Controversy seems to follow autism like the tail on a kite. From the earliest descriptions of autism, clinicians questioned whether this disorder was an independent entity. For instance, in 1959, Bender1 highlighted the imprecision of the term because many distinctly different etiological pathways could converge in the same disorder and concluded that autism was neither an etiological nor clinical entity. In contrast, in 1952 Van Krevelen’s skepticism about the existence of autism was reversed when he encountered his first patient who was “as much like those described by Kanner as one raindrop is like another.”2 Others argued whether autism was psychogenic or hardwired in biology and suggested that factors as diverse as mothering3 or the reticular formation4 could be the common etiologic denominator. More recently, controversy continues and is reflected in changes in diagnostic criteria and the increasing recognition of heterogeneity that incorporates the word “spectrum” into the diagnosis. With each new prevalence estimate that suggests inexplicable and substantial increases in the number of children diagnosed with the disorder,5 the urgency to better understand causation is amplified.
Although the evidence is already abundant that no relationship exists in the general population between measles, mumps, and rubella (MMR) vaccine receipt and autism spectrum disorder (ASD) risk,6 immunization rates remain low in certain populations and countries because of this inappropriate belief. Descriptions of autism contain histories of children who seemingly were suddenly affected by a catastrophic developmental event between 1 and 2 years of age—a time in proximity to a scheduled MMR immunization. It made sense knowing this temporal window to ask, “Could it be that, if all of the requisite genetic and other risks are present, MMR can lead to the development of autism?” If so, the population in which there might be such a signal would be families already affected by autism.
In this issue of JAMA, Jain and colleagues7 evaluated 2 questions in their large insurance claims database: does the incidence of ASD differ in younger siblings of affected children who are immunized with MMR vs those who are not? And, for the population as a whole, does the incidence of ASD vary as a function of MMR immunization status? The answer to both questions is no.
In the report by Jain et al,7 of 95 727 children with older siblings who were included in the study, 1929 had an older sibling with ASD and 994 children had ASD diagnosed. The relative risk of ASD at age 2 years was 0.76 (95% CI, 0.49-1.18) for children with older siblings with ASD and 0.91 (95% CI, 0.67-1.20) for children with older siblings without ASD.
Some parents of children with ASD may have chosen to delay immunization in subsequent children until they were certain any risk had passed. Such behavior, which arguably could enrich the immunization rate in the nonautism subgroup relative to the group that may have been showing early atypical development, might create the impression that MMR vaccine is actually reducing risk for ASD. Indeed, Jain et al7 report relative risks of less than 1.0. Even so, short of arguing that MMR vaccine actually reduces the risk of ASD in those who were immunized by age 2 years, the only conclusion that can be drawn from the study is that there is no signal to suggest a relationship between MMR and the development of autism in children with or without a sibling who has autism.
Taken together, some dozen studies have now shown that the age of onset of ASD does not differ between vaccinated and unvaccinated children,8,9 the severity or course of ASD does not differ between vaccinated and unvaccinated children,10 and now the risk of ASD recurrence in families does not differ between vaccinated and unvaccinated children.7
The study by Xiang and colleagues11 in last week’s issue of JAMA directs attention for autism risk elsewhere—to the prenatal environment. Accumulating data from various different sources, including genetic, neuropathological, electrophysiological, and even infant eye gaze preference studies, have suggested that the developmental pathways for autism are created much earlier than clinical symptoms are manifest—informing both the timing and the types of environmental exposures on which research should focus. This study leveraged the large Kaiser-Permanente database and asked 2 questions, building on findings that maternal diabetes increases autism risk.12 First, is the risk for ASD increased among offspring of mothers with type 2 diabetes during pregnancy, and second, for those mothers who develop gestational diabetes, does the time of onset during the pregnancy influence that risk or provide clues about critical periods of vulnerability?
Of the 322 323 children studied, 3388 were diagnosed with ASD, including 2963 unexposed, 115 exposed to preexisting maternal type 2 diabetes, and 310 exposed to gestational diabetes. The unadjusted incidences were 1.77, 3.26, and 2.14 per 1000, respectively. More than 99% of infants who were exposed to maternal diabetes in utero did not develop ASD. However, in adjusted analyses, the authors found an increased risk in the subgroup of children exposed to gestational diabetes at 26 weeks or earlier. The hazard ratio for preexisting type 2 diabetes was 1.21 (95% CI, 0.97-1.52) and for gestational diabetes at 26 weeks or earlier, 1.42 (95% CI, 1.15-1.74). Thus, the timing for this environmental exposure is isolated to early pregnancy.
Inasmuch as a number of the genes implicated in autism map to the insulin-signaling pathway,13 the finding is intriguing. Although insulin does not cross the placenta, elevation of fetal blood glucose levels as a consequence of maternal diabetes could increase fetal insulin secretion and perhaps inappropriately stimulate the fetal PI3K/Tor pathway at a critical period in development.14
It will be interesting to determine whether such genetic abnormalities are overrepresented in this subgroup of children with autism associated with an abnormal metabolic environment during pregnancy compared with the majority of similarly exposed but unaffected children. Given the increasing prevalence of obesity among women of child-bearing age, understanding the effect of insulin and maternal diabetes on pregnancy outcomes becomes more important.
In this issue of JAMA, Bearss and colleagues15 present the results of a multisite, 24-week randomized trial comparing 89 parents of children with ASD with behavioral problems who received parent training to reduce disruptive behavior with 91 parents who received simple education about ASD. In light of the controversial parent blaming that is a legacy in the field, the term “parent training” is arguably a misnomer and ought to be retired. In effect, the training helps promote becoming a better behavioral interventionist, not a better parent.
In this study, the largest of its kind to date, the parent-rated primary outcomes improved more in the training group than the education group, although neither met the prespecified statistical threshold for a clinically important difference. The secondary outcome based on clinician ratings found that 68.5% of children of parents who received behavior training and 39.6% of those who received autism education were rated much or very much improved at the completion of the trial, an outcome that rivals any medication studied to date. It is difficult to assess the clinical significance of parent-reported changes in problem behaviors ascertained through rating scales because parents could not be blinded to their group assignments. However, the use of clinician-rated global improvement measures and the maintenance of improvement at follow-up at 48 weeks are noteworthy.
Although specific behavioral training was superior, both groups reported considerable improvement over baseline, suggesting that even regular intensive education about autism provided value to parents and translated to perceived behavioral improvements in their children. Some challenges for the future include what can be learned about the children who did not respond to behavioral intervention and why some children of parents who were not educated about behavioral principles also improved. Autism is a diverse condition, and a better understanding of how psychosocial interventions work will be critical for determining how to personalize treatment.
As long as there are swirling winds, the kite will have a tail, and to be fair, controversy can be a stimulus for progress. These studies move the field forward toward a more focused and productive search for temporal and environmental factors that contribute to autism risk. They also provide information to support families affected by autism—either by allaying concern that MMR vaccine might be harmful or giving parents the knowledge and tools to better understand their child’s condition and manage his or her behavior. The field is long overdue for calm weather, and the forecast is increasingly promising.
Corresponding Author: Bryan H. King, MD, MBA, Seattle Children’s Autism Center, University of Washington and Seattle Children’s Hospital, PO Box 5371/M1-1, Seattle, WA 98145-5005 (firstname.lastname@example.org).
Conflict of Interest Disclosures: The author has completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
King BH. Promising Forecast for Autism Spectrum Disorders. JAMA. 2015;313(15):1518-1519. doi:10.1001/jama.2015.2628