The ethics of predicting autism spectrum disorder in infancy

Towards the end of a routine check-in appointment with your young patient—a 3 y/o boy recently diagnosed with autism spectrum disorder (ASD)—his mother shares concerns about his infant sister, currently 6 months old. The mother is aware that her daughter is at increased risk for ASD. She requests an MRI scan of her infant’s brain, based upon research she has read showing that MRI can be used to predict which infants will go on to develop ASD. The mother communicates that she is eager to know whether her daughter is going to develop autism so that she and her husband can prepare financially, and so she can place her daughter on the long waitlist for autism-specific services in her local community. As this family’s psychiatrist, how should you respond to her request?

This hypothetical scenario is anticipated in the next 5–10 years, as researchers work to develop and validate novel methods for predicting ASD prior to the onset of symptoms. Using MRI, for example, researchers have been able to detect structural and functional brain changes in the first year of life in infants at high familial risk for ASD that predict which of these infants will go on to develop ASD (positive predictive values 80–100%).¹ Novel prediction methods are also being developed using modalities such as EEG and eye-tracking.² Once replicated, these predictive methods could shift the timing of ASD detection and intervention into the first year of life, thus accelerating neuroscientific investigations into the early pathogenesis of ASD and eventually enabling randomized, controlled trials of presymptomatic interventions.^1,3,4 In anticipation of this shift, this commentary considers whether predictive testing for ASD should be done and for whom, and looks to the well-established genetics literature that over decades has drawn (and re-drawn) the boundaries for predictive testing in children.

Limits and recommendations for predictive genetic testing in childhood

The prediction of disorder or disease before the onset of symptoms has long been a goal of preventive medicine. In practice, however, it can be challenging to convey a predictive diagnosis in the absence of observable symptoms, and to manage the individual-level uncertainty about when (and to what degree) symptoms will eventually manifest. If a predictive test can allow an efficacious treatment to be started immediately, there is clear objective benefit to testing. However, when there is no immediate treatment available (as in the classic example of Huntington’s disease) the benefits of testing become more subjective. Consenting adults can decide for themselves whether to seek a predictive genetic test, even for disorders that cannot be treated; however, when predictive testing is considered for children, additional measures are put in place that limit the choices available to parents. Testing for untreatable adult-onset disorders (e.g., Huntington’s) is discouraged until a child reaches the age of majority to protect her/his decisional autonomy about whether or not to seek testing.⁵ However, the question of whether to allow parents the option to test for childhood-onset disorders with no immediate treatment is less clear.

Prior to 2005, international consensus statements/guidelines for genetic testing were ambiguous or inconclusive about whether predictive testing should be performed for childhood-onset disorders with no proven intervention.⁵ More recently, guidance has shifted towards allowing parents to decide whether to pursue predictive testing for their children (see Table 1). Parental discretion in testing decisions is advised most strongly in the context of familial disorders, where the uncertainty of whether a child will develop symptoms can be a source of worry and anticipatory distress. While these statements from professional organizations do not represent all perspectives on genetic testing in childhood, the apparent trend towards allowing parents more discretion in testing decisions provides useful precedent for the emerging context of predictive testing for ASD.

Table 1.

Statements Supporting Parental Discretion in Decisions about Predictive Genetic Testing for Children

Year	Organization	Statement
2009	European Society of Human Genetics	“In the case of presymptomatic and predictive genetic testing for conditions which become manifest in childhood and which can not [sic] be effectively treated or prevented…Genetic testing could be considered if this would be to the psychological or social benefit of the child and his family.”13
2013	American Academy of Pediatrics/American College of Medical Genetics and Genomics	“Parents or guardians may authorize predictive genetic testing for asymptomatic children at risk of childhood-onset conditions.” (Policy Statement)14 “Significant deference should be extended to parents regarding the timing of predictive genetic testing for childhood-onset conditions.” (Technical Report)15
2013	German Society of Human Genetics	“For diseases that present in childhood or adolescence without available therapeutic or prophylactic measures at the time of genetic testing…Genetic testing may be justified in those situations where the results of testing can be expected to avoid psychological or social harm to the child or teenager.”16
2014	Human Genetics Society of Australasia	“When the condition is always expected to have its onset at an age when the asymptomatic minor is too young to be involved in the decision to be tested…the decision about testing should be that of the parents after appropriate genetic counselling.”17
2015	American Society of Human Genetics	“Over the last two decades there has been a general shift toward greater parental discretion in the face of clinical uncertainty about the best interests of the child. This broad shift is not exclusive to genetics but has implications for genetic testing.”18
2017	American Medical Association	“In general, respect the decision of the patient’s parents/guardian about testing when the child is at risk for a condition with pediatric onset for which no effective measures to prevent, treat, or ameliorate the condition are available.”19

Note: Selected excerpts from recent international consensus and guideline statements promoting parent discretion in decisions about genetic testing for childhood-onset conditions with no proven intervention.

Benefits and risks of predictive testing for ASD

If decisions to pursue predictive testing for ASD are left to parents, what are the potential benefits and risks that a parent (and their provider) would want to consider? The most immediate benefit may be relief from anxiety caused by uncertainty. Most parents of children with ASD are aware that the disorder is heritable and are concerned about the recurrence risk to current and future children.⁶ Siblings of children with ASD have a 12-fold higher risk of developing the disorder, and parents may alter family planning and other life decisions based on these concerns. Therefore, predicting ASD may be similar to predicting other familial disorders (such as hereditary cancers), for which a clear benefit is reduced uncertainty about the diagnostic future.⁷ At the same time, any predictive test is unlikely to be 100% accurate, and it remains unclear whether tests for ASD will be able to predict dimensional outcomes such as symptom severity.² Lingering uncertainty would be a risk for parents even after receiving a predictive test result. Parents who learn their infant will not develop ASD may remain vigilant to developmental milestones despite this assurance. Parents who expect an ASD diagnosis may have enduring questions about which symptoms will emerge, when, and at what level of severity.

