Coincidental associations do not provide proof for the etiology of autism

In this editorial, we are challenging the opinions of the review “Theoretical aspects of autism: Causes—A review” by H. Ratajczak (J. Immunotoxicol. 8:68–79, 2011). There are a number of different points addressed by Ratajczak, which require critical examination for accuracy. Two of the most egregious include the unqualified accusations that vaccines and thimerosal (the organomercury compound added to vaccines because of its anti-bacterial activity) are implicated in the development of autism.

First, the scientific literature does not support childhood vaccines as a cause of autism or autistic spectrum disorders (ASD). In introducing autism prevalence as a means of establishing coincidental associations with vaccines, Ratajczak stated: “It is first important to assess the coincidence and prevalence data to get timelines that might help determine the major causes of autism.” This statement begins the convoluted reasoning that Ratajczak used throughout the manuscript in an attempt to build hypothetical “causes” for autism. Ratajczak neglected to note that autism in various forms predates vaccines and use of thimerosal as well as many of the other putative causes that she associated to autism. On p. 71, Ratajczak stated that autism was discovered in 1930, which coincides with the introduction of thimerosal in vaccines; however, this neglects that autism in various forms has existed for at least 100 years, when Swiss psychiatrist, Eugen Bleuler, in 1911 used the term to describe one constellation of symptoms of schizophrenia, childhood developmental disorders were only beginning to be studied in Western medicine. It was 30 years later that Drs. Kanner and Asperger identified behavioral types, which we now associate with ASD. Nonetheless, autism has likely been an affliction of children long before the invention of vaccines or thimerosal.

Controversy exists as to whether the prevalence and incidence of ASD is rising and on p. 69 Ratajczak cited a news article by Sullivan (2005) regarding a study conducted by a respected research group at the Center for Autism and Developmental Disabilities Epidemiology at the Johns Hopkins School of Public Health. These investigators surveyed State Department of Education data on a wide range of neurodevelopmental disabilities (Newschaffer et al., 2005). Under the autism classification, they found that the ASD prevalence increased with time among younger birth cohorts. Since they found no similar decrease in other cohorts of neurological disabilities, they concluded that the increased prevalence was not due to shifts in how individuals were classified in order to gain special education services; they made no conclusions as to the cause of the apparent increased prevalence. They went on in their discussion to cite the limitations of their data, in particular, the fact that the type of administration data they used is subject to more classification bias than otherwise seen in research studies. It is our opinion that Ratajczak failed to review the study and instead relied on news reports to make scientific assertions that are not intended by research of the Hopkins group. Similar inconclusive data regarding prevalence and cause have been reported from other large epidemiological studies of autism (Fombonne et al., 2006; CDC, 2009; Hultman et al., 2010; Windham et al., 2010; Kirby et al., 2011). Whether autism prevalence is indeed increasing across all segments of the population continues to be debated. These studies, including the Hopkins Group work, point to a very complex picture of autism prevalence based on geography, maternal age, ethnicity and gender. None of the studies point to vaccines or thimerosal as a factor impacting the prevalence of autism and Ratajczak’s use of these studies to manufacture coincidental evidence linking autism prevalence to vaccine or thimerosal is misguided.

In addressing autism incidence, the author again failed to cite relevant literature and relied on review articles (Ayoub and Yazbak, 2006) that only marginally relate to understanding changing incidence rates for autism. To paraphrase Gerber and Offit (2009), the manuscript by Ratajczak is typical of the anti-vaccine movement in providing a “tale of shifting hypotheses.” It is indeed unfortunate that the author did not review Ritter (2005), who provided a framework of criteria for evaluating epidemiological studies of incidence or prevalence for ASD. Had Ratajczak chosen to review the literature using Ritter’s suggestions, she would have found many of her citations and subsequent comments to be invalid. Specifically, the Ritter review highlights the significance of three large population-based articles concerning prevalence and incidence of ASD (Honda et al., 1996; Baird et al., 2000; Fombonne and Chakrabarti, 2001). These studies clearly point toward a myriad of confounding variables that may be impacting incidence studies, none of which are related to vaccines and thimerosal.

On p. 70, Ratajczak cited a meta-analysis study by McDonald and Paul (2010) of longitudinal cohort studies in various regions of the world. The conclusions of this article suggest that various geographical regions had different inflection dates for increases in the incidence of autism. These changes in the prevalence could be related to a number of factors, which were not specifically speculated. McDonald and Paul noted that the incidence spike for the United States was 1988–1989 and Ratajczak speculated that this rise in autism is linked to MMR II vaccine introduced in the United States in 1988. Ratajczak ignored the rising autism rates both prior to this date and failed to cite a growing body of evidence that is the rate of ASD in children who received older versions of MMR the same as MMRII, there have been no differences in rates of autism, regression and response to vaccination in children receiving, one, two, or no MMR vaccinations (Honda et al., 1996; Fombonne and Chakrabarti, 2001; Madsen et al., 2002). The scientific literature now strongly indicates that the MMR vaccine plays no role in development of ASD.

