Abstract
Due to heritability, next-generation genetic tests have the potential to affect family members beyond the patient being tested. Geneticists and genetic counselors, in dialogue with patients and their relatives, will need to establish for whom and in what way genomic testing results matter during the communication of testing results, indicating the spillover of presumed pathological variants. On the basis of video-recorded consultations of the return of exome results in a genetics clinic, we distinguish three different logics deployed to explain the relevance of the findings for the patient, extended family members, and unborn relatives. While geneticists tend to be cautious in interpreting findings for the patient and living relatives, the findings become more deterministic in the context of reproductive decision making. The presentation of results then establishes the causal role of variants and reflects back on disability as a state to be prevented, in the process establishing genetic ties between kin.
Genomic testing is a family affair. The proliferation of next-generation sequencing has changed the landscape of medicine in myriad ways ranging from susceptibility testing for conditions such as breast cancer, put in the spotlight by Angelina Jolie, to diagnostic testing to explain clinical symptoms. Due to heritability, the impact of a genomic test for an individual has the potential to spill over to other people. Parents who thought they were receiving a diagnosis for a symptomatic child may have to manage a perceived genetic vulnerability across their kin network or reconsider reproductive decision making. Social researchers have noted that families receiving genetic testing results incorporate these results into their preexisting notions of what symptoms and diseases run in their families (Freese and Shostak 2009:116–17). Thus, certain results confirm suspicions and become socially and medically actionable, while other results are filtered out (Rapp 2000).
Geneticists and genetic counselors establish the social and clinical relevance of genomic testing results when they inform patients of sequencing results (Pilnick 2002a). During this consultation, clinical staff and families delineate the different parties implicated by genomic findings, ranging from the patient to siblings, extended family, and future not-yet-conceived siblings. These parties do not hold equivalent stakes in the results: the test may reveal a diagnosis that the relative was unaware of or suggest an increased risk for future health issues. The opportunity to act on this information also varies: it may suggest treatments, may lead to prevention, or may be ignored. Genomic testing then renders living relatives and as-yet-unborn relatives differently genetically salient. This raises the research question of how families and geneticists in the clinic extend the same genomic testing result to living and future relatives.
In this article, we compare geneticists’ and parents’ perspectives of when and how results spill over from patients, to relatives, to future children. We argue that patients and clinicians modulate their logics about testing for different kin. Clinicians use the patient’s phenotype (symptoms) as a prerequisite for testing and thus treat the identification of a genetic cause as an end point in the patient’s search for a definitive diagnosis. In contrast, patients generally treat the identified genetic cause as a vulnerability that extends across the kin network, which then warrants testing of other family members, including prenatal testing for unborn children. Clinicians consider these relatives differently. Clinicians treat the logic that extended family should be tested as unjustified without a problematic phenotype. Yet, for unborn children, clinicians instead reason that the opportunity to prevent disability justifies testing, even if the test results remain intrinsically uncertain.
Our argument makes three contributions. First, we contribute to a growing literature bridging sociology of the family and sociology of health and illness. Under the guise of querying a molecular diagnosis, genomic testing explicitly reveals a shared genetic foundation among living and future members of the kin network. The impact of testing on families goes farthest when the results’ implications for reproductive decision making are broached. For patients, an inherited risk may be sufficient to reconsider a pregnancy or to engage assistive reproductive technologies. A collateral effect of diagnostic genomic testing, then, is the distribution of risk among family members.
Second, understanding patient and clinician logics of implicated beneficiaries reflects back on the role of disability as a driver for genomic testing. The logics to justify extending genomic testing results revolve around disability as a state to be explained and avoided. Genomic results indicating risk for disability are treated as sensitive and applied reluctantly to existing relatives but willingly to the unborn. We examine how this particular risk of disability, that historically in the context of genetics has been associated with eugenics, travels from living to future relatives.
Finally, as one of the most advanced clinical genetic tests available, exome sequencing results also index what clinical genetics can explain and the relationship of such a genomic account to alternative social and environmental explanations (Shostak 2013). Much of the recalibration of genetic variants as the cause of disease occurs in research laboratories, but because clinicians must link general genomic knowledge to the specifics of a patient, the clinic also emerges as a site of knowledge production (Latimer 2013). Patients and clinicians not only interpret genomic testing results over the course of the consultation, but the findings also shift from less to more causally certain depending upon different relatives.
Background
Clinical exome sequencing compares a person’s sequenced DNA with a referent—the standard human genome—and singles out variant genetic markers that may indicate pathology. An unusual genetic variant raises questions about whether it is new in the patient or inherited, whether it is shared by other living relatives, and whether it may be passed on. In the process of delineating molecular causes, clinical geneticists explain for whose health these variants are relevant and take a stand on disability and disease with implications for reproductive decision making.
