What Causes Autism?

It's genetic. It's environmental. It's both.

Autism has an incredibly powerful genetic component. This determines who is vulnerable to getting autism.

Autism has an incredibly powerful environmental component. This determines how many people get autism and why it keeps increasing.

So what is that environmental component, and is there a role for vaccines? Let’s dig in.

This is educational in nature and not medical or dietetic advice. See terms for additional and more complete disclaimers.

Autism is Genetic

Twin studies show a powerful effect of genetics in autism. Identical twins share 96% of autism risk, while fraternal twins share 38% of autism risk.

Even identical twins share an in utero environment. They are usually raised by the same family. They, like everyone else, shares the broader environment to some extent with their communities at various levels ranging from small groups to the entire world.

So, the impact of environment is definitely greater than the 4% of difference found among identical twins. The heritability estimates from these studies range from 64-91%. The estimate depends on how the prevalence of autism is modeled: if it is modeled as a large spectrum with high prevalence, genetics takes a stronger role; if it is modeled with a stricter definition and a lower prevalence, environment takes a bigger role.

Fraternal twins share everything that identical twins share except an extra 50% of their genetics. As such, the 2.5-fold greater autism risk shared by identical twins makes it indisputable that there is a powerful genetic component to autism.

It is very important that if the environment does not vary, its influence is masked.

For example, if governments the world over mandate that an important environmental cause of autism is put into everyone’s water supply, the impact of this influence will be completely muted because it is shared by everyone.

If the relevant genes had to be triggered by a single environmental factor whose historical exposure level was zero, then a government mandate putting that factor into the water supply could make the incidence of autism go from zero to 5% practically overnight. The genes would go from irrelevant to 91% of the explanation overnight.

If different municipal governments had control over how much of that factor was put in the water, variability would be introduced. Suddenly the environmental factor’s influence would manifest in the statistics, and the heritability estimates would fall.

Thus, the heritability estimates can only be considered representative of the environmental conditions that prevailed when they were measured.

Which Polymorphisms Predispose to Autism?

Genome-wide association studies (GWAS) are a way to look at broad associations of common polymorphisms in large groups. Rare mutations do not show up in these studies due to lack of statistical power.

The largest and most powerful GWAS to date was published in Nature Genetics in 2019. It developed polygenic risk scores made from common variants that were only capable of explaining 2.5-3.8% of the variation in autism risk.

On the one hand, scoring in the top ten percent of polygenic risk indicated a 2.8-3.6-fold greater risk of autism. That means it has useful predictive power.

On the other hand, the very small proportion of risk explained compared to the heritability shown by twin studies suggests that common variations do not explain much of the total heritability. That is, as useful as the polygenic score may be, there is something else going on that is far more powerful.

We have to be careful here, though. As with the thought experiment on the government-mandated environmental trigger, if we postulate a “missing heritable factor” it could be the thing that turns on and off the relevance of the polygenic risk score.

In other words, suppose the “Mystery Heritable Factor X” is found in ten people. It may be that that within those ten people the polygenic risk score from common polymorphisms has nearly complete explanatory power for which of them get autism and which of them don’t.

Some of the most important genes figuring into this score were NEGR1, a regulator of axon growth; PTBP2, a regulator of mRNA splicing; CADPS, a calcium-activated protein involved in releasing neurotransmitters and neuropeptides; KCNN2, a calcium-activated potassium channel involved in controlling neuronal excitability; KMT2E, an enzyme that uses methylation to control genomic stability; and MACROD2, an enzyme that clears away ADP-ribose in the nucleus made from the irreversible hydrolysis of the niacin (vitamin B3)-derived NAD+ during the reaction to cellular damage.

The featuring of two calcium-activated proteins in this score is interesting since oxalate is three times higher in autistics and binds to calcium. KMT2E is likely to interact with nutritional, genetic, and toxic factors governing methylation, and MACROD2 is likely to interact with nutritional, genetic, and toxic factors governing oxidative damage.

What Is the Missing Hereditary Factor?

My contention is that the gigantic unexplained heritability is dominated by rare inborn errors of metabolism, mostly that compromise the production of cellular energy necessary to fuel the brain at critical stages of development.

