8.4 Testing the MMR-autism hypothesis

Fitzpatrick, M. (2004) Chapter 8 ‘The Lancet Paper’ taken from MMR and Autism: What Parents Need to Know, London, Routledge. Copyright © 2004 Michael Fitzpatrick.

In the concluding ‘discussion’ section of their Lancet paper, Dr Wakefield and colleagues suggested that further investigations were needed to examine the syndrome of ‘autistic enterocolitis’ and ‘its possible relation’ to MMR (Wakefield et 1998: 641). They indicated two directions for further research. First, the authors observed that if there were a causal link between MMR vaccine and this syndrome ‘a rising incidence might be anticipated after the introduction of this vaccine in the UK in 1988’. They considered that published evidence was inadequate to answer this question, inviting further epidemio-logical research to clarify it. Second, they reported that ‘virological studies’ (presumably those later reported by the team headed by Professor John O'Leary in Dublin, Ireland) were ‘underway’. Let us now examine the outcome of attempts to substantiate the MMR-autism hypothesis through researches in these areas.

In its responsibility for vaccine safety, the Medicines Control Agency commissioned an epidemiological study to investigate the question of whether there was an increase in cases of autism in Britain following the introduction of MMR. Dr Wakefield's challenge to analyse any rise in incidence was taken up by Professor Brent Taylor, community paediatrician at the Royal Free Hospital, and a team including vaccine specialist Dr Elizabeth Miller and Open University statistician Dr Paddy Farrington. Their results were published in The Lancet in June 1999 (Taylor et al 1999a).

They identified all known children with an autistic spectrum disorder born between 1979 and 1998 in eight North Thames health districts 498 children in all – and studied their medical and vaccination records. They found that:

• although the number of cases of autism had increased steadily since 1979, there was no sudden ‘step-up’ or change in the trend line after the Introduction of MMR in 1988;

• there was no difference in age at diagnosis between the cases vaccinated before 18 months of age, after 18 months of age, and those never vaccinated;

• there was no clustering of developmental regression in the months after vaccination.

They concluded that ‘our analyses do not support a causal association between MMR vaccine and autism’ (Taylor et al 1999a: 2026).

The authors themselves acknowledged two limitations of their study. They could not verify the diagnoses of autism in all cases and they may have missed some cases. They relied on clinical notes of variable quality and many did not contain systematic or regularly updated information, which would have allowed independent validation of diagnosis. Despite making ‘substantial efforts’ to identify all cases, they may have missed some children who were not known to local health or education authorities. However, it is unlikely that these factors significantly affected the overall results.

In a letter to The Lancet, Dr Wakefield criticised the Taylor study on three grounds (Wakefield 1999). He claimed that the statistical methodology used (‘case-series’) was inappropriate to detect temporal associations between vaccination and conditions, such as autism, characterised by an insidious onset and delay in diagnosis. On the contrary, the authors replied, this method was particularly suitable for this sort of study, which has a good record of revealing rare adverse effects (Taylor et al 1999b). Dr Wakefield's second objection focused on the authors’ judgement that one finding that of a marginally significant raised incidence of parental concern between 0 and 5 months after MMR – was a statistical artefact. The authors claimed that one such finding (out of 14) might have been expected by chance, and that it could be explained by ‘the combined effect of approximate recording of parental concern at 18 months and a peak in MMR vaccinations at 13 months’. Finally, Dr Wakefield made the accusation that the authors had ‘failed to declare’ the fact that some of the children in the study may have received MMR as a result of a catch-up campaign. The authors’ rebuttal was that these children had been identified and that in all cases in which the age of first parental concern was recorded, it preceded vaccination.

If epidemiological studies failed to support the MMR-autism hypothesis, what about the virological studies? During 2002 two papers based on studies of intestinal biopsies on Dr Wakefield's ‘autistic enterocolitis’ patients by a team lead by Professor John O'Leary in Dublin were published.

In the first paper, published in February, the researchers claimed to have identified fragments of the measles virus in intestinal tissues of 75 out of 91 children with inflammatory bowel disease and developmental disorder (Uhlmann et al 2002). However, this study did not indicate whether the children had had measles or MMR. The authors did not indicate whether they had found whole measles virus, whether of wild or vaccine strain, or any other viruses, such as mumps and rubella. Many commentators wondered whether inadvertent sample contamination or some other technical error with the notoriously difficult reverse transcriptase polymerase chain reaction assays might explain these results (Afzal et al 2003). The study was also criticised on the grounds that the controls were not matched for age or time since vaccination. Others observed that, even if these findings were confirmed and replicated, the presence of measles virus fragments in the gut would not prove that they caused either inflammatory bowel disease or autism.

In response to the controversy generated by his paper, Professor O'Leary issued a statement insisting that he had ‘not set out to investigate the role of MMR in the development of either bowel disease or developmental disorder, and no conclusions about such a role could, or should be, drawn from our findings’ (O'Leary 2002a).

In a presentation in June 2002 to a US congressional committee Dr Wakefield claimed that a new study, due to be published by Professor O'Leary, had confirmed that the measles virus present ‘in the diseased intestinal tissues of children with regressive autism’ was indeed derived from the MMR vaccine (Wakefield 2002a). For Dr Wakefield, these studies constituted ‘a key piece of evidence in the examination of the relationship between MMR vaccine and regressive autism’. Professor O'Leary, however, promptly rejected Dr Wakefield's interpretation of his work, insisting that it ‘in no way establishes any link between the MMR vaccine and autism’. (O'Leary 2002b). Indeed, he strongly recommended that parents should give their children MMR¹.

An abstract (summary) of the new O'Leary study was duly presented at the annual meeting of the Pathological Society of Great Britain and Ireland in Dublin in July 2002. This was a pilot study designed to discover whether the measles virus RNA found in the guts of children in the earlier study originated in wild measles or from immunisation. The paper described a technique for discriminating between two closely related genome sequences, which the authors claimed could distinguish between wild and vaccine strain measles (by identifying a single nucleotide at position 7901 of the genetic code of the wild measles virus). They found vaccine-strain measles virus in the gut biopsies of 12 children with inflammatory bowel disease and development disorder (and confirmed wild measles strain in brain specimens of three patients with SSPE – a rare complication of measles). They concluded that ‘this pilot study corroborates our earlier findings of an association between the presence of measles virus and gut abnormalities in children with developmental disorder, and indicates the origins of the virus to be vaccine strain’ (Shiels et al 2002).

However, an immediate response to this study from the WHO collaborating centre for measles in the UK challenged the validity of the technique used by O'Leary's team. This indicated that the method used was not able to distinguish between wild and vaccine strains (it could result in several wild strains being incorrectly classified as vaccine strains). ‘Consequently’, it concluded, ‘the technique described does not reliably discriminate between wild and vaccine measles virus’ (Brown et al 2002). When presented with this information at the US congressional hearings on autism, Dr Wakefield accepted that if this method could not reliably make distinguish the two different forms of measles, then the Conclusion drawn by the paper was not justified. The first piece of evidence promising some support to the hypothesis advanced by Dr Wakefield in 1998 was thus discredited even before publication.

¹ It is interesting to note that Professor O'Leary's repudiation of the claims, made on his behalf by Dr Wakefield and his supporters, has never been acknowledged by the anti-MMR campaigners, who continue to cite O'Leary's research in support of the MMR-autism thesis, in explicit defiance of his statements to the contrary.