Introduction

Genome sequencing (GS) has moved into clinical practice as an effective diagnostic tool for rare Mendelian disorders.1 While successfully identifying variants that explain the patient’s phenotype for about 25% of those referred for testing,2 this diagnostic testing also produces results of which the clinical interpretation remains uncertain. The occurrence of uncertain results has become increasingly prevalent with the broader use of GS and the amount of information it generates.3, 4 A recent taxonomy delineates the possible sources and ramifications of uncertainty,5 but empirical work on its production, interpretation, communication, and impact on patients and clinicians is just beginning.3, 6, 7, 8

A critical focus is to characterize ways in which clinicians interpret uncertainty arising from GS, how they communicate it to patients, and how patients understand and respond to it. Some have questioned the value of reporting uncertain results owing to concerns about misinterpretation that could lead to psychosocial and other harms from downstream effects such as unnecessary interventions.9, 10, 11 Proponents who argue for reporting uncertain results contend that this information holds potential value in the future, that nondisclosure can also result in harms, and that harms can be mitigated by appropriate pre- and posttest genetic counseling.3, 12, 13, 14 While ambiguous results are not new in medicine,15, 16 the potential for uncertain results and any accompanying harms is much greater in GS owing to the scale of the assay. A recent review of seven genomic sequencing studies of patients with suspected genetic disease reported that, across these studies, 40% learned an uncertain result (range 8.6–51.8%).2 Key concerns remain about which uncertain results should be returned and how to communicate the inherent complexities to patients in ways that reduce potential misunderstandings and harms.17

Thus the expanded use of genome-scale sequencing as a diagnostic tool will inevitably lead to an increasing need to communicate uncertain results. Here we present data from an ethnographic study, embedded in a clinical sequencing project, that observed clinical geneticists’ communications about uncertain diagnostic results and interviewed patient/participants about the responses they gave on learning them. Findings from this study can inform the debate surrounding the impact of disclosing uncertain genomic information to individuals having diagnostic genome-scale sequencing.

Materials and methods

To provide empirical data on clinicians’ communications and patient/participants’ subsequent responses, we recruited 21 adult patients and 11 parents of child patients (hence referred to as “participants”) who learned uncertain results from exome sequencing (ES) conducted by the North Carolina Clinical Genomic Evaluation by NextGen Exome Sequencing (NCGENES; University of North Carolina, Chapel Hill, NC) study. NCGENES studied the diagnostic utility of ES in a clinically diverse set of patients suspected to have an underlying genetic etiology for their condition but in whom no diagnosis had been reached through the usual clinical care. NCGENES results were filtered by a diagnostic list, or lists, of genes selected based on each patient’s phenotype. An expert committee made up of clinical geneticists, genetic counselors, clinical laboratorians and PhD-level research scientists decided as a group whether the result to be disclosed should be categorized as diagnostic, possibly diagnostic/uncertain, or negative. Uncertain results were classified based on whether the case was that there was a variant of uncertain significance, an uncertain phase of two pathogenic variants (i.e., in cis or trans), an uncertain or incomplete phenotypic fit for the participant’s condition, or another reason or reasons for uncertainty. Diagnostic and possibly diagnostic/uncertain results were confirmed by Sanger sequencing analysis in the UNC CLIA-certified Molecular Genetics Laboratory and a clinical report was issued.

The 32 participants in this ethnographic study were purposively selected as far as possible to reflect the different disease cohorts and sociodemographic characteristics (gender, age, race/ethnicity, education) of the 169 NCGENES participants who were given uncertain diagnostic results (25.3% of the total 669 enrolled). Participants provided written consent for an ethnographer to observe the study visit at which NCGENES clinical geneticists and genetic counselors disclosed results, and for telephone interviews 4 weeks and 1 year after disclosure. The second interview was conducted because many participants had not yet been given the results from their family testing, nor had they met with a recommended specialist at the time of the first interview. All 32 participants completed the first interview, and 27 (18/21 adults and 9/11 parents) completed the second interview. Of the five participants not re-interviewed, two were lost to recontact, one declined, and two were unable to do the interview owing to illness. Interviews were semistructured and designed to elicit participants’ (i) understandings of their uncertain result, (ii) concerns or psychosocial responses related to it, and (iii) plans of action or accounts of actions they had taken based on the result, especially regarding recommendations for further testing or consultations. Study visits and interviews each lasted approximately one hour, and were audio recorded and transcribed verbatim. The university institutional review board approved all procedures and protocols.

