Universal screening for autism spectrum disorder (ASD) is recommended during pediatric primary care visits in the first 2 years of life. However, many children are missed by initial screening and not diagnosed with ASD until years later. Research efforts are underway to develop and evaluate new objective measures of risk for ASD that can be used in infancy, before symptoms emerge. Initial studies with these tests, particularly MRI-based screening for infants at high familial risk, have shown promise but have not yet been evaluated in clinical trials. We present the study design for a hypothetical clinical trial that would combine presymptomatic detection and intervention for ASD and consider, through commentaries from diverse perspectives, the ethical issues that should be anticipated in advance of beginning such trials. Commentators Drs Pruett and Piven address the social value of the proposed research and importance of researcher-bioethicist collaborations. Drs Estes and Wolff discuss the clinical potential and challenges of developing presymptomatic interventions for infants at risk for ASD. Dr Harrington takes a neurodiversity view of presymptomatic prediction and intervention and their implications for autistic identity and quality of life. Finally, Drs MacDuffie, Peay and Wilfond consider the potential risks and benefits that must be evaluated and weighed in the next phases of research on presymptomatic detection and intervention for ASD.
In the 13 years since the American Academy of Pediatrics recommended that all children be screened for autism spectrum disorder (ASD) at 18- and 24-month well-child visits,1 pediatrician offices have become the primary site for ASD screening in the United States. Unfortunately, the most widely used parent-report screening test (the Modified Checklist for Autism in Toddlers) has only modest positive predictive values (15%–18%)2 and misses the majority of children who are eventually diagnosed with ASD. Research efforts are underway to develop new screening approaches that could lower the average age of diagnosis. Some have the goal of detecting ASD in the first year of life, before the emergence of observable symptoms. If validated, screening tests that enable accurate presymptomatic detection will be poised to alter dramatically the landscape of ASD screening in pediatric primary care.
The focus with this Ethics Rounds is phenotypic tests under development that could identify infants at risk for ASD before the onset of symptoms. This is in contrast to genetic testing approaches to identify genetic syndromes frequently associated with ASD like tuberous sclerosis complex or Fragile X. Current research efforts apply methods like eye-tracking, EEG, and MRI to detect developmental alterations in the first year of life that can predict later ASD diagnosis, even for children with no identified or suspected genetic syndrome.3 In one recent example, the Infant Brain Imaging Study Network demonstrated that brain characteristics detectable on MRI as early as 6 months of age can predict later diagnoses of ASD in infants with an older affected sibling.4,5 The Infant Brain Imaging Study network is currently working to replicate their initial results, with the eventual goal of using MRI to detect high-risk infants for entry into a randomized controlled trial (RCT) of a presymptomatic intervention. In this article, we present a hypothetical study design to anticipate the ethical issues that should be considered when designing such a trial.
Infants with an older sibling with ASD will be recruited at 6 months of age to participate in an MRI scan conducted during natural sleep (no sedation required). A machine-learning algorithm applied to the MRI data is expected to yield a predictive result (positive predictive value ≥80%) for ASD in ∼20% of the infants. These infants will then be recruited into an RCT, in which they will receive either a novel behavioral intervention or usual care. The behavioral intervention will use play-based techniques to encourage the development of social communication and improve parent–child interactions. During the consent process, parents will be informed that there is limited evidence to date supporting the efficacy of the intervention for this age group. Primary outcomes at 24 months will include whether infants meet diagnostic criteria for ASD, the severity of autism symptoms, and measures of language and social communication abilities.
This study will be the first to combine predictive testing and delivery of presymptomatic intervention to infants before the consolidation of symptoms that define ASD. Given the complexity and novelty of this approach, what ethical issues are raised by such a trial?
Drs Pruett and Piven, Psychiatrists/Neuroscientists, Comment
This case prompts consideration of the potential social value and ethical challenges associated with prediction and intervention for later developing ASD. We are co–principal investigators of a new project (the inspiration for this case report) to replicate and extend findings predicting ASD from MRI scans in the first year of life.4,5 The presymptomatic detection of, and intervention for, those infants classified as ultrahigh risk for ASD is a worthy goal for ASD research, as long as we carefully consider the challenges inherent in treating risk (not disorder) and take into account the tremendous heterogeneity associated with ASD and the nature of ASD-associated behaviors that can be egosyntonic (ie, compatible with an individual’s values and beliefs about themselves).
