Autism is a behaviourally defined condition, with no definitive biomarkers, characterised by impairments in social interactions, verbal and non-verbal communications, and by restricted and repetitive stereotypical behaviours. Autism spectrum disorders are considered to be aetiologically and phenotypically heterogeneous. Until the mid–late 1980 s autism was a relatively rare condition with a prevalence of 5–10 per 10 000 of the population. However, over the last 15 years autism rates have increased such that it is now the most prevalent serious childhood developmental disorder in the developed world with a rate in the UK of 1 in 100(Reference Baird, Simonoff and Pickles1, Reference Harrison, O'Hare and Campbell2). However, there is considerable debate regarding the reasons for and the extent of the increase, with some authors claiming that changes are due largely to widened diagnostic criteria while others claim that the increased prevalence is a reality and perhaps underestimated(Reference Baron-Cohen, Scott and Alison3, Reference Blaxhill, Baskin and Spitzer4).
Over the past 10 years, a number of behavioural and learning disorders have been associated with suboptimal cellular fatty acid status including attention deficit hyperactivity disorder (ADHD), dyslexia, dyspraxia and autism(Reference Richardson and Ross5). Similarly, a number of neurological and neurodegenerative disorders including schizophrenia, unipolar and bipolar depression and postnatal depression have been linked to reduced erythrocyte and plasma fatty acid concentrations that may be related to changes in lipid metabolism(Reference Young and Conquer6).
The principal highly unsaturated fatty acids (HUFA) in neural tissues are DHA of the n-3 series (22 : 6n-3) and arachidonic acid (20 : 4n-6; ARA) of the n-6 series. The brain is about 60 % lipid, largely phospholipid (PL), and of this lipid the majority is made up of DHA and ARA that together represent more than 20 % of the brain dry weight and more than 30 % of retinal weight(Reference Horrobin, Bennett, Peet, Glen and Horrobin7, Reference Uauy and Dangour8). The essential HUFA, EPA (20 : 5n-3) is not generally enriched in neural tissues but has significant anti-inflammatory activity due to its ability to compete with ARA, and attenuate the production of ARA-derived eicosanoids in tissues and cell membranes(Reference Uauy and Dangour8, Reference Ruxton, Calder and Reed9). Classically, the essential fatty acids in humans are regarded as the C18 fatty acids linoleic (18 : 2n-6) and α-linolenic (18 : 3n-3) acids, although these require further metabolism by a series of desaturation and elongation reactions to form their functional HUFA, namely, ARA, EPA and DHA(Reference Sprecher10). However, while these pathways exist in humans the conversion of 18 : 3n-3 to DHA is very poor, with between 0·05 and 4 % being converted, which suggests that requirements for HUFA will not be met without significant exogenous supply(Reference Pawolsky, Hibbeln and Novotny11). Interestingly, this conversion is higher in women than men, particularly during pregnancy when conversion efficiency can increase to 9 %(Reference Burdge, Jones and Wootton12, Reference Burdge and Wootton13). This fact may be relevant to the sex ratio seen in autism where the ratio of prevalence is more than 4:1 in favour of males(Reference Constantino and Todd14).
As described above, the fatty acid composition of the brain and neural tissues is unique with respect to their high HUFA concentrations. Recently, many functions for these HUFA have been elucidated including regulation of signal transduction(Reference Kim15), neuro-inflammation(Reference Orr and Bazinet16) and cellular repair and survival(Reference Bazan17). The cell membrane PL are enriched in HUFA, especially in the sn-2 position, and these membranes are subject to constant breakdown and re-synthesis to maintain membrane integrity. The replacement HUFA for PL synthesis are obtained from the circulating plasma and may be directly from dietary sources or from the actions of synthetic pathways in vivo. Recent studies have found polymorphisms in the gene cluster associated with the fatty acid desaturase-2 gene (FADS2) for Δ6-desaturase in patients with ADHD(Reference Brookes, Chen and Xu18, Reference Scaeffer, Gohlke and Muller19). As Δ6-desaturase is the rate-limiting step in HUFA synthesis, reduced expression could reduce HUFA availability and a reduction in cellular concentrations.
