Newborn Sequencing in Genomic Medicine and Public Health

Diagnostic Sequencing in the NICU

Level III and IV NICUs care for many neonates with genetic disorders, including metabolic disorders²⁰ and congenital malformations.^21,22 Indeed, genetic disorders and congenital anomalies are the leading cause of death in the NICU.²³ Current approaches for diagnosis of suspected genetic disorders in NICU patients include karyotyping, chromosomal microarrays, single-gene testing, and gene panels. In many cases, the etiologic diagnosis remains elusive despite several rounds of genetic testing, and clinical genome-scale sequencing is used for only a small fraction of cases. If genomic sequencing were used earlier in the diagnostic process, it could provide more timely definitive diagnoses and thereby increase the precision of treatments, whether therapeutic or palliative, and provide answers about prognosis and family recurrence risk. In addition, sequence data could aid in interpretation of the false-positive NBS results commonly reported in premature infants because of their immature organ systems, liver enzymatic activity, long-term parenteral nutrition requirement, and other comorbid conditions.²⁴

Preliminary studies have shown potential cost reductions resulting from genomic sequencing in neonates with genetic disorders.^7,25,26 However, uptake has been impeded by concerns about clinical interpretation of ambiguous results, secondary findings and potential parental anxieties, economic factors including resistance on the part of third-party payers in the absence of definitive data on clinical utility and cost-effectiveness, and difficulty ordering tests due to institutional concerns over high costs. The process is also complicated by a need for detailed phenotypic information for optimal analysis by diagnostic laboratories. Provisional genomic diagnosis of genetic disease is technically feasible in as little as 26 hours.^27,28 However, cost is inversely related to the turnaround time, and the optimal balance of cost and clinical utility is unknown. Although some aspects of rapid turnaround time can be improved through technology, limits imposed by variant review and clinical interpretation standards will remain, including the need for communication between the clinical team and the laboratory for clinical genomic assessment. Parental acceptance of testing, given the unclear future implications for insurability and privacy, is not yet known. Finally, the responses of neonatologists and parents to genomic information warrant more study, and rigorous frameworks for translating such information into precision care plans in NICUs remain to be developed.

Preventive Sequencing in a Public Health Setting

Current NBS yields few false-negative results but does incur substantial numbers of false positives, with attendant emotional and financial costs.²⁹ For example, in a study of 176 186 specimens screened by MS/MS, there were 51 true positives, 2 false negatives, and 454 false positives that were ultimately resolved as nondisease after referral to a metabolic center.³⁰ Genomic sequencing could function as a multiplexed second-tier screen, increasing the specificity of current MS/MS screening tests by distinguishing false-positive results from true disease and aiding in the differential diagnosis of nonspecific biochemical profiles. Sequencing could also confirm conditions identified through other screening methods,^31–33 aid in providing prognosis and appropriate treatment,^34–37 determine the etiology of conditions identified through point-of-care testing,³⁸ and provide families with information about pathogenic variants that could be used for family testing. Such analyses could also increase knowledge of correlations between genotypes and phenotypes and might reveal possible genetic contributions to false positives, such as abnormal MS/MS screening data due to carrier status or hypomorphic variants. These goals will require longitudinal follow-up to obtain clinical data and examine genotype–phenotype correlations to ultimately determine the predictive capacity and clinical impact of genetic variants.

Genomic sequencing could also be deployed as a first-tier screen, particularly for rare disorders that currently lack methods for conventional biochemical NBS. In the public health context, selecting exactly which disorders to screen for requires careful consideration of factors such as age of onset, severity, penetrance, treatability, confirmatory testing, and opportunities for surveillance. Some genetic conditions will fulfill the original Wilson and Jungner criteria for screening,^39–41 but many will not. Thus, there is a need for scalable methods to determine which conditions would be appropriate for inclusion in a public health screening setting. In addition, the current practice of returning findings consistent with carrier status might be unsustainable given that nearly every person is likely to be a carrier for a handful of rare recessive conditions.

A significant challenge in the use of sequencing for NBS is the lack of data regarding the analytic and clinical performance of sequencing as a predictive test. Little is known about the positive or negative predictive value of genomic sequencing in asymptomatic people whose previous probability of disease is very small. Although it may seem self-evident that Mendelian diseases would be best identified from genetic sequence, this assumption remains to be demonstrated and may not be true. Monogenic illnesses are often influenced by additional and as-yet-unidentified genetic or environmental factors, whereas MS/MS measures analytes that are typically closer to relevant phenotypes. Even for the best-studied diseases, there are substantial challenges in interpreting rare variants. Variant selection algorithms that maximize sensitivity necessarily sacrifice specificity, leading to increased false-positive results and potential downstream harms due to unnecessary medical interventions. A strategy of excluding “variants of uncertain significance” (which by definition have poor predictive value) from genomic screening results may be necessary. Historically, the focus of NBS on preventable disorders has led to low tolerance for false-negative screening results. With genomic sequencing it might be necessary to shift the screening paradigm from finding all affected individuals to finding an optimal proportion of cases for a larger number of potentially treatable conditions.