The ethnographers (D.S. and K.A.R.) conducted a systematic content analysis of the observation and interview transcripts, employing data display matrices to summarize, categorize (i.e., code), and systematically compare and link findings within and across the 32 cases.18 We compared the content of clinician communications across all cases for commonalities and variations, and linked these to participants’ responses as recounted in interviews, noting especially both congruencies and differences between what clinicians communicated and what participants understood, as well as participants’ psychosocial and behavioral responses to their uncertain results.

Results

Participant characteristics

The sample of 32 participants included 21 affected adults and 11 parents of affected children. Sociodemographics of the sample are depicted in Table 1. Both parent and adult participants were predominately non-Hispanic white, with adults being more diverse by condition, gender and age. There was some diversity by education, with 56% having a college degree or higher Importantly, 78% reported having previous genetic testing with no molecular diagnosis.

Table 1 Sociodemographic characteristics of interview participants
Full size table

Congruence between clinicians’ communications and participants’ understandings

NCGENES clinicians had the task of interpreting the clinical reports of uncertain diagnostic results to participants. It is important to note that not all uncertain results were variants of uncertain significance (VUS). In their communications with participants, clinicians attempted to clarify the nature of the uncertainty, the reasons the result was considered an uncertain but possible answer for the condition, and the implications of this uncertainty for participants and their relatives.

A clinical geneticist and a genetic counselor met with each participant to discuss diagnostic results. Although the eight NCGENES clinicians who disclosed results differed stylistically, the order and content of their communications conformed to the same general pattern. This was largely due to their similar training as clinical geneticists and genetic counselors and their close collaboration as a research team since the beginning of NCGENES in 2011. During the project, they had frequent discussions at project and molecular sign-out meetings about difficult cases, leading to a consensus about the most effective ways to communicate different types of results. In cases in which uncertain results were disclosed, clinicians followed their description of the genetic result with an assessment of the degree to which they believed the variant was a plausible explanation for the patient’s symptoms. They described sources of uncertainty, including limitations of the technology, incomplete databases, and rudimentary knowledge of the contribution of pathogenic variants to disease. Clinicians often elicited additional information about personal or family medical history to better evaluate if these features matched the phenotype expected based on the reported genotype. In some cases, they recommended family testing or other clinical evaluations that could potentially provide additional evidence regarding pathogenicity of the variant. After summarizing key points and addressing questions, clinicians ended by reiterating that reanalysis and advancements in the field could potentially provide a more certain answer in the future. Thus, the premise they communicated was that an uncertain result should not be viewed as a final result but one requiring more information and research. They assured participants that they would continue to search for an answer and contact them with any new information.

Our analysis of participants’ understandings of these communications indicates remarkable congruence in almost all cases. Participants demonstrated a wide range of ability to recall specific details with a few participants having only a superficial understanding, but with one exception (discussed below), they understood that the result was uncertain and not a definitive answer. They regarded the uncertainty of the result in ways that were similar to how it was communicated: as being uncertain but likely, uncertain but possible, or uncertain but an unlikely explanation of their symptoms. Table 2 provides representative examples that illustrate this congruence.

