Sequencing Newborns: A Call for Nuanced Use of Genomic Technologies

Introduction and Summary of Recommendations

The cost of genome sequencing has plummeted over the last 10 years. Many scientists and doctors hope that affordable genome sequencing will lead to more personalized medical care and improve public health in ways that will benefit children, families, and society more broadly.¹ One hope in particular, which has been expressed on multiple occasions by the current director of the U.S. National Institutes of Health, Francis Collins, is that all newborns could be sequenced at birth, thereby setting the stage for a lifetime of medical care and self-directed preventive actions tailored to each child’s genome.² Indeed, it is often suggested that universal genome sequencing is inevitable.^{3, 4}

Such optimism can presume that discussing the potential limits, cost, and downsides of widespread application of genomic technologies is pointless, excessively pessimistic, or overly cautious. We disagree. Given the pragmatic challenges associated with determining what sequencing data mean for the health of individuals, the economic costs associated with interpreting and acting on such data, and the psychosocial costs of predicting one’s own or one’s child’s future life plans based on uncertain testing results, we think this enthusiasm deserves to be tempered.

Holding in mind the reasons for both optimism and caution, this report responds to the premise that sequencing should now be universally applied to newborns. It grows out of a four-year research process that was funded by the National Institutes of Health [see box 1], and is addressed to individuals and groups involved in the care of newborns and interested in the potential role of genetics in that care. These groups include federal and state policy makers and regulators, newborn screening programs, practice leaders in pediatrics and family medicine, insurers, genetic scientists, disease and disability advocates, public health practitioners, health educators, and journalists.

Our Ethics and Policy Advisory Board was large and heterogeneous. We did not reach agreement about everything. Some members find this report too cautious about the prospect of using sequencing in the newborn context. Some think it isn’t cautious enough. The six lead authors took responsibility for writing this report and are answerable for everything that is controversial or wrong in it. For what this report gets right, readers should credit the entire group, who generously contributed large amounts of time and rich insights to the discussions out of which this report grows.

In the analysis that follows, we distinguish between two reasons for using sequencing: (1) diagnosis of individual infants who have been identified as sick, and (2) screening populations of infants who appear to be healthy. By diagnosis, we mean identification of a disease or illness that explains a newborn’s symptoms. By screening we mean the systematic use of a test across large numbers of infants to identify those who have conditions like cystic fibrosis or sickle cell disease, that are currently detected through state-sponsored newborn screening programs, and those who are at elevated risk for developing conditions in childhood or later in life, such as pediatric or adult-onset cancers.

We also distinguish among three contexts in which sequencing for diagnosis or screening could be deployed: (A) in clinical medicine, (B) in public health programs, and (C) as a direct-to-consumer service. Each of these contexts comes with different professional norms, policy considerations, and public expectations.

Finally, we distinguish between two types of genome sequencing: (i) Targeted sequencing where only specific genes are sequenced and/or analyzed (gene panels are a kind of targeted sequencing) and (ii) whole exome or whole genome sequencing, where all of the DNA (whole genome) or all of the coding segments of all genes (whole exome) are sequenced and analyzed. Whole exome and whole genome sequencing generate large amounts of data and therefore a large number of results that could be returned to parents and families.

We say more below about the kinds of data that sequencing generates. For now, we note that while some sequencing data can clearly indicate that an infant has a rare pediatric disease or is at elevated risk to develop a disease in childhood or later in life,⁵ often the data are far more difficult to interpret. Knowledge of the human genome is still in its infancy, and the full range of gene/disease relationships is not known, nor is the meaning of many sequence variations yet clear. Even when scientists are certain that a given stretch of DNA is statistically unusual, its implications for the person’s health and well-being can be quite uncertain. The expression of most gene variants is, after all, heavily shaped by a complex system of genomic, epigenetic, and environmental forces that are different for each individual and thus impossible to predict precisely based only on the sequencing results. Interpretation of these variants can be especially difficult if the sequenced individual is currently symptom free, as many newborn babies are.⁶

While the complexities and uncertainties associated with the interpretation of sequencing data create challenges when the technology is used to diagnose symptomatic newborns (see 1, above), the Ethics and Policy Advisory Board (EPAB) concludes that these challenges can be managed, and they can be outweighed by the potential benefits to the child and family. In a symptomatic newborn, targeted or whole exome/whole genome sequencing can help guide other tests for diagnosis or specific treatment that is urgently needed. This kind of sequencing can be an iterative process, during which clinicians use the infant’s symptoms (or phenotype) to help interpret the sequencing data. These same complexities and uncertainties, however, limit the usefulness of whole exome or whole genome sequencing as a population screening tool. In an asymptomatic baby, there is no iterative potential as there are no symptoms that can help clinicians interpret the sequencing data. While we do see considerable benefit in using targeted sequencing to screen for or detect specific conditions that meet the criteria for inclusion in newborn screening panels (see 2a above), uses of sequencing as a more generalized screening tool for newborns (see 2b above) are at best premature. We present arguments for these conclusions below.

We do not see the widespread adoption of whole genome sequencing in newborns as inevitable. Nor do we assume that this—or any—new technology should be used indiscriminately or to its fullest extent in all cases. Our society has a responsibility to reflect, prospectively, on the potential benefits as well as the known costs and harms that genome sequencing could bring and to tailor adoption of the technology accordingly.^{7, 8} Our analysis is therefore responsive to the different reasons for using the technology, the different contexts in which it might be used, and the different versions of the technology available. Because we focused on newborns, we were sensitive to newborns’ inherent vulnerability and to the importance of the newborn period for families. We developed our recommendations with a clear understanding of the history and distinctive role of universal screening programs for newborns in American public health and the particular good that such newborn screening programs have achieved.^{6, 9} We conclude that sequencing technology can be beneficially used in newborns when that use is nuanced and attentive to context.

Commitments to Clarity and to Benefiting Infants, Family and the Public Good

In this report, we try to be clear about the nature of the technology, the complexity of the results that it generates, and what is known about sequencing’s benefits, limits, and costs for individuals, families, and the public. We also hew closely to the well-known ethical commitment to benefiting the health and well-being of patients—in this case infants, and to some extent their families.

Clarity about benefits and limits is particularly important for a technology, like sequencing, where substantial gaps still separate the hoped-for and the established implications of its use.^{1, 10} Some of these benefits and limits stem from the complex, varied, and uncertain nature of sequencing results. An increasing number of gene variants are known to be highly correlated with specific diseases or disorders, and some of these pathogenic or likely pathogenic variants have specific therapeutic interventions available.¹¹ But many have unclear health implications. Of particular importance, sequencing offers little if any predictive information regarding a person’s risk for most common diseases (e.g. heart disease, diabetes, and hypertension), which are due to many complex factors that include multiple genes but also include environmental and behavioral factors. These diseases do not follow the one-gene-one-disease model of traditional Mendelian genetics in the way that well known genetic disorders like Huntington disease or sickle cell anemia do. This gap between what sequencing results can reveal and the kinds of information most people need to improve their health, combined with widely publicized hopes for the revolutionary power of genomics, creates the very real risk that patients, research participants, health care professionals, policy makers, and others will have unrealistic expectations of what sequencing can achieve and little appreciation for its downsides.