Ratajcak’s attempt to link MMR to autism persists with a statement on p. 71: “Other reports have also used prevalence data that support an association of MMR vaccine with an increased prevalence of autism.” What other reports? Later on p. 71, it is stated that likely something “new” in MMR II vaccine or changes in schedules have resulted in the increased prevalence of autism. What studies demonstrate this link? Ratajczak provided neither scientific evidence nor a rationale to support considering such a hypothesis. Ratajczak’s comment “... changes in the type and increasing number of vaccines given at one time implicate vaccines as a cause of autism.” Once again simply is not supported by the scientific evidence; it appears to be a concerted effort on her part to convince the reader of her belief that vaccines are a cause of autism. This type of statement inappropriately assigns causation to simple temporal coincidences. Using this type of flawed logic would draw into question hundreds if not thousands of temporarily relevant exposures during the first months of life to the possible influences on the risk for autism.

MMR vaccines and thimerosal, which we discuss below, have undergone extensive scientific scrutiny with respect to autism; multiple ecological, case–control and retrospective cohort studies have failed to demonstrate a link between the vaccine and ASD (Fombonne and Chakrabarti, 2001; Madsen et al., 2002; Hviid et al., 2003; Stehr-Green et al., 2003; Verstraeten et al., 2003; Andrews et al., 2004; Heron et al., 2004; Hornig et al., 2008). A recent study of MMR vaccine in children with autism and enteropathy concluded that persistent measles virus RNA in the gastrointestinal (GI) tract or following MMR exposure was not associated with autism (Hviid et al., 2003). Ratajczak appeared to be unaware of recent studies that suggest the signs of autism and ASD can be observed in infants under the age of 12 months and thus prior to any MMR vaccination (Mars et al., 1998; Barbaro and Dissanayake, 2010; Ozonoff et al., 2010).

The issues that Ratajczak raised on p. 70 regarding vaccine antigens and autism are to us as immunologists beyond our comprehension. The vast array of antigens a newborn encounters during and immediately after birth indeed sets off an explosion of immunological activity. Recent studies assessing thymus activity using TREC assays in newborns (Stehr-Green et al., 2003; Verstraeten et al., 2003) emphasize the impact foreign antigens have on our developing immune systems. Present day knowledge about how our naïve immune system develops from birth to age 2 years are known to be quite complex; the evolving and maturing cellular components of the immune system represent heterogeneous populations with multiple functional and structural phenotypes. The amount of vaccine antigens is infinitesimal compared with those encountered in the environment. Upon entrance into the world, there is a natural antigenic expansion of the immune system, which is very apparent when conventionally raised animals are compared with those raised in gnotobiotic (germfree) conditions; immune cell numbers and activities are underdeveloped without the plethora of antigens awaiting us (Bauer et al., 1963; Olson and Wostmann, 1966). Although an infant’s immune system is rapidly developing, there is no evidence to support Ratajczak’s statement that “the immune system is particularly sensitive at 2 months-of-age.” In light of the multitude of stimulants in the natural environment and variances in our genetics that could affect susceptibilities, it is a wonder that disorders do not occur more frequently.

Numerous studies suggest that mutations across a wide range of signaling molecules and receptors that are critical for immune function can lead to aberrant immunity. Such genetic mutations and polymorphisms are not the result of vaccinations and instead point to an inherited set of traits that might predict aberrant responses to foreign antigens. There is no evidence to date to suggest that specific vaccine antigens play a role in the development of autism, nor is the evidence cited relevant to the question. The Farfel et al.’s (1999) and Megson’s (2000) articles on G_i proteins in no way implicate vaccine antigens as a cause of autism and the link to vitamin A in the later work is speculation that a genetic disorder (not a vaccine antigen) involving a mutation in regulation of the retinoid system could be a cause of autism. The speculation regarding pertussis toxin G proteins in a vaccine administered locally clearly does not make sense in light of infection with Bordetella pertussis in which infection and exposure to toxin may last weeks in the respiratory system.

Regarding the anti-bacterial preservative thimerosal that had been used in some vaccines, the scientific literature does not support thimerosal as a cause of autism. The scientific literature on mercury and thimerosal as a cause of neurodevelopmental disorders is extensive. In order to best understand how an environmental toxicant such as mercury can be a cause of neurological disease, one must consider three lines of evidence. First, epidemiological studies must provide a link between exposure to a toxicant and development of a specific neurological impairment; second, a relevant animal model is used to establish cause and effect providing a surrogate of human disease; and finally, there must be an established link between human exposure to the particular toxicant in question at equivalent levels and the development of disease.

