Nutritional and Metabolic Status of Children With Autism vs. Neurotypical Children, and the Association With Autism Severity

James B Adams; Tapan Audhya; Sharon McDonough-Means; Robert A Rubin; David Quig; Elizabeth Geis; Eva Gehn; Melissa Loresto; Jessica Mitchell; Sharon Atwood; Suzanne Barnhouse; Wondra Lee

Disclosures

Nutr Metab. 2011;8(41) 

In This Article

Results

Correlation of Autism Severity Scales

As shown in Table 4, the PDD-BI, ATEC, and SAS scales were strongly correlated with one another, R = 0.75–0.81, similar to the findings of a previous study.[37]

Comparison of Neurotypical and Autism Groups With Published Reference Ranges

Reference ranges for the neurotypical children in this study were calculated based on the 10th and 90th percentiles of their distribution. This is more exact than using +/- two standard deviations if the data is not normally distributed, which sometimes was the case. These calculated reference ranges were compared with published reference ranges for vitamins (Table 5), minerals (Table 6), primary amino acids, and secondary amino acids. Two primary sources were used for vitamins and minerals: 1) the National Health and Nutrition Examination Survey (NHANES) National Report on Biochemical Indicators of Diet and Nutrition in the US Population 1999–2002,[38] and 2) the Tietz Textbook of Clinical Chemistry;[39] both are generally viewed as highly credible sources for the US population. In some cases only adult reference ranges are available from those sources. Despite the differences in techniques and methodologies, the agreement with the NHANES reference ranges is very good, and the agreement with the Tietz reference ranges is reasonable, especially when comparing to pediatric values. The agreement with published reference ranges is a validation of our methodology and of our calculated reference range for neurotypical children, which we will compare with the autism group in the next section. The advantage of having our own reference range for neurotypical children is that it closely matches the age, gender, and geographical area (Arizona) of our autism group.

The amino acid reference ranges (10th and 90th percentiles) for the neurotypical group (present study) were compared with pediatric reference ranges by Lepage et al[33] and (where available) with values from the Tietz Textbook of Clinical Chemistry.[39] For Lepage et al,[33] two reference ranges are listed, one for 6 year olds and one for 16 year olds. In most cases the values from the present study are in reasonable agreement with the published values.

For the autism group, the average (mean) levels of their vitamins, minerals, and most amino acids were within the published reference ranges (where available). However, as will be discussed below, a t-test comparison of the levels of vitamins, minerals, amino acids, and other biomarkers in the autism group and the neurotypical group revealed many significant differences.

Comparisons of Biomarkers Between Autistic and Neurotypical Groups

Vitamins Table 5 shows the participants' levels of vitamins, vitamin-like substances, and biomarkers of vitamin status. Because we are making multiple comparisons (our hypothesis is "are the levels of any vitamins different in children with autism vs. controls"), we need to apply a Bonferroni correction (see statistical analysis section). For 21 comparisons, p values are defined as: "significant" = p < 0.002, "marginally significant" = p < 0.005, and "possibly significant" = p < 0.05. Figure 1 compares the levels of vitamin-related biomarkers that were different in the autism group compared to the control group.

Figure 1.

Vitamins and related substances which were significantly different between the autism and neurotypical groups, rescaled to the average neurotypical values. The average values and the standard deviations are shown. The number of asterisks indicates the p-value (* p < 0.05, ** p < 0.01, *** p < 0.001).

For the vitamins, the only significant difference was a 20% lower biotin (p < 0.001) in the children with autism. There were possibly significant (p < 0.05) lower levels of vitamin B5, vitamin E, and total carotenoids. Vitamin C was possibly slightly higher in the children with autism. Vitamin B6 (measured as the active form, P5P, in the RBC) had an unusually broad distribution in children with autism compared to controls (see Figure 1), with the levels in the children with autism having 3 times the standard deviation of the neurotypical children.

The levels of two vitamin-like substances, lipoic acid and choline (free and total) were also assessed. Levels of lipoic acid and free choline were similar in the two groups, but total choline was 17% higher in the autistic group (p < 0.0001).

The functional need for vitamins was indirectly assessed by measurements of several urinary metabolites, including FIGLU, kryptopyroles, methylmalonic acid, and n-methyl-nicotinamide. FIGLU and n-methyl-nicotinamide were somewhat higher in children with autism (possibly significant, p < 0.05), suggesting an increased need for folic acid and niacin, respectively. The average levels of urinary kryptopyroles were not significantly different in children with autism, but the children with autism had a much broader distribution.

