Among the 274 patients with ASD considered as nonsyndromic, we were unable to detect any patients with mucopolysaccharidoses, abnormalities of purine and pyrimidine metabolism, creatine deficiency syndromes or aminoacidopathies. The only abnormalities found were a non-specific increase in urinary creatine excretion and a 3-methylglutaconic aciduria. These results suggest that the prevalence of screened IMD in association with ASD is low (<0.5%) and thus indicate the weak cost effectiveness of a systematic metabolic workup in ASD considered as nonsyndromic.
Very few IMDs may begin with isolated ASD as a prominent feature. For example, phenylketonuria (PKU) concerns the rare neonates born in developed countries who escaped from neonatal screening or the unscreened neonates born in countries for which PKU screening is not systematically performed. These can exhibit an isolated ASD over months or years that is usually secondarily associated with behavioral disorders, epilepsy and severe ID 
. Similarly, children with classical homocystinuria (HCY) due to cystathionine-β-synthase deficiency may exhibit isolated autism 
. The clinical picture may later be enriched by the cardinal manifestations of HCY i.e.
lens subluxation, vascular thrombosis and skeletal abnormalities 
. Isolated ASD may also be observed at the onset of mucopolysaccharidosis type III (MPS III) or Sanfilippo disease 
. MPS III patients often present with severe behavioral disorders and subtle signs of slowly progressing cognitive impairment, primary including speech regression and loss of toilet training and further leading to a severe encephalopathy 
. In addition, rare cases of urea cycle disorders, especially ornithine transcarbamylase deficiency have been reported in patients with nonsyndromic ASD 
. In this case, ASD are usually part of recurrent bouts of neuro-psychiatric problems (behavioral disorders, ataxic gait) frequently associated with hepato-digestive abnormalities 
. These rare IMDs for which ASD is an initial sign could justify a systematic minimal metabolic screening in ASD patients considered as nonsyndromic, especially because PKU, HCY and urea cycle disorders are treatable conditions for which early recognition and treatment are essential. Plasma amino acids (PKU, HCY, and urea cycle disorders), post prandial plasma ammonia level (urea cycle disorders) and urinary MPS profile (Sanfilippo disease) could be proposed. In these disorders, however, the clinical picture comprises with time additional signs such as severe behavioral disorders, cognitive regression, ocular, liver or other neurological signs. Careful reappraisal of the clinical signs is therefore warranted and crucial because sometimes the initial diagnosis of nonsyndromic ASD should be reconsidered in the presence of subtle symptoms. At the Robert Debré hospital, due to published yet controversial recommendations 
, a systematic metabolic workup is performed in all patients with ASD, whatever the associated symptoms. These metabolic tests are not exhaustive and concern various diseases with syndromic and nonsyndromic ASD. Moreover, a complete screening for all of the IMD is not feasible in a standard hospital setting 
. Targeted testing could more effectively identify IMD that require early treatment. Plasma and urine amino acid profiles allow the detection of PKU and HCY, while urinary organic acid profiles can identify all of the organic acidurias, some of which include autistic signs 
. In creatine deficiency syndromes, autistic features are mostly not isolated but associated with ID, speech delay and epilepsy 
. A few inborn errors of purine (especially adenylosuccinase deficiency) and pyrimidine (especially dihydropyrimidine dehydrogenase deficiency) metabolism 
may include autism in association with psychomotor retardation, epilepsy and muscular wasting in adenylosuccinase deficiency 
The two abnormalities identified in our cohort are remarkable. The patient with persistent creatine urinary excretion may have had a mild and so far uncharacterized creatine transporter defect. Moreover, on brain magnetic resonance spectroscopy imaging, the patient had a low but discernible creatine peak whereas in creatine transport defects, the creatine peak is usually markedly decreased 
. Normality of creatine uptake as well as SLC6A8
investigations were not in favor of a primary disorder of creatine transport. Low creatine peak on brain spectroscopy has been reported in other neurogenetic disorders especially in abnormalities of white matter homeostasis 
. Elevated creatine urinary excretion in the patient is most probably a false positive result that might be related to excessive creatine dietary intake (mainly due to high protein intake) as previously reported 
. Unfortunately, we were unable to call the patient back for diet control. Three-methylglutaconic aciduria is a non-specific abnormality underlying a highly heterogeneous and poorly defined group of diseases named the 3-methylglutaconic acidurias 
. Besides the 3-methylglutaconyl-CoA hydratase deficiency, an inborn error of leucine catabolism, the whole group of 3-methylglutagonic acidurias is most probably seen in the context of various mitochondrial dysfunctions 
. Further analyses in this direction are underway in this patient.
