Risk Assessment for Primary Prevention

Risk Assessment for Primary Prevention AKA Some of You Won t Get a Second Chance to Make a First Impression Alternative Title: The Tension Between Population Medicine and Personalized Medicine Joshua Liberman, MD, FACC Governor-Elect, ACC-Wisconsin Chapter How can we assess and stratify risk? Risk Factor Evaluation Lab Testing Imaging Risk Factor Evaluation Current guidelines recommend office-based risk stratification of all individuals using multiple risk factor scores to determine global risk An expanding number of readily accessible assessments of global cardiovascular risk have been widely available to health care practitioners for decades. Framingham Reynold s Risk Score HEARTSCORE Risk Calculators AHA/ACC Pooled Cohort Equation (PCE) Framingham Risk Score The gold standard Developed in 1998 Uses: Age Total Cholesterol HDL-C Smoking Blood pressure History of diabetes mellitus 1

Other Risk Scores Estimates sex-specific risk of coronary heart disease (CHD) events over the subsequent 10 years Reynolds risk score Incorporate the use of biomarkers (hscrp) Family history of premature CHD. Each individual s risk can then be categorized as low, intermediate, or high (<10%, 10%- 20%, and >20%) European Scores: PROCAM6 HEARTSCORE Both are variations of the Framingham score ACC/AHA Pooled Cohorts Equation This new calculator incorporated: ACC/AHA PCE Different than Framingham: Sex Age Race Smoker? Diabetes? Total Cholesterol HDL Systolic Blood Pressure Rx for HTN? Added endpoint of Stroke to non-fatal MI and CHD death Expanded applicability to different ethnic groups Fact: Great. So what s the problem? There is solid evidence that if you modify the variables that comprise global CHD risk assessment tools, you will decrease CV disease. Smoking cessation Blood pressure control Physical activity Lipids 2

Fact: Surprisingly, there is no clear evidence that simply providing patients with global CHD risk information actually improves: Risk perception Adherence to therapy Improvements in cardiovascular outcomes The Problem with Calculators Despite the presence and use of these scores, in 50% of patients the initial manifestation of CAD is an MI or death In 1/3 it is sudden death So despite recommendations to use these scores, we are clearing missing many people who go on to have fatal events. The Problem with Calculators Part II The majority of CV events occur in patients deemed low risk by these calculators Data from NHANES suggests that low risk patients comprise 85% of the population Constitute 2/3rds of the population risk In the Real World Despite using risk prediction calculators, patients continue to suffer events A majority of patients hospitalized with CAD have LDL-c levels that are normal by current guidelines ~ 50% of patients with a first ACS event had LDL levels < 100 mg/dl 17% had LDL < 70 Sachdeva A et al. Am Heart J 2009; 157:111-117 3

The Problem with Calcultors Part III Patients continue to have clinical events Despite being at their recommended targets Despite having normal lipid panels Despite being on treatment Our current process of risk assessment is not adequate In substantial numbers We need improved identification of at-risk individuals A path to a more personalized risk assessment How else can we stratify risk? Risk Factor Evaluation Calculators How else can we stratify risk? Risk Factor Evaluation Calculators Lab Testing Lab Testing Imaging Imaging 4

Lipid Testing ApoB NMR Lab Testing for Risk Stratification CardioCRP/hsCRP CardioDx Corus CAD Test Lipid Testing ApoB NMR Lab Testing for Risk Stratification CardioCRP/hsCRP CardioDx Corus CAD Test But is this really risk assessment, or just more accurate lipid assessment? Lipid Testing ApoB NMR Lab Testing for Risk Stratification CardioCRP/hsCRP CardioDx Corus CAD Test hscrp Acute phase protein produced predominantly by hepatocytes Non-specific marker of inflammation Can be produced by inflamed endothelium as well Inflammation from a variety of different causes appears to increase CV risk May not be just a marker of inflammation, but also a participant Pearson, TA. Circulation 2003; 107:499 Relative Risk of CV Events According to Baseline Levels of hs-crp in Healthy Postmenopausal Women hscrp Just a marker, or is the level important? There is a gradient of increasing risk proportional to hs-crp beginning at extremely low levels of CRP Hs-CRP of 0.2 is better than 0.5 which is better than 1.5 Jupiter confirmed that patients who achieved lower hs-crp levels did better P for trend < 0.001 Source: Ridker 2000 Arima H et al. ATVB 2008;28:1385-91 Ridker NEJM 2008; 359:2195-2207 5

