required for both parties and just outlining the limitations and context of the file.” [GC3]
In ensuring that processes put in place are construed as supportive rather than obstructive, it is worth bearing in mind the interpretation that patients/families may place on being questioned about their intentions regarding use of the data. One parent observed:
I would probably interpret that, coming from a corporate background, as they want to keep this data. Maybe it’s very expensive to create this data and they don’t want it to be released for other parties who haven’t put in their resources to interpret it. Rightly or wrongly, I don’t know, but that’s probably how I would interpret it, because that’s how the business world often works. Then if people were asking me questions about why do you need it, I would feel that that was a way of blocking it. [Pat/Fam6]
Discussion
Researchers involved in genomic research studies are encountering increasing numbers of requests from participants for access to their raw genomic data. Ethical guidance on whether, when, and how raw genomic data should be returned must be informed by expert stakeholder opinion. Stakeholder views will not provide a comprehensive body of evidence to address the ethical issues inherent in decisions to return raw data but are a critical conduit to understanding the diverse viewpoints that must be considered in formulating policy guidance. We note that patients formed a relatively small proportion of our study cohort (< 25%), and that the views of patients are likely to differ significantly from those of professionals on the other side of the therapeutic alliance. This imbalance was corrected to an extent by the inclusion of patient advocates; taken together, patients and patient advocates comprised close to 40% of the study cohort. Throughout the following discussion we highlight where the views of patients/advocates diverged substantially from those of health professionals.
Legal or Ethical Right to Return
The first question guiding this research was “what minimum legal or ethical rights should study participants have to seek return of their raw genomic data?” Legislative regimes in multiple jurisdictions appear to include provision for research participants to request access to their raw genomic data,39 albeit indirectly via rights existing under respective privacy and data protection regimes. However, the legal position varies across and within jurisdictions, and existing legal duties to return raw genomic data are, not surprisingly, unfamiliar to research participants and researchers.
Australian research guidelines are also equivocal, compounding uncertainties around researchers’ obligations. These intricacies in Australian law and inconsistencies with ethical guidelines may not be well understood by members of Human Research Ethics Committees when considering study protocols.Reference Pysar40 Regardless, there are compelling arguments for providing participants with access to their raw data, including fundamental respect for participant autonomy and decision-making,41 the facilitation of further interpretation of data,42 and the enhancement of reciprocity and solidarity — stimulating public willingness to engage in research and ensuring participants perceive personal and social benefits in involvement.43
Our stakeholders strongly supported the return of raw genomic data to study participants. Interviewees articulated that patients able to access their data are likely to experience a sense of agency and may seek further analysis of data if desired. Return of data fosters a sense of reciprocity for participants and for institutions undertaking research, particularly for those participants for whom involvement has not yielded clinically beneficial outcomes. Our data indicates that although not every participant or family member expressed a desire to receive their raw genomic data, health professionals interviewed are clearly motivated to help patients where feasible. This includes providing access to raw genomic data where sought.
It is evident that many patients and families take research “into their hands” and view raw data as a conduit to allow them to maintain some control over their health journey. Facilitating return of data is an important aspect of promoting consumer-led health initiatives, hence we recommend a right of return be made clearer under relevant research ethics guidelines and be built into individual study protocols.
Safeguards for Return and Implementing a Process for Return
The second major question examined in this study was “What are the minimum safeguards for exercising this right, and how should they be incorporated in a process for return?” Almost two decades ago, McGuire et al provided recommendations for return of genomic data in the WGS research context, calling for individual projects to develop a data return and counselling policy.Reference McGuire44 The growth of WGS as the standard genomic sequencing technology has made this recommendation more pertinent, as WGS generates extensive data with potential clinical utility.45 As a result, the generation of project-specific policies detailing processes for return has become increasingly important.Reference May46 Development of generic guidance materials or protocols is also worthy of consideration to assist the development of policies tailored to individual projects.47 Not every interviewee advocated that a generic protocol for return of raw data would be useful; some thought that the distinctiveness of requests for return and individual research studies necessitated case-by-case treatment. However, we conclude that at the very least, a study-specific protocol provides a useful point of guidance in grappling with requests by individuals who may have varying reasons for making a request.
Our stakeholders recognized multiple issues in returning raw genomic data that should be addressed in a protocol for return. They confirmed the importance of being cognizant of issues surrounding return of raw genomic data at commencement of research studies utilizing WGS and building these into study design, emphasizing the harms that can arise if appropriate processes for return are not adopted. We focus here on four aspects of our reported findings that are worthy of further discussion.