Research on the use of genome sequencing for newborns highlights both the promise and the perils of this technology. Studies are showing that sequencing can aid in rapid diagnosis of selected hospitalized infants. These diagnoses can in turn can lead to changes in medical care, including use of different interventions or redirection of care, including to palliative care.^{12, 13} Other research shows that using sequencing as a screening tool in healthy newborns can pick up some, but not all, of the conditions that are currently screened for in state newborn screening programs.¹⁴ Use of sequencing as a screening tool has also been shown to detect some rare and serious health conditions that could conceivably meet newborn screening criteria but that are not included in those programs today because they cannot be detected with existing testing modalities.^{15,6, 16}

At the same time, researchers have raised concerns about the potential for screening uses of sequencing technology to lead to unnecessary interventions and costly long-term monitoring. Research also shows that parents are concerned that sequencing results might be used in discriminatory ways by employers and might cause children to be denied life, long term care and other kinds of insurance (US federal law prevents health insurers from using genetic test results). ^{6, 17} The newborn period is an important time for family bonding as well as a time of vulnerability for the still developing baby. These observations have led to hopes that sequencing would enable diagnosis of conditions that, if treated early in life, could significantly improve the child’s health. They have also led to concern that providing parents with variants for adult-onset conditions, variants of uncertain meaning, or variants that are not clinically actionable, would offer no immediate medical benefit to the vast majority of infants and families and instead cause parental distress and disrupt parent-infant bonding. Children who have a pathogenic or likely pathogenic variant could be labelled as “patients in waiting,” risking their stigmatization by both family members and health care professionals. ¹⁸

The ethical commitment to benefiting the health and well-being of infants and families is deeply familiar to healthcare professionals, captured by such familiar phrases as “first do no harm” and “the patient’s best interests.” It stems from and applies to the care relationship between health care professionals and patients and primarily concerns physical harms and health-related benefits, including psychosocial harms and benefits. When it comes to care of newborns, the primary and often sole focus of that care is the infant and what might benefit or harm that infant directly.

In some contexts, however, we are persuaded that it makes sense to extend this core commitment beyond the infant to take account of benefits and harms to families. Such an extension recognizes that infants are embedded in families, and that while the infant’s health and well-being is of the utmost importance to assessments of sequencing in newborns, it would be a mistake to sharply separate the health and well-being of a child from the health and well-being of that child’s parent(s) and family.¹⁴

In particular, while we recognize that decisions about whether to sequence a newborn’s genome should still rest on the needs of the infant, we recognize that there can be occasions when returning certain kinds of results from that testing could be justified on the basis of direct and significant benefit to a parent or sibling. For example, results might reveal that a parent has elevated risk for cancer or the newborn is a carrier for a recessive genetic trait, meaning that while the child is will not develop the condition, one of the parents must also a carrier and a subsequent child might be affected by the condition.¹⁹ This information might cause parents to undergo further testing themselves, and could inform their future reproductive plans. If a child is undergoing sequencing anyway, return of carrier results to parents might, under an expanded definition of benefit, aid the family, which of course includes the child. Allowing for a slightly expanded understanding of benefit means, on the flip side, that we must consider the potential for sequencing results to harm parents too, including by producing unnecessary anxiety in parents or damaging the parent-child relationship. Return of an infant’s carrier results could also offer no benefit to the family or the infant whose genome was sequenced, and could instead open that child up to stigma or discrimination when they reach childbearing age or result in additional but unnecessary costs to the health care system if the result is not well understood by the child or family over time.⁶ If nothing else, operationalizing an expanded understanding of benefit will require robust informed permission from parents and access to robust follow-up counseling and follow-up services for families, both of which are possible in clinical situations but very unlikely in screening contexts, including public health newborn screening programs.

A complicated and increasingly contested well-being related issue attending the use of sequencing in children, including newborns, is the impact of testing on the child’s future decision about whether or not to be tested. Parents make all sorts of choices that impact their children and open or foreclose that child’s future choices, including decisions about vaccination, school choice, and religious upbringing. Parents also learn about their child’s current and likely future health through various tests and screening programs. Yet many, if not most, of those decisions are a necessary part of parenting that cannot be postponed until the child is old enough to make their own decisions, or they are decisions that can be reversed. When it comes to religious beliefs, a child can eventually make their own decisions, and when health conditions are identified as part of newborn screening, they will have onset in childhood and require immediate treatment. But testing for health conditions that do not have their onset until adulthood, and for which there are no beneficial interventions during childhood, could be postponed until the child is older. Research has repeatedly shown that many adults who are at familial risk for such conditions—including Huntington disease, certain breast and ovarian cancers, and Alzheimer’s—elect not to be tested for them or seek to control the timing and circumstances of the testing,^20–22 decision that are foreclosed when parents agree to whole genome or whole exome sequencing or receiving results associated with adult-onset conditions. For these reasons, professional guidelines on pediatric genetic testing have long cautioned against testing children for adult onset disorders, a position that is sometimes described as maintaining the child’s “right to an open future.”^{23, 24} The language of rights may be too strong, but the future interest of the infant in deciding for themselves about whether and when to learn about their risk for various adult onset conditions deserves consideration.

Finally, we note that decisions about whether to use a technology like sequencing in one newborn can have implications beyond that infant and their family, which is why the broad commitment to beneficial action extends to the public good. For example, individual use almost always draws on shared healthcare resources, can have implications for evolving norms around parenting or for public attitudes about the meaning of particular disorders, disabilities, and other kinds of differences, and could contribute toward shared understandings of the rights and interests of children. Another public resource—newborn screening programs—could be affected by sequencing technology. Newborn screening programs are an effective way to bring beneficial services to newborns regardless of insurance status or their parents’ ability to pay, and for this reason they are, in the United States, one of the few programs that deliver guaranteed care to all.⁶ Many clinical genetics programs pride themselves on delivering not just screening, but also follow-up diagnosis and treatment services. They must operate within fixed budgets and are understandably loath to adopt technologies or expand the number of conditions they screen for if doing so would undermine current programs. If sequencing were instead used as a screening tool within routine pediatric care, attention must turn to the capacity for existing pediatric healthcare professionals to understand and appropriately respond to the volume and diverse nature of sequencing results, and for payers to cover the costs of delivering these services and follow-up care, potentially over the child’s entire lifetime.

Making good on these commitments to benefiting infants, families and the public good requires adequate funding for, and attention to, research on the implications of sequencing technology for infants, families, and the public good. This research need not seek a yes/no vote on sequencing: it is a rare technology that is always beneficial or always problematic. More likely—as research is already showing—for each purpose to which sequencing can be put, and across each context in which it might be used, and occasionally even from patient to patient, the benefits and burdens of using the technology will differ in both kind and magnitude. For example, even a small possibility that sequencing an undiagnosed baby in the NICU could improve that infant’s medical care—or provide answers to a distressed family—might justify the cost of analyzing the data, counseling the family, and the risk of generating secondary findings that could lead to unnecessary concern. Yet these same outcomes could be weighed very differently when considering the possible use of sequencing to screen all newborns.

In extending our understanding of the implications of sequencing to family members and the public, we acknowledge the concern that this kind of thinking can lead to policies or practices that prioritize the best interests of parents or society over those of the child, resulting in harm or loss for the child. These concerns will be familiar to those who have followed previous technologically-driven expansions of state newborn screening panels.⁶ Given the complexity of these expanded benefit debates, we want to make clear that we are not advocating for policies or practices that sacrifice the interests of the child for the interests of family members or for the greater good. Instead, we are asking for those other potential benefits or interests to be investigated and discussed as part of a nuanced exploration of the implications of sequencing in newborns—and we are keeping open the possibility that there might be rare circumstances under which decisions about sequencing an infant might be justified in some part because the results could aid the family and, somewhat more often, under which returning (or not returning) certain kinds of results to parents might be justified with reference to the family or indeed to the public good.

