1. D-Dimer levels over time and the risk of
recurrent venous thromboembolism:
An update of the Vienna Prediction Model
Sabine Eichinger, Georg Heinze, Paul A. Kyrle
Dept. of Medicine I & Center for Medical Statistics
Medical University of Vienna, Austria

2. Background I
•
Patients with unprovoked venous thromboembolism (VTE) are at
increased risk of recurrence.

3. Recurrence risk after unprovoked VTE
Kyrle, Lancet 2010

4. Background II
•
Patients with unprovoked venous thromboembolism (VTE) are at
increased risk of recurrence.
•
By use of the „Vienna Prediction Model“ patients with unprovoked
VTE can be further stratified according to their recurrence risk.

5. Nomogram to predict recurrence
Points
Sex
Location
DDimer (µg/l)
Total Points
0
10
20
30
40
40
60
50
60
70
90
80
90
100
male
female
proximal DVT
distal DVT
100
0
pulmonary embolism
150
200
50
250
100
12 months cumulative recurrence rate
0.02
60 months cumulative recurrence rate
400
750
190
150
0.04
0.1
500
200
0.06
250
0.08
24%
0.2
1000
0.3
1500
300
2000
350
0.1 0.12 0.15
0.4
0.5
Eichinger, Circulation 2010

6. Study aim
• To expand the “Vienna Prediction Model” in order
to assess the recurrence risk also from later time
points on

7. Patients
Inclusion
Exclusion
• > 18 yrs
• VTE provoked by surgery,
• First VTE
trauma, pregnancy, female
• OAC > 3 mo
hormone use
• Objective Dx of VTE
• AT-, PC-, PS-deficiency
• Lupus anticoagulant
• Cancer
• Antithrombotics

8. D-Dimer measurements and follow-up
D-Dimer
OAC
stop
3 w
baseline
3
9
months
15
24

9. Statistical analysis
•
Preselected variables: sex, location of VTE, D-Dimer
•
Competing risk regression model to predict cumulative incidence of
recurrence using all variables
•
Estimated a series of models to predict cumulative recurrence from
various time points after baseline
• Each model uses most recent D-Dimer values
 “Dynamic Vienna Prediction Model”

10. Patient characteristics (n = 553)
Age (yrs); median (25th, 75th P)
53 (43, 62)
Women; n
219 (40%)
Location of first VTE; n
PE + proximal DVT
464 (84%)
distal DVT
BMI (kg/m2); median (25th, 75th P)
89 (16%)
27.2 (24.4, 30.0)
F V Leiden; n
126 (23%)
F II G20210A; n
27 (5%)
Duration of anticoagulation (mo); median (25th, 75th P)
6.7 (6.2, 8.5)
Observation time (mo), median (25th, 75th P)
68 (46, 98)

11. D-dimer levels over time after anticoagulation
Time
Patients, n
D-Dimer (µg/L)
median (25th, 75th percentile)
Baseline
553
338 (226, 551)
3 months
534
339 (227, 551)
9 months
494
356 (239, 557)
15 months
457
363 (237, 572)
24 months
415
375 (245, 610)

12. Subdistribution hazard ratios (SHR) for recurrent VTE
Variable
Time point
Male vs. female sex
Baseline
2.4 (1.6, 3.8)
3 months
2.3 (1.5, 3.5)
9 months
2.0 (1.3, 3.0)
15 months
1.7 (1.1, 2.7)
Baseline
1.8 (1.0, 3.4)
3 months
1.7 (0.9, 3.1)
9 months
1.5 (0.8, 2.8)
15 months
1.4 (0.8, 2.7)
Baseline
1.3 (1.1, 1.6)
3 months
1.3 (1.1, 1.5)
9 months
1.2 (1.0, 1.4)
15 months
1.1 (0.9, 1.4)
Proximal DVT or PE
vs. distal DVT
D-Dimer (per doubling)
SHR (95% CI)

13. Nomogram to predict recurrence from
3 months
0
27
50
14%

14. Nomogram to predict recurrence from
9 months

15. Cross-validation

17. Summary
•
In patients with a first unprovoked VTE D-Dimer levels do not
substantially change over time after anticoagulation.
•
The effect of risk factors on the recurrence risk may change over
time (e.g., effect of male sex and location of first VTE weakened).
•
By integrating patient’s sex, location of VTE and serial D-Dimer
measurements the recurrence risk after anticoagulation can be
assessed not only after 3 weeks but also from later time points on.

18. Conclusion
•
The “ Dynamic Vienna Prediction Model” allows predicting the
recurrence risk from various later time points after VTE which
provides greater flexibility in counseling patients regarding their
individual recurrence risk and optimal anticoagulation.