The Causes & Treatments for Autism | Dr. Karen Parker

Huberman frames the discussion around the prominent theories and the current understanding of the biological basis of autism, as well as what remains mysterious and unresolved about its causes. He distinguishes a rise in incidence from a rise in diagnosis, stating that the conversation will address why incidence, not just diagnosis, has increased over the last 10 to 15 years, and he indicates that Parker will present brand-new findings from her laboratory pointing to a new understanding of causation and to a novel treatment. His opening question is whether the frequency of autism is genuinely increasing or whether medicine is getting better at detecting what was always there, and he asks whether clear answers exist.

Parker describes the answer as multifactorial. She attributes part of the change to improved detection: children were previously diagnosed at nine or 10 years of age, whereas clinicians can now reliably diagnose at two to three years of age; pediatricians use autism screeners during the first couple years of life; and awareness has increased. She cites a current figure of 1 in 36 US children with an autism diagnosis, notes that it was 1 in 44 a little over two years earlier, and Huberman recalls an earlier figure of 1 in 80. The figures are presented as reported prevalence rates rather than as a settled causal account.

if you've met one kid with autism, you've met one kid with autism

Asked whether the rate is averaged across boys and girls, Parker says autism is male-biased in prevalence and cautions that the studies vary. She describes autism as a highly clinically heterogeneous disorder, a caveat she returns to throughout the conversation. She reports that different studies show roughly three to four boys affected for every one girl, and explains that US prevalence figures come from the CDC's 11 monitoring sites, which follow children and release new prevalence rates every few years.

Huberman asks which diagnostic tools are used given that two- and three-year-olds have limited vocabulary, raising language, visual gaze and auditory processing as possible domains. Parker responds that autism is a behavioral diagnosis rather than one based on a blood test or comparable tool, made by an expert, usually a psychiatrist or psychologist. She states that, based on the DSM-5, clinicians look for two core features: pervasive social interaction challenges and the presence of restricted, repetitive behavior.

Co-occurring features Parker lists alongside the core criteria: anxiety; sensory challenges; seizure disorders; and sleep disorders. She notes that each person with autism has a distinct collection of traits, and that this collection is how the diagnosis is made.

On early intervention, Parker says a diagnosis of autism is usually required before behavioral interventions are used. She describes baby sibling studies, in which the high heritability of autism is used to enrich a prospectively followed infant population: because a family with one child with autism is at increased risk for subsequent children with autism, some at-risk infants are enrolled in behavioral studies even before a diagnosis.

Asked whether parents in engineering, mathematics or physics are more likely to have autistic children, Parker says autistic traits are continuously distributed across the general population. She names two measurement instruments, the Social Responsiveness Scale, a US-based instrument, and the autism quotient, a similar measure designed in England. She reports that work with the autism quotient indicates that individuals in intense STEM fields such as engineering, physics and math carry a greater burden of autistic traits even without an autism diagnosis.

Huberman uses the severity spectrum, illustrated by an account of a residential school and by self-identified autistic adults who drive, work and maintain relationships, to ask whether autism is better described as multiple spectra. He proposes dimensions of severity, of stereotyped behaviors, and of focused interests, suggesting there may be 50 or 60 such spectra, and asks whether the spectrum is multidimensional rather than a single line from severe to mild.

Parker replies that understanding the biological basis of behavior would clarify such dimensions, but that without it the field is left with the question of whether to lump or split. She states that autism is highly heritable, with estimates of about 40% to 80% genetic that vary widely, and that the common thinking is that about 50% of autism is associated with common genetic variants. She describes autism as largely an inherited polygenic condition in which many common variants are additive: a lower dosing of such variants might correspond to higher functioning, whereas a single-gene, highly penetrant disorder might correspond to severe intellectual disability. She adds that there is a great deal that is not known, and that without an underlying biological basis it is difficult to diagnose and treat, a situation she says is common across psychiatric and neurodevelopmental disorders.

are we lumpers or splitters?

Huberman asks whether what is called autism may in fact be 50 different disorders or conditions and what the diagnosis centers on. Parker calls this the $64,000 question and draws an analogy with cancer biology, where molecular profiling of a removed tumor can now guide personalized medications, in contrast to a behaviorally defined diagnosis that shows marked heterogeneity. She reports that some clinicians argue that autism accompanying fragile X, Prader-Willi syndrome or Timothy syndrome is not really autism, yet because the diagnosis is behavioral and criteria are met, the reasoning becomes circular. She concludes that sub-defining the condition will be difficult until autism itself is understood.

it becomes this weird circular argument

Huberman asks whether any specific neural network can be identified as different in autism, citing a study indicating that many of the genes altered in autism, or the proteins consequent to them, are at the synapse. He recalls a question raised at a Stanford talk: whether autism is known to be a brain issue at all, or whether it could instead involve the immune or cardiovascular system, and asks how one would establish that it is a challenge of the brain.