Genetic testing technologies then impact families through the revelation of shared genetic variants that tie certain blood relatives closer based on symptoms or disease risks (Armstrong, Michie, and Marteau 1998; Atkinson, Featherstone, and Gregory 2013; Bosk 1992). Rather than buying into genetic essentialism, an extensive interview-based social science literature shows that families render genetic information relevant within their preconceptions of who is at risk for disease and what “runs” in the family (Lock 2005; Markens, Browner, and Press 1999; Shostak, Zarhin, and Ottman 2011; Whitmarsh et al. 2007), a process referred to as the familiarization of genetics (Chilibeck, Lock, and Sehdev 2011). The subjective “badness” of the disease plays a big role in how socially actionable a variant becomes. Sharing a familial predisposition for hypercholesterolemia, for example, is easier to ignore than a susceptibility to Huntington’s disease when several relatives have died from this condition (Weiner 2011). Families resist and accommodate genetic knowledge (Rapp 2000). Even the decision to undergo genetic testing may reflect preconceived notions of family susceptibilities and may pale in light of more pressing concerns (Geelen, Van Hoyweghen, and Horstman 2011). Genetic information is rarely a deterministic force rearranging family relationships (Novas and Rose 2000) but is nevertheless a factor to be taken into consideration because it differentiates some relatives from others.
While we know that families qualify genetic results based on their understandings of who is or should be at risk, we know less about how geneticists and genetic counselors communicate genomic results to patients and families in the clinic. This initial presentation is critical due to the highly specialized nature and uncertainties in genomic testing. The notion of genetic determinism—that genes make us who we are in a traceable, unalterable biological design of humanness—has yielded over the past decades to the realization that the function of DNA is more complex, interactional, and marginal than expected and involves complications at the cellular and environmental level (Keller 2000; Landecker and Panofsky 2013). Consequently, within a paradigm of genomic testing, the interpretation of a variant as a cause or predictor of disease is marred by epistemic uncertainties related to, among other factors, the variant’s inheritance pattern, population frequency, and clinical association. Genetic causality not only varies from certain to probabilistic, but laboratory geneticists also report out variants with different levels of causality (see below). These classes of causality require much interpretive work by clinical geneticists to render abstract genetic information relevant to, and for, different implicated parties.
In this context of uncertain genomic causality, geneticists communicating genomic testing results will need to decide (a) for whom are the testing results relevant and, if there is differentiation, (b) on what logical basis are the results relevant for some relatives but not for others. In the most expansive orientation, genomic testing results have implications for patient and all relatives, including those as yet unborn. But clinicians and families may restrict the information for some parties. The logics used to limit or extend genetic findings could be clinical (the presence of a set of symptoms), genetic (the presence of a shared variant), therapeutic (likely interventions following from the genetic cause), or prognostic (either as medical surveillance or as a likely patient trajectory). The interpretation of implicated relatives could also be determined on voluntary (any relative who wishes to receive information), ethical (e.g., children may want to wait until they are adults to consent to genetic testing), or regulatory grounds (in France, for example, a law mandates that every positive genetic test result needs to be communicated to relatives; Derbez 2017). Because of their training and professional orientation, clinicians may be more likely to adopt a biomedical logic while patients and family members, following the literature on adaptation and resistance to genetic testing results, may opt for a logic privileging their understandings of risk for specific relatives.
It is unclear whether clinicians and families would prefer to limit or extend the results to various kin because implicating relatives in genomic testing results comes with known opportunities and drawbacks. A genomic diagnosis fixes a disease in ways that few other diagnostic procedures can. Especially for people who bounced between medical specialties in an attempt to figure out the reason for symptoms, a genetic diagnosis comes with the reassurance of knowing what is wrong and an institutional order (Jutel 2009). Locating causes at the molecular level, however, also raises concern about social stigma and discrimination (Nelkin 1996; Phelan 2005; Shostak, Conrad, and Horwitz 2008). Consequently, relatives may opt not to share genetic information (Hallowell et al. 2006). At the same time, research about patient preferences shows that a strong desire to find out genomic testing results is tempered with an aversion to learning about incurable conditions (Wright et al. 2014).
Susceptibility testing spreads those benefits and drawbacks of a genomic diagnosis in the family as prognostic knowledge, creating the opportunity for prevention but also for courtesy stigma (Goffman 1963). The guilt of passing diseases across the family may be attenuated by the responsibility to act on the opportunity of knowledge about risks and to reaffirm common bonds (Finkler 2000). Genomic testing may turn asymptomatic people into patients-in-waiting (Timmermans and Buchbinder 2010), living with the possibility of disease and with a knowledge—even if fundamentally uncertain and incomplete—that can never be “unknown.” Every relative may be at risk for this extension of the patient role.
The extrapolation of genomic findings to future relatives remains especially controversial due to a long historical entanglement of clinical genetics with eugenic practice (Comfort 2012). While disability activists have expressed concern that genetic information will be wielded as a eugenic tool for disability oppression (Parens and Asch 2000; but see Shakespeare 2006), some bioethicists have posited instead that there can be no moral objection to using genetic technologies for avoiding disability and enhancing human potential (Agar 2004). Clinical geneticists and patients recognize the potential of generalizing genomic findings to the unborn. Whether the possibility of a future child with disabilities prompts parents-to-be to take preventive steps preconception or during pregnancy depends to a large extent on the associations, familiarity, and experience prospective parents have with disability (Landsman 1998; Rapp 2000; Raspberry and Skinner 2011a). Here, the uncertainty of genomic information can be particularly distressing (Pilnick and Zayts 2014). Prospective parents must consider starting or discontinuing a pregnancy based on ambiguous prognostic information that could nevertheless be highly consequential. For parents of a child with disabilities, using prenatal genetic testing to decide whether to intervene in pregnancy often becomes a referendum on the child with disabilities. These parents are less likely to opt for future children rather than ending up in a difficult situation of having to make a reproductive “choice” about terminating a pregnancy (Kelly 2008; Raspberry and Skinner 2011b).