Predictive Sequencing of Newborns in Genomic Medicine

The broadest vision of genomic medicine, and potentially the most challenging for societal and practical reasons, involves the use of sequence data to guide a patient’s care throughout life. Genomic sequencing reveals information well beyond the scope of conventional NBS. Some of this information could result in medical action, but most will not, raising questions of exactly what information should be reported and when.^42,43 Variant data could conceivably be held for future diagnostic analysis in the event that the patient develops symptoms of a genetic condition. Currently, our ability to interpret genetic variants is largely confined to simple monogenic and oligogenic conditions. As we learn to use multifactorial models for risk stratification or management in a clinically useful way, additional information will increasingly be available. Within this spectrum is the potential for genomic information to alert clinicians to reconsider the family history or interpret physical examination findings in a new light, and potentially the ability to benefit other family members before they develop a disease. Pharmacogenomic variants could guide the real-time selection and dosing of medications, yielding safer and more effective treatments. Recessive carrier traits detected in newborns could alert parents to genetic risks that provide information valuable to reproductive planning. Common variation for complex conditions may motivate families to be more vigilant about diet and other lifestyle choices. Finally, there is the potential for voluntary personal exploration of one’s own genomic data.

If genomic sequence data are available, parents will need to make decisions about whether to learn about additional categories of information that may predict future events about their child with differing levels of certainty and ability to intervene, ranging from childhood-onset conditions that may not have direct interventions or preventive measures, adult-onset medically actionable conditions, or carrier status for recessive disorders. Studies suggest that parents are interested in their child’s genetic variants, even when that information has no defined clinical utility,⁴⁴ although these preferences have largely been elicited in hypothetical scenarios and may not reflect real-life choices.

Genomic information may enable families to become aware of otherwise unsuspected familial risks, including potentially actionable adult-onset conditions in the infant that a parent is unknowingly carrying. However, some findings may be at odds with professional guidance that genetic testing in asymptomatic minors should generally be done only when identification before adulthood is needed to prevent harm and directly benefit the child.^11,13,17 The potential to query genome-scale data in children for secondary findings has elicited vigorous debate over the ethical boundaries of return of results. The argument has been made that benefit to family (eg, a parent who is unknowingly at risk) may be a valid consideration in decisions regarding return of results for actionable adult-onset conditions in children.^45,46 Genome-scale sequencing and analysis in newborns would probably require modification of current informed consent procedures^41,47 compared with a targeted screen focusing on a restricted number of conditions. If implemented across the entire population, genomic sequencing would fundamentally alter contemporary public health NBS procedures, necessitating innovative approaches to facilitate parental decision-making. Alternatively, such testing could be implemented through voluntary, out-of-pocket testing.

Expanding genomic sequencing in newborns to include a broad range of conditions demands a close partnership between clinical providers and parents, similar to other areas of medicine in which shared decision-making is becoming the norm.^48,49 Determining criteria for disclosure of information will be a challenge for clinicians and policymakers and will require development of decisional supports to help parents determine the information they want to learn.⁵⁰ The clinical interactions needed for support of parental decision-making would move this activity beyond the realm of current public health service provision and into the clinical domain. Workforce shortages may present significant obstacles if genome-scale sequencing were to become widely available in the public health setting,⁵¹ suggesting that new and more scalable materials and procedures for communicating the potential benefits and risks of learning such information would need to be developed, validated, and deployed. Systems may need to be established that enable parents to request certain results from their child’s genomic information over time, allowing them to decide iteratively as their values and perceptions of risk change or as their child attains an age to assent or consent.

The complexities of genomic result interpretation currently demand trained geneticists and genetic counselors to provide guidance and follow-up management. As genomic medicine becomes more mainstream in health care, a broader range of health care providers will need to interact with genetic information, which is likely to be increasingly viewed as 1 of many risk factors influencing future conditions; reports will have to be constructed with clarity that makes them useful for pediatricians and primary care providers. The potential use of this type of genomic information over time would necessitate development of new infrastructure to manage reporting, reanalysis, storage, and integration with electronic health records. Such data could be used for iterative phenotyping of the individual to define the clinical relevance of genetic variants. However, thresholds for reporting or acting on potentially relevant variants would have to be calibrated against the possible harms of false-positive results or overdiagnosis, which would lead to unnecessary, dangerous, or costly medical treatments.⁵² The benefits of detecting true positives must therefore be balanced against the magnitude of harms.