hs-crp Reducing both hs-crp and LDL-C is beneficial PROVE-IT-TIMI 22 CARE REVERSAL A to Z JUPITER Lowered by: Weight Loss Increased exercise Smoking cessation Statins hs-crp Ridker, PM et al. NEJM 2005; 352:20-8 Ridker PM, et al. Circ 1998; 98:839-44 Nissen SE et al. NEJM 2005; 352:29-38 Morrow, DA et al. Circ 2006; 114:281-8 hs-crp Still somewhat controversial whether it is an independent risk factor No evidence yet* that lowering CRP lowers risk or improves outcomes independent of lowering LDL CIRT is a randomized, double-blind, placebo-controlled, multi-center, eventdriven trial funded by the National Heart Lung and Blood Institute (NHLBI) that will enroll 7,000 men and women from the United States and Canada. While inflammation plays a critical role in atherothrombosis, it is unknown whether inhibition of inflammation per se will lower vascular event rates. Evaluate whether or not low-dose methotrexate (LDM) will reduce rates of myocardial infarction, stroke, and cardiovascular death among stable coronary artery disease patients with type 2 diabetes or metabolic syndrome, conditions associated with an enhanced pro-inflammatory response. Lipid Testing ApoB NMR CardioCRP/hsCRP Genetic Testing Gene Presence? 9p21 Lp(a) Gene Expression? CORUS CAD 6

Corus CAD Integrates age, sex and gene expression to calculate a score that has been demonstrated to accurately assess the likelihood of obstructive CAD Production of RNA changes in response to disease states High Sensitivity (89%) High Negative Predictive Value (96%) How else can we stratify risk? Risk Factor Evaluation Calculators Lab Testing Imaging Designed for use in evaluation of pts with chest pain Rosenberg, S et al. Ann Int Med 2010;153:425-34 Risk Stratification via Imaging Stress Tests CT Angiography of Coronary Arteries Cardiac MRI Cardiac PET Scan Calcium Scoring A Word About Stress Tests Who thinks we should use stress tests to risk stratify primary prevention patients? A Word About Stress Tests What is the NNT to prevent death for performing stress tests in asymptomatic patients with risk factors? NNT = What is the NNT to prevent death for performing stress tests in asymptomatic patients with known CAD? NNT = DIAD Study 1123 asymptomatic diabetic patients randomized to screening with nuclear stress testing or to no screening. Ischemia was detected in 25% of patients Severe abnormalities in approximately 6% of patients. Patients were then followed for ~5 years. There were 15 events in the screening group and 17 events among those who weren't screened Statistically nonsignificant No difference in any of the secondary end points angina, heart failure, stroke, and revascularizations. Young LH et al. JAMA 2009; 301: 1547-1555 7

Why? Established noninvasive methods of evaluating CAD, such as stress testing, only identify patients with advanced atherosclerotic disease that are flow-limiting and cause myocardial ischemia But autopsy studies show that most fatal events come from non flow-limiting lesions Rupture of vulnerable plaques that were not previously flow-limiting We need a better way to identify who has sub-clinical atherosclerosis Risk Stratification via Imaging Stress Tests CT Angiography of Coronary Arteries Cardiac MRI Cardiac PET Scan Calcium Scoring CT Angiography of the Coronaries Pros: Available Accessible Able to see within coronaries Can see composition of plaque Cons Cost Radiation exposure Cardiac MRI Pros: Great for overall structure of the heart Pathology of myocardium Cons: Duration Cost Can t see within coronaries due to motion Cardiac PET Pros: The best stress test Excellent performance in detection of ischemia, viability Excellent in obese patients Cons: Duration Cost Availability Don t actually visualize within coronaries Can t see subclinical atherosclerosis 8