Equity and the Right to Access Raw Genomic Data
First, our stakeholders addressed the desirability of building automatic return of raw data into study protocols versus returning it to individual participants on request. Significantly, the legal rights of Australian research participants to access their raw genomic data do not require research organizations to preemptively offer up raw genomic data to participants; their exercise generally requires a proactive request from participants, generally contingent on participant knowledge that the data are available to access. While some participants may have sufficient genomic literacy to seek further information about their genomic information, many will not. This introduces a barrier to access for participants with lower levels of educational attainment and genomic literacy, those who do not know that raw data is available and how it might be usefully deployed, and those who may not think to ask for return of results due to the fact they are going through a difficult clinical journey. On the other hand, “opening up” access may lead to a significantly greater number of requests.
This issue of equity in access to results is an issue that has not been addressed to any degree in the published literature. Some scholars have proposed building in a default right of access on the basis that genomic sequencing initiatives have a duty to return results,Reference Angrist48 but a majority of studies do not currently provide such a default right. Some commentators advocate dealing with requests on a case-by-case basis.49 Empirical studies have demonstrated that there are very few (if any) circumstances where requests by participants are refused,50 and our interviewees confirmed this to be the case in their experience. Others have concluded that there are a limited number of feasible methods of returning data, and that the technological requirements inherent in these methods (at a minimum, broadband connection, laptop or desktop computer, capacity to access cloud-based infrastructure) are not accessible to all, setting up disparities across participant cohorts.Reference Crawford51
Our participants were mixed in their views on whether a default right of access should be built into processes for return. Even patients and patient advocates were not universally supportive of a default right. Issues of equity and inequality may arise in circumstances where participants are unaware of their legal rights to access or the nature of genomic data.Reference Khoury52 Use of research funds to provide access to raw data only to those who request it may be viewed as inequitable, preferencing those who are genomically literate, and who know that raw data is available and how it might be usefully deployed.
Our stakeholders also emphasized hazards in research organizations proactively offering raw genomic data to research participants, in that this may give a heightened impression that this data will provide useful health information. If this data is to be offered to all participants in the interests of equity and to address differentials in health literacy, it would need to be accompanied by careful communications explaining these limitations, and that it is being provided in the interests of fairness. Although different considerations are likely to apply in different studies, on a generic level, a plausible way forward is to ensure participants have some fundamental awareness of the nature of the data generated from their biological samples, and that this data is accessible upon request.
Support and Mitigating Risks
Second, a strong theme among our stakeholders was that raw genomic data should not be released to participants without sufficient preplanned support to counter risks associated with data release. These significant potential risks in permitting participants to access their raw genomic data have been well canvassed in the literature and include security and privacy risks to donors53 and their close genetic relatives,54 unrealistic expectations as to the utility of data,55 and misinterpretation of variants through secondary use,56 including through utilization of TPI services.57 The potential re-identifiability of raw genomic data, particularly if coupled with phenotypic data, underpins this concern.
Risks associated with security of data and associated privacy concerns were identified as concerning by our stakeholders. Again, our contention is that development of a standardized process that carefully considers these issues will assist in counteracting these risks and ensure secure data transfer and strict messaging about maintenance of privacy in respect to data files. The most careful consideration should be given to the impact of release of raw data in two situations. One relates to pediatric participants, where a parent seeks to access their child’s data on their behalf. In such cases, although return is not precluded,Reference Swaminathan58 greater complexities in this situation means that further protections must be built into processes for return. Data custodians must be certain that the request for access is in the best interests of the child, and that the child’s future health and well-being would not be adversely impacted by release.Reference Horris59 The other relates to release of raw genomic data that may impact genetic relatives. Many interviewees noted privacy risks to blood relatives inherent in the release of germline data. This challenge can be managed in a similar manner to the return of WGS findings more generally by ensuring that transparent guidance clearly explaining issues around privacy and implications for genetic relatives is provided.Reference Budin-Ljosne60 Limited privacy protection under current laws for relatives means that clear communication to the participant of the possible impact for relatives of on-sharing of data is paramount.
Misinterpretation of the data was another important factor for stakeholders, and one that was particularly cogent in prompting comments by interviewees that the return of raw data must be undertaken with adequate support processes in place. The basis of this perceived risk is participants’ (and in some cases clinicians’) lack of capacity to understand complex genomic concepts, and thus the resulting data. This gives rise to vulnerabilities that could lead to harm. Our evidence also underscores the significant concerns around TPI services and others providing erroneous advice by reporting variants that patients do not have or reporting on non-pathogenic variants. Particular concerns were identified in relation to somatic (tumor) data. A further concern voiced was the provision of advice from outside the research study recommending treatment regimes or drugs that are either contrary to those previously indicated (sometimes outside standard of care treatments), or unavailable. This highlights that experiential learning tailored to specialization is important for education of medical specialists in genomic medicine.Reference McClaren61 Our evidence does reveal that molecular oncologists engaged with the ZERO and APGI programs are well versed in interpreting and considering the implications of targeted treatment recommendations arising from genomic testing.
