Paul Shattock & Dawn Savery
Autism Research Unit, University of Sunderland, UK
Paper presented at the Durham Conference April 1996 [Modified October 2006]. Please note that several advances have been made since this paper was first presented.
Amongst the multitude of diseases and disorders, which can afflict mankind, pervasive developmental disorders such as autism remain one of the very few that are still defined in terms of observed symptoms. Thus it remains a "syndrome" with no clear explanation for its causation or existence at all. Over the years, there have been many attempts to determine the underlying causes but, as yet, there are no universally accepted explanations. Psychiatrists, psychologists, geneticists, anatomists and electro-physiologists and other groups have provided explanations based upon their own understanding. Each of these groups has provided valuable insights but an intellectually satisfactory explanation remains tantalisingly out of reach.
The existence of the syndrome seems to be the only factor which is universally accepted yet if it exists there must be a rational explanation for the coincidence of the strange behavioural, perceptual and psychological abnormalities which autism entails. If the sequence of events which may underlie the syndrome can be identified, it will be possible to work out ways in which we can intervene and either prevent the syndrome or, at least, ameliorate the problem. We would like to advance the case for autism being the consequence of a metabolic disorder.
We have been studying the urinary profiles of people with autism for over 10 years. We have employed a variety of separative techniques including gel filtration, capillary electrophoresis, mass spectrometry and a variety of High Performance Liquid Chromatographic (HPLC) systems which are constantly being modified as developments in instrumentation and technology occur and as our knowledge and understanding increases.
During this time, we have examined urine samples from approaching 12,000 subjects and certain characteristics are, at last, beginning to become apparent. Initially, we collected little clinical and background information but in our more recent studies such information has been collected. As we have accumulated this information, certain correlations have become evident and a consideration of these has permitted us to create a hypothetical model for autism. Our model is based upon our own studies but also relies heavily upon the work of other researchers and in particular that of Dr. Kalle Reichelt (Oslo, Norway) and Dr. Rosemary Waring (University of Birmingham, UK).
This model is based upon acceptance of the opioid excess theory of autism as expounded initially by Panksepp (1979) and extended by Reichelt et al (1981) and ourselves (e.g. Shattock et al, 1990; 1991). We support the hypothesis that autism (or at least some cases of autism) could be the consequence of the action of peptides (short chains of amino acids) of exogenous origin affecting neurotransmission within the Central Nervous System (CNS). We believe that these peptides result in effects which are basically opioid in nature (akin to drugs such as morphine) and that they may either, themselves, have direct opioid activity or that they may form ligands for the enzymes which would normally break down such opioid peptides that occur naturally within the CNS. In either case, the consequence would be the same. The CNS neuroregulatory role, which is normally performed by the natural opioid peptides such as the enkephalins and endorphins, would be intensified to such an extent that normal processes within the CNS would be severely disrupted.
The presence of this intense opioid activity would result in a large number of the systems of the CNS being disrupted to varying extents. Perception, cognition, emotions, mood and behaviour would all be affected. By the same mechanisms, higher executive functions would be disrupted (see Ozonoff et al 1994) and the many and diverse symptoms by which autism is described and defined would result. The presence of these opioid peptides would also affect the immune system in a variety of ways. Opioids are well known intermediaries in neuroendocrinimmunological processes but a detailed discussion of their precise role is beyond the scope of this paper. These opioids can, perhaps through this mechanism, also be involved in the pruning of CNS nerve cells, which occurs in-utero and in the early years of infancy. Elevated levels of opioid peptides at this critical time would result in excessive pruning of neurones within the CNS. The effects are likely to be similar to those seen in people with autism and reported in the cerebellum and corpus callosum by Courchesne (1994) and in the anatomy of the Purkinje cells by Bauman (1985). These effects have been previously described (Shattock et al, 1991).
