My discovery about abnormal organic acids in autism began as many discoveries do, as an accident. In the 1960’s, a great deal of progress had been made in discovering the biochemical abnormalities that caused a number of diseases called inborn errors of metabolism using a technology called gas chromatography-mass spectrometry. It seemed possible that this new technology might be applied to any disease. However, thirty years later, very little progress had been made in solving the mystery of a number of diseases like autism, schizophrenia, and Alzheimer’s disease.
In the field of metabolic diseases, urine samples are analyzed for their chemical constituents after extracting the chemical compounds from the urine using organic solvents such as ether and ethyl acetate. It is preferable to test urine over blood because urine is a filtrate of blood in which much of the water has been removed, so that the concentration of a compound in urine might be 100 times more concentrated than it was in blood. A very high concentration of characteristically abnormal chemical compounds would indicate the likely presence of a genetic disease. For example, when a child has PKU (or phenylketonuria), which is a genetic disease where a genetic mutation is present, very high concentrations of chemical compounds called phenylketones will appear in the urine. This mutant gene codes for an abnormal form of the enzyme phenylalanine hydroxylase that converts phenylalanine to tyrosine. Since the enzyme is defective, phenylalanine is not converted to tyrosine and phenylalanine builds up in the blood just as a logjam begins in a narrow part of a stream. If a child with PKU is treated with a diet low in phenylalanine as an infant, the child will develop normally. However, if the diagnosis of PKU is not made until the child is much older; the child may be significantly impaired and suffer mental retardation (1).
Let me give another analogy: You go into a bank, open an account and make deposits for several weeks. After about a month you go back into the bank and you say, “I’d like to know my account balance.” The teller looks at you and says with a straight face, “A lot”. You feel concerned about this lack of information and press her for more information, and she says, “You really have more than most people do”. That is still not satisfactory but there is no manager available so you walk away feeling confused and decide to go back later when a different teller is on duty. The next time you come in, you request the manager, asking again for your balance and this time the manager says, “Not much." Although this type of accounting may be adequate for comparing the assets of Bill Gates and a street person, it is not much help for those in the middle class. In essence, the majority of metabolic disease testing that was performed ten years ago was the “a lot” or “not much” variety, and this still exists in as much as 50% of the testing done today.
I suspected that many subtle changes in the body’s metabolism were being overlooked in using this kind of technology as a result of the “not much” and “a lot” kind of interpretations. What I set out to do was to quantitate the changes in the different molecules in the urine just as the bank accountants in a bank balance the money transactions. I was successful because of new computer software that allowed for the rapid quantitation of very complex data. If it were not for this particular software, my work would not have been possible.
This computer software had originally been designed for the environmental field. Our drinking water, sewage and ground water can contain many kinds of pesticides, herbicides, and industrial chemicals. Testing for all of these chemicals requires very sophisticated computer software and this software was ideally suited for doing metabolic disease testing. The goals I set out to achieve were (1) to identify every chemical that I could, and (2) to quantitate everything I possibly could and do it as accurately as possible.
If we could know everything possible about people, including what kinds of chemical compounds were normal for them, then it would be easier to identify what was going on in the metabolism of a patient with a particular disease. Prior to beginning testing, we sent samples out to another laboratory performing the “a lot, not so much” kind of testing and I was very surprised to see that about 98% of the samples came back with an interpretation of normal. It does not seem possible, in my opinion, that someone could have a devastating disease, and not have it alter their metabolism in some way.
After receiving this information and filing it away in my mind, I continued to remain skeptical of this common perception that microbial products were unimportant. The body did not have a metabolic segregation system in which human metabolites were allowed into certain areas of the body, while microbial products were separated into other compartments. All of these products were intermixed throughout the body. Several months after initiating my new laboratory service, Enrique Chaves, MD., a colleague of mine from the University of Kansas Medical School, and a pediatric neurologist who was also interested in biochemistry (a rather rare occurrence in physicians as a group) referred a woman to me who had two children with severe muscle weakness. Dr. Chaves, who had also been using the old technology in his laboratory, could find nothing unusual in the two brothers. The muscle weakness was so severe that sometimes, for several hours, these children could not even stand up. There had been an intensive search for the cause of this muscle weakness. When Dr. Chaves analyzed the lab results, he found no evidence of any genetic disease. Since I had this new technology, I was very interested in trying to find out what was going on. I told the mom that we would test samples of her children’s urine and see if we could figure out what was happening to them.
