A number of studies suggest that a subset of children with autism make significant social and language gains in the first year of life, and then experience a dramatic loss of skills. As infants, these children babble and make eye contact. However, those abilities suddenly disappear. This loss of skills is known as regression.

Some research suggests that these children may be a unique subgroup within the autism spectrum, distinct from those who show more gradual declines. The question of whether there is an abrupt change in only some children with autism has become an important topic for parents, clinicians and researchers.

On 13 February, SFARI hosted a workshop to explore whether regression is unique to some children affected by autism. Basic science and clinical researchers from around the world met to share and discuss current research and ideas. The participants concluded that regression is not a distinct classification within the autism spectrum because the majority of children with the disorder show a decline in skills starting as early as 12 months of age, with some having more severe regression than others.

All daily skills, such as the ability to grasp items or to follow a caregiver’s gaze, have a developmental trajectory, and autism symptoms may not be apparent until the trajectory is considerably off-course. For example, a skill such as speaking does not typically emerge until the second year of life, but it relies on all of the information acquired during the first year. If language development is impaired during the first year, a child’s difficulties may not be apparent until the age at which he or she is expected to talk. What presents as a dramatic decline in behavior may actually be a reflection of accumulating abnormal neurobiological functioning, rather than a sudden change.

Cathy Lord, director of the Center for Autism and the Developing Brain at New York-Presbyterian Hospital in New York City, presented data from a study that provided free assessments of autism and follow-ups to children 18 to 30 months of age whose parents or clinicians were concerned that they might have the disorder.

Overall, these children show gradual rather than abrupt changes in autism symptoms. About 21 percent show worsening of symptoms with time, but almost the same percentage demonstrate improvements. This suggests that there are different and gradual trajectories for autism symptoms.

The amount of time that infants who go on to develop autism spend looking at faces also declines with time, based on assessments at 6, 12, 18 and 24 months of age, according to data presented by Sally Ozonoff, vice chair for research at the University of California, Davis MIND Institute. These children also have better social skills compared with typical children at 6 months of age. These social abilities are at typical levels at 12 months of age and show a steady decline after 18 months.

The results suggest that this early heightened social behavior may make children appear to have drastically regressed, when instead their decline has happened gradually from a higher starting point.

Ami Klin, director of the Marcus Autism Center at Emory University School of Medicine in Atlanta, also showed data based on the gaze preferences of infants that suggest that infants with autism gradually diverge from typical children during development.

The participants also agreed to consider regression in autism as distinct from childhood disintegrative disorder, which is an autism spectrum disorder characterized by a very severe and rapid loss of skills after 2 years of age.

To determine if regression has occurred in young children, researchers rely on either retrospective studies, which look back at parental accounts of their children’s behaviors, or prospective studies in which children come into a laboratory at different ages and their skills are evaluated at different time points. Home videos are another source of information that is used to augment parental accounts or when a child cannot come into the laboratory to be observed.

According to Ozonoff, 19 of 22 young children who went on to develop autism showed a decline in gaze directed toward faces between 6 and 36 months of age. However, only three parents reported any regression in their child’s behavior, suggesting that parents are not necessarily sensitive to regression in their children.

Audrey Thurm, staff scientist at the National Institute of Mental Health, pointed out that we still need parent reports to develop a complete clinical picture of their child, but that the field is going to need to develop better methods for acquiring accurate information from parents. Future research should focus on testing strategies that may enhance parents’ memories about their child’s behaviors in different contexts.

Another important question related to regression is whether there are neural biomarkers that precede the behavioral symptoms in autism, and if these biomarkers can provide insight into whether there is a regression subtype in autism.

Joseph Piven, director of the Carolina Institute for Developmental Disabilities at the University of North Carolina at Chapel Hill, presented imaging data on white matter, the nerve bundles that connect brain regions, in the infant siblings of children with autism, who have a higher-than-normal risk of developing the disorder. The brain scans capture changes in measures of the flow of water through the brain, an indicator of the maturity of white matter connections.

His team has found that how white matter develops from 6 to 12 months of age in children who are later diagnosed with autism differs from those who do not develop the disorder. They have also shown an atypical pattern of connectivity 18 months before the appearance of the core behavioral symptoms of autism, supporting the theory that autism is a result of gradual rather than abrupt neurobiological changes.

The workshop participants concluded that brain changes likely occur before behavioral symptoms of the disorder emerge, and that these alterations should be considered along a continuum. Some children with autism may have a more severe deficit on this continuum, which may translate into what many call regression.

Biomarkers could help to predict the emergence of autism before behavioral symptoms appear and to clarify the role of regression in the disorder.

Pat Levitt, director of the Zilkha Neurogenetic Institute at the University of Southern California, discussed how researchers can use mouse models to understand the early brain changes in autism. Mouse models of regression would need to show early typical behavioral development, followed by a rapid decline.

There are mouse models with genetic mutations in which developmental changes such as puberty drive the emergence of a disorder or disease. But puberty comes much later than the age at which regression is described in autism, so other drivers are necessary.

Timing is a challenge for another reason as well: 6 to 36 months of age in humans is equivalent to the pre-weaning stage in mouse pups. Any assays in which a pup is separated from its mother introduce a confounding variable because separation is known to cause stress and to lead to lasting effects on behavior.

These challenges would need to be overcome before the type of regression seen in autism can be reproduced in mouse models.

Jeffrey Neul, associate professor of molecular and human genetics at Baylor College of Medicine in Houston, discussed Rett syndrome, a developmental disorder that overlaps with autism. Rett syndrome is rare, almost always caused by a single genetic mutation, and is present primarily in females. It is characterized by typical development, stagnation and then severe regression in all skills, including language, motor and social abilities.

While the regression symptoms in Rett syndrome are different from those described in autism, there is a loss of skills in both cases. However, in Rett syndrome, these lost skills, including language, can often be regained.

Understanding regression in Rett syndrome mouse models could help researchers better understand the changes to neural circuits and brain pathways that underlie regression in autism.

Sarah Spence, director of autism services at Children’s Hospital Boston, discussed other medical issues that can cause regression, including epilepsy syndromes that overlap with autism, such as infantile spasms, tuberous sclerosis and Landau-Kleffner syndrome. Children with these syndromes have high rates of autism, but they can be treated and the skills lost can be regained.

Metabolic disorders can also be associated with autism and present as a regression of abilities. Cerebral folate deficiency and mitochondrial disorders both have symptoms similar to autism, but are treatable.

Similarities between these disorders and autism could point the way to biomarkers and suggest common mechanisms that lead to regression, Spence says.

In conclusion, studies have shown that most children with autism show a gradual decline in skills rather than an abrupt loss of abilities.

More detailed biological research is needed in the first 6 months of life and even prenatally to uncover biomarkers that may identify children at risk for autism before their behavioral symptoms emerge. Early identification could also provide crucial opportunities for early intervention.

Research also suggests that home videos can provide important insight into the development of a child before they have received a diagnosis of autism. Using these videos or following at-risk children in the laboratory is more reliable than parent interviews, research shows, because parents do not always accurately recall changes in their children’s abilities over time.