A number of papers have reported changes in erythrocyte and plasma fatty acids in autism patients. Vancassel et al. (Reference Vancassel, Durand and Barthelemy20) observed significantly reduced n-3 PUFA and an increased n-6:n-3 ratio in plasma PL from autism patients compared with those who they designated as intellectually impaired patients. Similar changes were seen in erythrocyte PL from children with classical or early-onset and regressive autism compared with adult controls, although in addition there were decreased levels of total dimethyl acetals (DMA) and increased ARA:EPA ratios, 24 : 1 and 22 : 5n-6 in the autism patients compared with controls(Reference Bell, MacKinlay and Dick21). Stearic acid (18 : 0), total SFA and 18 : 2n-6 were significantly increased and 18 : 1n-9, total MUFA and ARA were significantly reduced in regressive autism patients compared with controls(Reference Bell, MacKinlay and Dick21). In addition, erythrocyte type IVA phospholipidase A2 concentrations were significantly higher in both autism groups compared with adult controls. In a recent study in California, erythrocyte PL fatty acid compositions were measured in twenty age-matched patients with regressive autism, early-onset autism, twenty patients with non-autistic developmental disabilities and twenty typically developing (TD) controls(Reference Bu, Ashwood and Harvey22). Their main findings were increased levels of 20 : 1n-9 and 22 : 1n-9 in regressive autism subjects compared with TD controls. In addition, 20 : 2n-6 was increased and trans-16 : 1n-7 was decreased in patients with regressive autism compared with early-onset autism patients(Reference Bu, Ashwood and Harvey22).
In the present study, it was planned to collect blood samples from three groups of children, who had respectively a diagnosis of autism or were developmentally delayed (DD) or were considered TD. Erythrocyte and plasma PL fatty acid compositions were measured in all three groups as well as in a smaller number of patients with autism who had consumed fish oils in the 6 months before sampling. The primary hypothesis tested in the present study was that children with autism have a lipid metabolic disorder that can be characterised by changes in circulating fatty acids, and in particular, but not exclusively, by reduced concentrations of essential HUFA (ARA, EPA and DHA) in erythrocyte and plasma PL. In the autism group receiving fish oil we hoped to establish whether such supplementation would result in increased membrane levels of the n-3 HUFA and reduced n-6 HUFA.
Materials and methods
Study participants
A total of 330 children were approached as part of the present case–control study. Of these, 198 had a diagnosis of autism, forty-five had DD and eighty-seven were classed as TD. Of these, forty-nine were recruited for inclusion in the study from the autism group, thirty-nine in the DD group and fifty-two in the TD group. Reasons for rejection included: taking fish oil supplements (thirty-five, two and fourteen subjects in the autism, DD and TD groups, respectively); taking steroids (five, one and seven subjects, respectively); and other reasons (fourteen, three and five subjects, respectively). The autism study participants were either existing patients of the Royal Hospital for Sick Children (RHSC), Edinburgh or were recruited from the Lothian National Health Service (NHS) catchment area. The subjects with autism met the criteria for autistic disorders by the Autism Diagnostic Interview – Revised (ADI-R)(Reference Lord, Rutter and Le Couteur23), which is a comprehensive investigative-based interview that covers most developmental and behavioural aspects of autism. The ADI-R has a diagnostic algorithm that maps to both Diagnostic and Statistical Manual for Mental Disorders, 4th ed. (DSM-IV) and International Classification of Diseases 10th Revision (ICD-10) criteria for the diagnosis of autistic disorder. In addition, the diagnosis of autism had been made by either a Consultant Paediatrician or Child and Adolescent Psychiatrist according to evidence-based guidelines(24) which included multidisciplinary specialist assessment, the employment of a classification system(25, 26) and the use of a published diagnostic instrument(Reference Harrison, O'Hare and Campbell27).