Facilitating Efficient Presymptomatic Intervention Trials
Although data are limited, researchers suggest that early ASD intervention is better than later.6 Presymptomatic intervention might be even more powerful, coming at a time when the brain is most malleable and before the consolidation of brain changes and behaviors that further complicate treatment. Treating risk before disorder is a well-worn path in medicine (eg, treating hypertension reduces the chance of a later heart attack). However, this approach is novel for ASD. Without accurate presymptomatic prediction, all high-familial-risk children must be enrolled in presymptomatic intervention trials, although the majority (80%) will not develop ASD. Our current work may, therefore, enable highly efficient trials of potentially efficacious presymptomatic interventions by enrolling children at ultrahigh risk for developing ASD. Although our current prediction algorithms reference a diagnostic category associated with significant impairment (ie, ASD), future clinical trials might predict more dimensional aspects of behavior that are related to ASD. These trials might include children with a range of impairment in the behaviors that define ASD (social communication and interaction deficits, restricted and repetitive behavior) and associated comorbidities (eg, aggression or language disorder). If we succeed in predicting quantitatively defined levels of impairment, referrals would then be possible for clinical trials of interventions for specific patterns and profiles of behavioral challenges, rather than serving only those children who are predicted to meet a dichotomous threshold for ASD.
Treating Impairment, Not Eliminating Diversity
The tremendous phenotypic heterogeneity in ASD is well known. Individuals with ASD are often classified as either “high-” or “low-functioning” and have various combinations of other-defining behaviors and associated comorbid conditions, including language deficits, motor coordination problems, self-injury, aggression, intellectual disability, anxiety, attention-deficit/hyperactivity disorder, and epilepsy. Few would question treating depression or hallucinations, because these cause great pain and discomfort. With ASD, however, there is a question about changing enduring traits that are more akin to personality and may be seen as defining who an individual is. Our goal is to mitigate the development of significantly impairing ASD phenotypes and not to limit the wonderful diversity of human expression, subjective experience, and achievement inherent in those individuals who find these characteristics syntonic with their view of themselves. We do not want to change behavioral characteristics for individuals with symptoms that are manageable and who are able to succeed in life in a way that meets their personal goals. However, just as there is no question that we should treat learning disabilities and depression, those ASD-associated behaviors that result in significant morbidity and impairment should be considered legitimate targets for intervention.
Collecting Social Science Data to Guide Future Research and Implementation
In our society, we allow parents considerable discretion to make treatment decisions for their children. Accurate presymptomatic prediction would give parents more power with respect to these decisions, but there is an urgent need to generate a knowledge base about the implications and various ways to best use this information regarding risk most effectively. It is essential to develop a program of ethical, legal, and social implications research parallel to research on predictive biomarkers and presymptomatic clinical trials, to ensure ethical conduct of this science and its future clinical translation. Bioethicists must collaborate with researchers studying risk and contribute to our understanding of the implications of this research. Researchers must listen to the broad array of stakeholders with an interest in predicting ASD outcome, including autistic individuals, their parents, providers, health economists, third party payers, and clinicians. And finally, we must commit to revising our scientific and clinical objectives on the basis of this feedback and the data generated in rigorously conducted ethical, legal, and social implication studies of presymptomatic detection and intervention for ASD.
Drs Estes and Wolff, Child Clinical and Educational Psychologist, Comment
Current diagnostic practice for ASD depends on recognition of core behavioral signs in the domains of social communication and restricted and repetitive behaviors.7 However, research has revealed behavioral and neurobiological markers that are present before core symptoms.4,5,8,9 Thus, as described in this case, it is conceivable that it may soon be possible to identify children who have a high likelihood of developing ASD in infancy. Current ASD interventions are designed to address delays, deficits, and behaviors that have already consolidated into the symptoms of ASD. There are not yet guidelines for usual care of infants at high familial risk. Pioneering ASD intervention research was based on children age 5 to 7 years10 ; much has been done to extend and adapt such interventions to younger children, as early as toddlerhood.11,12 But, because it is not yet possible to reliably identify infants presymptomatically in a clinical setting, interventions have not yet been designed or tested for infants before symptom emergence. As researchers involved with understanding early development and behavioral intervention for ASD, we recognize some of the unique opportunities (and potential challenges) that presymptomatic intervention presents.
