Skip to main content

Table 4 Results from the ROC curve analyses in EXPOsOMICS CVD

From: A blood DNA methylation biomarker for predicting short-term risk of cardiovascular events

Follow-up time

# Events

DNAmCVDscore

DNAmGrimAge

SCORE2

MRS

SCORE2 + DNAmCVDscore

AUC (95% CI)

Sensitivity; specificity

AUC (95% CI)

Sensitivity; specificity

AUC (95% CI)

Sensitivity; specificity

AUC (95% CI)

Sensitivity; specificity

AUC (95% CI)

Sensitivity; specificity

18 years

160

0.525 (0.461; 0.589)

0.477; 0.580

0.569 (0.505; 0.632)

0.647; 0.482

0.749 (0.696; 0.803)

0.719; 0.698

0.484 (0.420; 0.548)

0.523; 0.500

0.753 (0.700; 0.807)

0.732; 0.685

17 years

159

0.527 (0.463; 0.591)

0.520; 0.553

0.562 (0.498; 0.626)

0.669; 0.472

0.753 (0.700; 0.806)

0.753; 0.671

0.516 (0.452; 0.580)

0.545; 0.484

0.757 (0.704; 0.810)

0.740; 0.689

16 years

158

0.527 (0.463; 0.591)

0.513; 0.554

0.559 (0.496; 0.623)

0.583; 0.572

0.751 (0.698; 0.805)

0.718; 0.692

0.527 (0.463; 0.591)

0.449; 0.610

0.756 (0.704; 0.809)

0.705; 0.717

15 years

149

0.541 (0.477; 0.604)

0.518; 0.577

0.560 (0.496; 0.623)

0.681; 0.450

0.739 (0.684; 0.794)

0.741; 0.671

0.536 (0.472; 0.600)

0.509; 0.477

0.745 (0.691; 0.799)

0.687; 0.732

14 years

139

0.567 (0.504; 0.630)

0.483; 0.662

0.581 (0.518; 0.644)

0.494; 0.662

0.730 (0.674; 0.786)

0.705; 0.691

0.573 (0.509; 0.636)

0.682; 0.475

0.741 (0.686; 0.796)

0.688; 0.683

13 years

123

0.595 (0.531; 0.659)

0.625; 0.537

0.592 (0.528; 0.656)

0.526; 0.667

0.720 (0.663; 0.777)

0.662; 0.699

0.595 (0.531; 0.659)

0.531; 0.650

0.731 (0.676; 0.787)

0.651; 0.740

12 years

111

0.596 (0.530; 0.662)

0.529; 0.649

0.594 (0.529; 0.659)

0.510; 0.694

0.709 (0.650; 0.768)

0.588; 0.775

0.592 (0.525; 0.658)

0.588; 0.613

0.718 (0.661; 0.775)

0.598; 0.757

11 years

95

0.622 (0.554; 0.690)

0.536; 0.705

0.618 (0.549; 0.686)

0.568; 0.663

0.693 (0.630; 0.755)

0.582; 0.758

0.610 (0.540; 0.680)

0.536; 0.695

0.710 (0.651; 0.770)

0.600; 0.737

10 years

84

0.622 (0.551; 0.693)

0.580; 0.607

0.619 (0.548; 0.691)

0.576; 0.631

0.700 (0.636; 0.765)

0.550; 0.845

0.602 (0.530; 0.673)

0.541; 0.667

0.721 (0.661; 0.781)

0.554; 0.821

9 years

67

0.647 (0.571; 0.722)

0.601; 0.612

0.645 (0.569; 0.721)

0.649; 0.582

0.683 (0.611; 0.754)

0.609; 0.746

0.623 (0.546; 0.700)

0.581; 0.627

0.708 (0.642; 0.775)

0.577; 0.731

8 years

55

0.687 (0.611; 0.762)

0.569; 0.709

0.676 (0.601; 0.752)

0.627; 0.673

0.692 (0.618; 0.766)

0.623; 0.746

0.638 (0.554; 0.722)

0.635; 0.618

0.732 (0.664; 0.800)

0.612; 0.782

7 years

37

0.711 (0.626; 0.796)

0.615; 0.703

0.707 (0.625; 0.789)

0.554; 0.811

0.678 (0.587; 0.770)

0.622; 0.703

0.668 (0.566; 0.770)

0.669; 0.622

0.721 (0.641; 0.801)

0.741; 0.649

6 years

28

0.778 (0.702; 0.854)

0.673; 0.786

0.771 (0.688; 0.853)

0.634; 0.821

0.753 (0.668; 0.839)

0.686; 0.786

0.730 (0.616; 0.844)

0.669; 0.750

0.803 (0.740; 0.867)

0.676; 0.893

5 years

23

0.774 (0.695; 0.853)

0.623; 0.826

0.773 (0.694; 0.851)

0.634; 0.826

0.761 (0.665; 0.856)

0.723; 0.783

0.735 (0.618; 0.852)

0.688; 0.783

0.800 (0.741; 0.860)

0.757; 0.783

4 years

16

0.823 (0.746; 0.899)

0.799; 0.688

0.821 (0.736; 0.906)

0.632; 0.938

0.785 (0.653; 0.917)

0.722; 0.875

0.783 (0.649; 0.917)

0.689; 0.875

0.830 (0.755; 0.906)

0.773; 0.750

3 years

13

0.811 (0.720; 0.901)

0.642; 0.846

0.806 (0.724; 0.888)

0.652; 0.846

0.772 (0.620; 0.923)

0.772; 0.846

0.762 (0.628; 0.896)

0.768; 0.769

0.809 (0.726; 0.892)

0.705; 0.846

2 years

7

0.851 (0.767; 0.935)

0.799; 0.857

0.842 (0.745; 0.939)

0.737; 0.857

0.751 (0.575; 0.926)

0.656; 0.857

0.717 (0.562; 0.872)

0.779; 0.571

0.863 (0.766; 0.960)

0.766; 0.857

  1. For each composite biomarker, we report the AUC (95% CI), sensitivity, and specificity according to the best threshold (minimising the distance from the top left corner of the ROC curve) derived from logistic regression model adjusted for matching parameters (age, sex, and centre of recruitment). Predictive performance was evaluated at different time points, right-censoring the follow-up time in the range of 18 to two years, with one year interval