In the field of metabolic diseases, it is well known that some disease abnormalities only show up at the time the child is severely ill, i.e., if the child has a severe cold, flu or chicken pox. The biochemical pattern may be close to normal while the child is well. So when I spoke to the mom, I emphasized that we should get multiple samples rather than just a single one. Several months later, the mom came back with a whole armful of samples saved in her freezer, which were actually more samples than we usually tested in an entire month. I talked to my technologist Ellen Kassen and told her we would have to bite the bullet and get these tests under way as best we could. We began to test the samples. In each sample, I would see that there was no chemical compounds characteristic of any of the known genetic diseases, which are called inborn errors in metabolism. My overall impression however, was that these samples were still abnormal because there was a marked difference in the kinds of chemical compounds found in the urine samples of the two brothers compared to those found in the urine of normal children.
These compounds were the same ones that my colleagues said were not important because they were from microorganisms in the intestinal tract. I was now very curious about what was going on. By this time, my colleague Dr. Chaves had moved from across town to the same institution where I was located. I was able to walk across the hall and further discuss with him any possible explanation as to why these brothers had these abnormal concentrations of chemicals resulting from microorganisms. At that time, he also mentioned that, in addition to the profound muscle weakness, the brothers also had autism. When I looked at their medical charts, I saw that they also had a history of frequent ear infections which is similar to many children with autism. A brief description of the technology used for testing the samples is appropriate at this point.
Urine samples are extracted to obtain a purified extract for analysis by a gas chromatograph-mass spectrometer (GC/MS). Samples are loaded onto a sampler tray of the GC/MS. The sample is then injected into a hollow tube in the oven of this instrument called a column. The different molecules in the sample go around and around in coils of this column just like a group of horses going around a racetrack and then come out at the finish line. At the finish line the sample is bombarded by a beam of electrons that break the molecules into pieces of different sizes and shapes. The molecules can be identified because each molecule has a characteristic way of breaking up or fingerprint. The data from this fingerprint is then transferred into a computer. Then the computer analyzes all that data, makes sense out of it, identifies it and quantifies how much of each kind of molecule is in the urine sample. The increase in the capability of this technology has been phenomenal. When I first started in this field, the analysis of a single chemical compound would have taken most of the day. Now we can identify a thousand different compounds in a single afternoon.
Figure 1 shows a typical chromatogram for the analysis of the urine sample of a normal child. This profile is called a total ion chromato-gram. People who work in the field call each one of these blips that you see a peak. A peak is detected when identical molecules in the sample are swept by the pressure of an inert gas around the circular column and finish at a particular time. The size of this peak is proportional to how much of a particular kind of molecule there is. Small fast molecules cross the finish line faster than big slow molecules just as fast horses have the fastest race times. Fast molecules have the smallest transit time, which is called a retention time. The bigger the peak, the more of a compound is there. Conversely, the smaller the peak, the smaller the amount of compound.
What I found is that there was a consistent pattern of abnormally elevated chemicals in the urine samples of the two brothers with autism that were known to be derived from the intestinal microorganisms. So virtually all of the big changes that you see in the chromatogram of the child with autism (Figure 2) were due to the fact that they had much higher concentrations of the chemicals that were produced by microorganisms residing in their intestinal tracts.
Based on all the information that I had gathered, I reasoned that if the abnormal compounds from the intestinal tract had something to do with causing autism, then treatment of the microorganisms that produced these byproducts should improve the behavior of the child. I only had to wait a short time before I got the opportunity to test out my hypothesis. A child had been referred to the Neurology Department of the hospital to confirm a case of autism and organic acid testing had been requested. This child had the kind of history that is very frequent in autism.
After 68 days Bruce’s mother started running out of nystatin and began giving only 1/2 doses so that she didn’t run out of it completely. During that time the tartaric acid starting going back up and when she got the nystatin prescription refilled, the tartaric acid went back down. What this indicated to me was the fact that the nystatin was causing a marked reduction in this urinary tartaric acid. The other significant finding was that even after two months of nystatin, the biochemical abnormality would reappear within a short time of stopping the antifungal drug. In some cases, reports have been received of this same phenomenon in hundreds of other cases. Even after six months and after two or three years of antifungal treatment, there is often a biochemical “rebound” and loss of improvements after discontinuing the antifungal therapy. Several explanations are possible for this phenomenon:
What is tartaric acid and what is known about this product? A toxicology manual (3) indicates that tartaric acid is a highly toxic substance. As little as 12g has caused human fatality with death occurring within 12 hours to nine days after ingestion. Since this compound especially damages the muscles and the kidney (4,5) and may even cause fatal human nephropathy (kidney damage)(6), it was of particular interest to me since the two brothers with autism’s initial symptoms were extreme muscle weakness as well as evidence of impaired renal function. Gastrointestinal symptoms were marked (violent vomiting and diarrhea, abdominal pain, thirst) and followed by cardiovascular collapse and/or acute renal failure (3). (A gram is approximately the weight of a cigarette.)