Achieving Nuanced Use of Sequencing in Newborns in Three Contexts

Completion of the Human Genome Project, the international effort to sequence the three billion letters of human DNA, was accompanied by predictions that genomics would go on to transform medicine. In April 2003, the National Human Genome Research Institute’s website included the following promissory note:

(G)enome-based research will eventually enable medical science to develop highly effective diagnostic tools, to better understand the health needs of people based on their individual genetic make-ups, and to design new and highly effective treatments for disease. Individualized analysis based on each person’s genome will lead to a very powerful form of preventive medicine. Through our understanding at the molecular level of how things like diabetes or heart disease or schizophrenia come about, we should see a whole new generation of interventions, many of which will be drugs that are much more effective and precise than those available today.²⁵

At the same time that high hopes were expressed for the future of genomics-powered personalized medicine,²⁵ concerns were raised that mapping and classifying people’s genomes would undermine their privacy, lead to new forms of discrimination (e.g., by employers and insurers), foster the essentialist idea that people are their genes,²⁶ bolster attempts to interpret social identities in biological terms, and trigger depression, anxiety, suicidality, and worry in individuals whose genetic risk for certain conditions was determined to be high.^27–29 In addition, some cautioned that a genetics focus would lead to lessened emphasis on the social determinants of health and health disparities among underserved groups (e.g., racial/ethnic minorities³⁰).

Since then, researchers working in different contexts and with different populations have sought to document some of the hoped-for benefits of sequencing and the feared downsides. These diverse studies have investigated both clinical (diagnostic) and screening uses of sequencing, and have assessed a wide variety of possible consequences following receipt of sequencing results for individual patients, their families, and the public. Many studies have focused on the direct impact of sequencing results on the patient’s medical care–the “clinical actionability.” However, researchers have also investigated other direct and indirect, medical, and non-medical implications of sequencing information, such as the possibility for these results to end a patient’s search for a diagnosis, often characterized as the “diagnostic Odyssey,” even if the identified condition is untreatable, or to inform the patient’s subsequent life choices, including their reproductive decisions. Of particular relevance to our focus on newborns, investigations have looked beyond the specific patient whose genome was sequenced to assess the impact on family members and the public, whether in the form of increases or reductions in utilization of healthcare services and costs,³¹ or advances in the state of scientific understanding.

Overall, neither the fondest hopes not the darkest fears of sequencing have been fully realized.³² Instead, a nuanced and complex picture is beginning to emerge of the benefits, risks, and limits of sequencing results for the health and well-being of both individuals and families, and for contributions to the public good.³³ This nuanced and complex picture led us to analyze the possible use of two kinds of genome sequencing in newborn babies for two purposes and across three contexts: targeted and whole exome or whole genome sequencing, used for diagnosis or screening purposes, in clinical practice, including both the care of symptomatic newborns (e.g. in the NICU) and routine pediatric care, as well as in public health programs, specifically state-based newborn screening programs, and as a direct-to-consumer service.

Sequencing to Diagnose Sick Newborns Can Benefit Children and Families

In 2011, Worthey and colleagues reported an exemplar that represents the best-case scenario for use of sequencing in sick newborns. Traditional diagnostic methods had failed to diagnose the cause of severe bowel disease in a 15-month old child. The child’s clinicians suspected an underlying immune defect, but they had been unable to arrive at a definitive diagnosis. The child’s exome was sequenced leading to a previously unsuspected diagnosis of an X-linked inhibitor of apoptosis deficiency, which in turn allowed successful treatment of the child with a stem cell transplant.³⁴

This case report fueled hopes that whole exome or whole genome sequencing could be similarly useful for critically ill babies in neonatal intensive care units for whom diagnosis was elusive. Thus far, reports show that genome sequencing of this carefully selected clinical population identifies variants that are plausibly associated with some or all of the babies’ symptoms, leading to “molecular diagnoses,” in approximately 40%−60% of cases.^{13, 35, 36} Yet the clinical significance of these molecular diagnoses—their impact on the medical care of sick infants—has unfortunately only rarely mirrored the case reported by Worthey and colleagues. In that case, the diagnosis triggered a treatment that cured the child’s disease. In most subsequent cases of sequencing in symptomatic infants and children, a diagnosis—if one was identified—had other, less tangible, consequences for infants and their families.

For example, Yang, et al., reported that, “For the 25% of cases that received a molecular diagnosis, this information ended the Diagnostic Odyssey, provided more informed medical management, and allowed for precise determination of reproductive risks; however, relatively few cases resulted in specific treatment to reverse the condition.”³⁷ Solomon describes the benefits of genome sequencing as follows: “Knowledge of the genetic cause of a medical condition may be beneficial related to such issues as informed medical decision making, selection of optimal supportive care, prognostic considerations, and avoidance of unnecessary testing.”³⁸ Iglesias and colleagues report that a molecular diagnosis led to discontinuation of further diagnostic intervention, assistance with reproductive planning, and identification of other family members who were carriers. But they also pointed out that it only rarely led to changes in treatment.³⁶ Where there is diagnostic sequencing reduces the need for additional medical testing in the infant and refocus treatment, perhaps even informing decisions to allow an infant to die, it can be very beneficial to families. Such testing can yield information that can help the family understand what has happened to their child.

Downsides to and concerns about the use of sequencing in this context nevertheless exist. Results can be complex, uncertain, and voluminous, making them difficult to explain to families and possibly leading to worry. Often, because of the complexities of penetrance, and the rapidity with which new data is accumulating, it is hard to know whether a genomic variant is pathogenic or benign.A^a,39 Furthermore, because sequencing yields such a vast amount of information, there is a very low signal-to-noise ratio. Biesecker and colleagues recognized the problems of information overload, “A whole-genome or-exome result is overwhelming for both the clinician and the patient… [because variants] … range from those that are extremely likely to cause disease to those that are nearly certain to be benign, and every gradation between these two extremes.”⁴⁰ This reality runs counter to the near-magical narrative of preemptive precision medicine.

The possibility also exists that sequencing for diagnostic purposes in this context could impact healthcare utilization and costs, whether or not that subsequent care is delivered to the infant or to their family members.⁴¹ Explaining results and offering follow-up care will require expertise on the part of the infant’s and the family’s healthcare professionals, which suggests that indiscriminate use could drive up costs. Recently published research led by Stephen Kingsmore has shown, however, that use of sequencing in selected NICU patients can reduce the overall costs of that infant’s care.⁴²

Because direct benefits to the child are in most cases unlikely, it is imperative that sequencing be done with the parents’ well-informed permission, meaning that the full range of impacts of targeted or non-targeted sequencing are clearly and comprehensively explained including the limits and the possible impacts on parents and other family members. Biesecker and Green have reported that “pretest counseling is particularly important, to maintain realistic expectations for finding the causative variant and to alert the patient or family that in most cases, a positive result is unlikely to change treatment or management decisions or to improve the prognosis.” ⁵ Parents should be helped to understand that the most common consequence of a molecular diagnosis in a sick newborn is the confirmation of an unalterable, bleak prognosis, a result that is not the “clinically actionable results” “or “clinical utility” that parents and healthcare professionals might hope for⁴³ but that be actionable in a broader sense because it can allow parents to make decisions about medical care.

Parents should also be informed about the possibility that testing will generate other results unrelated to the infants underlying condition, and they should discuss what they do or do not want to know about such secondary or incidental findings. It is impossible to inform parents of every possible incidental finding. Instead, they need to be informed of the types of results that may be available. For example, they might be asked whether they would want to know about carrier status, or adult-onset diseases, information that can help the family avoid the birth of another child with the same condition. Use of sequencing for a diagnosis in the context of clinical care is a situation in which it is realistic and reasonable to expect that expert follow-up and support for the infant and their family will be available, further supporting its availability, and where provision of that expert care is not expected to overwhelm healthcare services or insurers simply because so few newborns will require this level of care.

