Biomarkers in Cardiovascular Diseases. Peter Ganz, MD. Chief, Division of Cardiology, San Francisco General Hospital

Biomarkers in Cardiovascular Diseases Peter Ganz, MD Chief, Division of Cardiology, San Francisco General Hospital Maurice Eliaser Jr. Distinguished Professor of Medicine, University of California, San Francisco DEPARTMENT OF MEDICINE DIVISION OF CARDIOLOGY SAN FRANCISCO GENERAL HOSPITAL

Biomarkers: Explosion of Interest

Examples of Biomarkers Pertinent to Atherosclerosis Blood Inflammatory Markers (e.g. hs-crp, LpPLA 2 ) Carotid Intima-Media Thickness (U/S) Coronary Calcium (CT) Getting More Information Out of Lipids: LDL-C C Fractions, apob 100, ApoA 1

What Are Biomarkers Used For Assess Cardiovascular Risk (and Motivate) Monitor Whether Treatment Is Effective Biomarkers Surrogate Endpoints for Pharma Understand Biology of Disease

What Are Biomarkers Used For Monitor Whether Treatment Is Effective Biomarkers Surrogate Endpoints for Pharma Understand Biology of Disease

54 year old male Entirely asymptomatic Treated for hypertension, BP = 134/70 mm Hg on ACE inhibitor, calcium channel blocker and diuretic Non-smoker Family history negative for premature CAD Adheres to a prudent diet, daily exercise, waist circumference 32 in. Labs: Fasting BS = 93 mg/dl; Total cholesterol = 206 mg/dl, HDL = 38 mg/dl, TG = 196 mg/dl, LDL = 129 mg/dl? Initiate statin therapy

NCEP/ATP-III Risk Estimation Goal: Match the intensity of treatment with the level of absolute risk Estimate 10-year risk for hard CHD (CHD death or non-fatal MI)

Truth About CVD Risk Prediction Health care professionals are good at judging cardiovascular disease risk.

ATP III: Major CHD Risk Factors Other Than LDL-C Cigarette smoking Hypertension: BP 140/90 mm Hg or on antihypertensive medication Low HDL-C: <40 mg/dl* Family history of premature CHD (1st-degree relative): Age male relative age <55 years female relative age <65 years male 45 years female 55 years Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. JAMA. 2001;285:2486-2497.

ATP III Framingham Risk Scoring Assessing CHD Risk in Men Step 1: Age Years Points 20-34 -9 35-39 -4 40-44 0 45-49 3 50-54 6 55-59 8 60-64 10 65-69 11 70-74 12 75-79 13 Step 2: Total Cholesterol TC Points at Points at Points at Points at Points at (mg/dl) Age 20-39 Age 40-49 Age 50-59 Age 60-69 Age 70-79 <160 0 0 0 0 0 160-199 4 3 2 1 0 200-239 7 5 3 1 0 240-279 9 6 4 2 1 280 11 8 5 3 1 Step 3: HDL-Cholesterol HDL-C (mg/dl) Points 60-1 50-59 0 40-49 1 <40 2 Step 4: Systolic Blood Pressure Systolic BP Points Points (mm Hg) if Untreated if Treated <120 0 0 120-129 0 1 130-139 1 2 140-159 1 2 160 2 3 Step 7: CHD Risk Step 6: Adding Up the Points Age Total cholesterol HDL-cholesterol Systolic blood pressure Smoking status Point total Point Total 10-Year Risk Point Total 10-Year Risk <0 <1% 11 8% 0 1% 12 10% 1 1% 13 12% 2 1% 14 16% 3 1% 15 20% 4 1% 16 25% 5 2% 17 30% 6 2% 7 3% 8 4% 9 5% 10 6% Step 5: Smoking Status Points at Points at Points at Points at Points at Age 20-39 Age 40-49 Age 50-59 Age 60-69 Age 70-79 Nonsmoker 0 0 0 0 0 Smoker 8 5 3 1 1 TM Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. JAMA. 2001;285:2486-2497. 2001, Professional Postgraduate Service www.lipidhealth.org

Current Risk Prediction The CHD Prevention Iceberg Estimated 10-Year Risk High ~8% Intermediate ~10-12% Low ~80% >20%, CHD or DM 10% - 20% <10%

Criticisms of Framingham Risk Score Uncertainty about performance Does not account for family history Does not incorporate novel risk factors that might be helpful Not useful for young adults; undertreats women; under-treats patients with extremely high levels of a single risk factor (e.g. genetic dyslipidemias) Applicability to other race/ethnic groups

How Well Does the FRS Perform? What does that mean?

