Skip to main content
. 2021 Mar 24;20(5):2410–2419. doi: 10.1021/acs.jproteome.0c00918

Table 2. ROC Curve Parameters of the Combined Multimetabolite Biomarker Model for Detecting Subjects with Obesity and IS and of the Individual Metabolites That Are Part of This Modela.

detection of OB IS sensitivity (%) specificity (%) AUC (95% CI)
Only in Obese Population
combined multimetabolite model 88.0 100 96.7 (91.9–100)
GPE 18:2 76.0 90.9 89.1 (78.8–99.4)
GPI 18:0 76.0 81.8 77.5 (56.7–98.2)
GPS 18:0 84.0 72.7 78.8 (61.1–96.8)
GPC aa 18:0/18:1 (GPC aa 36:1) 88.0 63.6 79.3 (61.0–97.6)
GPC aa 18:2/18:2 (GPC aa 36:4) 72.0 72.7 69.5 (48.7–90.2)
GPC aa 18:2/18:3 (GPC aa 36:5) 72.0 72.7 71.6 (50.0–93.3)
In the Whole Study Population
combined multimetabolite model 90.9 76.7 87.6 (79.0–96.2)
GPE 18:2 63.6 66.7 71.8 (56.5–87.2)
GPI 18:0 81.8 66.7 70.2 (50.3–90.0)
GPS 18:0 72.7 81.7 77.9 (61.9–93.9)
GPC aa 18:0/18:1 (GPC aa 36:1) 63.6 66.7 61.4 (43.3–79.5)
GPC aa 18:2/18:2 (GPC aa 36:4) 72.7 55.0 53.0 (35.1–70.9)
GPC aa 18:2/18:3 (GPC aa 36:5) 72.7 53.3 60.3 (39.1–81.5)
a

Their predictive power was evaluated in the whole obese subset and in the whole study population including nonobese individuals. Metabolites are sorted alphabetically. Abbreviations: aa, diacyl; AUC, area under the curve; CI, confidence interval; GPC, glycerophosphatidylcholine; GPE, glycerophosphatidylethanolamine; GPI, glycerophosphatidylinositol; GPS, glycerophosphatidylserine.