For most vitamins, children with autism have levels that lie within the neurotypical reference ranges defined by the 10th and 90th percentiles (see Table 5). However, there are some cases where more than 25% of the autism group lie below the neurotypical reference range (total carotenes) or above the neurotypical reference range (vitamin C, free choline, total choline, FIGLU).

Essential Minerals Table 6 shows the levels of minerals in whole blood (WB), red blood cell (RBC), serum, and urine (for iodine) for the study participants. (28 comparisons: "significant" is p < 0.002, "marginally significant" is p < 0.004, and "possibly significant" is p < 0.05). Figure 2 shows the levels of minerals which were different between the autism and neurotypical groups.

Figure 2.

Minerals which were significantly different between the autism and neurotypical groups, rescaled to the average neurotypical values. The average values and the standard deviations are shown. The number of asterisks indicates the p-value (* p < 0.05, ** p < 0.01, *** p < 0.001).

The largest difference was a much lower level of WB lithium (-53%, p < 0.006). Note that three of the controls had unusually high levels of WB lithium, and were from the same family, so the data is analyzed with and without their data; removing their results reduces the magnitude of the difference, but the significance of the result remains the same.

Iron status was measured in three ways: serum ferritin, serum iron, and RBC iron. The first two did not reveal any difference between the two groups, but RBC iron was slightly higher in the children with autism (+7%, p < 0.0005), with 42% of the children with autism having levels above the 90th percentile for the typical children.

There were small, possibly significant differences in several other minerals. There were possibly significant slightly higher levels of RBC potassium, RBC phosphorus, copper (WB and RBC), and RBC boron, and a possibly significant lower level of RBC calcium and magnesium (serum and WB).

For most minerals, children with autism have levels that generally lie within the neurotypical reference ranges (see Table 6). However, there are some cases where more than 25% of the autism group lie below the neurotypical reference range (urinary iodine, RBC calcium) or above the neurotypical reference range (RBC iron, RBC phosphorus, RBC boron).

We also investigated the correlations of levels of minerals measured in different blood compartments, as shown in Table 7. In some cases the levels correlate strongly, but in some they do not; in the latter case measurements for those elements need to be interpreted cautiously as different compartments will give different results. For magnesium, copper, zinc, manganese, and selenium there are strong, very significant correlations of levels between the WB and RBC, and a modest correlation for molybdenum. For calcium and magnesium, there is a significant correlation of levels in WB and serum. For calcium, there is a small negative correlation of RBC and serum levels. For potassium and phosphorus, correlations between RBC and serum are generally not significant, except possibly for a weak correlation for phosphorus for the neurotypical group. For iron, there are no significant correlations between levels in RBC iron, serum iron, and serum ferritin. In summary, interpretation of results for some elements (magnesium, copper, zinc, manganese, and selenium) is consistent across blood compartments, but for some elements it is not (calcium, potassium, phosphorus, iron).

Sulfation, Methylation, Glutathione, Oxidative Stress Table 8 shows the results for sulfation, methylation, glutathione, and oxidative stress markers. (11 comparisons, so "significant" is p < 0.005, "marginally significant" is p < 0.01, and "possibly significant" is p < 0.05). Figure 3 shows the results which were different between the autism and neurotypical groups.

Figure 3.

Sulfation, methylation, glutathione, and oxidative stress biomarkers which were significantly different between the autism and neurotypical groups, rescaled to the average neurotypical values. The average values and the standard deviations are shown. The number of asterisks indicates the p-value (* p < 0.05, ** p < 0.01, *** p < 0.001).

Free and total sulfate in plasma were very significantly lower in children with autism (-28% and -65%, respectively, p < 0.0001).

S-adenosylmethionine (SAM, the primary methyl donor in the body) was also very significantly lower in children with autism vs. controls. Although the percentage difference is not high, the normal reference range is very narrow, so this difference is very significant. The level of SAH was not significantly different, but it had an unusually broad distribution, with 27% of the children with autism having levels below the 10th percentile of the neurotypical group, and 24% had levels above the 90th percentile. The SAM/SAH ratio was 10% lower in children with autism (p = 0.006).

Uridine (in plasma) was very significantly higher in the children with autism (+93%, p < 0.0001). Uridine is believed to be a marker of methylation status, and in fact SAM and uridine were somewhat negatively correlated (R = -0.30).

Adenosine was slightly higher (marginally significant) in children with autism, which may indicate that some children have an impairment in adenosine deaminase.

Reduced plasma glutathione (GSH) was very significantly lower in the children with ASD. GSH is an important anti-oxidant and important for excretion of toxic metals.