Parker notes that some speak of autisms in the plural and says the field is at an early stage of locating the major player, allowing that for some people it may be more brain-based and for others involve the gut-brain connection. She lists barriers to progress: limited access to cerebrospinal fluid and brain tissue biopsies; difficulty placing affected children, who may be unable to sit still or may have sensory issues, into a brain scanner; and the constraints of model systems. She argues that a suitable model organism should have complex cognitive abilities, complex social skills, vision as its primary sensory modality and possibly consolidated sleep, and says control mice fundamentally lack many characteristics needed to model autism with fidelity. She attributes about 50% of preclinical failures, in which an intervention that works in an animal fails in a human clinical drug trial, to poorly selected animal models, and presents more sophisticated models as a point of entry.

we're at the infancy of thinking about this

Huberman defines preclinical models as non-human models, naming mouse, non-human primate, flies and worms, and notes that the conversation will turn to non-human primate models as well as work in humans. He frames genes and environment as always interacting rather than as nature or nurture, and asks about the role of the environment in both the frequency and the presentation of autism.

Parker refers to multiple epidemiological studies and lists factors associated with increased risk: advanced parental age, prematurity and severe prematurity, and maternal illness during pregnancy. She presents these as associations rather than as established causes.

Huberman revisits a hypothesis from Pasco Rakic, who studied neuronal migration in early development, in which neurons are born at one location and migrate to a final resting place before forming connections. Rakic proposed that the increased incidence of autism might be correlated with increased early prenatal ultrasound, with mouse-model papers in high-profile journals including Proceedings of the National Academy and Science reporting more migration errors with each successive ultrasound. Huberman notes that the idea seemed not to progress into clinical guidance, that ultrasounds remain critical in pregnancy, and that other theories, including two engineer parents, food-environment toxicity, and vaccines or vaccine adjuvants, have surfaced in the literature and popular press, with some papers retracted. He asks whether one major environmental risk factor has emerged from this body of theories.

perhaps, and I really want to emphasize perhaps

Parker explains why a single risk factor is hard to isolate: every person entering a trial has a different genetic background, whereas a stratified trial would manipulate only one or two variables at a time, and experimental studies in developing children are very difficult, so the field relies on large epidemiological studies. She discusses neurogenetic syndromes with high penetrance for autism, such as fragile X, noting that because clinical presentation is diverse a medication that helps a handful of children may leave a trial without the statistical power to detect it. She describes the current state as seeing only where a flashlight is pointed in a dark room, and argues for genetically defined subgroups in which gene-by-environment interactions and medication effects can be tested.

think about being in a dark room, and you have a flashlight, and you only see where you shine the light

Huberman raises Timothy syndrome, associated with a mutation in an L-type calcium channel that is important in neurons and in other tissues including heart tissue, so that affected children have cardiac issues as well as autism. He observes that co-occurring gut, heart or musculoskeletal problems are often assumed to be consequences of autism, whereas the autism may instead be secondary to, or independent of, another condition. He returns to the question of whether autism is a spectrum or many distinct conditions.

Huberman introduces oxytocin and vasopressin as two neuropeptides he expects are subject to misunderstanding. Parker describes them as small peptides nine amino acids long that differ by only two amino acids, ancient molecules hundreds of millions of years old and highly evolutionarily conserved, with precursor forms such as vasotocin or mesotocin in other species. She says they are involved in social behavior across phylogenetic taxa, including egg-laying and reproduction, and that oxytocin and vasopressin are pervasive in mammalian species, though the brain areas and tissues in which they are made are similar but not completely overlapping.

Parker notes that because the two are structurally similar they can have similar effects, and that they bind to four receptors, which makes it difficult to disentangle which peptide is acting, at which receptor and where in the brain; she uses a key-and-lock analogy for hormone and receptor. She recounts that both were discovered for peripheral effects: oxytocin for uterine contractions and milk letdown during lactation, historically regarded as a female hormone, and vasopressin for urinary output regulation and blood pressure. She describes naming conventions, noting oxytocin was named once, from Greek for quick birth, while vasopressin was named twice, as arginine vasopressin or antidiuretic hormone, and that genes named twice once led researchers in different fields to work without communicating. She reports that neuroanatomists observed projections back into the brain from the hypothalamus, and cites a 1971 paper by Duke zoologist Peter Klopfer, Mother Love: What Turns It On, on precocious sheep and goat offspring, which proposed that oxytocin co-released into the brain during milk letdown turns on maternal bonding.

Oxytocin is orchestrating all these events of motherhood.

In sheep and goats, Parker says, passage through the vaginal canal activates oxytocin receptors, and an oxytocin antagonist that blocks the receptor prevents bonding to the offspring; lactation has a comparable effect. She stresses that this is species specific: humans and many primates evolved in extended family groups with biparental care and adoption, so the same mechanism would not necessarily be expected. She offers a personal account of a C-section, postpartum complications, deep vein thrombosis and pulmonary emboli that prevented vaginal delivery and lactation, while emphasizing the bond with her child. She frames bonding through evolutionary selective pressures and describes social interaction as rewarding from infancy, suggesting that many disorders in the species are characterized by a lack of social connectedness, citing autism, drug abuse, and social isolation or loss of a loved one as a predictor of stress-related depressive and anxiety disorders.