Table 2 Congruency between clinician communication and participant understanding of an uncertain result
Full size table

For the majority of the 32 participants, clinicians presented the result as an uncertain but possible answer. For example, a genetic counselor began the study visit with an older male with neuropathy by saying: “We do have something potentially to tell you about or at least to puzzle you with…we did find a genetic change in a gene that we’re suspicious about. It’s never been seen, and so we call it a variant of uncertain significance.” She then described symptoms associated with changes in the gene (BSCL2), the variability of the phenotype, and the lack of knowledge about whether or not symptoms of neuropathy were associated with variants in the portion of the gene where the variant was located. She ended this explanation, stating,

It’s enough to make us say, “Well, it’s definitely uncertain in its significance.” It might have absolutely nothing to do with your neuropathy. One of the ways we can potentially find out about that and try to learn a little bit more is to see if this gene change occurs in individuals in your family who have the symptoms.

In the follow-up interview the participant, who had a science background, said he had learned that they had found “an undiscovered variant in my genome.” He did not remember the name of the gene, but understood that it was an uncertain result for which family testing might provide more information.

For five participants, clinicians presented the result as an uncertain but likely cause, as in the case of a girl who had kidney problems and skeletal dysplasia. The clinical geneticist began the study visit with the father: “We actually did find something we think explains what is going on.” He added that it was a rare change, not reported before, and thus there was insufficient evidence to prove its pathogenicity but that he was reasonably confident it was the answer “because of what we know about this gene [WDR19] and because of what we know about [your daughter] and how well that those match together. That makes it much more likely this is the cause.” The father reported in the first interview that he and his wife believed this result was very likely the answer for their daughter’s condition, and they were looking for more information on the gene and associated syndromes. As this case suggests, many participants engaged in further activities that could shed light on their results, such as searching the internet for additional phenotypic aspects that might or might not match their own experience. These efforts were encouraged by NCGENES clinicians, who told participants that they were “a part of the team” and welcomed participants to inform them of any new information that could potentially help resolve uncertainty.

At the time of the first interview, only one participant had imbued an uncertain result with a more definitive interpretation than the one which the clinicians presented. Clinicians reported a novel heterozygous missense variant in SLC1A3 as a possible but uncertain cause of the adult participant’s seizure-like episodes. However, the participant talked about it as “a life-changing result”:

Even if there is no treatment for it, it’s extremely significant to me because it’s redemptive after all these years of not knowing. I’ve had a lot of good doctors, but I’ve had a lot of just horrible treatment by doctors too, and just saying, “Wow. This is what I have.” Just having a name to something when you’ve suffered for so long is just an amazing relief …it’s been really redemptive with all that I’ve gone through.

This case is a useful reminder that even when clinicians stress the uncertain nature of a result, participants make meaning of diagnostic results in light of their own experiences and their need for an explanation. Yet this participant’s understanding of her result changed after family testing showed her unaffected father had the same variant, making it much less likely that it was causative. In the second interview, she interpreted the result more in line with the clinicians, saying, “It is still the case there is no answer.” As this example indicates, both clinician and participant interpretations can change because of new information such as results from family testing, additional clinical evaluations, and reanalysis and reinterpretation using improved databases.

Behavioral and psychosocial responses

NCGENES participants could give or withhold their consent to the placing of clinically confirmed diagnostic results into their electronic medical record (EMR). Almost all of the participants enrolled in NCGENES elected for results to be placed there. Of the 32 participants in the ethnographic study with uncertain results, 18 of 21 adults and all 11 parents consented to placement. Of the 3 who declined, one was concerned about insurance discrimination, another about misinterpretation, and the third saw no advantage to having it in his EMR. Three-fourths of those who consented did so primarily because they perceived that the result might be valuable to their doctors now and/or in the future. Five participants specifically cited that a desire to have the result documented and accessible to their physicians was the key reason for consenting.