All three markers of oxidative stress, namely oxidized glutathione (GSSG), the ratio of oxidized to reduced glutathione (GSSG:GSH), and plasma nitrotyrosine, were very significantly higher in children with autism.

For sulfation, 36–56% of the autism group have sulfate levels below the neurotypical reference range (see Table 8). For SAM, SAH, and SAM/SAH, 25–39% of the autism group have low levels, and 60% have elevated uridine, another marker of methylation status. Adenosine was elevated in 33% of the autism group. For reduced glutathione, oxidized glutathione, the ratio of GSH:GSSG, and nitrotyrosine, 30–53% of the autism group have abnormal values.

ATP, NADH, NADPH, CoQ10 Table 9 shows the results for ATP, NADP, NADPH, and CoQ10. (4 comparisons: "significant" is p < 0.012, "marginally significant" is p < 0.025, and "possibly significant" is p < 0.05). Figure 4 shows the results which were different between the autism and neurotypical groups.

Figure 4.

ATP, NADH, and NAHPH were significantly different between the autism and neurotypical groups. The average values and the standard deviations are shown, rescaled to the average neurotypical value. The number of asterisks indicates the p-value (* p < 0.05, ** p < 0.01, *** p < 0.001).

The primary function of mitochondria is to produce ATP, the primary energy source in the brain and in the body. CoQ10 is an important co-factor for mitochondrial function. We found that children with ASD have levels of plasma CoQ10 that are very similar to the neurotypical group. Levels of CoQ10 did not significantly correlate with levels of ATP or with autism severity. The autism group had much lower levels of plasma ATP and of NADH (RBC) and NADPH (RBC), which are the precursors to ATP, and 36–51% of the autism group had levels below the neurotypical reference range. The level of ATP, NADH, and NADPH were all highly correlated with one another (r = 0.67–0.69, p < 0.001).

Plasma Amino Acids: Primary The levels of primary (proteinogenic) plasma amino acids are given in Table 10. Note that these are free, not total, amino acids in plasma. (20 comparisons: "significant" is p < 0.0025, "marginally significant" is p < 0.005, and "possibly significant" is p < 0.05). Figure 5 shows the results which were different between the autism and neurotypical groups.

Figure 5.

Amino Acids which were significantly different between the autism and neurotypical groups, rescaled to the average neurotypical value. The average values and the standard deviations are shown. The number of asterisks indicates the p-value (* p < 0.05, ** p < 0.01, *** p < 0.001). The standard deviations for beta-amino-isobutyrate and "homocystine + homocysteine" are outside the margins of the figure.

A few samples had abnormally low ratios of glutamine/glutamate (< 4) and also asparagine/aspartate (< 5). These amino acids are especially sensitive to shipping conditions, and abnormalities in both ratios suggest that some thermal degradation occurred during shipping/processing, resulting in conversion of some glutamine to glutamate, and asparagine to aspartate. This was the case for five autism samples and three control samples. The results for those samples were not included in the analysis.

The autism group had significantly lower levels of tryptophan, a precursor to serotonin, (-19%, p < 0.001) and higher levels of glutamate, an excitatory neurotransmitter (+18%, p < 0.001). There were smaller changes in other amino acids that were possibly significant (p < 0.05), including slightly increased serine, and slightly decreased tyrosine and phenylalanine.

For most primary amino acids, children with autism had levels that generally lay within the neurotypical reference ranges (see Table 10). However, there are some cases where more than 25% of the autism group lie below the neurotypical reference range (asparagine, tyrosine) or above the neurotypical reference range (histidine, glutamate).

Plasma Amino Acids: Secondary The levels of secondary plasma amino acids and amino acid metabolites are given in Table 11. Cystathionine was also measured, but all the measurements except 1 were below the detection limit of 0.05 umoles/100 ml, so those values are not listed. (21 comparisons: "significant" is p < 0.002, "marginally significant" is p < 0.005, and "possibly significant" is p < 0.05). Figure 5 shows the results which were different between the autism and neurotypical groups.

The autism group had a higher level of beta-amino isobutyrate (+39%, p = 0.004, marginally significant). They had a possibly significant lower level of taurine. They also had a possibly significant much higher level of "homocystine + homocysteine"- however, it should be noted that 71% of the autism group and 89% of the neurotypical group had levels below the detectable limit, so the "homocystine + homocysteine" results should be interpreted with caution.

For the secondary amino acids, children with autism had levels that generally lay within the neurotypical reference ranges (see Table 11). However, there were some cases where more than 25% of the autism group lay below the neurotypical reference range (citrulline, phosphoethanolamine) or above the neurotypical reference range (phosphoserine).