Surprisingly, the Food and Drug Administration lists tartaric acid in the Generally Recognized As Safe or GRAS category (9), which means this product, can be freely used as an additive in processed foods. Unless a food additive is put on the GRAS list, the food company using the product may have to spend thousands or even millions of dollars to prove its safety. Therefore, the political pressure to get a product on this GRAS list is intense. Tartaric acid is a byproduct of the wine industry and a tremendous amount of tartaric acid sludge has to be removed from the wine after the grape juice yeast fermentation. This sludge is the primary source of tartaric acid used as a food additive.
Tartaric acid is available as a food additive in baking powder, grape and lime flavored beverages, and poultry. It may also be found in grapes and grape products. Cream of tartar, which may be used for baking, is nearly pure tartaric acid. It is used in the food industry as a firming agent, flavor enhancer, flavoring agent, humectant, acidity control agent, and sequestant (9). There is no evidence that any mammals can produce it, so it is most likely, purely a yeast by-product. Tartaric acid may only be formed in the absence of oxygen and it is an analog of the Krebs cycle compound malic acid (Figure 4). An analog is a chemical compound that closely resembles but is not identical to another chemical compound. The atoms that differ in the two molecules are shaded in gray. The reason an analog is important is that the analog may prevent the normal biochemical from completing its normal biochemical function.
I would use this analogy to explain the analogs. You live in a neighborhood in which the same builder used the same locksmith who put a lock in each house that is just a little different. There have been a few burglaries in your neighborhood recently, so when you go to visit your neighbor next door, you decide to lock your door before going to your neighbor’s house. When you arrive at your neighbor’s house, your neighbor hands you a cup of coffee and you put your key on the kitchen counter right next to your neighbor’s key. You drink the cup of coffee, chat for a while and when you decide to go home, you unknowingly reach down and pick up your neighbor’s key. Then you take your neighbor’s key, which looks almost exactly like yours, go back to your house and put it into the lock. It goes in but when you start to turn it, nothing happens.
On a molecular level, the same kind of thing happens. Probably in some of the cases, the analog or false copy of the molecule breaks off and is stuck in the biological keyhole, which may be the critical part of an enzyme or cell receptor. These analogs then prevent the biochemical functioning from occurring. In some cases, the key eventually comes out and the right one is able to perform its biochemical function, however, your metabolism has experienced some degree of delay and lacks efficiency. This lack of efficiency can have a big impact if a high percentage of your metabolic processes are being affected simultaneously. Organs like the brain with a high rate of metabolism may be affected more than other organs. Think of how your TV set runs during a brownout when the supply of electricity is too low. If your metabolic processes are not efficient and are not producing sufficient energy, the brain may not process information efficiently.
Let’s return to tartaric acid and its specific role as an analog. Tartaric acid inhibits the enzyme fumarase (10), which is important in the function of the Krebs cycle, the biochemical process that produces most of the body’s energy. In addition, the inhibition of fumarase also decreases the supply of malic acid for other functions of the cell. The proper function of the Krebs cycle depends on a continuing supply of malic acid. If malic acid is not provided in sufficient quantities, the Krebs cycle is short-circuited.
A large percentage of patients with the disorder fibromyalgia, who have high amounts of tartaric acid in the urine, respond favorably to treatment with malic acid (11-13). I presume that supplements of malic acid are able to overcome the toxic effects of tartaric acid by increasing deficient malic acid. Fifty percent of the patients with fibromyalgia, who also have elevated yeast metabolites, also suffer from hypoglycemia (low blood sugar) even though their diet may have adequate or even excessive sugar (14). The reason for this may be due to the inhibition of the Krebs cycle by tartaric acid. The Krebs cycle is the main provider of raw materials such as malic acid that can be converted to blood sugar (Figure 5) when the body uses up its supply. (The technical name for this process is gluconeogenesis or “new formation of glucose”.) If sufficient malic acid cannot be produced, the body cannot produce the sugar glucose which is the main fuel for the brain. Therefore, the person with hypoglycemia feels weak and their thinking is foggy because there is insufficient fuel for their brain. If adults with elevated values of tartaric acid in the urine have foggy thinking, little energy, and are so depressed that they may seek out Dr. Kevorkian, imagine what a similarly affected young child, who has yet to form a clear concept of the world, must feel like.