Across Contexts, Use Sequencing for Screening Purposes is Rife with Challenges

Below we consider use of sequencing as a screening tool, whether in public health newborn screening programs, as part of routine primary care, or as a direct to consumer service. We conclude that, at present, sequencing might be an appropriate tool for population screening of newborns when it is limited to genomic targets associated with conditions that meet existing criteria for inclusion in newborn screening programs. Whole exome or whole genome sequencing, however, does not meet the ethical and legal conditions for use in newborn screening programs. Neither targeted nor genome-wide sequencing can be well integrated into routine pediatric care, and the use of sequencing as a direct-to-consumer service should be discouraged.

Although specific considerations attend each particular context, some considerations regarding screening uses recur. In particular, we repeatedly come back to the variety of difficulties associated with the large volume and varied, often complex, nature of sequencing result. As long as laboratories and clinicians err on the side of returning a wide range of results, explaining these results to parents and counseling them on next steps in any context will be time consuming and therefore costly. Many results will trigger follow-up care, including additional testing and ongoing monitoring of at best uncertain utility, again resulting in costs to the health care system or to individual families. Some results, in particular uncertain results, have been shown to cause significant distress and ongoing worry.⁴⁴ In the absence of symptoms that justify diagnostic use of sequencing, identification of pathogenic, likely pathogenic and uncertain variants could transform healthy infants into the “pre-sick, ”or “patient in waiting,” inappropriately medicalizing infants.^{18, 39} The results of sequencing might also be available to and used by insurers and employers to deny coverage, increase premiums, or limit opportunities. Some of these risks or downsides might be temporary (i.e. medicalization risks could be alleviated if understanding of genomics improves), but others are likely to be permanent (e.g. potential for sequencing results to be used by insurers and employers) as well as the potential for uncertain results to cause ultimately unnecessary worry.^b

Targeted but Not Whole Exome/Genome Sequencing Could Assist Public Health Newborn Screening Programs

For over fifty years, the public health departments of US states have run newborn screening programs. Beginning in the 1960s, with the development of a simple blood test to identify phenylketonuria (PKU), states instituted population-based screening for all newborns. Over time, these programs have expanded to include other conditions that can be identified in the first days of life and, in most cases, can be effectively treated. Early detection of these conditions has in most cases allowed families to initiate interventions that can improve health and developmental outcomes for these newborns.⁴⁵ Similar programs exist in more than 50 countries around the world.⁴⁶ In the US, these programs annually identify around 12,500 newborns with a small number of serious and rare conditions.⁴⁷

In the US, each state decides which conditions to test for and which tests to use. A committee appointed by the US Secretary for Health and Human Services (the Secretary’s Advisory Committee on Heritable Disorders in Newborns and Children) curates a list of disorders that the Department of Health and Human Services then recommends be part of state newborn screening programs. This list is known as the Recommended Uniform Screening Panel (RUSP). A condition can only be added to the RUSP if the committee agrees that it meets three criteria, similar to those developed in 1968 by Wilson and Jungner to assess all screening programs.⁴⁸ In this report, we refer to these criteria as “newborn screening criteria.” They are: (1) evidence supports the potential net benefit of screening for this condition (identification of the condition early in life, before it would ordinarily be clinically detected, has been shown to benefit the infant), (2) states have the ability to screen for the disorder (a test of appropriate sensitivity and specificity is available, and states have the capacity to adopt that test and provide the necessary follow-up services), and (3) effective treatments are available (early detection, followed by intervention and treatment, has demonstrated benefits for the infant and those interventions or treatments are available).^{49, 50}

The most important criterion for Wilson and Jungner was the availability of an effective treatment: “Of all the criteria that a screening test should fulfil,” they wrote, “the ability to treat the condition adequately, when discovered, is perhaps the most important.” That principle has become a flashpoint for many advocates of greatly expanded screening, who argue that treatability is too narrow a criterion (or alternatively, that its meaning is too narrowly defined). As is evident from our discussion above, we have some sympathy with the notion that implications for families and the public need to be taken into account when assessing use of sequencing in newborns. However, we are cautious about applying that idea to the public health context, in which millions of babies would be screened each year. Most states screen for the majority of disorders on the list; and some states also screen for additional disorders. California, as one example, mandates that its Department of Public Health follow RUSP guidelines.

Newborn screening has been effectively mandatory in the US, with some states allowing limited exemptions based on religious or other objections, since the 1960s.⁴⁵ One reason is that newborn screening was established under the aegis of public health, where parental permission has never been the norm. Another was that, in the 1960s, deferral to expert knowledge was commonplace, clinicians who cared for newborns were expected to act on that knowledge. But also crucial was the fact that the programs targeted PKU, a disease with devastating consequences that could only be effectively treated if diagnosed soon after birth. Given the existence of an effective therapy for PKU, the decision to reject testing posed a risk of harm to the child and could never be in his or her best interest. Undiagnosed PKU was associated with enormous suffering for individuals and their families and also – because most of those affected were institutionalized in state hospitals – substantial expense to the state. It was only in the 1970s, with the rise of rights-based movements such as the womens’ movement, civil rights, and the emergence of bioethics as a discipline, that lack of express parental permission became an issue. But clinicians and public health practitioners warned then that requiring parental permission could undermine programs’ cost-effectiveness, that many parents would refuse permission for their infants to be tested, that the provisions would fail to achieve meaningful parental choice, and that they might be extended to other routine procedures — all arguments still invoked today in support of maintaining the status quo. Since then, justification for the mandatory nature of NBS has shifted, to center on the potential harms that could occur to a newborn if they were not screened, thus placing child welfare above parental decision making. These child welfare considerations, combined with the fact that the administration of newborn screening is maintained by state health departments as an essential public health service, has led to the routinization of newborn screening within the US health care system. In recent years and in light of scientific developments, proposals have been made to modify the criteria for inclusion of a condition in newborn screening panels. ^{6, 51}

Newborn screening programs rely on a combination of state funds and fees, which are often paid by insurance companies. Referred to as “kit fees,” these funds allow states to maintain screening and follow-up services. When adding new conditions to panels, programs must assess the impact that adding a single condition or set of conditions may have on their programs, given the technological and other laboratory resources necessary to perform the screening as well as any additional program costs that may be needed to implement the screening statewide. In addition to financial resources, programs must also assess the communication and educational resources needed to implement screening, including any changes to update state educational materials, clinician education and outreach efforts, or other programmatic resources needed for appropriate follow up.^C

One of the foundational principles of newborn screening programs is the universal accessibility of screening to all families in the US. State programs strive to assure that all newborns will be screened regardless of parental ability to pay and whether the infant is born in the hospital or at home. Overall, states are successful in promoting universal access to screening, and there is evidence that state newborn screening program can reduce health disparities.⁵² As the number of conditions screened for grows, however, maintaining universal access is an increasing challenge. For example, while state programs generally can assure that every infant is screened, the accessibility of diagnostic and follow-up services can be more difficult, and can lead to possible disparities, especially within communities with decreased access to health care. When weighing costs, programs must examine how adding conditions or new technologies to newborn screening programs affects their ability to promote equitable and universal access to screening.^{53, 54}