Clinical geneticists preparing to communicate genomic testing results to a family face the challenge of deciding for whom the results are relevant and in what way. The same genomic findings with varying levels of uncertainty may implicate various relatives differently and for different reasons. Because of their clinical orientation, we would expect that geneticists privilege biomedical information, while relatives may rely on their emic understanding of genetic risk to limit or extend the test results. Considering the convoluted history of genetics with eugenics, we may also expect geneticists to be cautious in expanding uncertain genomic test results to the as yet unborn. The different logics clinicians and families adopt may then facilitate testing additional family members, perpetuating a genetic worldview among some but not all relatives.
Data and Methods
Exome sequencing allows geneticists to sequence all coding regions of the human genome to identify variants associated with specific disorders. The exome constitutes 1.2% of the human genome involved in an estimated 85% of disease mutations. The test produces about 20,000 variants in every person, which the laboratory reduces to a handful of variants implicated in the patient’s phenotype. The diagnostic yield of exome sequencing is 26% (Klein et al. 2014). To capture the broad interpretive range of exome findings, the American College of Medical Genetics and Genomics recommends five test result categories depending on whether the variant causes the patient’s symptoms: a variant is either pathogenic, likely pathogenic, of uncertain significance (VUS 1 ), likely benign, or benign (Biesecker and Green 2014). Few labs report benign or likely benign polymorphisms since these variants have little clinical value. The inclusion of VUS results is challenging because the category suggests that there was insufficient evidence to classify the molecular change as detrimental or neutral. In a rapidly changing field, laboratories use the category to allow the clinician to monitor the variant.
We draw from a corpus of 38 video-recorded consultations of families with children living with a range of serious disabilities (such as seizures, cognitive delay, and physical symptoms) and one child with cancer only. Even if these children had had various clinical diagnoses for years, the geneticists initiated exome sequencing to explain the children’s specific constellation of symptoms and to provide more precise diagnostic, prognostic, and treatment information. These children had undergone exome sequencing by geneticists at a large academic clinic and visited the clinic to receive their results. When families were scheduled to return for the results, a genetic counselor alerted our research team. We introduced the research project and obtained informed consent from the families prior to the counseling session. We also gathered basic demographic and contact information at that point. Four families declined to participate in our study. An institutional review board approved all procedures.
The data set included consultations with six geneticists, all of whom were male. For this analysis, we focused on a subset of 22 cases involving child patients and the delivery of at least one variant that was interpreted on the laboratory report as of uncertain clinical significance, likely pathogenic, or pathogenic. 2 Of the patients, 52% were male (n = 11). The average age of the child was 8.6 years. Thirty-three percent of these parents self identified as non-Hispanic white, 29% as Latino, 24% as Asian, and 14% as African American. Fifty-seven percent of parents reported having graduated from college or having a graduate degree, while the remainder had less than a college degree. Sixty percent of parents reported having an income of $80,000 or more in 2012, with the remaining parents reporting incomes between $20,000 and $79,999. Finally, all but two families reported having English as their native language (90%).
Exome sequencing is an emergent technology, and while we may expect that the issue of how much information to reveal to whom is a common concern across academic and commercial testing centers, the generalizability of this analysis is limited due to a single research site at an academic center and our focus on genetic testing of children with disabilities.
We approached our analysis utilizing a combination of conversation analysis and ethnographic methods (Maynard 2003). This involved coding the data in a modified grounded theory approach (Timmermans and Tavory 2012) to distinguish the broad categories of genomic news delivery and their underlying rationales. For instance, we coded all cases for how the geneticist presented the genomic findings for the patient and then used this as a baseline to see how the results were adapted for different relatives. We coded whether clinicians or parents initiated these extensions of the results and how the interpretations changed over the course of the consultation in light of parent questioning. Once we found patterns in the news delivery, we analyzed relevant moments of the consultation for their interactional order, focusing on the consistency with which, for instance, parent questions shaped clinicians’ interpretations. The result was a conversation analytically informed microethnography. We identified themes present across the consultations. We showed transcripts of relevant segments of the consultations that were meant to be particularly clear examples of the behavior we were analyzing but, unless noted, did not otherwise differ markedly from other examples in our corpus.
Results
When a geneticist provides positive results of a patient’s exome sequencing, both parents and clinicians treat the findings as having implications beyond the patient’s genetic diagnosis. Yet, clinicians draw tighter boundaries around the results’ scope than parents. For geneticists, a genetic cause implies a molecular diagnosis and represents an end point in the patient’s diagnostic odyssey, a term used to refer to the extensive search across medical specialties to diagnose and explain a child’s symptoms. For parents, the genetic diagnosis also constitutes an explanation of their child’s symptoms. At least as importantly, however, parents treat the result as a genetic vulnerability shared across the extended family network, and they take the result as a warrant for inquiring about testing members of the extended family, such as siblings and cousins. Clinicians, in response, try to contain the scope of results and explicitly avoid unnecessary geneticization. The key issue is whether other relatives display symptoms consistent with the patient’s phenotype. If not, geneticists consider testing not indicated. Geneticists’ reticence to interpret the results beyond the patient has one important exception: they view the result as an opportunity to prevent a similarly affected future pregnancy and will extend the findings to as-yet-unborn siblings. In the name of prevention, the genotype of the unborn child is presumed to trump any consideration of symptoms.