Citramalic acid, like tartaric acid, is another analog of the normal compound malic acid. Citramalic acid is exactly the same (Figure 4) as malic acid except it has an extra CH3 group (called a methyl group) on it. Presumably, citramalic acid acts like tartaric acid in inhibiting the production of malic acid. There are two different types of citramalic acid called isomers. Both types of citramalic acid are probably in the urine of children with autism (2).
Figure 6 shows the chemical structure of a compound called arabinose, which is a sugar. (Arabinose is not an organic acid but is a chemical that we detect with our test.) This is not the same kind of sugar as kitchen table sugar but it is chemically very closely related. Like all sugars, it is sweet, which is what makes it a sugar. I found that, in the two brothers with autism, some of the values were much higher than in normal children.
In a study that was reported in the journal Science (a magazine in which experts report their findings to one another in highly technical language), Kiehn (15) reported information about a very closely related sugar called arabitol. Normal individuals have very low values of arabitol in the blood serum, but as people got sicker (or colonized) with the yeast, the values of arabitol increased. As the colonization worsened to a state called invasive candidiasis, the arabitol values could get extremely high: over a 1000 times the values found in the normal or control individuals. Many other papers have confirmed that high levels of this compound in both humans and animals were associated with Candida overload (16-18).
Women with vulvovaginitis due to Candida were found to have elevated arabinose in the urine (20) and restriction of dietary sugar brought about a dramatic reduction in the incidence and severity of the vulvovaginitis. Thus, antifungal drug therapy for children with autism could be useful to reduce the concentration of a yeast produced abnormal carbohydrate that cannot be tolerated by the child with defective pentose metabolism. Arabinose tolerance tests should be able to rapidly determine if such biochemical defects are present in children with autism.
Elevated protein-bound arabinose has been found in the serum proteins of schizophrenics (21) and in children with conduct disorders (22) and arabinose’s ability to alter protein function might be another mechanism by which arabinose might affect biochemical processes in autism and other diseases.
Arabinose may be found in some other foods in small quantities but the most significant source of dietary arabinose appears to be apples and pears. Arabinose values may be very elevated after drinking apple or pear juice or products such as applesauce or pear sauce (Figure 8). Therefore, apple and pear products should be restricted for a couple of days prior to testing. Several parents have reported severe worsening of autistic symptoms within a short time after their children ate apples. It is likely that the arabinose from apple products is responsible for this reaction.
Arabinose may also be formed from the breakdown of the sugar glucose (23) and antioxidants such as glutathione may inhibit this conversion (24). The breakdown of glucose also results in the formation of an aldehyde called glyoxal, which can also react with and modify protein structure and function. Glyoxal may be converted in the body to glycolic acid, glyoxylic acid, or oxalic acid. (Figure 8)
The aldehyde group of arabinose can react with the extra amino chemical group (called an epsilon amino group) of an amino acid called lysine that is present in a wide variety of proteins. This combined arabinose-lysine molecule may then form cross-links with an amino acid called arginine in an adjoining protein (25), forming a compound called a pentosidine (Figures 9 A, B). The formation of a pentosidine may cross-link different proteins (Figure 10) and may alter both the biological structure and function of a wide variety of proteins (25). The effect on all of the body’s functions may be devastating.
Let’s use the LA freeway analogy again to understand. Suppose that on a very foggy day during rush hour, gremlins hiding under your moving car and those of your neighbors took strong steel bars and welded them to the frames of the moving cars. The steel bars stick out at a perpendicular angle for about three feet from the side of your car without you or any other driver noticing, because of the fog. Arabinose would be like the steel bar and the proteins would be like the cars. Next, the gremlins welded the other end of the steel bars to the frames of neighboring cars and they welded a bar between that neighboring car and a third car and so forth. Furthermore, some cars might be welded to each other by their bumpers in addition to their sides. Now imagine what will happen when one or more of the drivers wanted to exit or change lanes. Chaos and carnage would ensue. The combined molecule of arabinose, lysine, and arginine is called a pentosidine and is like the two cars welded together. The undoing of these cross-links will become a major challenge in the future for treating older individuals with autism in which many of these cross-links have already been established.