Building Effective Presymptomatic Interventions
Presymptomatic intervention would, by definition, not target autistic features directly. Existing interventions for ASD could not simply be adapted for infants. Instead, presymptomatic intervention would need to target behavioral precursors of defining features of autism, such as visual orienting to promote the development of joint attention and expressive language. The process of intervention would also differ, with procedures deftly responsive to change to ensure infants reach developmental milestones across multiple domains. New interventions might also include targets not necessarily distinct to autism but with known pervasive positive effects on child development, such as supporting parent–infant interaction, as proposed in this case example. However, it is unlikely that improving parent–infant interaction alone would significantly reduce the prevalence of ASD or attenuate its most severe cognitive and adaptive outcomes because the parent–infant relationship is not a causal factor. Indeed, implicating parent behavior as mechanistic in the development of ASD recalls long discredited psychogenic theory and underscores why a careful, evidence-based approach is essential. A strong parent–infant relationship may improve quality of life, an important consideration for all interventions, or may mediate longer-term communication or social outcomes, but should not be confused with a primary target of presymptomatic ASD intervention.
Potential Barriers to Implementation
Aside from whether an efficacious presymptomatic approach can be developed, the acceptability of presymptomatic identification and intervention and its potential impact on families is poorly understood. It is critical to incorporate diverse parent perspectives on feasibility and acceptability from the inception to ensure effectiveness, not just efficacy. Parents may find presymptomatic intervention of lower priority than other responsibilities; caring for an infant is demanding even for the “easiest” children. For some families, it may come down to a choice between addressing clear developmental needs in an older child with autism versus taking action on subtle alterations in a healthy baby sibling. Although some parents feel better prepared for a second child with ASD, others experience increased distress when faced with the possibility that a younger sibling may be developing ASD. Coupled with the emotional impact of confronting the likelihood of having a first or second child with autism, this calculus may result in the decision to take a “wait and see” approach.13 Even if parents embrace presymptomatic intervention, the potential for iatrogenic effects and the psychological and financial costs of false-positives would need to be considered. If intervention is implemented without a holistic approach, taking into consideration family adaptive functioning, lasting negative impacts are possible.
Clinically Promising, Structurally Challenging
Despite current diagnostic procedures that can identify ASD as soon as symptoms consolidate, usually by 24 months,14 and evidence of cost effectiveness and long-term efficacy of intervention,15 many children are not identified until years after their symptoms emerge and many receive no ASD-specific, evidence-based intervention.16,17 Even for families who do receive early diagnosis, access challenges and limited availability of trained providers can result in long periods of time spent on waitlists or otherwise attempting to secure services. These same structural challenges would need to be addressed for presymptomatic intervention to reach its potential; if children are not identified shortly after birth, the window for presymptomatic intervention will pass. Taking advantage of the presymptomatic window also requires that interventions be immediately available. These challenges are not insurmountable, but they require significant investment and buy-in from intervention scientists, policymakers, and other stakeholders from the outset.
The exciting potential of presymptomatic intervention is consistent with the maxim “an ounce of prevention is worth a pound of cure,” taking advantage of neuroplasticity in the first years of life and ensuring infants do not miss the learning opportunities needed to develop skills vital to achieving future developmental milestones. Although there are now a handful of preliminary studies of presymptomatic intervention (eg, Green et al,18 Rogers et al19 ), issues relevant to implementation, as well as others such as content, timing, and dosage, remain open to inquiry. Caution and systematic evaluation are imperative as we move forward, but presymptomatic intervention represents a potentially transformative approach to supporting children with autism.
Dr Harrington, Clinical Psychologist/Autistic Self-Advocate, Comments
This case proposes a new method for identifying ASD in infancy, paired with a novel behavioral intervention. The previous commentary addressed ethical ramifications if the proposed intervention does not have the intended outcome. As a licensed psychologist who is also autistic, I will address an issue that may not occur to nonautistic researchers: ethical concerns if the intervention does have the intended outcome and causes change in a child.