The first control group was of TD children who were matched on chronological age and who were visiting outpatient clinics for investigative procedures or undergoing elective surgery. They donated part of the blood sample being drawn for routine investigations. These controls measured within the normal ranges on the Strengths and Difficulties Questionnaire(Reference Goodman, Ford and Simmons28). The second control group comprised children with learning disabilities and/or generalised DD who were matched to the children with autistic disorder on the Vineland Adaptive Behaviour Scales(Reference Sparrow and Cicchetti29). Not all cases or controls were able to be matched. However, thirty-eight pairs of autism cases were age- and sex-matched with DD controls where the age matching in this case was done using developmental age as assessed by Vineland and forty-five pairs of autism cases were age- and sex-matched to TD controls using chronological age. In addition to the original research aims, and due to the high number of children with autism who were taking fish oil supplements, it was decided to analyse samples from twenty-one subjects with autism who had taken fish oil and compare them with the forty-nine autism subjects described above. The present study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human patients were approved by the Lothian Research Ethics Committee (LREC reference 04/S1103/51). Written consent was obtained from all parents. The sampled population included forty-nine children with autism, aged 7·5 (sd 3·5) years (89·6 months), including five females, thirty-nine children with DD, aged 6·0 (sd 3·3) years (72·2 months), including four females and fifty-two TD children, aged 7·5 (sd 3·6) years (89·9 months), including three females.
Blood collection
Whole blood (4 ml) was collected by phlebotomists at the RHSC Edinburgh, into tubes containing EDTA as anticoagulant (Teklab Medical Laboratories, Durham, UK). The blood was centrifuged (1000 g; 5 min.) and the plasma removed and retained. In addition, the buffy coat and the top 2 mm of erythrocytes were removed using a Pasteur pipette and placed in a separate vial. The remaining packed erythrocytes were split into two samples and stored at − 70°C at the biochemistry laboratories at the RHSC. Samples were frozen at − 70°C in a freezer within 60 min of collecting the blood sample. Each sample was assigned a numbered code that was known only to project staff at the RHSC so that samples were analysed blind by staff at Stirling for polar lipid fatty acid compositions of erythrocyte and plasma. Samples were collected from the RHSC every 2 months and delivered to Stirling on dry ice where they were stored at − 70°C until analysed.
Lipid extraction, preparation of phospholipid fraction and preparation of fatty acid methyl esters in erythrocytes and plasma
The lipid extraction was a modification of the method of Bligh & Dyer(Reference Bligh and Dyer30). The erythrocyte pellet was thawed and 0·25–0·50 ml placed in a 15 ml stoppered tube. Then 5 ml methanol–0·01 % butylated hydroxytoluene (BHT, w/v) was added and then mixed on a vortex mixer. After 20 min at room temperature 2·5 ml chloroform–BHT was added, mixed and left at room temperature for a further 40 min. The tubes were centrifuged at 500 g for 5 min and the supernatant fraction decanted into a clean 15 ml tube after filtering though Whatman no. 1 paper. A further 2·5 ml chloroform–BHT and 2·5 ml of 0·88 % KCl (w/v) were added, mixed and centrifuged as above. The lower chloroform layer was removed and filtered into a 15 ml tube and evaporated under N2. The lipid was taken up in 0·8 ml chloroform–methanol (2:1, v/v) and placed in a weighed glass vial. This was dried under N2 and desiccated for 16 h. After recording the lipid weight the sample was re-suspended in chloroform–methanol (2:1, v/v)+BHT, at a concentration of 10 mg/ml and stored at − 70°C until analysed.
Plasma samples (0·5 ml) were extracted by the method of Folch et al. (Reference Folch, Lees and Sloane Stanley31) in 10 ml chloroform–methanol (2:1, v/v), washed with 0·88 % KCl (w/v) and filtered and dried as above. Final lipid extracts were re-suspended in chloroform–methanol (2:1, v/v)+BHT, at a concentration of 10 mg/ml and stored at − 70°C until analysed.