Risk Stratification via Imaging Stress Tests CT Angiography of Coronary Arteries Cardiac MRI Cardiac PET Scan Calcium Scoring Calcium Scoring Calcium Scoring: What actually is it? ECG-triggered, low-radiation, non-contrast CT scan 3mm Slices Allows detection and quantification of the amount of coronary artery calcium (CAC) Coronary calcium is defined as a lesion above a threshold of 130 Hounsfield units, with an area of 3 adjacent pixels (at least 1 mm 2 ). Coronary Calcium Coronary Calcium Scan Measured by Agatston score Equivalent mass and volume Reproducible, semiautomatic computer method Result compared to population norms Calcium in LAD Calcium Scoring is Accurate The presence of calcium in coronary arteries is pathognomonic of atherosclerosis Coronary calcium is 100% specific for coronary atherosclerosis In common practice, the scores are expressed as a percentile score based on age and sex Use one of the many available databases MESA, Framingham, etc. Budoff MJ et al. 2006. Circulation, 114:1761 91. Keelan PC, et al Circulation 2001;104:412-417 Budoff MJ et al. JACC 2008 Nov 18;52(21):1724-32 9

Accurate for Atherosclerosis Atherosclerotic plaque burden and the extent of CAC are closely correlated Intravascular ultrasound Post-mortem autopsy Histopathology The total CAC score measured represents an anatomic measure of overall cardiac plaque burden Mintz GS et al. 1997. JACC 29:268 74. Baumgart D et al. 1997. JACC 30:57 64. Budoff MJ et al. 2006. Circulation, 114:1761 91. Rumberger et al Circulation 1995;92:2157-62 Keelan PC, et al Circulation 2001;104:412-417 Budoff MJ et al. JACC 2008 Nov 18;52(21):1724-32 Background CAC is proportional to the extent of atherosclerosis The extent of coronary atherosclerosis, rather than the severity of an individual stenosis, is the most important predictor of death due to CV causes Schmermund et al 1997 Why Extent and Not Severity? Because most fatal events come from non flowlimiting lesions Rupture of vulnerable plaques that were not previously flow-limiting CAC provides the most accurate available method of the early detection of atherosclerosis The Bethesda conference The more lesions you have, the higher risk for rupture of one of them Taylor, AJ et al. JACC 2003; 41:1860-2 Calcium Scoring is Prognostic Calcium Scoring for Risk Stratification There is a linear relationship between CAC and CHD events The higher the score, the worse the prognosis Scores in the highest quartile predict a significantly increased risk of a cardiac event > 2x for women ~10x for men Even after adjustment for NCEP ATP III category Erbel RA et al. ACC Scientific Sessions 2009 Shaw LJ, et al Radiology. 2003;228:826-833 10

Negative Risk Factor cimt <25 th percentile Absence of carotid plaque Brachial Flow-Mediated Dilation >5% change ABI >0.9 and <1.3 hscrp <2 Homocysteine <10 ntpro-bnp <100 No microalbuminuria No FamHx of premature CAD Absence of metabolic syndrome Healthy Lifestyle The strongest negative risk factor was CAC = 0 Blaha et al. Circulation 2016; 133:849-58 Prognosis for patients with zero scores The Power of Zero Sarwar et al, JACC Imaging 2009; 2:675-688 Calcium Scoring is Prognostic CAC = 0: Patients have a very low rate of CV death or MI (~0.4%) Across ethnicities Up to 12 years out CAC = 0 The Power of Zero In > 64,000 asymptomatic patients Only 146 of 25,903 patients (0.56%) had a CV event during a mean follow-up period of 51 months Overall 15-year mortality rate was ~3% Budoff MJ, et al. JACC 2007;49:1860-1870. Erbel RA et al. ACC Scientific Sessions 2009 Shaw LJ, et al Radiology. 2003;228:826-833 Georgiou D, et al. JACC 2001;38:105-110 Greenland et al. JACC, 2007; 49:378-402 Taylor AJ, et al. JACC 2005 Sep 6;46(5):807-14 Arad Y, et al. JACC 2005;46:158-165 Detrano A, et al. NEJM 2008;358:1336-1345 Shaw LJ et al. Ann Int Med 2015 7;163(1):14-21. Sarwar et al, JACC Imaging 2009; 2:675-688 11