These priorities reflect three different causal logics that are applied to different family members. Whereas clinicians use a phenotype-first logic as the warrant for testing the patient to achieve a diagnosis, geneticists maintain this clinically driven logic as a rationale for not testing asymptomatic relatives. A phenotype-first logic specifies that exome testing is indicated only when the patient exhibits symptoms consistent with a possible molecular cause. The logic prioritizes the clinical picture and uses genomic testing to explain symptoms. Parents, however, deploy a genotype-first logic as a rationale for testing asymptomatic relatives, orienting to the test not as diagnostic of a phenotype but as a susceptibility screen to reveal risk throughout the family pedigree. A genotype-first logic starts from the known presence of a genetic variant in the patient and aims to establish whether this variant causes symptoms or risks among blood relatives. Parents apply this logic to both born and unborn relatives. Geneticists strongly resist the extension of the genotype to other relatives with the important exception of the recurrence risk in future children. In discussing the reproductive implications of exome findings, geneticists rely on a genetic determinism logic that holds that genes determine phenotypical outcomes. This long-standing logic has been repudiated in much of genetics over the course of the twentieth century (Keller 2000), but as we will show, it survives in reproductive genetic counseling.
The Patient as Primary Beneficiary
When clinicians present positive findings of the exome test, they—not surprisingly, since that was the reason for ordering the test—treat the patient as the primary stakeholder, focusing on the degree of fit between the genomic result and the patient’s phenotype (Pilnick 2002b). Prior to discussing the results, the geneticist reviews with the parents any change in symptoms, medication, or therapies while consulting the inevitably thick medical file containing a long history of test results and interventions from various medical specialties. This review of changes in disabilities and other conditions, such as the onset of childhood cancer, forms the backdrop of the results discussion.
Geneticists commonly present the child’s phenotype as a puzzle to which the test provides a solution. This is particularly clear when the finding is classified as pathogenic or likely pathogenic. In Extract (1), after explaining how the exome test works, the geneticist makes explicit that the test gives a clear “answer.” He not only adds stress to the word answer but also underscores the point by stating explicitly that this “puts an end to what we’ve been looking for” (lines 2–3). Transcripts use conversation analytic conventions (Hepburn and Bolden 2013).
(1) Case 2 52:04
Approximately half of all test results involve variants of uncertain (clinical) significance (VUS), which can be subsequently treated by physicians in the clinical encounter as more or less likely to be causal (Stivers and Timmermans 2016). In Extract (2), the clinician indicates that the variant can cause liver complications. He then (lines 7–9) states the gene is a VUS, “meaning it may or may not give a symptom” (lines 11–12). Yet, even here the physician reveals that his goal is to identify a genetic finding that explains the patient’s symptoms. The clinician affirms the patient as the test’s primary beneficiary.
(2) Case 25 16.00
In VUS cases, the geneticist qualifies the exome results. The genotype may reflect variants that are too common in the population or may not have been linked specifically to the patient’s symptoms (Timmermans, Tietbohl, and Skaperdas 2016). This leaves the clinician unable to offer a clear cause. Yet, across the consultations—in VUS and more certain pathogenic variants—physicians reflect a phenotype-first logic: they are testing to establish a causal connection between the phenotype and the genotype but use the phenotype as the reference point.
While they do not use the term phenotype, parents also orient to explaining the patient’s symptoms as the primary purpose of the consultation. For instance, in Extract (1), the parents stated that they had waited 18 years to find out what “went wrong” with Richard. In other consultations, parents express excitement or anxiety about the prospect of finding out exactly what caused their child’s symptoms, even if implications for treatment remain unresolved.
Parent versus Clinician Orientations to Additional Relatives
Parent questions following the primary news delivery represent a window into their range of concerns and into their testing logic. Across the 22 cases of positive results, we observed that parent questions focused on five topics: family heritability, diagnosis and diagnostic certainty, prognosis, treatment, and alternative causes of the condition. Parents asked questions about family heritability in 73% of consultations (n = 16). Parents commonly initiated these questions immediately following the delivery of the main news, suggesting that they treated these questions as important. In contrast, over the course of the discussion, clinicians rarely broached siblings or extended family outside of parent questions.
These parent questions about heritability across the kin network posit a different logic than the phenotype-first logic where exome sequencing is used to explain disability. A genotype-first logic presumes that if a variant is present in one relative and is inherited, relatives may carry it too. Thus, it may be useful to know whether they are at an increased risk and use the genetic knowledge for preventative steps. This reasoning starts from knowledge about a genetic variant and wonders about different phenotypes that the variant may cause. Parents use genomic knowledge to differentiate members of their family tree.
In Extract (3), we see a series of questions concerning inheritance. The first question comes following the physician’s reminder that although the parents were tested, only results relevant to the child patient will be included in his results presentation (lines 1–9, 11–15). The mother then requests clarification of whether this means that they cannot tell their carrier status (lines 19–22, 23–25). This question extends the implications of the results to the parents.