The role of opioid peptides on the immune system is well documented. It is suggested that the abnormalities in the immune system, which have been reported in people with autism (e.g. Marchetti 1990) could be the consequence of excessive opioid activity. The effects of such opioids are difficult to predict as they are concentration dependent and the effects may be to enhance or depress various elements of the system depending upon the concentration of peptides. We believe that these peptides are derived from an incomplete breakdown of certain foods and, in particular, gluten from wheat and some other cereals such as barley, rye and oats and from casein from milk and dairy produce. There is an increasing body of evidence, both circumstantial and direct to support these hypotheses and much of it is presented elsewhere [in these 1996 conference proceedings] by Reichelt and colleagues. A selection from our own data will be presented in this paper.
Our model can be represented diagrammatically as in Figure 1 (a - d)
Figure 1(a) represents the situation in a clinically "normal" subject. Each "star" represents one peptide molecule with biological, in this case opioid, activity. It is clear that when any proteins are broken down in the gut, peptides will occur as intermediate compounds which will be broken down further into their amino-acid components. It has also been shown that, even in normal, healthy individuals, a proportion of these may cross from the intestines into the bloodstream. For example, 10% of the peptides may cross through the normal intact gut wall and appear in the blood stream. If 10% of this complement crosses the blood brain barrier (in both cases the amounts are reasonable for small peptides), then 1% of the total peptides present in the gut will have reached the CNS. Once there, they may directly regulate transmission in all of the main neurotransmission systems or, alternatively, may form ligands for the enzymes which would normally break down the opioid peptides which occur naturally in the CNS. In either case, the consequence would be an increase in opioid activity.
In this normal situation, the levels of peptides in the gut are comparatively small and the quantities reaching the brain are minimal so the net effects are negligible. In the example shown in figure 1 (b), there are vastly increased levels of peptides in the gut and, given the same order of leakiness of the gut wall and the blood brain barrier as before there will be an increased quantity of peptides reaching the CNS and perhaps clinically significant consequences.
The reason for the increased levels of such peptides in the gut may be the inadequacy of the enzyme systems, which are responsible for their breakdown. For example, there may be genetically determined deficiencies of the required endopeptidase enzymes. There may be shortages of the co-factors, such as vitamins and minerals required for the enzymes to function. Alternatively, the pH in the relevant areas of the guts may be inappropriate for the specific enzymes to act.
Figure 1 (c) represents the situation in which the levels of peptides in the gut are normal but, for some reason, the gut wall is excessively leaky so that vastly increased quantities of the peptides will cross the gut wall and enter the blood stream. Thus, there will be an increased level of peptides in the CNS and possible clinical consequences. Increased permeability of the gut wall has been demonstrated in a high proportion of, but not all, children with autism (D'Eufemia 1996).
There are a number of factors, which could result in increased leakiness of the gut. There may be damage caused by purely physical action such as a surgical operation or some natural flaw. The deficiencies in the Phenyl Sulphur Transferase (PST) systems, as described by Waring (1993), would lead to increased permeability of the gut wall. Normally the proteins lining the gut wall are sulphated and, in this state, form a continuous protective layer over the surface of the gut wall. Where there is insufficient sulphation, the proteins clump together and the layer becomes discontinuous. The net result is an increased permeability of the gut wall. If this were the case, the passage of peptides across the gut wall would be greatly enhanced.
Over the years, some parents have claimed that the autism, which they have seen in their child, only became apparent after an immunisation programme of some sort. The traditional and orthodox response has always been to refute these suggestions and point out that the benefits of these programmes far outweigh any potential problems. It is usually claimed that the coincidence of the occurrence of autism (or epilepsy or certain other conditions) and the use immunisation programme is of no significance. However, these parental reports are so persistent and numerous that they should not be ignored. Wakefield (Balzola et al 1995) has recently demonstrated that the Measles element of the MMR vaccine will produce fairly gross abnormalities in the gut wall. Sometimes the damage is very severe and he suggests that Crohn's disease could be the ultimate result. A dramatic increase in the leakiness of the gut wall would be a logical consequence. Given that the immune system may already be compromised by the presence of limited amounts of exogenously derived opioid peptides and that the vaccines frequently consist of living, attenuated (weakened) strains of the disease, these consequences would no t be surprising.
