Fig. 4

Correlations of clinical biomarkers and metadata with metabolites and lipoprotein alterations. (A) A negative correlation was found between blood alanine and CSF alpha-synuclein, (B) the Glyc/SPC ratio parameter shows a significant positive correlation with BMI index, (C) lipoprotein characteristic HDL-4 apolipoprotein Apo-A2 had a negative correlation with human total tau levels, (D) DMGly and citrate metabolites have positive correlations with LEDD, (E) blood 3-HB levels show a nonsignificant tendency to correlate to MoCA score (within non-PD Controls and PD recessive groups of individuals’ samples only), (F) blood ApoB100/ApoA1 ratio correlate positively with BDI-II, (G) area-under-curve analysis (receiver operating curve building) demonstrates a good statistical differentiation of Sporadic PD early and Sporadic PD late patient groups (n total = 139) for the LEDD available entries only, (H) area-under-curve analysis plot with a logarithmically normed box plot of blood citrate indicating of Sporadic PD early and Sporadic PD late patient groups (n total = 139) for the significant discriminatory characteristics tested separately based on the LEDD available entries only patient samples, (I) tyrosine shows a positive correlation with LEDD for PD sporadic (early, late; n = 139), (J) pyruvate has a positive correlation with LEDD for PD genetic groups (GBA, recessive, LRRK2, and double mutation carriers; n = 92). The software used places LEDD in the number one position, with a correlation of value plus one with itself (segments I and J). The rates shown on a (“classical”) Receiver Operating Characteristic (ROC) curve, often so, are also plotting the true positive rate (the sensitivity) against the false positive rate (1-specifcity) for different threshold settings of a diagnostic test, as performed in MetaboAnalyst²². The red separator line is representing an “optimal cutoff” that is according to located point furthest from the diagonal and according to the Youden’s method²³.