A PL fraction was prepared from 0·5 mg of total lipid applied to a 20 × 20 cm silica gel 60 TLC plate (VWR, Lutterworth, Leics, UK) and developed in iso-hexane–diethyl ether–acetic acid (80:20:1, by vol.) and dried for a few minutes at room temperature. The plate was sprayed lightly with 2, 7, dichlorofluorescein (0·1 %, w/v) in 97 % methanol (v/v) and the PL bands on the origin scraped from the plate and placed in a 15 ml test-tube. Fatty acid methyl esters (FAME) were prepared by acid-catalysed transesterification in 2 ml of 1 % H2SO4 in methanol at 50°C overnight(Reference Christie32). The samples were neutralised with 2·5 ml of 2 % KHCO3 and extracted with 5 ml isohexane–diethyl ether (1:1, v/v)+BHT. The samples were then re-extracted with 5 ml isohexane–diethyl ether (1:1, v/v) and the combined extracts dried and dissolved in 0·3 ml isohexane before fatty acid analysis.
Measurement of erythrocytes and plasma fatty acids
FAME were separated and quantified by GLC (Fisons 8160; Carlo Erba, Milan, Italy) using a 60 m × 0·32 mm × 0·25 μm film thickness capillary column (ZB-WAX; Phenomenex, Macclesfield, Cheshire, UK). H2 was used as the carrier gas at a flow rate of 4·0 ml/min and the temperature programme was from 50 to 150°C at 40°C/min then to 195°C at 2°C/min and finally to 215°C at 0·5°C/min. Individual FAME were identified compared with well-characterised in-house standards as well as commercial FAME mixtures (Supelco™ 37 FAME mix; Sigma-Aldrich Ltd, Gillingham, Dorset, UK).
Statistical analysis
To determine if there were any differences between fatty acid compositions in autism cases and age- and sex-matched TD controls, or DD controls, paired t tests were used. To compare the fatty acid compositions of the autism group (n 49) with autism patients who were excluded from the main trial due to consumption of fish oils (n 21) a two-sample t test was used. The level of significance was set initially at P < 0·05.
Results
In total, 330 children were identified from the Lothian National Health Service (NHS) trust catchment as being potential participants in the present study. Of these, 140 children agreed or were suitable for participation. The reasons for non-suitability were varied and included consumption of fish oil supplements in the previous 6 months and taking steroids. The percentage of participants already consuming fish oils represented 18 % of the autism group, 16 % of the TD group and 4 % of the DD group.
When comparing the erythrocyte polar lipid fatty acid data of the autism and TD groups (Table 1) only a few fatty acids or ratios were different (P < 0·05). α-Linolenic acid (18 : 3n-3), the precursor for the longer-chain n-3 HUFA, was reduced in the autism group compared with the TD group while the ARA:EPA ratio was increased. The plasma polar lipid fatty acid compositions in the pair-matched autism and TD groups showed very few differences (Table 1). The minor SFA 14 : 0 and 15 : 0 were reduced while the n-6:n-3 and ARA:EPA ratios were increased in the autism group compared with the TD group.
Table 1 Fatty acid compositions of erythrocyte and plasma polar lipids (weight % of total fatty acids) in autism and typically developing (TD) subject groups (n 45, pair-matched)
(Mean values and standard deviations)
* P < 0·05.
When comparing the erythrocyte polar lipid fatty acid data between pair-matched children from the autism and DD groups (Table 2) the SFA 20 : 0, 22 : 0 and 24 : 0 as well as the MUFA 16 : 1n-9, 20 : 1n-9, 22 : 1n-9 and 24 : 1n-9 were all increased in the autism group compared with the DD group. In addition, the 18 : 1n-7 DMA fatty acid was increased in the autism group compared with the DD group. By comparison, the total n-6 fatty acids and the total PUFA were reduced in the autism group compared with the DD group.
Table 2 Fatty acid compositions of erythrocyte and plasma polar lipids (weight % of total fatty acids) in autism and developmental delay (DD) subject groups (n 38, pair-matched)
(Mean values and standard deviations)
LOQ, limit of quantification; DMA, dimethyl acetals.
* P < 0·05.
Relatively minor changes were also seen in the plasma polar lipid fatty acid compositions when comparing pair-matched autism and DD subjects (Table 2). The SFA 20 : 0 and total saturates were reduced in the autism compared with the DD group while 18 : 1n-9 DMA and total DMA were increased in the autism group compared with the DD group.
