Wolfgang Koenig, MD, FACC

1
1st Vulnerable Patient Satellite
Symposium,American Heart Association Orlando,
USA, November 11, 2003
Is the Framingham model sufficient for
prediction of coronary events? Should CRP be
added to Framingham Risk Score? How about
calcium score?

• Wolfgang Koenig, MD, FACC
• Dept. of Internal Medicine II - Cardiology
• University of Ulm Medical Center, Ulm, Germany

2
CHD Risk Assessment in Asymptomatic Patients
Selective Use of Noninvasive Testing
Modification of Probability Estimates of CHD by
Non-invasive Testing

• Assessment by multivariable
• statistical models e.g.
• Framingham Risk Score or
• PROCAM score
• Clear guidelines for high or low
• risk subjects, but not so for
• those at intermediate risk

Post-test Probability of CHD Event in 10 Yrs
Pre-test Probability of CHD Event in 10 Yrs
modified after Greenland et al. Circulation
20011041863-1867
3
Can CRP Change Our Practice?
C-Reactive Protein Modulates Risk Prediction
4
C-Reactive Protein Modulates Risk Prediction
MONICA/KORA Augsburg Cohort 1984-98
Methods Patient Population and Assays

• 3,435 men aged 45-74 years, participating in the
  three MONICA surveys 1984/85, 1989/90, 1994/95
• Exclusion of prevalent CHD
• Standardized assessment of cardiovascular risk
  factors Total cholesterol, HDL-C, blood
  pressure, smoking, BMI, physical activity, social
  class, diabetes mellitus, alcohol consumption.
• Endpoint determination according to the MONICA
  protocol (fatal and non-fatal MI and sudden
  cardiac death)
• Determination of CRP by a hs-IRMA (Hutchinson et
  al. Clin Chem 2000) with a detection limit of
  0.05 mg/L (CV lt 12).
• Determination of total cholesterol and HDL-C by
  routine enzymatic methods (CV lt 4)

Koenig et al. AHA 2003
5
RR of CHD According to the Estimated 10-Yrs Risk
Alone and in Combination With CRP MONICA
Augsburg Cohort
(N3,435 Men 45-74 Yrs 191 Events FU 6.6 Yrs)
Population at risk 809 914 650
526 536
8
P0.09
8
P0.02
7
7
6
6
AIC 2776
AIC 2789
5
5
Multivariable Relative Risk
4
4
P0.03
3
3
P0.26
2
2
P0.20
1
1
18 32 35 50
56
0
0
lt 6 6-10 11-14 15-19 ?20
lt 6 6-10 11-14 15-19 ?20
Framingham Estimate of 10-Year Risk ()
Koenig et al. AHA 2003
6
Risk of a First Coronary Event by Cox Model w/o
and With CRP for the FRS With 3 and 5 Categories
Factor Events/n Events/n HR (95CI) P-value HR (95CI) HR (95CI) P-value
FRS 1 lt6 18/809 18/809 Ref. Ref. Ref.
() 6-19 117/2090 117/2090 2.81 (1.71-4.62) 2.39 (1.45-3.94) 2.39 (1.45-3.94)
?20 56/536 56/536 6.19 (3.64-10.54) lt0.0001 4.85 (2.82-8.33) 4.85 (2.82-8.33) lt0.0001
AIC 2816 2797 ?AIC 19 ?AIC 19
AUC 0.713 0.740 0.740 0.0077
FRS 2 lt6 18/809 18/809 Ref. Ref. Ref.
() 6-10 32/914 32/914 1.63 (0.91-2.90) 1.46 (0.82-2.61) 1.46 (0.82-2.61)
10-14 35/650 35/650 2.70 (1.53-4.77) 2.35 (1.32-4.16) 2.35 (1.32-4.16)
15-19 50/526 50/526 5.61 (3.27-9.62) 4.50 (2.59-7.80) 4.50 (2.59-7.80)
?20 56/536 56/536 6.21 (3.65-10.57) lt0.0001 5.01 (2.91-8.62) 5.01 (2.91-8.62) lt0.0001
AIC 2789 2776 ?AIC 13 ?AIC 13
AUC 0.735 0.750 0.750 0.0163
AIC, Akaikes Information Criterion ?AIC, AIC
(model without CRP) AIC (model with CRP) AUC,
Area under the curve
Koenig et al. AHA 2003
7
Coronary Calcification and Atherosclerotic
Cardiovascular Disease Events St. Francis
Heart Study

• Prospective, longitudinal, population-based study
  of asymp-tomatic men and women aged 50 to 70 with
  no prior history, symptoms or signs of
  atherosclerotic CVD
• Subjects on or with indication for lipid-lowering
  therapy excluded
• Coronary calcium measured by EBCT scanning,
  Agatston method
• Events verified by independent Endpoints
  Adjudication Committee, blinded to coronary
  calcium score
• A total of 5,585 subjects were scanned
• Risk factors measured in 1,817
• 4.3 years follow-up, 96 complete
• 122 subjects (0.6/year) with ? 1 atherosclerotic
  CVD event

Arad et al. ACC, Chicago 2003
8
Prediction of CVD Events by Coronary Calcium
Score St. Francis Heart Study

• Baseline Calcium Score
• and CVD Events
• Event 584 ? 775
• P lt 0.0001
• No event 142 ? 381
• Coronary Calcium Score (?100
• vs lt100) and CVD Events
• All CVD 122 9.5 (6.5-13.8)
• All coronary 105 10.7 (7.1-16.3)
• MI/coronary death 43 9.9 (5.2-18.9)

RR
Events N RR (95 CI)
Calcium Score
Arad et al. ACC, Chicago 2003
9
Prediction of CVD by Coronary Calcium Score vs
Framingham Risk Score St. Francis Heart Study
Calcium score vs Framingham risk index
prediction of coronary events Area
under ROC curve P-value Calcium score
0.81 ? 0.03 lt 0.01 Framingham 0.71
? 0.03
per year (observed)
per 10 years (predicted)
Arad et al. ACC, Chicago 2003
10
Summary and Conclusions

• The addition of CRP to a prediction model of the
  FRS resulted in a better fit of the model
  containing CRP and significantly improved
  prediction of incident CHD for the calculated FRS
• The latter was particularly true for those at
  intermediate risk (10-20 over 10 years)
• Thus, CRP measurement modulates coronary risk and
  may therefore modify the physicians
  interpretation of the patients risk status
• Calcium scoring also seems to improve prediction
  based on the FRS
• However, these findings have to be replicated in
  other populations