Fig. 1: Neurocognitive outcomes in PA are associated with biomarkers linked to the mitochondrial dysfunction and the severity of PCC deficiency.

Biomarkers shown here as natural log-transformed values were linked via regression to the underlying severity of PA and mitochondrial dysfunction and showed similar associations with full scale IQ and Vineland Adaptive Behavior Composite scores. Red and blue lines illustrate an unadjusted linear regression for the outcomes (Y-axis) full scale IQ (blue) and Vineland Adaptive Behavior Composite (red). The log-transformed predictor variable for each panel is named in the header. The population-average standard score range (85–115) is shaded in gray. Parameter estimates corresponding to these lines are shown in Table 3 and test statistics in Supplementary Table S3. The distributions of untransformed continuous biomarkers are shown in Supplemental Figure S2. A Plasma propionylcarnitine (ln-transformed). B Plasma total 2-methylcitrate (ln-transformed). C In vivo whole body 1-¹³C-propionate oxidation (ln-transformed). D Plasma FGF21 (ln-transformed). E Plasma GDF15 (ln-transformed). F Plasma glycine (ln-transformed). G Plasma glutamine (ln-transformed). H Urinary concentration of glutamine normalized by creatinine (ln-transformed). I Serum erythropoietin (ln-transformed). J Affected gene, PCCA vs PCCB. K Comparison of neurocognitive outcomes in participants with two (biallelic) loss-of-function PCCA or PCCB alleles vs all other genotypes (e.g., two missense alleles or one missense plus one nonsense allele).