The leading medical textbook on this topic, Saudubray, Inborn Metabolic Diseases: Diagnosis and Treatment, contains 76 uses of “autism” or “autistic” to describe autism as a consequence or comorbidity of these diseases.

Autism has been associated with the following inborn errors of metabolism:

• Phenylketonuria, a defect in the conversion of phenylalanine to tyrosine that leads to secondary depletion of most of the major neurotransmitters as well as the brain’s supply of methyl groups.

• CBS deficiency, which leads to toxic accumulation of homocysteine and deficient provision of downstream products such as cysteine, taurine, sulfate, and glutathione.

• SAH hydrolase deficiency, which blocks the methylation system prior to homocysteine formation and leads to a severe deficiency of methylation as well as toxic accumulation of the amino acid methionine and its aberrant alternative metabolites.

• Branched-chain alpha-ketoacid dehydrogenase kinase deficiency, which leads to the depletion of branched chain amino acids. Autism has also been associated with impaired transport of BCAAs into the brain.

• Urea cycle disorders, which lead to the accumulation of neurotoxic ammonia during conditions of protein catabolism.

• Hartnup disease, which leads to a deficiency of tryptophan, niacin, NAD+, and NADPH.

• Propionic aciduria, a defect in the biotin-dependent clearance of propionyl CoA, a metabolite of odd-chain fatty acids, BCAAs, methionine, and threonine, which leads to toxic accumulations of propionyl CoA and depletion of the vitamin B5-derived CoA pool, necessary for the metabolism of several other amino acids, and of fatty acids and glucose.

• L-2-hydroxyglutaric aciduria, which is a defect in the riboflavin-dependent enzyme that is necessary to rescue alpha-ketoglutarate — a critical intermediate in the citric acid cycle — from a wasteful side reaction.

• Smith-Lemli-Opitz-Syndrome, a defect in the synthesis of cholesterol, the rate-limiting factor for synapse formation.

• Succinic semialdehyde dehydrogenase deficiency, an impairment in the ability to clear GABA by having it enter the citric acid cycle. This compromises brain energy metabolism and also causes GABA to accumulate and spill over into the production of gamma-hydroxybutyrate (GHB, marketed commercially as Xyrem). Overdose of pharmaceutical GHB is often associated with agitation and “bizarre self-injurious behaviors.”

• Adenylosuccinate lyase deficiency, which leads to the accumulation of toxic intermediates in purine metabolism.

• Lesch-Nyhan Syndrome, a deficiency in the recycling of adenine nucleotides and their conversion to guanine nucleotides. The mechanisms of how this leads to the disease phenotype are considered unsettled.

• Cerebral creatine deficiency. This is a deficiency of creatine in the brain. Creatine is the ambassador of the mitochondria, spreading the impact of its ATP production throughout the cell.

• Creatine synthesis disorders, which impair the synthesis of creatine and lead to whole-body creatine deficiency.

• Cerebral folate deficiency, which compromises methylation and DNA synthesis in the brain by depriving it of folate (vitamin B9).

• Severe MTHFR deficiency, which compromises the use of folate in methylation. This is is not C677T or A1298C — this refers to rare defects that are much more severe.

• Lysosomal storage disorders, including those that compromise the clearance of sulfated carbohydrates; those that interfere with cholesterol trafficking; and those that impair the clearance of seed oil (PUFA)-derived oxidation products.

• Cerebrotendinous Xanthomatosis, a defect in the conversion of cholesterol to bile acids that leads to excessive cholesterol accumulation.

• Wilson disease, which interferes with copper handling in a way that prevents copper from fulfilling its beneficial roles and increases the likelihood that copper causes oxidative stress.

• Acute intermittent porphyria, which leads to a deficiency of heme and toxic accumulation of intermediates in the heme synthesis pathway. Possibly playing a role in vampire legends, it causes photosensitivity and can be treated with heme (which could be obtained from blood) to shut down the impaired pathway.

• Pantothenate kinase deficiency, which leads to a deficient ability to convert pantothenic acid (vitamin B5) into its cofactor form, mainly coenzyme A (CoA) which plays a critical role in nearly all aspects of energy metabolism, and 4’-phosphopantetheine, used for fatty acid synthesis. It also leads to secondary accumulation of toxic levels of iron in the brain.

• BPAN deficiency, a deficiency in autophagy.