Although 38% saw no disadvantages to this placement, 40% said they were aware of possible insurance discrimination but did not believe it would happen to them, or said that the result would not increase the risk because their disorder was already detailed in their EMR. Others mentioned the possibility of job discrimination or “leaks” of the information, but thought the advantages of having the result in the EMR outweighed potential disadvantages. With two exceptions, participants said they had informed their specialists and/or primary care providers about the result, and had sought their doctors’ opinions about whether the result fit with their or their child’s condition and what implications it might have for care. Two of the three individuals who did not put the result in their EMR nonetheless reported talking with their doctors about it.

Family testing was recommended for 8 children and 12 adults, to help with interpretation of uncertain results. At least 1 parent of the 8 children, but only 5 of the 12 adults, completed this testing. This difference was in part due to convenience, as parents could provide consent and a sample at the time the results were disclosed, but could also be attributed to their expressed interest in finding an answer. Adults faced more barriers to pursuing family testing. Participants gave a variety of reasons for not following up on family testing. Some thought it unlikely that such testing would result in the acquisition of any significant information, or “it fell off the radar”, or they did not want to pressure family members. Others were keen to pursue it, but their relatives refused to be tested, or they were not able to have blood drawn owing to illness or other problems.

Those who pursued family testing and/or consultation with specialists said they were not worried while waiting for any additional results. Several parents explicitly stated that there was no reason to be concerned until they knew the outcome of the testing because they had been through this process before. For example, one mother learned that her son had a variant that could indicate Coffin-Siris syndrome, but she was awaiting the results of parental testing and a consultation with her pediatric geneticist before coming to any conclusions. She commented that uncertainty was nothing new; her son had previously received an uncertain result from a microarray analysis.

It is important to note that, although a few voiced disappointment at not getting a definitive molecular diagnosis, only one of the participants expressed distress related to the uncertain result. The majority were not concerned with the uncertainty surrounding the result, but rather with the challenge of not getting a definitive diagnosis, and therefore remaining uncertain about the progression of the disease and the risks for their children. Participants were aware of and accepted an uncertain result as a possible outcome of study participation and, except in those few cases where it was returned as a likely answer, they appropriately viewed it as “not an answer,” “inconclusive,” “an unsolved mystery,” or “not very significant.” No one expressed regret at learning the uncertain result and, instead, many noted its value and potential usefulness in the future. It was a “piece of the puzzle,” something “to file away” until more became known. Most expressed sentiments that were similar to this adult who learned an inconclusive result: “They can return anything they want to me. Where I would be disappointed is if they didn’t return it. What about in seven years if they find something?”

This potential of genomics to find answers in the future was a significant factor in shaping participants’ valuing of and reactions to the uncertain result. They were optimistic about future answers and valued their ability to contribute to the research that could lead to these advancements. They recognized that even if the uncertain result was not informative to them now, the information could help build data- and knowledge bases. One participant summed up this widely-shared view of the promise of genomics:

Once exome sequencing becomes more available I think that a lot of other people or a fair number of people will be diagnosed with it, and then when there’s more people diagnosed with it, then there will be more money put into research for not only what I have, but for things that affect other people because then more money will be spent for research, and treatments will be developed, and there will be more studies done.

In summary, participants understood uncertain results in ways that were congruent with the NCGENES clinicians’ communications. They followed the clinicians’ recommendations for further consultation, although family testing that might have resolved the uncertainty was not always carried out. Participants were prepared for an uncertain or uninformative result, and after learning it, did not act in ways that caused harm. They did not express regret for having learned the uncertain result; rather, most regarded it as potentially valuable in the future.

