Upload
others
View
20
Download
0
Embed Size (px)
Citation preview
Novel Screening Tool
for Abdominal Aortic Aneurysms
based on Plasma D-Dimer and Myeloperoxidase
Wolf Eilenberg, Medical University Vienna
Wolf Eilenberg
Department of Surgery, Division of Vascular SurgeryMedical University of ViennaGeneral Hospital, Vienna, Austria
SPECIAL RESEARCH PROGRAMME SFB-54: “INTHRO”Cellular Mediators Linking Inflammation and Thrombosis
Disclosure
Wolf Eilenberg, Medical University Vienna 2
Speaker name: Wolf Eilenberg
I have the following potential conflicts of interest to report:
NONE
Neutrophil depletion inhibits AAA formation in mouse models• Eliason et al. Circulation 2005
Neutrophil-released proteases and ROS contribute to media destruction• Houard et al. FASEB J 2009
Neutrophil myeloperoxidase (MPO) is elevated in blood of AAA patients• Vega de Ceniga et al. Ann Vasc Surg 2014
Neutrophil involvement in AAA
3Wolf Eilenberg, Medical University Vienna
Diagnostic marker potential of MPO for AAA compared to the established marker D-dimer
4Wolf Eilenberg, Medical University Vienna
p<0.001p<0.001
D-DimerAUC=0.830p<0.001
1 – Specificity
Sen
siti
vity
MPOAUC=0.829p<0.001
1 – Specificity
Sen
siti
vity
Exploration set:Plasma measurements of MPO and D-dimer in AAA patients (N=41) andhealthy controls (N=38) matched for age, sex, BMI and smoker status
Development of a diagnostic scorebased on combined D-Dimer and MPO values
5Wolf Eilenberg, Medical University Vienna
Plot of cross-validated predicted probabilities vs. true outcome status:
Calibration plot (observed event rates vs. predicted probabilities in quartiles):
- Logistic regression model by stepwise backward elimination of clinical and laboratory parameters
- Only D-dimer and MPO show significant independent importance for AAA diagnosis - D-dimer: odds ratio 3.96 and 95% confidence interval 1.65-12.81 - MPO: odds ratio 1.23 and 95% confidence interval 1.08-1.43
Plot of cross-validatedpredicted probabilitiesversus true outcome status
ROC curve:
The area under the ROC curve and the discrimination slopes was 0.85 and 0.415 respectively.
Model 3: model with marker and demographic variables
## ORs 2.5 % 97.5 % pvals ## (Intercept) 0.000 0.000 0.014 0.001 ## Smoker2 19.711 1.608 492.095 0.040 ## DDimer 3.130 1.269 10.755 0.034 ## Monos_Sys 1.001 0.999 1.003 0.120 ## CD16monos_per 1.087 0.985 1.219 0.115 ## MPO 1.318 1.132 1.614 0.002
AUC=0.834Coefficient ofdiscrimination(R2)=0.404
Diagnostic linear score
derived from the logistic model
according to the estimated regression coefficients
SCORE = -3.442 + 1.375 * D-dimer + 0.205 * MPO
Validation of the diagnostic scorebased on combined D-Dimer and MPO values
6Wolf Eilenberg, Medical University Vienna
AUC=0.934Coefficient of discrimination=0.489
Validation set:
Plasma samples from AAA patients (N=30) [median aneurysm diameter 54.6 mm (IQR=10.3)]
and unmatched healthy controls (N=29)
DICH- positive predictive value of 95.7% (78.1%-99.9%) - negative predicted value of 77.8% (60.9%-89.9%)
SCORE Sensitivity: 73.3% (54.1%-87.7%)Specificity: 96.6% (82.2%-99.9%)
SCORE predicts probability as PROB=1/(1+exp(-SCORE)
A simplified classifier: dichotomized SCORE (DICH)SCORE < 0 ... DICH=0SCORE >= 0 .... DICH=1
DICH * Disease Crosstabulation
Disease
Totalhealthy AAA
DICH 0 Count 28 8 36
% within DICH 77,8% 22,2% 100,0%
1 Count 1 22 23
% within DICH 4,3% 95,7% 100,0%
Total Count 29 30 59
% within DICH 49,2% 50,8% 100,0%
Acknowledgement
7Wolf Eilenberg, Medical University Vienna
Laboratory work
Christine Brostjan
Branislav Zagrapan
Sarah Hetzer
Katharina Münch
Suriya Prausmüller
Paimann Nawrozi
Felix Juster
Patrick Kirchweger
Johannes Klopf
Marie-Therese Grasl
Hubert Hayden
Katharina Seif
Luca Martelanz
Renata Rajic
Alexandra R Szewcik
Alexandros Potolidos
Katharina Dosch
Clinical work
Christoph Neumayer
Ihor Huk
Biomedical Statistics
Georg Heinze
Alexandra Kaider