The number of conditions on the panels of individual states grew significantly in the 1990s, following the introduction of a new testing method called tandem mass spectrometry. This expansion, and growing concerns over large discrepancies among state screening panels, led to the creation of the RUSP and the Secretary’s Advisory Committee on Heritable Disorders In Newborns and Children. There has been concern that a number of the conditions added do not meet newborn screening criteria and that the original review process that led to the creation of the RUSP was problematic. History cautions against premature expansion of panels.^{53, 55, 56} As previous policy analyses of this process have shown, some of the conditions added in the past two and a half decades do not have accepted treatments, or have only interventions that fall outside “traditional medical treatments.⁵⁷ Others have significant phenotypic variability, this means that some newborns who screen positive actually have a very low probability of disease, and those who do have a particular condition can have widely differing ages of onset, varying disease severity, and varying response to treatments. In some cases, a positive screen exposes children or families to harms, including unnecessary diagnostic procedures, painful yet usually unsuccessful treatments, and increased stress and anxiety for parents and families as they deal with future uncertainty.^d One example of a condition that raised many of these issues is Krabbe A disease, which has been included in New York State’s newborn screening panel since 2006, following lobbying by patient advocates. The inclusion of Krabbe A disorder on screening panels has been roundly criticized because of the screening test’s low positive predictive value, defined as the number of positive results that are true positives, and the risky and rarely successful treatment options, such as stem cell transplantation.^e The incremental expansion of these panels is sometimes cited as another illustration of what social critics call the “technological imperative”: i.e., the tendency for professionals to experience a moral urgency to use new tools when they become available regardless of their conformance to accepted standards.^58–60

As state newborn screening programs consider how and whether to integrate genome sequencing, they are wary of these same problems.⁶¹ A “genomic expansion” of newborn screening could also move programs away from the original intentions of public health newborn screening, thereby calling their legal and ethical mandate into question, and could stretch states’ ability to provide the screening and follow up services. Quite simply put, the resources needed to provide genomic analysis and follow up services could break the newborn screening system.^53,6 For example, while existing newborn screening can be done for typically under $100 per infant, adding sequencing would potential add hundreds, if not thousands of dollars to the screening fees. Costs of sequencing along could overwhelm and ultimately undermine the benefits of the newborn screening system. Additionally, there would likely be concerns about the impact of use of genome sequencing on public trust in the newborn screening system if parents worry about long-term storage and potential sharing of sequencing results that include their child’s unique DNA signature.^{62, 63}

The very existence of newborn screening programs is threatened by expansions of those programs that undermine the legal and ethical basis of the programs or that might, as a practical matter, overwhelm the programs or prevent them from carrying out their commitment to providing universal screening and follow-up services. With this threat in mind, our workgroup identified two ways that genome sequencing could be used by newborn screening programs: to sequence one or more specific genes (targeted sequencing) or to sequence entire genomes (genome-wide sequencing). Use of targeted sequencing—as a primary or secondary test—in newborn screening programs can be justified provided it meets existing newborn screening criteria. Because genome-wide sequencing does not now meet those criteria, its use cannot be justified at this time.

Targeted sequencing would essentially be a new way to identify conditions that meet newborn screening criteria but are not yet screened for because they cannot be adequately detected using existing screening technologies. For example, genetic analysis is necessary to make a distinction between early and late-onset Spinal Muscular Atrophy, even while it is not generally needed when screening for conditions such as PKU. Given this fact, the proper question to ask is probably not, “Should all newborns have their genomes sequenced at birth?” but “Are there conditions for which all newborns should be screened and, if so, when is genome sequencing a better way to do such screening than other available tests?” If targeted to appropriate conditions, sequencing would meet the legal and ethical justification for mandatory screening, including waiver of parental permission, and the use of public health funds. Genome-wide screening (which, we refer to as “non-targeted sequencing”), by comparison, would return many results that do not meet newborn screening criteria, making it impossible to justify under existing legal and ethical criteria for public health newborn screening programs.

Targeted sequencing itself could be used in at least two ways. Targeted sequencing could also be a first line-screen, bundled in with other first-line newborn screening modalities (e.g. metabolic tests, hearing tests, tests for severe combined immunodeficiency, etc.). Or it could be a secondary screen that is used only following a positive or ambiguous initial result or a first-line test.

Use of sequencing as a first line screening modality would allow programs to screen newborns for conditions that cannot be detected with existing biochemical screening tests but that otherwise meet the criteria for inclusion in public health newborn screening programs. In addition to the other criteria for expansion of newborn screening—net benefit from early detection and availability of effective interventions—the ability of programs to administer the test and to adequately follow up on results is a key consideration given the volume of results that even targeted sequencing could yield. The addition of even targeted sequencing could disrupt the current benefits these programs provide to newborns, if such sequencing overwhelmed the financial and personnel capacities of NBS programs. Interviews with state health officials conducted by one of the working group members reveal great concern about potential burdens.⁶¹ Targeted sequencing cannot yet—and might never be able to—replace all existing NBS methods, since sequencing cannot identify many of the conditions that metabolite analysis and other kinds of tests can identify (because identifying such conditions requires the observation of some metabolic effect or other phenotypic trait, such as accumulation of a harmful amino acid in PKU). This means that programs must consider the utility of maintaining multiple test modalities.

Secondary use would allow programs to identify genotypic variations, or particular mutations, that could help reduce the “diagnostic Odyssey” many parents experience once a positive screen is found. Additionally, having more detailed genomic information may help refine potential treatment plans by giving primary care professionals and specialists more accurate genotype-phenotype information related to disease characteristics, such as severity or age of onset. One example is California’s use of sequencing to follow up a positive screen for Cystic Fibrosis, by sequencing the full CFTR gene.⁶⁴ Furthermore, some genetic conditions are not revealed by tandem mass spectrometry, the currently dominant technology.

A number of states have already implemented targeted genome sequencing as a secondary test where a better understanding of the newborn’s genotype is likely to improve treatment decisions and promote better outcomes for the newborn. The CFTR gene is sequenced both to confirm the initial screening test and because information about the variant might inform potential phenotypic outcomes for that newborn and refine treatment decisions.^f Others have raised concerns about false positive results, including situations in which parents are given results, the infant is monitored, but symptoms do not manifest. For example, sequencing reveals information about a new “condition” called CFTR-Related Metabolic Syndrome.⁶⁵ An educational booklet notes: “We cannot clearly predict the future health of your child, although he or she is likely to remain healthy.”⁶⁶

In addition, the concerns of some parents about the impact of genomic data on their children must be considered. These concerns include the potential psychosocial harms that arise when results are uncertain or ambiguous. This issue is magnified in children, particularly in variants associated with phenotypes difficult to identify in infants, including developmental delay and intellectual disability. Additionally, parental concerns about genetic discrimination or reservations about government agencies having genetic information about their newborns could prompt them to opt out of screening altogether. Addressing these issues could require changes in law and policy, and will require thoughtful consideration by public health officials and clinician leaders about what kinds of genomic information should be returned to parents and how best to communicate that information.

Genome-wide sequencing could enable newborn screening programs to look for a much wider range of conditions than they screen for today, including increased risk for later-onset conditions and conditions that are untreatable.⁶⁷ The potential for expansion is breathtaking; resources such as Online Mendelian Inheritance in Man (OMIM) catalogue over 15,000 genetic disorders. And new applications, such as pharmacogenomics testing, are part of the call for Precision Medicine. Use of the technology in this way would amount to a substantial—indeed unprecedented—expansion of newborn screening programs. This expansion cannot be legally or ethically implemented without major alterations to state-mandated newborn screening programs because whole exome or whole genome sequencing would produce results that do not meet the criteria needed to justify dispensing with parental permission.^g,68 Altering programs to require formal parental permission raises concerns that parents will opt out, thereby undermining the universality of the newborn screening system.^{53, 69} While US studies of the impact of requiring parental permission for NBS have not shown a dramatic impact on participation⁷⁰ and while an increasing number of jurisdictions worldwide require permission – in some cases, such as Scotland, even explicit written permission – without any demonstrated negative impact on participation, those studies and programs are not screening for the huge number of conditions that whole exome or whole genome sequencing could be used to detect. In addition, a change in US policy to require parental permission would require staff time and documentation, adding cost to programs and increasingly the possibility that some newborns would receive no screening whatsoever.