• Biotinidase deficiency, which causes impaired recycling of the B vitamin biotin.

• Carnitine synthesis deficiency, which prevents the synthesis of carnitine, otherwise found only in meat, which is necessary for fatty acid oxidation and the detoxification of metabolic byproducts of macronutrient metabolism.

These disorders are incredibly diverse. They include many B vitamin-related or -responsive disorders and several impacting methylation, but also disorders of too much or too little cholesterol, accumulation of toxic metabolites, defects in autophagy, and defects in neurotransmitter production.

How common are these disorders as a cause of autism?

• In a series of 187 children with autism in Greece, five (2.7%) were diagnosed with inborn errors of metabolism.

• In a series of 179 children with autism in Turkey, six (3.3%) were diagnosed with inborn errors of metabolism.

• In Iran, a series of 105 children found 13 (12.4%) had inborn errors of metabolism.

Rates of inbreeding are highest in Iran and lowest in Greece, likely explaining the variation.

However, analysis of metabolic biomarkers suggests vastly higher rates of impaired energy metabolism in autistics:

• 17% have elevated lactate, indicating impaired pyruvate clearance or impaired respiratory chain activity.

• 41% have elevated pyruvate, indicating impaired pyruvate dehydrogenase activity.

• 28% have an elevated lactate-to-pyruvate ratio, indicating impaired respiratory chain activity.

• 19% have elevated acylcarnitines, indicating that CoA-requiring pathways such as fatty acid oxidation or the oxidation of BCAAs, lysine, or tryptophan, are causing sequestration of the CoA pool.

• Autistic children have three times more oxalate in their urine than controls. Oxalate is a powerful mitochondrial toxin and also a byproduct of impaired metabolism with a number of enzyme mutations that can raise its levels.

Heterozygosity for two metabolic disorders in closely enough related enzymes or pathways can often cause clinically relevant disease without it fitting into a diagnosable pattern. This is known as synergistic heterozygosity. While this gets almost no attention at all, it is mathematically impossible for it not to be far more common than diagnosable metabolic disease because the number of possible interactions among closely related enzymes and pathways is one or more orders of magnitude higher than the probability of two severe impairments occurring in the exact same gene.

Further, heterozygosity for a single metabolic disorder could become clinically important if some other stress hurts the function of the healthy copy of the gene sufficiently. For example, if someone is heterozygous for a rare, severe MTHFR mutation and also is homozygous for C677T and is also deficient in riboflavin and folate, at some point the cumulative stress on the allele that is not affected by the rare mutation will be so substantial that the person might as well be homozygous for the rare mutation.

If 2.7% of autistics in societies with low rates of inbreeding have diagnosable inborn errors of metabolism, then nearly all other autistics likely have clinically relevant heterozygosity for these disorders.

The resulting profound impairment in metabolic function then may or may not need to be co-present with some of the common polymorphisms in NEGR1, PTBP2, CADPS, KCNN2, KMT2E, and MACROD2 to result in autism.

It is likely that the underlying metabolic dysfunction has to be addressed early to prevent autism and may have much more limited utility if it is found afterwards. For example, two brothers had a deficiency of the biotin-recycling enzyme biotinidase. One was diagnosed at age four after having already been diagnosed with autism. Ten milligrams of biotin per day did not resolve his autistic behavior. His younger brother was identified as having the same disorder in newborn screening and was proactively supplemented with ten milligram of biotin per day. He never developed autism.

Why Is Autism Increasing?

The incidence of autism appears to have been increasing globally by about 0.06% per year for decades. In the US, autism prevalence estimates increased from one in 150 to one in 44 between 2000 and 2018, to one in 36 by 2020, to — just released days ago — one in 31 by 2022. This indicates a particularly intense trend in the United States that persists in recent years despite enormous consciousness around the possibility for diagnostic changes to be occurring over time. While we cannot say for sure how diagnostic inflation could be impacting these estimates, it is almost certain that there is a real increase happening.

If the most important genetics tend to be in energy metabolism, then we would expect the following environmental modifiers of their influence:

• Nutritional status, since vitamins and minerals are often cofactors of these enzymes.

• Toxic exposures, since metals and other toxins can inhibit these enzymes.