A California epidemiological study has shown that even after removal of thimerosal from early childhood vaccines in 2001 the prevalence of ASD has continued to rise (Schechter and Grether, 2008). Ratajczak’s discussion of thimerosal restates common discredited hypotheses and several well-controlled study designs have failed to demonstrate a causative link between mercury contained in thimerosal and ASD (Immunization Safety Review, 2004; Thompson et al., 2007). A summary of available research, updated in 2004 by the Institute of Medicine, could have provided the author with a more expansive list of studies and opinions regarding ASD and vaccine issues from experts in the field (Thompson et al., 2007). A second level of review has occurred in our courts where scientific data must provide legal arguments justifying causation. In several levels of judicial review, thimerosal has not been found to be a cause of autism (Mead v. Secretary of Health and Human Services Case No. 03-215V; Snyder v. Secretary of Health and Human Services Case No. 01-162V; Cedillo v. Secretary of Health and Human Services Case No. 98-916V). Therefore, large-scale epidemiological studies have failed to demonstrate a link between thimerosal and neurodevelopmental disorders, including autism and ASD.

Experimental animal models have also shown little effect of ethylmercury (EtHg) at levels far greater than those obtained through vaccination. Even in larger-dose exposure studies using mouse models in which neurological effects are seen, the damage to neurological tissues are generally reversed over short periods of time, a reversal that does not exist in human autism (Hornig et al., 2004); additionally, even the initial affects of thimerosal have not been replicated (Berman et al., 2008). In 2004, based on an average birth weight (BW) of infants at 2, 4, 6 and 12 months, the total thimerosal exposure for an 18-month-old child conservatively would have been a total of 26.5 µg/kg BW over 72 weeks. As of 2010, the thimerosal exposure due to vaccines would be significantly lower and yet the prevalence of autism remains high and may even be more prevalent. The statement on p. 70 that “thimerosal is indeed harmful” could be made for every chemical. A quote often used by toxicologists is “the dose makes the poison.” An excess of anything, including oxygen and water, which are essential for life, can lead to detrimental health outcomes. When considering the toxicity of a chemical, the exposure dose, the duration of the exposure, the dosage per kg body weight, the exposure route and the age, genetics and health status of the individual being exposed must be taken into consideration. Thimerosal (which is rapidly metabolized in vivo to EtHg and thiosalicylate) as well as other forms of mercury are emphasized by the author without any discussion of these critical parameters. With regard to causation of autism, emphasis has been placed on thimerosal only because it is known that mercury is a neurotoxicant; however, for assessing the risk associated with any toxicant, there must be consideration of the above-mentioned variables. These variables were not mentioned when Ratajczak made the conclusion that the thimerosal in vaccines is harmful.

Dietary exposure to methylmercury (MeHg) has been well-documented to cause neurodevelopmental deficits, but the exposure dose and duration of the exposure are critical determinants. Inorganic mercury is highly toxic especially to the nervous system and the kidney (National Research Council, 2000; Clarkson et al., 2003) and due to its well-known neurotoxicity there has been a public health focus on prenatal exposure since most toxicants can cause more damage during fetal development. The tragic MeHg poisonings in Japan and Iraq due to consumption of contaminated fish and grain, respectively, emphasize the detrimental effects of mercury on the developing nervous system (Bakir et al., 1973; Harada, 1995). In these examples, fetuses and newborns were exposed to relatively high doses of MeHg for months to years and even after these relatively long-term high-dose exposures, some individuals exposed had minor to no apparent neurological problems and symptoms often became less severe with time. Autism is not a transient disorder and thimerosal (or EtHg) is not the same as MeHg exposure, two points that the author fails repeated to understand or discuss when proposing thimerosal as a cause of autism. Additionally, EtHg from thimerosal is cleared more rapidly from the body, has lower penetrance to the brain and decomposes faster than MeHg (Aschner and Ceccatelli, 2010); combining these parameters suggest that thimerosal would be less neurotoxic than MeHg. Dose, duration of exposure and speciation differences among the various forms of organomercury were not considered by Ratajczak.