ROC (Receiver Operator Characteristic) Curves ROC curves Function of the sensitivity (true-positive rate) and 1-specificity (false-positive rate) AUC = summary statistic (c-statistic) AUC represents the predictive accuracy, or the ability to discriminate (future) cases from non-cases Developed during WWI to differentiate German vs. Allied planes by radar

ROC Curves Sensitivity 1 0.75 0.5 0.25 0 Perfect Prediction AUC=1.0 AUC=0.75 Good Prediction AUC=0.5 Toss of a coin 0 0.25 0.5 0.75 1 1-Specificity Pepe et al, Am J Epidemiol. 2004;159:882-90

How Well Does the Framingham COHORT Risk Score Perform? AUC- Men AUC - Women NHANES I 0.73 0.84 NHANES II 0.78 0.80 Framingham 0.80 0.86 Tecumseh 0.81 0.88 Honolulu 0.80 -- LRC Follow-Up 0.83 -- HDFP 0.73 0.79 Ref: Heart 2002; 88: 222-228. Above models all include only age, SBP, cholesterol, smoking, and diabetes

Comparing ROC Curves Adding a new test 1 new marker + current model 0.75 Sensitivity 0.5 0.25 current model AUC=0.85 AUC=0.75 0 0 0.25 0.5 0.75 1 1-Specificity Pepe et al, Am J Epidemiol. 2004;159:882-90

Family history Examples of Proposed Additional Risk Markers Blood biomarkers CRP Multimarker approaches Subclinical disease biomarkers Coronary calcium score Carotid intima-media thickness DNA profiling

CVD Events / 1000 over 8 Years Offspring CVD Event Rates by Premature Parental CVD Status No Parental CVD 150 125 100 75 50 36 31 + Parental CVD 141 102 25 2 6 10 13 12 0 Q1 Q2 Q3 Q4 Q5 *P<0.05 **P<0.001 Quintile of Multivariable Predicted Risk * ** 95 Lloyd-Jones, JAMA 2004; 291:2204-2211

Additional Predictive Value of Premature Parental CVD Information Multivariable Model* Without Parental CVD Including Parental CVD AUC (C-statistic) for ROC Curve Models Predicting Offspring CVD Men Women Combined 0.80 0.81 0.82 *Includes offspring age, total/hdl ratio, SBP, antihtn therapy, diabetes, BMI, smoking. Lloyd-Jones, JAMA 2004; 291:2204-2211

Additional Predictive Value of Premature Parental CVD Information Multivariable Model* Without Parental CVD Including Parental CVD AUC (C-statistic) for ROC Curve Models Predicting Offspring CVD Men Women Combined 0.80 0.81 0.82 0.81 0.82 0.83 *Includes offspring age, total/hdl ratio, SBP, antihtn therapy, diabetes, BMI, smoking. Lloyd-Jones, JAMA 2004; 291:2204-2211

C-Reactive Protein (CRP) and Atherosclerosis 1930-1955 CRP is produced in response to inflammation. It is detected as a precipitate with pneumococcus C polysaccharide Tillett WS, Francis T. J Exp Med 1930 Abernathy TJ, Avery OT. J Exp Med 1941 Anderson HC, McCarty M. Am J Med 1950 CRP levels are elevated in acute myocardial infarction attributed to inflammation in the necrotic myocardium Goldner F, Meador C. South Med J 1951 Kroop IG, Shackman NH. Proc Soc Exp Biol Med 1954 Elster SK, Levinger EL, Levy HT. Clin Res Proc 1955