• Energetic demand from growth, pregnancy, exercise, and stress, since these can either lead to a deficiency of energy or the driving of macronutrients into impaired pathways that generate toxic byproducts (like ammonia or BCAA acyl CoA esters).

• Increases in body heat, since many defective enzymes are thermolabile, meaning even body heat is too hot, making them lose their shape or lose their binding ability for a nutritional cofactor, where any additional heat makes the defect much worse.

• Inflammation, which represents a special combination of the above: all inflammation signals energetic demand to deal with an internal crisis; this comes with increased nutrient demand to drive the higher rate of metabolism and the differentiation of immune cells; it is often accompanied by something toxic, such as a microbial toxin of an infection or vaccine, or the adjuvants of a vaccine; and it often increases body heat.

Nutritional status is highly variable across the population, but will still have trends to look for over time.

Toxic exposures will vary according to individual lifestyle choices, but will also vary over time at a population level due to regulatory and industrial trends.

Energetic demand is going to be highly variable within an individual, and will mainly be determining the timing of the onset of a disorder within that individual and to some degree cumulative energetic demand and the peak magnitude of energetic demand will influence whether that individual becomes affected.

Inflammatory stress is likely to have major waves in a population according to infectious epidemics and vaccine policies.

The cumulative burden of these factors and how they temporally cluster in a person’s life can determine both the penetrance of a mutation causing a metabolic impairment and its timing of onset.

The penetrance is the degree to which a mutation causes disease within a population. If the penetrance is higher in the population, then any given individual with the mutation is more likely to be clinically impacted.

The timing of onset is impacted by when the cumulative burden or the peak magnitude of stress reaches the threshold required to render the impaired metabolic pathway clinically important.

Autism could be increasing due to one or both of the following:

• The stresses listed above are increasing in cumulative burden, such that they are increasingly likely to cross the threshold required to render a metabolic impairment clinically significant, and thus increasing the penetrance of the metabolic disorders.

• The stress listed above are becoming increasingly distributed in a temporal architecture that pulls the metabolic stress into period of time where the brain is vulnerable to autism pathogenesis.

The current model of autism pathogenesis separates autism into two epochs:

• Epoch 1 starts early in pregnancy and continues till just before birth. This primarily involves a dysregulation in the formation of neurons, and the proliferation, migration, and fate of brain cells.

• Epoch 2 starts late in the second trimester but continues until about 14 months of age. This primarily involves a dysregulation of synapse formation.

Other authors divide this up into a few more stages and extend the process to age four:

Thus, if the stresses listed above are temporally clustered into the period from the first trimester of pregnancy through 14 months of age (or four years of age), they will be far more likely to result in autism.

It is definitively not the case that there is a single point of failure where autism is suddenly switched from off to on. Rather, there is a period of 23 (or 58) months where autism is cumulatively progressed toward, hence the “spectrum” of autism disorders that can result.

However, there can still be a single point where metabolic dysfunction precipitously deteriorates in response to a trigger and pushes the person over a final threshold of dysfunction that manifests to parents or caregivers as symptomatic onset.

Could Vaccines Be Increasing Autism Rates?

The framework I advocate herein suggests that both infectious illnesses and vaccines could contribute to autism.

The Saudubray textbook only mentions vaccines contributing to the onset of metabolic disorders once, in the case of glutaric aciduria, a sometimes riboflavin-responsive disorder of lysine and tryptophan metabolism:

At a median age of 9–10 months, the majority of untreated patients suffer an acute brain injury, usually associated with febrile infectious disease, but this acute encephalopathic crisis may also be precipitated by any other episode that induces catabolism, including undesirable reactions following routine immunisations.

A 2022 review identified case reports in the literature of 15 children for whom vaccinations precipitate the onset of inborn errors of metabolism.

One of them was a 3-month old girl whose glutaric aciduria was precipitated by a pentavalent vaccine for Hib, hepatitis B, tetanus, diphtheria, and pertussis, who died of cardiac arrest four days later.

In another, a polio vaccine precipitated the onset of glutaric aciduria in a pair of identical twins but they survived.