Notably, a recent study pointed out that mitochondrial dysfunction may be involved in ASD 
. In this report, plasma lactate determination was performed in a restricted sample of 69 patients with ASD. Twenty per cent of them displayed hyperlactatemia and 7% fulfilled the criteria for a disorder of oxidative phosphorylation (OXPHOS) 
. This initial study has limitations as the autistic clinical phenotype was not well defined. More recently, a retrospective study of 25 patients with a primary diagnosis of nonsyndromic autism who were further determined to have enzyme- or mutation- defined OXPHOS deficiency showed that 96% of these patients actually exhibited clinical symptoms differentiating them from idiopathic autism 
. These results suggest that careful clinical and biochemical reappraisal is warranted in patients with ASD initially considered as nonsyndromic, but also confirm that ASD patients with OXPHOS dysfunction often exhibit other symptoms such as microcephaly, marked motor delay, sensorineural deafness, oculomotor abnormalities, exercise intolerance, cardiomyopathy or renal tubular dysfunction. Accordingly, we decided not to screen for hyperlactatemia in our nonsyndromic ASD population. Furthermore, normal plasma lactic acid concentrations do not exclude the presence of a mitochondrial disorder 
. Recent reports emphasize a putative association between mitochondrial dysfunction and autism (for review see 
) and highlight the role of brain energy metabolism dysfunction as an important target for future studies 
. In ASD, as already emphasized for several neurodegenerative disorders 
, mitochondrial dysfunction could be regarded as a secondary defect in brain energy metabolism.
With the metabolic workup conducted in the sample of 274 patients diagnosed with nonsyndromic ASD, we were unable to detect any of the screened IMD in any of our patients but one, who exhibited persistent urinary excretion of 3-methylglutaconic acid. Our data showed negative results in 99.5% of cases, which did not differ from the estimated prevalence of IMD in the general population 
. Moreover, the causative relationship between the unspecific 3-methylglutaconic aciduria and ASD is unclear and remains to be proven. These results strongly call into question the utility of a systematic metabolic workup in these patients. Few studies have previously explored the prevalence of IMD associated with ASD. All of them have been performed on restricted cohorts of ASD patients. One study reported that none of the 53 patients with ASD, screened for ammonia, amino acids, lactic acid and pyruvic acid in blood and urinary organic acids, exhibited abnormal results 
. Also, Wang et al
. compared the urinary creatine and guanidinoacetate to creatinine ratios between 57 ASD patients and 49 unrelated control samples. They did not detect any significant difference between groups suggesting that creatine deficiency syndromes are not more frequent in patients with ASD than in controls 
Several limitations of this study should be underlined. First, we considered as nonsyndromic ASD patients with ID. We assumed that altered mechanisms in the central nervous system in ASD could involve a wide variety of clinical diagnoses crossing categorical boundaries, specifically ID 
. At the opposite, some authors consider that the co-occurring ASD in association with ID is the consequence of reduced compensatory capacity, and suggested that ASD patients with ID should be regarded as syndromic ASD 
. Second, the retrospective study design is a limiting factor. Finally, the failure to detect any IMD in our group of patients might be due to the limited sample size; much larger study groups would be warranted to detect extremely rare IMD.
Conclusions and Perspectives
The data reported here strongly support the view already stressed by others 
that systematic metabolic investigations are not contributive to the etiology of nonsyndromic ASD. On the other hand, early diagnosis and proper therapeutic intervention for some metabolic disorders causing nonsyndromic ASD may significantly improve the long-term cognitive and behavioral outcomes 
. Therefore, a careful clinical evaluation, with cautious reappraisal of clinical signs, is crucial. Such a medical practice appears more reasonable than a costly systematic workup. Finally, a large population based prospective study assessing the benefits of routine metabolic screening in nonsyndromic ASD would be of great interest in the future to confirm our results.