Medication Effects

45% of the children with autism were taking one or more medications (see Table 1). Some of those medications may have affected their levels of vitamins, minerals, or other biomarkers. Some psychopharmaceuticals (such as risperidone and clonidine used by several of this ASD population) and anti-convulsants (valproic acid and topiramate used by two of the ASD subjects) are known to interfere with nutrient levels and gastrointestinal function in a variety of ways. Anticonvulsants may interfere in energy production[40] and folate metabolism and for some in levels of vitamin D. Valproate, a known teratogen[41,42] and hepatotoxin[42] as studied in humans and animals increases GABA in the brain and is a folate antagonist with absorption not affected. In animal studies[43] valproate has been shown to increase oxidative stress as methionine and vitamin E mitigate teratogenic effects. Valproate inhibits histone deacetylases, increasing accessibility of DNA to demethylases resulting in altered gene expression. In humans valproate may increase plasma ammonia, homocysteine and glutamine and decrease carnitine.[42] Topiramate may create metabolic acidosis, decrease glutamate and increase GABA. The effect upon specific nutrients has not been as well studied for the psycho-pharmaceuticals.[44]

To investigate if medication use had a significant effect on results, a t-test comparison was made between the autism group taking medications (45%) vs. the autism group not taking any medications (55%). The only differences with a p-value less then 0.01 were lower RBC copper (-9% lower, p = 0.001) and higher plasma methionine sulfoxide (+35% higher, p = 0.002) for the autism medication group compared to the autism no-medication group. So, aside from those two differences, it appears that medication use had little effect on the results.

Correlations With Autism Severity

The correlations of each biomarker with each of the three autism severity scales were calculated. Table 12 lists the biomarkers which had the highest correlation with autism severity (r > 0.34 in absolute magnitude, corresponding to a p value of 0.01 or lower). The biomarkers had p < 0.01 for only one autism severity scale at most. Given multiple biomarkers, the cut-off for significance is below p = 0.001. So, none of the results are significant, but some have p < 0.01 and are worth further investigation.

Regression Analyses

Regression analysis develops an equation that relates one or more "independent" variables (such as metabolic biomarkers) to a single "dependent" variable (such as severity of autism). The regression equation has coefficients that minimize the differences between observed values of the dependent variable and those predicted by the equation. The standard measure of how well a regression performs is R2, which is the proportion of the variation in the dependent variable that can be explained by the regression. (If R2 = 1, the regression equation fits the dependent variable perfectly; if R2 = 0 the independent variables provide no useful information about the dependent variable).

Vitamins The regression analysis yielded a significant result for all three autism severity scales (adj. R2 of 0.25–0.57), with the highest adjusted R2 for the PDD-BI. Vitamin B6, Vitamin C, N-methyl-nicotinamide, and Vitamin K were the most consistently significant variables.

Minerals The regression analysis yielded a significant result for all three scales (adj. R2 of 0.22–0.38), with the highest adjusted R2 for the PDD-BI. Calcium (RBC), Iron (RBC), Zinc (WB and RBC), and Potassium (RBC) were the most consistently significant variables. Note that almost all of the most consistently significant variables were in RBC; ie, it is the RBC levels that seem to be most strongly associated with autism severity.

Sulfation, Methylation, Glutathione, Oxidative Stress The regression analysis yielded significant results for all three scales, with all three severity scales having modest adjusted R2 (0.15–0.24). Free Sulfate was the most consistently significant variable, followed by Oxidized Glutathione and SAM.

ATP, NADH, NADPH, CoQ10 The regression analysis yielded significant results for only one severity scale (the ATEC), with only a modest adjusted R2 (0.15). NADH and ATP were the two significant variables.

Primary Amino Acids The regression analysis yielded significant results for all three scales (adj. R2 of 0.22–0.39), with the PDD-BI having the highest adjusted R2. Proline and Serine were the most consistently significant variables.

Secondary Amino Acids The regression analysis yielded significant results for all three scales, with modest adjusted R2 (0.18–0.26). Ethanolamine and Beta-amino-isobutyrate were the most consistently significant variables.

Overall Analysis This analysis involved starting with all variables from the previous analyses that were significant in one or more of the subgroup analyses (p < 0.01), determined individually for each autism severity scale. The Overall regression analysis yielded highly significant results for all three scales (p < 0.002 or better for all cases), with the highest adjusted R2 for the PDD-BI, followed by the SAS and then the ATEC. Different markers were significant for different autism severity scales.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....