However, when CAC > 0: Even in older patients: Those with absent or low CAC are at a significantly lower risk of mortality compared with the general population In patients with any detectable calcium, their 3-5 yr CHD risk of events is nearly 4-fold higher CAC 1 to 10 (1.06%) CAC >10 (3.96%) May not be quite high enough for secondary prevention, but clearly should have aggressive risk factor modification Nakanishi R et al. Eur Heart J Cardiov Imag. 2016;17(11):1305-14 Greenland et al. JACC, 2007; 49:378-402 Georgiou D, et al. JACC 2001;38:105-110 Arad Y, et al. JACC 2005;46:158-165 Blaha et al. JACC Img 2009; 2:692-700 Coronary Calcium Scoring Among individuals at intermediate risk, a calcium score of >100 ( > 75th percentile for age) would yield a post-test probability of >2% per year High risk ~ CHD risk equivalent population Requiring secondary prevention strategies High Scores = High Risk Patients with increased plaque burdens are ~10x more likely to suffer a cardiac event over the next 3 5 years. For women with a CAC score 400, the 15-year mortality was 23.5% Pletcher MJ, et al Arch Intern Med 2004;164:1285-1292 Detrano R, et al. JACC. 1996;27:285-290. Keelan PC, et al Circulation 2001;104:412-417 Georgiou D, et al. JACC. 2001;38:105-110. Erbel RA et al. ACC Scientific Sessions 2009 Shaw LJ, et al Radiology. 2003;228:826-833 Kelkar AA et al. Circ Cardiovasc Imaging. 2016 Apr;9(4) What about Very High scores? Asymptomatic individuals with a very high CAC score ( 1000) followed for 17 ± 11 months. Patients with scores 1000 experienced an annual event rate of 25% for hard cardiac events (!!!) MI and death Calcium Scoring for Diagnosis/Identification Wayhs et al (2002) 12

Population vs. Personalization Calculators work for Populations, but do they work for all individuals? And do our calculators even work for populations? The Ideal Risk Assessment Tool Identify patients supposedly at risk, but really at low risk To save them from unneccesary medications and procedures Identify patients thought to be low risk, but whom are really high risk To institute appropriate preventive therapy Is CAC better at identifying risk? Low Risk Low Risk Women (by ACC/AHA PCE) <7.5% risk by Pooled Cohort equation 36% had positive CAC scores. ~3-fold higher risk of cardiac event Is CAC better at identifying risk? Intermediate Risk ~40% of Intermediate Risk patients (5-7.5% risk) by ACC/AHA Pooled Cohorts had CAC = 0 18% of patients >75yo had scores = 0 Scores >100 found in 10% of patients Very high CAC scores ( 400) found in 3-5% Kavousi et al. JAMA 2016;316(20):2126-34 Mortensen et al. JACC 201668(9):881-7 Isma eel H et al. Am J Cardiol 2016;118:1480- Pusnani et al. JAMA 2015; 314:134-41 Is CAC better at identifying risk? Intermediate Risk Intermediate Risk Patients 15-25% of patients can be reclassified as high risk Possibly up to 40% depending on cutoff 90 th percentile vs. 100 HU score Is CAC better at identifying risk? High Risk Only 22% of patients deemed High Risk by ACC/AHA Pooled Cohorts had CAC>300. Only 42% had CAC>100 ~ 60% could be shifted to low risk Rosner Preis, S et al. AJC 103(12):1710-5 Erbel RA et al. ACC Scientific Sessions 2009 Isma eel H et al. Am J Cardiol 2016;118:1480-5 13