Intervening to Change Autistic Brain and Behavior
Interventions that aim to change the autistic brain to be more “neurotypical” risk erasing a valuable form of human diversity. Autism, once conceptualized as a deficit, has been reimagined by the autistic community and researchers as a neurologic variation associated with valued strengths ranging from visual processing20 to resistance of peer pressure.21 Many autistic people take deep joy in their distinctive interests, values, and sensory experiences. The autistic community compares the desire to “cure” autism with past efforts to “cure” homosexuality or gender nonconformity,22 which often had similarly caring but misguided goals of alleviating a patient’s suffering in an unsupportive world.
Interventions that aim to mold autistic behavior toward neurotypical norms can create an illusion of “normality” but can leave a child struggling with unrealistic expectations and insufficient support. Behavioral intervention has often taught autistic children to ignore their own needs and perceptions while reinforcing desired compliance and communication behaviors. For example, increasing eye contact is often a goal of treatment despite autistic accounts of physical pain from eye contact.23 “Treating” low eye contact can be akin to “treating” a sprained ankle by teaching the child to walk on it without wincing. Known as “masking” or “camouflaging,”24 the suppression of natural autistic behavior is associated with negative outcomes such as suicidality.25 Conversely, feeling accepted as an autistic person is associated with improved mental health.26
Intervening to Meet the Goals of Autistic Individuals
Identification during infancy is advisable only if paired with an intervention that will promote flourishing rather than impose normality. This is the primary goal of the neurodiversity movement, a political effort that overlaps with the adult autistic community and advocates for respecting and accommodating diverse neurologies rather than trying to “fix” them.27 The movement is sometimes criticized as overlooking autistic people with intense support needs such as intellectual disability and destructive behavior. It is true that many autistic people and their families currently experience significant challenges and suffer from lack of support. However, the movement sees this as a failure of society rather than the individual and imagines a world of supports and accommodations robust enough that autistic people can be safe, happy, and included regardless of their language levels or cognitive scores.28
An example of an existing intervention that would be consistent with neurodiversity goals is the collaborative and proactive solutions/collaborative problem-solving (CPS) model, an empirically supported intervention for children with oppositional behavior that can reduce severe outcomes such as self-injury and need for restraint.29 The CPS model conceptualizes aggressive and self-injurious behavior as an indication of a mismatch between the child’s needs and the environment. Rather than reshaping a child’s behavior to fit inflexible norms, CPS aims to reduce or eliminate destructive stress responses by tailoring expectations to the child’s current level of skill development. The child is then included in problem-solving at whatever level their cognitive and communication skills allow, by using a prescribed method that promotes trust and respects autonomy while building skills such as flexibility and perspective-taking.
CPS is an example of the type of intervention that could be fruitfully combined with identification in infancy. Through earlier identification, parents could prepare to have more flexible expectations for their child. Psychoeducation around neurodiversity and contact with autistic adults could help dispel myths and fears that parents hold about autism and prepare them to understand and appreciate their child’s unique perspective. Parents could proactively implement strategies like sensory accommodation, visual supports, and alternative communication tools (eg, sign language and picture cards) and, eventually, behavioral interventions like CPS if needed. Improved communication and reduction in stress could then help the child be more available for learning and able to cope without destructive behavior.
Toward an Inclusive Future
Similar to the way LGBTQ+ acceptance and racial integration seemed unimaginable only decades ago, disability inclusion as advocated by the neurodiversity movement will be challenging but is not impossible. Effective disability inclusion often indirectly improves the environment for nondisabled members of communities as well.30,31 Researchers studying early identification and intervention can help make this vision a reality by incorporating ways to uplift diverse minds rather than homogenizing then.
Drs Macduffie, Peay, and Wilfond, Research Ethicists, Comment
This case and the above commentaries convey the complex pragmatic and ethical challenges confronting researchers seeking to develop and test tools for presymptomatic detection and intervention for ASD. These commentators have so far highlighted the importance of 2 benchmarks of ethical clinical research32,33 : forming collaborative partnerships with relevant community members to determine appropriate intervention targets (ie, parents, autistic adults) and maximizing the social value of this research for children and families. Here, we focus on a third benchmark: achieving a favorable risk/benefit ratio for families (and not solely for infants) via careful attention to specific aspects of the study design.