The erythrocyte polar lipid fatty acid compositions in autism patients compared, using two-sample t tests, with autism patients who had consumed fish oil supplements showed a number of differences between treatment groups (Table 3). Of the twenty-one autism patients who had consumed fish oils, fifteen had taken Eye Q™ (Equazer,Wallingford, Oxfordshire, UK), one had taken Haliborange-3 (Seven Seas Ltd, Hull, UK) and one had taken Eskimo-3® (Whaley Bridge, Derbyshire, UK) while four did not specify the fish oil product. Among the MUFA, 18 : 1n-7 was decreased in the autism compared with the autism+fish oil group. In the n-6 fatty acids, ARA, 22 : 4n-6 and 22 : 5n-6 were reduced in the autism+fish oil group compared with the autism group. The reductions in n-6 HUFA were reversed in the n-3 HUFA which had increased EPA, 22 : 5n-3, DHA and total n-3 HUFA in the autism+fish oil group compared with the autism group. In addition, the n-6:n-3 ratio and the ARA:EPA ratio were both reduced in the autism+fish oil group compared with the autism group (Table 3).
Table 3 Fatty acid compositions of erythrocyte and plasma polar lipids (weight % of total fatty acids) in patients with autism who met the agreed participation criteria compared with those excluded due to consumption of fish oil supplements in the preceding 6 months†
(Mean values and standard deviations)
* P < 0·05.
† Two-sample t test comparing autism (n 49) with autism patients who were taking fish oil supplements (n 21).
The plasma polar lipid fatty acid compositions in autism patients compared, using two-sample t tests, with autism patients who had consumed fish oil supplements showed a number of differences similar to those seen in the erythrocyte data described above (Table 3). The MUFA 18 : 1n-7 was increased, as seen in erythrocytes, but 18 : 1n-9 and 20 : 1n-9 were decreased and increased, respectively, in the autism+fish oil group. ARA was not different in plasma but 22 : 4n-6, 22 : 5n-6 and total n-6 fatty acids were all reduced in the autism+fish oil group while 18 : 3n-3, 20 : 4n-3, EPA, 22 : 5n-3, DHA and total n-3 fatty acids were all increased in the autism+fish oil group (Table 3). As in erythrocytes, the n-6:n-3 and ARA:EPA ratios were both decreased in the autism+fish oil group compared with the autism group.
Discussion
The results of the present study do not suggest that the autism population in general have a problem with abnormal PL metabolism although there are subtle changes in specific fatty acids and ratios that might indicate problems with some individuals in the study groups. While we initially set the level of significance at P < 0·05 this would not take account of multiple comparisons and, thus, if a stricter level of significance set at P < 0·0025 were applied then most of the values presented in Tables 1–3 would not be significant. Correction for multiple comparisons prevents accidental significance being assigned when combining and comparing data from unrelated sources, for example, data from different tissue or cell types. However, when considering membrane fatty acid analysis from one cell type or plasma fraction, a level of significance set at P < 0·0025 is probably not necessary. This is supported by studies in the literature involving fatty acid datasets from similar pair-matched studies(Reference Bu, Ashwood and Harvey22, Reference Aupperle, Denney and Lynch33, Reference Clayton, Hanstock and Hirneth34).
The most notable difference between the erythrocyte fatty acids from the autism and TD study groups was an increased ARA:EPA ratio in the former. A similar increase in this ratio was also seen in plasma PL fatty acids in the autism group compared with the TD group. This increased ratio was also seen in an earlier study where children with autism were compared with adolescent and adult controls(Reference Bell, MacKinlay and Dick21) but no difference in the ARA:EPA ratio was seen in the study by Bu et al. (Reference Bu, Ashwood and Harvey22). In the latter, the mean erythrocyte ARA:EPA ratio varied from 66 to 72 which is much higher than the values of 23 and 17 seen in the autism and TD groups in the present study. This may be due to higher n-6 PUFA intake, relative to n-3 PUFA, in the US diet or perhaps differences in methodology used in the two studies. By comparison, the ARA:EPA ratio in two different groups of UK adults, conducted in our own laboratories in 2006–7, was 15·8 and 16·7(Reference Cyhlarova, Bell and Dick35) (JG Bell, EE MacKinlay and JR Dick, unpublished results).