In figure 1 (d), the blood brain barrier is less effective than normal so that any opioid peptides in the blood stream would easily pass into the CNS and exert their full range of actions. The blood brain barrier is a complex system, which is partly physical and partly biochemical. The biochemical element consists, in part, of enzymes, which should destroy potentially harmful substances such as exogenously derived peptides. Since, according to these hypotheses, the peptidase activity in the individual with autism may, in any case, be depressed, the barrier may be somewhat more permeable than normal. Once again, there may be other environmental factors, which could exacerbate the process either slightly or dramatically. There are occasions where physical damage, perhaps following surgical intervention, appears to have initiated the appearance of the autism. It could be that the autism also becomes apparent after the child has experienced a bout of encephalitis or meningitis. In both cases, one could reasonably expect damage to the barrier and the entry of larger amounts of these peptides and other compounds. More controversially, the role of immunisations must also be considered in this context. The large numbers of parents who can pinpoint the time of the appearance of autism to within days or, in some cases, hours of an inoculation of some sort can no longer be dismissed as mere cries from those seeking someone to blame for the plight of their child. Whilst the role of vaccination programmes in the causation of autism remains unproven, the mechanism by which it could occur does exist. Within their early years, most children in the Western world are immunised against Measles, Mumps, Rubella, Diphtheria, Pertussis and Tetanus and many of these vaccines use living, attenuated forms of the disease. Given the compromised state of the immune system of the child with autism, there could be risks inherent to this procedure.
Thus in many people who are completely asymptomatic there could exist comparatively small quantities of these exogenously derived opioid peptides. When something seriously affects the permeability of either the gut wall or the blood brain barrier or perhaps both, the consequences could be very severe. Of course, this is not to say that such environmental factors are always necessary prerequisites for the appearance of autism in an individual. They could convert a genetic predisposition into a clinically hugely significant and pervasive condition.
Allergy or Toxic Response?
It should be stressed that this hypothetical model involves the occurrence of effects, which are primarily the consequence of toxicity rather than allergy. Thus, in the vast majority of cases, tests for allergy, which rely upon the presence of specific antibodies to casein or gluten or wheat products (IgE), will prove negative or at least only minimally positive. Confusion often exists on this issue and the effects of wheat on people with autism are ruled out on the basis of blood tests, which may not be of relevance. Since the peptides will affect the immune system, in a way, which is variable in extent and nature, there may be evidence of an overactive immune system in certain respects so that multiple allergies are frequently reported where such testing has been carried out.
There is no doubt that a small proportion of the population do have a definite Coeliac Disease condition, which underlies the autism. Since the vast majority of people with autism have not been tested for this condition, it is not possible to produce reliable statistics but we would suggest that our own data and that of Reichelt (both unpublished) would suggest that the condition exists, in 3 -4% of the population with autism.
In the majority of people with autism, there also seems to be some evidence of allergic responses and often to milk and gluten containing products. The immediate consequences are probably minimal but should be born in mind when taking a full clinical picture.
Some other considerations
There are other factors, which may be of relevance in the causation of autism or at least in the appearance or exacerbation of the symptoms and these are accommodated within the theory. Candida has often been reported as being involved in autism but its role is unclear. Shaw (1994 and 1995) has reported the presence of metabolites of fungal origin in the urine of people with autism but this is indicative of the existence of such organisms and does not necessarily imply direct involvement in the causation of the autism. Although there is a lack of supportive data, it is often suggested that an unusually high proportion of people with autism suffered from ear infections in infancy. This may or may not be the consequence of the abnormalities in the immune function referred to above but it does often result in the use of antibiotics. The consequence of the use of these antibiotics would be the reduction in the extent of the bacterial flora of the gut and an increase in those organisms such as Candida, which are not affected by these antibiotics.