Is CAC better at identifying risk? High Risk Approximately 1 out of 5 patients with a PCE prediction of >15% ten year risk actually had CAC= 0 The Number Needed to Screen (NNS) in the High Risk (>15% PCE) group to find 1 patient with a CAC= 0 4.5 20% of patients thought to be high risk were really extremely low risk In the Intermediate Risk group (7.5%-15% PCE) NNS = 2.6 Mortensen et al. JACC 201668(9):881-7 Re-Classified Patients 1 out of 7 patients had significant disagreement/discrepancy between risk predicted by CAC and by Pooled Cohorts. Re-Classified Patients In the patients who could be re-allocated to a low-risk group because of a score of 0: The event rate was just 1% over five years Truly low risk In the group reclassified as high risk: There was an event rate of >8% over five years Truly high risk Isma eel H et al. Am J Cardiol 2016;118:1480-5 Erbel RA et al. ACC Scientific Sessions 2009 Is CAC better at identifying risk in patients with DM? Diabetes Is DM truly an anginal equivalent? 25-30% of diabetics have no evidence of CAD Risk Stratification Patients with DM and no evidence of CAD by coronary CTA did not exhibit an increased risk of death compared with the propensity-matched nondiabetic subjects Calcium Scoring vs. Risk Calculators Rana et al. Diabetes Care 2012;35:1787-94 Blanke P et al. JACC Card Imag. 2016 Nov;9(11):1280-1288a Raggi P t al. J Am Coll Cardiol 43:1663 1669 14

Why? Because Risk Factors Stink The poor discriminatory power of risk factors was demonstrated in 542,008 patients presenting with a first myocardial infarction: 14.4% had 0 RFs 34.1% had 1 RF 31.6% had 2 RFs 15.4% had 3 RFs 4.1% had 4 RFs Will a Calcium Score Change Management? Canto JG et al. JAMA 306:2120 7 Change in Management Strategy For patients previously considered high risk with negative scores Can downgrade therapy through Shared Decision- Making process For patients previously considered low risk that have positive scores Counsel accordingly and start therapy CAC = 0 Data suggest that aggressive management in this cohort is not warranted downgrade therapy What does that mean? Up to one-third of the statin-eligible population would do well without chronic medical therapy and would be effectively treated with lifestyle changes alone Nasir K, et al. JACC 2015;66:1657-68 15

High Score A score of >100 (or >75th percentile) would yield a post-test probability in the majority of patients >2%/yr Event rate 20% over 10 years Re-classifies the patient within the range of a CHD risk equivalent population Requires secondary prevention strategies Patients care about their score CAC has been shown to positively affect initiation of and adherence to medication and lifestyle changes In 505 asymptomatic patients, statin adherence >3yrs after visualizing their CAC scan: 90% in those with a CAC score >400 44% in those with a CAC score of 0 Greenland, P. et al. J Am Coll Cardiol 2007;49:378-402 Patients care about their score EISNER Trial CAC-directed care produced significant improvement in systolic blood pressure, low-density lipoprotein cholesterol, weight, waist size, and FRS compared with usual care, without an increase in downstream testing. 980 asymptomatic subjects followed for 3 years Significant increases seen in: Exercise Aspirin therapy initiation Dietary changes Highest in patients with CAC scores >400 Rozanski A et al. JACC 57:1622 1632. Kalia NK et al. Atherosclerosis 185:394 399. Orakzai RH et al. Am J Cardiol 101:999 1002 Calcium Scoring is Cost-Effective Primary Prevention Setting: Calcium scoring in at-risk men (age 40-50yo), was modeled to cost $37,633 per quality-adjusted life year saved. Assuming a 30% improvement in survival Usual benefit of a statin Among those with 10-year estimated risk of 6% to 20%, decision to avoid treatment among those with CAC=0 versus treat-all approach with generic statins is cost-effective, as long as CAC testing is priced < $235. Most patients would prefer to pay for an imaging test than take a pill every day for life Taylor AJ, et al. JACC 2005 Sep 6;46(5):807-14. Hutchins R et al. Circ Cardiovasc Qual Outcomes 2015;8:155-63 Roberts ET et al. PLoS One. 2015;10:e0116377 16

Calcium Scores and Stress Testing ACC/ASNC appropriateness criteria state that a zero calcium score precludes the need for nuclear stress testing Represents a potential huge cost savings for the US health care system Fast No contrast Calcium Scoring: Pros Low radiation dose Cheap $150 Cheaper than NMR, Stress Tests, etc. Just not covered by insurance Brindis RG, et al. J Am Coll Cardiol 2005;46:1587-1605 Sarwar A et al. JACC Imaging Calcium Scoring: Cons Costs of Potential Downstream Testing Radiation Exposure A Word About Radiation Average Radiation Doses Avg Background radiation in US Chest x-ray Mammogram Chest CT Nuclear Stress Test Low Dose Lung CA screen CT 64-slice Calcium Score 3.0 msv/yr 0.05 msv 0.8 msv 7 msv 15 msv 1.5mSv ~1 msv If we REALLY care about radiation exposure Thompson RC. J Nuc Card 2006: 13(1); 19-23 Massenger B et al. Int J Cardiovasc Imaging. 2016 Mar;32(3):525-9 17