CRP (mg/l) Elevation of CRP in Active Coronary Artery Disease 25 20 * 15 10 5 0 non-ischemic stable CAD unstable CAD (% above normal) (20%) (13%) (90%) Berk et al, Am J Cardiol 1990; 65: 168-172

CRP Level and Mortality Risk in Acute Coronary Syndromes: FRISC-II Cumulative Probability of Death (%) 20 10 0 P=0.001 P=0.29 0 6 12 18 24 30 36 42 48 CRP > 10 mg/l (n=309) CRP 2-10 mg/l (n=294) CRP < 2 mg/l (n=314) Months Lindahl et al, NEJM 2000; 343:1139

CRP (mg/l) C-Reactive Protein in Various Populations 20 10 Acute Coronary Syndromes } elevated levels CRP assay 0 Stable CAD Apparently Healthy } normal range hscrp assay

hs-crp and Risk of Future MI and CVA in Apparently Healthy Men P Trend <0.001 P Trend <0.01 3 2 Relative Risk of MI 2 1 0 1 2 3 4 Quartile of hs-crp Relative Risk of Stroke 1 0 1 2 3 4 Quartile of hs-crp Ridker et al N Engl J Med 1997;336:973 979.

Twenty-Two Two Prospective Studies of the Association of CRP Concentrations with the Risk of CHD in Essentially General Populations, Grouped According to Several Study Characteristics Variable Date of publication Reykjavik (current) Study Between 2000 and 2002: 11 studies Before 2000: 11 studies Study size 500 Patients: 4 studies <500 Patients: 18 studies Location Western Europe: 11 studies North America: 11 studies Study sample Population or general practitioners register: 11 studies Other: 11 studies Sex Male: 12 studies Female: 3 studies Not reported separately: 8 studies Mean duration of follow-up 10 yr: 8 studies <10 yr: 14 studies Plasma or serum storage temperature -20 o C: 7 studies <-20 o C: 13 studies No. of Cases of CHD 2406 2794 1953 4107 2961 4520 2548 4477 2591 4272 1325 1471 4174 2894 3847 2905 1 2 4 Odds ratio for CHD Danesh, J. et al. N Engl J Med 2004;350:1387-1397

Multivariable-Adjusted Relative Risks of CVD: CRP With Framingham Risk or LDL-C C-Reactive Protein (mg/l) C-Reactive Protein (mg/l) Multivariable Relative Risk 25.0 20.0 15.0 10.0 5.0 0.0 <1.0 1.0 3.0 >3.0 0-1 2-4 5-9 10 Framingham Estimate of 10-Year Risk (%) Multivariable Relative Risk 3.0 2.0 1.0 <1.0 1.0 3.0 >3.0 0.0 <130 130-160 160 >160 LDL Cholesterol (mg/dl) Ridker et al., N Engl J Med. 2002;347:1557-1565 1565

AHA/CDC Guidelines for Appropriate Use of hscrp in Primary Prevention CRP is an independent marker of risk CRP cut-points ( 1 mg/l, 1-3 mg/l, >3 mg/l) The measurement of CRP should help direct further evaluation and treatment in those judged at intermediate risk by Framingham global risk factor assessment (10 20% tenyear risk) CRP may be useful in motivating therapeutic lifestyle changes (exercise, dietary changes, smoking, wt. loss) Pearson et al, Circulation 2003: 107:499-511

Treatment: Life Style Modification?

Life Style Modification? Not Always Effective

Area Under the Curve (AUC) for CV Risk Factors vs. CV Risk Factors + hscrp Study Design Sex MV-adj RR, Q4:Q1 AUC Traditional CVD RFs AUC CVD RFs + CRP Women s Heath Study Prospective W 2.3 0.81 0.81 Rotterdam Study Nested C/C W/M 1.2 0.773 0.778 MONICA Germany Prospective M 2.2 0.735 0.750 Reykjavik Cohort Nested C/C W/M 1.4 0.64 0.65 Framingham Offspring Prospective W/M 1.9 0.74 0.74 Framingham Heart Prospective W/M 1.6 CHS (6 novel RFs) Prospective W/M N/A 0.78 0.78 0.73 0.72 Conclusion: CRP does not add predictive value to the traditional risk factor models Lloyd-Jones et al, Ann Intern Med 2006; 145:35-42

