Comparison of Three Clinical Trial Treatments for Autism Spectrum Disorder Through Multivariate Analysis of Changes in Metabolic Profiles and Adaptive Behavior

Several studies associate autism spectrum disorder (ASD) pathophysiology with metabolic abnormalities related to DNA methylation and intracellular redox homeostasis. In this regard, three completed clinical trials are reexamined in this work: treatment with (i) methylcobalamin (MeCbl) in combination with low-dose folinic acid (LDFA), (ii) tetrahydrobiopterin, and (iii) high-dose folinic acid (HDFA) for counteracting abnormalities in the folate-dependent one-carbon metabolism (FOCM) and transsulfuration (TS) pathways and also for improving ASD-related symptoms and behaviors. Although effects of treatment on individual metabolites and behavioral measures have previously been investigated, this study is the first to consider the effect of interventions on a set of metabolites of the FOCM/TS pathways and to correlate FOCM/TS metabolic changes with behavioral improvements across several studies. To do so, this work uses data from one case–control study and the three clinical trials to develop multivariate models for considering these aspects of treatment. Fisher discriminant analysis (FDA) is first used to establish a model for distinguishing individuals with ASD from typically developing (TD) controls, which is subsequently evaluated on the three treatment data sets, along with one data set for a placebo, to characterize the shift of FOCM/TS metabolism toward that of the TD population. Treatment with MeCbl plus LDFA and, separately, treatment with tetrahydrobiopterin significantly shifted the metabolites toward the values of the control group. Contrary to this, treatment with HDFA had a lesser, though still noticeable, effect whilst the placebo group showed marginal, but not insignificant, variations in metabolites. A second analysis is then performed with non-linear kernel partial least squares (KPLS) regression to predict changes in adaptive behavior, quantified by the Vineland Adaptive Behavior Composite, from changes in FOCM/TS biochemical measurements provided by treatment. Incorporating the 74 samples receiving any treatment, including placebo, into the regression analysis yields an R(2) of 0.471 after cross-validation when using changes in six metabolic measurements as predictors. These results are suggestive of an ability to effectively improve pathway-wide FOCM/TS metabolic and behavioral abnormalities in ASD with clinical treatment.