A subsidiary concern was the reputational risk for the research study should variants not forming part of the study pipeline subsequently be discovered. However, provided the research endeavor itself is not jeopardized,62 such concerns should not ground denial of access and can be dealt with via clear advice to a participant.
Effective Communication Through the Process of Returning Raw Genomic Data
Third, interviewees emphasized the importance of communication and education, because learning why the participant wants to access the raw data enables a research team to provide education to provide an informed, autonomous choice. This theme was prevalent particularly among health professionals: some felt that exploring reasons for release would assist in providing this support; patients and patient advocates considered the reasons for desiring access less relevant. An approach that supports the provision of general information to accompany the return of raw genomic data has been previously propounded,Reference Winkler and Knoppers63 and would address concerns related to knowledge imbalance and risks associated with security of data/privacy and future misinterpretation of data.
While almost unanimously of the opinion that raw genomic data should be returned if requested, stakeholders’ comments on what information should form part of that return were more nuanced. Interviewees considered that project teams should explore the reasons for requests with participants, the fundamental basis for this being that it would facilitate the provision of information and education. Our conclusion, however, is that a standard protocol provides a useful starting point for having these conversations and that although participants may have varying reasons for wanting to access their raw data, the processes for returning it are likely to take a comparable path.
The Issue of Resourcing
Fourth, our stakeholders emphasized the resourcing implications of providing access to raw genomic data and the lack of availability of staff to guide participants through the process of data return. Genetic counsellors are in short supply and high demand,Reference Dragojlovic64 as are molecular oncologists. Their time is a finite resource and excessive time spent with one participant removes them from time on other clinical or research priorities. Scholars have also pointed to this consequence as a nontrivial outcome of providing access to raw data that may threaten the viability of research studies.65 This concern was reinforced by those we interviewed.
Our recommendation here is that the adoption of standard processes/protocols to guide the process of return will assist in managing these concerns. Carefully planned and coordinated systems, with accompanying materials designed during the planning phase and incorporated into research studies, will have the effect of limiting workload associated with ad hoc data return requests and permitting more efficient resource allocation. Front-loading this work will also ensure adoption of coordinated workflows and technical/logistical procedures. We suggest that participants receiving access to their raw genomic data do not require further interpretation of raw data or extensive genetic counselling66 provided they are given adequately informative preprepared materials.
A smaller number of interviewees (including patient advocates) discussed the viability of “flipping the model on its head” and storing data centrally with inbuilt patient access. On the assumption that donor empowerment is key to equitable notions of reciprocity and equity, such a model would alleviate many issues around resourcing return of data and promote patient-led sharing. It might remove some perceived risks of returning data to participants, minimize duplication of research effort, and guarantee data quality, but also raise challenges in terms of interoperability and conformity with ethics protocols. Some research initiatives currently under development are explicitly embracing this model (eg Count Me In study).67 It is a radical departure from the current research model but has potential value in terms of expanding use of research-derived germline data.
Limitations/Future Research
While this study strongly supports the position that participants’ raw genomic data should be returned, a key consideration going forward will be structuring processes for return of raw genomic data. Interviewees were divided on whether raw genomic data should be automatically returned or available only on request. Clearly, the development of return processes hinges on this question and future work might further explore which strategy is preferable. Processes for return must incorporate provision for participant support and resourcing, and further research might also investigate whether building in funding to facilitate return should practicably be undertaken at study-, institutional- or sector-level.
Conclusion
There remains significant ambiguity around whether an obligation to return raw genomic data exists, but our evidence supports a conclusion that research participants are entitled to access their raw genomic data. Practical considerations guided our stakeholders’ opinions on precisely when raw genomic data should be returned, in that a majority considered return should be predicated on a request, rather than returned preemptively. Nevertheless, we recommend that some communication to study participants that raw genomic data may be accessed would promote equity. We revealed strong support for a protocol/guidelines to steer policies around return of raw genomic data, given the widely held view that accompanying support structures are essential. There are feasibility and resourcing implications for research studies prepared to return raw genomic data, underscoring the need to plan the process of return early in the research project lifecycle.
Acknowledgements
The authors thank the editors and reviewers for their invaluable feedback on drafts of this article.
Disclosures
CEW is supported by the National Health and Medical Research Council of Australia (APP2008300).
Funding Statement
This project was funded by the Medical Research Futures Fund (Genomics Health Futures Mission) N76758.
Open access