Multiple Biomarkers and CVD Prediction Framingham Offspring Study AUC- : 0.76 AUC+: 0.77 10 biomarkers: CRP, BNP, N-ANP, aldosterone, renin, fibrinogen, d-dimer, PAI-1, homocysteine, urinary alb/creat Wang et al, NEJM 2006; 355:2631-2639

The C-statistic (AUC) Cook et al, Circulation. 2007;115:928-935

ROC AUC (C-Statistic) Is Insensitive in Evaluating Risk Prediction Models In the Women s Health Study, once age, smoking and blood pressure are accounted for, the addition of LDL-C, HDL-C, or CRP has minimal effect on the c-statistic. We should not discard CRP any more than LDL-C or HDL-C measurements. It is more appropriate to determine whether a new test re-classifies individuals into different risk categories (calibration). Cook, Ridker et al, Ann Intern Med 2006: 145:21-29 Cook, Circulation 2007; 115:128-135

Best Fitting Model A Age HbA1c %, if diabetic Ln(SBP) Current Smoking Ln(hsCRP) Parental history of MI < age 60 Apo-B 100 Apo A-I [Lp(a)-10] if Apo-B 100 > 100 Clinically Simplified Model B (Reynolds Risk Score) Age HbA1c %, if diabetic Ln(SBP) Current Smoking Ln(hsCRP) Parental history of MI < age 60 Non-HDL-C (TC-HDL-C) HDL-C -- Not in the final model: Obesity, exercise levels, alcohol use, creatinine, homocysteine, fibrinogen, sicam-1

Risk Reclassification: ATP-III vs Reynolds Risk Score Intermediate Risk 10-Year Risk Reynolds Risk Score (%) 40 30 20 10 5 0 Intermediate Risk Ridker et al, JAMA 2007;297:611-9 0 5 10 20 30 10-Year Risk ATP-III (%) 40

Risk Reclassification: ATP-III vs Reynolds Risk Score Intermediate Risk 40 10-Year Risk Reynolds Risk Score (%) 30 20 10 5 0 4% 27% 21% Intermediate Risk Ridker et al, JAMA 2007;297:611-9 0 5 10 20 30 10-Year Risk ATP-III (%) 40

Risk Reclassification: ATP-III vs Reynolds Risk Score Intermediate Risk 40 10-Year Risk Reynolds Risk Score (%) 30 20 10 5 0 4% 27% 16% 21% 20% 25% Intermediate Risk Ridker et al, JAMA 2007;297:611-9 0 5 10 20 30 10-Year Risk ATP-III (%) 40

Risk Reclassification: ATP-III vs Reynolds Risk Score Intermediate Risk 40 10-Year Risk Reynolds Risk Score (%) 30 20 10 5 0 4% 21% 2% 27% 20% 16% 4% 25% Intermediate Risk Ridker et al, JAMA 2007;297:611-9 0 5 10 20 30 10-Year Risk ATP-III (%) 40

Risk Reclassification: ATP-III vs Reynolds Risk Score Intermediate Risk Nearly half of the intermediate risk subjects were reclassified! 40 10-Year Risk Reynolds Risk Score (%) 30 20 10 5 0 2% 21% 27% 4% 56% 55% 20% 96% 16% 4% 75% 25% Intermediate Risk Ridker et al, JAMA 2007;297:611-9 0 5 10 20 30 10-Year Risk ATP-III (%) 40

Evaluating Biomarkers: Tug of War ROC Curves Reclassification

JUPITER Rosuvastatin in the Primary Prevention of Cardiovascular Events Among Individuals with Low LDL-C C and Elevated CRP No History of CAD Men >55, Women > 65 LDL-C <130 mg/dl CRP >2 mg/l 4 week Run-in Rosuvastatin (N =7500) Placebo (N =7500) MI Stroke Unstable Angina CVD Death CABG/PTCA Screening Visit Randomization Visit Safety Visit Bi-Annual Follow-Up Visits End of Study Visit LDL CRP FHS Lipids hs-crp LFTs HbA1C Lipids hs-crp LFTs Lipids hs-crp HbA1C Early termination of JUPITER due to marked benefit is expected to make CRP screening widespread