Potential Benefits and Risks
The clinical benefits (if any) of the proposed presymptomatic intervention are unknown, and this state of equipoise can justify randomizing a subset of the sample to receive usual care, assuming robust informed consent. The potential for direct benefit from the intervention may not, however, be the only type of benefit valued by parents. Enrolled infants may benefit from periodic evaluation and close surveillance for non-ASD developmental delays (which are more common for high-risk siblings)34 and associated referral for standard early intervention services (ie, publicly-funded services like speech or occupational therapy). Parents who receive a presymptomatic prediction of later diagnosis may benefit from more time to prepare emotionally, financially, and logistically (eg, by moving closer to extended family or an academic medical center).35 Previous work with families of children with Fragile X syndrome suggests parents place great value on learning health information about their child even if it does not lead to proven treatment.36 As investigators with expertise in pediatric bioethics and genetic counseling, we hypothesize that the same will be true for many parents of children likely to develop ASD and that parents who value these types of ancillary research benefits will be more likely to enroll their children in the research.
That said, the disclosure of presymptomatic predictive results also has potential to cause harm. Risks include the potential for false-positive or false-negative results. Parents of children with a positive result who do not develop ASD (false-positive) may expend substantial time, financial resources, and effort on interventions that are unnecessary and supplant other developmentally valuable experiences. Worth noting, however, is that the risks associated with unnecessary behavioral intervention are minimal; current approaches for young children use play-based techniques to teach social communication skills and encourage responsive 1:1 caregiver interactions that could benefit any child. Infants with a negative test result who do develop ASD (false-negative) could be delayed in receiving a diagnosis and miss out entirely on the window for early intervention. Given that false-negatives may have a greater impact on child outcomes, it therefore may be appropriate to prioritize test sensitivity over specificity. Determining optimal testing parameters will require empirical data collected over time on the impact of predictive results on parents, parent–child interactions, and child outcomes.
The proposed trial provides an opportunity to closely monitor families who receive a positive predictive test result and who are enrolled in either in the experimental or usual care group. Research teams should also follow families who receive negative test results and are not recruited for the intervention phase. Research staff should be trained in education and counseling approaches informed by genetic counseling, and longitudinal follow-up visits for all families should be implemented to monitor parent wellbeing and family and child outcomes. Finally, in light of emerging data that many parents choose not to enroll their high-risk infants in presymptomatic intervention when offered,13 the research team should also follow families who receive positive results even if they are unable or decline (for personal, logistic, or other reasons) to continue their participation into the intervention phase. Longitudinal data collected on all families who receive predictive results from novel testing modalities will allow calculation of the risk/benefit ratio to guide future research.
In coming years, new screening approaches to predict ASD presymptomatically (whether by MRI, EEG, eye-tracking, or other behavioral assays) are anticipated to move into subsequent stages of clinical translation and be tested in RCTs in combination with novel interventions. Here we have considered one potential trial design and explored, along with commentators, steps that should be taken to promote the ethical conduct of such trials. Unfortunately, the broader context for this research is an inequitable landscape for ASD screening, diagnosis and intervention. In US primary care settings, children of color tend to be screened less often, with less accurate results, and are diagnosed with ASD later compared to white children.37,38 Particular attention must be paid to potential inequities before, during, and after presymptomatic detection and intervention trials. An unfortunate outcome would be the creation of new and expensive methods for predicting and mitigating symptoms of ASD that exacerbate existing health disparities. We believe this outcome is not inevitable and must be actively avoided. Investment of federal and philanthropic research funds in ASD services research could improve current service systems and prepare for the necessary expansion to implement presymptomatic testing and intervention. Although some might view services research related to not-yet-validated predictive biomarkers as premature, in reality, the simultaneous advancement of biomarker and services research is essential for new predictive approaches to achieve their clinical potential and to avoid further disadvantaging children and families who already face challenges accessing diagnostic and intervention services for ASD.
The authors thank Don Bailey and John Lantos for their helpful input on the conceptualization of this manuscript.
Drs MacDuffie and Wilfond conceptualized the overall manuscript, conceptualized and drafted their commentary, and reviewed and revised the overall manuscript; Drs Estes, Harringon, Peay, Piven, Pruett, and Wolff contributed to conceptualizing, drafting, and revising their respective commentaries; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: Supported by National Institute of Mental Health grant F32MH118689 (to Dr MacDuffie) and National Institute of Mental Health grant R01 MH118362 (to Drs Pruett and Piven). Funded by the National Institutes of Health (NIH)
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.