Second, non-targeted sequencing would produce results that are not associated with childhood disorders amenable to medical or other interventions, thereby undermining one critical justification for spending public health funds on these programs. Third, unless significantly more money were devoted to newborn screening programs, states would be unable to conduct genome sequencing and, even if the cost of sequencing drops, unable to afford to provide follow-up counseling and other interventions to families for all the conditions that would be identified.

We recognize that our recommendations are much more caution than at least one alternate view, which sees the advent of genome sequencing as an opportunity to completely revisit the criteria used to govern newborn screening policies and practices and the laws that underpin them. On this view, the programs could be radically changed or a separate newborn sequencing program could be established, perhaps as a component of primary care (see below). Express parental permission could be sought and significantly more public funds could be apportioned so that universal—or near universal—whole exome or whole genome sequencing could be offered. Such an expansion would move programs away from their current ethical justifications and the legal bases for their near mandatory nature. Our view is that we are far from this moment. At this point, the major risk is that premature or overly broad use of sequencing technology in newborn screening programs could overwhelm existing programs and undermine their public and political support, thereby placing at risk the good work those programs accomplish today.

Sequencing Should Not Become a Routine Part of Infant’s Primary Care

Perhaps sequencing could instead be used as a screening tool in routine pediatric care? Research at the University of North Carolina is testing such an approach, one that could be implemented by pediatricians or nurse practitioners in the context of routine care. Even though there is greater potential in this context to obtain fully-informed permission from patients, concerns persist about the use of scarce medical resources, the availability of a suitably trained workforce, and the likelihood of causing confusion and worry for parents caution against the use sequencing as a screening tool in primary care.

Some of these concerns will likely change over time. In particular, the results of sequencing will undoubtedly be better understood, and more pediatric healthcare professionals will likely become knowledgeable about and able to communicate the meaning of results. Yet, sequencing will need to show significant medical benefit in order to justify the significant resources that this kind of screening entails (e.g., an insurance company will require real clinical benefit from sequencing even if the results are no longer as uncertain as they are today).⁷¹ Coverage might be spotty and uneven, leading to health disparities. In addition, ongoing concerns about the impact of sequencing results on the child’s access to educational, employment and insurance opportunities will need to be addressed. More subtle concerns, including the potential medicalization of a person’s life based on genomic information and the consequences of increasing surveillance of each persons genome and health, must also be addressed.⁷² Behavior that previously might have been found simply odd, might be attributed to faulty genes; the genomic embodiment of parental expectations may subtly shape a child’s future possibilities.

Furthermore, the psychological situation of the parents is very different. While a sick baby creates a family crisis, with parents usually anxious and stressed, none of that is true in the context of a clinical visit for a healthy baby. Screening use of sequencing would not address an existing problem or alleviate worry and concern due to actual symptoms. Instead, it could create anxiety where there was none. In addition, screening use of sequencing is likely to require significant societal resources-including follow-up diagnostic testing and patient counseling about results – that are in scarce supply. Other benefits of sequencing, which factor into our endorsement of sequencing in diagnostic situations and possible return of an expanded set of results, such as a result that suggests genetic risk in a parent, are not enough to argue for introducing routine sequencing in healthy infants, especially in light of the resources needed to do so. For all these reasons, we do not endorse the use of genome sequencing in healthy newborns and caution against its widespread use as a screening tool in pediatric care.

Clinicians Should Counsel Parents Against Direct-to-Consumer Sequencing of Newborns

Direct-to-consumer (DTC) genetic testing emerged in the early 2000s as a market-based option to obtain genetic information without a clinician intermediary and as part of a broad drive for consumer-driven health care. Initially marketed by firms purporting to tailor lifestyle and nutritional guidance to an individual’s genetics, the offer of direct-to-consumer genetic testing for health risks was met with controversy and regulatory uncertainty.^73,74 Over the intervening years, researchers have begun to examine concerns over the reliability, usefulness, and psychosocial and health-system implications of DTC testing.^75–78 While regulatory agencies involved, such as the US Food and Drug Administration (FDA), initially took action to oversee the quality of the tests being offered,⁷⁹ they have more recently sought to ease access to tests.⁸⁰

Today, DTC companies offer health-related genetic risk predictions for adult-onset conditions such as Alzheimer disease, Parkinson disease, and hereditary thrombophilia, as well as paternity and carrier testing, ancestry testing, sports aptitude testing, expanded newborn screening, and the provision of raw (un-interpreted) genomic data that customers can take to third-parties for analysis. These companies use a range of technologies, from single nucleotide polymorphism (SNP) genotyping, which examines variations at a specific locus of a select gene, to whole genome sequencing. Some provide access to genetic counselors or other clinicians for assistance in the interpretation of results. New hybrid models of direct-to-consumer genetic testing may be on the horizon, including firms that incorporate consumer-initiated genetic testing with expert interpretation, and place the opportunity for genomic data sharing more explicitly in the hands of the consumer.⁸¹ Demand for these services appears to be increasing, with the value of the direct-to-consumer testing market, including genetic testing, projected to rise from $15 million in 2010 to $352 million by 2020.⁸²

Companies’ policies toward testing infants and children vary widely, with some websites explicitly stating that tests were intended for adults only, while others, ostensibly targeted toward adults, remained silent on whether samples from minors would be processed if submitted. Still other companies note that while their services are not directed toward minors, parents or guardians can set up an account and order a test on behalf of their child, assuming full responsibility for information security and accuracy.⁸³ In practice, companies are probably unable to maintain a chain of custody over samples from collection to processing and have no way of verifying the age of the person whose sample they are testing.

Proponents of DTC genetic testing frame the service in terms of democratization, autonomy, and consumer empowerment, arguing that individuals have a right to their genetic information. When it comes to testing children, some argue that responsible parents should avail themselves of all available information to assure their children’s health.⁸⁴ Critics argue that users of DTC testing – and particularly children – are placed at risk of, “inaccurate results, inaccurate interpretations, potentially harmful interventions…altered family dynamics…issues of privacy, self-determination, and disclosure vis-à-vis parents and children,” when they access test results without the involvement of clinicians.⁸⁵ They question whether DTC genetic testing can deliver – scientifically and technologically – on its promises to both accurately screen individuals for disease risk and move towards population-level preventative genomics. They also warn that the market narrative surrounding DTC testing might undermine trust in clinicians and create a “risk society” based on genetic surveillance, where individuals bear the burden of managing uncertain genetic information across their lifetimes.⁸⁶

The implications of direct-to-consumer genetic testing in the pediatric population are still unknown. Empirical research in adults indicates that DTC genetic testing may be neither as beneficial nor as harmful as originally thought, and that receiving personal genetic results may in fact temper the high expectations that some consumers have of the tests.^{76, 87} Study findings on the percentage of adult consumers who seek genetic counseling following direct-to-consumer genetic testing range from 1–14%.⁸⁸ However it is unclear how true any of these findings will be for children, and studies have not yet examined parental motivations for using DTC or the downstream health system implications of parents accessing these services.^{76, 88, 89}