The role of Candida is still very controversial but if it is present in the gut it will undoubtedly affect the gut wall and increase its leakiness. Its invasive potential would be greatly enhanced by the deficiencies in the protein layer lining the gut, which were referred to earlier as being consequent upon the deficiencies in the Sulphur Transferase systems. Given that the Sulphur Transferase system is, in any case, operating at greatly reduced efficiency, anything which utilises the system would be likely to amplify the effects of the deficiency. Foods or drugs rich in phenolic components, have often been reported by parents as having a deleterious effect on their children. Thus drugs such as paracetamol (acetaminophen), foods such as chocolate or drinks derived from apples or citrus fruits could, indirectly, result in a leakier gut and an increase of symptoms.
What Causes Autism?
If the opioid excess theory of autism is accepted there can be no one simple answer to the question of what causes autism. The condition could be described as the result of a metabolic disorder with many individual contributing elements. There is no doubt that there is a genetic component; although this genetic predisposition must become apparent through some process or processes. Thus there may be deficiencies in at least two genetically determined elements. The peptidase system and the Sulphur Transferase systems are non-functional so either of these could be considered as causative. The opioid peptides come, in the main, from gluten and casein so it could be argued that these substances are the cause. It would appear that Candida, various phenolic compounds and various vitamins and minerals have a role to play but on their own, these elements would not result in autism. The picture is basically a simple one although there is a complex interplay between a large number of factors. Reducing the effects of any of these contributory factors would be helpful in ameliorating the symptoms of autism but, in some cases, the effects are minimal when taken in isolation. Since, in this hypothetical model, the opioid peptides have a central role in the story, the remainder of this paper will concentrate upon these substances and our recent researches in this area.
If peptides are present in the blood, they will tend to be collected by the kidneys and dumped in the urine. Thus, the peptide content of the urine will, to some extent, be reflective of the content of the blood. Given that this method must still be regarded as experimental and that the removal of blood samples, from children, with or without autism, can cause distress, we have concentrated upon the urine rather than blood for our studies. Our preliminary results together with the details of the methodology have already been published (Shattock et al, 1990). Although there have been certain minor alterations the process remains essentially the same.
Briefly, a fresh sample (5ml) of urine is subjected to a preliminary clean-up process (SPE), which separates peptides and other similarly sized and charged molecules from the small, water soluble molecules and the larger molecules. The peptide fraction is then injected onto a column and a solvent passed along the column in a process known as gradient elution High Performance Liquid Chromatography (HPLC). The products are detected as they emerge from the column and the results recorded in graphical and numerical form.
Figure 2 & 3: HPLC results from control and autism samples.
Figure 2 shows a typical urinary profile (chromatogram) obtained from a normal medication free subject with no evidence of autism or other related pathological condition. The scale along the bottom represents the time taken for each material to elute (in minutes). The area under the peak gives some indication of the amount of each component, which is present in the extract. The peptides with biological activity tend to appear in the region, which lies between about 18 and 30 minutes (in this particular system). It has not yet been proven that all of the peaks between these points are due to peptides but we speculate that some are due to the results derived from previous research studies undertaken. Although there are slight variations from day to day that are reflective of transient dietary or other changes, the patterns for each individual retain characteristic features. The control sample shows a number of peaks in the relevant area but nothing of any major consequence.
Figure 3 shows the profile of a sample taken from a subject with autism. Generally speaking, the peaks in the relevant area are larger and more numerous. In particular, the vast majority of subjects show their major peak at around 20-21 minutes*. At this point in time, it is not possible to assign a structure to the substance causing this peak.
* [Update: The structure of this peak has been subsequently elucidated as being trans-indolyl-3-acrloylglycine (IAcrGly); a more thorough account of this compound and its potential relevance can be found here]
In a smaller subgroup of subjects (less than 10%) a major peak appears in the profile at 25-26 minutes. This peak co-chromatographs precisely with bovine beta-casomorphin 7, one of the principal opioid peptides obtained from the breakdown of cow’s milk, in several different chromatographic systems. Reichelt (Knivsberg 1990) has also demonstrated the presence of bovine casomorphin using an immunoassay technique.