Does the patient even need another test? We can prevent unnecessary radiation exposure by using info from available studies before ordering new ones ~ 600,000 3 mm ECG-gated CT scans done in the United States annually 40-45% of patients undergoing CT Chests had potentially clinically significant cardiovascular findings and coronary calcifications that went unreported. >7.1 million 6 mm lung CT scans are done annually for other clinical indications Berrington de Gonzalez A et al. Arch Intern Med 169:2071 2077 Woodring JH et al. Australas Radiol. 1989 Feb;33(1):79-83 Sverzellati et al. Radiol Med 2016 Mar;121(3):190-9 Johnson C et al. AJR Am J Roentgenol. 2014 Apr;202(4):725-9 CAC >0 was present by expert reader interpretation in 108 of 201 (53%) nongated noncontrast CT scans in patients without suspected CAD But was reported by the radiologist in only 69% of the 108 positive scans CAC scores on standard 6 mm chest CTs are strongly correlated with 3 mm ECG-gated CTs Similarly predict mortality in community-living individuals. Uretsky S et al. Am J Cardiol 115:1513 1517. Hughes-Austin JM et al. JACC Imaging 2016; 9(2) A Question: Chest CTs performed for other clinical indications may provide an untapped resource to garner CVD risk information without additional radiation exposure or expense. Is there another test that almost half the population gets on a yearly basis that can help us risk stratify our patients? Hughes-Austin JM et al. JACC Imaging 2016; 9(2) 18

Mammography to Screen for CAD? There is a strong quantitative association of Breast Arterial Calcification seen on Digital Mammography with CAC. positive predictive value of nearly 70%-80% for identifying women with presence of CAC BAC is superior to standard cardiovascular risk factors. BAC is equivalent to both the FRS and PCE for the identification of high-risk women A Thought Experiment Assume a conservative estimate of 10% of mammograms with Breast Arterial Ca 2+ Approximately 4 million women nationwide undergoing screening mammography will exhibit Breast Arterial Ca 2+ 2 to 3 million of them will likely have signs of premature CAD. Margolies L et al. JACC Cardiovasc Imaging. 2016;9(4):350-60 Mostafavi L et al. PLoS One 1:e0122289 Margolies L et al. JACC Cardiovasc Imaging. 2016;9(4):350-60 Mostafavi L et al. PLoS One 1:e0122289 Conclusion Risk Calculators are useful for populations Individualized tests like Calcium Scoring tell you accurate, prognostic information about the patient in front of you They are cost effective Minimal radiation Motivating Final Thoughts Editorial Comment #1 Population Medicine Population vs. Personalization Do we even need risk assessment at all? 19

Population Medicine: Empiric Treatment of all patients? According to the 2013 ACC/AHA guidelines, everyone with optimal levels of risk factors will qualify for statin therapy if they live long enough (using 7.5% threshold). AA men at age 66 AA women at age 70 Non-AA men at 63 Non-AA women at 71 Some have suggested universal therapy with statins starting at age 55yo Wald NJ, Law MR BMJ 2003;326:1419 Population Medicine In the setting of low cost/well tolerated statins, there is little role for Calcium Scoring Currently Calcium Scoring has a Class IIb recommendation as a result Editorial Comment #2: A Curiosity of our Culture Atherosclerotic vascular disease accounts for more death and disability than all types of cancer combined Cancer screening is recommended, calcium scans are not Colonoscopy: 0.1-0.3% serious adverse event rate $4.9 Billion on Cancer Research by NCI annually $2 Billion spent on CV research by NIH annually Pletcher et al. Circ Cardiovasc Qual Outcomes 2014;7:276-84 Fisher et al. GI Endoscopy; 74(4):745 The End The greatest danger to any society is a failure to embrace new ideas Franklin Delano Roosevelt 20