(3a) Case 10 10:26
The mother’s concern about inheritance extends beyond her own carrier status to that of other members of her extended kin network—her mother, grandmother, and aunts—as shown in (3b).
(3b) 22:15
For the mother, the shift from a discussion of her child’s genes to her own, her mother’s, grandmother’s, and aunts’ reflects a concern with the spread of the condition across her family pedigree. This is similarly true when parents ask about the implications of the genetic finding for their other children (the patient’s siblings). In Extract (4), before the clinician can fully articulate the inheritance pattern, the mother offers her understanding in overlap with the doctor that the relevant variant would have been inherited from her (line 6). As soon as she receives confirmation, her next question concerns the boy’s half-brother (lines 8–9).
(4) Case 13 39.07
Thus, parents’ questions certainly are concerned with explaining their child’s symptoms, and in that sense there is an overlap with the physician’s orientation to a fit between the exome result and the patient’s phenotype. Parents’ questions about different relatives reveal, however, that they are treated as topically coherent. Switching to a genotype-first logic, parents move fluidly from questions about the patient to the effect of the variant in other relatives. For parents, the patient represents one node in the family network, and it is a brief hop to other kin. Across our data, parents ask about the implications of genomic results for themselves, other children, siblings, nieces, nephews, aunts, uncles, parents, and grandparents.
In response, physicians across our study actively argued against extending the genomic knowledge to relatives and discouraged the testing of extended family, treating these potential patients as outside of their jurisdiction. Clinical geneticists typically assert that in the absence of symptoms, it would be pointless to test the family member. Often the child tested had serious disabilities apparent at birth, and without the presence of similar symptoms in kin, finding the same variant would have limited clinical utility. Variants do not necessarily express their pathogenicity in the same ways, however, even among blood relatives. Geneticists are aware that one relative may be affected at birth while another relative may develop symptoms only later in life or that a variant may produce a range of symptoms, some of which may not be readily apparent without further testing or monitoring. Even without symptoms, clinicians could thus have made a case for testing a single variant in selected relatives.
Still, they consistently countered the parents’ genotype-first logic with a phenotype-first logic—the same logic that led to testing the patient. Extract (5) offers an example involving eight-year-old Ben, the child in our study not with multiple disabilities but with only the onset of cancer (another child had cancer and disabilities). This case is analytically highly relevant because the sudden, unexpected onset of the cancer could be a strong clinical rationale to find out if other relatives shared his risk. Exome sequencing discovered a heterozygous variant in the CHEK2 gene (which the father carries) prevalent in the Ashkenazi Jewish population, the ethnicity of Ben’s father. The father inquires about whether his daughter should be tested as well, given that the variant was inherited from him. The geneticist immediately disarms this question with the lack of symptoms, invoking a norm that geneticists do not test without symptoms (lines 15–19), even though the father notes that his son had no symptoms either prior to developing cancer (lines 10–11).
(5) Case 19 29:00
In this interaction, we see the two different logics clashing. The geneticist sticks to the prevailing logic of phenotype-first: in an absence of symptoms (i.e., cancer) checking for a variant would not be informative and could cause undue anxiety. Testing is done to solve the puzzle of a symptomatic phenotype—a puzzle not present in Ben’s sister. In Ben’s case, genomic testing could have explained his cancer. Even if his sister shares the variant, she is currently asymptomatic and could remain so for the rest of her life, similar to her father. The lack of a phenotype then makes testing of relatives moot. In contrast, when the father raises the possibility of testing Ben’s sister, he implies a logic that prioritizes the genotype and uses that to screen for susceptibility across the family. Testing his daughter makes sense to him because Ben’s cancer occurred unexpectedly, suggesting that his daughter may be similarly at risk for a sudden onset of cancer. In countering the argument, the geneticist points out that it may be worse to burden an asymptomatic sibling with this kind of fundamentally uncertain genetic knowledge. The geneticist elaborates the reasoning with reference to the sister’s autonomy to make a decision as an adult regarding genetic testing.
In considering how parents and clinicians approach the delivery and reception of exome sequencing results, we have observed that both participants orient to the phenotype-first logic with respect to the patient. Yet the participants diverge when considering further importance of the results. Parents consider the extended kin network to be another set of stakeholders who, using a genotype-first logic, might benefit from testing. This logic transforms exome sequencing from a diagnostic test to a susceptibility screen. Clinicians maintain a phenotype-first logic and consistently point out that if a member of the kin network is asymptomatic, then he or she should not be tested. Their response does not differ based on whether the variant is likely to be causal or of uncertain clinical significance. In contrast to how they presented the result for the patient, they may thus downgrade (i.e., consider less causal) a pathogenic variant for the extended family. The two exceptional situations in our study where a geneticist agreed to test a sibling confirm this phenotype-first logic. In the first, the patient’s brother had extensive developmental delay and other symptoms similar to the patient that could plausibly be explained by the variant. In the second, the patient’s younger sister also had similar symptoms to the patient, also indicative of joint looseness, and the geneticist—noting that “knowledge is power”—agreed to test the sister but only for the specific variant.