ARA and EPA are competitive substrates for the production of eicosanoids, including, inter alia, PG, thromboxanes and leukotrienes (LT) that are involved in the generation and modulation of inflammatory responses(Reference Calder36). Compared with ARA, EPA is less favoured for eicosanoid synthesis by the cyclo-oxygenase and lipoxygenase enzymes that generate PG and LT, respectively, but EPA can inhibit production of the more bioactive PG and LT at a number of steps in their synthetic pathways(Reference Galli, Marangoni and Petroni37, Reference Jump38). Thus, the dietary balance of n-6 and n-3 fatty acid intake can affect the concentration and bioactivity of these cellular mediators. The tissue ARA:EPA ratio is regarded as an inflammatory index since ARA produces eicosanoids that stimulate aggregation and inflammatory responses via 2-series PG and 4-series LT while the 3-series PG/5-series LT, derived from EPA, have opposing activities(Reference Allessandri, Guesnet and Vancassel39).
Comparing the erythrocyte PL fatty acids in autism and DD groups, the latter had reduced concentrations of 20 : 0, 22 : 0, 24 : 0, 20 : 1n-9, 22 : 1n-9, 24 : 1n-9 and 18 : 1n-7 DMA and increased concentrations of total n-6 fatty acids and total PUFA, compared with the autism group. The long-chain fatty acids 24 : 0 and 24 : 1n-9 are important components of myelinated nerve axons and synapses in brain white matter where they can comprise up to 48 and 16 % of fatty acids in glyco- and sphingolipids, respectively(Reference Ledeen, Yu and Eng40). Thus, these changes may reflect suboptimal myelin development amongst the DD children. The reductions in 20 : 0, 22 : 0, 20 : 1n-9 and 22 : 1n-9 might indicate that these fatty acids are being utilised to synthesise 24 : 0 and 24 : 1n-9 in the DD group. The study by Bu et al. (Reference Bu, Ashwood and Harvey22) observed an increase in 20 : 1n-9 and 22 : 1n-9 in patients with regressive autism compared with TD controls. These MUFA are minor components of membrane lipids but are found at higher levels in neutral lipids where they are substrates for energy production by β-oxidation. It was suggested(22) that this elevation of MUFA might reflect an impairment in this energy production pathway as seen in patients with mitochondrial dysfunction which has been recorded in psychiatric disorders, including autism(Reference Clark-Taylor and Clark-Taylor41, Reference Shao, Martin and Watson42). However, a similar elevation was not observed in the present study.
The final part of the present study compared treatment-naive autism patients with autism patients who had taken fish oil supplements over the previous 6 months. Of the n-6 fatty acids, ARA, 22 : 4n-6, 22 : 5n-6 and total n-6 PUFA were reduced in fish oil-supplemented patients compared with the unsupplemented group. The reduction in ARA and its elongation–desaturation products in the fish oil-supplemented group is due to competition between EPA and ARA for the sn-2 position on PL, displacement of ARA by EPA and reduced synthesis of C22 HUFA. This is seen in most studies with EPA-rich supplements(Reference Stevens, Zhang and Peck43, Reference Panchaud, Sauty and Kernan44). In the erythrocyte n-3 fatty acids, fish oil intake increased EPA, 22 : 5n-3, DHA and total n-3 fatty acids and reduced n-6:n-3 and ARA:EPA ratios. As with the reductions in n-6 HUFA, these changes in n-3 fatty acids are similar to those recorded in other studies using fish oil supplementation(Reference Stevens, Zhang and Peck43, Reference Panchaud, Sauty and Kernan44, Reference Katan, Deslypere and van Birgelen45).
The changes in plasma fatty acids in autism patients who took fish oil were similar to those in erythrocytes, except that ARA was not reduced. In ADHD patients given fish oil, plasma ARA remained unchanged or increased following supplementation(Reference Katan, Deslypere and van Birgelen45, Reference Stevens, Zental and Deck46). Of the n-3 fatty acids, the same fatty acids were increased and decreased as in erythrocytes but 18 : 3n-3, 20 : 4n-3 and total PUFA were also increased, reflecting the abundance of these fatty acids in fish oil. These changes are consistent with other studies where patients have taken fish oil supplements including schizophrenia, ADHD, cystic fibrosis and autism(Reference Bell, MacKinlay and Dick21, Reference Stevens, Zhang and Peck43, Reference Katan, Deslypere and van Birgelen45, Reference Puri, Richardson, Peet, Glen and Horrobin47).