In others, a disorder of cholesterol trafficking was precipitated by a BCG vaccine; a disorder of a salt- and potassium-dependent BCAA and ketone metabolism enzyme was precipitated by a Japanese encephalitis vaccine; a disorder of a B12-dependent enzyme in the metabolism of amino acids and odd-chain fatty acids was precipitated by a flu vaccine; a case of familial hypertriglyceridemia was precipitated by a Pfizer COVID vaccine; a disorder of carnitine-dependent fatty acid oxidation was precipitated by the AstraZeneca COVID vaccine; and a disorder of hereditary fructose intolerance was precipitated by a rotavirus vaccine.

They cite a paper from 2011 as showing that vaccination does not increase the amount of hospitalizations in children with inborn errors of metabolism. But that paper also shows that the “sickest” of these children have a 4.5-fold increased risk of hospitalization after vaccinations, and it includes two children who died in the month or two following a vaccine-associated metabolic deterioration.

It is more accepted that illness with a fever can trigger metabolic deterioration. For example, the Saudubray textbook mentions fevers in this context 86 times.

Fevers often go hand in hand with increased energetic demand, but the body heat itself can trigger the onset of a disorder caused by a thermolabile enzyme.

Vaccines can cause fevers, so if fevers can precipitate metabolic disorders, vaccines can precipitate them too.

Vaccines happen to be unique from other sources of fever in several respects:

• All the common vaccines are injected. They purposefully skip over the potential for a non-inflammatory reaction to a low dose of pathogen in the mucosal membranes in favor of producing a high-dose systemic inflammatory response.

• Vaccines usually require extra toxins to help make this happen, such as heavy metals.

• Sometimes five or more vaccines will be given at once. Few babies in history have otherwise ever gotten sick with five more illnesses at the same time.

Further, a lot of vaccines are given at prime autism pathogenesis time. CDC recommends pregnant women get vaccines for whooping cough, flu, RSV, COVID, and sometimes a few more. CDC recommends babies receive 25 or more vaccine doses in the first 14-15 months of life. These include 5-8 doses given at two months and 5-6 doses given at four months. Another 7-15 doses will be given by age four.

Peter Hotez cites this study as showing no link between maternal flu vaccination and autism. However, the data show that flu vaccination in the first trimester increased autism risk by 26%:

They adjusted this for confounders till it dropped to 20%, then they multiplied the P value by a factor to account for making multiple comparisons of different trimesters, which made it lose statistical significance. But the straightforward interpretation of this study is that a flu vaccine in the first trimester mildly increases autism risk.

An oft-cited 2014 meta-analysis reviews the totality of the evidence that vaccines don’t cause autism. However, none of the papers look at cumulative vaccine burden and its relation to autism, nor timing of vaccine burden and its relation to autism. They instead look at thimerosal, mercury, or MMR, compared to their non-thimerosal, mercury-free, or monovalent measles, mumps, and rubella counterparts. This includes the massive Dutch study that was concerned with the MMR versus the monovalent versions of measles, mumps, and rubella.

One paper that kind of tried to tackle cumulative burden looked at total antigen exposure. However, because the whole-cell pertussis vaccine has far more antigens than any other, the study wound up just looking at whether someone had gotten the whole-cell pertussis vaccine instead of other pertussis vaccines.

Hotez and many others cite a macaque study that administered vaccines to monkeys without causing autism. However, the incidence of autism at the time of the study was one in 70. Each group had 12 monkeys, which gave it a 0.17% chance of observing one case of autism if the prevalence and mechanism were similar. Further, the macaques have a comprehensive genetic management plan to maintain their genetic health, but autistic children are enriched in inborn errors of metabolism. Thus, the study is too small and the population not vulnerable enough.

As noted in the Appendix, two case series and three case reports suggest that autism can onset very soon after vaccination in some children. This temporal association between vaccination and symptom onset has never been addressed in the many studies showing that “vaccines don’t cause autism” by which is meant “MMR and mercury vaccines don’t cause more autism than any other vaccines.”

Data from medical practices with liberal attitudes toward vaccination choice suggest parents who choose not to vaccinate have children with 4-fold to 5-fold lower rates of autism. In a self-described “control group” rejecting vitamin K shots and vaccines for both mother and child, zero cases of autism occurred in over 1000 children, whereas you would have expected 32 to occur according to the current average. That doesn’t mean unvaccinated children never develop autism. One reader commented on my Facebook page that she has a fully unvaccinated daughter diagnosed with level one autism. These are not random samples of the population, and there could be diagnostic bias exaggerating the trend.