Leveraging Genetic Causes for the Unborn
Even though geneticists work hard to stem any spillover of genetic variants to asymptomatic relatives, they equally adamantly encourage testing for the prevention of disability in unborn siblings either preimplantation or during pregnancy. Geneticists skip the presence of a symptomatic phenotype in the unborn as a necessary prerequisite for genomic testing, instead using a genetic deterministic logic that gives even more power to genetic variants than a genotype-first logic. A genetic cause—even if tentative—in an older sibling (the patient) is sufficient to warrant genetic testing of the unborn and to counsel prospective parents on the availability of reproductive interventions. Where the phenotype prevailed in arguing against testing living relatives, in the discussion of recurrence risk for the unborn, the genotype becomes deterministic.
The exceptional position geneticists take related to the unborn becomes apparent in who initiates the topic of reproductive implications. Not every visit has potential unborn siblings since in some cases the parents have made it clear that they will not be having more children. Strikingly, however, when reproductive decisions are presumed relevant, physicians typically initiate discussing recurrence risk. This sharply contrasts with geneticists’ avoidance of discussing the implications of testing results for extended family members.
Consider the case of Michaela, who lost developmental milestones at six months of age and received a clinical diagnosis of “global developmental delay” with “refractory seizures.” At the time of exome sequencing, Michaela was six years old but functioned cognitively at the level of a two-year-old. The family opted for exome sequencing to figure out what caused Michaela’s intractable symptoms. Exome sequencing identified two genes: a de novo variant in SCN2A which has been associated with infantile seizures and a variant in MYO5A that is usually associated with skin and immunodeficiency issues but may also have neurological ramifications. The team thought that SCN2A explained the phenotype and reported the variant as pathogenic, but since MYO5A might also be involved, they listed it as a VUS.
The parents met with the geneticist and genetic counselor while an aid accompanied Michaela into the waiting room. As is typical, the geneticist interpreted the pathogenic result as a solution to the puzzle of what was ailing Michaela. He announced that “the exact same variant” (in SCN2A) had been found “in one patient with neonatal infantile seizures” and handed the father the journal article where this case was reported. The geneticist concluded that the “genetic variant [SCN2A] . . . makes sense. We’re in rather firm territory here because this exact variant has been reported before. . . . I feel rather comfortable telling you that it’s likely to be the cause.”
The geneticist brought up the relevance of the finding for reproductive decisions. He noted that the de novo inheritance pattern suggested that a future child would be unlikely to have the same mutation.
(6a) Case 3 15.12
The geneticist expanded on the reproductive implications discussing chorionic villus sampling, amniocentesis, and preimplantation genetic testing. These technologies are presented as means to prevent the onset of the associated disabilities, even though the test had established that the variant in Michaela was de novo. The mother’s question, “Can we? on this particular one?” implies that the common goal here is to avoid a recurrence of this disability in a next pregnancy.
With this line of reproductive counseling, geneticists have abandoned their standard that genetic testing requires a puzzling phenotype. The egg, the sperm, amniotic fluid, and the fetus do not yet have an easily interpretable phenotype, but they each have a sequenceable genotype, and selecting the fetus based on this genotype may prevent recurrence of disability. With their suggested course of action, geneticists also confirm that autonomy in reproductive decision making resides with parents and not with the unborn child: whereas with an older child, issues of autonomy become relevant regarding the decision to get tested, a fetus has no say in the decision (Casper 1998). The unborn child is considered unique because disability can be prevented. Prevention, however, requires a strong commitment to genetic determinism in which the disabled older sibling’s phenotype signals both the outcome to be avoided and the fetus’s likely future. When considering recurrence risk, geneticists are no longer treating exome sequencing as a diagnostic test or as an open-ended susceptibility screen but as a prognostic test.
This reproductive genetic exceptionalism is further underscored by geneticists’ willingness to upgrade (i.e., attribute more causality) variants of uncertain clinical significance. This happened in Michaela’s case when the geneticist addressed the second gene classified by the laboratory as a VUS. “We found something else in a different gene called MYO5A, which has been associated with Griscelli syndrome.” After asking the family about skin and immune issues but not finding any, he added that the variant could be relevant but geneticists just do not know enough about it. While explaining, he realized, however, that the implications for reproductive decision making are critical because the MYO5A variant is compound heterozygous, meaning that the recurrence risk is relatively high at 25% since both parents are carriers for the variant.
(6b) Case 3 27:13
Whereas in the case of living relatives and patients the geneticists carefully qualify their consultations based on the variant’s heritability and causal ambiguities, these uncertainties are played down when discussing recurrence risk. The geneticist’s realization that the implications of a VUS result differ for Michaela and for her future sibling pits Michaela’s interests against those of her parents and unborn siblings. Finding two genes did not affect Michaela’s diagnosis. As the geneticist stated, the second variant “may be relevant” and is therefore part of the discussion, but the initial gene is treated as primary and causal. The VUS gained predictive currency, however, in the context of reproductive decision making. The implications of the genetic results for the parents contemplating another child and for Michaela are thus given different causal weights.