I suspect that autism was reversed in the children of Pam Scott and Karyn Seroussi because they started therapy at a very young age. However, there have been many reports of improvements in people with autism in their twenties after antifungal therapy. Antifungal therapy cannot undo any of the existing cross-links, but can only prevent the formation of new cross-links by reducing the production of yeast arabinose. The tissue concentration of this combined molecule is almost linearly related to age (25); the increase in crosslinks (steel bars) in this molecule is one of the main reasons we lose flexibility as we age.
One child with autism with a very high urine arabinose (1144 mmol/mol creatinine) was examined by MRI (a type of brain scan) and found to have diffuse demyelination (loss of myelin) of the white matter of the brain. (Values as high as 4000 mmol/mol creatinine have been found in children with autism who have not been eating apple products.) It is possible that pentosidine formation could account for this demyelination. Myelin is the material that covers the axons of the brain in much the same way that plastic insulating material is wrapped around copper electrical wire. Without an intact myelin cover, the nerve impulses in the brain are short-circuited just like an electrical wire with torn insulation. Most children with autism are not examined by their physicians with MRI but, on a research basis such an examination of children with high urine arabinose values might be helpful to prove a link between high arabinose and demyelination. A summary of the possible adverse effects of pentosidine is given in Table 1.
The epsilon amino group of lysine is a critical functional group of many enzymes to which the vitamins pyridoxal (vitamin B-6), biotin, and lipoic acid are covalently bonded during coenzymatic reactions (26); the blockage of these active lysine sites by pentosidine formation may cause functional vitamin deficiencies (Figure 11) even when nutritional intake is adequate. In addition, the epsilon amino groups of lysine may also be important in the active catalytic site of many enzymes.
Protein modification caused by pentosidine formation is associated with crosslink formation, decreased protein solubility, and increased protease resistance. The characteristic pathological structures called neurofibrillary tangles associated with Alzheimer disease contain modifications typical of pentosidine formation. Specifically, antibodies against pentosidine react strongly to neurofibrillary tangles and senile plaques in brain tissue from patients with Alzheimer disease (27). In contrast, little or no reaction is observed in apparently healthy neurons of the same brain. Thus, it appears that the neurofibrillary tangles of Alzheimer’s disease may be caused by the pentosidines. The modification of protein structure and function caused by arabinose could account for the biochemical and insolubility properties of the lesions of Alzheimer disease through the formation of protein crosslinks. Similar damage to the brains of children with autism might also be due to the pentosidines and neurofibrillary tangles have also been reported in the brain tissue of an individual with autism (28). It has been reported that frequent urinary tract infections are associated with more severe Alzheimer’s disease (29). The use of antibiotics to treat urinary tract infections would of course lead to yeast overgrowth. I have found that urine arabinose is elevated in some cases of Alzheimer’s disease and have received a report of a favorable response from antifungal therapy to treat Alzheimer’s disease from a woman with a child with autism and a father with Alzheimer’s disease.
Glutathione has been reported to inhibit pentosidine formation (24). Supplementation with the vitamins biotin, pyridoxal (B-6), and lipoic acid (whose function at protein epsilon amino groups may be blocked by pentosidines derived from arabinose) might also be beneficial. Addition of vitamin B-6 derivatives or vitamin C to proteins helps to prevent pentosidine formation (30). In fact, I suspect that the beneficial effects of vitamin B-6 in autism reported in multiple studies (31) may be mediated by prevention of pentosidine formation. Pamela Scott used high amounts of vitamin B-6, for her child who recovered from autism prior to starting antifungal therapy. I suspect that this reduced somewhat the effects of the yeast die-off reaction. One way to test this idea would be to do a formal study to see if vitamin B-6 supplementation was less effective in treating autistic symptoms after antifungal therapy compared to supplementation before antifungal therapy.
Other compounds called furans that are occasionally elevated in the urine of children with autism are probably derived from fungus such as Aspergillus (32-34) rather than yeast although it is possible they may be produced by yeast as well. The names of these compounds are called 5-hydroxymethyl-2-furoic acid, furan-2, 5-carboxylic acid, and furancarbonylglycine. The concentration of furan compounds in the urine also dropped markedly in children with elevated values after nystatin therapy, indicating to me a probable yeast and/or fungal origin of these compounds. Other investigators (35, 36) noted that these compounds increased after sugar consumption and assumed that these compounds were sugar products of human metabolism but neglected to take into account the Japanese work and the role of gastrointestinal microorganisms in modification of sugars in the food. My interpretation is that these compounds may be derived from sugar but that they are converted to these furan products by the metabolism of yeast and/or fungi in the intestinal tract.