However, the total lack of rigorous data addressing whether total vaccine burden and timing of vaccine burden is associated with autism makes it absolutely necessary to entertain these preliminary reports until this question is addressed the way it finally should be.

The Study We Need

Here is the research program that needs to be done to settle the vaccine question:

The United States HHS should organize a national randomized cluster trial of alternative vaccine schedules. Since there are other countries that vaccinate less than we do with no evidence of iller health, and since there are older US vaccine schedules at times that were not marked by increased childhood mortality, the study should compare more minimalist vaccine schedules to the current schedule. The unit of randomization could be hospitals, schools, or municipalities. Parents should have total choice about whether to withdraw from the study and follow the current schedule or no schedule. Those that deviate in either direction should have the opportunity to be studied as a free-choice side arm, and if they choose to be studied, details about their health choices should be tracked to asses for any cofounders in comparisons between groups.

Other Research Priorities

All children with autism should undergo comprehensively metabolic screening and whole genome sequencing that has as its intent not necessarily to diagnose them with a metabolic disorder but rather to identify the most actionable limiting bottlenecks in their metabolism.

Newborn screening for inborn errors should expand to look for signals of non-diagnosable partial metabolic impairments at birth and their ability to predict future autism.

Research should examine the role of all nutritional and toxic factors in the development of autism alongside all inflammatory factors (including infections and vaccines), and should place special emphasis on how these factors interact with rare metabolic disease genes as well as the common polymorphisms correlated with autism.

Prevention is best initiated at the beginning of a pregnancy. Examining the primary metabolic impairments in a first child with autism and intervening to optimize nutrition around them starting in the next pregnancy has the potential to radically reduce the degree to which first-degree relatives share autism risk. This should be studied prospectively.

Appendix: The MMR Controversy

In 1996, Fudenberg looked at 22 autistic children enrolled between 1984 and 1987 and examined the relation with maternal viral infection in the second trimester, multiple infections, especially ear infections, in the first 15 months of life, and the relation of onset to immunisations.

Fifteen out of them developed symptoms within one week of the MMR vaccine. Three of those fifteen had reacted to the vaccine with high fevers (up to 106F) and convulsions.

Twenty of the 22 had antibodies to myelin; half had rubella titers that were more than ten times normal; almost half had anti-thyroid antibodies; half six had increased toxic metals, especially aluminum, and decreased trace metals in their hair; and several had evidence of having had measles more than twice, indicating an impaired immune response.

In two families, two of three siblings had autism, and chronic ear infections were found in the two affected siblings but not in the unaffected sibling.

Fudenberg proposed a model wherein some individuals are genetically predisposed, who are then exposed to high antibody titers in the mother that cross the placenta and are present at four weeks, and then are exposed tor a given virus, live virus vaccine, or DPT vaccine that causes an aberrant immune response, often driven by interaction between the antibodies obtained from the mother and the newly injected toxin present in the offending vaccine or viral infection They suggested that one strategy would be to identify the genetic predisposition, and then delay some of the most problematic vaccines to three years of age, since the immune system would be more developed and competent at that time in many children,.

Fudenberg had his medical license taken away in 1995, the year before his paper was published, for unrelated drug-hoarding charges that he denied. His paper was never retracted.

Later in 1998 Wakefield and 11 co-authors famously reported that in 12 autistic children, eight had their onset within 24-48 hours, one week, or two weeks of the MMR vaccine, while one had a temporal association with measles infection and another had a temporal association with an ear infection.

Six years later, in 2004 ten of the original twelve authors retracted the raising of the question of the possibility of a causal link between MMR vaccines and autism in view of the “major implications for public health” that made it “the appropriate time” to formally retract the question.

Then, twelve years after the original paper was published, in 2010, the editors retracted the paper on the basis of the UK General Medical Council’s Fitness to Practice Panel on January 28, 2010, finding that the children studied in the paper were not “consecutively referred,” nor were the investigations “approved” by the local ethics committee.