Timmermans and Buchbinder’s concept of “patients-in-waiting”, developed in relation to expanded testing in public health newborn screening programs, provides a lens through which to examine the uncertain utility of pediatric DTC genetic testing.¹⁸ While the services espouse autonomy – the individual’s (or in this case the parents’) right to access and utilize genetic information, the result could be what Siddhartha Mukherjee describes as increased societal anxiety due to “constant diagnostic surveillance,” leading individuals to feel that they are “ under siege from the future.”⁹⁰

Our assessment of the use of DTC sequencing in newborns mirrors the issues we addressed around the use of sequencing in other contexts, with the added complexity that parents would receive sequencing results outside the healthcare system. While the rhetoric of precision medicine and the vision of population genomic screening advertised by direct-to-consumer firms have encouraged participation, we note that the value and benefit of genetic risk screening in a healthy population remains largely hypothetical.⁹¹ We also note that use of DTC genetic testing in children conflicts with clinical and professional guidelines and recommendations, which focus on the best interest of the child and recommend limiting testing to clinical contexts and for conditions that manifest during childhood.^{14, 85, 92} While counseling and follow-up on results would almost certainly require assistance from a clinician who can examine the infant or with whom they have an ongoing care relationship, those services are generally not built into the direct-to-consumer experience. The potential for results to generate worry and confusion is therefore exacerbated, and the chance of benefiting health and well-being is lessened. If and when parents do turn to their healthcare professionals, interpreting and managing results will generate additional burdens on the health care system. Recognizing that the value of direct-to-consumer genetic testing may truly be in the aggregation of data across large numbers of customers, and given the shifting regulatory terrain governing genomic big data, we recommend that health care professionals discourage use of DTC sequencing by parents.

We are not arguing for a law against parents accessing DTC sequencing of their newborns. Such a law may be inconsistent with the broad discretion that parents have to make decisions for their children, and would in any case be a blunt and overly intrusive way to manage this new technology. Rather, we recommend that parents pursue sequencing only if their clinicians recommend it and only inside the care relationship, so that they may be properly counseled before and after testing, and so that follow-up services may be integrated into their infant’s care.

Parents may need help to understand the potential downsides of sequencing. Market forces and the technological imperative almost guarantee that the benefits of sequencing will be hyped and the downsides will be discounted by at least some DTC companies,⁹³ creating a significant risk that parents—and clinicians—may be misled. In light of these market forces, we further recommend that clinicians refuse to advertise or otherwise promote sequencing to parents (for instance, clinicians should refuse to display advertisements for sequencing in patient waiting rooms). Questions from parents about DTC services are an opportunity for clinicians to understand and address the concerns or hopes that are driving parents’ interest, and to educate the parents about evidence-based ways to address those concerns. Perhaps diagnostic sequencing will be indicated due to family history or symptoms in the child, in which case the clinician could recommend targeted sequencing or whole genome or whole exome sequencing as part of clinical care. Often parents do not understand what is already screened for as part of newborn screening or routine pediatric care, in which case the clinician can educate the parent about these aspects of pediatric care, to reassure them that their child is receiving appropriate care.

We recognize that strictly precautionary admonitions may prove insufficient in the face of rapid industry growth and consumer demand. At the same time however, the era of Big Data has brought to the foreground longstanding issues around privacy and data control that society must grapple with – and which are particularly sensitive when decisions are being made on behalf of children unable to assent to testing. If parents ignore this advice and proceed with DTC testing, they will at the very least need significant guidance from testing companies and very likely from their child’s health care professionals, so that they can understand and act on their results. Insurance companies and other payers will rightly question the appropriateness of these DTC driven uses of medical resources. By highlighting such challenges, we intend for our recommendations to give consumers, clinicians, parents, and firms the opportunity to pause to consider the societal implications of sequencing newborns as part of commercial testing.

Conclusion

A nuanced approach to the nascent technology of genome sequencing is responsive to the different reasons for using sequencing, the different contexts in which that use can occur, and the emerging data about how those different uses impact newborns, families, and the public good. This approach resists the naïve notion, fueled by both unbridled scientific optimism and venture funding, that genomic information is an unqualified “good.” Instead, we recognize that genomic sequencing is a technology like all others. It is not a panacea. It has harms as well as benefits. If used carefully, it may help improve the health and well-being of newborns and their families. If used carelessly, it may disrupt family dynamics, waste medical resources, and undermine public trust.

We should not exaggerate the risks or the benefits of its use for different purposes and in different contexts. Rather, we should strive to get the balance right. At the present time, that balance clearly supports the careful use of either targeted sequencing or whole exome/whole genome sequencing to assist in the diagnosis of sick infants. It does not support genomic sequencing of all babies at birth.

In reaching this conclusion, we took careful note of both the limitations of sequencing results when used for asymptomatic babies and of the ways that sequencing technologies might affect existing and beneficial state-run newborn screening programs. We concluded that the introduction of whole exome or genome sequencing could destabilize these programs. It could thus lead to two specific and worrisome harms. One would be to families who were given ambiguous or misleading genomic results. The other would be to the existing programs that might crumble under the weight of a new mandate and thus deny families the long-standing benefits of these programs. For newborn screening programs, even narrowly targeted use must follow long-standing norms and values that protect infants and families and that strive to assure equity in implementation and follow-up.

Furthermore, we agreed that based on today’s knowledge base, admittedly in rapid flux, clinicians should not offer non-targeted predictive sequencing to the parents of healthy newborns. Based on what we know today, we believe that, for healthy newborns, the potential risks of false positive or merely ambiguous or uninterruptable results outweigh the potential benefits of probabilistic predictive information. This is even more true for direct-to-consumer testing, in which families do not even have the benefit of clinical expertise in interpreting results.

Many people will, no doubt, find our recommendations unduly pessimistic. Proponents of universal testing often articulate what is, in essence, a research agenda. They acknowledge that we don’t yet know enough to warrant such testing, but claim that the only way to gain the knowledge is to do such testing. We agree that testing of healthy newborns should, at this point, be considered a valuable research project. But the results of the research are not yet in. We are thus convinced that, at present, a nuanced approach to this technology is required. It is precisely because of our deep respect for the power and promise of a “precision” approach to the care of newborns that we advise further reflection and debate as we determine, through careful research, the best applications of sequencing technology in the clinic and in public health.

BOXED TEXT

Some Detail About Sequencing and the Interpretation of Results

Genome sequencing is the process of determining the order of base pairs in an organism’s DNA. Today’s genome sequencing relies on next-generation sequencing technologies, which allow for rapid sequencing of large amounts of DNA. Mapping the sequence of the base pairs that make up DNA is an important first step in an enormous task: understanding the relationship between genes, particular variants, and traits. That understanding, while certainly more advanced than it was when the first drafts of the human genomes were sequenced fifteen years ago, is still very much incomplete. Indeed, comparing different people’s genomes, analyzing the relationship between their genetic differences and differences in their health, and using that information to improve medical care is the ongoing work of translational genomics research.

Sequencing the whole exome or whole genome of a human yields a huge amount of data. The human genome contains 3 billion base pairs and its exome (the subset of DNA that codes for proteins) has about 30 million base pairs. In its raw form, a whole exome or whole genome sequence is just a huge string of genetic letters. The accuracy of the test—how precisely it reports the individual’s actual genetic make-up—is known as “analytic validity.”⁹⁴ Establishing analytic validity is just the first step, however. To begin to make sense of that string of DNA, the sequencing data must be interpreted or analyzed. The individual’s sequence is compared with one or more reference genomes, and any differences—known as gene variants—are then investigated by consulting one or more constantly growing databases that contain reports and studies linking different gene variants to health conditions or other traits. Publicly accessible databases ensure that all known cases are shared as widely as possible. For some variants, hundreds or thousands of cases have been studied and the impact of the variant is well understood. For others, just one or two cases are recorded, while some gene variants are observed for the first time. Where the meaning or significance of a variant is unknown or highly variable, the implications of that variant for the person who carries it are uncertain. Such uncertainty can be especially acute when the person sequenced is an infant who is not yet showing any signs or symptoms.