Discussion

In this study, we found that that there was general congruence between how clinicians explained an uncertain result and participants’ interpretations of this uncertainty. Two characteristics of participants in this study probably played a large role in this high level of congruence. First, the majority (78%) reported having had previous genetic testing that had not revealed a molecular diagnosis for their or their child’s condition, leaving them uncertain about its etiology and family implications. Participants were thus accustomed to ambiguity, as they had already lived with and experienced uncertainties related to their own or their child’s medical conditions. Second, during the informed-consent process, genetic counselors specifically addressed the likelihood of obtaining uninformative or uncertain results, thus tempering participants’ expectations of getting a definitive answer (details on this process are provided in Supplemental Information online). Others have advocated that managing participants’ expectations about results in this way should be generally recommended as best practice.19

That participants’ interpretations largely coincided with clinicians’ communications may seem contrary to what has appeared in the existing literature about how patients misunderstand genetic information,20, 21 but may be explained by our specific focus on participants’ understanding of the significance of their result and what actions, if any, they should take. Although some participants’ comprehension of other genetic information, such as genetic risk and inheritance patterns, may not be precisely congruent with that of clinical geneticists, they grasped the key concept that the diagnostic result was uncertain and should not be used to guide health-care or other decisions.

Uncertain medical results are often viewed as harmful, but this uncertainty can hold hope and promise.22 In NCGENES, genetic counselors prepared participants at the time of enrollment for the likelihood that their diagnostic results would be uncertain or uninformative.19 When results were disclosed, clinicians not only explained the possible reasons for and sources of the uncertainty, they also stressed that, by disclosing variants with uncertain meaning, they were enabling those variants to be reconsidered in future interpretations. As found in a sociological examination of the central role of potentiality for NCGENES participants who learned negative results6 as well as in other studies,3, 7, 23, 24 such explicit communication helped participants contextualize the uncertain result and mitigated adverse responses to it, while promoting optimism for potential certainty in the future.

In addition to the clinicians’ nuanced explanations about an uncertain result, this framing results in a future promise of genomic discovery was found to be key to participants’ understanding.25 NCGENES clinicians promised to keep looking for the answer, adding new genes to diagnostic lists, and reanalyzing novel variants. Most participants referenced this promise in their interviews. They understood that uncertainty was part and parcel of this new enterprise, driving the research that might resolve uncertainty in the future. Some participants may not have had a realistic timeline of when science could produce those answers, but hoped that their children and others would benefit if answers did not come in time for them. They viewed the continued efforts of genomic researchers and clinicians as the best chance of finding a definitive genetic diagnosis, and took seriously the clinicians’ promise to “keep looking” for the answer.

Findings from this study indicate that uncertainty in diagnostic GS results is accepted and understood by patients when carefully and adequately explained. Concerns have been raised about the potential harms of returning uncertain results from GS4; however, genomic medicine is not exceptional in this regard. Much as in other fields of medicine, the biggest risk of returning an uncertain result is likely to be its misinterpretation as a true positive result and subsequent downstream consequences, such as risks of overtreatment. We would argue that greater harm is caused when laboratories overcall variants of uncertain significance as “likely pathogenic” (classifying them as “positive” results) than in characterizing a variant as uncertain and disclosing it appropriately to the patient. However, there is also the concomitant risk that uncertain diagnostic results will be interpreted as negative and be ignored, thus preventing appropriate clinical action.26 It is important in large-scale GS to convey to participants the potential for uncertain results and to clarify that these are neither positive nor negative, but will remain uncertain until more knowledge is attained.

There are limitations to the current study. Findings may not be generalizable to other clinical or research settings that differ by population characteristics, the level of participant education, and clinical communication practices. The participants in this study had already undergone diagnostic odysseys, with the majority having exposure to the genetic testing process, and many had previously experienced uninformative or uncertain results. Their expectations of definitive results had been tempered by the NCGENES consent and pretest counseling process, and by clinicians’ contextualization of the uncertain result. Participants’ optimism about future answers could reflect a bias specific to research participants and may not be representative of patients undergoing clinical GS. However, even clinical GS candidates are likely to have had similar experiences. Longitudinal studies in other contexts are needed to determine if and when harms arise from uncertain diagnostic results, and the factors that amplify or ameliorate such harms.

In spite of these limitations, the current study suggests that uncertain diagnostic results from GS can be relayed to patients in ways they can understand and consistent with providers’ interpretations, without causing undue harm.