RESUMO
BACKGROUND: The rationale of the current study was to test the clinical utility of the folate pathway genetic polymorphisms in predicting the risk for autism spectrum disorders (ASD) and to address the inconsistencies in the association of MTHFR C677T and hyperhomocysteinemia with ASD. PATIENTS AND METHODS: An artificial neural network (ANN) model was developed from the data of 138 autistic and 138 nonautistic children using GCPII C1561T, SHMT1 C1420T, MTHFR C677T, MTR A2756G, and MTRR A66G as the predictors of autism risk. A neuro fuzzy model was developed to explore the genetic determinants of homocysteine. Meta-analyses were carried out on 1361 ASD children and 6591 nonautistic children to explore the association of MTHFR C677T and homocysteine with the risk for ASD. RESULTS: The ANN model showed 63.8% accuracy in predicting the risk of autism. Hyperhomocysteinemia was observed in autistic children (9.67±4.82 vs. 6.99±3.21 µmol/l). The neuro fuzzy model showed synergistic interactions between MTHFR C677T and MTRR A66G inflating homocysteine levels. The meta-analysis showed MTHFR to be a genetic risk factor for autism in both fixed-effects (odds ratio: 1.47, 95% confidence interval: 1.31-1.65) and random-effects (odds ratio: 1.57, 95% confidence interval: 1.16-2.11) models. The meta-analysis of nine studies showed hyperhomocysteinemia as a significant risk factor for autism in both fixed-effects (P<0.0001) and random-effects (P=0.026) models. CONCLUSION: Genetic polymorphisms of the folate pathway were moderate predictors of autism risk. MTHFR C677T and hyperhomocysteinemia have been identified as risk factors for autism worldwide. Synergistic interactions between MTHFR C677T and MTRR A66G increase homocysteine.