Born and Unborn Implicated Relatives
How can geneticists who strongly adhere to a clinical logic of phenotype-first and rebuff a genotype-first logic still adopt genetic deterministic arguments? Rather than opposing or incompatible logics, we argue that the logics support each other in the sense that a genotype can become deterministic only for the unborn in the context of prior genotype-phenotype links. The presence of an obviously disabled or seriously sick patient with a (potential) genetic cause is necessary for the interpretive leap to future risks. When the geneticist establishes that a variant causes or even may cause a severe phenotype, then the mere recurrence possibility in the patient’s future sibling is sufficient to trigger a discussion of reproductive interventions. The deterministic logic then feeds off the phenotype-first logic: for the genotype to be deterministic in future siblings, the possibility of causality in a patient first needs to be established. Put differently, simply finding pathogenic variants unrelated to the patient’s phenotype will not lead to preventive measures (this would imply a genotype-first logic). Actually, the fact that such incidental variants would have been observed in an older sibling without a corresponding phenotype will likely render the findings nonreportable.
We turn to a final case to show that even within the course of a single consultation, clinicians fluidly treat different implicated parties according to contrasting testing logics. The switchback between different logics depends upon the mere possibility that a VUS may cause the phenotype. Although only a VUS for the child patient, this relatively weak causal connection is sufficient to render the variant deterministic in the discussion of future children.
This case involves eight-year-old Adrian, who has a history of infantile spasms and seizures and is gastronomy tube dependent with little cognitive development. The team located a hemizygous, maternally inherited, X-linked VUS that may indicate iron accumulation in the brain. The geneticist explains causal ambiguity when stating that the genetic change “is 100% sure. What is not 100% sure is whether this change will cause the condition that he may or may not develop based on this gene, right?” For the purpose of explaining the patient’s phenotype, the variant offers only a qualified answer. Adrian’s disabilities may be caused due to iron accumulation, but there is no way to tell until he undergoes repeated MRIs for the next years to locate and assess the buildup of iron deposits in his brain. The exact causal role of the variant remains mired in uncertainty.
The tentativeness of the causal nature of the results diminishes, however, when the geneticist broached reproductive implications.
(7a) Case 10 20:41
The explicit reference to “having another child like Adrian” contains the clue to how geneticists can switch from their strong alignment with a phenotype-first logic to a genetic deterministic logic. Although this consultation is unusual for its explicit reference to having another child with disabilities, many consultations (including [6b]) implicitly allude to the disability as an unwanted state. Adrian, with obviously apparent disabilities, embodies the outcome to be avoided. The geneticist’s use of the tag question, “right?” presumes that the mother would agree and pursues her agreement. Exactly because the geneticist considers the phenotype so severe, the uncertainties of the variant’s causality matter little. If there is even a small chance that reproductive technologies could avoid having a similar child affected with a fatal condition, geneticists feel it worthwhile raising the preventative steps. The VUS gains causal power in light of the undesirable phenotype of a child with known disabilities. The phenotype-first logic then enables genetic determinism.
In line with our previous findings, the mother next brought up the implications of this result for her immediate relatives. After she asks whether her other sons should undergo testing, the geneticist counters that they are asymptomatic and discourages testing, in the process deflating the variant’s causal relevance. Thus, even though the geneticist has deployed a genetic deterministic logic to recommend pre- and postnatal testing two minutes earlier, he now disarms the mother’s genotype-first logic.
(7b) Case 10 23:10
The mother challenges the physician’s stance, citing that one of her other sons has obsessive-compulsive disorder, severe attention deficit disorder, and anger issues. The geneticist concedes: “We can go ahead and [test], but it’s something that I would rather not create more confusion . . . because we’re really not sure for Adrian if this change really is the reason of what’s going on.” The genetic counselor adds, “And rather than going down this road where we find this change in all these different kids and then we get overconcerned, might be, may create more anxiety.”
The geneticist and the genetic counselor try to talk the mother out of going on a testing spree by qualifying the causal power of the genomic finding. They fall back on the difference between the siblings’ and the patient’s phenotype: even though the brothers also exhibit behavioral issues, they do not share the same symptoms. Consequently, according to the geneticist and genetic counselor, knowing that the siblings have a mutation would not provide medically actionable knowledge. With the reference to undue anxiety (see also Extract [6b]), they hint at the social power of diagnoses in organizing a medical trajectory and disease identity, possibly leading to stigma. In contrast to the recurrence risk discussion earlier, finding a genetic cause may lock the siblings in on a diagnostic path that does not fit them right now.
At this point of the consultation, we see how the three different logics allow geneticists to draw different implications from the exact same variant for various relatives. In light of the patient’s phenotype, WDR45 is at best a hypothesis that will not be testable until the comparison of MRIs taken over multiple years. In contrast, the mere possibility of preventing a severe phenotype imbues the uncertain variant with deterministic causal power in the context of reproductive decisions. Yet, in the opinion of geneticists, the variant is meaningless for screening asymptomatic, or in this case differently symptomatic, siblings. Geneticists then consistently use the patient’s phenotype as a touchstone for causal inferences. The mistake that parents make, in the eyes of geneticists, is to divorce the exome results from the phenotype and give the genetic variant independent causal power. Thus, “another child like Adrian” justifies genetic determinism for discussing recurrence risk, while the prospect of “more anxiety” in asymptomatic relatives argues against a genotype-first logic.