Those children whose primary problem appears to be milk related tend to have this or other casomorphin peaks as present in their sample. Although it is difficult to be precise, it does appear that these subjects had a form of autism, which appeared to be evident from a very early age. The vast majority of children in our study are reported as having developed fairly normally in their early days and then, at age 2-3, to have developed the symptoms of autism. Profiles from these subjects tend to have the major peak (IAcrGly), at about 20-21 minutes, which we loosely associate with gluten. It is significant that infants are not normally introduced to dietary gluten until they are aged about 12-18 months. Commercial and infant baby foods tend not to contain wheat based products because of the possibility of Coeliac Disease.
During the course of our investigations, we have encountered a number of subjects who give what could be classed as "false positive" results. That is, the "20 minute" peak (IAcrGly) is the largest within the specified area. Many of these subjects are, as far as can be ascertained, completely devoid of clinical abnormalities. In some cases, however, there is a clear indication of some other abnormality. For example, subjects with dyslexia consistently give a "positive" result. We have also found subjects who show some of the behavioural and psychological abnormalities, which may be within the same spectrum of disorders. These include Attention Deficit Disorder (ADD), Attention Deficit Hyperactivity Disorder (ADHD) and also some types of Obsessive Compulsive Disorder.
Although, as yet, we have been unable to reduce the peaks to meaningful numerical data, we are able to make certain observations about the profiles. For example, the profiles given by the groups mentioned above are much more similar to the profiles of subjects with Asperger Syndrome than with the more severe, classical forms of autism. Profiles from subjects with ADD and ADHD are now being studied in a more systematic way.
During the course of these studies, we have been examining urine samples from other members of families where autism exists. On the whole, even those subjects who do give a positive result exhibit no clinical abnormality but in many cases, parents (or the subjects themselves) do report evidence of mild difficulties. In one of our earlier series of studies (unpublished) some 46% of mothers; 42% of fathers and 56% of siblings showed such abnormalities. This must be seen against a figure of 25% for control populations.
Upon detailed questioning it is evident that some (but certainly not all) of these siblings and some of the parents do show symptoms, to a greater or lesser extent, of the conditions described above. There is considerable anecdotal evidence, which suggests an increased incidence of these conditions in the families where autism exists but as far as we are aware, little in the way of hard, supportive data.
Consequences for Therapy
For many years, parents of children with autism and associated spectrum disorders have been investigating the effects of diets from which gluten and casein have been removed. On the whole these efforts have been treated with cynicism, scepticism or outright hostility by many healthcare professionals. There are however, physicians, dieticians, teachers and other healthcare professionals who are prepared to investigate these ideas for themselves but, on the whole, parents are not encouraged to pursue such interventions. The Norwegian studies, referred to earlier (Knivsberg et al, 1991), do provide supportive evidence for the effectiveness of such interventions but until there is further replication, professionals find it difficult to offer wholehearted endorsement to such efforts (see published references).
The anecdotal reports from parents are more than encouraging and we are aware that literally thousands of parents are intervening in this way with or without the support of their physicians. We are aware of a limited number of reports where very dramatic improvements have been reported by parents of children who are self-injurious. We are also aware of a number or cases where very young children have, apparently, shown very dramatic and rapid improvements but these cannot, at this stage, be considered as conclusive. As far as we are aware, the first case where a physician was prepared to prescribe gluten-free products on the National Health Service (NHS), for autism, was in March 1995. Although we have not collected appropriate data, we would estimate that in over 50% of the cases in which we have been involved, physicians are sufficiently satisfied with the theoretical basis and with the observed results to prescribe these products in this way. Clearly, it is time that appropriate clinical trials are initiated so that the effectiveness of such interventions can be gauged.
Although, in an effort to ameliorate the symptoms of autism in their children, parents have been investigating the usefulness of diets devoid of gluten and casein, for many years these methods have not been accepted by many orthodox medical practitioners. This is largely because there has been no hypothesis that has been sufficiently able to combine all the relevant evidence into a single, comprehensible series of events. We believe that the data and explanation we have provided, together with that provided by our colleagues will go some way to filling that requirement. As well as continuing our investigations into the nature, identity and origins of the abnormal complement of materials evident in the urine of people with autism (and related disorders), we are initiating a research programme to study the effectiveness of the removal of dietary gluten and casein in the treatment of autism.