This panel purported to refute the statement that the children were “consecutively referred to the department of paediatric gastroenterology with a history of a pervasive developmental disorder with loss of acquired skills and intestinal symptoms (diarrhoea, abdominal pain, bloating and food intolerance).” They said this implied a “routine referral process” in which the children were referred because of gastrointestinal symptoms and in which the investigators played no part, when in fact four children were referred to look at the link between MMR and autism and Wakefield played a role in the referral of two of those children and two others. They considered this “irresponsible,” “misleading,” and “contrary to your duty to ensure the information in the paper was accurate.”

The disputation of ethics approval focused on the fact that 1) the recruitment process for several children had evidence of being de facto initiated prior to the approval date of December 18, 1996, and 2) several children’s gastrointestinal problems were not sufficient to merit a diagnosis of gastrointestinal disease and therefore did not meet the inclusion criteria of “symptoms and signs of intestinal disease or dysfunction namely pain, bloating, alternating constipation and diarrhoea, steatorrhoea and failure to thrive.”

Investigative reporting by Brian Deer — who was largely responsible for provoking the convening fitness to practice panel — published in the British Medical Journal identified additional discrepancies. Three children didn’t have autism: two had Asperger’s or possible Asperger’s, one turned out not to have any diagnosis. Five had variable evidence of behavioral abnormality or metabolic dysfunction predating the vaccine. In one case, the child had a facial deformity, indicating metabolic dysfunction that may have started in utero. Deer talked to two parents who indicated the onset of autism was substantially different from what was in the paper: one wanted Wakefield’s license revoked; the other complained that Deer was using “gutter press” tactics to get her to change her story. Most of the parents had some relation to anti-vaccine ideas, with 11 of 12 having blamed the MMR vaccine themselves in the hospital.

The panel also argued Wakefield failed to disclose conflicts of interest, including a lawsuit and business ventures related to his research, and that his group subjected children to unnecessary medical procedures. These findings, however, were not included as reasons for the Lancet retraction.

There are certainly legitimate questions around Wakefield’s procedural appropriateness, conflicts of interest, and motivations, and it is unlikely all of the data in the paper are completely accurate. However, the retraction of the paper had nothing to do with the temporal association between the autism onset and the MMR vaccines.

Deer’s contention that the temporal association falls apart when there is evidence of preexisting metabolic or behavioral dysfunction in five of the children does not hold up if we use a multifactorial model that involves various opportunities to cross thresholds of metabolic dysfunction, where crossing a certain threshold at the right time of development results in autism.

Fudenberg’s paper had a larger sample size and never had a retraction at all. Deer gives Fudenberg a small paragraph in his BMJ series and never questions the details of his paper.

So, there are two case series suggesting this temporal association and no clear refutation of it.

A spattering of case reports continued to roll in suggesting vaccines sometimes precipitate autism onset:

• A 2008 case report noted onset of autism and precipitous metabolic decline “within 48 hours after immunizations to diphtheria, tetanus, and pertussis; Haemophilus influenzae B; measles, mumps, and rubella; polio; and varicella (Varivax).”

• A 2013 case report noted autism onset with language regression from the 4-year to 18-month level ”immediately after her 4-year-old well child visit.” While this paper doesn’t say anything about vaccines, the four-year well-child visit typically involves DTaP, polio, MMR, chicken pox, hep A, and now according to CDC COVID and varicella vaccines.

• A 2017 case report notes that a child had no developmental problems at the 14-month well-child visit but the parents reported regression over the course of months 15 to 19, which doesn’t have a clear “immediate” relation to the 14-month visit, but was in the wake of a visit that likely included vaccinations for RSV, hep A and B, DTaP, pneumococcus, polio, MMR and varicella.

Much has been made of the Wakefield retraction as if the entire idea of a temporal association between the MMR vaccine and autism onset was proven fraudulent. But the Wakefield attraction had nothing to do with the temporal data, even though Deer questioned it in his article, and no one ever refuted or retracted the Fudenberg data, which made the same suggestion two years earlier with a larger sample size.

The possibility for a vaccination to acutely provoke autism onset is consistent with my model where a certain magnitude of stress can push an individual across the threshold into metabolic deterioration. This was not the singular start of autism. Events beginning prenatally and continuing up to that point built up the predisposition. The vaccine was just the straw that broke the camel’s back. But its occurrence within the window of time where autism onset is relevant gave it permanent significance to that individual’s health trajectory.