What Are Biomarkers Used For Biomarkers Surrogate Endpoints for Pharma Understand Biology of Disease

Finding the Non-Responders to Statin Therapy: Dual Goals Reducing both CRP and LDL-C is important for: Retarding progression of coronary atherosclerosis REVERSAL study, Nissen et al, NEJM 2005; 352:29-38 Achieving favorable clinical outcome PROVE IT study, Ridker et al, NEJM 2005; 352; 20-28 A to Z study, Morrow et al, Circulation 2006; 114;241-248

What Are Biomarkers Used For Monitor Whether Treatment Is Effective Biomarkers Surrogate Endpoints for Pharma Understand Biology of Disease

AHA's Coronary Calcium and CT Angiography Statement

AHA Scientific Statement on Coronary Calcium Score In asymptomatic individuals with risk factors, start with the Framingham Risk Equation. In intermediate-risk patients (10-20% risk), it may be reasonable to use EBCT or MDCT to measure coronary calcium score to refine clinical risk prediction and to select patients for more aggressive target values for lipidlowering therapies Why such a low-level recommendation? (class IIb, level of evidence: B).

CVD Risk Assessment by Measuring Coronary Calcium Score (EBCT) Circulation 2006; 114:1761-1791

Characteristics and Risk Ratio for Follow-Up Studies Using EBCT An independent, multivariable odds ratios in excess of 3.0 is typically required to increase the c-statistic, and hence improve discrimination of cases from noncases by 5%. Most new risk factors [including CRP] do not achieve this level of added risk above standard risk factors. Pepe et al, Am J Epidemiol. 2004;159:882-90 Circulation 2006; 114:1761-1791

Characteristics and Risk Ratio for Follow-Up Studies Using EBCT Cohort vs. Referred or self-referred subjects Circulation 2006; 114:1761-1791

Promise of the Coronary Calcium Score for Risk Assessment No uniformity for calcium score cut-offs scores among these studies; Only 4 studies -? Adequate representation of ethnic groups, ages, genders; (Class IIb, level of evidence: B) Prospective, Population-Based Cohort Studies of Apparently Healthy Subjects; Risk Factors Were Measured; Multivariable Analysis Was Applied Lot of promise but fine-tuning is needed n age F/U yrs Calcium Score Cut-off Risk Factor Assessment St Francis Heart Study Relative Risk South Bay Heart Watch Prospective Army Coronary Calcium Project Rotterdam Study Arad et al, J Am Coll Cardiol 2000;36:1253 Greenland et al, JAMA 2004;291:210 Taylor et al, J Am Coll Cardiol 2005;46:807 Vliegenthart et al, Circulation. 2005;112:572 Circulation 2006; 114:1761-1791

Multi-Ethnic Study of Atherosclerosis (MESA) 6722 man and women free of apparent CVD; age 62 years; Unadjusted Kaplan Meier Cumulative-Event Curves N=833 N=752 N=1728 N=3409 Detrano et al, NEJM 2008; 358:1336-1345

MESA Results Hazard Ratios Adjusted for Standard Risk Factors Area under the Curve for Risk Factors Alone and for Risk Factors plus Coronary-Artery Calcium Score Detrano et al, NEJM 2008; 358:1336-1345

Simvastatin With or Without Ezetimibe in Familial Hypercholesterolemia: ENHANCE trial Carotid IMT Kastelein, et al, N Eng J Med 2008; Now dismissed Just a surrogate

Carotid Ultrasound Study Double lines on near and far walls Patient position Common carotid artery Longitudinal view ASE Consensus Statement, Journal of the American Society of Echocardiography cardiography 2008; 21:93-111 111