Based on the information available in reference databases and in the individual patient’s medical record, laboratories then classify each variant. Classification systems vary somewhat. In general, when the evidence is strong that a particular variant is associated with a disease or disorder, it will be classified as “pathogenic” (note that this term is used even when the variant increases risk for, rather than guarantees development of, a particular disease or disorder). Sometimes the evidence is strong that a variant is “benign.” Where understanding of the impact of a variant is less strong, it can be classified “likely pathogenic” or “likely benign.” Where the evidence is unclear, the variant will be classified as “uncertain/unknown significance”.⁹⁵ Although the classification of certain variants is widely agreed upon, research shows that different labs classify many of the same variants differently.⁹⁶ The evidence base underlying variant classification is constantly evolving. Variants may be re-classified, raising the need for laboratory policies on re-analysis and re-contact.⁹⁵

The degree to which a particular variant actually is associated with a particular disease or trait is known as that result’s “clinical validity.” A number of factors contribute towards the uncertainty that surrounds the meaning of many sequencing results. While a rare number of genetic differences are widely agreed to lead to particular syndromes, conditions, or diseases, there is still variability—and therefore uncertainty—in exactly how the individual will be affected by that genetic difference. This variability in how a gene is expressed, referred to as “variable penetrance,” can include differences in symptoms, variable age of onset, variable severity, and variable response to treatment. And in some cases the disease might never develop. Further challenges for variant interpretation result from genetic, epigenetic, and environmental interactions in complex conditions.⁹⁷ The clinical validity of a given genetic result—the degree to which that particular variant actually is associated with that particular disease—will vary based both on how the test is used (e.g. diagnostic versus screening test) and on characteristics of the disease and population tested (e.g. penetrance of the genetic trait and prevalence of the disease being tested for in the population). ⁹⁸

In a final step, the results are returned and the person who receives the interpreted sequence (often known as the “results” or “findings”) considers what can be done based on the results. In clinical contexts, a physician or other healthcare provider will consider whether the results might change the treatment plan for that the patient. The degree to which a molecular diagnosis leads to a beneficial change in management is known as that result’s “clinical utility.” While different studies classify clinical utility differently, most report that about 30–50% of results have some clinical utility.^{13, 99}

Various results might be considered useful by the patient or their family to inform non-medical decisions (an idea captured by the term “personal utility” and related to our discussion of benefit above). In the US, the American College of Medical Genetics and Genomics (ACMG) has recommended that a subset of results, which indicate the presence or risk for certain clinically actionable conditions, be returned whenever a person has their genome sequenced as part of clinical care.^{11, 100} ACMG explicitly noted, however, that their recommendations do not apply to use of sequencing in newborn screening or any screening use of sequencing in healthy children.

Why interpreting sequencing data is so challenging and how it is evolving as more sequencing is done

Three elements of genomics lead to many of the ambiguities in interpretation.

1. Many variants are rare or unfamiliar and must be classified based on very limited data. Sometimes the function of a gene is not well understood, meaning that every variant of that gene is potentially of unknown significance. Furthermore, as new data become available, classification may need to be revised. Variants that were once thought to be pathogenic sometimes turn out to have been misclassified, and vice versa.

2. Many variants, including those classified as pathogenic or likely pathogenic, have reduced or incomplete penetrance, which means that not all those who carry the variant will develop the disease or trait associated with it.

3. Many variants have variable expressivity, which means that people with the same genetic variant have a range of signs and symptoms of different severity.

4. Data bases used for interpretation of variants are derived primarily from individuals of European ancestry, who are overrepresented in studies, but do not represent the general population.¹⁰¹ Individuals from diverse ancestral backgrounds often have variants, both benign and disease-associated, that cannot be easily interpreted given that gene frequencies vary across human populations.

5. In newborns, interpretation of variants is particularly challenging because for most babies very little phenotypic information is available.

About this Project and the NSIGHT Ethics and Policy Advisory Board

In 2010, the NIH sponsored a workshop to plan future research on newborn screening in the genomic era.¹⁰² In response to the workshop’s report, in August, 2012, the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) and the National Human Genome Research Institute (NHGRI) issued a funding opportunity announcement for studies designed to “explore the implications, challenges and opportunities associated with the possible use of genomic sequence information in the newborn period.”¹⁰³ The institutes specified that each proposal should have three components: acquisition and analysis of genomic data about newborns; clinical research to advance understanding of specific disorders identifiable by DNA-based analysis; and study of the ethical, legal and social implications (ELSI) of the possible use of genome sequencing in newborns. The institutes’ overall goal with this program was to better understand how genome sequencing could improve both clinical care of sick newborns and the screening of healthy newborns for genetic diseases. Four projects were funded; a consortium, including the four sites and NIH staff, was established.¹⁰⁴ Called the NSIGHT consortium (Newborn Sequencing in Genomic Medicine and Public Health) two of the four sites focused on the use of genomic testing in clinical contexts, while two focused on questions that arise when sequencing is used for population screening.³⁹ The UCSF project included a working group to examine the ethical and policy issues arising from the application of genome sequencing in newborns. Comprised of representatives from each research site, as well as invited experts from outside NSIGHT, the “Ethics and Policy Advisory Board” (EPAB) held monthly conference calls and three (two-day) face-to-face meetings conducted over three years. A range of expert stakeholders and a parent representative made presentations designed to provoke discussion. The EPAB also made a series of presentations to the steering committee of the NSIGHT consortium, to obtain feedback from scientists and clinicians. Through an iterative process of reflection and writing about a complex and rapidly changing field, the EPAB debated the issues. This paper reflects the results of that process. We offer it, along with comments by many of the NSIGHT investigators, as a way of informing the policy discussion around genome sequencing in newborns.

Table 1:

Summary of EPAB recommendations

Clinical Contexts
1.	Targeted or whole exome/genome sequencing may be used to assist in diagnosis of symptomatic newborns (e.g. infants in neonatal or pediatric intensive care units or under the care of specialists), with parental permission and with access to genetic counseling and follow-up services.
	Results unrelated to diagnosis of the infant may be returned to families if those results could benefit family members.
2.	Targeted or whole exome/genome sequencing should not be used as a screening tool given the limited usefulness of sequencing in healthy populations, concerns regarding storage of results, potential discriminatory or insurance uses, and the potential for results to generate unnecessary distress and require health resources for follow-up care, monitoring, and counseling.
Public Health Context
3.	Whole exome/genome sequencing should not be used in state-sponsored newborn screening programs.
4.	Targeted sequencing may be used in two ways:
	• Following a positive screening test: as a secondary test to help diagnose conditions that meet existing newborn screening criteria: i.e., conditions that are best identified in the newborn period, for which newborn screening programs can afford to provide screening and follow-up, and that have effective and available treatments.
	• As a primary test to screen for conditions that meet existing newborn screening criteria but are currently not included in newborn screening programs, where sequencing is either the more appropriate or only method for screening for that particular condition.
Direct to Consumer Context
5.	Parents should not use direct-to-consumer sequencing (targeted or whole exome/genome) for either diagnosis or screening of their newborn.
	Healthcare professionals should discourage use of direct-to-consumer newborn sequencing services by parents.