The relatively minor changes observed in erythrocytes and plasma fatty acids in children with autism compared with TD children does not support a general ‘phospholipid spectrum disorder’ in the autism population. Similarly, there is no evidence, based on the results of the present study, for providing all autism patients with fish oil supplements. However, a number of noteworthy points arose from the present study. First, there was a relative reduction in n-3 fatty acids and/or a relative increase in n-6 fatty acids in the autism population compared with TD children, as evidenced by the increased ARA:EPA ratio in the former. The ranges for this ratio in the autism and TD groups, 11·2–100·2 and 11·8–46·8, respectively, suggests there are children in the autism group with very high ratios that may indicate problems in addition to poor dietary intake. We were not able to obtain a prospective dietary history from the children in the study because ethically we had to obtain an opportunistic blood sample from the children in the control group. If we had chosen to gather this information from the children in the autism group in advance of the blood sampling, then we may have introduced confounders by drawing the families' attention to the HUFA content of their child's diet. However, recent work by Herndon et al. (Reference Herndon, Di Guiseppi and Johnson48) suggests that children with autism differ from controls in their lower intake of dairy products and Ca and so it would appear that whilst we cannot discount a possible contribution of diet, this is unlikely to fully explain the differences seen. It was also apparent that the n-3 HUFA in all the groups studied were significantly lower than values reported in UK adults(Reference Bell, MacKinlay and Dick21, Reference Cyhlarova, Bell and Dick35). There is also evidence that n-3 HUFA are significantly higher in neonates and infants(Reference Pontes, Torres and Trugo49, Reference Smithers, Gibson and McPhee50) compared with the children in the present study. However, we do not know whether levels in children would be expected to be lower than those in infancy and adult life but the fact that they rose to similar levels following supplementation suggests that the levels are suboptimal. In addition, autism patients given fish oil had erythrocyte DHA levels which were similar to both UK adult and neonate values. Therefore, when we compare these findings with animal studies, and n-3 HUFA in neural development in young humans(Reference Uauy and Dangour8, Reference Ruxton, Calder and Reed9, Reference Allessandri, Guesnet and Vancassel39), the low levels seen in Scottish children in the present study are concerning. Whilst the evidence for efficacy of HUFA supplementation in conditions such as autism and ADHD still needs to be fully delineated(Reference Bent, Bertoglio and Hendren51), we have demonstrated that it is effective at a biochemical level in promoting what is generally considered to be a more optimal ratio(Reference Clayton, Hanstock and Hirneth34, Reference Katan, Deslypere and van Birgelen45, Reference Stevens, Zental and Deck46, Reference Moriguchi, Greiner and Salem52, Reference Richardson and Puri53). Future studies should consider these areas as well as using molecular techniques to establish whether specific problems with HUFA retention and metabolism exist in a cohort of children with autism.
Acknowledgements
The present study was supported by funding from the Chief Scientist Office of the Scottish government (project CZB-4-256).
Thanks are also due to Dr Livar Frøyland and Dr Pedro Araujo of the Norwegian Institute for Nutrition and Seafood Research for their advice on statistical analysis.
J. G. B. was the principal investigator and wrote the manuscript with contributions from all other authors. D. M. was the research nurse responsible for patient recruitment, collection and processing of blood samples and administration of clinical aspects of the study. D. J. M. and R. M. B. were responsible for analysis of phospholipase. E. E. M. and J. R. D. were responsible for all aspects of fatty acid analyses. S. C. was responsible for recruitment and assessment of all DD patients. A. E. O. was the joint principal investigator responsible for all clinical aspects of the study. C. G. was responsible for most of the statistical analyses of data derived from the project. All authors read and approved the findings of the study.
None of the authors had a conflict of interest.