The article by Schechter and Grether (2008), which Ratajczak indirectly cited, demonstrated no change in autism prevalence following the elimination of thimerosal in vaccines. A follow-up study by the same group points out data suggesting the introduction of thimerosal-containing influenza vaccine in 2004 could temporally account for rising autism prevalence. However, Ratajczak failed to recognize many of the larger cohort studies to date, which serve to refute this argument. In particular, Hviid et al. (2003) retrospectively studied 467,450 children in Denmark; 2,986,654 person-years after accounting for lost to follow up and revealed no link between thimerosal and autism. This study is significant, as in Denmark− since 1970− only the whole-cell pertussis vaccine administered to children contained thimerosal; during the period 1992–1997 no vaccines contained thimerosal; and since 1997 acellular pertussis vaccine has been used, which also does not contain thimerosal. These differences allowed the research team to compare autism rates in children who received no thimerosal with those children who had received one (25 µg EtHg), two (75 µg EtHg), or three (125 µg EtHg) doses of thimerosal-containing vaccine. There was no increased risk of autism in children who received thimerosal vaccines versus those who did not, nor was there a dose response to thimerosal administration and autism. Verstraeten et al. (2003) also conducted a tiered study of over 200,000 children in the United States and found no associations between thimerosal-containing vaccines and autism.

One final twist in the Ratajczak opinion that vaccines are a cause of autism is her suggestion that vaccines, which contain some human DNA, could hypothetically get into developing cells and undergo homologous recombination with the genome. Such a process could potentially alter the normal development of a cell type, tissue, or organ. However, there have been numerous studies attempting to purposefully induce homologous recombination for the development of gene repair therapies and the studies have indicated that relatively high copy number is needed and even then successful recombination has not been efficient (Richardson et al., 2002; Nakayama, 2010). This process is not a possible mechanism of autism using any of our current vaccines.

The Ratajczak review takes us through a confusing array of possibilities with none leading to the ultimate solution of the riddle: “what causes autism?” From a gene and environment perspective, the causes of ASD are indeed a complex web of interacting factors. As recently reported with regard to the genome alone, there is a sizable pool of genome copy number variations that have been identified as candidate ASD risk loci and this study “only considered ASD variants not found in controls, which may oversimplify the biological role that certain variants might contribute to ASD etiology” (Gai et al., 2011). The environmental biological, chemical and psychological stresses, which can affect gene expression, add further complexity to elucidating the causes of autism. However, continued attempts as in the Ratajczak review to implicate factors already considered and scientifically ruled out as general causes only delays our progress toward understanding the root causes of autism.

Our genes are best considered information warehouses that are constantly looking for signals to tell them what to do. As an example of the importance of genes and environment in human disease, Sørensen et al. (1988) evaluated genetic influences on premature deaths across disease categories; they compared siblings raised in adoptive environments to those raised by biological parents. Infectious disease and cardiovascular causes of premature deaths were strongly predicted by inherited parental genes versus environment; the latter had more influence for cancer-related deaths. Such studies regarding “nature versus nurture” are ongoing for autism and preliminary data indicates that the inherited component of risk will be a critical factor in whether a child develops autism (Pinto et al., 2010). Environmental stimuli that trigger release of genetic information are likely to use common mechanisms, such as oxidative stress, to set off the cascade leading to disease.

The most challenging clinical research question ahead of us is whether a “susceptible” individual can be prevented from developing autism. This question is highlighted by the recent case reported by the United States CDC of a child with a mitochondrial disorder who developed autism months after vaccination. Was this individual genetically “destined” to develop autism despite vaccination? (Offit, 2008). We have addressed two concerns raised by Ratajczak with regard to autism; it is neither vaccination nor thimerosal that causes autism. The aftermath of the Wakefield manuscript and the impact of anti-vaccination movements on vaccination rates and childhood mortality due to preventable infections have been significant (Fair et al., 2002; McBrien et al., 2003; Parker et al., 2006). In 1802, the British caricaturist James Gillrary published the cartoon entitled “The Cow-Pock− or− the Wonderful Effects of the New Inoculation!” which depicted individuals growing cow-like appendages following vaccination with cowpox virus to prevent smallpox. As Gillrary’s cartoon points out in 1802, the anti-vaccine mindset based on irrational beliefs existed even as deaths from smallpox still raged across the globe. The Ratajczak review reminds us as scientists that we must continue to challenge beliefs and opinions based on coincidence and speculation. We must work with the public to recognize that when diseases of children, such as autism, elude our best efforts to understand the cause, it is not acceptable to substitute our emotional prejudices for the truth. Failure to do so will risk many unintended harmful consequences to the public health.

David A. Lawrence

Jeffrey S. Kennedy

Notice of correction:

This paper published online on 16 June 2011 was incorrectly classified as an Editorial. The current version labelled as Letter to the Editor is correct. The publisher would like to apologize for any inconvenience caused.

Acknowledgement

The authors thank Dr. Michael Aschner (Vanderbilt University) and Dr. K. Michael Pollard (Scripps Research Institute) for their helpful suggestions.

Declaration of interest

The authors report no declarations of interest.