Another potential benefit of predictive testing would be targeted access to intervention. Infants for whom a predictive test shows high likelihood of later ASD could be enrolled in intervention in one of two ways. In the first—what we refer to as on-time intervention⁸—infants could be frequently monitored for emerging ASD symptoms or delays and enrolled in individualized behavioral intervention as soon as their development begins to diverge. This approach could significantly accelerate treatment delivery compared to the standard approach of waiting to obtain services only after receiving a formal diagnosis. A second approach would be to enroll infants in a presymptomatic intervention which could be delivered immediately, before the emergence of symptoms. Current evidence-based interventions exist to support the on-time model; however, there have not yet been any large-scale studies of presymptomatic interventions delivered in infancy. For either approach, it has been argued that anticipated harms are minimal, given that the frontline modality—behavioral intervention (even if erroneously delivered)—is unlikely to be harmful.⁴ If pharmacologic intervention are eventually developed for presymptomatic infants, the risk-benefit calculus could be different. However, with behavioral intervention the financial and time demands of presymptomatic intervention could be a burden for families, particularly those with an older child with ASD. A recent study found that 44% of parents declined the opportunity to enroll their high-risk infant in early intervention, with logistical factors such as parent work schedules and distance to clinics emerging as primary barriers.⁹

In addition, there could be societal harms of introducing a new group of children with a predictive ASD diagnosis into a service system that is already stretched to the limits. For example, one study found families using Medicaid waivers experienced a delay of 3 years between ASD diagnosis and accessing early intervention treatment.¹⁰ To avoid exacerbating this problem with an influx of families seeking presymptomatic ASD intervention, the development of predictive testing must be accompanied by a feasible plan to equitably increase the capacity of federally-funded early intervention services (e.g. those provided via Part C of the Individuals with Disabilities Education Act).

Some harms associated with predictive testing are less relevant when applied in this context and age range. For example, a concern associated with presymptomatic evaluation for behavioral disorders that manifest prior to adulthood (e.g., psychosis) has been the potential for predictive diagnostic label to result in stigma/discrimination against an emerging adult who is not yet showing symptoms.¹¹ Given the very young age at which predictive testing would occur for ASD and the current services landscape, a predictive diagnosis would likely open more doors for children than it would close (i.e., allowing insurance coverage or entry into early intervention/special education programming). Within a family, however, it is possible that parents could adjust their expectations and interactional style with an infant based upon a predictive diagnosis, with potentially positive (but also potentially negative) impacts. It is worth noting, however, that concerns about adverse psychosocial impacts of predictive genetic testing have generally not been empirically supported,¹² suggesting that families may be better able to cope with predictive results than scholars initially anticipated.

The relative weight of the benefits and harms of ASD prediction may also depend on who is holding the scale. Parents, clinicians, researchers, autistic adults, and neurodiversity advocates all bring different perspectives and priorities to the issue of early identification of ASD.³ Efforts to develop presymptomatic interventions can be perceived as devaluing the identity and experience of autistic individuals, depending on the goals and targets of the treatment. Engaging a broad set of stakeholders will be critical for understanding the perspectives of those who will be most impacted by a temporal shift in the age of identification and treatment of ASD.

Setting the stage for future research

Adoption of a new predictive test (in any modality) requires cumulative evidence that the benefits of testing outweigh the harms. Carefully designed, prospective research studies are the best way to gather such evidence. Therefore, while clinical application of predictive testing for infants at high-risk for ASD is currently premature, research efforts to develop and evaluate such tests are ethically warranted. Carefully-developed informed consent procedures for such studies will be critical to ensure parents understand the potential for lingering uncertainty even after testing, and the unknown efficacy of presymptomatic intervention. Resulting evidence from this research will inform future deliberations about whether predictive testing for ASD in these infants offers sufficient benefit to justify widespread clinical use. Even if testing for high-risk infants is shown to have clinical benefit, generalizing these methods to infants at population risk for ASD (currently estimated to affect 1–2% of US children)² will require both validating the predictive methods in this group and re-examining whether the benefits of predictive testing outweigh the risks for families with no prior knowledge or experience of ASD.

The recent trend in predictive genetic testing towards a more discretionary set of policies acknowledges that the benefits of predictive information for families often goes beyond strict clinical utility (i.e., enabling access to available, proven treatments) to include a more inclusive and subjective notion of personal utility (i.e., improving well-being in areas that are valued by particular individuals.)⁷ Family knowledge about heritable health risks can cause anxiety for which predictive tests offer steps toward resolution. Like the mother in our opening vignette, many parents want to know even non-actionable health information about their children, and for such parents a predictable future may be preferred over an uncertain one. Other parents may choose to wait and seek evaluation only after observable symptoms emerge. The next 5–10 years will reveal whether predictive testing for ASD can offer personal and/or clinical utility for high-risk infants and their families. A forward-thinking research agenda should address both child-level outcomes and the broader implications of predictive testing for intervention service systems and societal attitudes about ASD.