Carotid IMT Measurement Pixel resolution ~ 0.11 mm when imaging at 4 cm depth; Multiple measurements of several extended segments lenghts permit subpixel resolution lumen-intima interface CIMT media-adventitia adventitia interface Kastelein et al, Eur Heart J 2008; 29:849-858; 858; ASE Consensus Statement, Journal of the American Society of Echocardiography cardiography 2008; 21:93-111 111

Carotid IMT is Associated With Risk of CV Events in Healthy Individuals Myocardial Infarction Atherosclerosis Risk in Communities (ARIC) n=13,204 Cardiovascular Health Study (CHS) n=4476 Rotterdam Study n=2267 Malmö Diet and Cancer Study subcohort (MDCS) n=5163 Carotid Atherosclerosis Progression Study (CAPS)n=5052 Total I 2 for heterogeneity 42.5% n=30,162 Stroke Atherosclerosis Risk in Communities (ARIC) n=14,165 Cardiovascular Health Study (CHS) n=4476 Rotterdam Study n=5479 Malmö Diet and Cancer Study subcohort (MDCS) n=5163 Carotid Atherosclerosis Progression Study (CAPS)n=5052 Total I 2 for heterogeneity 28.2% n=34,335 0.9 1.0 1.1 1.2 1.3 1.4 Hazard ratio (95% CI) per 0.1 mm difference in cimt* *Adjusted for age and sex Lorenz MW,et al. Circulation. 2007;115:459-467

Carotid IMT much work remains to standardize the results; technically challenging; relatively expensive

A New Risk Estimator? Risk Factor Units Value Notes Gender Age Total Cholesterol HDL Systolic Blood Pressure m years 45 mg/dl 230 mg/dl 40 mmhg 135 Treatment for Hypertension {Only if SBP>120} Current Smoker yes (y) or no (n) yes (y) or no (n) n n Time Frame for Risk Estimate 10 years Lifetime Your Risk 5% 46%

Identifying More High Risk Individuals by Extending the Time Horizon Estimated 10-Year and Long-Term Risk High Intermediate Low Lloyd-Jones Curr Opin Lipidol 2006

Applications of Lifetime Risk Estimation Public health perspective: Best example of utility of lifetime risk data: Lifetime Risk of breast cancer for women at age 40 = 1 in 8 Published in early 1990s, widely disseminated Annual mammography rates, US women over 40: 1987 25% 1994 60%

Disease Lifetime Risks for Common Diseases at Age 40 Men Women Any CVD* 2 in 3 >1 in 2 CHD 1 1 in 2 1 in 3 AF 2 1 in 4 1 in 4 CHF 3 1 in 5 1 in 5 Stroke 4 1 in 5 1 in 5 Hip fracture 5 1 in 20 1 in 6 Breast cancer 6,7 1 in 1000 1 in 8 Prostate cancer 6 1 in 6 -- Lung cancer 6 1 in 12 1 in 17 Colon cancer 6 1 in 16 1 in 17 Diabetes 8 1 in 3 1 in 3 * Unpublished FHS data. 1. Lloyd-Jones, Lancet 1999. 2. Lloyd-Jones Circulation 2004. 3. Lloyd-Jones, Circulation 2002. 4. Seshadri, Stroke 2006, Age 55. 5. Cummings, Arch Intern Med 1989. 6. SEER cancer statistics review, 1975-2000. 7. Feuer, J Natl Cancer Inst 1993. 8. Narayan, JAMA 2003.

Would Treating Cholesterol Early in Life be More Effective than Starting in Middle Age? Atherosclerosis Risk in Communities (ARIC) study PCSK9 mutation (example of Mendelian randomization): 28% lifetime reduction in LDL-C 88% CHD risk reduction!!! Cohen et al, NEJM 2006; 354: 1264-1272

Summary For the foreseeable future Novel markers may be useful for further risk discrimination among intermediate risk (hscrp, coronary calcium score) Do a better job with what we already have Use the Framingham Risk Score Control rates of HTN, lipids, DM, smoking Control weight!!!

Summary Lifetime risk estimates can provide an important adjunct to 10-year risk estimates Identify younger individuals with low short-term term but high lifetime risk Help with population strategy Communicate risk Motivate adherence to therapies

It Has Been a Pleasure to Speak to You!