Discussion
For whom do genomic testing results matter and in what way? If a strict clinical rationale were the only factor driving the interpretation of testing results, geneticists could extend results to all relatives manifesting similar symptoms, to those with a common genetic inheritance, or to all wishing to take preventative measures. Technologically, it is relatively straightforward, cheap, and safe to test living relatives for known variants after obtaining a blood sample. Yet, geneticists and families abide by testing logics that render the same variant differently actionable for various relatives.
While both clinicians and parents agree that the primary beneficiary for exome sequencing is the patient, they differ on the implications of the findings for living relatives. Embracing a genotype-first logic, parents treat the variant as a familial risk factor and advocate for finding out whether the variant puts others at risk. Geneticists, in contrast, explicitly advocate against medicalization of symptom-free relatives using a phenotype-first logic. This disjuncture in logics may be explained by the experience of parents devoting their lives to taking care of a child with disabilities and their desire to prevent similar challenges for others. The test result may offer the hope that even if the genetic results do not profoundly change the care for the child with disabilities, knowing about the genetic risk in the family may help prevent the onset of medical problems. As such, the parents’ expansive attitude would be consistent with the motivation of parents with disabilities to join disease advocacy movements to prevent disabilities through national screening programs, even if their own child will not benefit from screening (Grob 2011). Some parents in our study indeed expressed such altruistic desire for prevention.
The clinician’s perspective, in contrast, is marked by the difficulty of interpreting genomic variants. Children with disabilities should be among the easiest patients to locate genetic causes for because they have obvious, well-studied symptoms. Yet even in this population, the diagnostic yield of exome sequencing remains low. In the observed consultations, geneticists referred to most patients as “unique,” meaning that there have been few other patients described in the literature with similar genotype-phenotype matches. The field is not ready to test asymptomatic relatives because geneticists could not even imagine what the actionable clinical value of such testing would be. When parents raise the issue of testing living relatives, geneticists and counselors aim to dissuade them, and as they remain gatekeepers for such testing, no asymptomatic relatives in our study underwent testing.
This reluctance makes the geneticists’ volunteering the use of exome findings for prenatal testing remarkable. Research has shown that parents of children with disabilities are less likely to want more children than parents with nondisabled children (Raspberry and Skinner 2011b). Few parents in our study brought up reproductive concerns, and most showed little reaction when the geneticists and genetic counselors launched extensive discussions about using the findings to test future pregnancies. When we discussed this puzzling finding with an experienced genetic counselor, she hypothesized that her colleagues’ attitude toward reproductive testing might be a holdover from earlier times, when the field of counseling started and counselors realized that they often could not help an affected child but might be able to prevent similar symptoms in a future pregnancy. The crystallization of a cultural script for genetic reproductive counseling decades ago could also explain the staying power of a genetic deterministic logic in this context.
In light of a well-known and controversial history of eugenic practices fueled by genetic deterministic theories, the exception to the cautious attitude regarding genetic labeling in the reproductive context is nevertheless surprising. Still, since we did not study exome sequencing in a prenatal clinic or in ongoing reproductive decision making, we did not observe actual eugenic practice. Instead, we observed how geneticists recommended assistive reproductive technologies and prenatal testing, but they did not link them to state population-control policies for human enhancement and procreation deterrence. Befitting contemporary neoliberal medical practice (Kerr and Cunningham-Burley 2000), the reproductive implications are discussed in the context of medically mediated individual choices. Yet, these options originate from a test that looks at thousands of variants about which often little is known. Next-generation testing is then altering both the quantity and quality of genetic information to be considered, rendering it a different kind of eugenics that has be called eugenomics (Aultman 2006). Eugenomics pivots on how clinicians and parents render uncertainties meaningful in clinical and family situations. With a rapidly evolving epigenetic and gene–environment literature, it is unclear whether genetic information will lose predictive power. We find, however, that geneticists are willing to overlook epistemic uncertainties for the sake of prevention. This observation that geneticists inevitably engage with eugenic thought in oblique or more explicit ways is consistent with the work of historian Nathaniel Comfort (2012), who mapped eugenic thinking over the course of the twentieth century. Future research, however, will need to explore how these discussions affect actual reproductive decisions.
Research in epigenetics and gene–environment interactions has rendered the received Mendelian view of finding a single gene for a disorder the exception rather than the rule. Very few of these insights, however, have trickled down to the clinic. Exome sequencing is currently not capable of capturing gene–gene or gene–environment interactions. In this sense, the search for a single or couple of genes to explain complex disability phenotypes looks increasingly anachronistic in light of the epistemic edge of genomic science (Keller 2000). Yet, the causal role of variants is qualified in the interaction between parents and geneticists. Geneticists downgrade the causal role for asymptomatic patients, even if parents wonder about susceptibility, and upgrade the same variant in light of recurrence risk. Within the limitations of single-variant causality, it is not just the anticipatory future use of the variant that establishes its specific relevance for different relatives, but genetic causality also emerges as an interactional achievement out of the consultation. Regardless of how families incorporate the relevance of this information, the consultation session then establishes an irreducible social and genetic connection between some and not other relatives.
Footnotes
Acknowledgements
We are grateful for the comments of Neil Gong, Eleni Skaperdas, Iddo Tavory, Caroline Tietbohl, the reviewers, and members of the University of California, Los Angeles Institute for Society and Genetics.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by a National Science Foundation grant (SES-1256874).